Professional Documents
Culture Documents
http://hdl.handle.net/2027/uc1.b4118385
UNIVERSITY OF CALIFORNIA
LIBRARY
DAVIS
REACTOR ANALYSIS
Consulting Editor
Bonilla
Nuclear Engineering
Hoisington
Nucleonics Fundamentals
Price
Schultz
Reactor Analysis
Stephenson
REACTOR ANALYSIS
ROBERT V. MEGHREBLIAN
Oak Ridge National Laboratory
DAVID K. HOLMES
Toronto
London
1960
LIBRARY
UNIVERSITY OF CALIFORNIA
DAVIS
REACTOR ANALYSIS
Printed
Copyright 1960 by the McGraw-Hill Book Company, Inc.
This book, or
All rights reserved.
in the United States of America.
parts thereof, may not be reproduced in any form without permission
of the publishers.
Library of Congress Catalog Card Number: 59-15469
THE MAPLE PRESS COMPANY,
41328
YORK,
PA.
PREFACE
is,
placed on
is
is
is
it
is
it
dis
it
is
is
is
is
vi
PREFACE
In
many
instances
cal examples.
Finally, the authors thank Mrs. Yvonne Lovely for her expert assist
ance in the preparation of the manuscript.
Robert
David
V. Meghreblian
K.
Holmes
CONTENTS
v
Preface
Chapter
1.1
1.2
1.3
1.
Introduction
Chain Reactions
Nuclear Reactors
The General Reactor Problem
Chapter 2.
Probability
1
17
22
29
2.1
29
2.2
31
Chapter 3.
3.1
3.2
3.3
3.4
Multiplication
Collision Density
Neutron Balance
Infinite-medium Criticality Problem .
to Safety Storage Calculation
Application
49
50
51
55
of a Chemical-processing
Plant
60
Neutron Flux
Problems
Slowing-down
40
50
3.5
Chapter 4.
36
63
68
69
4.1
69
4.2
85
Chapter 6.
5.1
5.2
5.3
5.4
5.5
5.6
98
102
112
118
125
148
157
160
160
180
189
198
223
230
viii
CONTENTS
237
244
5.8
Chapter
6.
262
Reactor
268
268
6.2
273
Chapter
7.1
7.2
7.3
7.4
7.5
7.6
7.
302
308
321
327
Transport Theory
330
Introduction
The One-velocity Transport Equation
The Boltzmann Equation with Energy Dependence
Methods of Integral Equations: One-velocity Model
Applications of Spherical-harmonics Method
Computational Methods for the Neutron Age
Chapter 8.
Generated for wjivans (University of Florida) on 2015-09-22 17:17 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
286
330
331
352
366
387
399
Reflected Reactors
Introduction
The One-velocity Model
8.3
Reactors with Spherical Symmetry : Serber- Wilson Condition
The Two-group Model
8.4
Sample Computation Using Two-group Model
8.5
Completely Reflected Systems
8.6
Feynman-Welton Method
8.7
Multigroup Treatment of Reflected Reactors
8.8
Problems
418
418
8.1
422
8.2
Chapter 9.
9.1
9.2
9.3
9.4
9.5
9.6
456
470
476
486
520
543
Reactor Kinetics
546
Problems
Chapter 10.
10.1
10.2
10.3
10.4
10.5
10.6
440
547
556
566
577
590
610
624
Heterogeneous Reactors
626
626
645
660
692
698
.
704
ix
CONTENTS
Chapter
11.1
11.2
11.3
11.4
Control-rod
Theory
721
Chapter
12.
Hydrogenous
Chapter
13.1
13.2
13.3
13.4
13.5
13.6
13.
Perturbation
727
738
743
748
12.2
12.4
721
Systems
12.1
12.3
11.
....
.
Theory
748
750
752
759
763
Theory
Medium
of the Bare Reactor
of a General System
....
....
763
771
776
781
785
787
Name Index
791
Subject Index
795
CHAPTER
INTRODUCTION
1.1
An understanding
of the properties
and behavior
of nuclear
chain
various
nuclear
the rates of the reactions involved in the chain are directly dependent
upon the density of the chain carrier, in the present case the neutrons.
Thus in order to determine the various properties of a reactor, such as
the power-production rate and the radiation-shielding requirements, it is
necessary to obtain the fission reaction rate throughout the system and,
In fact, all the basic nuclear
therefore, the neutron-density distribution.
and engineering features of a reactor may be traced back ultimately to a
knowledge of these distribution functions.
The subject of reactor analysis is the study of the analytical methods
and models used to obtain neutron-density-distribution functions.
Since
these functions are intimately related to various neutron-induced nuclear
reactions, a knowledge of at least the basic concepts of nuclear physics
is essential to a thorough understanding of reactor analysis.
of nuclear
It is intended
reactor physics and the analytical methods of attack.
that the last section be used primarily for purposes of review and to aid
the reader in orienting the various topics with regard to the over-all
structure
REACTOR ANALYSIS
the
physical
features
[CHAP.
as a convenient
of neutron
phenomena
reference for
relevant
to
symbolically by
XA + n1-*
It
may be represented
(1.1)
{XA+1}*
thus,
XA +
(X^1)* XA+l + Ey
(1.2)
Now consider the situation in which the nucleus XA+l in the unexcited
(ground) state acquires an energy Eh just large enough to separate a
neutron.
The appropriate reaction would be
XA+l +
E-*
{XA+1}*
XA + nl
(1.3)
SEC.
INTRODUCTION
1.1]
that
Ey = Eb
(1.4)
The binding-energy
~c2(Am-A/)
- Nn) - M]
= c2[JVn(m
mp)
+ mpA
- M]
(1.5)
where the approximation follows from the fact that mn ~ mp. In order
to obtain the average binding energy per nucleon Eb, we divide through
by
^ ~ c*(mn - ^
(1.6)
This is the average energy required per nucleon to separate the nucleus
The value of Eb varies from about 1 to
into its constituent particles.
In the mass range of interest to
9 Mev over the entire mass scale.2
reactor physics (A > 70), Eb decreases monotonically from 8.6 Mev at
A = 70 to 7.5 Mev at A = 238. It is this variation which determines
It is shown later
two fundamental features of the fission reaction.
that this indicates (1) that there is a positive energy release if a nucleus
with A > 85 is caused to disintegrate and (2) that nuclei in this range are
theoretically unstable with regard to the fission process.
Consider first the question of the energy released when a nucleus
As an example, let us take the case of the neutrondisintegrates.
induced fission of the nucleus XA which results in the formation of two
On the basis
fragments YAl and ZA' of masses M i and M2, respectively.
of the compound-nucleus concept, this reaction may be written
nl +
(1.7)
ZA>
is,
In general, the binding energies of the fragments will differ from the
the combined masses
binding energy of the original nucleus; that
of the fragments will not equal the mass of the fissioned nucleus XA+i.
The difference appears as an energy release Es which may be determined
mv
Eb m
Fig.
3-5, p. 63,
McGraw-
REACTOR ANALYSIS
E,
In the present
- (Mx + Mt)]
cWc
[CHAP.
case
(1.8)
- =U
(1-9)
Ef ~
{A
- ] + AJfi* - El]
l)[Ep
(1.10)
Mass number A
Fig.
1.1
Fission energy
of the mass number.
is
2),
Precisely
relation
is
could
conceivably
shows, however,
undergo
Experience
spontaneous
disintegration.
fission does not occur at anything
that spontaneous
SEC.
INTRODUCTION
1.1]
like the rate one might be led to expect on the basis of the fission ener
The explanation rests with the fact that a nucleus cannot
getics alone.
disintegrate until it has first acquired a certain activation or threshold
Thus there exists an energy barrier between the state of the
energy.
whole nucleus and the fragmented state.
If
r2
1.2
radiation.
r,+
Fig.
it,
J.
elementary
treatment
66, 426
(1939);
J.
Frenkel,
given by D. Halliday,
"Introductory
Sons, Inc., New York, 1950.
&
'An
J.
Phys.
(1939).
is
N. Bohr and
(U.S.S.R.),
125
1,
max
Nuclear
REACTOR
ANALYSIS
[CHAP.
nucleons and the electrostatic repulsion due to the charge on the protons.
When the nucleus is set into oscillation, any departure from its original
shape results in an increase in its potential energy because of the "surface
tension."
Such distortions tend, however, to separate the proton popu
lation and thereby create centers of electrostatic repulsion; these forces
the potential energy of the system and increase the distortions
If the distortional oscillations lead to a "dumbbell-like"
further.
decrease
instant of separation.
If
Nuclear Fuels.
b.
If
M{A,Z)
the mass of
is
is
SEC.
INTRODUCTION
1.1]
nucleus containing
mass units,1
M(A,Z)
= 0.99395.4
0.00084Z + 0.0141
A' +
0.000627^
0.083
2)2
(A/2A~
&
It
5,
the
is defined
when A is odd
= 0
0.036
A*
(1.11)
0.036
Ai
(1.12)
Eb =
c\M
mn-
Mc)
(1.13)
to thermal
In fact, the
REACTOR
ANALYSIS
[CHAP.
than the threshold energy (~7.1 Mev). Evidently, neutroninduced fission is not possible with these nuclei when the neutron kinetic
This limited fissionability makes these
energy is less than 1.8 Mev.
isotopes much less attractive as the primary fuel component for a steady
chain reaction, although their presence in a chain supported by any of the
"thermal fuels" can lead to significant augmentation.
The energy released from the fission reaction of any of the isotopes
This
mentioned above may be computed from the equation for Ef (1.8).
expression, which applies for the two-fragment divisions, will suffice for
is less
most situations
of interest
since a three-fragment
A rough
a very low probability.
estimate of the fission energies generally available from such nuclei
The resulting
may be obtained from a sample computation for U236.
value will be representative since the various mass numbers differ but
little, and the principal dependence on binding energy occurs in the
ni + u23i-->
|U236
)*-* Kr"
Ba142
(1.14)
The average binding energy per nucleoli Eb for a nucleus of mass num
ber 236 is about 7.5 Mev, and for nuclei of mass numbers 94 and 142,
8.6 and 8.3 Mev, respectively.
The approximate relation (1.10) yields
for E,:
Ef
= (236) (8.6
Thus, roughly
200
7.5)
+ (142) (8.3
Mev is released in
- 8.6)
= 217
Mev
a heavy nucleus.
interactions
fer of some of the kinetic energy of the fragment to the struck nucleus.
A series of such collisions eventually slow the fragments to thermal
equilibrium with the environment.
Approximately 85 per cent of the
fission energy is liberated in this way and must be removed by a suitable
Halliday,
SEC.
INTRODUCTION
1.1]
The mass
the target nucleus and the energy of the captured neutron.
distribution of the various fragments resulting from the fission of U235,
For a given nucleus, the shape
U238, and Pu239 is shown in Fig. 1.3.
of the yield curve is highly sensitive to the energy of the incident neutron.
The figure shows clearly the characteristic grouping of the fission frag
ments into two distinct regions of the mass scale for the case of a thermal
It is seen that such reactions most frequently
neutron-induced fission.
produce two fragments, one of mass number around 95 and the other
Although other subdivisions can occur, their
of mass number 140.
very improbable.
Of these alternative schemes, the division
of the compound nucleus into two fragments of equal mass occurs in
likelihood
is
Table
1.1
Energy, Mev
165 + 15
fission
197 15
REACTOR ANALYSIS
10
[CHAP.
Of
>
Mass number
Fig.
Yields from fission of U2", U"8, and Pu2" versus mass number A.
[Re
printed from C. DuBois Coryell and N. Sugarman (eds.), "Radiochemical Studies: The
Fission Products," Part 3, National Nuclear Energy Series, Div. IV, vol. IX, McOrawHill Book Company, Inc., New York, 1951; Uin from Paper 219 by Steinberg and
Freedman, and Uut and Pu*n from Appendix B.]
1.3
SEC.
INTRODUCTION
1.1]
11
the distinction between the neutrons that appear instantly and those
which appear at later stages in the decay processes of the fragments.
If we denote by v the total number of neutrons which appear from fission,
then it is clear that the nuclei Y and Z will contain fewer neutrons by
exactly
this amount;
we now
therefore,
write
reaction
(1.7) in the
form
representative
n1
\XA+l\*-^
YA*
tj236
(U236)
m1
(1.15)
we might have,
for example,
ZA~A>-'+1
- Kr93 +
Ba140
3n
(1.16)
We have assumed here that three neutrons are emitted by the fragments.
The number of neutrons emitted from fission is dependent upon the
In first approximation, it is permissible
energy of the incident neutron.
to neglect this dependence and to assume that the average number of
neutrons produced per fission is a function only of the target nucleus.
It should be recognized that this is not a true picture and that, in fact,
Table
1.2
Neutron
Fissionable
nucleus
Us
Un
2.54
2.46
2.88
Pu"
0.04
0.03
0.04
2.31
2.08
2.03
0.03
0.02
0.03
the average number of neutrons increases with the energy of the incident
For many reactor calculations, however, this dependence is
particle.
neglected and some average number is selected for the entire energy
range.
for
a given fissionable
by any one reaction is of little interest in reactor physics, and one would
certainly prefer an average value which could be assigned to a large
number of such reactions.
Measurements of this type have been made,
and the presently acceptable values of v (the average number of neutrons
per fission) for the three principal nuclear fuels are listed in Table 1.2.
c. Nuclear Reactions and Neutron Economy.
So far in our discussion
of the chain-reacting system, we have focused attention on the fission
Fission is but one of several neutron-induced nuclear reactions
reaction.
REACTOR
12
ANALYSIS
[CHAP.
(2)
XA +
n1
inelastic
processes
are(l)
The radiative-capture
scattering.
by the equation
\XA+l)*-* XA+l + Ey
(1.2)
The symbol Ey here denotes energy liberation in the form of gamma rays
(electromagnetic radiation of wavelength in the order of 10-10 cm) hav
ing about 1 Mev energy. In this process, a neutron is captured by the
fissionable material and forms a compound state.
However, instead of
releasing its excess energy by fissioning, it does so by the emission of some
This emission may remove all
radiation immediately after the capture.
the excitation energy of the compound nucleus and thereby leave it in a
If this initial emission leaves the nucleus with some
state.
residual energy, further emissions must eventually follow until a stable
stable
configuration is achieved.
reaction could be written
For the
case of U235,
ni + u5 |U2!6!*->
the radiative-capture
+ Ey
U236
(1.17)
ically remove neutrons from the system, it merely degrades (as a rule)
the kinetic energy of the neutron population.
For the nuclei of primary
to reactor technology (heavy nuclei), the inelastic-scattering
process is conveniently described with the aid of the compound-nucleus
In these reactions the nucleus "captures" a neutron at one
model.
interest
it at another; thus,
+ XA-> {XA+i\*-> \XA\* +
A compound nucleus
is not
always
formed,
n1
however;
(1.18)
some inelastic-
by
nl + XA-> \XA}* +
n1
(1.19)
SEC.
1.1]
collision
INTRODUCTION
13
occupy
an increase in the internal energy of the nucleus, is precisely the energy
Thus, the sum of the kinetic
loss suffered by the incident neutron.
energies of the emitted neutron and the struck nucleus (after collision)
will differ from the total kinetic energy of the system before collision by
exactly the amount of the excitation energy of the residual nucleus.
A third nonproductive neutron-nucleus reaction which is of some conse
Of
potential-type scattering extends over the entire energy spectrum.
primary importance to the present discussion is the fact that the nuclear
fuels are generally of large nuclear mass, and elastic collisions with
neutrons produce little change in the kinetic energy of the neutron.
Thus, elastic-scattering collisions with fuel nuclei effect little or no
change in the neutron population, either by altering the neutron energies
or by serving as neutron removers.
Inelastic scattering and radiative capture are the two processes which
compete with the fission reaction.
The extent of this competition is
determined by the relative magnitudes of the rate at which nonproduc
tive reactions occur as compared to the rate at which fissions occur. If
the nonproductive reactions for a particular fissionable material constitute
only a small fraction of all neutron-nucleus reactions, then the possi
bilities of this substance as a nuclear fuel are greatly enhanced. The
deciding factor, however, is the number of neutrons released by fission.
a given substance, the likelihood of fission taking place may be very
For
In the laboratory system of coordinates. The kinetic energies in the center-ofmass system are conserved.
1 See, for
example, Halliday, op. cit., pp. 396-402; also, Blatt and Weisskopf, op. cit.,
chap. VIII, especially pp. 325-329.
1
REACTOR ANALYSIS
14
[CHAP.
high, but for the purposes of a chain-reacting system, this is of little value
if on the average the number of neutrons released per fission is negligibly
Thus it is the combined characteristics, reaction rates of non
small.
productive processes and number of neutrons per fission, which determine
the merits of a fissionable material as a nuclear fuel.
The focus of attention in reactor analysis is upon the fate of the
neutrons produced by the fission reactions, rather than upon the fission
process proper with its resultant fragments and electromagnetic radia
The
core:
(if
coolant materials.
3. The shield: An outer shell which serves as a stopping medium for
the biologically dangerous radiations from the core and the reflector.
any) constitute the principal
The core and the reflector region
is
is
is
is
is
is
SEC.
INTRODUCTION
1.1]
15
be used to describe
not strictly true since there are such nuclear reactions as the (n,2n) reaction
can serve as secondary
REACTOR ANALYSIS
16
[CHAP.
will escape from the system, and the remainder will be captured by vari
Thus absorption or escape is the only eventuality open to a
ous nuclei.
neutron.
If
we define
nuclei
=
fraction
of neutrons which experience radiative capture by nonfo
fuel nuclei
then of the available n neutrons, nfa are absorbed and nf. escape; there
fore,
1
where, by definition,
/. + /.
(1.20)
/.-//+/+/.
(1-21)
Of the
nfa neutrons absorbed, nff cause fission, and from these fissions,
neutrons
are emitted.
If the population n is to remain at steady
vnff
state, we must require that
(1.22)
/.+/.
This relation may
be
/,+//('/.+/.
For convenience,
(1.23)
(1M>
=
-*
a
(1.25)
T,
a-p-P-T^M
(1.26)*
The importance of this relation rests with the fact that for
a given fuel,
the average number of neutrons per fission, must be at least as large
os 1 + a if the chain reaction is to be self-sustaining.1
Thus, the number
v,
reactions.
sources,
i.e.,
no sources
SEC.
INTRODUCTION
1.2]
17
It should
on the right in (1.26)].
be noted that the fraction of neutrons which escapes from the reactor is
At best, / = 0, a situation which
dependent on the size of the system.
Clearly, even for an infinite
is possible only in an infinite geometry.
system, some losses (radiative capture) still occur.
Much of the attention in reactor analysis is devoted to the calculation
A knowledge of these quanti
of the fractions /a, / etc., defined above.
distributions
ties forms the basis for determining the neutron-density
will
later
derive
suitable
for
procedures
computing
We
in the reactor.
these quantities, not only as a function of the spatial coordinates, but
properties
collisions ;
What
we
a neutron
may have and the probability of each such life history ; then
we may take
an average over these probabilities, to find some macroscopic property
of the system which is of interest.
Inasmuch as these problems involve
large numbers of particles and time intervals long in comparison with
their average free flight time (time between collisions), the deviations
in the behavior of the assembly from the average expected behavior
will be very small. So long as these conditions hold, we may replace
the actual assembly by a model in which all particles behave according to
This approach is used over and over again in
some average pattern.
reactor analysis; consequently, the phrase "on the average" is heavily
worked and becomes such a burden that it is often omitted as understood.
1.2
a. Classification
Nuclear Reactors
of Nuclear Reactors.
18
REACTOR
ANALYSIS
[CHAP.
SEC.
INTRODUCTION
1.2]
19
are born into the system with energies of a few Mev, but are subsequently
slowed to thermal speeds by a succession of elastic- and inelastic-scatter
ing collisions with the surrounding nuclei. Thus neutrons are distributed
If one were
over the entire energy scale from thermal to several Mev.
1.4 shows
REACTOR ANALYSIS
20
[CHAP.
(a)
(b)
(c)
Neutron energy
Fig.
1.4
Neutron-energy
reactors.
(6)
intermediate,
be grouped
or industrial areas.
Package-power applications, however, refer to power
In this classification, one might include
production for limited facilities.
power plants for areas in unusual climatic conditions, advanced bases,
or small, isolated establishments.
The application of nuclear reactors
to power mobile units can conceivably include any device which is
designed for terrestrial operation or, for that matter, space craft as well.
The possibilities in this area of application are as yet hardly known.
The application of nuclear reactors for research purposes includes the
production of high-intensity neutron and radiation fields and the produc
tion of radioisotopes.
The use of radiation and neutron fields for experi
mental research in physical and biological sciences is well established,
and the construction of nuclear reactors for use in university and research
laboratories is already under way.
The production of radioisotopes has,
likewise, become an important application of nuclear reactors, no doubt
stimulated by the considerable success of radiotracer techniques.
SEC.
INTRODUCTION
1.2]
21
The third, and the earliest, application of nuclear reactors was for the
The first of these devices were the
production of fissionable materials.
Hanford reactors which were built for the production of plutonium by
Nuclear reactors which
the conversion of the U238 in natural uranium.1
material are generally classified as breeders or con
has become the custom to distinguish between these two
reactor types by designating as breeders those reactors which produce
fissionable material of the same kind as the actual nuclear fuel which
produce
verters.
fissionable
It
is
is,
natural uranium as the fuel. The important fact in any case is that some
fissionable material is produced from the fertile material added to the
The U238 in the Hanford reactors is such a material, as is Th232
reactor.
In either of these fertile materials, the nuclear
yields
(which
U233).
is
involved
the
capture of neutrons by the nucleus and the subse
process
Although breeders and converters
quent formation of fissionable nuclei.
provide only a portion of the fissionable material required in present
reactor technology, they are, nevertheless, an important source. Another
application of these reactors which is being given increasing attention in
recent times is the combined production of fissionable material and power.
The idea of a power source which is capable of producing its own fuel
of course, highly attractive from the economic standpoint. Although
many such devices have been proposed, their practicability
not yet
entirely established.
Many questions in both nuclear physics and
economics must be answered before such applications can become engi
neering realities.
In
is
is
interdependence
to the nuclear properties of the reactor constituents.
A simple illustra
tion of this point
found in the case of the thermal reactors.
Great
pains are taken in the design of these systems to obtain efficient moderator
materials
reason
fission
to thermal energy.
The
that the fission characteristics
neutrons
is
is
desired
of the principal fuels are superior in the thermal-energy range. As will
be shown later, this results in
minimum requirement in fuel mass.
would be important in a design situation which
limited by the scarcity of nuclear fuels.
We have here another reason
for the selection of the thermal-reactor type for the earliest designs.
The radiative capture of a neutron in U"8 yields
is
This consideration
they consume.
Pu"'
[cf.
Eq.
(1.17)].
22
REACTOR
1.3
ANALYSIS
[CHAP.
and time.
a reactor
usually
various
confronted.
It
is,
likewise,
to perform.
The nuclear fuel mass required to sustain a chain reaction in a reactor configuration
of specified composition.
SEC.
1.3]
INTRODUCTION
23
A knowledge of these
and of paramount interest to the reactor analyst.
distributions determines the design of the cooling system and provides a
Once the density of neutrons
detailed count of the neutron population.
known at every point in
of fuel breeding and conversion, and the flow of neutrons out from the
This last factor determines, in part, the shielding
surfaces.
requirements and the intensity of neutrons beam which may be available
for research purposes.
The problems of control effectiveness deal with those considerations
reactor
which enter into the design and the operation of the nuclear elements
which control the neutron population.
The mechanical configuration
forms;
of these elements may take many
the determining factor is
usually design and operating convenience.
The customary device is a
rod or some other cylindrical unit which contains a material that is a
By some appropriate motion of the unit, this
strong neutron absorber.
material is brought into a region of the reactor which contains a high
By either withdrawing or inserting this neutron
density of neutrons.
sink, the neutron balance can be quickly altered and the population
caused to rise or fall, thereby accelerating the chain reaction or shutting
it off. Other control devices that have also been successful include the
1 In the order of 10 to 15 per cent accuracy.
REACTOR ANALYSIS
24
[chap.
tions in the neutron population and power production, but these quickly
The prediction
decay and the system returns to steady-state operation.
of the thermal response of the reactor to these disturbances belongs to a
All these involve questions of dynamic
related class of problems.
response and stability, and in the majority of the cases of engineering
interest
1 The
Except in the
case of
fissions or not.
SEC.
1.3]
INTRODUCTION
25
of the controls (shim rods) which will introduce additional fuel into the
reactor and thereby maintain the chain reaction.
The rate of fuel consumption and the amount of control adjustment
available
analyst.
for this purpose are among the problems which confront the
Closely related to the fuel-burnup problem is the production
convenient
information will serve best as an outline to locate and orient the various
ideas and methods within the over-all structure of the subject.
The general field of problems described above, except in some special
areas, may be treated by the well-known methods and analytical models
of mathematical
It has already been noted that the most
physics.
general description of the neutron population usually starts with a
neutron-balance relation of the Boltzmann type.
The Boltzmann equa
tion was developed in connection with the study of nonuniform gas
mixtures, and the application to the neutron problem represents a con
siderable simplification of the general gas problem.
(Whereas in gas
problems all the particles are in motion, in reactor problems only the
neutrons are in motion.1)
The fundamental equation of reactor phys
ics, then, is already a familiar one from the kinetic theory.
Further,
many of the most useful neutron models obtained from approximations
to the Boltzmann equation reduce to familiar forms, such as the heatconduction, Helmholtz, and telegraphist's equations.
These simplifica
tions result from the elimination of various independent variables in the
1 This
is not strictly true because the nuclei themselves are in motion, but in many
problems the neutron speeds are very great, and nuclear motion can be neglected in
comparison.
REACTOK ANALYSIS
26
[CHAP.
is,
of course, that the geometric confines of the real problem are ignored and
attention
focused on other aspects of the situation.
The infiniteis
is
is
is
it
is
is
is
is
is
it
is
SEC.
INTRODUCTION
1.3]
include
a reflector.
the solution
27
at best, require
problem in order to
of a two-region
If
taken into account, the Fermi age model may still be used; however, the
analysis of multiregion systems which include the neutron-energy spec
The
trum usually requires the application of numerical methods.
so-called mulligroup methods have been developed for this purpose.
Many of these methods are based on the continuous slowing-down model
One very useful model which has been devised for
Various forms
treating multiregion reactors is the two-group model.
of this model have been used for analyzing reactors of both the thermal
of Fermi age theory.
encountered
in reactor analysis
is
In a
(e.g., the core, the reflector) is a homogeneous medium.
heterogeneous reactor, the core consists of a lattice of many small cells,
each of which contains both moderator and fuel.
The moderator and
the fuel exist as separate regions; however, the scale of the nonhomo
An
geneity is much smaller than in the general multiregion problem.
analysis of this type of reactor requires a detailed study of the nuclear
itself
characteristics
determine the space and the energy distribution of neutrons in these cells.
From this information, he can readily predict the over-all nuclear char
semiempirical
equation.
All
models or higher-order
Exponential
REACTOR ANALYSIS
28
[CHAP.
only
problems which include the coupling between the reactorpower output and the neutron population will, in general, involve a
nonseparable relationship between the time and the space variables.
However, in many instances, the important features of the problem can be
determined by using simple physical models which do separate out the
Such solutions will not yield an accurate
space and the time variables.
dependent
description of the neutron physics, but they can give valuable insight
into the basic characteristics of the system in question.
The analytical treatment of reactor-physics problems is derived in
main part from the fundamental Boltzmann equation.
These models
of this equation to various degrees of accuracy.
are approximations
Which model is selected for a particular reactor problem is determined
by the accuracy required and the nature of the phenomenon to be
When the problems deal with the bulk, or integral, neutron
analyzed.
characteristics of the reactor, relatively crude models can usually be
Nuclear-fuel-mass
calculations fall in this category.
On the
applied.
This
other hand, more detailed studies require more elegant methods.
relatively
of
the
small
dimensions
which
are
is a consequence
involved
In these situations, many of the simplifying assump
in these problems.
tions about the neutron-density distribution and the composition of the
media (such as isotropy) are no longer valid, and more precise descrip
Problems which involve regions close to material
tions must be used.
boundaries and heterogeneous cell calculations are problems of this type.
Accurate descriptions of the physical situation in these problems can be
This is not to say, however, that
given only by the Boltzmann equation.
Again, cruder
the Boltzmann equation must be solved in all such cases.
models have been devised which can be applied even to these problems,
and previous experience can always reduce appreciably the work required
One of the purposes of the present work
to obtain adequate solutions.
is to explain some of these approximations which have proved to be use
ful and to demonstrate their application to typical situations.
CHAPTER
2.1
Neutron-Nucleus Interactions
of the problems of
reactor physics require
instance a description of the average
behavior of the neutrons in the reactor.
The average behavior is
obtained through an understanding of the probabilities of the various
a.
Detailed
analyses
in each
neutron-nucleus
or will increase.
trajectory of
If
will either remove the neutron or alter its energy and direction.
1 In reality, this will seldom be the
case.
In most situations, the reactor will be
surrounded by the atmosphere, and some neutrons will always be scattered back to
the reactor by collisions with air molecules.
But, these events are negligibly few in
number and can be ignored in so far as their effects on the chain reaction are concerned.
* The range of nuclear forces is in the order of
10" cm.
29
ANALYSIS
REACTOR
30
[CHAP.
For the present, we will not concern ourselves with the details of the
various collisions which can occur. It will suffice to distinguish between
If the neutron is scattered on its first collision,
scatterings and captures.
Again,
it will lose some energy and will move on in a new direction.
will
is
that
this
new
direction
allow
the
neutron
to
possibility
there
a
reach the boundary of the reactor without further collisions and so escape
from the system. On the other hand, if a collision occurs, there is once
Fig.
2.1
This scattering process will continue until the neutron is eventually lost
to the system.
For any single neutron, the spatial trajectory will be
some complicated series of broken straight-line paths which extend from
its point of birth to the point at which it is eventually absorbed or leaves
the boundary of the reactor (refer to Fig. 2.1).
Number of scatterings
Fio.
2.2
2.2.
Note that each scattering collision causes a jump change in the
neutron energy.
As we have already mentioned, the history of any one
neutron is of little importance to the over-all neutron problem.
We are
concerned only with the average behavior of the neutrons, and this can be
sec. 2.2]
PROBABILITY
CONCEPTS,
CROSS SECTIONS
31
Fission
The various scattering interactions
2.
are included
general heading:
Scattering: All neutron-nucleus reactions which produce a neutron and
a nucleus that differ from the reactants only in their energy content
(kinetic and/or internal). Scattering reactions are of two types:
Elastic scattering: The products of the reaction are a neutron and the
struck nucleus in its original internal-energy state.
2. Inelastic scattering: The products are a neutron and the struck
1.
2.2
a. Average
nucleus
Probability of Collision
The calculation of the number of neutronthat occur per unit volume per unit time at each
Behavior.
collisions
neutrons.
1 The
In any calculation
numbers
nuclei are generally in motion, and the relative speed of the neutrons is given
by the magnitude of the vector difference between the velocity of the neutron and
that of the nucleus.
32
REACTOR ANALYSIS
[chap.
1 dn(x)
dx n(x)
= 2
(2.1)
is easily seen to be
n(x) = ntfrZx
(2.2)
(2.3)
SEC.
PROBABILITY
2.2]
CROSS SECTIONS
CONCEPTS,
33
The physical content of Eq. (2.2), that the number of uncollided neutrons
falls off exponentially with distance, may be subjected to experimental
verification directly, and the fact that the form of (2.2) is found to fit
the data very well is evidence for the validity of the assumption expressed
byEq. (2.1).
The result (2.2) may also
The number of
neutrons
the number
of neutrons
surviving to
starting
distance x
at x = 0
will
not
suffer a collision
|_in traveling distance
(2.4)
x_
If
a neutron
a neutron
flight of =
will suffer
a first collision
probability that
a neutron
Further,
- w(x)
the
before
first collision
(2.5)
e~Xx
e~Zl
(2.6)
fix)
^-Fix)
ax
= Xe-Z*
(2.7)
condition,
composition,
REACTOR ANALYSIS
34
[CHAP.
is
is,
along its path before it encounters a nucleus is now measured from the
the neutron
That
as good as new and can be considered
position x.
The mathematical statement of this
to have originated at point x.
F(x)
probability that
(2.8)
<
PU
X)
is
is
is
where
a random variable and F(x)
the corresponding distribution
function. For this calculation, we require also the conditional probabil
defined as
ity function. The conditional probability
between x and x +
it
x),
occurs
(2.9)
= occurs after x(
>
y\ >
x)
< x
P(x <
y\t
reaches
before
collision
it
makes
[i.e.,
chaps.
y)
P(
<
>
x)
<
P(x
<
Or,
Eq. (2.8)].
first observed at
is
it
if
is
The "as good as new" hypothesis can be stated with the aid of these
known to have survived to x, then the
Thus,
the neutron
moves an additional
probability that the first collision occurs before
neutron
identically the probability that
distance
[i.e., Eq. (2.9)]
functions.
13,
(2.10)*
14,
sec.
PROBABILITY
2.2]
35
statement
from which
CROSS SECTIONS
CONCEPTS,
P(*
<
*f
V>
* >
*>
(2.11)
The function P(x < < x + y, { > x) is the joint probability that the
collision occurs in the interval (x, x + y) and that the collision point is
If we substitute this relation into (2.10), we obtain
beyond x.
Pit
P(x
<
< x + y,
> x) =
P(x
<
< x
y)
(2.12)
F(y){l
It
is convenient
distribution
- F{x)]
we suppose that,
for
each x,
= l-
y)-
[F{x + y)
Fix) has
We apply the
functions.
a difference
i F(y)[l - F(x)\
If
F(x +
F(x)
(2.13)
equation.
If
- Fix)]
a derivative,
we divide
(2.14)
of (2.14) tends to F'(x) as y tends to zero; the left-hand side must there
fore also converge.
Thus we find
S[l
- Fix)]
F'ix)
(2.15)
lim
v o
^
y
= F'(0) = S
As noted, this limit must exist, and since it does exist, F(0)
the identification (2.16).
The solution to Eq. (2.15) is
Fix)
(2.16)
= 0, hence
- e-s*
Ibid., p. 166.
REACTOR ANALYSIS
36
[CHAr. 2
F(x)
- e-s*
2e-Sl
(2.17)
fjf(w)dw
(2.18)
f(x)
f*^f(w)dw
The general form of these functions is shown in Fig. 2.4 where, for
convenience, the parameter 2 is taken as unity. The origin of the x axis
is the point at which the uncollided
neutron is first observed. The func
tions F(x) and f(x) are identically
zero for all x < 0. The distribution
function F(x) approaches unity in
the limit as x becomes arbitrarily
large, i.e., for very large x, the prob
o
ability that the neutron has made
frequency
a collision
is nearly
certainty.
f(xi) dx
(2.19)
It
is easy to show that f(x) is indeed the probability of collision per unit
path of travel around x by means of the following computation: The
probability that the first collision per unit path traveled occurs around x
is given by
< x + Ax)
Ax
ax^o
< < x + Ax) = F(x + Ax)
- F(x)
< + A*)
- Fix)
lim P(x
But,
P(x
<
Therefore,
Um
P(s<f
AX
Al-0
2.3
Hm
AI-.0
Fix + Ax)
Ax
Eq. (2.1)
is
SEC.
PROBABILITY
2.3]
Jfm
x =
CROSS SECTIONS
CONCEPTS,
dx = 2
xf(x)
JO
37
dx = ^ =
2i
(2.21)*
(2.22)*
may
\_dn
dx n
= 2 =
neutrons at x.
Or, rewriting
(2.23)
Since the derivative at the left of this relation gives the change in the
neutron density n(x) over the distance dx, it must correspond also to
the number of collisions which occur in dx. Thus a verbal statement
of (2.23) might be as follows
Number of
number of
number of collisions
collisions in
uncollided
traversing dx}
neutrons at x
unit path
(2.24)
If
Ae-A< dt
p. 174.
is obtained by integrating
g(t).
dt
(2.25)
REACTOR ANALYSIS
e-A<
t,
is
- G(t)]
= JV(0)[1
N(t)
nuclei present at
If
g{d)dd
[CHAP.
(2.26)
G(t) =
J'_m
38
and N(t)
= iV(0)e-A'
(2.27)*
not
dN
dt
AN(t)
dN
dt
(2.29)
is
present at time
to
ta,
is
it
is
A,
a
a is
is
N(0)c-*"
(2.30)
ta
0;
is
0)
thus,
As before, N(0) denotes the number of nuclei present at =
=
of
at which the number
nuclei
the time instant (measured from
We require, therefore, that
exactly half the initial amount.
present
is
obtained
iN(0)
(2.31)
result
is
The half-life
t0
N(U)
The
is
time
is
is
q.
(2.32)*
The half-life, decay scheme, and energy of radiation (or of the emitted
particle) for some radioisotopes of interest to reactor technology are
The various known decay schemes for these nuclei
listed in Table 2.1.
are given in the third column.
The symbol a denotes radioactive decay
of the unstable isotope by the emission of an a particle; /3~, emission of a
negative electron; y, emission of electromagnetic radiation; K, orbital
SEC.
PROBABILITY
2.3]
Table
Isotope
6B
Half-life
CONCEPTS,
CROSS SECTIONS
Radioactive Isotopes
2.1
Decay
Energy,
scheme
Mev
Decay
Energy,
scheme
Mev
43 days
12
0.022 sec
Half-life
Isotope
pr
1.85
0.50
1.6
.C"
,N
5100 years
/s-
7.35 sec
0.15
6JTe135
3.5, 10
..I1"
<2 min
6.7 hr
p~
1.40, 1.00,
0.47
8o
28 sec
6.2, 6.7
(3-
2.9, 4.5
i,Al
14.8 hr
0-
>.K
1.8
.7Co
years
10.7 min
10
250 days
1.7
0~
3.01
1.8
hPiii1"
47 hr
/?-
1.9
,oTh!
0.93
0.25
0-
1.35
1010
or
3.98
0.32
(")7
1.11
Pu
108
4.56, 4.40
0.162
a
(e-)y
24,110
years
8.91
years
0.056, 1.5
K(/3+)
1.6
1.1
years
1.54
1.39
0.247
r
r
hr
.oNd1"
1.38, 2.76
,oZn
(")7
P~
1.39
7
2.30 min
9.2 hr
1.6
i.Na"
.Xe
electron emission;
and
e~,
5.15
0.42, 0.20
internal con
of the radioactive-decay
process is entirely
general and applies to all the decay schemes mentioned above.
Some
of the isotopes listed in Table 2.1 may decay by the emission of more
than one type of particle.
For example, Al28 decays by either /3~ or
y emission.
The half-lives of nuclei that possess multiple decay schemes
are based on the combined probabilities of the various possible events.
Thus, in the case of Al28, the 2.30-min half-life means that a given sample
of this material will be reduced to one-half its initial concentration in
2.30 min by either
emission or y emission.
Although it was not specifically stated, the decay process selected for
the above derivation implied a single decay scheme. The analysis is
more general than this, however, and can be changed to account for the
possibility of additional modes of decay. To continue our example,
consider the case of Al28.
Let Gp(t) denote the probability that an Al28
nucleus will decay before time
6,(0111
- Gy(t)]
_
=
t
t
by
by either
f}~
not decayed
or y emission
(2.33)
1A
detailed description of these processes can be found in any of the standard texts
on nuclear physics.
An introductory treatment is given by F. Bitter, "Nuclear
Physics." chap. 4, Addison-Wesley Publishing Company, Reading, Mass., 1950.
REACTOR ANALYSIS
40
[CHAP.
- G0(t)][l - Gy(t)]
= e-<Ao+V
(2.34)
The number of Al28 nuclei which survive to time t (i.e., have not decayed
by either 7 or /3_ emission) is the product of this joint probability and the
initial population iv*Ai(0); thus
=
NM(t)
It
vrrr
+
Afi
(2.35)
ArA1(0)e-<A+V (2.35)
is
<2-36)
Ay
by three
or more modes.
2.4
tron will collide with any nucleus whose center falls within a distance 7?
If N is defined as the
measured perpendicularly from the path line.
number of nuclei per unit volume in the medium, then along unit length
of path a total of ttR2N nuclei fall within the required range for collision.
Thus on the basis of this simple model
*
2 =
Thus 2
is
(tR*)N
"
TfQ
jf
(2.37)
a neutron
found
r + dr from
a given
(2.38)
PROBABILITY
sec. 2.4]
CONCEPTS,
CROSS SECTIONS
41
Then,
--
21
In
--"1
-*(r)*
JW
(2.39)*
the last step in (2.39) the integral of the weighted probability func
tion over all space is designated by the symbol a. Equation (2.37) may
also be written in this form with a particular choice for the function h(r).
Thus, quite generally, the expected number of collisions per unit
distance traveled 2 is proportional to the nuclear density through a
will
its dimensions
for nuclei
sections
veniently
measured
in
are con
barns, with
v',
tron of
v,
of the collision problem (Sec. 2.2), we were not concerned with the phys
ical process which took place between the neutron and the nucleus at
Instead, we confined our attention to determin
the instant of collision.
REACTOR ANALYSIS
42
[CHAP.
It has become
the various
It
2/
N<ra =
N<7y
Nof
subscript.
cross section
macroscopic absorption
The
(2.40)*
cross section
inelastic-scattering
It
is meaningful
following question: What is the probability that the neutron will travel
a distance x before it is either absorbed or scattered?
We can answer
this question by means of a calculation much like the one we used for the
radioactive-decay problem of Al28.
The two processes elastic scattering and absorption
and their respective distribution functions are
F,(x)
- e--
Fa(x) =
are independent,
e-1-*
(2.41)
These functions give the probability that the particular event denoted
by the subscript occurs before the neutron reaches x. Thus, the prob
ability that the neutron remains uncollided by the time it reaches x is
[cf. Eqs. (2.28) and (2.33)]
[1
- F.(x)][l -
Fa(x)}
- F.(x)
e-*
(2.42)
1
F(x).
- e-
(2.43)
SEC.
2.4]
PROBABILITY
CONCEPTS,
CROSS SECTIONS
43
/.(*)
with
2 = 2
Soe-^
(2.44)
We conclude from
+ 2..
2, +
22
2B
(2.45)*
gives the total number of collisions of all n types which the neutron
Any partial sum, e.g., 2i + 2S,
experiences per unit path of travel.
gives the number of collisions of types 1 and 2 per unit path of travel.
Macroscopic cross sections, then, are additive. This operation, how
ever, offers some conceptual difficulties in connection with the absorption
process. The idea of a single neutron being absorbed more than once
appears to be in contradiction to the definition of the absorption process.
In dealing
And, indeed it is. But, this is not what Eq. (2.45) implies.
with the absorption process (and with any of the others, for that matter),
it is more meaningful to talk about the mean free path to absorption
A physical interpretation
Xa, which is the reciprocal of 2 [cf. Eq. (2.21)].
of (2.45) is better made, then, in terms of the corresponding mean free
Thus, we write
paths.
X*
y'
-j
-+-++Xi
X2
(2.46)
The total mean free path is the average distance along its path that a
neutron travels before any one of the n events occurs.
Note that, if the
mean free path to any one of these individual events approaches zero
(i.e., the probability of the event occurring increases to unity), the total
mean free path also goes to zero.
The functions F0(x) and/(x) derived above may be used to describe the
neutron population and distribution of collisions in the example con
If n(0) neutrons leave the origin of the x axis, moving in the
sidered.
positive x direction, then the number of neutrons which reach x without
collision is [cf. Eq. (2.2)]
n(x) = n(0)e~*-
(2.47)
REACTOR ANALYSIS
44
[CHAP.
Thus
(2.48)
As(x) = n(0)2.erx*
Absorptions:
Aa(x) = n(0)Se_I!r
a;
is
(2.4ft)
Note that the neutron population and the density of events all decrease
exponentially from the origin.
It has become the practice in dealing with combined events to define
The
a suitable set of symbols for some of the important reactions.
standard practice is to call the total cross section 2, of a given material the
Thus
sum of the cross sections of all the possible types.
2, = 2 + 2. + 2in
(2.50)*
In
process is unimportant,
situations wherein the inelastic-scattering
As previously remarked
the usual definition is to write 2< = 2 + 2,.
is
cross
section
given
by
[Eqs. (2.40)], the absorption
2 = 2T
+ 2/
(2.51)*
In
the case of nonfuel materials (2/ = 0), the absorption cross section
is just the radiative-capture cross section.
We found in Sec. 2.4a that the
c. Measurement of Cross Sections.
macroscopic cross section 2 could be described in terms of the nuclear
first principles.
is required,
we
must rely
sec. 2.4]
PROBABILITY
CROSS SECTIONS
CONCEPTS,
45
Although the basic problems of reactor physics are not generally con
cerned with the analysis of cross-section experiments, there are some
special areas of reactor analysis which involve the same techniques and
analytical models. The analysis of radiation experiments in research
reactors and criticality studies are examples of such problems.
There is
one cross-section experiment, however, which is extremely simple to
study from the analytical point of view and which will serve very well to
demonstrate some of the ideas presented in this section.
This problem
has the additional merit that it gives some further insight into the
The experiment in question is the
physical concept of a cross section.
measurement of the total cross section of a given material.
The essential features of this experiment are as follows: A collimated
beam of monoenergetic neutrons is allowed to strike a slab of the material
of interest.
If n(0) is the number of neutrons crossing a unit area of the
slab surface per unit time at x = 0 (see Fig. 2.6), the number of uncollided
neutrons crossing a unit area per unit time at dis
~~
*
tance x inside the slab is [cf. Eq. (2.47)]
n(x) = n(Q)e-z-z
about
Fig.
J*
2.6
Collimated neubeam inident
*
(2.52)
The numbers n(0) and a are known, and n(a) is measured by the detector.
Thus, one can compute S,
S, =
knowledge
N<rt =
- In
a
n{a)
(2.53)
yields
the
REACTOH
46
ANALYSIS
[CHAP.
urements for specific neutron reactions are much more difficult to make
and will not be discussed here.
Extensive compilations of such data
have been made, nevertheless, for many materials of interest to reactor
These
technology and are now available in several standard sources.1
data for the various elements are customarily presented in the form of
Also available in
cross-section curves as a function of neutron energy.
The thermalmany sources are the thermal-neutron cross sections.
neutron cross sections refer to the cross section for neutron-nucleus inter
action in a system wherein the kinetic energy of the neutron corresponds
to the equilibrium temperature of the material.
This reference tempera
ture is customarily taken as 20C (0.025 ev).
Table
Material
Density,
g/cm*
H
D
<r,
0.330
0 .46 mb
Li
0.534
1.85
Be
Generated for wjivans (University of Florida) on 2015-09-22 17:17 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Thermal
2.2
B
C
N
71
10
mb
755
1.60
3.20mb
1 .88
<0.2
Na
Al
0.971
2.70
0.87
7.86
mb
38.0
Mo
7.0
Cd
1.4
In
Xe1"
7.0
4.0
4.8
10
4.2
0.505
0.230
4.0
1 .97
1 .5
1.4
Sm>"
Th
U"
U"
U"8
I'll"9
Pu"1
H20
Fe
Co
2.53
37.8
11 .0
8.9
7.0
D,0
Ni
8.9
4.6
17.5
BeO
Density
10.2
8.65
7.28
a.
2.5
2550
2.2
190
2 .7
7.7
11.3
18.7
18.7
18.7
10
50,000
7
12.6
585
687
2.75
1065
533
10
8.2
9.6
580
750
1100
1.0
1.10
3.01
0.66
0 .93 mb
9
mb
108
14.3
10.0
All
cross sections are given in barns except where noted as millibarns (mb).
Refers to thermal energy (20C or 2,200 m/sec).
The thermal-absorption,
elastic-scattering, and
speed is 2,200 m/sec.
fission cross sections for some common reactor materials are listed in
Table 2.2. The density of the material at 20C is also included.
The
millibarn notation (mb) used in the table is defined
as one-thousandth
of a barn.
d. Calculation of Macroscopic Cross Sections.
In order to determine
the various neutron-nucleus reaction rates, we require the appropriate
Since these probabilities are related to the
probabilities for collision.
collision cross sections 2 (or the mean free path X), the reaction rates
may be computed from a knowledge of the microscopic
cross sections a
for example, Donald J. Hughes and John A. Harvey, in U.S. Atomic Energy
Commission, "Neutron Cross Sections," McGraw-Hill Book Company, Inc., New
1 See,
York,
1955.
sec. 2.4]
and
PROBABILITY
the nuclear
relation :
densities
nuclei
vol
CONCEPTS,
N.
CROSS SECTIONS
The densities
47
are obtained
from
the
= mass/ volume
= nuclei/mole = Avogadro's number (0.6023
= mass/mole = atomic (or molecular) weight
density
N
The macroscopic cross section may
1024)
N and obtain
&
be
written in terms of
(2.54)*
these
quantities
2 =
Thus, 2
is immediately
(2.55)
is
specified.
The magnitudes
The frequency function for collision is still of the form given in Eq. (2.17),
but the cross section 2 is now to be computed from the properties and
densities of all nuclei present.
Consider A; fictitious media, each of which
consists solely of one of the nuclear species of the final mixture.
Let the
nuclear density of the medium composed of the ith nuclear type be the
Then, for each of these
same as the density of that type in the mixture.
ANALYSIS
REACTOR
48
[CHAP.
_
=
probability that
such that
with
a neutron collides
a nucleus of type
before reaching x
e~sli)x
(2.57)
2W> is to be computed
from the properties of this ith
Now
medium, using the known <r(i) and the density JV. from Eq. (2.54).
imagine these various media to be combined to give the resultant mix
ture.
Furthermore, let us assume that collisions with the various nuclear
The parameter
is
some function
distance x without
Fm(x), where
2<
[1
k
- Fk(x)]
(2.58)
e-zt,, = e-2""'1
2'"'
- F1(x))[l - F,(x)]
f]
(2.59)
i-i
The last equality follows from our usual definition of a. The frequency
function fm(x) for collision in the medium may be computed from the
distribution function Fm(x) by means of the result (2.58).
=
F'm(x)
= 2'-'e-sl"1*
Cross
Sections
of Compounds
AV(1>
(2.60)
is
/(*)
f.
AV
(2.61)
The
calculation
is
is
with
to survive
k [1
neutron
is
If
it
is
given by
Ft(x).
collision in the mixture,
is,
sec.
2.4]
PROBABILITY
CONCEPTS,
CROSS SECTIONS
49
is determined
speed of
ev, the target nucleus
responds as a free particle because at these energies the chemical binding
forces are negligible in comparison to the collision forces, and so the bonds
are disrupted.
At lower neutron energies, internal vibrational states
the neutron.
At neutron energies in
excess
of
of the molecule can be excited, and the neutron will experience inelastic
collision with the molecule.
However, there exists some neutron speed
below which the lowest vibrational mode of the molecule cannot be
excited, and the target nucleus responds to the collision with effectively
the mass of the molecule.
In addition to these considerations, the
PROBLEMS
2.1
Monoenergetic neutrons arc being emitted isotropically from a point source
in an infinite medium at a rate qo neutrons/sec.
If the nuclei of the medium have an
absorption cross section <ra for neutrons of the given energy, a negligible scattering
cross section, and a nuclear density Ar, find the number of neutron absorptions occur
ring per cubic centimeter per second at each point of the infinite medium.
2.2
Suppose that in the same medium as in Prob. 2.1 neutrons arc emitted per
pendicularly from a line source at a rate qt neutrons/sec per centimeter length of line.
Find the number of absorptions per cubic centimeter per second at each point in the
infinite medium.
2.3
Suppose that the neutron source in the infinite medium of Prob. 2.1 is a per
pendicular-emitting plane with a strength qt neutrons/cm* of source area per second.
Find the number of absorptions per cubic centimeter per second for all points in the
medium.
2.4
Suppose that the neutron source in the infinite medium of Prob. 2.1 is a plane
which emits neutrons such that from each side of a small area dS, q3 dS neutrons are
emitted per second with an angular distribution given by
q(0) de dS = 2q, dS sin 0 cos 0 d0
where 0 is measured from the normal to the plane (0 having the range 0 *-/2 for
each side). Show that the number of absorptions per cubic centimeter per second for
all points in the medium is given by
CHAPTER
MULTIPLICATION
3.1
Collision Density
The importance of the nuclear reaction rates (the number of neutronnucleus collisions per unit volume per unit time) to the problems of
reactor physics was discussed in the preceding chapters.
The purpose
requirements
Expected
number of
collisions
in time
dt in
volume dt
be
expected
number of
distance
number of
collisions
traveled
per neutron
by neutron
in time dt
neutrons
in volume
per unit
distance traveled
[cf.
(3.1)
dt
Eq. (2.22)]
by neutrons of
speed
per
(3.2)
SEC.
If
MULTIPLICATION
3.2]
CONSTANT
AND
NEUTRON FLUX
51
we define, also,
,
n(X'V't}
_~
F(T,V,t)
(3.3)
F(r,v,t)
or
2(i>) v dt
= 2(t>)
n(r,v,t) dt
n(r,t;,<)
(3.4)
The quantity F(r,v,t) is called the collision density for the nuclear reac
tion described by the cross section 2(y). No subscript has been assigned
to 2 so as to keep the derivation general.
Equation (3.4) shows clearly
the dependence of the reaction rate (or collision density) on the neutron
This relationship was stated previously in the intro
density n(r,v,t).
duction to Chap. 1.
for example,
24(y)
2<,(i;) v
n(i,v,t)
n(i,v,t)
(3.5)
the actual number of collisions will vary around the expected number
with a variation which will depend upon the neutron and nuclear densities.
The macroscopic cross sections 2(y) may be written in terms of the
nuclear densities in the usual way [cf. Eq. (2.39)]. In general, these
densities will be spatially variant functions and so should be properly
written in the form N(r). The resulting equation for the collision density
is then
F(T,v,t) = <r(y) N(t) v n(i,v,t)
(3.6)
a. Infinite-medium
Model.
Neutron Balance
The actual
situation in an operating
cycle of neutron production
REACTOR ANALYSIS
52
[CHAP.
through
the boundaries
of the system.
a neutron
In the time
that the "microscopic" picture would be much the same as the over-all
picture of the reactor system, with neutrons of all energies continually
Thus, at any given time
streaming in and out of the little volume.
instant, there would be present in this volume neutrons of all energies,
corresponding to different stages in their life history. Some of these
energies less than that with which they entered.1 If the volume element
we selected were in the core, the situation would be further complicated
As in the case of neu
by the continual appearance of fission neutrons.
trons which arrive in this volume by passing through its boundaries
(1)
(2)
(3)
(4)
speed v
(3.7)
1 These
reactions
SEC.
3.2]
MULTIPLICATION
CONSTANT
AND
NEUTRON FLUX
53
We recognize
neutron population in going from one point to another.
then that the neutron-density function n will not include spatial coordi
nates in its argument [cf. Eq. (3.3)]. Assumption (2) further restricts
omit the speed variable v, it being understood that all the neutrons in
Although we specified a similar
this medium have the same speed.
in
of
situation
the discussion
Sec. 3.1, it was convenient to retain the
independent variable v in that instance in order to emphasize the func
Assumption (3) is intro
tional dependence of the quantities involved.
duced so as to simplify the problem and allow attention to be focused on
other aspects of the physics.
The energy changes resulting from scatter
will
be discussed in detail in Chap. 4.
ing collisions
Assumption (4)
is again a simplification, but it does not impose any great limitation on the
As in the case of the neutron density n, the
usefulness of the results.
cross sections will refer to neutrons of speed v, but the symbols will not
include the argument v for the reason given above.
b. Neutron Density in Infinite Medium.
Even though
the system
described above is a highly fictitious one, we can nevertheless acquire
from it a considerable insight into the nature of a chain-reacting complex.
We approach this objective by asking the fundamental question of all
reactor problems: What is the distribution of neutrons in the system?
Because of the nature of the limitations we have imposed, this distribu
tion will be expressed in terms of the only remaining variable, the time t.
Thus any balance condition we write for the neutron population must
On the basis of the assumptions (3.7),
be in terms of its time dependence.
all the materials are uniformly distributed throughout the space.
Fur
thermore, the only sources of neutrons are the fission reactions, and the
No neutrons can be lost by escape
only sinks, the absorption reactions.
of
this
boundaries
reactor.
The neutron-balance condition
through the
of these competing
is as follows:
processes.
Number of
neutrons gained
per unit volume
|_per unit time
number of
number of
neutrons produced
neutrons lost by
absorption
per
REACTOR ANALYSIS
54
[chap.
3.1.
If
T n{t)
v2fvn(t)
- 2avn(t)
(3.8)
Nf
2
No
2^0)
(3.10)
all nonfuel
materials
JSVi1'
2^2)
2^*"
sa>
+"
2 =
_|_
(3.11)
Eq.
follows
from
It
SEC.
3.3]
MULTIPLICATION
CONSTANT
AND
NEUTRON FLUX
55
until time
then for all t >
t 0.
At
= 0
all
dt
n(t) = (vZ,
- Xa)vn(t)
(3.14)
is
(3.15)*
Fio.
3. 1
Time dependence of neu
tron density in infinite medium.
3.3
n0
moves through the system and makes collisions with the various species
of nuclei present; the results of these collisions will be: (1) change in
direction of motion of the neutron, (2) loss in kinetic energy of the neu
REACTOR ANALYSIS
56
[CHAP.
will be those of the moderator, the coolant, and the structural materials,
all of which have cross sections for nonproductive capture of neutrons.
Only the fraction of neutrons absorbed in the fissionable fuel can be
Furthermore, not all captures in fissionable materials result
productive.
in fission and, hence, more neutrons, because each fissionable element has
also a cross section for radiative capture.
In any event, a certain fraction
of the no will cause fission, and these fissions in turn will yield, say,
If the reactor system is self-sustaining, the ni neutrons
rai neutrons.
of the second generation will be identical in number with the no neutrons
of the first generation; thus, the population is maintained.
The ratio of the neutron population in two successive generations,
such as n0 and n\, is a useful number for describing the characteristics
This ratio is called the multiplication constant
of the chain reaction.
of the reactor and is formally defined as
,
ni
n0
one generation to
(3.16)*
fc
1.
In this
case,
reproduces
Thus, if
conditions
It
is convenient
here to
of the previous
of the neutrons introduced into the unit volume are actually absorbed
there, every one of the neutrons which leave this volume element to be
SEC.
MULTIPLICATION
3.3]
NEUTRON FLUX
AND
CONSTANT
57
- fS -
(3I8)*
is
is,
2/>
2/ + 2<f
vNF<r,
Ntfn 2/)
20>
is
77
t\
is
is
is
?;
t)
is
^_
(3.20)
follows that
*
and
it
a =
ff7T^>
The quantity
if".
(3"21)*
is
is
it
is
is
and
is
if
in the reactor.
Note that
If this notation
used in
1.
No
nuclei
0,
if
>
0;
likewise,
0,
r,f
where
^^
(3-23)*
we
If
thus, the
REACTOR ANALYSIS
58
[CHAP.
It
was remarked in Sec. 3.26 that the scattering cross section did not
play a role in the analysis of the infinite-medium model, and indeed
This is entirely a consequence
the result (3.22) is independent of 2,.
of the assumptions of an infinite medium and no change of speed in
scattering.
If
the nonfuel
coolant, etc.) are
of the reactor.
dimensions is identically
is,
1.
discussed later.
1
quantities
MULTIPLICATION
SEC. 3.3]
AND
CONSTANT
NEUTRON FLUX
59
is
-*"
nJN+n^
(3-24)
(3-25)
second relation
-I"
(3.26)
Vi
and
V{ =
actual volume
reactor
-y
of component
(3-27)
t as
it is to appear in
(3.28)
Ni
Pa
~
(3.29)*
In this expression, P,
fraction; thus,
Pi
The normal density
temperature
p<
= PiVi
(3.30)*
It
of material
follows that
at the reactor
(3.31)
pf
pm
The simultaneous.
REACTOR ANALYSIS
60
[CHAP.
pmQ
Am
[(ij
s
A*
m =
- l)m/sp
1]
(3.33)
*P
l)m/sp +
o-,(A"
Pa/
,(')
(3.34)
PF
3.4
Ar[(v ~ l)m/P +
1.
N*
where
PfGL
(ij
Nr
In
is,
is
the present section, we carry out sample calculation for the safety
The calculation
chemical-processing plant.
storage requirements in
based on the infinite-medium one-velocity model.
Our consideration of nuclear reactors, thus far, has included only the
of course, only
principal features of the reactor proper. The reactor
and the
operation
The
complex.
one element of an extensive engineering
of this device require many additional component systems.
*
COOLING
REACTOR
CHEMICAL
PROCESSING
DILUTION
To reactor -
Fig.
3.2
Chemical processing
New fuel
elements
cycle.
is
determined by the
Precisely what supporting components are required
Reactors designed for indus
type and the design purpose of the reactor.
trial or military application are primarily used for the production of power
These applications are frequently made on
or fissionable materials.
is
Under
large scale; consequently, the supporting facilities are also large.
such circumstances, the preparation and supply of the reactor fuel ele
This
particularly true
mammoth undertaking.
ments can become
in the case of reactors which utilize fissionable materials in solid form.
Such large quantities of solid fuel elements can be handled only in large
chemical-processing
plants.
is
problem
When dealing with large amounts of fissionable material,
which inevitably arises
the criticality consideration of the system.
important to the chemical-processing cycle for two rea
The problem
is
sons: (1) the safe storage of the fissionable material at each stage of the
process, and (2) the optimization of the batch size handled by the process.
Both of these factors can enter into the design considerations of the
The principal
number of intermediate stages.
process at any one of
stages in the chemical processing of solid fuel elements are shown in the
a
maintenance
SEC. 3.4]
MULTIPLICATION
CONSTANT
AND
NEUTRON FLUX
61
block diagram of Fig. 3.2. The first stage is denoted cooling and cus
tomarily involves the storage of the spent fuel elements in a suitable
vault or canal wherein the radioactivity of the fission fragments present
in the elements is allowed to decay to tolerable limits. The second stage,
here called dilution, is the preparation of the fuel elements for processing
(e.g., dissolving by some suitable acid solution). The final stage involves
Fuel cluster
Core
A-A
Spent slugs
Fio.
3.3
REACTOR ANALYSIS
62
[CHAP.
Slug length
Slug diameter
Slugs per fuel-element cluster
Enrichment (% U")
Total number of slugs
12
in.
0 . 4 in.
500
3
150,000
Cross-sectional
The densities of the various materials in the homogenized system are com
puted from Eq. (3.30), using pF = 18.7, p.*., = 7.8, and pHl0 = 1 g/cm8
(cf. Table 2.2).
Pf
pfVf
/V,
= 0.843
= 2.53
a<a26)
a?
use
= 687
= 0.66
SEC. 3.5]
MULTIPLICATION
CONSTANT
NEUTRON FLUX
AND
63
For the 3 per cent enriched uranium slug, we obtain the absorption
cross sections [cf. Eq. (3.29)]
Nf&m
= (14.7)(0.6023)(687)(0.03)
= (0.843)(0.6023)(2.53)
5o
= (0107)(0.6023)(0.66)
2(26>
/ [cf.
vf
2<,t,el)
is,
n ocrj
2|H!0>
by (3.22),
is
is
Eq. (3.19).
3.5
Neutron Flux
is
density
through
the definition
(v
is
it
population,
frequently more convenient in reactor calculations to
work with another function, called the neutron flux. The neutron flux
= neutron
speed)
= nv
(3.35)*
if
is
is
If
4>(T,v,t)
= n(i,v,t)v
(3.36)
(3.23)]
^5
2<28)
?m
cm_,
V(26)
-|-
^
_^^
^^
(14.7)(0.60BK2.75)(0.97)
S<n, =
REACTOR ANALYSIS
64
[CHAP.
of speed dv.
n(i,v,t) dv dr =
dv
(3.37)
dv dr =
is
(3.38)
(r, v,t),
if the deviation is
in question approach v.
is customary to call
the total track length per unit volume per unit
time.
The basis for this terminology is evident from the definition (3.35).
We write this relation in an expanded form
It
(T,v,t)
dv
di
= vn(i,v,t) dv dr
or
(t,v,t)
distance
number of
traveled
neutrons
per unit
per neutron
per unit
volume per
time
unit
speed
speed
interval J
rvoiume]
This definition
(3.39)
(3.40)*
is frequently
(3.41)
It
can
SEC.
MULTIPLICATION
3.5]
variable v. Thus, if
(cf. Fig. 3.4)
AND
CONSTANT
dE
n(i,v,t)
n(r,,<) = n(t,v,t)
or
65
we use the
n{r,E,t)
NEUTRON FLUX
dv
(3.42)*
(3.43)
n()
. "00
mv
_ n(V2lA)
= n[v(E)]
VWm
y/2Em
(3
u)
The
where we have omitted the variables r and t for the sake of clarity.
corresponding expression for the flux per unit energy range is obtained
Fig.
3.4
Equivalent
a similar calculation.
The fundamental relation between the expres
sion for the flux in speed space and energy space is derived from (3.35)
by
using (3.42)
If
v(E) n(i,E,t) dE
n(r,v,t) dv
(3.45)
v(T,E,t)
then
(r,E,t) dE
v(E) n(T,E,t)
= (t>(r,v,t) dv
(3.46)
(3.47)*
This equation corresponds to the relation (3.42) and states that the total
track length per unit volume per unit time around r at time t is conserved
in going from speed space to energy space.
If we introduce the Jacobian
dv/dE, it is easily shown that (again omitting r and t)
,<A')-4^g^-:m (v^)
\\ m J
It has
cms.
REACTOR ANALYSIS
66
[CHAP.
and the flux per unit energy solely by the argument following the flux
Thus, if 4> denotes flux and n denotes density, then
symbol.
^(
4>(E)
(3.49)
31
dE
Jr 4>(i,E,t)
=
(3.51)
(0
(3.50)
3l(#,0
Generated for wjivans (University of Florida) on 2015-09-22 17:17 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
di dE
jE *(E,t) dE
fg dE
jy
of neutrons
4>{r,E,t) dr
of all
(3.52)
In
(3.53)
F(t,E,t) = Z(E)
=
4>(t,E,t)
(3.54)*
S a\.(E)
(t
<p\ , -, t)
dF>
E + dE
(3.55)
SEC. 3.5]
MULTIPLICATION
CONSTANT
AND
NEUTRON FLUX
67
and so on. This relation should be compared to (3.5). The form (3.54)
for the collision density is especially useful in writing neutron-balance
conditions such as (3.8). In the present notation, this equation takes
the form
I ~at (<)
*2/*(0
- 2.0(0
3.5
Collimatcd
Fig.
is
of
is
is
if
is
Unit area
ANALYSIS
REACTOH
68
[CHAP.
Thus,
(a typical value),
Xa
= 6.54
109
the rate a
absorptions/cm3/sec
= 4.54
107
neutrons/cm3
PROBLEMS
Consider an infinite reactor composed of a uniform mixture of beryllium and
5.1
The atomic ratio of isotopes in natural uranium is U"8:Um =
natural uranium.
993:7. Assume that the one-velocity model applies, and use the microscopic cross
sections and densities of Table 2.2.
o. Compute the atomic concentration of U*" required to make the reactor critical.
6. Show that, if it is assumed that the average density of the U-Be mixture can be
approximated by the density of pure Be, then, in this case, the error in the Um con
centration is of the order of 1 per cent.
c. Find the mean number of scattering collisions made by a neutron during its
lifetime.
3.2
It is proposed that a reactor of pure UF be constructed, the fluorine to provide
the moderation and the U"' in the uranium to serve as the nuclear fuel. A first
estimate of the critical fuel concentration for such a system can be obtained from the
one-velocity infinite-medium model.
a. If the uranium in the UF8 is composed of U"5 and U"*, determine the atomic
Use the data
ratio of U*" to U"* required to make the infinite medium critical.
t)
2.08
(U")
fo=9mb
for fluorine
CHAPTER
SLOWING-DOWN
4.1
a. Scattering
phenomena.
All three types of reactions enter into the determination of the spec
trum. The fission reaction determines, of course, the average energy
The scattering and absorption
of the neutrons borne into the reactor.
reactions together determine the detailed shape of the spectrum.
The
largest influence is due to the scattering interactions which establish the
The effects of the absorptions are some
over-all shape of the spectrum.
what secondary, and in most reactor situations these tend primarily to
REACTOR ANALYSIS
70
[CHAP.
ground state, and these states are spaced widely apart (in the order of
1 Mev).
If the struck nucleus is of large mass, however, the first excited
state is closer to the ground state (10 to 100 Kev), and the adjacent states
are much more closely packed1 than in light nuclei.
These general properties of light and heavy nuclei are illustrated in
4.1.
energy
appreciably
possesses
threshold
energies
(the separation energy between the first
excited state and the ground state) . For
JL*_P_
higher
Fig.
e0
4.1
Nuclear energy levels:
(a) low mass number; (6) high mass
number
as 2 and 3
-
Mev.
.,
as effective
limitation.
The elastic-scattering
process, however,
for example,
pp. 365-374,
Nuclear Physics,"
SEC.
4.1]
BLOWING-DOWN
PROCESS
71
of the energy change does depend, however, on the initial energy of the
As will be shown shortly,
neutron and on the mass of the struck nucleus.
the light nuclei are the most efficient moderators from the viewpoint of
the elastic-collision process.
It is clear from the above discussion that a reactor which is designed
to utilize the superior low-energy properties of its nuclear constituents
Thermal reactors
must contain large amounts of moderating material.
In accord with the remarks above, these
are systems of this type.
reactors contain materials of low nuclear mass which slow down the high-
At the higher
energy fission neutrons by means of elastic collisions.
neutron energies (above threshold) the inelastic scatterings with the
heavy nuclei in the structural components of the reactor, and with the
Although these con
fuel itself, can assist in this slowing-down process.
tributions are generally small, they can be significant in some instances.1
Inelastic scatterings are most important in systems which contain large
amounts of high-nuclear-mass materials, such as iron, liquid metals, and
Reactors which utilize materials of this type tend to have
uranium.
Sometimes this property is deliberately designed into
fast spectrums.
In
such
systems good moderator materials are kept to a
reactor.
the
minimum in order to reduce the slowing down of neutrons as much as
In
In
1 The
HEACTOK ANALYSIS
72
[CHAP.
(E)
dE
v(E) n(E) dE
total track length per unit volume per unit time of neutrons with kinetic energies between E and E + dE
is characterized
v
by neutron cross
is the average number of neutrons
ME)
= 2T(,')
(4.1)
(4.2)
+ Xf(E)
the nucleus
8EC.
4.1]
SLOWING-DOWN PROCESS
be treated
73
velocity
velocity
Vi = velocity
V*i = velocity
W for the
and nucleus velocities, respectively.
The subscripts 0 and 1
apply as before.
The immediate objective is to determine the relationship between the
initial and final energies of the neutron, i.e., the energy loss suffered by
For this
the neutron of mass m in colliding with a nucleus of mass M.
use the symbols w and
neutron
+ MV0
= mv,.
MVi
(4.3)
balance we have
imvl + MVl
lmv\ + \MW\
(4.4)
(M +
il/Vi
(4.5)
v0
w0 = v0
c =
-ry
m
- - TX
(v0
Vo)
(4.6)
(v,
- VO
W.-Vi-c-j^OTt-TO
1
1
(4.7)
REACTOR ANALYSIS
74
[CHAP.
it is
total
that
= mw,
+ MW0
mwo
so the
seen
MWi
= 0
(4.8)
momentum
imw* +
pfW
*h
i
Wl =
w0
Wi
Velocities
niass system.
4.2
(4.9)
Z
-^w,
(4-10)
ff
Fig.
p/W?
imvfl +
(4.9) yields
in eenter-of-
w\ = w\
W\
W\
(4.11)*
Nuclei.
This assumption
initially at rest in the laboratory system of coordinates.
simplifies the analysis somewhat ; however, it may introduce appreciable
error in the study of collisions between low-speed neutrons (in the order
of thermal energy ~0.02 ev) and nuclei.
The speed of the nuclei in
these situations is comparable to the speed of thermal neutrons, and the
approximation is therefore poor. For the present we will not consider
collisions by neutrons of electron-volt (or lower) energies and will limit
our calculation to the portion of the slowing-down process above the
is [see
Eq. (4.5)]
C
: J>
+ M
Vo
(4.12)
SEC. 4.1]
tion
v\
75
Our
4.3.
namely, %mv\.
[see (4.7)]
= (Wl
= w?
+ c)s
c2
+ 2w!
(4.13)
we have
(^TT/)2v
(4.14)
Center of mass
Fio.
4.3
system for
2^
be
m)
A/)
(-^)2
(415)
(^X/)2
vl
* (jT?
vl
u*
(iff
r\
/3
Wl
'
C
/S.
<
2mMvfa
(m
+ M)2
(4.16)
Let
(4.17)
If
E*
(A +
[chap. 4
ANALYSIS
REACTOR
76
by
1)2
m and
(A* +
the definitions
(4.17) are
2^ + !)
(4-!8)*
It
known.
is clear
that,
varies from a maximum value of E0, the initial energy, to the minimum
value
EmiB
= E0
a =
where
(x^pi)2~
= aE
|Y
(4.20)*
In order to determine
d. Frequency
(4-19)
it in another way,
/S.
it
f,
it
we require
determined
with the
or theory.
suitable
is
is
result in scattering
and
+ drj
tj
dr}
f(jj)
AnPn(.v)dv
(4.21)
Pifo)
Po(u) =
/3.
it
i?
The argument
is
the distribution of final energies between the limits E0 and Emln [see
Eq. (4.19)], there must be some additional information as to the distribu
Or putting
tion of the scattering angles in the center-of-mass system.
171-181,
pp.
SEC.
SLOWING-DOWN
4.1]
PROCESS
77
This vari
the argument i) alone is sufficient to describe the scattering.
able takes on all values in the interval ( 1,1), the very interval over
which the functions P are denned.
If the function f(7j) is specified, the coefficients A are readily com
Thus we determine the coefficient A, by
puted by the usual methods.
multiplying both sides of (4.21) with the polynomial P,(rj) and integrating
over the interval (1,1):
P.M
^
Kv) * =
A,
Pn(v) P.(v) dv =
A.
or
2s
2A,
2s
(v) Kv) dv
(4.22)
A0 =
i f^PoWCdr,
Kv)=i
scattering
follows that
-1<V (4.23)*
(4.23)*
$u
should
d"
be noted
is is
It
and
/_',
= C
= A0 =
Therefore
dr>
it
/_". K")
This point
discussed
Wtf
(4.24)
(^J
2V
wlV)2]i =
Vl
(^j
(c
Vi = [w\ sin2
is
where
Mo
it
is
is
d0
do
it
system,
in the laboratory system. The
pute the cosine of the scattering angle
shown in Fig. 4.4.
An alternate view
angle
The
given in Fig. 4.5.
cosine of 80, call
mo, may be computed from the geometry of Fig. 4.5 by
using the relation
The simplest
(4.25)
KEACTOK ANALYSIS
78
[CHAP.
Wo
(4.26)
Mo
(4.27)*
1)1
This expression relates the cosine of the scattering angle in the laboratory
system to the cosine of the scattering angle in the center-of-mass system.
In order that a scattering collision be isotropic in any reference frame,
it is necessary that the average value of the cosine of the scattering angle
This condition is already met for the (C) system by the func
be zero.
(4.28)
M.(l) f(l)
M0O7),
*>=!/_, (a|
(4.29)
1)
dr>
is
mo
1*1
Ja
=
f0) *>
/_\
/_!,
rj
i
=
*J
and find
/_i
7.
as expected.
f(l)
(4"30)*
is
is,
it
is
direction.
0,
is
is
mo
it
nuclei, however,
It
it
(>Kev).
is
Thus we compute
/_\
tional form of
SEC.
SLOWING-DOWN PROCESS
4.1]
79
fWrf^o
i(l
-1
ar,) dr,
<
= 0
i<-i
i?
<
>
(4.31)
is
0,
is
if
is
is
if
is
\a\
<
and
the cosine of the scattering angle in (C).
As compared to the frequency function for isotropic scattering (4.23),
seen to have an added term %ar\.
This situation
the above function
described physically by saying that there
component in the
Note that a > the
scattering, and this component has magnitude a.
predominantly forward, and
predominantly
<
scattering
where 0 <
backward.
energy.
is
is
is
is
is
to
(4.32)
<
t)(E;Eo)
is
tj)
it
is
if
it
rj
is
The energy E0 denotes the initial energy of the scattered neutrons, and
a
Now, to each final energy E there
an associated
given by (4.20).
to
each
increment
scattering angle cos-1
Furthermore,
[cf. (4.18)].
Thus
change in tj, call
we
d-q, there corresponds a change dE in E.
scattered
relate Eton and dE to dr\, then the probability that a neutron
into
final energy range in dE about
[see (4.32)] must be exactly the
be scattered into di\ about 17. In other words, we
probability that
require that the probability that a certain event occur be the same
used to describe it.
Thus
regardless of the independent variable {E or
we require that
\KE;Eo) dE\
algebraic signs of
dE and
that
(4.33)*
|f() d\
so as to account
positive increment
for the
dE corre-
We use the notation \)(E;Eii) in order to keep in mind that the initial energy of the
scattered neutron was 0.
1
If
the frequency
directly from the energy-angle relationship (4.18).
function
specified in terms of either angle or energy, the corresponding
readily computed by the usual trans
function in the other variable
formation of variable.
demonstrated in
The procedure to be followed
We define the frequency function in energy by
the following calculation.
ANALYSIS
REACTOR
80
[CHAP.
drj
Kv)
(4.34)
dE
dy
dE
from
(4.18).
dy
dE
(4.35)
2AE0
Note that in the present calculation E, the final neutron energy, cor
If this result is used in Eq. (4.34), the
responds to Ei of Eq. (4.18).
frequency function in energy for the case of isotropic scattering in the
center-of-mass system is
(A +
t)(E;E0) dE =
\YdE
iAE0
dE
=
(1
- a)E0
aEo <
< Eo
(4.36)
Thus t)(E;E0) does not depend upon E, so that the kinetic energy of the
scattered neutron falls with equal probability into any unit energy inter
val in the range aE0 < E < E0- It is of interest to compute also the
distribution function for this process. If we define
&(E)
probability
that final
neutron
- a)E
de
&(*;) d( =
aE, (1
(1
- aE9
- a)EB
(4.37)
(4.38)
where the lower limit of the integral is given by Emin according to (4.19).
Incident neutrons of energy E0 cannot emerge from an elastic-scattering
collision with kinetic energies <a,0; thus &(E) m 0 for E < aE0.
The
maximum
i)(E;E0)
0
aEo
sE
E
^ Eo
> Eo
(1
- c)Eo
0
E
(1
aEo
- a)Eo
1
(4.39)
SEC.
4.1]
SLOWING-DOWN PROCESS
81
(1^) .
UE;E0) dE
(^n2)
(o,
is
kw E
The
a,
a-
/;.
(4-41)
and distribution
will
it
If
it
a) a
is
released in
and ^(1 +
is
be
El
(l^j
E\
(ly Eo
= 0.67
Mev
is,
{.
is
REACTOR
82
ANALYSIS
[CHAP.
The number
E*
(4.43)
7 8 910
4.7
30
20
Mass number
Fio.
40 50 60
80 100
200
dE
(4.44)*
SEC.
4.1]
83
From the definitions of and u, it is clear that is the average gain in the
lethargy of a neutron per collision.
From (4.42) we have
= In
Eq
-g
= In
E*
- In ^,
E+
= u
(4.45)
w0
energy E0
The history of these neutrons may
Eth (thermal energy).
be described in the following way: (1) Neutrons appear at high energy
E0.
(2) Each neutron makes a first collision with a nucleus of the
medium, and this collision may be either an elastic scattering or an
(4.46)
defined by (4.42).
This is not properly the average number of
collisions to traverse an interval Aw; however, we will use the form of this
result and assume that
with
M2
Ui
made by a neutron
in
w2
An analysis of the behavior of the neutrons in a medium wherein the nuclei are in
thermal motion is given in Sec. 4.7.
' One form of the distribution function for collision from which this result was
obtained is given by E. U. Condon and G. Breit, Phys. Rev., 49, 22<J (1936).
1
84
REACTOR
ANALYSIS
[CHAP.
U\ =
= 1), namely,
in passing
Needless to say, similar calculations for the general case (A > 1) are
difficult to carry out and have led to complicated expressions for which
However, it is
simple approximating formulas have not yet been found.
reasonable to assume that, if u2 tti is small compared to In (1/a) (the
maximum jump in lethargy per collision), Eq. (4.47) should hold, at least
approximately.
(Here we are considering single-element media only.)
that
this
condition
corresponds to the situation in hydrogen wherein
Note
In (1/a) is infinite and any (u2 Ui) is small in comparison.
For media
in which A > 1, and w2 U\ ~ln (1/a), Eq. (4.47) is commonly used
even though it is at best only an approximation for which the limits of
validity are not known.
Since the results of using (4.47) over the extended
lethargy range have generally been good, we shall accept the approxima
tion in this work; thus, for all A and for any lethargy interval i * u2,
however large, we will take (u2 tti)/f to be the average number of
collisions in traversing the interval (2 Ui) and recognize that the
approximation is particularly good for
M2
U\
(4.48)
tered emerges
where
Mo
= In
p-'
(4.49)
Then
(4.50)*
is
(4.51)
SEC.
It
SLOWING-DOWN PROCESS
4.2]
85
P(w
< u)
Uol
(4.52)
u =
In;
Fur
(4.53)*
u = oo
E = 0
4.2
These
given in terms of the neutron energy alone.
functions may also be described in terms of the neutron speed, or as we
will shortly demonstrate, in terms of the lethargy.
Which one of these
density
functions
\*{E) dE\
where
(E)
(v)
\(v)
dv\
(4.54)
..
REACTOR ANALYSIS
86
(E)
[CHAP.
(v)
and
be
If
we
define
.
then,
\4>(u) du\
\(E)
dE\
(4.56)
(4.57)
Z,(w)
2/(m)
{u)
{u)
(u)
(elastic
number of \
=
<
du
neutrondu
nucleus
(4.58)
letnarS>es m du
I
J about u
/ fission
\
It
That
is,
2()
Typical
neutron cross-section
JK
S(u)
(4.59)*
.c
be specified,
r
4.9
Fig.
curve.
= 2(y) =
I(tt)
du
SEC.
SLOWING-DOWN PROCESS
4.2]
87
If
which has a unique form for a given neutron history. Three such
Although no two can be alike, most
histories are sketched in the figure.
of the neutron trajectories for a given medium follow closely a general
In fact, it is entirely meaningful to talk about an "average"
form.
Whatever the
trajectory or an average neutron slowing-down track.
form of this function, it will describe the average behavior of a neutron
in the slowing-down process. As in the case of the neutron density, the
function for every neutron is not required (or
specific slowing-down
For
of reactor analysis, a function which describes
the
purposes
desired).
the average behavior is sufficient.
Urrtz;
Distance
Fig.
4.10
0(source energy)
traveled
by neutron
Slowing-down
track.
Distance
Fia.
traveled
by neutron
Slowing-down
track
medium of heavy nuclear mass.
4.11
in
model.
REACTOR
88
ANALYSIS
[chap. 4
the
q(E) = slowing-down
(4.60)
Thus this function gives the rate at which neutrons are slowing down
It should be observed that this concept is
past some given energy E.
similar to that of a flow (current) in a fluid.
One may think of the
function q as giving the flow rate of neutrons down the energy scale.
In the fluid analogy, q could be used to specify a continuity condition
on the neutron population in terms of the factors which affect the slowingdown process.
This we propose to do shortly.
The slowing-down density may also be defined in terms of the neutron
As in the case of the neutron cross sections [cf.
speed or lethargy.
Eq. (4.59)],
= q(v) = q(u)
q(E)
It
is
is
v,
is,
if
scales
it
is
it
is
is
is
is
uniform.
In this model the slowing-down density
different from zero for all
< E9, but
identically zero for all
energies
> Ea [since by (4.39)
no neutrons are speeded up by scattering collisions, and no neutrons
All neutrons introduced into the medium
are introduced at
> E0].
That
(4.61)
q0
(4.62)
SEC.
4.2]
SLOWING-DOWN
PROCESS
89
^ q(E)
q(E)
and
If
we
(4.63)
= constant
= 0
(4.64)
q0
(u)
du
**a^
be
du
If
it follows that
q(u) = {2,(tt)
(u)
(4.66)
This relationship
materials.
neutron-absorbing
It is of interest to compare the expression (4.65) with the correspond
The symbol denotes
ing relation given in terms of the neutron energy.
in
the average gain
lethargy per collision; its counterpart in energy space
is
(1
a)E/2
[see
Eq. (4.40)].
i \ du
?(")
-j
- a)E
(4.67)
90
REACTOR ANALYSIS
[chap. 4
This statement follows directly from the physical fact that the number
of collisions which occur in traversing dE must be identically the num
ber which occur in traversing du if dE implies du. But, by (4.61),
q(u) = q(E), and by (4.44), du = dE/E; therefore,
q(u) du
and
it follows that
2q(u) du
1
a
(4.68)
0.6
forms given in
(4.67) provided that a is near unity.
Note that if this condition is to be
OA
satisfied,
l-a-"
condition (4.48)
as
0.2
and
mass large,
Density by ColliIntervals.
sion
We turn our atten
tion now to a more realistic treat
c. Scattering
0.2
Fig.
4.12
0.4
0.6
Approximation
0.8
for
l.O
.
SEC.
4.2]
SLOWING-DOWN PROCESS
91
aE'
Fig.
4.13
REACTOR ANALYSIS
92
[CHAP.
will produce neutrons with energies <E. But this number is precisely
the probability that a neutron which suffers a collision at energy E'
will emerge with some energy <E. We obtain this from the distribu
tion function (2?) [see Eq. (4.39)]; thus the fraction of neutrons scat
tered to energies <
is
P( <
E)
(f
(4.69)
The total number of neutrons scattered from dE' which emerge with
is
'1
4>{E')
"f
X.(E')
f.f
dtf'
(4.70)
In order to obtain the total contribution to q(E) from all such intervals
qo(E
exactly go
number, a
The total
to q(E)
is
is
- aE0)
- a)E0
(4.71)
m- j>*W)[^]^ ^f>
(4.72,
is,
<
This expression gives the slowing-down density q(E) for all energies
in the interval aE0 <
E0, that
> aE0.
(4.72) applies for all
< aE0, the source cannot contribute directly to q(E) and the
When
omitted; thus,
5o term
q(E)
[B/a2.(E')4>(E')\
^L]dE'
for
is
is
integration
< aE0
(4.73)
SEC.
4.2]
SLOWING-DOWN PROCESS
B/u
93
E/a < E0, and the flux is therefore defined over the entire interval.
The above results involve two unknowns 4>(E) and q(E). In order to
obtain a complete definition of the neutron distribution in energy,
This can be obtained from the neutronConsider, therefore, an energy interval dE about E
The number of neutrons disappearing from this interval
Fig. 4.13).
Z.(E)
4>(E)
dE
fE/a2.(E')
JE
(E') \n
\_\\
dE
- o-)E
dE'
for
<
aE,
(4.74)
A'o
Z.(E)
HE) dE
pME')
HE')
y^jp]
dE' +
(r^fo
(4.75)
The results (4.74) and (4.75) define the neutron flux (E) for all values
The solution of these integral equations when substituted
of E < Ea.
into the expressions (4.72) and (4.73) yields the appropriate forms for
q(E) over the entire range 0 < E < E0.
We demonstrate the procedure for solving these equations by carrying
out the calculation for the first two collision intervals, namely, aE0 Ea
For convenience call the scattering density 2,< in the
and a2E0 aE0.
first interval Fi{E) = X,(E) i{E). Then (4.75) may be written
/m
F1(E)
94
REACTOR ANALYSIS
This is
[CHAP.
F1(E0)
(j?)
(4.77)
E-*E,
?0
jT
U
)#0
E -*
as
E0
= Fi(Eo)
Substitution into (4.77) yields the scattering density in the first collision
interval (aE0 < E < E0)
=
me) me)
^a)-Eo($y
l/(l-a)
(4.78)*
write
Fm
(1
(1
Note that the integration over the interval E to El a has been separated
into two parts so as to accommodate the
change in the functional form of 4>{E) as the
proceeds from the second collision
integration
otE0
E
a2E0
Fig.
4.14
--
mF,{E)
Second collision
interval.
Ft(E)
ME) ME)
(i
(1
(4.78).
a)E
- a)2()1/(I
goy'
(1
(4
K
)+l
80)}
(r^ - (irj +
[1
"
al,(1-)]|
(481)*
(1
a)haaini
a>
SEC.
4.2]
SLOWING-DOWN PROCESS
d. Asymptotic
densities Fi and
Fi{aEo)
It
Solution.
F 2 at E
~
(1
95
is of interest
= aE0.
- a)E^-<,
Fl(aEa)
(1
- a)E^'^
Fi(aE0)
- a1"1-']
Fi(o#0)[l
(E)
at
= aE0; there
the discontinuity
as the neutrons
progress down the energy scale, the collision densities of any two adjoin
ing collision intervals, say F(E) and Fn-\(E), which describe the energy
distribution of the neutrons, approach the same analytical form, and dis
similarities between the two functions recede to higher and higher deriva
tives.
As the ratio E/E0 * 0, F(E) approaches a limiting form.
This
limit can
a"2?0,
Thus
tJK -
JB
(1
Ot)h
+ ("
Ja'-'Eo
(1
F(E)
|.
(4.82)
Ct)E
The col
~ F_i(i?)
F{E)
(4.84)
qiE)
_~ fE/"F{E')(E JE
(1
aE')
a)E'
dE>
[*/- (E
]E
(1
- aE') dE' _
- a){ET
The last equality follows from the fact that no neutrons can
~ qo
(4-85)
be lost in a
REACTOR
96
[CHAP.
ANALYSIS
F(E)
or, if we use q(E) =
energies below
solution
(E)
the source
thus,
<7o/;
||
(4.86)
q0,
q(E)
The asymptotic
Z.()
50,
= a2E0,
&.(E)
4>(E)E
(4.87)*
80
1
70
60
50
1
\
40
, Asymptotic
o
30
20
\\
1
1
N
1
^
10
0.2
0.1
n-
O0
4. 15
of source
4th
a*
0.3
3rd
4-
, Collision density
F(E)Ea/qa
1
1
1
1
Jgttl
Fig.
form
0.4
0.5
' 0.6
2nd interval
0.7
0.8
0.9
1.0
1st interval
i0 = energy
material
of mass A
= 7.
F(E) E0/qo in
This function
is
a medium of pure
plotted in terms
SEC.
4.2]
PROCESS
BLOWING-DOWN
97
of the energy ratio t E/E0, where E0 is the energy of the source neu
The discontinuous form of the collision density is clearly dis
trons.
It is of interest to note that,
played in the first two collision intervals.
in the third interval and beyond, F{E) "smooths out" and rapidly
approaches the asymptotic solution E0/E% (given by the broken-line
In the preceding discussion of the asymptotic solution, it was
curve).
remarked that this form could be used as an approximate description of
F(E) even in the first and second intervals, especially for the heavier
It can be
(4.51) and making a suitable change of variable from E to u.
shown that the slowing-down density q{u) is given by [see also (4.53)]
q(u) =
OC
fU2.(u')
Jut
4>{u'){e"'-"
- a) du' +
~
go(e"^
1
"}
Of
u < o + In
=
j /
"
- a) du'
"
,
S.(m') 4>(w')(eu'-u
u > uo + In
(4.88)
with
u0
a source of monoenergetic
F(u) =
"
! / F(u')
a Jo
^
1
e'- du'
= T-^ I
, F(u')
1 a /-ini
eu'
-"
du'
u < In
u > In
(4.89)
integral equations (4.88) and (4.89) for q(u) and F(u) may be
solved by the same procedure as used in the case of the energy variable.
It is easy to show, for example, that the collision density in the first
These
interval Fi(u)
is given
by
Fl(tt)
intervals.
q*_a
It
is of interest
(4.90)
to determine
the
REACTOR ANALYSIS
98
[chap. 4
asymptotic solution for F(u) in the lethargy range far removed from the
source at u = 0. In this range we use the form
1
1
fCn-l)
a yM-in
In
du'
"
/
l-aj(n_i)inl
e-'- du'
Fn(u')
(4.91)
"
F,(u')
/
i-yu_ini
.
e'- du'
in the nth collision interval. The approximate form used at the right
follows from the same arguments as used in the calculation of F(E).
The solution to Eq. (4.91) is
L4
F,(u>
II
by direct substitu
Thus in the asymptotic range,
as can be verified
1.3
tion.
Ail'"
1.2
1.1
(4.93)
l"a
Fig. 4.16 Collision density F(u) in pure
= 2.
scattering medium of mass
is
3lno
1ns
Inj
if
4.3
a.
is
is
(4.92)
Collision dansl
*.<) =
= 1.38.
LB
cross sections.
SEC.
4.3]
SLOWING-DOWN
PROCESS
99
tion losses and fission sources [see Eqs. (4.74) and (4.75)].
As a first step in deriving the expressions for the neutron density in an
absorbing medium we consider the relatively simple case of an infinite
The
medium of pure hydrogen, for which a rigorous solution is possible.
model for this calculation is defined by the following set of assumptions :
(1) infinite medium of pure hydrogen (A = I); (2) 2a() j 0; (3) go
neutrons of energy E0 appear per unit volume per unit time throughout
all points of the medium; and (4) the hydrogen nuclei are stationary.
We begin the analysis with the calculation of the total collision density
F(E) defined by
F(E)
- ME)
4>{E)
ME)
*{E)
[z.(E) + ME)]
condition
ME') (E')
(E)
rr,
+p
E <E0
(4.94)
The term at the left gives the rate at which neutrons are removed, by
both absorption and scattering collisions, from the unit energy interval
around E.
The integral, as in Eq. (4.75), gives the rate at which neu
trons enter unit energy interval about E by being scattered down from
Note that the frequency function for scattering in
energies above E.
hydrogen takes the simple form 1/E' if E' is the neutron energy before
collision.
Because a = 0 for hydrogen, a neutron can be scattered from
Consequently, the neutron source can
Et> to any energy 0 < E < Ea.
contribute directly to the neutron population at every energy from Ea
down to zero. Thus the term qo/Eo appears in the balance relation for
This form for the source term is only an approximation,
all E < E0.
Actually,
all neutrons that appear from the source do not
however.
REACTOR ANALYSIS
100
[chap.
enter into the slo wing-down process; some source neutrons are absorbed
at the first collision, and therefore the go factor should be corrected to
account for these losses.
It would be better, then, to use the form
Number of neutrons
slowing down from
E0 per unit volume
per unit time
..
J^.
q{E)
q{Eo)
Z,(g0)
f"
[za(E0) + S.(^)
(4.95)
But in many
cases, ^,a{Ea)/Z,{E0)
and
<5C 1,
q(E0)
?0
a differential
equation.
--mi-m
f(E)
-^^jrj
F(E)=F(E0)exp[JE
But from (4.94), F(E0)
W) -g exp [ - ^
= q0/E0; therefore,
q(E)
J*
X.(E')
4>(E')
= go exp
d~ +
^-
EF(E)
f*'
be
(4.97)
written in terms
- qoPn(E;E0)
(4.98)*
not absorbed
between
and E0.
Thus the
identified as the
nonabsorption probability:
is
verbal
where the subscript H has been added to denote hydrogen.
statement of this result might be expressed as follows: The number
of neutrons which succeed in slowing down from E0 as far as
given
multiplied
the
by
starting
of
neutrons
at
(namely,
by the number
g0)
E0
is
SEC.
4.3]
SLOWING-DOWN PROCESS
101
(4.99)*
2,(E')E'
= probability that a neutron will slow down from E0
to E without being absorbed (in a medium of pure
hydrogen)
If
\- ]o -^y-J
we
(4-100)
absorbed.
It can be demonstrated by means of a crude probability argument
that these results (4.99) and (4.100) are indeed the nonabsorption prob
abilities defined above. For convenience, we will carry out the calcula
tion in terms of the lethargy variable.
Suppose that the interval 0 to u
Ph(u;0)
[Jp-
i-i
m-i
(4.101)
ith interval
-n
(4-102)
T
-f-"w-i
pj1
Pi =
subintervals
_o_n
therefore,
If
./V
Fio.
4.17
Lethargy sub
intervals.
Am cannot depend
REACTOR ANALYSIS
102
[CHAP.
2/
AT
limit
4'
-4-Sr]
2("i)
Aw,
may be replaced by
lnpH(U;0)
Jq
>
0,
If
we pass to the
i-1
^]
['-
ft
pH(M,0)
(4-103)
I1[1--1^J
i-l
Ph<.-0)
-^y-
escape
probabilities.
These
functions
treatment
is
and p(u;0)
We turn now to
treatment of the more gen
in
down
medium
which contains absorbing
eral problem
For the present we consider a medium of a single nuclear
nuclei.
later analysis the problem of slowing down in
species and leave for
In this more general treatment both a and
mixture of nuclear materials.
of slowing
Integral Equations.
a.
q0(e~"
du' +
/_ln
r^
a
f"
u < In
S.(m') *(')("'-"
a)
2,(u') (u')(e"'-"
Jo
r-1
a
q(u) =
is
is
2a(E) are different from zero. As before, we assume that the nuclei
are stationary and that the neutron source (g0 neutrons per unit volume
monoenergetic (source neutrons have lethargy u0).
per unit time)
identical in form
The integral equation for the slowing-down density
=
If we take for convenience, Wo
then
to the relation (4.88).
is
- a)du'
(a)
(4.104)
u > In
(6)
SEC.
4.4]
SLOWING-DOWN PROCESS
103
The collision
F(u) =
S(w)
*(u)
["
Jo
S.(m')
(u')
'- du' +
u < In
"
a
(a)
(4.105)
/u-lni
> In
(6)
The term at the left gives the rate at which neutrons are removed from
the unit lethargy interval about u by both scatterings and absorptions.
The terms at the right give the rate at which neutrons are added to this
interval by scattering in from above.
b. Solution in Asymptotic Range.
A general solution of Eqs. (4.104)
and (4.105) has not been found; however, an approximate solution may
be obtained which is valid for small absorption.
This approximation
is based on the assumption that the asymptotic solution applies for all
values of u > 0. We select then Eq. (4.104b) for the slo wing-down
This is one equation in the two unknowns q(u) and (u). As
density.
a second relation we introduce
^q(u)
-Z(M) *()
(4.106)
with the initial condition q(0) = q0. This relation is entirely satisfac
tory from physical requirements inasmuch as the only mechanism for the
loss of neutrons from the slowing-down stream in an infinite medium
is the absorption collision.
However, even with this simplification the
set of equations (4.1046) and (4.106) cannot be solved rigorously.
There
fore we introduce the next approximation, which is to regard the absorp
tion as small so that we have nearly the solution for the pure scattering
case.
Under these restrictions there can be no rapid changes in (u);
such changes could arise only from the high rate of neutron removals
due to the presence of strong absorbers.
Thus we assume that the
scattering density 2(u) (u) changes only slightly over each collision
interval u
In
to
w.
it becomes pos
2.(u') *(')
~ S.(m)
*() + (u'
u)
[Z.(u) *(u)]
(4.107)
REACTOR ANALYSIS
104
[CHAP.
~ fS.(u) *()
7=1-
where
As
a first
"
and
2.(u) (u)
q(u)
This equation
The result is
(4-108)
(4.109)*
^ g()
derivative [2,(u)
2,(u) >()
(u)]'
[2.(u)
(")]
in (4.108); thus,
- 7 ^ ?()
*(")]
- a)
ln*
2(1
yields,
(4.1046)
approximation to (4.108)
g()
We
- t{ [2.()
{u)
(4.110)
which relates the slowing-down density q(u) to the flux (u) in a medium
of light absorption.
One of these functions can be eliminated by using
this relation in (4.106). If we solve for q(u), we obtain the expression
It
^J
- Jq
(4.111)
0 both
slowing-down
density
approaches
[see
Eq. (4.98)].
Equations
and 7 differ by
quantities of the order of 25 to 50
per cent of their values (see Fig.
Thus, although
4.18
Comparison of
and y.
7, so that
[2(u) + Z.()]
= 24()
()
(u)
(4.112)*
SEC.
4.4]
SLOWING-DOWN PROCESS
105
[if
Neutron energy
(a)
cross
section,
(a)
Resonance structure;
(6)
is
is
which included the resonance peaks (see Fig. 4.19). Since the absorption
very large in the region of resonance peak, the require
cross section
ment that the scattering collision density be a slowly varying function
over a collision interval (In 1/a) will not be satisfied, and separate con
sideration must be given to the problem of evaluating the nonabsorption
probability in regions which include an absorption resonance. It would
appear, then, that the term "resonance-escape probability"
misap
plied to p(ui;u0) however, in many practical cases (4.113) may be used to
good approximation even in the presence of absorption reso
obtain
medium
nances, since the macroscopic absorption cross section of
a
particularly
large
consisting of several nuclear species may not exhibit
Morevalue at
resonance of any one of the nuclear types present.
a
a.(E)
REACTOR ANALYSIS
106
[CHAP.
(in energy units) the resonance at a height of one-half the maximum value
On the basis of a physical argument,
of the tr() curve (see Fig. 4.19a).
it is clear that, if the resonance is high and broad,1 many neutrons will be
captured in this energy range (see Fig. 4.196).
Consequently, the neu
tron density, and therefore the flux, will be highly depleted in this vicin
ity, and the assumption on which the result (4.111) is based would be
Thus (4.113) could not be expected to give a good estimate of
violated.
the resonance-escape probability in such a system.
On the other hand,
if the absorption resonances are "sharp," (4.113) may still be applicable.
The sharpness of a resonance is determined by the ratio A,/rf, where
Ai?i denotes the average loss in the energy of a neutron per collision
around the resonance energy 2J<. The resonance is sharp if this ratio is
By this definition the resonance at energy
much greater than unity.
Et in Fig. 4.19a is sharp inasmuch as A^3/r3 y> 1. The resonance at
E4, however, is broad because in this instance AEt/Tt ~ 1.
Equation (4.113) is applicable to systems which exhibit sharp reso
nances due to the fact that in these situations a large fraction of the
neutron population will never "see" some of the resonances.
So long
as the average jump in energy per collision at the resonance energy is
much larger than the resonance width, there is a strong likelihood that
the neutron will be scattered from some energy above the resonance to
some value below without ever occupying the energy band in which the
Thus many of the neutrons in the
large resonance cross section appears.
system will avoid capture by never having been exposed to the resonances
and thereby enjoy a greater opportunity to reach any specified energy
level below the source energy.
However, if the resonances are broad
in comparison to the average energy jump in the vicinity of the resonance
energy, no neutrons will succeed in jumping clear across the resonance in
one collision, and all neutrons will spend some time in the resonance
As a result, the probability for capture increases, and the
energy band.
resonance-escape probability for a system which possesses broad reso
nances will be significantly less than one for a system which has sharp
resonances.
are not meant to imply, however, that the result
(4.113) for the resonance-escape probability is rigorously correct for sys
tems which have strong but sharp resonances.
The intention here is to
These arguments
And the concentration of absorber large, i.e., when 2a is of the order of S,.
SEC.
4.4]
SLOWING-DOWN PROCESS
107
offer some justification for the use of this formula in systems which con
In this connection it is appropri
tain appreciable amounts of absorber.
ate to note that (4.113) takes the rigorously correct form in the limit as
Furthermore,
1, i.e., in a system of pure hydrogen [cf. Eq. (4.99)].
{
we know that the presence of a high sharp resonance in the cross-section
curve does not act as an impenetrable barrier to the slowing-down process.
Many neutrons will get through such a barrier in a single collision.
Equation (4.1 13) shows that, although the appearance of such a resonance
will decrease the resonance-escape probability (the integrand increases
1,000
c
h -1} \h
\
100
-i
'
'
\f\*
1
\ Tl ff
10
0.1
0.01
0.1
1.0
10
10*
10*
10*
10*
10*
10T
106
10'
(a)
1,000
n
A
--
loo
10
0.1
0.01
f
0.1
1.0
10
102
10J
10*
105
Energy, ev.
(b)
Fig.
4.20
Cross-section curve for a nuclear fuel:
(f>) fission cross section for U"s.
(a)
U"s;
is
sharp resonances.
It was pointed out previously that the preceding analysis was inappli
cable to media which contained heavy absorbers (2 ~S.), especially
large, broad resonances appeared in the absorption cross-section curves.
if
Energy, ev.
REACTOR ANALYSIS
108
[CHAP.
is
Em
Energy
Fig.
4.21
e0
absorptions.
A detailed discussion
SLOWING-DOWN PROCESS
SEC. 4.4]
109
for the special case of a medium of small absorption cross section. Excep
tions to this restriction were allowed for media which possessed large
absorption
nances.
and (4.111)].
Two problems of importance to reactor physics which were not treated
in these analyses are the slowing-down distribution of neutrons in the
energy range close to the energy of the source neutrons and the general
These calculations
slowing-down problem in strongly absorbing media.
cannot be carried out by elementary analytical methods.
The experi
mental approach mentioned in connection with the calculation of the
resonance-escape probability (see Sec. 4.4c) is an alternate method for
problems,
however,
extremely
REACTOR
110
ANALYSIS
[CHAP. 4
pendent of energy, and (5) the absorption cross section is inversely pro
portional to the relative speed between the neutron and the nucleus.
These assumptions are similar to the set used in Sec. 4.2c. The principal
is
(if
it
is
is
is
this decision
scattering
is
if
if
is
the event
an absorption, a new neutron
introduced at the source
the event
a scattering,
speed and the procedure repeated; (4)
decision must be made regarding the speed of the target particle and the
relative angle of approach of the neutron to the target.
Having made
(3)
is
it
it
neutron
thus terminating a history. By tracing many such histories
possible
to compute the average number of the various events (such as scatterings
and absorptions) per energy interval for all energies per neutron intro
Thus the collision densities, neutron flux, and
duced into the system.
the slowing-down density can be obtained.
Note that the resonance-
/.
Nuclear Energy,
2,
is
is
probability
an especially easy quantity to compute by this
the number of neutrons which survive to
method; clearly, p(E)
divided by the total number of histories examined.
specific calculation for the scattering collision density was carried
out for the first two collision intervals using a slowing-down medium of
escape
153-167,
(1956).
*
is
SEC. 4.4]
SLOWING-DOWN PROCESS
111
mass number
Fig. 4.22.
Sv
=
Vt
Vt
m
/:U-7'xA*
(4.114)
and
open circles in the figure represent the points computed by the Monte
The discontinuity in the collision density at the lower
Carlo method.
400
Fig.
420
460
440
480
4.22
500
1 /v
J 520
absorber.
This function
1/t absorber.
9o
(1
where
Tt a 293K,
- a)x,
fxV
\ZTN)
\N.)
\x0)
\x
is
+ KiV* -a)/(l-a)
+ k5
(4.115)
fx0
4.15).
(4.116)
a.
is
the nuclear
is
9o
It
)/(l-o)
{x)
(1
Z.(-r)
(4.117)
a)j.-0
is
is
is
= 8.6275
lO"5
is
Jo, where a
given by Eq. (4.20).
By using a relation of the type (3.43).
Xi =
a'
>Jfc = 1.3804
380
ev/K.
The
REACTOR ANALYSIS
112
[CHAP. 4
method is not limited, however, by the value of k, and the flux distribu
tion in more highly absorbing media can be obtained as readily as in the
It is seen from the figure that the Monte Carlo data showpresent case.
some scatter; this is due to the poor statistics in the high-energy range.
The essential features of the collision density in this range are well
exhibited, nevertheless.
4.5
a. Asymptotic
slowing
Equations.
down of neutrons
from a monoenergetic
consider the
source in an infinite
we
which
is a
calculations.
Generated for wjivans (University of Florida) on 2015-09-22 17:18 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
The appropriate
N species)
...
If we
iih
nuclear species
range
u > In
q(u) =
2,(m) 4>(u) =
2>(u')
(') (e"'-u
- 0 du'
(4.118)
(4.119)
SLOWING-DOWN PROCESS
SEC. 4.5]
113
ability that these collisions increase the lethargy of the neutron to a value
above u must be calculated separately for each nuclear type; finally, the
combined contribution from all scattering collisions must be summed up
with the length of the scattering interval (In l/a<) varying from one
Similarly, Eq. (4.119) is a balance condition for the
nucleus to another.
u.
lethargy
unit interval around
In the special case wherein all nuclei
b. Mixture of Pure Scatterers.
in the mixture are pure scatterers, Eqs. (4.118) and (4.119) have a simple
The integral equation (4.119), with
solution.
[so
Si" n 0
has the solution
are
where the
C,
2<()*(u)
= C,
set of constants.
2,
that 2,
iThe
t-i
1,2,
use of
Z]
,N
of (4.118) gives
N
i-1
"'
If
fc2>(u)
(u)
(4.120)
AT
for a mixture by
N
(4.121)
be
written
q(u) = ?(")2,(w)
*() =
(4.122)
<l(0)
is
is
is
The last equality follows from the physical requirement that, in the
absence of absorbers, q(u) = q(0).
mixture of pure scatterers
treated, then, as a single scatterer, with the quantity
replaced by the
This definition
average defined in (4.121).
plausible from
physical
standpoint since
effective in slowing down neu
given nuclear species
trons only in proportion to the fraction of the scattering collisions which
It should be noted that, since 2i')(u) lethargy
are due to that species.
is
is
is
tends to be less
dependent, (w) must be also; but the dependence of
Thus
particular
sensitive because of its functional form.
usually
chosen to represent the entire lethargy range of interest.
a
JV
obtained
directly
ANALYSIS
REACTOR
114
[CHAP. 4
9(0)
\{u) 2. (it)
Note that if one assumes that all the scattering cross sections have the
same functional dependence, then
|(u) =
= constant
(4.124)
this
case of
independent of lethargy;
that all
then
.V
are
(4.126)
The simplest
is,
2.(it)
[2<()
yields
i=
*()
fcZ>(tO
q(u)
If
is
y, ;m
where
-^
ln2
[2'()
*()]
(4.127)
of Eq. (4.118)
*()])
a<
(4.128)
(4.129)
neglected, we obtain as
*(')
u)
2<"(')
first approxi
We
obtain
9() =
is
>-l
f,2''>(u)
*(M)
"
7"(u)
^9(M)
(4130)
SLOWING-DOWN PROCESS
SEC. 4.5]
y(u) = <~
where
(4.131)
[S*(u) *()]'
^q(u)
115
case, we assume
-2(u) *(u)
(4.106)
Thus
q(u) =
fc2>()
*() +
2.() *()
7(m)
=
is often used
1()~7(M)
Generated for wjivans (University of Florida) on 2015-09-22 17:18 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
so
that
(4.132)
that
(4.133)
()
(4.134)*
This relationship between q(u) and 4>(w), which seems to be related to the
accurate equations only through several very limiting assumptions, has
been used in many calculations on actual reactors with very good results.
It will be generally assumed throughout this work.
An explicit expression for the slowing-down
substituting (4.134) into (4.106). The solution
q(u) = ?(0) exp
wherein
absorbing
the exponential
medium.
[L
density
is
obtained
by
is
- P
(4.135)*
Jo {(') 2,(u')J
probability in an
(4.132) in
necessary
from the
(4.131)].
Mu)
*(M)
*(0)
exP[-
f MO*q
(4.136,
from (4.121).
This expression for the flux may
now be used to compute the first estimate 7i(u).
We obtain from (4.131)
where
%(u)
is computed
Yi()
W _
[2<<>(u)
second estimate of
From (4.132),
gn(")
Z.(u) + 7i() Z.(u)
,.A
g(0)
Jo
X exp
f(0 2.(uO
f.(')2.(')J
for
(4.137)
011
be used
[CHAP. 4
*,()]' f.T*
[2<*(u)
i-l
This relation may
ANALYSIS
<.
REACTOR
{,
116
is
is,
d. Moderating Ratio.
It was shown in Sec. 4.1 that the effectiveness
of a material as a slowing-down medium was determined in part by its
value of
the average gain in lethargy per collision.
Another factor
which can affect this characteristic of a substance
of course, its macro
scopic scattering cross section 2,. The product of these two parameters
gives an indication of the moderating properties of
This
material.
quantity
sometimes called the moderating power:
= moderating
2a
power
(4.139)
is
Thus the larger the moderating power of material, the greater its effec
tiveness as a slowing-down medium.
This number, however,
not a
complete measure of the slowing-down effectiveness.
better quantity
the moderating ratio, which takes into account the adverse effect of
The definition of this quantity
high-absorption cross section.
is
is
'
= moderating
ratio
(4.140)
is
The substitution of this result into (4.131) yields the second estimate
7u(w), which in turn gives m(u), and so on. This procedure should
converge quite rapidly. As pointed out previously, however, for most
practical systems of interest, Eq. (4.134) should be entirely adequate.
SEC. 4.5]
BLOWING-DOWN
PROCESS
*()
0 correspond
117
[see Eqs.
(4.111), (4.113),
9.Pl(;o)
= q0pi(u;0)
q2(u)
- exp r
(
<w
p.(;0)
["WW
du'~\
Jq
....
(4.142)
If
we call m, = (2,/20),-,
g2(u)
(4.143)
= e-u(l/m,-l/m,)
(4 144)
2.
is,
<5C
2 and
we have
|2.>
2a,
2.
if
> q2(u)
the moderating ratio of medium
Good moderating materials, then,
greater than that of medium
It should be recognized
possess large
(small A), large 2,, and small 2.
that, in using the moderating ratio as a criterion for moderating effective
necessary to establish unambiguously the cross sections which
ness,
enter into the calculation.
It
evident from (4.143) that the proper
is
is
it
is
suitably averaged
range
is
a rough estimate
sufficient,
has become customary to compare moderating media in terms of
these properties at thermal energy (0.025 ev).
The moderating ratios at
and
it
4.1
Material
Moderating
Ratios
Moderating ratio
H,0
164
D20
13,200
Be
144
236
[CHAP. 4
REACTOR ANALYSIS
118
4.6
Infinite Homogeneous
Multiplying Media
Arbitrary Source Distribution. Our considerations of the slowingdown problem so far have been limited to infinite media with monoa.
^ q(u)
q(u) =
(4.145)
&,(u) *()
(4.112)
any change in q(u) must be due to the combined effect of the removals
S(uJ
Fig.
4.23
Source
function S(it).
4>{u)
initial conditions
5(0) = 0
5(0) = 0
(4.146)
which imply that the lethargy scale is so selected that the origin is well
above all neutron sources (see Fig. 4.23) ; thus no neutrons appear in the
system at energies above the value which corresponds to u = 0. The
solution to (4.112) and (4.145) is readily shown to be
9() =
/"
S(u'Mu;u') du'
(4.147)*
SEC. 4.6]
BLOWING-DOWN
where
p(u;u>)
PROCESS
exp
119
[-
(4.148)
u = 0 (corresponding
S(u) = q5(u)
(4.149)
du =
jo"
SM
S(") du =
(4.150)
q0
exp
jf"
qB
9(u) =
jf"
is
is
Thus j(w)
i(u) du =
f~
{,
c.
(4.152)
is
is
/_".
be
REACTOR ANALYSIS
120
normalized;
[chap. 4
dE =
i(u) du
(4.153)
z{E)
f~
is fQ"
thus,
some experimental
One of the initial assumptions (2) was that the nuclei were stationary
thus tbe neutrons may slow down all the way to u =
The source
1.000
750
500
Fig.
=
E, Mev
energy
4.24
9
3
}
Neutron
Jo"
therefore
S(u)
is
2,0')
4>{u')
be obtained
du'
(4.154)
(4.155)
if
is
fQU
q(u) =
f0"
(4.147).
If
du
Jo"
statement
fQ"
S(u) du
per unit
introduced
/number of neutrons
of neutrons
fo"
250
2/(w')
(u')
du'
(4.156)
If
PROCESS
BLOWING-DOWN
du
Jo S(u)
1
f"
1,
121
then
and
j(tt) du
f* 2,(u')
{u')
du' =
fQ"
SEC. 4.6]
2,(u')
du'
4>(u')
(4.157)
is
if
by virtue of
(4.155) yields
/"
()
(4.158)
f"
Jo/
jzm
f2<(w)
(4.159)
('
"
Jo
For convenience
^7^r
f'lW ^)u')
f2,(u) Jo
(4-161)
fi(w)
dw~]
j(m') exp
y(u) du
Jo"
(4.160)
Ku) s W&j
|fe)
f*
du'
we define1
y{u)
du'
(4.162)*
and
is
condition,
Condition (4.162)
requirement
is
no thermal group.
it
tion, there
is
is
This
of the
is
it
multiplying medium.
We will have more to say of this later.
The fission spectrum z(E), foi the particular fuel Um, has maximum
Mev. For many purposes
adequate to group all
value at about
neutrons from fission at some one energy in the neighborhood of the posi
be confused with
j(u) = *(u)
Not to
0.
is
(4.163)
ANALYSIS
REACTOR
122
[CHAP. 4
y(u) exp
y(u) p(w;0) du
Jq"
Jq"
- fjp(u') du'] du
(4.164)*
Since the cross sections which appear in (4.162) and (4.164) are generally
obtained from experiment, simple analytical representations usually are
not available for these quantities, and the evaluation of the integrals must
be done numerically.
Fig.
4.25
That
fraction of
collisions
at u which
successfully
|_are fissions_
reaches lethargy
|_w
_time at u
neutron at u
probability that
Number of neutrons
produced per unit
volume per unit
average
average
number
number
collisions
neutrons
y(u) du
f*
fQ"
km
is
is
There
an alternative way of regarding (4.160) which may be more
given generation
instructive. The ratio of the number of neutrons in
to the number in the preceding generation
the multiplication constant
[see Eq. (3.16)]; then
(4.165)
is
is
where kx
defined as the multiplication constant of the infinite medium.
If the composition of the medium specified, then all the cross sections
are known [and therefore y(u) and /3(w)] and the integrals may be evalu
is
if
is
is
y{u) du
Jo"
|Q"
r.
is
supercritical.
greater than unity, the reactor
Alternatively, whatever the value of the right-hand side, some number
may be found such that
value
ve
critical
if
If
ated.
is
is
is,
(4.166)
SEC. 4.6]
SLOWING-DOWN PROCESS
123
i.e., for any composition of the infinite medium, there exists some (ficti
tious) average number of neutrons per fission which would make the sys
tem critical.
be more
a slowing-down
probability that
(4.168)
process consisting
of these
= S[u'
- {u
{)]
du
(4.169)
where 5 is again the Dirac delta function. The physical statement of this
function is that the scattered neutron will emerge from the collision with
lethargy in du about u only if u differs from u' by exactly the amount .
REACTOR ANALYSIS
124
[CHAP. 4
(u)
du =
2,(u')
4>(u')
I(u;w') du' du
/"
2,(u)
+ "i(")
(4.170)
2/(m')
}()
fo"
2.(u')
+ 2,(
(')
du'
(4.157)
4>{u')
)*(
- (
{)]
du'
(4.171)
is
0,
If
we
unknown function
=
<
for all u
define the origin of the lethargy scale such that j(u)
then the appropriate initial condition to be used with (4.171)
is
This relation
f~
{)
this relation
(4.172)
<K0)
- we
*<
if
[.<ae
?fjP]
(*')
can be
" Mathematical
Theodore v. Karman and Maurice A. Biot,
Methods in Engineering," chap. XI, McGraw-Hill Book Company, Inc., New York,
1
If
is
is
The first integral gives the rate at which neutrons are scattered into the
interval du about w, and the second, the rate at which they appear from
fissions.
The system described by (4.170)
at steady state; therefore,
the criticality condition must be satisfied:
1940.
SEC. 4.7]
SLOWING-DOWN PROCESS
*(u .- 8
*()
- '()
125
(4.175)
and use the first two terms on the condition that {<'() <3C {u). The
substitution of this expression into (4.171) reduces the difference equation
The result is (2, is taken to be inde
to a first-order differential equation.
pendent of lethargy)
2,(u)
()
as j(u) + 2.*(u)
{2.
'()
(4.176)
It
is seen that this result is in good agreement with the relation (4.159).
Note that the resonance-escape probability, given by the exponential of
(4.177), is defined in terms of the scattering cross section, whereas the
form in (4.159) involves the total cross section. The discrepancy here is
not significant however, inasmuch as (4.159) applies to media with slight
This is also the case in the present calculation ; otherwise we
absorption.
could not have assumed that the flux was a slowly varying function over a
single collision interval.
4.7
"Statistical Mechanics,"
pp-
REACTOR ANALYSIS
126
[chap. 4
collision used in the previous calculations [see Eqs. (4.39) and (4.50)] were
defined for stationary nuclei; these functions do not apply, therefore, to
scattering collisions by thermal neutrons (neutrons of energy around kT).
In fact, a scattering collision experienced by a thermal neutron will result
in an increase in its kinetic energy in a large fraction of the collisions.
Because of the thermal motion of the nuclei, some energy can be trans
ferred from the nucleus to the neutron, thereby increasing the neutron
This is in contrast to collisions between neutrons and stationary
speed.
nuclei which always result in a decrease in the energy of the neutron (and
an increase in the energy of the nucleus).
and nuclei.
cated, and detailed descriptions are difficult to specify and even more
difficult to apply.
The behavior of neutrons at the lower end of the energy scale is much
Because of
different, then, from the behavior in the high-energy range.
the basic difference in the nature of the two phenomena, slowing-down
and thermal motion, it is convenient to separate the neutron trajectory
in energy space into two parts.
down process from the source energy down to the thermal range, and the
SLOWING-DOWN PROCESS
SEC. 4.7]
127
V
n(v) dv
(4.178)
The derivation of the functions m and 2J? is developed from the neutronbalance relation for all reactions which can affect the density of neutrons
in a differential element dv. For this purpose we require two additional
functions, namely,
g
= v
|p|
collision
g'.
is
is
REACTOR
128
ANALYSIS
[CHAP. 4
calculation
p<t9J?
u>;
thus,
Number of neutrons
scattered out of element =
dv per unit time
Jv
m(y) dv
po{p,u>)
9D?(V)
dV da
(4.180)
',
-v,
',
Fio.
Velocity
4.26
vectors before and after
collision in a laboratory
system of coordinates.
and
to'
v' =
v'(v.W)
(4.181)
J[V',<o'
d*'
This
(4.182)
Note that
a(p>')
do>'
angle
Also
= <r(pV)
. . , .
(4.183)
(4.184)
which states the physical fact that the scattering cross section to rotate
the vector p (corresponding to v and V) through angle to into p' must be
the same as that which rotates p' (due to v' and V) through w' into p.
In the present generalized notation, u = u'.
SEC. 4.7]
SLOWING-DOWN PROCESS
im/imjirj
129
a change of integra
number
scati >\
/
, . ,
,
/f
.. ..
m(v') dvp<r(p,u) 3fl(V) dVdw = .
Jv,a
tered into dv per unit time
of neutrons
,OE.
(4.185)
Note that we still denote velocity vectors before collision by the symbols
v' and V; however, the integration variables which correspond to the
that
dv' dV' du' = dv dV
do>
(4. 186)
that
speed
between neutron
= w
-W
e' =
w'
(4.187)
If we
square each of these expressions, subtract p- from p'", and apply the
results (4.10) and (4.11), it follows immediately that
P
= p'
(4.188)
[m(v') 9D?(V)
- m(v) 2H(V)]p
If
(4.189)
is
m(v) 9fl(V)
= In m(v)
dv dV dw = 0
(4.190)
functions
F(V) = In 9)i(V)
/ and
such
(4.191)
it follows that
(4.192)
which states that there exists some scalar function of the velocities before
The only scalar
and after collision which is unchanged by the collision.
invariant in the present system is the total kinetic energy of the colliding
REACTOR ANALYSIS
130
Therefore,
[CHAP. 4
we set
/(v) =
F(V) = a(p/F2)
a(mt>J)
(4.193)
where a is some constant, m the neutron mass, and M the nuclear mass.
Finally, we substitute these expressions into (4.191) and obtain
m(v) = 6e*
The coefficients
and
2K(V)
Be^>
(4.194)
If we use
50{,
we
obtain
(set)
is
is
is
a is
where
medium.
is
a m
j,
dynamic
SEC. 4.7]
SLOWING-DOWN PROCESS
131
the definitions of the distribution functions which accounts for the effects
of the absorption losses on the neutron spectrum.
The principal effect
of the absorption collisions is to depress the neutron population over the
energy interval wherein the absorption cross section is relatively large.
Since for most materials the microscopic absorption cross section varies
more or less as 1/v, this means that the neutron densities are increasingly
reduced, from the pure scattering media values, as one progresses down
the energy (speed) scale. As a result, the neutron distribution m is dis
torted and tends to peak at higher values of the neutron speed correspond
ing to the range wherein the absorption cross section is relatively small.
This shift in the peak, and distortion in the shape, of the neutrondistribution function is called hardening of the spectrum, and it is the more
These trends are illustrated in
pronounced, the greater the value of 2.
Fig. 4.27.
Purely scattering
0
Neutron
Fig.
energy
medium
One way in which this shift in the neutron density may be taken into
account is by introducing an effective neutron temperature T.
Thus we
define some fictitious temperature Tn which, when used in the function
m of (4.196), gives a new Gaussian that yields a better fit to the distorted
It
perature Tn which describes the thermal state of the nuclei in the system.
Thus a suitable set of functions for the neutron and nuclear densities in
an absorbing medium may be given by
m(v) dv =
veloci
(4.197)*
REACTOR ANALYSIS
132
cm/m
nr
2K(V)dV-v
It
[CHAP. 4
V and
motion2
If
a.
. .
be noted
mWdv =
we define
and
dv
..
(4-199)
then
m(v) dv =
I m(v) dv
Ja
v2
dv
I m(v) da
Ja
= iirv*
Also, if
,
n,h
(^ftJ
e~mvmkT" dv
(4-20)
, ,
N
n(v) dv = nt m(v) dv =
of neutrons per
unit volume
y
,
,
.
,.
whose speeds he between v and v + dv
number
(4.201)
model.
length per unit time (in some specified volume) of the thermal group, then
a convenient definition of the group cross section is given by
2th, =
ntb
m(v) dv
<r"(\v
- V|)|v -
V|7v~
9K(V) dV
(4.202)
where N is the number of nuclei per unit volume, nth the number of
thermal neutrons per unit volume, and 2Ji and m are given by (4.197) and
The symbol <r(|v V|), which represents the neutron-nucleus
(4.198).
1
Ibid., p. 14.
' The symbol Q denotes a unit vector in actual space (i.e., a
direction) and can be
specified by giving two angles, a co-latitude i? and an azimuth ^; therefore, a = Q(^,^),
and the differential solid angle (i.e., area on surface of unit sphere) is da m sin t? dd d\f>.
SEC. 4.7]
SLOWING-DOWN PROCESS
133
(4.203)
= n.h
= J~
m(v) dv
- V|)|v -
<r-(|v
m(v) dv
(4.204)
V| 2TC(V)
dV
m(v) dv
(4.205)
- V|)|v - V| 2N(V) dV
/v<r-(|v
(4.206)*
cr'"
/ iv (v) m(v) dv
= J-^-
(4.207)*
m(v) dv
m(v) dv =
yV(!>) dv
Jt iv(v)
JQ"
m(v) da =
iv(v) m(v) dv
(4.208)
is
Jo
iv (v)
v
^th
If
where we have used the definition (4.200) for the neutron distribution in
similar operation
speed space m(t>).
performed on the denomi
nator, then (4.207) reduces to
m(v) dv
m(v) dv
(4.209)*
sections
va(v) =
134
REACTOR
ANALYSIS
[CHAP. 4
ax(\v
V|) through the integral equation (4.206). Actually, the precise
forms of the functions <s" are not generally known.
The only information
about a specific reaction rate which is obtained from a cross-section experi
ment is the integral quantity <r(v).
In fact, the experiment itself consists
in performing the integration indicated in (4.206). This may be more
easily recognized if one recalls the experimental procedure involved in
A brief description of this pro
making a cross-section measurement.
cedure was given in Sec. 2.4c.
Such measurements are performed for
neutrons of various speeds (energies) by irradiating a sample of the mate
rial in question with a monoenergetic beam. For each neutron speed,
the experimenter obtains a measurement of the beam intensity, with and
without the specimen. The difference between these intensities, which
is usually given in terms of the reaction rate R(v), where
R(v) = va(v) =
fa"(\v
- V|)|v - V| 2K(V)
dV
(4.210)
determined
of all these interactions is the quantity <r(v), and the graphic representa
tion of these functions is the cross-section curves to which we have made
Finally, it should be pointed out that Eq. (4.206) is a
frequent reference.
statement (based on the model of a two-body collision) of
the physical process involved in performing the experiment.
We have noted that the reaction rate R given by (4.210) is dependent
mathematical
model, the influence of the temperature upon the reaction rate is seen
through the distribution function 9)i(V), which appears in (4.210). Thus,
SEC. 4.7]
SLOWING-DOWN
PROCESS
135
able procedure for generating these data from the laboratory measure
ments is discussed in Sec. 4.7e.
d. Effective Neutron Temperature. The neutron distributions around
thermal energy in both velocity and speed space involved an effective
neutron temperature Tn. This number appears as a parameter in the
functions (4.197) and (4.200) and establishes the position and shape of
the Maxwell-Boltzmann distribution which has been assumed to describe
the thermal-neutron
The calculation of this quantity has
population.
received considerable attention from various investigators.
For the
present discussion we draw from the work of Coveyou, Bate, and Osborn
previously cited in connection with the calculation of collision densities
near source energy (see Sec. 4.4d).
The work of Coveyou, Bate, and Osborn was aimed at determining the
effect of moderator temperature upon the neutron flux in an infinite
capturing medium.
The calculations were carried out for a uniform
medium
of unbound
l/v
absorption cross section. It was assumed that the nuclei were dis
tributed according to the Maxwell-Boltzmann relation (4.198) and that
neutrons were introduced
The details of the slowing-down process and the scattering around thermal
A
energy were determined with the aid of the Monte Carlo method.
brief description of the mechanics involved in the use of this method was
given in Sec. 4.4d. Again, we will not enter into a discussion of the
method but will confine our attention to the use of the results which were
obtained.
The Monte Carlo calculations carried out for the infinite-medium model
above were used to determine the distribution of neutrons
around thermal energy. These distributions were obtained by tracing
The density of neutrons
the individual life histories of many neutrons.
described
REACTOR ANALYSIS
136
results obtained
[CHAP. 4
As in the
i.t
/
/
>
l
/
/
/
1
1
|
1
1
/
1
5 10
100
1,000
1 2
5 10
100
\
1,000
1 2
5 10
100
1,000
Fig.
4.28
presented contain the same total number of neutrons per unit volume;
thus, in the thermal range, the distortion of the solid-line curves relative
to the broken-line curves is a measure of the effect of the (1/tf) absorber
on the neutron speed distribution.
A useful measure of the flux distortion is provided by the concept of
We define this temperature as
the effective neutron temperature Tn.
that number which when used in (4.197) gives the best least-squares fit
of a Maxwell-Boltzmann distribution to the computed flux (solid-line
The ratio of the effective neutron
curve) in the range1 0 < x < 35.
temperature to the moderator temperature TK/TN is indicated in the
The ratio has been omitted from the last
figure for the first two cases.
=
case (A
9, k
2~') because the flux was so severely distorted from a
1 The
the
8EC. 4.7]
SLOWING-DOWN PROCESS
137
(4.211)*
This
correlated fairly well1 all the cases reported in the reference paper.
at which
the available
cross-section
data were
We begin by stating the integral relation for the reaction rate in the
more general notation previously mentioned; thus, in place of (4.210), we
introduce
R(p;Ts)
where
/v
- V|)|v - V| 2R(V) dV
(4.212)*
we define also
<r(|v
< s Sfsji
S(p) = pa"{p)
then with
p m[\9\
|v
- V|
(4-213)
(4.214)
f S(P)
g(v
e, Os) dp
(4lrb/PS(p)eHT"e'V49Vrfe
^
6'N
In deriving the relationship bet ween the reaction rate at one tempera
ture R{v;df/) with that at a higher temperature R(v;6'N), where
> Bn,
5
To within
per cent, for the cases considered. This relation may be used for
systems wherein .4k < 0.2.
This relationship was pointed out to the authors by R. R. Coveyou.
*
ANALYSIS
HEACTOR
138
[CHAP. 4
we will work with the Fourier transforms of these functions rather than
with the actual functions.
For this treatment we require the three-
dimensional
the relation
fix) e*dx
${/(*)}
where
is the transform
If
variable.
(4.216)
s is the three-dimensional
s are given
trans
by
s =
(4.217)
transform of
R is
t R(v;BN)("dv
ff{fl(t>;M) = <R(s;M =
(4.218)
If
Wift L S{p) dg
exp
"
ib
|v
- e|J + ts
v]
dv
(4.219)
by
9(a,b,c)
(4.220)
6)2
aY +
"
*fx]
(x
- cy +
**]
["
{z
wN
v =
18 '
e|2
*w]
"
{y
|v
- ib
a)1
+ ix
(4-221)
mathe
dx
-(4,^)1
exp
(x
/_>p[-4i
is
it
If
is
which
readily obtained by completing the square in the exponent and
this operation
repeated
applying the definition of the error integral.
for the and integrations,
follows immediately that the v integral of
y
p|s
is
dv = (4x0*)' e'e-*"
(4.
-~ -
/exp
(4.219) reduces to
|v
^s'0n)
222)
BLOWING-DOWN PROCESS
SEC. 4.7]
139
Tou /
-S(p) *-d9
(4.223)*
is,
It is of interest to note that the above integral is in fact the Fourier trans
form of pa"{p), the reaction rate corresponding to the fundamental inter
action function.
The procedure outlined above may be repeated for the reaction rate at
the higher temperature R(v;8'N). The Fourier transform of this function
obviously,
Si
If
S(P)e^dg
(4.224)
<R(a;0's)
(4.225)*
)
,
ff-M/tt)l
If
J"
is
K& e~illdt
JF-1{(R(s;M e
(4.226)
to (4.225), we obtain
R(v;8'N)
(4.227)
is
is
- 0n)V
expI-q'
n'
[4r(ft
- Ml)
i
[-(s2 +
y*
* fexp
f'X^
- M]
demonstrated
z2
and
s2
This result
f2.
>;2
x2
y2
if
where
v1
(2x)
(4.228)
is
space, i.e.,
easily
SF-'{e-'l
=^j"
e-o* dt
(4.229)
REACTOR ANALYSIS
140
theorem:1
[CHAP. 4
j"
h{&) m
g{t) h{&
=
e-^df
f"
H (x
G(x')
- x') dx'
We now apply
d'N
RWs)
where 9Ji
is
-'{911(8,
9m)
8H)\
- (2^|
=
e'N
then by (4.228),
(4-231)
(4.232)
BN)
R(pM
9K(v
e,
e'N
d'N
where g() and h() are the transforms of G(x) and H(x).
this operation to (4.227).
Let
3K(*,
(4.230)
eN) d9
R(vv"*>
U^iT'
Jg
14^(0^
flTii
dN)\*
(4.233)*
dP
a material at the
to
its
cross
section
at
lower
temperature
TK. As in all
T'N
the preceding analyses,
assumed that the nuclei are distributed
The above equa
according to the Maxwell-Boltzmann relation (4.198).
tion may be used then to compute the cross-section curves for a reactor
is
it
temperature
is
if
is
is
is
it
is
simple result.
|[^rryj
9)?
\/v,
<4-234)
normalized
d* =
Thus
function.
the absorption reaction rate is
is
a
m vaa(v;e'N)
is
R{vfi's)
BLOWING-DOWN
SEC. 4.7]
PROCESS
141
V'RivM
~ R(v;8N)
(4.235)
in Chap. 6 (in particular, compare with the Fermi age model of Sec. 6.2a).
Finally, it should be pointed out that the general result (4.233) may be
used to compute the temperature coefficient (i.e., temperature depend
This has considerable
ence) of a Breit-Wigner resonance (see Sec. G.5g).
R(v;8N) +
MN
00 N
fl(y;0.v)
But, R
is a spherically
R(v;6'N)
8^
(4.236)
by utilizing (4.235).
R(v;8N) + A6N
l-
^ [t>i2(t>;M]}
(4.237)
where the derivative term may be obtained from the cross-section data at
the temperature corresponding to 0.v. In actual practice,2 the applicafor example, H. S. Carslaw and J. C. Jaeger, "Conduction of Heat in Solids,"
Oxford University Press, New York, 1948.
1 A Monte Carlo reactor code is available for carrying out this computation.
See
R. R. Coveyou and W. E. Kinney, "Neutron Reaction Rate Codes,'' Neutron Physics
Division Annual Progress Report for Period Ending Sept. 1, 1959, Oak Ridge
National Laboratory, ORNL-2842, pp. 144-145, Nov. 9, 1959.
1 See,
chap.
XIII,
[chap. 4
REACTOR ANALYSIS
142
cr.M
(4.239)
If
defined by (4.200).
(4.238)
m(v) dv
m(v) dv
(4-240)
v
m(v) dv
Co
<
f"
m(v) dv
is
normalized function.
The denominator of this result
since m
clearly the expression for the average value of the speed (v) corresponding
to the Maxwell-Boltzmann distribution.
That
(4.241)
is
m(v) dv =
is
If
fo"
is
is
a
found to
be
*
where m(v)
is
/"
vaa{v) m(v) dv
v
/
Jo
T.
c0
2(^Y
(4.242)
2.2).
First we write
(4.239)
(4.243)
SEC. 4.7]
SLOWING-DOWN PROCESS
aa
co^-j
143
(4.244)
a.(*r,)
(^J
2?
c.(kTn)
(4.245)*
No*
(4.246)
substance
is
g.
(j
(4.240) yields
not
In the
case
group cross section may be computed from the general relation (4.209).
This calculation requires, of course,
detailed knowledge of the cross-
is
For
portionality
of the fissionable materials U233, U23s, and Pu239, some of the preliminary
computation has already been performed.
Thus for these fuels data are
available which permit one to compute the thermal-group cross sections
by applying
suitable correction factor to the values obtained directly
from the cross-section curves. One definition of such
factor has been
See, for example, Donald J. Hughes and John A. Harvey, in U.S. Atomic Energy
Commission, "Neutron Cross Sections," McGraw-Hill Book Company, Inc., New
1
is
is
it
York,
1955.
REACTOR
144
ANALYSIS
[chap. 4
the so-called
factor (here represented by
This quantity is defined so that
a(kTn) F(Tn)
(4.247)
where a(kTn) denotes the value to be read off from the cross-section curve
at the neutron energy corresponding to the temperature Tn.
The func
tion f (Tn) denned by (4.247) may be written in the form
KTn)
ffth
(r/4)ff(*T)
(/4)ff(fc7i0
The application of
(4.241)
a(v) m(v)v dv
[*m(v)vdv
yields
jtmhm
jo
a{v)
miv)v dv
(4-249)
This correction accounts, then, for the non-l/v part of the cross
(4.21 1)].
section and is based on a Maxwell-Boltzmann distribution for the tem
Curves of this function have been computed for the
perature T.
principal fuels and are available in Hughes and Harvey2 and in the refer
ence paper.
The data for U233, U235, and Pu239 have been reproduced in
Fig. 4.30.
An alternative formulation of
behavior is given by Westcott.3
f factor given above by
ff(T) =
a suitable
This
correction
is the g
,(^)^(^)J
(4.250)
'Ibid.
J.
1,
a second correction
Jan.
25, 1957.
SLOWING-DOWN PROCESS
SEC. 4.7]
145
kT range where the l/E portion can be significantly larger than the "fast
tail" of the thermal group (refer to Sec. 4.8). Values of the l/E correc
tion factor have been obtained for several important elements and are
tabulated in the reference report.
The purpose of the present discussion
h. Scattering from Molecules.
O.990 I
200
400
600
200
400
600
200
400
Effective
Fig.
4.30
neutron
temperature
L_
600
reactor design.
Since these data are usually best obtained from crosssection measurements using the actual materials, the analytical treatment
of the process is of little immediate concern to the reactor physicist.
However, a discussion of the principal phenomena involved in this inter
action may be of some value in achieving an accurate interpretation of
Moreover, the process itself is of considerable
the experimental data.
interest to the nuclear physicist since the diffraction of slow neutrons by
for example, E. Fermi, Ricerca set'., 7, 13 (1936); W. Lamb, Phys. Rev., 66,
R. G. Sacks and E. Teller, Phys. Rev., 60, 18 (1941); A. Messiah, Phys.
Rev., 84, 204 (1951); A. C. Zemach and R. J. Glauber, Phys. Rev., 101, 118 (1956),
and 102, 129 (1956).
1 See,
190 (1939);
146
REACTOR ANALYSIS
[chap. 4
determined
is
1
internal quantum states of the molecule are comparable to the translational energy of the neutron.
Depending upon the temperature of the
medium, a certain number of these rotational and vibrational states of
the molecule will already be excited, and when a collision occurs, some
fraction of the neutron energy is imparted to the molecule.
This energy
transfer may result in the transition of the molecule to higher rotational
In some
states and/ or the excitation of various vibrational modes.
cases, depending on the neutron energy, the initial internal-energy state
of the molecule, and the mass of the struck nucleus, the energy transfer
may be in the opposite direction, and the neutron may emerge from the
These considerations
collision with a speed greater than its initial value.
become increasingly important as the neutron energy approaches the
thermal range.
Finally, there are the interference effects (4) noted
These are generally very complex, and arise when the neutron
wavelength is comparable to atomic spacing.
The influence of these several factors on neutron-nucleus interactions is
above.
transferred depends
SLOWING-DOWN PROCESS
SEC. 4.7]
147
from the high- to the low-energy range. The cross-section data are shown
in Fig. 4.31.
The steady rise in a. from the 20.36-barn value in the elec
tron-volt range to around 80 barns at 0.004 ev is due to the combined
10
'
'
0.01
0.02
0.04 0.06
1
0.1
'
0.2
0.4
0.6
1
1.0
2.0
4.0
6.0
10
Energy, ev
Flo.
4.31
'
'
very slow neutrons the scattering cross section (o-) B of a hydrogen nucleus
bound to a molecule of mass Am (in units of neutron mass) is given by
WB
4(rrx:y
where (a,) F is the cross section of the hydrogen nucleus in the free (atomic)
The generalization of this result to the case of a bound nucleus of
state.
mass A is
<
148
REACTOR
4.8
ANALYSIS
[chap. 4
Thermal Group.
The analysis of Sec. 4.6 dealt with the energy
distribution of neutrons in an infinite multiplying medium wherein neu
trons could slow down to zero energy.
We found in Sec. 4.7, however,
that in all media at temperatures above absolute zero the neutron popula
tion at the very low energies acquired a peaked distribution around some
energy of the order of kT which was determined by the thermal motion
(temperature) of the moderating material.
We showed also that this
distribution could be well approximated by a Maxwell-Boltzmann dis
tribution if the medium was not too highly absorbing.
In fact, we were
able to specify an effective neutron temperature Tn for this purpose which
a.
was dependent upon the actual moderator temperature and its absorption
properties [see Eq. (4.211)]. This temperature could be used to define the
neutron distribution [see Eqs. (4.197) and (4.20O)] in the thermal range.
distribution
in the low-energy
absorption.
Thus our model replaces the actual neutron distribution by a slowingThe thermal group consists of
down distribution plus a thermal group.
neutrons of one speed, this speed to be defined by the effective neutron
The actual thermal distribution, which was approxi
temperature T.
mated by the Maxwell-Boltzmann relation (4.200), is replaced then by a
"spike" or "group" at E = Elb (see Fig. 4.32). The selection of this
thermal energy EA is discussed shortly.
The mathematical statement of the model described above is (cf. Sec.
4.6a)
(u)
(4.145)
(4.112)
SEC. 4.8]
SLOWING-DOWN PROCESS
wlh
use
Si"th
to Eth.
corresponds
lim q(u)
149
m g(u(h)
(4.252)
fa)
(b)
I
/The rmal
n(E)
Fig.
4.32
Distribution
group
Slowing-down
distribution
So,
E0
(6) model.
2J,h<th.
In the
group
unit volume
The absorption cross section for the thermal group is computed by means
of Eq. (4.238). As already demonstrated, in the event that the absorber
has a 1/y cross section, the integral yields an especially simple form; this
result is given in Eqs. (4.245) and (4.246).
The absorption cross section
REACTOR ANALYSIS
150
[CHAP. 4
of many materials has this l/v dependence, fortunately, and these rela
tions may be used directly to obtain the 2*\
An appropriate definition of the thermal flux which is consistent with
the thermal-cross section formula is given by [cf. (4.204)]
th
J0
n(v)v
dv = ntbv
(4.253)
tribution,
vo
is found to
be1
vo
= v(kTn) =
(2*y
(4.254)
This number represents the position of the maximum point in the Max
well-Boltzmann distribution m(v) [see Eq. (4.200)]. Note that t>0 cor
responds to the energy kTn so that we immediately identify Eth as
E*
imvl = kTn
(4.255)*
It
must be recognized that this model, which treats the neutron dis
tribution by two separate regions (one for the high-energy range and one
for the thermal group), only approximately represents the actual distribu
tion, since the slowing-down process just above the thermal range is incor
S(u)
p(Mlh;M) du
(4.256)
f" SW
V(u;W) du'
(4.257)
ORNL CF
II,
SEC. 4.8]
If
SLOWING-DOWN PROCESS
= 0,
151
(4.257) reduce to
**
^p(tW))-^
(4.258)
PM
2,(u)
(4.259)
&t(u)
p(u)
p(u;0)
= resonance-escape
m = 0 to u
(4.260)
probability from
It is
should be noted that Eqs. (4.258) and (4.259) are not accurate.
with
zero
and
with
emerge
that
the
neutrons
appear
lethargy
imagined
<So
directions of motion isotropically distributed about the point of birth.
It
As these neutrons move away from this point, there are certain prob
abilities per unit distance of travel that they will make both absorption
If the neutrons are to contribute to the slowand scattering collisions.
ing-down density for lethargies u > 0, their first collisions must be scatter
ing collisions; thus in reality, only the fraction 2,(0)/2,(0) of the <S0 neu
trons should be counted as a source for the slowing-down process. We
will continue to neglect this effect {first-flight losses from sources), since
It
the absorption cross sections are often very small at higher energies.
is understood, then, that at any time it is felt to be important the source
strength may be reduced by the ratio 2,/2<.
b. Criticality Equation. Suppose that the infinite homogeneous
medium contains fissionable material and that the only sources of neu
Then, as above, if the system is
trons present are the fission reactions.
to operate in steady state, the number of neutrons produced from fissions
must equal the number of neutrons introduced into the medium, namely,
S(u). For this case the source function has the form
5(u)
[/o",h
(4.261)
The fission spectrum j(w) has the shape shown in Fig. 4.25, but for the
present calculation we set the origin of the u axis at the far left so that
Note that the integration in (4.261) extends
j(0) may be taken as zero.
The last term 2}h<,h
up to but does not include the thermal group.
accounts for the thermal fissions.
We choose to introduce one neutron per unit volume per unit time by
requiring that
|oUth
S(u) du =
(4.262)
Therefore,
du
[fQUlh
[CHAP. 4
= 0 to ulk, we obtain
[|o"th
2,(u')
4,(u') du'
(4.263)
2}^th]
r
v I
/"uth
r
I j(m')
y(u) du
(4.256) and
p(u;u') du'
S
I
in Jo
h(u) p(uth;u)
du]
But,
ANALYSIS
REACTOR
152
(4.264)
+ ^4
,lh
y(u) du
JoI
h(u) p(uth;u) du
into (4.167),
th
\fast fissions/
(4.266)
number of
number of
of neutrons
per fission
average number
ka =
h(u)p(ulh;u)du\
It
is
iI
Jo I r
(it) du
Jo
vc
(4.265)
uti.
km
i J0
Jo
(4.167)
Vc
vc
uth
vc
if
K
We compute
is
the system
not critical, the multi
Now,
plication constant kx may be computed analogously to the reasoning
Thus we employ again the concept of a fictitious
leading to Eq. (4.166).
neutron yield per fission vc and the relation
is
if
is
is
\thermal fissions/
(4.267)
where we use the expression fast fissions to denote the number of fissions
SEC. 4.8]
SLOWING-DOWN PROCESS
153
per unit volume per unit time produced by neutrons with energies above
Etik, and thermal fissions to denote those produced by neutrons in the
thermal group represented by the energy E = E^. Thus we may write
^
If
thermal
fissions
(4.268)
total fissions
(4.269)*
thermal fissions
then (4.266) takes the form
)
be
written in
p(ulh;u) du
(4.270)*
introducing an
for neutrons released according to
escape probability
pth
fQUtb
i(u) ?(*;) du
(4.271)
*.
(4-272)
How
any large error, on the basis of known values for similar reactors.
ever, in general (4.270) is to be used in calculating kx.
If it is a good approximation that all neutrons from fission are intro
duced at one energy (chosen at some appropriate energy near the peak
of the fission spectrum), then by setting the zero of lethargy to correspond
to this energy,
J()
= ()
(4.163)
in the infinite medium
(4.266) becomes
y(u) p(u) du +
v.:
(4.273)
(4.274)
REACTOR ANALYSIS
154
[CHAP. 4
(4.275)*
>?/Pth
This equation
y*
1/
Sa
2th
-L
/"""'
^Jo
j(u) p(utll;w) du
(4276)
Jo
z.-p /S*
(4-277)
is,
This relation may be used to follow the neutron production cycle at steady
If we begin the cycle by introducing one thermal neutron into the
state.
multiplying medium, then a fraction (SJ,h) p/2? is captured by the fission
able material.
Of these, a fraction S}b/(S^h)f causes fission. Thus the
single thermal neutron produces /2}h/(2h)? fissions.
However, if we
consider that in general the fuel material will also possess some nonzero
fission cross section in the fast range, then for each thermal fission there
will be e thermal fissions plus fast fissions. The total number of fissions
in the system
Each fission produces neutrons
therefore, e/2}h/(2J,h)jr.
Thus qtf fast neutrons appear for each thermal neutron
on the average.
added to the system.
These fast neutrons are eventually slowed down
to thermal energy by the various scattering collisions with nuclei.
How
only
certain fraction survive the entire slowing-down process since
ever,
some neutrons are removed from the system by fast captures.
The
therefore,
fraction reaching thermal
fast
neutrons
are
eventu
p,;
yfep,,,
the system
ally thermalized.
Clearly,
to remain at steady state,
is
if
is
77/ep,h
SLOWING-DOWN PROCESS
SEC. 4.8]
1,
so as to reproduce the
155
source (4.163).
_ jo*
+(>
du
2/0th
fj? ggO
() *u
<r}hti,
These may be
written
M-lH,
For
<4-280)
REACTOR
156
ANALYSIS
[CHAP. 4
is,
tant, is the fact that both 2((u) and 2J,h decrease with decreasing fuel con
Again, this gives an increase in both the fast and thermal
centration.
fluxes; moreover, the thermal flux will in most cases gain relative to the
fast flux since the fuel-absorption cross section will usually be a larger
part of the thermal-absorption cross section than that of the total fast
cross section.
(Note that in the fast range the total cross section is due
The combined effect of these
primarily to the scattering cross section.)
variations is to decrease the ratio (4.279) as N? decreases.
Now consider a moderator-fuel mixture in which the ratio Np/NM is so
low that the fast fissions may be completely neglected, that
in Eq.
(4.275), e~l.
Then
*.^H/Pih
The quantity
(4.281)
is
is
is
0.002 0.003
Fig.
0.004 0.005
Nr/Nu
Infinite-medium
multipli4.33
cation as a function of the atomic ratio
(4.282)
Um, for
is
0.001
0.0047
of fuel to moderator.
jo
exp
05,(u)
is
an inte
(4-283)
2,(u)
is
pth =
it
NFc,
_ UNM*iM)(u)
Further, the scattering cross section of the moderator will often have so
little variation over the fast range that some representative value may be
taken over the entire range; then, Eq. (4.283) may be approximated by
^"STO* P^)du]
(4.284)
SLOWING-DOWN
SEC. 4.8]
PROCESS
The integral
157
resonance
integral:
futh
=
o-(w) du
/
Resonance integral =
[E
jEt),
JO
aa{E)
(4.285)
Cj
Values of this integral have been obtained for various fuel materials and
are given in the literature.1
Equation (4.281) may be written in terms of these results; thus,
=
<
i1
% W\X exp [-
dfe* r^,(u)
du]
*-
(4.286)
t.v,
is
1)
(t
it
as high as
N F/N M
= 0.005.
>
-1
<
0<a
<
-1
i(l+av)
f(i) =0
<
is
Let the frequency function for the cosine of the scattering angle in the center4.1
of-mass system be f(ij), as given below, for the case in which
particle of mass A/i
initially at rest in the laboratory system:
collides with particle of mass Mi which
Ml
M\
2Af,A/s
(A/, + A/,)1
- A/A'
- cos
a In a
a
the re
1,
is
E,
is
As compared to the frequency function for the case of isotropic scattering in the
This situation
called
scat
center-of-mass system, ((?;) has an added term ^ar).
Note that the relation between the neutron energies before and after collision
tering.
component;
unaffected however by the addition of the
respectively,
Eo and
that is,
/A/,
U/,
MtJ
0 (1
1,
1,
J.
r-=-
a)i.o
Th-232,
dE
>
Et
aEo <
j.
= -7=
f)
(';) dE =
is
If
4.2
the probability that
neutron of kinetic energy Eo emerging from a scatter
ing collision with energy in dE about
given by
PROBLEMS
< Eo
< E0
REACTOR ANALYSIS
158
where a = [{A
I) /(A +
1)]* and
[CHAP. 4
It
2,(u>)
(w)
dw.
is the
Find
2u./|2,
the expression
1).
z(E) dE
Jge-E/B. dE
El~
SLOWING-DOWN PROCESS
159
2(E)
(E)E
p(E;Ea)
Generated for wjivans (University of Florida) on 2015-09-22 17:18 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
E,
= 2(0' = constant
= 2< = constant
= 2th = constant
= exp
<
Elh <
E < Et
E < E,
= 2?,h
and
fB,2JE') dE'l
JE t2,(E')E' J
Define
*
i
7i =
2?
f2,(<)
"th"
= 0, 1, or
Find
the general
v
2.46
E0 = 10 ev
Am =
Ei
12
= 10 ev
AF
Elh
= 235
= 0.025 ev
Take
Moderator,
the
Energy range,
ev
10^
c. What percentage
1
1.0
10'
0.025
0.025
10
10.
650
0.5
2.0
550
a.
a.
5.0
5.0
5.0
0.001
0.003
0.010
Note that the integral of the approximation for j(u) is not unity.
5.0
5.0
5.0
average
CHAPTER
ONE-VELOCITY REACTOR
5.1
a.
One-velocity Model.
inhomogeneous
anisotropic
medium
of interest
in
(the angular
transport-theory formulation.
The basis of the following analysis, then, is the one-velocity approxima
This model was introduced in Chap. 3, but our attention there was
tion.
confined to the infinite medium.
In the present treatment we will
1 The calculation of the total
track length <p(v) took into account the contributions
of all neutrons of speed t>irrespective of their direction of motion.
1 The Fermi age model.
160
DIFFUSION THEORY
SEC. 5.1]
161
the nature
methods.
The model we use for the present calculation has the following features :
(1) the system is a homogeneous and isotropic medium of finite extent;
(2) all neutron sources and sinks are uniformly distributed throughout
Thus all neutrons in
the system; and (3) the one-velocity model applies.
the system, including those which appear from sources, have the same
speed
v.
n(r,<),
where
.
.
'
We will also
.
'
i\
flux
4>{r,t)
n(r,0
(5.3)
REACTOR ANALYSIS
1G2
[CHAP. 5
is,
ment in unit time by passage across the boundaries of the element will
That
there
not, in general, be compensated by a corresponding gain.
will be in general net "leakage" (gain or loss) from the volume element.
,
,
,n
.
number of neutrons absorbed
If
cross section
for neutrons of
20
(5.4)
speed
r
I
Jr
and
|i
at
n(r,<)
di
S(r,t) dr
i-i
is
c/
24>(r,0 dr
2{r,t) dr
(5.6)
The function ?(r,<) dr denotes the net losses from the volume element dr
due to transport of neutrons through the boundaries.
lost per unit
net number of neutrons of speed
volume around point per unit time at time due to
passage through boundaries of the volume
v
sa
%(r,t)
(5.7)
is
set of boundary
conditions
and an ana
is
lytical expression for the leakage term 8(r,<)- The remainder of the
devoted to the derivation of useful relations for defining
present section
these quantities.
is
it
DIFFUSION THEORY
SEC. 5.1]
163
of time (i.e.,
;0;
te
that its
surface is normal to the direction of
the axis (see Fig. 5.1). At any instation x and
dA
f*
I-
so oriented
F[Q
Differential arca dA
(x) dA =
in x counted as positive and those moving nega
tively in x counted
g.
as negative
The approximating form for this function is chosen by the analogy of the
In the case of par
over-all neutron movement to a diffusion process.1
ticles wandering through a medium in such a manner that they suffer
many collisions in traveling over distances of interest, and emerge from
each collision with a nearly isotropic distribution in direction, it has been
found that the assumption that the net current be proportional to the
negative gradient of the particle concentration leads to a correct descrip
tion of many aspects of the process. In the one-dimensional neutron
problem under consideration, then, we might describe the behavior by
the assumption
Jx(x)
- Dv
n(x)
(5.9)
-D^(x)
(5.10)
p. 173,
REACTOR ANALYSIS
164
[CHAP. 5
= vn. Note that the use of (5.9) or (5.10) implies that in a region
in which the neutron density is spatially constant there is no net current,
since
i.e.,
Jx(x)
= 0
for n(x)
= n0, a constant
(5.11)
shows that
(5.12)
(5.13)
DIFFUSION THEORY
SEC. 5.1]
2* ~ 2,(1
mo);
165
~
1
> X.
(5.14)
AK>
which tells us that the transport mean free path is at least as large as the
scattering mean free path and increases with increasing values of /i0, the
Since
average cosine of the scattering angle in the laboratory system.
find
that
the
large values of mo correspond to small scattering angles, we
flow of neutrons is aided by forwardness in the scattering process. Thus
one would expect, in very general terms, that low-nuclear-mass materials
would support larger currents for a given neutron flux (density) gradient
[see (4.30)].
5.1
Thermal
Density,
Material
Diffusion Lengths
g/cmJ
Water
Heavy water*
1.00
1.10
and
Diffusion Coefficients"
Diffusion
length L,
cm
2.85
171
0.16% H20
116
1 .85
2.69
1.60
GBF
20.8
29.0
54.4
52.0
AGOT
Diffusion
coefficient
D, cm
Absorption
cross section,
0.16
0.66 barns
0.800
0.883
0.477
0.300
0.93 mb
9 mb
1 .043
4 .4
1.040
4.8 mb
mb
data were obtained from U.S. Atomic Energy Commission, "Reactor Hand
1.5.4, p. 487, Document AECD-3645, March, 1955.
should be noted that L for D20 is a very sensitive function of the H20 content.
These
It
Eq. (5.11).
Net loss of
neutrons in
unit time
from di
Thus,
net current in posi
tive x direction
through face
\_dy dz at x + dx
tive
x direction
through face
(5.15)
dy dz a t x
1 Further discussion of these quantities will be given in Sec. 5.5 in connection with
the diffusion-length concept.
REACTOR ANALYSIS
166
Or, in differential
[chap. 5
notation
dy dz
J +1JNJ,
dx)
T
z{x
T/st.
ay dz
Jz(x)
\-L z
,
(5-16)
a constant]
n D-j2(x)
d2
(5.17)
dx1
Equation (5.16) shows that, in order that there be a net leakage different
from zero, there must exist a spatial variation of the net neutron current.
Fig.
5.2
function of all
three space coordinates; thus, for a Cartesian system, we would define a
In this situation there would be a net
density function n = n(x,y,z).
of
As
movement
neutrons along each of the three coordinate axes.
before, we would compute a net current in the x direction /i(r) ; but now
there will also be, in general, components in the y and z directions.
These
If Jy{t) and
we compute in the same way as we did the x component.
be a
J,(i)
a single vector
function
J(r)
(see
DIFFUSION THEORY
SEC. 5.1]
Thus the
Fig. 5.3).
J(r)
167
J is defined
(5.18)
z directions,
of (5.10); then
J(r)
+
-D
DV(i)
k(r)
(5.19)
In this way
we
Fig.
obtain
5.3
This expression
V-
is the generalization
= -DV24>
(5.20)
dt
*(r,0
= ,S(r,0
- So*(r,0
DV*4>(r,t)
(5.21)*
S(r)
(5.22)*
HEACTOR
168
ANALYSIS
[CHAP. 5
J.(r)
where
^ j+(t)
Therefore we define
x.
- j_(r)
(5.23)
functions j+ and
j-
The
r.
Specifically,
number of neutrons which pass per unit time from
=
left to right (irrespective of the incident angle)
j+(r)
through a unit area normal to x axis at r
It
(5.24)
important to note that every neutron that passes through the unit
area is counted if its velocity vector has a component in the positive
x direction, regardless of the angle which its trajectory makes with this
area. The partial current j-(r) defines the movement toward negative x
in an analogous way.
The two functions
and /_ account for all the
neutrons which pass through unit area perpendicular to the x axis, and
Jx the measure of the net movement toward positive x. We now com
pute the partial current j_(r).
Consider
differential area dA in an infinite medium supporting a
Let the center of dA be the origin of
one-velocity neutron flux.
Car
tesian coordinate system such that the
axis
normal to dA.
Define
also
meas
polar coordinate system such that the co-latitude angle
ured from the positive
axis and the azimuth angle
from the positive
axis (see Fig. 5.4). Then, by definition, j-(0) dA
the number of neu
trons which pass through dA per unit time having made their last scatter
We can com
ing collision in the upper half-space in which x
positive.
pute this number by summing all the contributions from such collisions in
the half-space.
Consider then an elementary volume dx at space point
= r(r,0,^) (see Fig. 5.4).
There are 2<(r) dr scattering collisions per
Of the neutrons leaving dr from these collisions,
unit time in this volume.
directed toward dA.
We will estimate this fraction
certain fraction
on the assumption that the neutrons scattered in dr emerge with an iso
tropic distribution in angle; i.e., for the medium in question elastic
scatterings are isotropic in the laboratory system.
It
difficult to improve upon the isotropic assumption with the
exactly at this point that the
present method of approach, and
In
lost.
information about the angular distribution of the neutrons
(L),
not
in
and
the
neutrons
scattering
isotropic
the
neutron
general,
they impinged upon dr
scattered in di will emerge isotropically only
Thus,
net transport of neutrons in the region
there
isotropically.
surely invalid, and the
of dr, the assumption of isotropic scattering
6
is
\f/
is
is
is
is
is
is
a
if
if
is
is
is
it
is
is
,;'+
is
Suppose,
DIFFUSION THEORY
SEC. 5.1]
169
Not all the neutrons headed for dA actually reach dA, however.
Some
of these neutrons suffer further collisions along the path and are thereby
removed from the stream either by being absorbed or by being scattered
into a new direction which does not pass through dA. We estimate this
number on the assumption that the principal source of removals is
Thus we are
scattering collisions; i.e., we neglect absorption losses.
implying that the medium in question is only slightly absorbing.
We will
show later that this is in fact the only condition under which diffusion
theory yields a good approximation of the spatial distribution of neu
If we assume that scattering collisions are the only mechanism by
trons.
which neutrons can be removed from the group directed toward dA, then
by the concepts developed in Sec. 2.2 we have [cf. Eq. (2.17)]
P({ > r) =
Thus
- P(f < r)
- - e---r)
(1
tr**
(5.25)
170
REACTOR ANALYSIS
If
[chap. 5
we define a current
(5.26)
and dr =
dr dfi
r2
we have1
d\f/,
turz'^.4>{rM) dr dA
j_(0;r,M,^) dt dA =
(5.27)
4bt2
The integral of this quantity over all r yields the contribution to the
current through dA from all neutrons which made their last scattering
collision somewhere in the differential solid angle dp df (see Fig. 5.5).
We define a second function
number of neutrons which pass through
'
(0-'
\b) d ** d\l>
dA
^er
angle dp
In
2g\
scattering
and obtain
j-{0uijt)
dA
d/x
# dA
df
>L*!W^
dAjj tr^rjijf)
dr
(5.29)
half-space.
a general
sm$d$d\l/
It
so detailed
a knowledge
of
d>.
Fig.
(,#)
is sufficiently
expanding
ment
the
5.5
is
region
of
few
4> in
consistent
slightly absorbing.)
,f>(r)
= 0(0)
+ r
V*(0) + |r2V2(0) +
For the present calculation the first two terms will suffice.
1 The
mean-free-path
parameters r,
p,
and
(5.30)
DIFFUSION THEORY
SEC. 5.1]
171
V(0)
r|V0(O)| cos
(5.31)
/3,
of (5.30) is the flux at dA ; the second term is the first derivative correction
for the displacement r. If the angle between the vectors r and V> is
then
designated
Fig. 5.4). The substitution of the first two terms of (5.30) into
(5.29) along with the expression (5.31) yields
fi
(see
dn dyf/Zs
dA
dA
dn dip
|V(0)| cos
*(0) +
4x
cos
re-*\ dr
g"
[(0)c~z-r + |V>(0)|
></;
4-K
(5.32)
2.
is
x)
\p
dA
(5.33)
y.
'
J-K(0-
'j-{0)v,
j*
by
cos
4>y
we denote
and
sin
sin
cos
(\f>v
yp)
= cos
cos
8Z
V (see
V.
cos
(5.34)
M|o2'co80d^]
and
If
We can write
of
^[2x*(O)
8Z
rfM
/3
i_(0;M)
*)
and
(5.35)
dfl
ym
M^cosj,
d,
;_(0;M)
{5M)
Equation (5.36) gives the current through unit area normal to the x axis
(located at the origin) due to all last scatterings in a cone of width dp
about the x axis. This result will be very useful for computing the cur
azimuthally symmetric.
rent in systems in which the neutron density
is
number
passing through dA
Per um* time which
REACTOR ANALYSIS
172
[chap. 5
j_(0;M) d
0(0) +
JT1
j_(0)
|V0(O)| cos
0X
x.
Thus,
(5.37)
Fio.
5.7
involved.
J+(0;r,,*) dA = Z.*(r,M,*)e-
1^
be taken in the
dA
(5.38)
4irrJ
y.
is
to
The contributions from the solid angle dfi d\j/ are
(see Fig. 5.7).
1
(5.39)
dn
0(0) +
|V0(O)|m cos
6.
Im|
j+(0;n) d/i
*(0)
rfM
|V0(O)| cos
t>2.
cos
0,^0(0)
s.
ex
0(0)
|M|
j+(0) =
Finally
is
The computation of the partial current j+{0) may be carried out in the
same way.
The procedure
straightforward; however, some care must
(5.41)
DIFFUSION THEORY
SEC. 5.1]
173
32.
9X
(5.42)
where
If
cos
|V<(r)|
Bz = 0,
and
(x);
<(r)
= 4>'{x), and
moreover, |V0(r)|
/,(r)
Jx(x)
- r
4>'{x)
(5.43)
A comparison of this result with (5.10) shows that the elementary model
used in the calculation of (5.37) and (5.41) yields an expression for the
net current which differs from that obtained by a more elaborate theory
through
the constant of proportionality.
Whereas the transport theory gives for the
diffusion coefficient [see Eqs. (5.12) and
Generated for wjivans (University of Florida) on 2015-09-22 17:18 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
(5.13)]
3(2,
2,/io)
(5.44)
(5.45)
32.
Fiq.
5.8
Gradient of neutron
Thus although diffusion theory is able to pre
flux.
dict the functional form of the net current,
it fails to give a correct estimate of the diffusion coefficient.
However,
the estimate (5.45) is fairly good for media of large mass number and small
absorption cross section; if both 20/2, and 1/ A are much less than unity,
then by (5.44)
D
32.(1
- 2/3A) J_
32.
1
no)
32,(1
(5.46)
In all subsequent work we will use the relation (5.44) for calculating the
diffusion coefficient.
The result (5.42) for the net current can be used to construct the
vector net current at point r.
If 8 and 02 denote the angles which
makes with the y and z axes, respectively (see Fig. 5.8), then
Jx(r)
J(i)
J,(t)
-DjV</)(r)|
-D|V(r)| cosfl,
-D|V(/>(r)i cosfl,
cos
8X
(5.47)
REACTOR ANALYSIS
174
[chap. 5
J(r)
J,(t)i
J(t)j + 7,(r)k
Z>|V^(r)|(i cos
cos 0
+ k cos
If
<(r)
j+(T)
=j*(r)
j_(r)
be
DV(i)
written in terms
use
- ^|V*(r)|cos*.
(5.48)
ex
0(x), we have
j+(x) =
j_(s)
It
0Z)
as is to be expected.
of
- *'(*)
i *(x)
(5.49)
*'(x)
point, i.e., if
Generated for wjivans (University of Florida) on 2015-09-22 17:18 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
+ j
6Z
The
(i)\
j_(r)
if
(5.50)
i4>(i)
Sec. 3.5.
For the remainder of this chapter we will use the following set of equa
tions in discussing the one-velocity approximation:
-DV*(t)
n
J(r)
+ 2,(r)
=
3Str
3(2t
(a)
(6)
- 22./3A)
1
S(r)
-Z)V(r)
a set of
M
W
(5.51)*
boundary conditions
A complete state
General Statement of Boundary Conditions.
in the application
which
to
be
used
are
ment of the boundary conditions
e.
of the one-velocity
have a more detailed description of the neutron flux than presently avail
It is possible nevertheless to establish an appropriate set of
able.
A detailed description of the boundary conditions to be used in the one-velocity
model involves the methods and results of transport theory (see Chap. 7).
1
DIFFUSION THEORY
SEC. 5.1]
175
(5.52)
distributions
(2)
(5.53)
interface
(3) At
(5.54)
These conditions are perfectly general and must be met, at least in approx
imate form, by whatever function is specified for the neutron-density dis
tribution. The neutron flux defined by the elementary diffusion theory
cannot meet all these conditions.
The function <(r) is much too coarse a
description of the neutron population in that it gives the track length per
unit volume per unit time of all neutrons of a single speed.
Conditions
such as (2) above require, however, a much more detailed statement, and
be replaced by less exacting requirements.
Before inquiring
the
into
nature of these approximations, let us first examine the implica
tions of the three conditions as stated.
so must
Consider condition (1) : it is clear from this requirement that all regions
of a medium which are in the vicinity of an elementary source (such as a
point source) are to be excluded.
Equation (5.51a) does not apply then
in the neighborhood of a singularity; everywhere else, however, the flux
of course true only if the outer boundary of the system is a simply con
nected and nonreentrant surface.
REACTOR ANALYSIS
176
[CHAP. 5
ticular direction.
exact boundary
(5.51) are:
conditions
which
with Eqs.
(3)
At
(5.55)
(5.56)
(5.57)
J'f(0) =i?.+tf>(0)
Jll)(0)
= *<7o+jL2,(0)
('XM>
If we add these equations and introduce the expressions (5.49) for the
partial currents,
we obtain
J*(0)
- Ji(0)
= go
(5.59)
Thus if the source is removed (g0 = 0), the net current through the inter
face must be a continuous function, and it is discontinuous only in the
presence of a neutron source go ^ 0 (see Fig. 5.9). The subtraction of
Eqs. (5.58) yields
*(0) = 0,(0)
(5.60)
Therefore,
DIFFUSION THEORY
SEC. 5.1]
177
It
serves
x
0
Flo.
5.9
Flux and
net
current around
Vacuum interface,
Flo.
5.10
<(r)
interface.
been suggested,1
geometries.
accurate and easily applied relation.
Boundary condition (3) is the approximation to the physical require
ment (5.54) that no neutrons return to the system from the vacuum along
This requirement cannot be rigorously satisfied in the
any direction.
present case in which we wish to apply the diffusion approximations
Therefore we impose the less stringent condition that the partial
(5.51).
current from the vacuum be identically zero.
Hence
;-(0) =
(5.62)
where the origin of the coordinate system is placed at the physical bound
This condition may be written in
ary of the medium (see Fig. 5.10).
1 The
Sec. 8.3).
Serber condition
is frequently
used in spherically
symmetric systems
(see
178
REACTOR ANALYSIS
m~~2D
[CHAP. 5
We have by (5.49)
(5-63)*
Equation (5.62) gives an acceptable boundary condition at a mediuminterface for use with the diffusion equations.
However, it
should be realized that this relation places a condition on a function which
was developed on the basis of the diffusion approximation. It will be
recalled that, in the derivation of Eq. (5.37) for the partial current, it was
assumed that the flux was a slowly varying function at the point in ques
tion and therefore could be represented by two terms of a Taylor series.
This requirement is hardly satisfied at any interface, least likely at a
In such regions higher-order terms in the series would
vacuum interface.
be required, and the use of only the first two terms, i.e., the diffusion
approximation, is at best a first estimate.
While Eq. (5.63) is quite easily applied in most problems, it has been
shown that there is an even simpler condition which affords a very good
This approximating condi
approximation in a great many useful cases.
be
on
the
basis
of
the
developed
following
tion may
observation.
Within
the interior of a medium supporting a neutron gas, for example, a reactor,
the neutron density (flux) is relatively high, since in general the neutron
sources (fissionable material) are located in this region.
Neutrons pro
duced by these sources are eventually lost by capture collisions or escapes
Thus a close inspection of the neutron population in
from the system.
the medium would reveal that there is a continuous net movement of
Moreover, this
neutrons toward the outer boundaries of the system.
net outward flow becomes more pronounced as the physical boundary
In the limit, at the outer surface, all
(vacuum interface) is approached.
the neutron velocity vectors will be directed out into the surrounding
Thus at each station, progressing outward from the interior,
vacuum.
vacuum
Thus
the system are supplied only by the neutron sources in the center.
the neutron density falls off rapidly as the surface of the system is
approached; however, it does not vanish entirely inasmuch as there is
always a finite number of neutrons passing through
the system.
It would appear then that one could establish a reasonable approxima
tion to the vacuum interface condition by requiring that the function
representing the neutron density vanish at some appropriate point near
THEORY
DIFFUSION'
SEC. 5.1]
179
surface such that the resulting statement of the boundary condition best
Let us call this distance e, and
approximates the exact requirement.
suppose that in a typical problem the general solution to the diffusion
is known and that one of the arbitrary constants has been
eliminated by an appropriate boundary condition at some interior point.
equation
constant.
If
the solution
to eliminate the
above is 0(x), then we
be applied
mentioned
(5.64)
where for convenience we have placed the origin at the surface (see Fig.
If the approximating form of the flux (x), which is now deter
5.10).
mined everywhere in the medium, is to be a good estimate of the actual
flux, then
at x = 0.
it
is necessary to satisfy
condition (5.63)
But since
as closely as possible
is sufficiently
(5.65)
-7
0'(O) =
0(0)
(5-66)
2D
$\u
e,
under the
(5.67)
from (5.12).
We may summarize
as
may
be replaced
(5.68)
by the condition
0(zo + |Xtr)
=0
or
0(xo
= 0
(5.69)
It
is important to note that (5.69) and (5.68) are in fact different condi
tions and, if applied separately to the same system, will yield different
solutions for the flux 0. Although these solutions will differ only slightly
An illustration
REACTOR ANALYSIS
180
[chap. 5
For many
xo
(5.70)
The estimate obtained above for e (5.67) is based on the diffusion theory.
Other estimates can be obtained with the application of more accurate
A somewhat better number than
models, such as transport theory.1
(5.67) is
= 0.7104X,,
(5.71)*
Further, Eqs. (5.67) and (5.71) are each specifically calculated for planar
boundaries; but they are approximately correct for boundaries of small
For boundaries of increasingly greater curvature, e monocurvature.
tonically approaches a limiting value of ^X(r.
We conclude the present section with a summary of the diffusion-theory
boundary conditions which we will generally apply throughout the
remainder of this work. These conditions are the statements (5.55),
(5.61),
and
(5.69):
5.2
Plane Source.
1 S.
Frankel and E. Nelson, "Methods of Treatment
Equations," AECD-3497, September, 1953.
of Displacement
Integral
DIFFUSION THEORY
8EC. 5.2]
181
The
sidering infinite media which contain various elementary sources.
results obtained in these calculations are applied later to neutron-diffu
sion problems which involve more complex source distributions and to
Initially, for the sake of simplicity, we con
systems of finite extent.
sider only steady-state problems and systems in which the spatial dis
tribution of the flux can be expressed in terms of a single space coordinate.
Thus, we begin with the diffusion of neutrons from isotropic plane, point,
and line sources.
In the
medium.
is,
If
of
and z.
L2 ^
(x)
be
= Ae*'L
(5.74)
by
(5.75)*
4>(x)
4>"(x)
0,
is x
The concept of
plane distribution of isotropic
sources reduces the problem to finding the spatial dependence of the flux
in terms of a single coordinate, in the time-independent
case.
denotes the coordinate normal to the infinite-source
plane (which
>
= S(x,y,z)
= -D^(x,y,z)
+ ?a{x,y,z)
J(x,y,z)
-DV*4>(x,y,z)
Be-'11-
(5.76)
REACTOR ANALYSIS
182
[CHAP. 5
We have, then,
(x)
= Be~zlL
to x ^ 0 is
is applicable
J(x)
(5.77)
%q0
yields,
\x\
as
(5.74)
(5.516)
We compute
4>(xA
*(*)
to
(5.78)*
2D
If
Source plane-
Fig.
Flux distribution
5.11
to plane source.
due
24>
sin2
case,
is
4>
(5.79)
and ^;
fir2
[r*(r)]
(5.80)
F"(r)
*(r)
=0
1<P
function F(r) =
F(r)
>
(r),
. 4>\
ae{sm6de)
sin
7^
so
r2
>
0.
for
dr2
r2
is
is
sion
is
The solution
Lq0/2D.
therefore,
r(r)
(5.80)
and obtain from
(5.81)
DIFFUSION THEORY
183
is,
sec. 5.2]
there
(Ae"L
<p(r) =
fore,
Be-"L)
(5.82)
so as to satisfy
required that
lim 4irrV(r)
In the limit
-D*'(r)
J(r)
(1
is
J(r)
(5.83)
q0
r-*0
where
>
is
If
is
is
it
boundary condition
determined by the condition on the source
The constant B
(5.72a).
strength.
go denotes the rate of neutron emission from the origin
(neutrons per unit time), then the total net outflow of neutrons from the
source
given by the limit of the product of the net current in the radial
direction and the area of sphere as the radius 0. Thus
As before,
r/L)e-"'
qa.
Hr)
(5.84).
q-~
is
There
qo=
For the chosen flux
(r)
dr =
Jq"
is
loss of neutrons
= Be~TlL/r, we
20< dr (5.85)
obtain
g0
dr
(5.85)
= 4irSoL2B = <brDB,
is
c. Line Source.
third form of the one-dimensional problem of
neutron-flux distribution may be derived by considering
line distribu
For
of
convenience
we
tion
take the diffusion equation
isotropic sources.
(5.51a) in cylindrical coordinates and define <(r) = (p,<p,z).
pdp\pd;)
is,
0,
?W+M-L*-0
(5-86)
For an infinite line of uniform sources and axial symmetry, (5.86) reduces
to
REACTOR ANALYSIS
184
[chap. 5
where p denotes the radial coordinate measured from the line of sources
If we define a new variable x = p/L, this equation takes
at the origin.
the standard form of the modified Bessel equation1
4>"(x)
- <p'(x)
- (x)
= 0
(5.88)
is
4>(x)
AI0(x) + BK0(x)
(5.89)
where Iq(x) and K0(x) are the modified Bessel functions of order zero of
Complete tabulations of these
the first and second kind, respectively.
For convenience we
functions are available in the standard sources.2
Aw
Fig.
12
5.12
K0(x)
T~
Q\
|
3
give below the relation between these functions and the regular functions
and Y of the first and second kind.
In general, for the nth-order
functions,
7,(0
iJ.(U)
Kn(t)
\ i+Vn(ti)
iYnffl]
- \ *+lH<M
where H{u{it) is the Hankel function of the first kind and t is real.
The application of the boundary condition (5.72a) shows that the arbi
trary constant A must be taken identically zero, since /o(<) becomes
arbitrarily large as / * oo (see Fig. 5.12). The value of B is obtained by
imposing the source condition [cf. Eq. (5.83)].
lim 2icpJ{p) =
p-0
g0
(5.91)
DIFFUSION THEORY
SEC. 5.2]
185
It
= -D4>'{P)
J(P)
*(p) =
(5.92)*
2*D
Kernel
it is possible to de
termine the steady-state flux from
any distribution of sources in an in
As an introduction
finite medium.
to the technique for calculating the
flux for more general source distri
butions,
mined
if the source
*.(r)
If the
point
9, alone
-\t-tl\/L
4x/)|r
(5.93)
ti\
is
if
interactions
it
assumed that
the presence of a point source does not affect the diffusion properties of
the medium in its immediate neighborhood, the behavior of the neutron
flux from either source will be unaffected by the presence of the other.
neutron-neutron
is
these results
(r)
,(r)
4>2(r)
(5.94)
REACTOR ANALYSIS
186
[chap. 5
, i,
Formally, this result follows from the fact that the diffusion equation, of
which
and #2 are solutions, is linear; therefore, the solutions may
be superimposed.
Since the above line of reasoning for two point sources would not be
altered for three, four, or any number of point sources, it is concluded
that, if N point sources located at points r,(i = 1, 2, 3,
N) and of
...
...
strengths qt(i = 1, 2, 3,
N) are present in the infinite medium, the
total flux at a field point r due to all N would be given by
if
4*D\r
is
if
then we define
r'|
r',
due
(5.96)
r'
can compute the total track length per unit volume per unit time at
due to all sources in the medium by integrating (5.96) over all space.
is
is
be constructed
ing (5.97).
In order to demonstrate the application of this general method to neu
tron diffusion in infinite media, the plane-source and line-source dis
tribution problems (Sees. 5.2a and 5.2c) will be treated using (5.97).
a
Plane-source Distribution.
The plane source of neutrons may be
uniform distribution of point sources over an infinite plane.
regarded as
Consider then the case in which isotropic point sources are distributed
Part
chap.
7.
I,
0,
*(r'f dl
ii\
eralized to
continuous
number of neutrons produced
about space point
4*>|r
gr^-lr- r.l/L
sec.
DIFFUSION THEORY
5.2]
point x may
on the plane
be obtained
(see
187
Fig. 5.14).
Fig.
5.14
Plane-source distribution.
(x;s) dA
dA
e~,IL
q<>
= 0.
dA
J>(x;) dA
n-
e-'ILp dp
dO
we write
reduces to
qo_
f"
*(*)
sJ
(5.98)
AwDs
is
_
=
If
ID
2D jM
JM
e~'IL ds
Lqoe~ \z\IL
2D
if
is
S(r) = A5(x)
(5.99)
S(x) dx
where 8(x)
is
is,
area dA whose polar coordinates are (p,0) in the x = 0 plane and which
is a distance s from the field point
from (5.84),
dy
dz =
q0
REACTOR ANALYSIS
188
= g0.
...
4^5
&(x)
g0
\r=T\
r',
[z2
^DJJ
[x*
(tf
7/')2
y'Y +
{z
- xT +(y-
[(x
exp
-z')>]i
dx'dy'dz'
-z'Y
- z')*]l\
- 2'p]T
(f
r'|
|r
f
]
go
*(r)
where
(Z
and
[CHAP. 5
dy' dz'
S(r)
5(x) 8(y)
(5.100)
if
is a
go
exp
A
p'
sin 8'Y~+Jz
sin 0')2
l-*.[p>+(z-z')*)^dz'
zyv
[p2
+(z- z')2]*
p'
sin
sin
8
(p
("
8'Y +
(p
p'
cos
cos 0')2
(2
= g0, yields
*(p') dp'
[(p cos
dz'
p'
Up cos
6
exp
{
fu
f"
*(pA*)
jlm
Jo
lirp
Jq
J*dz
S(y) dy
Jo
6(x) dx
S(t) dr
J'
go
line, then
Jo
is
*')2]}
d8'
DIFFUSION THEORY
-* (p)
If
be recognized as the
exp
cosh
"
t)
zero:1
H\,"(z)
we
eft
However,
by (5.90),
ffi(ix)
(p)
?-
order
*)
(p,e,z)
= (z
189
(_
we define sinh
If
/_
SEC. 5.3]
K0(x)
is
which
a.
q0K0(p/L)
2*D
5.3
sec
5.
is
This calculation
geometry due to a plane source placed on one face.
performed for two different outside boundary conditions on the sourcefree face of the slab, namely, condition (5.57a), which requires that the
at an outside face of
system.
homogeneous slab of width a in the x direction
a
Consider,
conditions
a
to the boundary
Therefore
Complex Variable,"
and
REACTOR ANALYSIS
190
infinite in the
[chap. 5
At the plane
uniform distri
5.15
width a.
Infinite
slab
*"(*)
The general solution may
4>{x)
= AieZIL
be
- 1 *(x)
= 0
written
Ble~xlL
or
4>(x)
A j sinh
y + Bi
cosh j-
(2)
lim j+(x)
x->0
*?
(5.101)
We use for the partial currents the expressions (5.49). It is easily shown
that the application of these conditions to the second of the general solu
tions given above yields the flux i(x):
a. r~\
*<*)
~
x)i
- ^ rshm *( ~ j) + 2Dx csh x(q
[~(2^ + l)V- {2Dx- \ye- \
c=
u\o\
(5-102)
the symbol
x =
(5.103)*
0(a) = 0
(2) lim
T 0
(5.104)
sec.
DIFFUSION THEORY
5.3]
191
is,
(5105)
sinh*(a-z)
the results
t>i
and
comparing
is
if
We can obtain some insight into the physical requirements which must
the diffusion theory
to be valid in
given situation by
be satisfied
obtained
above.
Specifically,
let us
is
1.
compare the expressions (5.102) and (5.105) in the case when 2Dx
and 2D /a <.
It readily shown that, under this restriction, both of
*,(*)
sinh x(d
...
(5-106)
1.
Thus the solutions of the present problem based on the accurate and
approximate statements of the vacuum-interface condition are in good
Let us examine now
agreement provided that 2Dx <3C and 2D/a
the consequences of these requirements.
We have from (5.12) and (5.75)
(5.107)
3Sa
2D
(2)
(5.108)
1
(1) 2Dx
Therefore
if
is
is
is
2q0
x)
REACTOR ANALYSIS
192
approximate
are equivalent
[CHAP. 0
The approximate
1.0
0.8
.-J
\V
.
V
0.6
E
(/>0.4
D=l
i,
cm
20 = 0.01 cm"1
1-
ADx
0.2
si nh >a
20
10
30
a, cm
Fio.
Transmittance
5.16
of slab.
Let
For
case 2, the results of the above analysis apply with a source strength
to replace the factor q0/2 which appears in (5.106) [see (2) in (5.101)].
If the number of neutrons transmitted is counted as those received by a
noncollimated counter on the far side of the slab, then for case 2 this is
n0
precisely the net current from the back face. By assuming that the
extrapolation distance is small in comparison with the slab width [which
is implied in the inequality (2) of (5.108)], the net current calculated at
the extrapolated back face will very closely approximate the actual
Then, in the two cases, the transmittance T is computed as
current.
follows
1 :
r,
Case 2:
Tt
=.
Case
Woe
-s.
no
e~z'
-D'(n) _
ADx
sinh xa
ADx
sinh xa
sec.
DIFFUSION THEORY
5.3]
193
Differential equation:
DV2<(r)
+ 20*(r)
= 0
(5.109)
Boundary conditions:
(1)
= 0 everywhere
medium
(2)
at that]
point
Table
(5.110)
a few
systems.
separation
of variables.
REACTOR ANALYSIS
194
[chap. 5
its
use
Table
5.2
is also
Solutions
Geometry
Slab
Description of
system
Source
strength
Cylinder
Slab with planar sourceon the midplane Infinitely long cylinder with an
axially placed source line
Radius of sphere is Ao
Source point
Explicit
(1) *(3o)
*(z)
statement of j
boundary
= 0
- *(-x)
conditions
(2)
General solu
lim
[-2V(*)1
Or) A sinb
tion to
lim (-2rpD0'(p)]
p0
- + B cosh -
*(p)
"
- go
lim [-iwr*D4t'(r)]
-.0
Ah(P/L) + BK9(fi/L),
*<0
rL
- 90
4sinh--f
Boouh-
L]
differential
Slab:
Cylinder:
*()
Sphere:
*(r)
/So 1
2D cosh oo/L
2<rfl/o(j/i)
[/,(/) K,(p/L)
'1
rirD
,mh{-r)
tinh He/1.
x >0
- K,Q,/L)
Mp/L)\
sinh I
(*?)
case.
DIFFU8ION THEORY
SEC. 5.3]
195
is,
...
region
return
current from
.
. ,
incident currentL into the region
,_
(5.111)
Albedo =
is
it
especially useful for treating very thin regions, such as plates, foils, or
Thus, for example, one might treat containers and pressure
shells.
vessels in
reactor.
knowledge of the albedo of these components
theories for other major portions of
concept
extremely effective for
is
tions.
Consider,
is
(5.112)
[see (5.49)]
j+(0) =
= Cc~ltL
*(0)
"thin"
*'(0)
9,
(x)
If
of strength
if
it
is
is
a
(5.113)
ANALYSIS
REACTOR
196
[chap.
*>
The partial neutron current
iw
is,
;_(0)=^(0) + ^'(0)
Albedo
(5.115)
on
(5.116)
This form for the albedo applies only to the semi-infinite medium; for
other configurations, the albedo may be computed within the limits of
denotes the position
diffusion theory from the general relation (where
of the boundary)
It
is
should
The essential
involved in computing
parameters
the
albedo
[see
(5.117)] are the diffusion coefficient in the medium and the ratio of the
This
slope of the neutron flux to the value of the flux at the boundary.
ratio will depend upon the geometry, the diffusion coefficient, and the
of the material.
the semi-infinite
medium we
2D
2A
3
absorption
case of
fh\
\D)
_~
01
It
- 2Xlr(Sa/>)V3
(5-U8)
+ 2Xtr(V>)V3
is
0)
the albedo
unity. It
evident that for total reflection (2 =
should be noted, however, that the result for heavy absorption (Sa large)
cannot be obtained from the above relation because this result was
is
||^|g
of the semi-infinite
boundary is
sec.
DIFFUSION THEORY
5.3]
197
S.,cm_l
Fig.
5.17
Albedo
for
Interface
semi-infinite
Fig.
5.18
is
is
/S.
is
/3
it
is,
/3.
is
is
(n
p2
p3
0)
0) a)
(1
po =
p,
medium.
REACTOR ANALYSIS
198
In
- a)(o0)
o(l - 0)(a/3)
general,
p(* = 2n) = (1
p,(v = 2n
1) =
[CHAP. 5
n =
0, 1, 2,
(5.119)
These relations may be checked by computing the sum of the joint proba
bilities p for all v. The sum should be unity; thus,
00
n-0
P. = (1
00
(a/3)"
<0
+ a(l - 0)
n=0
00
(a/3)" = (1
n-0
DO
a/9)
(a/3)"
n=0
(5.120)
We compute the average number of crossings n from the relation
= 2(1
n-0
a/3)
n(a/8)" +
a(l
- 0)
n=0
n =0
oo
(c0)"
n=0
(5.121)
r-^
a
(5.122)
/3,
n =
5.4
Application
as
?iL+4)
ap
1
n =
is
a.
cylinder.
(d
is
and
is
The first series is easily identified as a|8(l a/S)-J and the second
the extrapolated
DIFFUSION THEORY
sec. 5.4]
199
These specifications
display the motivation for selecting the particular form of the timeindependent solution in the cylinder and sphere problems.
This model in which the neutron source is taken to be vJ,f4> (which
implies that neutrons from fission appear at the one velocity of diffusion)
is not expected to apply directly to operating reactors; however, the
techniques to be used later are well illustrated by this formulation of the
problem, and the results are useful, if properly adapted, to more general
situations. We take then as our neutron-balance relation the timedependent diffusion equation (5.21) along with the assumed source term
S(r,0 = v2/#(r,0
The general one-velocity
diffusion
is then
-DW+(i,t) +
(5.123)
- \ J (i,t)
Z.*(r,<) =
4>(x,t)
+ BH(x,t)
=^jt *(*,)
2a
o(x)
(5.125)
(5.126)*
in which
The appro
the definition
B* m
(1)
(5.124)
4>(a,t)
= 0
is some specified
(2)
(x,0)
[see (5.72)]
0o(i)
(5.127)
initial distribution.
(5.128)
REACTOR ANALYSIS
200
[CHAP. 5
*^M +
+"B>
F(x)
_c.
0
DvG(t)
(5 129)
^AZ[)>
F"(x) +
G'(t)
(B2
=
c2)
= 0
F(x)
(5.130)
-Dvc*G(t)
(5.131)
and if we call
02
= B* +
Go ^ G(0)
(5.132)
c1
(5.133)
F(x)
cos fix
+ C sin fix
(5.134)
F(a)
F(-a)
A
A
cos fid
cos fia
+ C sin fia
C sin fia
(5.135)
yield the
2C sin fia = 0
(5.136)
| (2n
fid =
t(ti +
+
1)
1)
= ana
7a
where a =
^fl
(2n +
where 7 = x (n +
1)
(5.137)
1)
= An cos anx
+ C sin
ynx
Equations
in Physics,"
(5.138)
Moreover,
we
chap. V, Academic
DIFFUSION THEORY
SEC. 5.4]
201
X2
when
f)
B2
a2
B2 = w'
when
/3
7s,
c2
then
it
/3s
note from the definition of /3 [see (5.133)] that for each value of n there
takes on values an
takes on values yn
0,
cos ax
Go(Ane-""
(?(<)
is
Fn{x)
C*-H
1,
sin ynx)
(5.140)
4>{x,t)
Fn(x) Gn(t)
(5.141)
n=0
0(x)
= Go
if
is
is
4>(x,0)
(5.142)
71=0
f*
J-s
of the trigonometric
cos anx dx
(5.143)
*o(x) sin yx dx
problem
Given
and any initial dis
now complete.
^rr
alio
4>o(x)
is
j0^3
/
Cn =
An =
/3
if
is
is
The analysis leading up to the result (5.140) has been concerned only
with boundary conditions (1) of Eq. (5.127). The requirement
that,
this condition
to be satisfied by the spatially dependent portion of
(x,t), then may take on only the values indicated in (5.137). This
in turn places a constraint on the constant
because of the relationship
limited to the values (5.139). Given, then,
(5.133); consequently,
Note, how
value of n, o, yn and in turn X and oi may be computed.
ever, that the evaluation of X and un involves the parameter B2 which
determined by the nuclear properties of the system.
(We will consider
this matter in detail shortly.) The remaining undefined quantities in
(5.140) are the coefficients A and C. These are determined from the
application of the initial condition (2) of (5.127). Thus
we compute
(x,0) from (5.141) using (5.140), we obtain
REACTOR ANALYSIS
202
[CHAP.
| 1
n=0
An cos anx
(5.144)
oo
(x,l)
4>(x,t)~Go
>
>
(4n
3)
jY
[^i)]-_[I(?i+i)]'
(5.145)
Ae-fl'x""
cos anx
(5.146)
n-0
is
is
is
is
is
B> m
X* m a*
(2/
It
e-B<"
y"X
S1"
...x..7,_.
cos anx
Y Ane-D^"
= Go
*(z,Q
C.%
tribution for all subsequent time from the general solution (5.141). The
coefficients involved in this expression are to be computed by (5.143),
and the various eigenvalues, from the relations (5.137) and (5.139).
Let us examine now the distribution of the neutron population in the
present system for large values of t. We inquire, then, into the behavior
of the flux (x,t), given the nuclear properties of the system and a knowl
Therefore, assume that
edge of the flux at some time instant long past.
=
0; then, for all subsequent
the flux is known at some arbitrary time t
time (during which no changes are introduced into the system), the flux
is given by
- S.)
(5.147)
sec.
DIFFUSION THEORY
5.4]
203
such that a particular member of the set X vanishes; i.e., let us take a
such that XJ, = 0, where m ^ 0. Then by (5.147)
P^f-iW-W
All other
in (5.147)]
. p*Li>]'
It
follows that if
(5.H8,
(*)
[(2 +
. _
(2m
(5.149)
m > n
m = n
m < n
X* > 0
<0
= 0
(5.150)
Thus for
33
(fi)2
("S/
1,
(5-151)
So)
cos
Ao =
sr.
ccox
= G0Ao cos
(5.153)
Mx)
cos
(i)
/-a
{x,t) ~(7oi4o
dx
(5-154)
a
if
=
multiplying slab
We conclude, therefore, that
at time
we start
with neutron flux of the form o(x) and choose the composition and the
"Fourier
is
REACTOR ANALYSIS
204
[CHAP. 5
(0 -
*w
then after some time, the flux distribution in the slab will approach
steady-state condition given by
ast
>
cos
(2a)
)
(x)
<(x,0
00
(5.156)*
>o(x)
cos
^r-1
(5.157)
0
is
if
is
is
is
it
a0cos(^)
At
the composition
is
(5.158)
<o,
a{x)
>
(0
<5159
0.
the slab
is
is
if
0 is
0,
Ar
>
0,
|NCLUPCf7i
0,
if
it
is
is
= AdG0 and
given by the initial conditions.
Furthermore,
no other choice than (5.155) will lead to steady-state operation [see the
discussion following (5.150)].
It true that
were possible to start
where
sec.
DIFFUSION THEORY
5.4]
205
= aoc-fl'x" cos
4>{x,t)
This function
(2d)
given by
(5.160)
t.
is
The initial
shown in Fig. 5.19 for various values of
magnitude of the flux was characterized by the amplitude Cto; thus to
achieve any other higher flux level we
(x,t)>
need wait only until the desired in
achieved by the factor exp
crease
reached at
this level
( Dv\lt).
~^\2
=
suitable readjustment in the
1
is
t',
If
is
o<*i
1%
Fio.
5.19
Flux distribution
in in
finite-slab reactor.
condition such
multi
is
plying system by external means and thereby exercise control over the
the basis of all reactor control methods (see
neutron population. This
Sec. 1.3b).
is
2,
it
l)wx
2d
0,
(2n
1,
cos
is
is
a a
a a
Conditions of the
the desired level.
and
types (5.155)
(5.159) can be de
rived for other reactor geometries.
REACTOR ANALYSIS
206
[chap. 5
>)
will predominate
in the series solutions for both the sphere problem and the cylinder
Moreover, if it is required that either of these systems oper
problem.
ate at steady state, there exists only one term from each of these series
as
one term
slab problem.
with
Eq. (5.124)]
V20(r) + fiJ(/>(r) = 0
(5.161)*
B2 defined
differential
[cf.
-Mr)]"
r
+ B^(r)
= 0
(5.162)
(r) =-{A
r
sin
Br + C
cos
Br)
(5.163)
This solution must satisfy the boundary conditions (5.72a) and (5.72c).
If the outer boundary of the sphere is given by r = R, these boundary
conditions take the form
(1)
(2)
(r)
JU)=0
._
....
(5-164)
The condition that the flux remain finite for all r, appropriate here since
there are no localized sources, yields the requirement that C = 0. The
values of B needed to meet condition (2) are
~
R
n = 0,
1,
(5.165)
Note that we have retained the symbol B to denote the various eigen
As indicated
values, instead of the notation a used in the slab problem.
in Eqs. (5.137) and (5.165), the eigenvalues form an infinite set; however,
only the first of these, corresponding to the fundamental mode, is required
It is understood, then,
since this mode gives the steady-state solution.
that the B which appears in the argument of
refers to the lowest eigen
value.
Since the quantity B was originally defined by (5.126), this
DIFFUSION THEORY
sec. 5.4]
207
j j
sin
(5.166)*
Note that as in the case of the slab problem there still remains an arbi
trary constant A which can be determined only from the initial conditions
of the system.
The solution for the cylindrically symmetric system is obtained from
(5.161); the appropriate form for the present application is [see Eq.
(5.86)]
- [p*'(p)]'
p
+ B2*(p)
= 0
(5.167)
AJ0(pB) + CFo(pB)
(5.168)
(ft
The Bessel function of the second kind F0(pS) has a singularity at the
origin; therefore, the constant C must be taken to be zero so as to satisfy
the requirement that
remain finite.
The boundary condition at
requires, in addition, that
=
(5.169)
JoCM)
The Bessel function J0 has an infinite
As in
Let its
*4
is
(R
See
Hildebrand,
where
Vl = 2.405,
etc.
(5.170)
op. ext., p. 126, for a sketch of this function and p. 166 for
zeroes.
1
0(r)
[chap.
REACTOR ANALYSIS
208
A7,
(p)
(5.171)*
It
is of interest to compare the shape of the neutron flux for the three
The normalized solutions (5.156),
elementary geometries considered.
(5.166), and (5.171) are shown in Fig. 5.20.
0.4
0.2
Fig.
0.6
a\
0.8
1.0
5.20
In each of the
Criticality Condition and Multiplication Constant.
three geometries for single multiplying media discussed above, it was
found that a certain relationship was required between the composition
of the medium and its dimensions if the system was to operate as a stable
This relation was expressed in the form
chain reactor.
c.
(2,
- 2)
5'
(5.126)
and
B2 =
\2&)
*'=(!)'
(5.172)*
sec. 5.4]
DIFFUSION THEORY
209
For each geometry, Eq. (5.126) and one of Eqs. (5.172) must hold simul
taneously if steady-state operation is to be achieved.
Note that by
imposing these requirements on each system the composition or size is
specified; thus, if the size of the system is selected, then
the corresponding value of B2, and the substitution of this
yields
(5.172)
number into (5.126) yields an expression for the composition (note that
completely
(1)
X?
= 0
(2)
Xg
> 0
toward
(3)
2) ("S/
~
"
]/v^*-A
Sa)
(5-151)
< 0
The situations
as an example)
exponential
Xg
Geo
namely,
Generated for wjivans (University of Florida) on 2015-09-22 17:19 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
(0
-D"(x)
+ 2*(x)
v2,(x)
(5.173)
(x)
cos
Bx
(5.174)
Sc
REACTOR ANALYSIS
210
[chap. 5
DB*(x) +
20(x)
r2f4>(x)
Integration of this equation over the entire volume of the reactor gives
(in the case of the infinite slab or cylinder the integration is carried out for
a unit height of the system)
DB2* + 2a*
= y2/I>
(5.175)
dx
(5.176)
$ =
4,{x)
be stated
(Total leakage
+ (total absorption
losses)
verbally
as
losses)
The left side of (5.175) is the total number of neutrons lost from the reac
tor (per unit time), and these give rise to y2/<p neutrons on the right side;
thus the multiplication constant for this case is
ZXB2*
~~
DB1 + 2,
(5.177)
of the
-Z)V^(r) +
20(r)
= v2,4>(f)
(5.178)
This
where for the steady-state case vc = v, the experimental value.
equation, then, describes a fictitious steady-state system which could be
achieved by using the fuel concentration of the nonstationary system
(implied in the macroscopic cross sections) if the neutron yield per fission
Clearly, if the flux in this system
were vc (in general, different from v).
is to be time independent, then the solution to (5.178) must satisfy a
relation of the type (5.177); but, the solution to (5.178) is of the form
(5.174); we have, therefore, the immediate result that
(5.179)
DIFFUSION THEORY
sec. 5.4]
211
but, only
v2/4>
Multiplication constant =
is,
of neutrons
(5.180)
is
A;
reactor are known (i.e., all the cross sections, including the fuel, have
been specified), and the geometric configuration and size of the system
have been selected.
The calculation of requires first the evaluation of
B2 from the appropriate relation in the set (5.172).
The substitution of
ve.
The
is
this result, along with the cross sections, into (5.179) yields
multiplication constant
then computed from (5.180).
It
is
*-(f)0+xr
km
is
+ LJ2)-1
(5.183)*
is
is
Again note that B2 obtained from (5. 172). The multiplication constant
customarily called the "effective multiplication
for the finite medium
the infinite-medium multiplication constant kK
constant," since
multiplied by
reduction factor
the one+ L2B2)~l; this factor
velocity nonleakage probability and will be discussed in detail later. We
summarize the above results by combining the information from (5.151)
(1
is
is
it
A2, X2,
>
is is is
corresponds to
0, 0, 0,
\\
corresponds to
<
<
>
(3)
(2)
corresponds to
(1)
and (5.183):
A;
critical
subcritical
supercritical
d.
is
REACTOR ANALYSIS
212
[chap. 5
the critical fuel concentration and total fuel mass for this reactor as a
For convenience in computation we will assume that
function of its size.
the moderator density Nm in the system is unaffected by changes in the
NF;
fuel concentration
good
Fig.
40
20
80
60
Radius
5.21
100
it cm
120
140
one-velocity model.
g/>]
approximation,
we must have,
from
- NMgM
D
This relationship is the requirement for criticality.
equation for Np gives the critical fuel concentration
=
Nf _
Figure
5.21 shows a
Nmo*>
/(,
+ DB*
1)
r)
= vo/fof1)
(5-185)
as 0.124
1024
It
N F(R)/N M-
sec.
DIFFUSION THEORY
5.4]
213
- i)
MF
of
Um may
be obtained
as a
function of R by the
MF(R)
4* A,
[isaa^(v
1)
fl[2(*>
(5.186)
This function
through
decreases and
minimum
then
as
rises again
to
by (5.186) because
the one-velocity model does not pro
vide an adequate estimate of all the
neutron losses resulting from escapes
through the boundaries of the system.
e. Rectangular
Block and Finite
Cylinder. The methods used in the
analysis of the infinite slab and cyl
is not predicted
inder
Fig.
Rectangular
5.22
block reactor.
V*{x,y,z) + B*(x,y,z)
where B2 = (c2/ 2)/Z).
may be written in the form
We look for
4>(x,y,z)
F(x)
= 0
a solution of
G{y)
H(z)
(5.187)
(5.188)
REACTOR ANALYSIS
214
[CHAP.
F"(x)
H"(z)
F{*)
It
Ki
G{y)
H(z)
C5,19U;
An appropriate
set of
k\ + k\ +
fcf
Bi
(5.191)
be stated
in terms of the
extrapolated
+ &,y,z)
0,
outer
is
flux
(x,y,z)
example,
4>(
The neutron
(1)
for
4>(x,y,z)
surfaces
where
is
flux
neutron
x,y,z).
of the
(2)
The
flux
vanishes
at
the
rectangular
the extrapolated
J,(x,y,0)
= Jv{x,0,z)
Boundary condition
is
Jx{0,y,z)
Thus,
directions
of the x, y, and
sin kxx +
F{x) =
cosg
cos kxx
and
F(x)
is
the necessary correction will be small and, furthermore, that the region of
primary importance in the reactor
near the center of the block.
With
this approximation in mind we can confine the requirements on the
sec.
DIFFUSION THEORY
5.4]
215
(x,y,z)
cos
2d
cos
26
cos
(5.193)
2C
We have selected for the eigenvalues fci, k2, and k3 the values correspond
ing to the fundamental modes. It follows that (5.191) may be written
(5.194)
This
rec
k =
1.
fej
(1
+ L2B2)-'
The value of
(5.196)
B2 to be used in
this relation
is
given by (5.194).
Another geometric configuration which is of practical importance in
Consider therefore a
reactor design is the cylinder of finite length.
in
form
the
of a right circular cylinder
homogeneous multiplying medium
of height 2h and radius (R (see Fig. 5.23).
symmetry, and the appropriate statement
equation is
\h
It
yp
*(p,2)
*(p,2)
B**^
= 0
(5-197>
source is given by
216
REACTOR
ANALYSIS
[CHAP. 5
be
(5.198)
written
[PR'(P)Y +' %Q
R(p)1""""1
Z(z)
-B*
(5.199)
As before, let
1
R(P)
[PR'(p)Y
V"/J
m-k*'
Z(z)
-i -*
= k\
k*
(5.200)
Z(z)
(5.201)
where
v\
Jo.
= x,
k.K =
(5.202)
solution
(p,z)
^(jf)C0SfE
boundaries.
C5-203)
The criticality
DIFFUSION THEORY
sec. 5.4]
three systems k =
1,
217
we must have
Right cylinder
Criticality condition:
h.
ft
XT
77
(1
A;
1,
is
is
is
is
(R
V2
Vyh
a =
is,
is
it
is
is
is
is
(5.206)
for cylinder
1.08/t
tion for
a large reactor),
then
it
Sphere:
B3
ylinder:
_
~
= T23h\
2B3
=
reasonable approxima
Cube:
(a
is
4t<
B*
~~
3B*~
161
150
B*
130
B>
(5.207)
/x\
sin
J008**
/2.405V
&
Radius
xz
and B1
<R, height 2h
/2.405P\
*H
cos
2C
fey
2b
cos
TZ
(i)1
X
23
Radius
2c
ry
Buckling,
rX
(s)'
cos
2b
Sphere
Rectangular block
2a
Reactor
Summary op
5.3
Table
R
:
is,
The actual critical volume of the block and cylinder will depend on which
which ratio a:b.c and
of the possible proportions is chosen, that
In each of these two cases, however, the ratio may be specified so as to
REACTOR ANALYSIS
218
[CHAP. 5
It
is seen that the least total volume of material is used in the spherical
Thus the sphere yields the minimum critical fuel mass for
configuration.
This result may ultimately be traced to
a given concentration of fuel.
the greater leakage of neutrons from the cubical and cylindrical forms,
which have greater surface-to-volume ratios.
The question of the most
for
is
efficient shape
a bare assembly
one of making the best use of all the
material involved in the reactor.
For example, the material at the
corners of the cube is very inefficiently used, since neutrons readily escape
from such regions, and in a homogeneous reactor there is as much fuel
invested (per unit volume) in the corners as at the center.
Another measure of the efficiency of use of the material in the reactor is
often taken to be the "peak-to-average" flux ratio.
If the flux profile
is relatively flat over the reactor volume, all regions are contributing
fairly equally to the total reaction rate; however, if the flux is strongly
peaked at the center of the reactor, the fuel in the outer regions is being
poorly used. Some measure of this is the ratio mentioned, that of the
highest value of the flux to its average value over the reactor.
In the
case of the bare reactor configurations
(0)
= V- =
>
Thus
Peak-to-average
Vr denotes
where
ratio =
fy.
..
ratio
*W df
It may
be easily shown
(5.208)
that
3.88
3.65
~~
3.29
Block:
Cylinder:
Sphere:
?l*
(5.209)
,
.
reak-to-average
considered here,
DIFFUSION THEORY
SEC. 5.4]
219
medium
material.
In that
instance we identified the source term for the thermal neutrons as the
In the
slowing-down density at thermal energy q(E,h) [cf. Eq. (4.252)].
present case we require a somewhat more generalized function which takes
Thus
into account the spatial distribution of the slowing-down density.
a suitable description for the source term in the diffusion equation for
thermal neutrons would be
5(r) = q(r,Elh)
(5.210)
where q(T,Etb) denotes the number of neutrons slowing past Elb per unit
Unfortunately,
volume per unit time around space point r in the reactor.
the analytical machinery so far developed for treating multiplying media
does not provide us with a detailed statement of the function q(i,Elb);
For
this approach to the estimation of <S(r) will be deferred until later.
the present we will apply an intuitive formulation which will draw upon
Let us define, then, the following source
the concepts already developed.
function :
S(r) = vZ/(T)plbe
number of thermal neutrons which appear per unit volume per unit
time around space point r
(5.211)
Note that, if
fissions
eE/tp is the rate at which fissions occur due to neutrons of all energies.
REACTOR ANALYSIS
220
[CHAP.
(5.211) fails to account for any losses in the neutron population because
Thus in (5.21 1) we have made the tacit
of leakage during slowing down.
assumption that there is no leakage Of high-energy neutrons from the
reactor during the slowing-down process and always move about in such
way that the space form of the neutron density (flux) at all energies is
the same as at thermal energy.1
In all the preceding calculations in this section the one-velocity
-S(r)
= i7/ptlle(r)Z0
kxSa(r)
(5.213)
-Z)V^(r) +
1 See
2a<Kr)
A^.*(r)
(5.214)
sec.
DIFFUSION THEORY
5.4]
be
221
#2
2)
(*2
- 2)
B1 as
J3 o c
ictlfJ
&
(5-215)
a".
I 1
with
L2 given
by (5.75).
k =
2i
(3.16)
n0
In our
neutrons if the system had been infinite and let pNL be the nonleakage
probability for a neutron in the finite system, then it follows that nf pNL
is the number of second-generation neutrons available to the finite system
Thus, rii
k =
n^
Mo
nf pNL, and
= A.pKL
(5.216)
+L2fl2
(5.217)*
By definition,
We can derive this result by an alternate approach.
probability must be given by the fraction of neutrons
Let us compute, therefore, the ratio of the total
absorbed in the system.
number of neutrons absorbed per unit time in the reactor to the total
the nonleakage
J
use
F-g if
ANALYSIS
REACTOR
Pnl
[chap. 5
2.*(r) dr
jr j-f
222
(5.218)
S{x) dr
If
where the indicated integrations extend over the volume of the reactor.
we use for <S(r) the equation (5.22) and recognize that the cross sections
are independent of the spatial coordinates for homogeneous system, then
,1B,
(5.219)
dr
(r)
*(r) dr + DB*
1 ;
2a
*(r) di
]y?
2a
Jv ?
Pnl
neutron
px.,
is
We may further conclude from these results that the probability that a
- Pnl
L2J52
(5.220)
+ L2g2
is
If
is
it
\**
may define
0
__
2J,hyth
uth
(5.221)
is
is
is
mean speed
Va,
(2^
Xh
is
In
vth
and
where
it
2i!")t>th
(5.222)
is
We identify the cross section for escape 2'j; (i.e., the number of escapes
Evidently,
from the relation
per unit distance traveled)
(5.175).
=
DB2.
This concept of DB2 as the expected number of escapes per
2
unit path
The
applicable only to finite single-medium systems.
ratio 2^/2^
obtained by computing the total number of escapes and
absorptions in the system by means of the method used for determining
is
Pl
given by
sec. 5.5]
Pnl
DIFFUSION THEORY
(2f + DB*W
(5.222) reduces to
2>lh(l + L'5)
= hpt,h
(5,223)
is,
223
is
is
Diffusion Length
5.5
is
is
is
it
is
is
it
is
is
it
Two physical parameters required in every application of the onevelocity model are the diffusion coefficient
and the diffusion length L.
So far, our interest in the one-velocity model has been confined to descrip
tions of the spatial distribution of thermal neutrons, and these parameters
have therefore been defined in terms of the thermal cross sections.
We
will show later that similar quantities can be defined in the treatment of
multivelocity systems by the energy-group method, or for that matter,
in any system wherein
convenient to describe the "diffusion prop
erties" of a particular group of neutrons in terms of single speed.
The diffusion coefficient and the diffusion length are fundamental
macroscopic properties of a material which are useful in the one-velocity
formulation. Both quantities can be measured directly in the labo
ratory by suitable experiments.
The direct measurement of the diffusion
coefficient, however, entails the use of a pulsed beam of neutrons.
Inas
much as this experiment involves
time-dependent
phenomenon,
discussion of the experiment will be deferred until after
suitable model
has been developed for the analysis of nonstationary problems in neutron
An experiment for the direct determination of the diffusion
diffusion.
based on
length, however,
steady-state phenomenon, and the impor
tant features of this experiment can be displayed by means of the models
and concepts already developed.
Because of the close relationship
between the parameters
and D,
would be desirable to examine the
techniques for their measurement simultaneously.
This
not possible
of
because
the complications mentioned above; thus for the present we
confine our attention to the study of the diffusion length and an experi
ment for its measurement.
It should be noted that separate measurement of both D and
not
Because of the relationship Ll = DfZa,
always required.
clear that
a knowledge of one parameter yields the other, provided of course that
known.
This
precisely the situation encountered in the
experi
2
ment; thus, in order to determine D we require
separate measurement
of 2. The diffusion-coefficient experiment based on the pulsed-neutron
224
REACTOR
ANALYSIS
[CHAP. 5
M
Fig.
5.24
(b)
tends to be small.
It should be noted that a proper
"crow-flight"
measure of the penetration is the
distance from source to
this penetration
point of capture.
for the plane and line source, but the basic relationship between L and
the crow-flight distance will prove to be the same in all cases.
Consider
then the diffusion of neutrons in an infinite medium from a point source of
DIFFUSION THEORY
sec. 5.5]
225
If /(r)
is
absorbed in
is
if
is,
(5.224)
dx
(5.225)
f5
then we define
a fr*f(r)
r,
is
is
The frequency function can be computed from our knowledge of the dis
Now
tribution of absorption reactions throughout the entire medium.
number
of
unit
in
per
total
time
this
absorptions
the
system and
qo
2>(r) dr the absorptions per unit time in dt at r. Thus the probability
of a neutron being absorbed in dr at
namely, the function f(r) dr,
simply
/(r)dr
MAr
(5.226)
is
|f
D
rV-"
dr = 6L>
(5.227)
4a-D
J0
Jv
is
go
"
radial
(0
(5.228)
is
f,
it
if
(5.229)
is
is
a
is
In
is
is
calculation of this type one selects suitable neutron source for an infinite
medium of the material in question and traces the detailed trajectories of
a large number of source neutrons from birth to capture.
From these
a
is
it
is
226
REACTOR
ANALYSIS
[CHAP. 5
strated
Fig.
5.25
Diffusion-length
experiment.
describing
the spatial
medium is therefore
- DV^(x,y,z)
+ 2a(x,y,z)
=0
> 0
(5.230)
(1) j+(x,y,0)
(3)
(5)
4>(x,h,z)
(+x,y,z)
We look for
1 Conditions
x and y.
*S(*,y)
= 0
= 4>{-x,y,z)
(2)
(4)
(6)
(a,y,z)
= 0
(x,y,C) =
{x,+y,z)
0
=
4>(x,y,z)
(x,-y,z)
(5.231)
The
(5) and (6) follow from the specification that S(x,y) is symmetric in
DIFFUSION THEORY
SEC. 5.5]
227
F(x)
ftl
G(y)
-k\
where
- k\ + k\
set
+/Cj
tf(z)
l>
'
(5.233)
F(x)
Ai
Aj
G(y) =
(5.234)
(5.235)
(5.236)
H(e)
C3 =
It
A3e~u'1.
C3e
(5.237)
H{z) = A3e-""[1
- e-tt->]
(5.238)
= C, cos
G(6) =
and in, n = 0,
= 0
with
km =
C2 cos fc25 = 0
with
fc2m
The corresponding
^
^
(2n +
1)
(5.239)
(2m +
1)
(5.240)
eigenf unctions
are, there
fore,
Cicosfci.r
Gm(y)
-t
(5.242)
k\n
*L
corresponding
k3mn such that
(5.243)
fcL,
is
is
a
It
(5.241)
that
k\
we have
A:,2
k\
Fn(x)
(x,y,z).
Thus we
4>(x,y,z) =
write
n,m
which yields
e-*--'[l
e~2* <*-')]
(5.244)
REACTOR ANALYSIS
228
[CHAP. 5
it
In ^(0,0,2)
Fig.
M
5.26
(b)
will suffice to examine the general form of the flux distribution along the
central (z) axis. This functional dependence is given by
*(0,0,) =
Anme-"-" [1
n,m 0
- e--]
(5.245)
We note however that the quantities k3mn increase rapidly with increasing
values of m and n; thus the higher terms of the infinite series diminish
more rapidly than the first.
Therefore, in first approximation we can
take
0(0,0,2)
~ ^ooc-l"[l
e-2*'"<'-J]
(5.246)
The factor in the brackets represents the "end effects" of the parallele
Note that for very long media (large c) this factor approaches
piped.
unity. Thus the function exp ( k30oz), which gives the usual exponen
tial attenuation in a nonmultiplying medium, is combined with a correc
tion factor which accounts for the end effects in the finite medium.
How
in
is
the case of a short medium (c small), the end effect
not
ever, even
significant until the values of z in question approach the magnitude of c,
i.e., when (C z) * 0. Only then is the end effect important. Figure
5.26a shows a typical graph of the approximate function given in (5.246).
It is evident from (5.246) that there exists some intermediate range of z
sufficiently far removed from both ends of the test medium that the
axial flux distribution in this region varies essentially as the exponential
A region of this sort is indicated by the interval AB in
exp ( k3oaz)-
DIFFUSION THEORY
sec. 5.5]
229
"1"
^loo
~t"
jji
(5.247)
it
The quantities
and
are known from the size of the specimen and fc30o
from the slope; thus, (5.248) yields the diffusion length L.
In practical situation may not be possible to obtain a purely expo
The specimen may be too short
nential region such as AB of Fig. 5.26.
for such region to exist. In this case the end effects on the flux distribu
a
tion can be removed by recognizing that these are confined to the quan
Thus flux measurements toward the outer
tity in the square bracket.
If because of
end of the specimen can be analyzed by means of (5.246).
is
not accessible
and the flux measurements must be made near the source plane, then the
end effect might well be negligible; however, the one-term expansion for
0(O,O,z) will no longer be acceptable and some higher terms in the series
must be retained.
In this case
will be necessary to find the best fit to
some practical
it
REACTOR ANALYSIS
230
the experimental
one might use
[CHAP. 5
4>(0,0,z)
Aooe-1""' + Aioe-k" +
.W-*""
(5.249)
Anm.
some of the ideas developed in connection with the one- velocity model.
It must be emphasized that this computation, which is based on the onecannot be expected to yield accurate results.
velocity approximation,
This model, in its present form, has severe limitations, and some of these
were discussed in Sec. o.4g. The most serious deficiency of the model is
that it fails to account for neutron losses during the slowing-down process.
Even if the reactor in question is a highly thermal system, the one-velocity
model in its present form is inadequate for describing the neutron physics.
The difficulty is due to the form of the source term used in the diffusion
It was pointed out in Sec. 5.4g that this term should properly
equation.
include certain factors (e and p) to account for some of the processes
which can affect the neutron population at thermal energy through their
But these factors alone do not
influence on the slowing-down process.
suffice. There is still a third factor which must be introduced to describe
the losses in the neutron population due to leakages during slowing down.
This factor is especially important in the case of small reactors wherein
the surface-to-volume ratio of the system is relatively large.
At this stage we are not yet prepared to take all these factors into
Thus the example which follows is not suggested as a proper
account.
that these calculations are realistic and therefore can conveniently intro
The
duce further approximations which will simplify the computation.
most important of these is the assumption that the source term in the
diffusion equation is of the form vXf4>; thus, in general, we will ignore all
"fast losses."
As an example of the application of the one-velocity model to the
analysis of nuclear reactors, we will determine the critical fuel mass and
maximum flux level in a light-water-moderated reactor which is designed
to produce 20 megawatts of heat.
Let the core of this reactor be a
60- by 32.3- by 55.4-cm parallelepiped, and for the present analysis assume
DIFFUSION THEORY
SEC. 0.6]
231
that no reflector is provided; thus the reactor from a purely nuclear view
point is a simple bare block geometry of the type studied in Sec. 5.4e.
Assume further that there are three basic ingredients in the core, namely,
water for moderation, aluminum for structure, and U2'6 for nuclear fuel.
We specify that the fuel elements in this system consist of flat plates of
Each fuel element is made up of 18 such
aluminum alloyed with Um.
plates, spaced 0.297 cm apart, and the core contains 28 of these elements,
The light-water moderator occupies the gap between
or subassemblies.
these plates and by the process of
Single fuel
natural convection serves in the
element
dual capacity as the coolant for the
Figure
5.28
gives
60 cm
of side plates.
In this calculation
we confine our
and
the
operating
of
by
Our nuclear
We will
use this
in Chap.
10
REACTOR
232
ANALYSIS
which follow.
multiplying
principal specifications
medium
of the
[CHAP. 5
configuration
Fig.
follows:
in
shown
5.27.
The
20 megawatts;
7.61
Fig.
0.0508 cm fuel
bearing region
I 0.0508-cm
-1
Al clad
M
5.28
(b)
Cross section of fuel element: (a) fuel element; (6) single fuel plate.
Cross-sectional
a.
6.
2. Cross-sectional
cm2
19.2 = 37.0
cm1
vH,0= 0.604
"mo tal
61.3
cm1
t\.t.i = 0.396
properties
5.10 cm1
= 19.2
Total:
The appropriate
Eq. (4.211)].
= 0.656
of the thermal-group
DIFFUSION THEORY
8EC. 5.6]
233
In
system.
be examined later.
In the
case
that
cross section1
<ra
[see
=
C'(333)
Eq. (4.245)].
0(TJ
= 0.66
(J^)]1
= 0.549 barn
wherein we have tacitly assumed that the absorption cross section for
For convenience, we will extend this assumption to the
water is l/v.
well.
The macroscopic absorption cross section for pure
aluminum as
Generated for wjivans (University of Florida) on 2015-09-22 17:19 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
is2
= ^H,0>(333)p(333)a/^[Il0
cm"1
Note that the mass density of pure water at 140F (333K) has been used
for the reactor. The transport cross section, which
is required for the calculation of the diffusion coefficient for the mixture,
is found to be
- Zg"o(293)p(333)
p(293)
sh,o)(333)
2tr
(333) =
In this calculation
0-985
we have assumed
(0 48)(1)
-=
^-US
cm
is
For
U236,
7J(333) = feYv(333)f.(333)
7/(333)
oy(333)/7(333)
10 barns
t,
= 2.08
^J =
<r,
= 2.46
1 In order not to complicate the notation, we have used 9 to represent the thermalgroup properties usually given by ath.
'See, for example, U.S. Atomic Energy Commission, "Reactor Handbook," vol. 2,
p. 39, AECD-3646, May, 1955, for density of water as a function of temperature.
REACTOR
234
ANALYSIS
[chap. 5
f function,
The
obtained
from
= <ra(293)
^(333)
r~M
(4) (333)
a, = 1.4 barns
they
appear
In the
= 0.396).
2<,H''
vH,o
= 0.604
and
S(h,o) = (2.05)(0.604)
= 0)
v,a.,
2-
at
-ai, _
~ ZS
Pa.(Wa1)
(2.7)(0.396)(0.6025)(0.191)
= (2-7)(0.396K0.6025)(1.4)
^^
= 0.00456 cm"1
The transport cross sections for the water and aluminum may be used to
This is an accept
determine the diffusion coefficient of the core mixture.
able approximation in the event that the fuel concentration is relatively
low; this is the case here. Thus we compute for the mixture
=
327r
3[2'> +
=
2[AI)]
(3)(1.24 + 0.0334)
= '262 Cm
The critical fuel mass is computed from the criticality condition for the
The appropriate relation is (5.181) with k = 1. If
one-velocity model.
we solve this equation for 2'aF), the macroscopic absorption cross section
of the Um, we obtain [cf. (5.185)]
+ DB>
~ 1
V
Zj,0)
where 2^' denotes the absorption cross section for the nonfuel components
in the mixture, the aluminum and water.
The quantity B2 is obtained
from (5.194).
If we neglect the extrapolation dimensions for all three
surfaces (i.e., if we use o ~; a, for example), then for a = 27.7, b = 16.15,
and
c = 30
cm,
BH0 (0
+
(0'Om
DIFFUSION THEORY
SEC. 5.6]
If
235
and
i?
= 2.08
into (5.185),
=
2J,0) =
0.0155 cm-1,
obtain
we
0.0155 + (0.262)(0.0152)
1 .Uo
Mmmrl
The corresponding value of the critical fuel mass MF is obtained from the
relation (where V = 107,000 cm3, the volume of the core)
M"
P" =
VS/M, _
~0^>
_-
L346 kg
(5-250)
-zt
(4116)
it will
be
recalled
=
2.(293)
2<H''(293)
2iA1,(293)
+ 2i'>(293)
where
2(H,o,(293)
^H,0)(293)p(333)a
<4h,o
= 0.01314 cm"1
2-(293)
S(293)
= (0-28)(2.7)(a396)(0.6025)
0i00549
= (001808)(687)
^^(333)
561
m_,
=
REACTOR ANALYSIS
236
= 20.36 barns.
o-<H)
18
Thus
(2) (293)1*
and (using A =
t>H,o
= (2K0.985)(0.604)(0.6025)(20.36)
Nrf*
2.
[CHAP. 5
(3) (333)
/0.0408\
0.811
V.
0386
1)
TH = TN(1
+ IAAk)
347K
4>.
eZ/^F
(5.251)
dx dy dz
(-)
The
(5.252)
use
is
it
If we
o, etc., then
abc
(5.253)
= 20 megawatts,
s'
^ (tt),
and
m ana
= -01808
(650) (0.974)
,M,,:",:' ,m
For
which
close
approximation
= 1513
^(li--S^)(13,420)
a
is
we have
10,5
-utrons/cmVsec
Eq. (5.193)].
In the event that this reactor were used as
research instrument,
would be of interest to determine the neutron current at the outer surfaces
of the system.
Presumably, in such an application, the test specimen or
to
sample would be placed as near to the core as possible so as to expose
One can estimate the neutron current at
the highest neutron density.
it
[see
it
4>(0,0,0)
,z)
where
DIFFUSION THEORY
sec. 5.7]
237
written
as
(5.255)
wherein we have selected the center of the x = a face of the core as the
Note that the above expression yields the maximum current
test point.
for any point on that face. If (5.193) is used in this relation, it may be
shown that
= 1.513
1015
neutrons/cm2/sec,
j'+(a,0,0) = 2.27
It
distance
use
for
then
X 10u neutrons/cm2/sec
should be noted in closing that the flux level and current obtained in
The diffusion
of neutrons
from elementary
trons is
in some
cases wherein the actual source yields fast neutrons, the source
term used in the one-velocity equation for the thermal neutrons must be
altered to account for the behavior of the fast neutrons prior to reaching
thermal energy.
Some aspects of this problem were discussed in Sec. 5.4g
in connection with the selection of a suitable fission-source term for the
diffusion equation [see Eq. (5.211)]. In this latter application it was
observed that fission neutrons appear with large kinetic energies and that
during the subsequent slowing-down process some of these fast neutrons
are lost to the system by resonance capture and leakage.
The discussion
of Sec. 5.4g included a few remarks about estimating some of these fast
losses, and
a more complete
neutron first appears in a medium it travels outward from its source point
until it makes a collision.
At relatively high energies the absorption
cross section is small (varies as l/v) so that this collision, as well as a
number of those following, is likely to be a scattering event.
In many
HEACTOR
2.38
ANALYSIS
[chap. 5
As a
ings occur within a short distance of the point of first collision.
reaches thermal energy in the vicinity of the firstcollision point. We can draw, then, the following crude picture of the
slowing-down
We use this quantity as the source term in the diffusion equation; thus,
for a spherically symmetric system,
- DV2(r)
+ 20*(r)
=
4aT2
(5.257)
DIFFUSION THEORY
SEC. 5.7]
(5.257)
239
*(r)
D&rT1
/-
P2|r
(5-258)
e\
5oS.
j*
exp
[-*(/-
(r2
p2
dd
[2*
and
*(r)
sinh xp e~2" dp
Q<2'
\c-.r frsl
WDxr [
Jo
+ sinh ht
Jr
(5.259)
It
r22</)(r)4Tr! dr
we
use
(r)
for
(5.260)
S*(r)4xr2 dr
the result
(5.84)
r2
fo"
r2
If
we calculate
cases
f"
two
is,
that
neutrons);
r2 = 6L2
P
where
trons.
Fig.
5.29
Coordinate system.
(5.227)
X,
we compute,
2/xZ\%^-
2K +
6L2
(5.261)
1/S the scattering mean free path for the fast source neu
first-scattering
the term 2X2.
Gaussian Source.
it
b.
described above
is
REACTOR ANALYSIS
240
a very
satisfactory
[chap. 5
treatment
of the slowing-down
the appropriate
6.
relations.
We recognize, then, that the assumption that a fast neutron is thermalized very near the point at which it suffers its first scattering collision is
As noted previously, a neutron may experience con
extremely crude.
siderable additional travel during later stages of slowing down as it passes
from one scattering collision to the next.
For a medium of the heavier
nuclei, the total distance traversed by a fast neutron can in fact become
quite large in comparison to the distance it covers as a thermal neutron.
The treatment of Chap. 6, which is based on the Fermi age model, shows
that this wandering about during slowing down produces a highly spreadIn the case that the source of fast
out distribution of thermal neutrons.
neutrons is a point source, this dispersion of thermal neutrons is given by
S(r)
qPf_~Zlh
(5-262)
where
If
diffusion equation, then the thermal flux may again be found by using the
Thus if (5.262) is substituted into (5.97), it can be
kernel method.
sinh
sinh xr
rtf-"'1***-"'
dr^j
D*7$k
*M
{*-
shown that
(5.263)
(4r,h)V8x
V*
Vr
source (5.262)
3
slowing-down
is
The mean square distance traveled by the neutrons from the source to
the point of absorption as thermal neutrons may be obtained by using
Eq. (5.260) and the flux given by (5.263); the result for the Gaussian
the relation
Vr
(4rth)i/4x =
(4rth)/4x
As compared with Eq. (5.227), this result shows the greater spreading
out of the neutrons which occurs during the slowing-down process.
c. Isotropic Flux in Slab from First-scattering Collisions.
We demon
strate some of the ideas developed above by computing the isotropic flux
DIFFUSION THEORY
sec. 5.7]
241
q0
flux as the quantity of interest because the effect of the slab is to convert
This effect
the parallel beam of neutrons into an isotropic distribution.
materials
most
in
slabs
of
high-nuclear-mass
would be
pronounced
because of the fact that a beam neutron scat
tered by the slab would emerge with a nearly
isotropic distribution (in the laboratory system of
Plane of
sources
isotropic
"(x)
and
+{a)
(2) J+(s)
(1)
(3)
<Ms)
- x*(x)
= 0
= J~{s)
=
= 0
_(-a)
+
>-
= 0
(5.265)
~ a for
-(x;s)
sinh x(a
s) sinh x(a
Dx sinh 2xa
x)
+ x)
x >
x <
(5.266)
where (x;s) denotes the flux at x due to a plane source of unit strength
at x = s. The subscripts used in (5.266) refer to the notation mentioned
These expressions give a complete description of the kernel for
above.
the slab geometry throughout the region a < x < o.
242
[chap. 5
ANALYSIS
REACTOR
J'a
4>(x)
S(s)
(x;s)
ds
(5.267)
ds
S(s)
S(s) >+(*;*)
JJ
4>(x)
f*a
is
it
s,
(5.268)
S(x) dx
probability that
number of neutrons
which enter slab per
slab surface
2,
2i
[Z,6-z<
beam
0,
dx about x
= qoS.e-^I+") dx
dx]
(5.269)
e^'
sinh x{a
a)
+
s)
sinh x{a +
x)
sinh x(a
x)
Dx sinh 2xa
is
<?o2,
(*)
is
is
ds
ds
(5.270)
(x;s)
properly,
the kernel
o)
is
is
is
is is
DIFFUSION THEORY
SEC. 5.7]
243
causes some error, it does simplify the integration and yields more com
pact results.
It
^ sinh *( "
2xa
x)
DW -I*'')
x)
*(X)
sTnh
(5.271)
2
,
OA
=1.145 cm"
D =0.357
K =0.771
0.3
0.51
is
cm
"
cm
-2 -1
-3
-5 -4
0.1
x, cm
Fio.
5.31
is
A(x\
*W ~
3o
(2)
*(o)
(5.272)
j+(-a)
*(a ~
sinh 2xa + 2Dh cosh 2*a
4q sinh
x)
This result
(1)
is
(5
273)
{^'6)
the microscopic thermal cross sections are taken to be <ra = 2.53 barns,
= 7.86 g/cm3 and
= 56.
and <t, 11 barns, and
The correspond
=
ing macroscopic properties are: iV
0.0846 X 1024 nuclei/cm3, S, =
0.932 cm-1, 2a = 0.213 cm"1, 2, = 1.145 cm"1, x = 0.771 cm"1, and
p
= 0.357 cm.
0.2
data.
REACTOR ANALYSIS
244
5.8
[CHAP. 5
By definition, an isotropic
placed in an "isotropic" neutron flux field.
of
the
surface
of the sample there are
flux is such that on an area dS
dS cos 6 sin 6 dd neutrons per unit time whose velocity vectors
incident l
DIFFUSION THEORY
sec. 5.8]
make angles between
[cf.
'(
and
dS
Eq. (5.36)]
) d
dS
dd
245
^ a SUI^ace (m an isotrpic
whose
of attack have cosines
angles
Aux)
between
and n + dp
(5.274)
n an area
were isotropic, the very presence of the sample would introduce perturba
However, in many reactor situations
tions in the angular distribution.
the assumption of an isotropic flux will afford a good approximation to
Precisely, the simple isotropic distribution
the actual distribution.
would be realized near the center of a spherical surface of large radius on
which isotropically emitting sources were uniformly and densely packed.
model:
In this
it
(5.275)
later that the transmission coefficient (or transfrom a physical standpoint, with our
It is appropriate,
then,
HEACTOR ANALYSIS
246
[CHAP. 5
will, of course,
be related
"**
Vl
= AL =
,T
N<T,(5.279)
,_
(5.279)
is the volume
o-e((
where
<r
we have
= <rj0
(5.280)
absorption
material.
The flux correction factor:
Cfff
(5.281)
(5.282)
Jo
V2(r)
1 See also Sec. 7.4b,
particularly
- x2(r)
Eq. (7.227).
= 0
is applied in the
form
(5.283)
DIFFUSION THEORY
sec. 5.8]
247
\ *(R)
+ ~
[n
V(R)]
\a
(5.285)
where n is the normal vector to the surface and R represents the coordi
is completely specified.
From
we
nates of the surface S; thus,
(i)
(r)
calculate
j(R)
dS
(5.286)
JSi
where
Sl
is the
also
/.
/o =
\ *(R)
17^
V
L<PSL
-|
[n
V*(R)]
/ (r) di
JVl
J-r I
*(r) di
(5.287)
(5.288)
(5.289)
where 4>sl is the average value of the flux over the surface Sl of the lump.
2. The First-flight Transport
Model.
The calculation of the spatial
variation of the flux in this approximation is not so straightforward as in
In general,
model, and no fixed expression can be given.
the procedure for finding the flux at any point inside the lump is to choose
an infinitesimal volume element dV at that point and calculate the total
the diffusion
track length per unit time in dV due to all neutrons impinging on the
The
lump which succeed in passing through dV before being captured.
1
with
sources.
REACTOR ANALYSIS
248
[CHAP. 5
(5 290)*
^-
4>*n
dn
(5.291)
direction
Expression
with (5.274).
approximation
into the lump and over the complete surface area Sl of the lump is ^/Sl4>oSince the probability that a neutron will pass through a distance s of
material without being absorbed is exp ( 2s), we have for the trans
Generated for wjivans (University of Florida) on 2015-09-22 17:19 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
mission coefficient
= cTT
ol<Po
=
Jsl
dS(R)
4r Jsl
/ dS{R)
ir>L
P d
Jo
\
Jo
fi dn
jfof
Jo
e-*-*<*'*M
Jo
rff
d+
(5.292)
Jo
sin
cos
d8
<r
Jo
(5.293)
-a
(^P)/o
2as/
(5.294)
sec. 5.8]
That
DIFFUSION THEORY
249
through
2as
(5.296)*
Diffusion approximation:
(r)
'
1
1
sinh xR
oR
2D
-^Tff.
(*R coth xR
/.
2.
(xR coth xR
2D(xR coth xR
+ 2D(xR coth xR
3
f.
+ 2D{xR coth xR
1)
(5.297)
l)/R
l)/R
(5.298)
1)
(5.299)
- l)/R
(5.300)
where Ro
t1
(Rl
- rDiEiiRo -
2R
cos
r0)
- E^R*
r0
r0)]|
(5.301)
Also,
(5.302)
(po
Po
+ l)e-"]
- tr* +
Po
pi + 2Poe-' + 2e-'
po
= 2Saff
- 2\
(5.303)
(5.304)
(5.305)
and1
E(x)
I:
e-" dy
(5.306)
1 The
E(x) functions are denned according to K. M. Case, F. de Hoffmann, and
G. Placzek in "Introduction to the Theory of Neutron Diffusion," vol. I, p. 153, Los
Alamos Scientific Laboratory, 1953 (U.S. Government Printing Office).
The func
tions E<>to E, are tabulated on pp. 155-161.
REACTOR ANALYSIS
250
[chap. 5
a comparison
small absorption, while the transport results should hold at large absorp
tion.
Note that the diffusion result is totally incorrect beyond a certain
value of 2, actually giving negative values of a.
The infinite slab:
1.
Diffusion approximation:
a cosh
xx
+2Dx
2Dx tanh xa
tanh xa
xa coth xa
f.
2.
face)
+ 2Dx tanh xa
*(*)
8(0,+)
\ 0{E2(Xax)
- *)]}
x > 0
2a
cos
e~z
"
(5.311)
(5.312)
a =
+ Et[X.(2a
XofT* + XqEi{Xo)
Xo
= 22ao
(5.313)
(5.314)
DIFFUSION THEORY
sec. 5.8]
251
Diffusion approximation:
(xR<,
(5.316)
- 2Dx) coth xT
2Dx) coth xT
xT cosh xT + (x'-RRo
sinh xT
cosh
xT
(xRo
2DxRo/R
2Dx)
+
2D/R + 2Dk*R0) sinh xT
(5.318)
= xA
(5.319)
(1
Rl)
(5.317)
*2(R3
- 2DxR0/R
2DxR0/R
(xRB +
2Dx2Ro
in
fo
2Dx2R<,
-~2Dx^
2Dx ^ + 2Dx
- 2D
sinh xT
R, and
3fl
(1
|2
a =
j,
2.
Po
vT^p)
J
4
sv
'
P07)
72)2
a =
7
2/4][(l
- |2T +
- 42aT
*o(l
1
4>i
72o
JT4[(1
= 22ft
/VI
72)]
po
873
1)
(1
U
fl
v2Pe-Po(l
6XP
7)
with
3*o
[1
(1
(1
72)
(1
Pi
1
,
- + - T)po]e-'.-^
- - 7') exp (-p Vl - +
ie7)2
Po,, ~
Po(l
7*)*c-"-')
/i
- (~Po Vl
4(1-7)
- 72)2[i(P0 - - #i(po Vl -
2J)x2fio
- 2D/R
+
2D/R
" _
with
(5.315)
Ro)\
R0x
+ sinh x(r
R0)
xT
{ cosh
1)
A = Rto
x(r
= [xRo cosh
0(r)
^R0x
1.
72)
(5.320)
(5.321)
2/2]e-'"'2
dy
(5.322)
(5.323)
(5.324)
REACTOR ANALYSIS
252
[chap. 5
Diffusion approximation:
1.
<Kp)
a =
f.
4>oIo(xp)
(5.325)
- 2DxIl(xR)/I0{xR)
Io(xR) + 2Dxh(xR)
1
1
(5.326)
+ 2DxI1(xR)/h(xR)
2I1(xR)
(5.327)
xRI0(xR)
/o
/.
2DnI,(xR)/h(xR)
(5.328)
0.8
0.6
^Sla D
8 0.4
-Cylin der
0.2
-;
Spher e
0.8
0.4
1.6
1.2
2.0
2.4
2.8
3.2
2. s
Fig. 5.33 Comparison of transmission coefficients for an infinite slab, infinite cylinder,
and sphere in the first-flight transport approximation.
2$ 4Z<,Fz,/Sl.
obtained
transmission
coefficient is
a =
with
[z2
AR3
+ 2R\1
f~
Jo
dz
fx
Jo
The transmission
r*(l
cos
J.
calculation.1
by numerical
-a
2ZaR
eos
dfi
(5.329)
(5.330)
W.
1, 1950;
sec. 5.8]
DIFFUSION THEORY
253
Bartels.
slabs, cylinders,
and
2K
2K
- 42aa-' In y2Xaa - f
with
~
a
22a + f
+
- l(22.fl)
a ~
+ i(22afl)2 +
1
(22aa)2
7 = 1.781072 .
22ff2
+
.
(5.331)
(5.332)
(5.333)
but numerical
shell.
calculations
j(n)
dfi =
<om
dn
%D4>'nn2
(5.334)
dfi
the direction
part of this
flux vanish;
may be true
that the second term of (5.334) is negligible in comparison with the first.
In the following presentation we first neglect the second term in (5.334),
then try to take it into account in simple cases.
In both calculations,
however, we retain the assumption that diffusion theory holds outside
the absorber; for small, highly absorbing lumps this assumption is in
error.
1. Assumption
of an Isotropic Flux Incident on the Absorber from the
Moderator.
As a first model for the calculation of the flux depression
near an absorber we assume that (a) diffusion theory holds everywhere in
calculated
for the
REACTOK ANALYSIS
254
[chap. 5
absorber in Sec. 5.8a; for example, the fraction of the incident neutrons
transmitted through the absorber will, on this basis, be the a calculated
in Sec. 5.8a, which is appropriate for the geometry and neutron proper
tances from the absorber approach its value in the absence of the absorber.
Such an assumption might very well represent the case of an absorber
placed in a large moderator region in which neutrons are being thermalized fairly uniformly over the region.
However, the results which
have a wider applicability than might be suspected from the
For any case in which
restrictive choice for the neutron-source function.
the absorber is placed in a large moderator region not near any strong
localized source of neutrons and in which the introduction of the absorber
follow
J'+W
j_(R)
and
lim *(r)
- xV(r)
0(R) ~ 2ZV(R)
PRW(R)
=
(5.335)
W(r)
moderator
is,
(-336)
(5.337)
where So
the uniform source intensity of neutrons, 4>0 the flux in the
region of the absorber prior to its introduction, and a the transmission
coefficient calculated on the basis of the diffusion-theory estimate for the
partial currents [see Eqs. (5.49)]. We find then, as indicated earlier, that
is
sec. 5.8]
DIFFUSION THEORY
255
mission
coordinate
*(R)
(5.338)
a is the original
(x)
0[1
De-"'*-"']
x > 0
(5.339)
=[l+2Dx(j-|;)J"1
(5.340)
where asl (for the uniform slab) is obtained from Eq. (5.308) or (5.313),
depending on the neutron properties of the slab.
*>-*>[l-aS$i]
cylinder axis)
=[i+2c*is(;^)r
(5-341)
REACTOR ANALYSI8
256
[CHAP. 5
where aP (for the uniform cylinder) is obtained from Eq. (5.326) or (5.329),
depending on the neutron properties of the cylinder.
Sphere (r is distance from center of sphere)
*(r)
= *o
[l
- ^ -"- j
(5.343)
where aSp (for the uniform sphere) is obtained from (5.298) or (5.303),
depending on the neutron properties of the sphere.
of the mathematical
Two special cases of great practical
problem.
interest, the finite cylinder and the finite slab, are omitted from the dis
cussion here since no general analytical solution is known (to the authors) ;
each case is treated by approximational methods deemed appropriate.
As an example of the use of these concepts, suppose that the (thermal)
absorption cross section of some material were to be measured by putting
Al
tions
is now to be calculated
AL
\(R)(l
is
(R)/l;
since
- a)iirRH
(1
a) of
(5 345)
From the definition of Eq. (5.344), the flux at the surface of the sphere is
related to the flux depression factor D and the flux in the absence of the
absorber
o by
4>(R)
= 6,(1
2D)
If it
(5.346)
is appropriate to use the first -flight transport model for the material
of this sphere, then Eq. (5.303) may be used for a, and we have finally
[using Eqs. (5.338) and (5.344) for 2D, and denoting moderator quantities
by the index
by the index A]
sec. 5.8]
DIFFUSION THEORY
Al
257
2D.(x.v,ff + 1)(1 +
= irR2(t>ot
1
+ 2DM{xMR +
1)
(i
\1
aA)/R
+ A l
- A/fl
(5.347)
Since
incident upon the absorber and consider the angular distribution given in
Eq. (5.334) which is appropriate to the diffusion approximation. We
consider first the results when diffusion theory is also assumed to hold in
the absorber region and subsequently discuss the first-flight transport
model for the absorber.
In the case that diffusion theory holds in both the moderator and the
absorber, and assuming a uniform distribution of neutron sources as
In
In
the
the
moderator:
V24>m(t)
(5.349)
L>M
= 0
absorber:
where
At the moderator-absorber
(5.350)
(5-351)
condi
- DM[n
<MR)
= 4>a(R)
V*(R)]
- DA[n
V^(R)]
(5.352)
(5.353)
addition
In
we require
(3)
Symmetry or nonsingularity on
(4)
lim
Mr)
= 0 =
A.
<ki(r)
(5.354)
(5.355)
This model gives precisely the same results for the flux depression as
that presented in the preceding discussion [Eqs. (5.335) through (5.337)]
if the a is calculated on the basis of the diffusion model [Eqs. (5.283)
In the following, these quantities are given with the
through (5.285)].
dependence on the absorber properties explicitly shown along with the
spatial dependence of the flux for the three simple geometries previously
used. In this case, of course, the results are applicable only to uniform
absorbers.
REACTOR ANALYSIS
258
Infinite slab:
m(x)
0O[1
Infinite cylinder:
1
4>o
(5.356)
cosh xaZ
cosh x.4a
Dmxm
1)
4>m(p)
x>0
JDe-fr-->]
_l_
[chap. 5
-r
(5.358)
^o(x.ifp)
tfo(x.ft) +
(5.357)
DMxMKx{xMR)h{xAR)
(5.359)
1
4>a{p)
00
PAXAI1(xAR)K0(xl,R)
DMxMh{xAR)K1(xMR)
DaxaIi{xaR)Ko(xuR)
Sphere:
u(r)
= 0o
Mr)
= 0o
-v(r-)
A,/
X^
ft COth
DA
x^
(5.360)
(5.361)
(5.362)
j
X.4
+
+
/O(XXP)
/o(x^fl)
X.tftf
R coth
x^
ft
/?
1
or
sinh xAr
sinh xxi?
(5.363)
(5.364)
DIFFUSION THEORY
sec. 5.8]
259
is,
For the absorber, the general form for the transmission coefficient (for
such geometries that the path length through the absorber is a function
only of the angular coordinate of the direction of motion with the normal
in the
to the surface, and not of the particular position on the surface)
notation of Eq. (5.293),
6.
/,
where we define
=
- \*d+
T
Jo
_/0
Me-s-c.*) dn
d+
JoI
'
^-
fi
At
(5.367)
^e-2"'"-*'
dp
(5.368)
is
be applied
Thus
(j^fy *(R)
(5-369)
(yqH;) *(*)
(5-370)
is
it
is
comparison of Eqs. (5.369) and (5.370) that the results for the
flux in the moderator and the flux depression are as presented for the
three simple geometries in Eqs. (5.339) through (5.344), except that the
from
(1
is
/2
it is
is
is
/j
/2
is
/i
(1
a)
is
ratio
to be replaced by the ratio
/i)/(l +.A).
a)/(l +
The quantity
just the a previously used and
defined in Eq. (5.367)
However,
defined in Eq. (5.293).
new quantity which must be
Note that
the fraction of
computed for each geometry of interest.
flux of neu
neutrons transmitted by
incident
sample upon which
trons whose angular distribution
an isotropic distribution weighted
with
For the cases of
cosine factor which makes
more "forward."
an infinite slab and a sphere, the quantities
are presented below; no
(5366)
2Z)*S
Jo
a =
= cos
For the distribution j(n) given by (5.334), appropriate
with
for the diffusion approximation, a may be put into the form
REACTOR ANALYSIS
260
[CHAP. 5
simple solutions and no numerical work for other cases are known to the
authors.
Slab, half-width a:
ft
= e~
Xo '--
Sphere, radius
x0)
po
+ xlEi(x9)]
(5.371)
22a
R:
f%
If
- ix0[e-"(l -
-.[2Po
2(1
P)e-"
p?e-']
(5.372)
= 22i?
fraction of the neutrons incident upon an absorber are nonisotropic in distribution, the expressions developed in the first part of
this section for determining the flux inside the absorber, and therefore
also the self-shielding factors, will be invalid. The correct calculation of
these quantities, although straightforward, is laborious.
effort has been directed toward
(c) The foil problem: Considerable
improving the models presented above for the particular case of an
absorbing foil in a diffusion medium, i.e., geometrically, a slab whose
thickness is small in comparison with its extent, but whose extent is
a sizable
G(r)
1
W. Bothe, Z. Physik,
- 0*(r)
(5.373)
DIFFUSION THEORY
sec. 5.8]
261
distribution T(r) is to
/3
-fVsr(r)
G(i)
= |80(r)
be
(5.374)
requires
computations; however,
a simple
case
will
be
con
form
T(r) = cF^r) +
V2F,(r)
where
C2F2(r)
VJF2(r)
- ^ *(r)
(5.375)
= 0
(5.376)
i.e.,
T(R)
If
(5.377)
T(p) = ci +
c2
In
h(xAp)
p
T (
R\ Tl
i*xaK)
41-
Ii(*aR) K0(x\fR)
^aHa
umxm
h^R)
Kx{xuR)
(5.378)
T(
iW
T
'o
fx,
\(*AR)Kt>{xMR)
\ 1 4I" j^d^di
^
DMxMh{xAR)Kl(xMR)\
("
[L
Io(xAp)
h(xAR)
(5.379)
In this result, we
use
0 = esy1'
where
Sy11
(5.380)
KKACTOR ANALYSIS
262
[CHAP. 5
PROBLEMS
S.l Consider a sphere of radius R which has at its center a point source which is
emitting neutrons of one velocity at a rate q0 neutrons per unit time.
a. Find the flux distribution within the sphere, using the boundary condition that
no neutrons return from the surrounding vacuum (evaluate all constants).
6. Show that the flux computed by a vanishes at approximately r = R + 2D.
c. Find the flux distribution within the sphere, using the boundary condition that
vanishes at r = R + 2D.
d. Compare the two expressions obtained from a and c for {r), and show that the
1.
expression of part c is approximately equal to that of part a when D/L and L/R
e. Using D/L = and R/L = 2, compute the net current of neutrons at the surface
of the sphere for cases a and c, and find
(r)
/(e)
Jia)
(r)
(see
Table 1.1).
DIFFUSION THEORY
2G3
(r)
- (*. -
1)
Note that the flux may have a singularity as r 0 because of the uniqueness of the
center point of the sphere.
b. Determine the expression for
in the following special cases:
(r)
d.
B.
b.
What fraction of all neutrons produced per unit time in a finite, critical, one
velocity, cylindrical, bare reactor escape across the curved surface?
6.7
264
REACTOR
ANALYSIS
[CHAP. 5
rod
Absorber
DIFFUSION THEORY
265
c. Apply the boundary conditions to eliminate one of the arbitrary constants in the
How would you evaluate the others?
general solution for each region.
d. If all the constants are evaluated (not required here), one can show that the flux
has the form (Z moderator < Z core) shown in the sketch.
(a)/
REACTOK ANALYSIS
2(56
Part A.
[chap. 5
General Solution
1. Set up the
balance, and the
fuel component
2. Show that
to the form
appropriate equations which describe the neutron-balance, energyequation of state for the gas mixture which consists of a nonfuel and a
List all boundary conditions.
(fissionable material).
the neutron-balance and energy-balance equations can be reduced
Neutron:
VJ0(r) +
= 0
(P5.1)
Energy:
VT(r) +
= 0
(P5.2)
where
'
V'
kD(n +
1)
|w/
- [no
(P5.3)
and {r) is the flux, T(r) the absolute temperature, k Boltzmann's constant, e the
energy release per fission, and n = No/Nr (the ratio of concentration of nonfuel to
fuel). Let
Ko +
jr^r
OX,r
(P5.5)
Here, N(r) denotes the total number of nuclei (of all kinds) in unit volume at r, a. is a
suitable microscopic scattering cross section (may be approximated by nonfuel
components), and a is some constant (say 0.7104).
3. With these definitions,
a. Show that R = fto(l + m)
6. Find the criticality condition
c. Determine the temperature function and show that it may be written (x = r/Ro)
T(x)
T*
+ it)
[(1
+ ,.)'
- *JI
(P5.6)
d. Find also the expressions for the flux <(x) and the nuclear density jV(x), eliminat
ing all arbitrary constants by means of the boundary conditions.
4. Sketch the functions T(x), (x), and N(x).
a. What is the condition on n if the diffusion equation is to yield a reasonable
representation of the neutron physics?
b. Find an approximate expression for the temperature at the center of the sphere
1.
T(0) which is valid when n
5. Critical fuel concentration.
Let D s Xtr/3 l/3ff,.V,,v with A/*.v = p/kTn and
T = 67V Thus we compute a diffusion coefficient in terms of some average tem
If we call
perature of the system 7'v which is defined in terms of TV
2ba.
(P5.7)
"=aW>
then
a. Show that the critical fuel concentration ratio n is
(2
+ M)(,
where v m m//af/\
b. Sketch n as a function of j. for various
$.
- 1)
ft.
267
DIFFUSION THEORY
6. Derive expressions for the total power output of the reactor by two separate
approaches.
Part B.
Computations
7. Assume that a gaseous reactor is a sphere of the type analyzed in Part A, and
Find the critical
let the material be UF6 so that the fluorine acts as the moderator.
enrichment of U*".
a. Derive expressions for a. and <rj,0) on the premise that the US3S enrichment is
negligibly low.
b. Compute these cross sections using the following data:
a.
rjin
F"
U"5
8.2b
3.9
10.0
aa
2.756
0.01
687
"1
0
0
580
8. Compute n, T(0), $, n and the enrichment N13i/N-<3a for the critical system,
Use f = 1,200 ergs/(cm) (sec) (C),
using To = 1000K, ft0 = 10m, and p = lOOatm.
e = 3.2 X 10-4 ergs/fission, a = 0.7, and b = 2.5.
9. Compute the total power output of the reactor.
10. Compute and plot (x), N(x), and T(x).
CHAPTER
6.1
The Combined
velocity approximation.
The
attention to be
focused on the diffusion process, and the application of the model to the
determination of the spatial distribution of neutrons in a finite system
was demonstrated.
use of this model allowed
268
SEC. 6.1]
269
tinual degradation
REACTOR
270
(1)
(2)
ANALYSIS
[CHAP. 6
scattering
cross section
sys
unity
Except for statements (6) and (7) we offer no explanation in support of
this particular choice of assumptions.
Statements (1) through (5) will
be justified in Chap. 7 wherein the Boltzmann equation is reduced to the
It is appropriate to remark on the validity and usefulness
present model.
of assumptions (6) and (7) here inasmuch as we will use the continuous
slowing-down model to describe the trajectory of the fast neutrons in
energy space. This model was introduced in Sec. 4.2b in connection with
It will be recalled from
the slowing-down process in an infinite medium.
that application that the actual trajectory in energy space is a discon
tinuous function; however, if the decrements in energy per collision are
small, the slowing-down track may be approximated by means of a con
tinuous function (see Fig. 4.11). It is clear that this would be a good
approximation to the physical system only if the slowing-down medium
Thus- in the present
consists primarily of large nuclear-mass materials.
Furthermore, the require
study we exclude hydrogenous substances.
will also tend to satisfy condition (6) [see Eq. (4.30)].
ment that
It is known from experiment that the scattering in the (C) system is
nearly isotropic for materials with large A, provided that the neutron
Below the ev range, thermal effects
speeds are in the ev to Mev range.
become important; above the Mev range, scatterings tend to be forward.
In the work which follows we will assume that all collisions are isotropic
in (C) and moreover are purely elastic. We will not consider inelastic
Actu
collisions in the present treatment in the interests of simplicity.
diffusion
model
the
combined
slowing-down
ally, those applications of
Such systems
which are most useful involve highly thermal systems.
include relatively small nuclear-mass materials and, except for the struc
tural members within the reactor, do not serve as a significant source of
Al
inelastic scatterings
271
SEC. 6.1]
b.
4>(x,u)
Consider a small volume dt of material in a medium supporting a multivelocity neutron flux at steady state, and let us confine attention to those
It is possible to write an
neutrons whose lethargies lie in du about u.
expression for the neutron balance
in the differential
volume dt du in
Sources-
leakage
/
1
c
du
+U
q(r, u+du)
Neutron
losses
+ ?(r,u) di du
(6.2)
u
r
u
r
[chap. 6
REACTOR ANALYSIS
272
This compli
identified only through the use of the Boltzmann relation.
cation is similar to that encountered in selecting a suitable leakage term
for the one-velocity model (see Sec. 5.1c). In fact, it is from the onevelocity result that a hint is obtained as to the form to be chosen for the
Further, the
proportional to the gradient of the neutron density.
constant of proportionality would be identified as in (5.51c) so that,
finally, we could write for the leakage
be
- D(u) V20(r,u)
8(r,) =
(6.3)
assumed form
lim
du->0
-D{u)
be
[q{r,u
+ du)
q(i,u)]
q(r,u) du
(6.4)
ou
written
- ou g(r,u)
(6.5)*
{Zt
(u) *(r,u)
rX*
(6.6)*
be zero.
In an actual case, of
course, the slowing-down process is not continuous; a given neutron takes
on only a limited number of lethargy values in slowing from fission
For a small (but finite) lethargy interval, it may be
energy to thermal.
sec. 6.2]
273
true that most neutrons which slow down through that interval never
take on lethargies in that interval. From this viewpoint it is rather sur
prising that (6.5) gives good results in application. That the result (6.5)
is justifiable to a given approximation by consideration of the energydependent Boltzmann equation indicates that (6.5) is useful because we
are usually dealing with very large numbers of neutrons.
Thus while
the slowing down and diffusion of any one neutron are by jumps, the
behavior of the total neutron population closely approximates that of a
continuous flow (in space and lethargy).
We conclude the present remarks on the combined slowing-down diffu
sion model with the reminder that the set of equations (6.5) and (6.6) are
to be applied
to those situations
which
satisfy
the conditions
(6.1).
boundary conditions,
have been used in situations wherein one or more of these requirements
Although good results can be obtained in some
have been violated.
such instances, usually requiring considerable modification to the original
equations, such applications generally yield erroneous results and must
be used
with caution.
6.2
generalization
quantities, each being defined for the real, positive range. Furthermore,
the neutron-balance relations for these two situations are mathematically
-DV^(r,t) +
2a>(r,<) =
4>(i,t)
- |n(r,)
= vn{r,t)
(6.7)
(6.8)
REACTOH ANALYSIS
274
[CHAP. 6
n(r,t)
_,
D(u)
VJg(r,M) +
2(w)
.
.
\Ar q(T,u) =
-DV*n(r,0 + Hjm(r,t)
- q(t,u)
- Jn(r,0
d
(6.9)
(6.10)
Q(i,u) exp
2al'd'j
(6.12)
and
exp
[- f^HaPdl']
absorbed.
The substitution of (6.11) and (6.12) into (6.9) and (6.10) yields
H ^(M) - I
=
DvV*N{t,t)
Q(r,)
JV(r,0
(6-13)
(6.14)
equa
sec. 6.2]
275
to the function
tj
of
by
(6.16)*
For the transformation u > r we write Q(r,u) > Q(r,r), and for t> ij,
iV(r,<) > iV(r,ij).
(This is a commonly used notational convenience
The application of these
if
misunderstood, can lead to mischief.)
which,
to (6.13) and (6.14) gives the relations
V!Q(r,r)
VW(r,)
transformations
Q(r,r)
(6.17)*
N(t,v)
(6.18)*
first of
"initial condition,"
REACTOR ANALYSIS
276
[chap. 6
Qfr/T
^U"1
0,
oo
r >
< x <
(6.19)
oo
Q(x,t)
slowing-down
t >
Q(x,t)
= 0
(6.20)
density
(3)
Q(-x,r)
dx = g0
is
JQ"
with
r(u)
(6.21)
0(x,s) = {Q(x,r)}
JQ"
(6.22)
= lim Q(x,t) =
(6.23)
r->0
r>0
6(x,s)
(6.24)
e(jt,8)
<
is
Q(x,
0)
is
@(x,s) =
(a) exp
(-x
(x
(3)
go
y/s)
(6.25)
sec. 6.2]
277
as general functions
of the
Jo
e(*,) dx
= 9o
^
*8
(-x,a)
= (x,s)
(6.26)
J0
exp ( x
y/~s)
dx + B(s)
be taken
Jo
exp (x
V s) dx
~
4$
Q(x, u)
O<u0
Fia.
6.2
With
density remain finite (for all positive values of x).
=
y/s,
this choice the above relation yields the result A (a)
and
qo/2
(6.25) may be written
slowing-down
e(
goexp(-xv^)
2
Vs
=-H@(x,s)\
9-^-
(6.27)
is obtained
q(x,u)
In Fig.
is
[4rr(u)]
(6.28)*
as defined
REACTOR ANALYSIS
278
[CHAP. 6
We observe from this figure that q(x,u) has two important features.
First, we note that for any given value of lethargy the slo wing-down
density is greatest at the source plane x = 0; and second, the effect of
the resonance-escape probability p(u) is always to lower the magnitude
of q(x,u) for given x. It is clear from the figure that the high-energy
neutrons which have suffered, on the average, only a few collisions are
concentrated in the vicinity of the source. This effect is in keeping with
the continuous slowing-down model which allows the neutron a degrada
tion of energy that is continuous in time as it wanders through the
medium.1 It is to be expected, then, that the neutrons of lowest energy
Slowing-down
6.3
will be found farthest from the source (see, for example, the curve u = ut
in Fig. 6.2) and will have distributed themselves more uniformly through
out the medium.
However, regardless of the neutron energy in question,
is
(u
0,
|x|
is
low,
over such a large volume that their density in any given locality
and (2) the presence of neutron-absorbing materials causes the continual
removal of neutrons from the slowing-down stream and progressively
fewer neutrons are available in the lower-energy regions.
Let us examine now the analogous problem of the distribution of onevelocity neutrons in the space-time system. Consider, then, an isotropic
v,
is
is
Fig.
<
be satisfied by N(x,rj)
<
are
=
(6.30)
4*
it
Again we find
convenient to specify the source strength in terms of
function denned for an absorber-free medium, N(x,ji). This function
related to the actual neutron density through Eq. (6.12), namely,
n(x,ri) = N(x,ri)e-''"
is a
nm dx
/;
(3)
go
0,
x
=
=
neutrons per unit area at
(2) No sources in the medium except at x =
(1)
(6.29)
oo
1>>
N(x,n)
rj
dr\
0,
N(x,r,)
dx2
is
that
279
is,
sec. 6.2]
(6.31)
(6-32)*
(4*L>*
as a parameter.
At small values of
n(x,t)
would
show
of time
I,
x
is
0.
sharply peaked form much like the shape of q(x,u0). This curve would
represent the spatial distribution of neutrons soon after their release from
=
the source plane at
As increases, the density function n(x,t)
flattens much like the curves for ui, w2, etc., of Fig. 6.2, indicating that as
time progresses the neutrons wander farther from the source plane and
tend to spread more evenly throughout the medium.
One can also sketch
from Eq. (6.32) a density-time function for given x analogous to the curve
of Fig. 6.3.
Such a curve would show how the density at
specified sta
tion would vary as the initial neutron burst passes by. In this case the
interpretation of the curve would be as follows: At short times after the
burst, the neutrons have not yet had time to reach point x, and therefore
the density would be low; very long after the burst, the neutron density
has fallen everywhere because of spreading out and absorption.
Thus
from the viewpoint of an observer at x, a pulse of neutrons passes some
time after the initial burst at the source.
The time at which the maxi
x
is
= r)(t) = Dvt.
where x2 = 2/Z) and
The solution to (6.29) which
satisfies conditions (6.30) can be obtained by the procedure used for the
The result
lethargy-space problem.
easily shown to be
(6.33)
REACTOR ANALYSIS
280
[CHAP. 6
will outline briefly the analysis of the lethargy problem for a line source.
This system is axially symmetric, and the appropriate differential equa
tion is
1
d_
St
P dp
Q(p,r)
> 0, r > 0
where
(6.34)
The function
go
fv
Q(T,r)
dl
S[?^,vUr)U-c)
= qo
If
0 is the transform of Q,
as before
Hp
p 1
pFpe(p's)
S@M
= 0
S(p,s)
A(s)
7o(p
Vs) + B(s)
Ko(P
Vs)
The coefficients A(s) and B(s) are obtained from condition (3).
is
these integrals
1/s.
+ B(s)
j*
Z(p V~s)p dp
is
A(s)
The first of
The
relation is found to be
jT"
appropriate
Ko(P
Vs)
divergent;
(3)
(6.35)
2d ed., p. 388,
sec. 6.2]
281
The Laplace inversion yields the function Q(p,r) from which we obtain
qM,^p
(6.37).
It
(q0
neutrons at time
= 0),
9(r'Wo)
(6-41)
[L(u;u0W
r(u;0)
- r(.j0)
= r(u)
- r(u0)
(6.42)
In the
n(r,^.A)=-
tQ
[4rDv{t
r0,
(6.43)
r0,
282
REACTOR ANALYSIS
[CHAP. 6
r0, u0
.
S(r0)Mo)
p(m;mo) e-lr-riv^(w)
=
yr,
IA
[4ht(u;w0)]1
(6.44)
< .
is
for
In
du
Jr
T\
5(r,u^0)Mo)
jf
/[
|fl"rfw0
jy
g(r,u) =
zero.
du
[^(ujwo)]1
a
'
*g(r,w^0,Mo)
neutrons appearing at
is
is,
S(t,u)
g0
(6.46)
SEC. 6.2]
283
S(i,u) dx dy dz du
where we have selected the z axis as the source line.
is the number of neutrons which appear in the volume element dx dy dz
per unit time and which have lethargies in du about u. The number of
g0
dz
J"^
range
J"^
dy
dx
i(x)
6{y) 5(u) du
=
neutrons/time/length
q0
0;
is
00
5(u) du =
5(u) du =
/ t
/CO
).
js
t/o)2
(z
x0)s
p(w;wo)e-r,/4r
du0
[4ht(m;m0)]?
/_".
(y
(x
rj
where
dx0
dy0
dz0
dy
The lethargy
Zo)2.
integra
x0
and
yo
integrations
dZo
&{Xo)
8^e~",l,rW
dx
The
[$$1
/_".
9(x,J/'z'M) =
/_".
tion yields
[4irr(u)]l
and
Zo
is
The
y1.
it
p1
= x1 +
where we have defined
in the form of the error integral,
In integral
is
**>
/*
(6.47)*
\U
seen
D(u') du'
~kwv
is
&t
I,
JO
commonly
J0
is
t(m;0) = r(u) =
,
,
q(x,y,z,u)
iB .Cx
(6-48)
REACTOR ANALYSIS
284
[chap. 6
The presence of
is,
2<
in which the
t give nearly
the same result; however, for consistency, the definition (6.47) will be
used here and in the remainder of this work.
is
is
is
it
it
is
it
" GWe"r"4r<U)
|V(u)]t
<6-49)
neutrons
r2
q{r'u)
at
u by means of the
r2/(r;w) dr
neutron
(6.50)
acquires lethargy
u in dr
fQ"
r*{u)
relation
This probability
obtained by taking the ratio of the number of neutrons
slowing past u in the spherical shell dr to the total number slowing past u
Thus
throughout the entire medium.
is
^,u)^
dr
(6
JQ
Kr;u) dr
5(r,M)4xr2 dr
51)
The substitution of this relation along with (6.49) into (6.50) yields the
result
=
(6.52)*
r(tt)
is
a
is
0)
sec. 6.2]
285
It is seen from Eq. (6.48) that roughly the square of the scattering
mean free path appears in the numerator of the integral for the age and
This leads us to the gross observation
appears in the denominator.
that a large age, and therefore a large slowing-down length, is associated
a large scattering mean free path and a large nuclear mass.
A large
scattering mean free path clearly implies long free flights between scatter
ing collisions; thus for a given energy loss per collision, a neutron in a
medium of large X, need travel a greater distance, on the average, to slow
with
Table
6.1
Age
Material
HsO*
Age to indium
resonance, cm!
Estimated age
from indium
to thermal, cm'
26.7
....
....
....
Be (1.85 g/cm)
BeO(3.0 g/cm3)
C(l .60 g/cm)
80.
17.2
Age to
thermal,
cm'.
27.7
109.
93.
78.
97.2
105 10
311 3
53
364
Progress Report for Period Ending Sept. 1, 1959, Oak Ridge National Laboratory,
ORNL-2842, pp. 109-112, Nov. 9, 1959; a summary of the most recent measurements
is given in U.S. Atomic Energy Commission, Power Reactor Technology, 3 (4), 31-33
(1959).
through
E. I.
tends to lengthen
Fermi
age.
the slowing
BEACTOK ANALYSIS
286
[CHAP. 6
from the results obtained above that the choice of moderators is limited
to those materials which have the smaller values of the Fermi age (from
fission energy to thermal).
The intention here, of course, is to slow down
the fission neutrons in as small a volume as practical ; this can be accom
plished only if the characteristic dimension of the moderator is much
larger than the slowing-down
length (i.e., if \/r <K moderator size).
Table 6.1 lists the Fermi age (to thermal1) of some materials of interest
to reactor technology.
6.3
The Unrefiected
(1)
(2)
fission sources
(3) The nuclei of the medium are at rest
(4) The slowing-down process terminates at thermal energy
(5) Neutrons in the thermal range can be treated as a single
group
/
>
(6.53)
"Age
to
thermal energy.
sec. 6.3]
287
on the neutron population at the higher end of the lethargy scale, i.e.,
the thermal group.
Practice has shown that this model is entirely ade
-D(u)
S(r,u)
- A q(j,u)
0 < u <
g(r,u) = 2,(m)
- AbVV.h(r)
(t,u)
Mth
(a)
(6)
+ Z**,(r)
= q(t,uth)
(6.54)*
(c)
Note that the coupling equation which involves the total cross section has
been selected. Experience shows that this relation gives satisfactory
results for the systems of interest here.
The last of these equations
the
of
in the thermal range
describes
distribution
the
neutrons
spatial
(c)
by means of the one-velocity model. It is clear that the source term1 in
this equation is indeed the rate at which neutrons slow down into the
thermal range from above (direct contributions to the thermal group
from the fission process are negligibly few). For the case of the monoenergetic fission neutrons, the source term in the slowing-down
may be written in the following form:
S(i,u) du
=
|outh
2,(u')
4>{r,u')
du' +
^^(r)]
() du
relation
(6.55)
This relation gives the number of neutrons which appear per unit volume
and time around r with lethargies in du about u.
Note that we have
used the delta function S(u) to represent the fission spectrum; the origin
u = 0 may be selected so that it corresponds to the peak of the j(u)
curve.
The factor in the bracket gives the spatial distribution of fission
neutrons; the first term of this expression represents the number of neu
trons produced by fast fissions, the second term, the number produced by
thermal fissions.
The delta function &(u) gives the lethargy dependence
of S(t,u) and states that neutrons which appear from fission enter the
system only at u = 0. Note that the above description assumes that
the space and lethargy distributions of the source neutrons are independ
ent ; thus all fissions in this reactor model have the same energy spectrum,
regardless of where they occur in the system.
The distribution of fissions in the reactor as a function of lethargy, at a
given space point, is given by the function /(r,w), see illustrative sketch,
Fig. 6.4. Thus/(r,w) du gives the number of fissions, per unit volume and
time at r, due to neutrons with lethargy in du about u. Evidently, the
integral of this function over the entire lethargy range gives the total
1 Cf. Sec. 5.4g,
288
REACTOR
ANALYSIS
[chap. 6
thermal
at
ttth.1
the
Clearly,
medium.
An appropriate
Fig.
6.4
f(t,u),
as a function of lethargy.
(T,u) =0
(?)
0 < u <
= 0
Mth
(a)
(6)
(6.56)
Thus properly f is a
function. However, we will assume here that the
lethargy-dependent
extrapolation length is the same for neutrons of all lethargies;2 in so
depends on the neutron
1 The
SEC. 6.3]
289
doing, we separate the space and energy dependence of our boundaryIt should be recognized that this is a gross
condition statements.
assumption and can be justified only if the predictions of the resulting
f.
is
is
fast flux
for q(r,u).
The appropriate
o(r,u)
< u
<
p(u) q(r,u) =
is
-a() VJ9(r,) +
wth
(6.57)
abbreviations
is
is
it
is
if
is,
REACTOR ANALYSIS
290
any conditions
- F (r) Q(u)
- F(r)*,h
g(r,u)
*th(r)
[CHAP. 6
- F(r) *()
+(r,u)
with *(u)
(6.61)
F(r)
()
() Q()
we obtain the
differential equations
=0
V2F(r) + B2F(r)
=0
(6.63)
(6.64)
therefore F{f) =
(6.65)
() =
F(f)
yf>
The functions
[-5s
exp
(')
/; Jo"
exp
aV]
r) *.<]
joU
e-fl*
. pM .
(6.66)
[see (6.47)]
(6.67)
iUty
JXPrgyTobf
(6.68)
tfJJ
Q()
Q(0) p(u)e-*M
(6.69)
these results.
[- AhV2F(r)
+ 2?F(t)]
F(r)
Q(,h)
sec. 6.3]
If
291
V'F (r) +
'
l)p*rf%ta'
mPn*:
F(t) = 0
(6.70)
p,h
+*-xf+BW
>
(6-71)
All the functions which appear in Eqs. (6.54) are now completely identi
fied.
below:
M _
(a)
Q(0)pt<rBW(r)
The volume integral over the reactor of the above functions yields the
relations
q(r,u)
di
Q(0)
p(u)e-B*
f F(t) di
(a)
REACTOR ANALYSIS
292
[CHAP. 6
The critical equation for the bare homogeneous reactor with monoenergetic fission neutrons may be developed from the solutions (6.72).
By the usual definition of criticality, we require that the number of neu
trons introduced into the system be equal to the number produced by
Thus the source function (6.55)
fissions in the subsequent generation.
The lethargy integral
may be used to establish the criticality condition.
of (6.55) yields
fQUlb
S(t,u) du
[/oUth
2,(u) *(r,u) du + Z?
*tt(r)]
This expression gives the total number of neutrons of all lethargies intro
duced at point r in the reactor per unit volume per unit time; but, this is
precisely the number slowing past lethargy zero because, by assumption,
neutrons from fission appear only at lethargy zero.
g(r,0) =
F(r)
Q(0)
S(r,) du
Thus
2,(u) *() du +
2}V,h] F(r)
= "
[Jo
7(M) P(M)
e_BVCU) dU
(6J5)*
+ DB*\
This relation
where the common factor F(r) Q(0) has been canceled out.
the critical equation for the bare reactor with monoenergetic fission
is
Utb
J-
it
2,() *(r,) du +
dr
given by
jR f
is
= J
is
It
is,
neutrons.
2}" *,h
(r) dr
*th (r) dr
Zf
(6.76)
= /0U'h
y{u)
written
du
p(j<)e-"r du + X? Pth<r***
(2? + Z^B2)"1
Pth<r***
+ X?
2>h
pue-""* (2? +
(2?
Dth B2)-1
is
where the integrations are to be performed over the volume of the reactor.
for
Note that this expression for
finite system and therefore differs
If we
from the definition developed for the infinite medium in Sec. 4.8b.
(6.74)
Z^B2)"1
(6"?7)*
sec. 6.3]
293
Note that this definition of the fast effect differs markedly from the cor
The difference is due
responding result (4.298) for the infinite medium.
to the additional factors in (6.77) which involve the size parameter Bi.
This result, however, is entirely consistent with (4.298) since in the limit
for infinite media B2 0, and the two expressions become identical.
If the relation (6.77) is used in (6.75), we obtain
K
2ih(l+L252)
0)
Qth
we use
the
as
interpretation.
into
the
reactor, then
lethargy zero are introduced
If
6.3e.
__.
= Tje/p,hfflhPNL
be given a physical
(6.80)*
If
n0
neutrons of
ftopthffth
woPthfftbPNL/ije
It is important
k = ije/pti.g'tbPNi.
(6.81)
The
at critical [see discussion accompanying Eq. (5.183)].
k is different from unity
for example, the reactor dimen
sions are too small for a given fuel concentration (i.e., the neutron leakage
Thus, given a fuel, (6.80) defines the reactor dimension for
excessive).
which appears in pNL, 0th, and . If
criticality through the parameter
where
fc
if,
multiplication
is
in Sec.
be demonstrated
REACTOR ANALYSIS
294
[chap. 6
(6.82)
slow to thermal
j7,h
and
k = rj/pthpNL
(6.83)
1)
(6.85)
fc,0thPNL
1)
(t
5.
In this
1].
(6.86)
is
is
-AhV,h(r)
=
by (6.74); but, when
(i.e., no fast fissions) this expression reduces
to the form F(r) Q(0) = vZf <lh(r).
If we use this relation in (6.72a),
then (6.54c) may be written
Z?*u,(r)
rtfprt^i)
fc.SItothfcbto
(6.87)
k = 17/pNL
sec. 6.3]
295
ing to (6.87) the system acts as though neutrons slowed down at the same
space point at which the parent fission reaction occurred but with losses
This is by no means what
due to fast leakage and resonance absorption.
actually occurs, since individual neutrons migrate from the point of birth
during the slowing-down process; but, in a single-medium system, wherein
the problem is so defined as to yield a solution by separation of variables,
neutrons are found to diffuse in the same space mode at all lethargies,
including their behavior in the thermal range.
b. The Use of the Fission Spectrum.
With the above model as an
elementary but useful approach to the problem of the homogeneous bare
reactor, we turn now to the more general systems which include the dis
tributed fission spectrum.
All the assumptions established for the
analysis of the monoenergetic fission spectrum model will be extended
to the present
spectrum
itself.
f"*
i(u) du
(6.88)
[** i(u)
JO
Note that u
If
du =
J(Ez(E)
Bth
= 0 corresponds
dE
|j()
or
to E0, and
th
du\
\z(E) dE\
(6.89)
to Eltt.
S(r,u) du
[ foU,h
X,(u')
(r,u')
du' +
2^lh(r)]
j(u) du
(6.90)
S(r,u) =
i(u)
[ph 2/(m')
(6.91)
REACTOR ANALYSIS
[CHAP. 6
is,
296
is
we assume
Again it will be convenient to normalize this function; that
In order for the chain
introduced into the reactor.
that one neutron
reaction to perpetuate itself, the yield from the resulting fissions must be
exactly one neutron; therefore,
jR
where
S(t,u) du
fRdr
[f S,()
*{u) du +
2}^.h]
/RF(r)
dr
du
3(w)
(6.92)
is
(6.93)
fRF{T)di
be
2,() *() du +
2}^]
(6.94)
S(t,u)
F(t) )()
in
(6.95)
-(u)
-F(r)
Q'(u)
(6.96)
m eSo^'
du>
Q(0)]
(6.98)
0(u).
determined from the requirement that [see (6.95)]:
S(r,0)
- g(r,0)
F(r)
/"
(6.99)
Q(0)
<S(r,0)
0;
j(u0c/o'-
is
is
it
[J
reduced to
therefore,
(6.100)
= exp
g(u;u')
p(u;u')
m exp
B2
a(w) dw + B2
/8(w)
du>
297
a(w) dw
= e'BKrW-Hu')]
/}()
where we define
dw]
sec. 6.3]
(6.101)
(6.102)
Note that denotes the fast nonleakage probability and the resonanceescape probability in the usual way [see
A-U=0Eqs. (4.260)]. Clearly, the probability
that
neutron will not be absorbed in
the lethargy interval (0,w)
will
the product of the probability that
not be absorbed in the interval (0,u') and
it
is
traversing
it
is
it
Thus
(6.104)
g(u;u')
p(u;u')
interval (w',w)
probability that neutron will not be captured
while slowing through lethargy interval (u',u)
(6.105)
(6.106)
Note that with these definitions one can also obtain (6.100) directly by
physical arguments.
Thus to find the slowing-down density past u, we
compute first the increment dq arising from the neutrons born in du'
about u' which slow down successfully as far as u without being captured
or escaping from the system (see Fig. 6.5)
;
dq(u;u')
Then, to obtain the total slowing-down density past u we sum the con
tributions from all neutrons born in the range (w' < u). If we call this
REACTOR ANALYSIS
298
[chap. 6
/"
Q() =
dq(u;u')
j*
is
-AkV**(r) + 2?^,(r)
F(r)
|o",h
(6.107)
We have used the slowing-down density g(r,M,h) as the source term since
few neutrons are contributed to the thermal group directly by the fission
As in the analysis of Sec. 6.3a, we take the relation (6.60) for
reactions.
the condition on ^th which allows that solution.
and
determine
0th(r)
The substitution of (6.60) into (6.107) yields an equation for F(i), which
when compared to (6.63) shows that
uth
=
2th
+ Ahfl2
?(">;")
Jo
^th
du
(6.108)
(a)
/(')?(;') P(;')d'
(&)
ffr)
2th
(6.109)
uth
+ AhB2
- wik
2,(u)
*(r>u)
Jo
F(r)
jT"
g(r,u) =
Jo
results
("";) p(u.;u) du
(c)
0.
The space function F(r) satisfies the differential equation (6.63) and the
Note that, except for a constant, the
boundary condition F(T) =
if
y(u) du
is
LiBij
g(-Uth>u')
2lb(l
"th
Jo
2'h
"I
J
+
If
/"uth
fu
is
expanded by introducing the solution for yp{u) from (6.100), using (6.61),
It easily shown that the critical equation
and ^th from (6.108).
v
du
(6.110)
introduce the definition of the fast effect and the thermal utiliza
tion/, then this equation may be written in the alternate form:
t
we
dis
The solutions to the case of the bare homogeneous reactor with
tributed fission spectrum may be summarized in terms of the above
= 7?e/pNL
J i(u)
g(uth;u)
p(u,h;u) du
(6.111)*
sec. 6.3]
299
group treatments
are transcendental
trial-and-error procedure.
One can either assume a size and compute the
critical concentration, or vice versa. We will do the latter.
The next
step then is to evaluate the macroscopic cross sections of the reactor
medium for the entire lethargy range, i.e., 20(m), S/(m), D(u), and , as
well as the thermal-group cross sections at the operating temperature of
the system.
These data may now be applied directly to the computation
of the various parameters which appear in the critical equation, i.e., rth,
In the first trial calculation it is convenient to assume a
pu>, v, f> and L.
value for the fast effect (perhaps = 1) and check it later.
All the neces
sary data having been collected, B2 may be computed from the appropri
REACTOR ANALYSIS
300
[CHAP. 6
This
ate critical equation by trial and error [see Eqs. (6.80) and (6.111)].
first estimate of B2 may be checked by applying it to the calculation of c,
If this value of
which can be obtained from Eq. (6.81), for example.
with the assumed value, the first estimate of B2 will be ade
quate; if not, the computed value of e may be inserted into the critical
As a rule e is very nearly
equation and a second estimate of B2 obtained.
agrees well
equation
physical significance.
d. Migration Area.
no
having
The critical
Fio.
6.6
Critical fuel mass and reactor
size as a function of fuel concentration.
reactors.
B2
<C 1 ,
and the fast effect e, which is relatively insensitive to B2, may be computed
as though the reactor medium were infinite [i.e., Eq. (6.77) reduces to
Thus (6.80) reduces to
(4.298)].
1
Moreover, if also
written
1
rthB2
1,
7//(p,h0ti,pNi.
then
glh
kKgthpNL
(1
^00
=
(1
where
+ B-rtb)(l + L2B2)
M* = rth + L2
(6.112)
^"00
+ B"-M2
be
(6.113) *
(6.114) *
down and diffusion length so that M* is the mean square distance from
Note,
the point of birth as a fast neutron to death in thermal capture.
sec. 6.3]
301
reactor.
g(iii + Aw,;ttj)
(probability of escape in
Am<
Am;)
Aw( is given
from system
Am,
jy
(6.115)
J(r,w.)
g(r,M,)
dS
di
where
Js
and
jv
indicate
integrations
Now
respectively.
J(r,Mi)
g(r,U<)
= Z((m,)
-D(Ui) V*(r,,)
D(m.) *(m.)
f (<) F(r)
VF(r)
We substitute
where we have used Eqs. (6.59) through (6.61).
expressions into (6.116) and apply Gauss's theorem to obtain
- D(ui)
Am,
V0(r,Ui)
Js
- D(iu)
dS
4>{ux) Am<
&(,)
&(,) fy *(r,,) di
*(m>)
fv
fy F(t)
F(r)
these
rfr
dt
The application of (6.63) to the fraction at the right yields B2D{ut) Am*/
Thus the probability of nonleakage while the neutron is in Aw, is
Z,(Mj).
g{Ui
+ AUi;Ui)
2,(u-)
(6.117)
The probability that the neutron will not escape in traversing the interval
(0,w) is obtained by taking the product of the individual probabilities for
- BWjuj)
Am,]
_~ -
YiIn
t\
iIn g(u) =
n [i - 5^*]
1l"i
a
and
OA- ,w [
(6.116)
B*D(Ui)
Am,
REACTOR ANALYSIS
302
[chap. 6
(We have assumed here that the neutrons lost by escape from the system
are few in comparison to the total population.)
In the limit as Aut 0,
the summation goes over into an integral, and we obtain
(6.120)*
6.4
The application
of the general results developed in the preceding sections of this chapter
may be demonstrated by means of two elementary calculations which
have some practical importance.
The first of these is the estimation of
the diffusion length of thermal neutrons in a given material from a
knowledge of the thermal-flux distribution in a block of this material.
A
simplified treatment of this problem has already been given in Sec. 5.5b ;
however, the calculation presented there was based on the assumption
that the neutron source was thermal.
We now discard this restriction
and consider the more general situation wherein the source yields monoenergetic neutrons of some energy E0 > Etb. The present calculation is
of interest, therefore, when "thermal-neutron columns" are not immedi
ately available and fast-neutron
sources must be employed for the
a.
experiment.
is,
is
We
complicated problem.
will avoid the mathematical difficulties involved in solving the more
more simple
general situation described above and consider instead
In place of the block, which may be the actual
geometric configuration.
geometry used in the experiment, we select an infinite medium (of the
which contain localized neutron sources
it
allows
con
proposed because
Eo- This particular configuration
siderable simplification in the analysis without distorting the final results,
is
Applications
sec. 6.4]
303
namely, the expression for the axial distribution of thermal neutrons far
Thus, although this elementary model avoids
the mathematical complications of the actual geometry, it does yield an
-D(u)
4>(z,u)
+ 2.(u) *(,)
q(z,u) = 2,(u)
-Ah ^
(z,u)
S^Mz)
lim Q(z,u)
- ^ ?(*,)
(6.122)
= g(z,.)
(6.123)
requirements
are
= 0
fQ"
(6.124)
Q(z,w) dz = iqo
is
is
in (6.123):
5.
is
only one space variable z. Thus although the general form (5.97)
by no means necessary, and, in fact, some labor can
applicable here,
kernel relation [see Eq.
be eliminated
by using the one-dimensional
it
(5.78)], namely,
D^/Zf.
2AT
S(r)
is
where
+*{e)
/_".
0
(2)
(6.121)
<r"l'~rl
df
(6-126)
EEACTOR ANALYSIS
304
be
[CHAP. 6
(Xe~-'ds
(6.128)
qol^'
^-xV^
[2.
Mz)
cosh xz
e~"
- e"
erf^O
erf(/32)]
(6.129)
VTth
where
Vrih
+ x Vt^
(6.130)
0.
0,
ertw
1-^(1-27.+
--)
(6-13"
e~" erf(/3i)
e"
erf(/32) =
The difference of the two error functions which appear in (6.129) may
written in terms of this expansion, thus
be
sinh xz
x2rth)
This result
is
Ml)(m^y..
(6,32)
to be compared to
^(z) ~ Ae-^oo'
which
(6.133)
is
dimensional
it
>
is
clear that
[see
fcjoo
so
Eq. (5.248)]
= x
(6.134)
sec. 6.4]
305
Thus, diffusion-length
is,
exponential behavior will not appear until much farther inside (see Sec.
5.7c and Fig. 5.31).
Finally, it should be noted that the result (6.132) is generally valid,
that
the thermal flux has the same functional dependence at large dis
is
is
0)
density vanish at the extrapolated boundary and that the initial condition
In the case of
(corresponding to t =
be satisfied at the source plane.
the infinite media, the boundary conditions, of course, did not appear,
and the initial condition was easily satisfied by means of simple integral
>
is
s)
(t
it
of the transformed
differential equation.
The solution obtained
in this way
not available in closed form and must be expressed by
is
is
tances from the source regardless of the energy of the source neutrons.
b. Method of Images
Infinite Slab with Plane Source.
The second
REACTOR ANALYSIS
306
[CHAP. 6
Q(x,r) =
(3) Q(a,r)
(2)
dxj
Q(xfT)
q0
(6-135)
Q(-x,t)
=
|"
(1)
Jim
is
The last relation follows by virtue of the fact that the slowing-down
proportional to the fast flux through the coupling equation
density
2S
Fio.
6.7
_j
Superposition of images.
is,
(6.6), and, in the usual way, we have required that the flux at all energies
satisfy the extrapolated boundary condition. The general solution to
in the case of an infinite medium,
(6.19)
QM
aW"
(6-136)
The conditions required for the slab geometry may be met by distribut
ing an infinite set of source planes of strength +90 to either side of the
slab, as shown in Fig. 6.7. The effect of these alternating "positive"
on the flux
and "negative" sources of neutrons at + 26, + 4a,
Note, for example, how the negative source at
evident from the figure.
inadequate to correct
+25 just cancels the flux at boundary +a but
for the condition at a. The major correction at a
provided by the
in
flux
the
at a due to the
negative source at 2a.
The small error
source at +23
now corrected by
positive source at 45, and so on.
The slowing-down density at due to any particular source of t = neu
is
is
is
...
is
sec. 6.4]
QM
'
be
n = 0, 1
is obtained
(6.137)
by superimposing
the
BO
written
g^i
307
^j
Q-(x,r)
(_i).e-|^*,.a|vr
(6.138)
B as
tion to the problem, we must show that it satisfies the boundary condi
tions (6.135).
The first condition is seen to be satisfied if we observe that
rt >
1
= Km
>
g-(*-2no)'/<r 1
n-1
and that
lim
jrr^r /
J -a
Condition
fT**'** dx
,-,0 L(4irr)
image system.
Q(fi,r)=^~ ^ (_l)ne-|3-23|./4,
nm
=
77^[e~5'/4r
e-s,/4'
e-<SS)''4'
is canceled
(6.139)
out by
one
= 0
The result (6.138) satisfies the differential equation (6.19) and the condi
tions (6.135) and is therefore the solution to the present problem.
In carrying out a calculation it is convenient to approximate
A reasonable estimate
[e-r'Hr
e-U+23)'l4r
the
e~(r-2)<l*r]
(6.140)
REACTOR ANALYSIS
308
[CHAP. 6
for x >
0.
Note that the terms in n > 2 have been neglected since
these terms decrease at least as fast as e~"'liT as n increases.
6.5
Reactor Temperature
Coefficients
a. Temperature
Changes and Reactor Stability. A problem of con
siderable importance to the control of nuclear reactors is the evaluation
of the temperature coefficient of the reactor system.
It is customary to
define the temperature coefficient as the change in the multiplication
turn, will cause the rate of energy production to increase and further
raise the temperature of the system.
Such a system would be unstable
and would require a positive and continuous influence of a control system
to maintain itself in the steady-state condition. On the other hand, if
the coefficient is negative, the effect of a temperature rise on the neutron
population (and therefore on the power production) would be to decrease
the number of neutrons (and power), so that the temperature would then
For
systems such as heterogeneous media and reflected configurations.
these systems, numerical methods are more useful, and changes in multi
plication constant are obtained by making repeated calculations for k for
successive small changes in the temperature-dependent
parameters of the
Here, attention is confined to the bare homogeneous system,
system.
and the emphasis is placed on presenting some of the fundamental con
cepts involved in the computation of temperature coefficients.
The
material presented in this section is meant to serve only as a demonstra
tion of a few of the more important ideas and approaches which can be
utilized in such calculations.
Furthermore, the present methods devel
oped for homogeneous systems may be useful, at times, for obtaining a
preliminary understanding of the temperature response of a particular
reactor system wherein accurate calculations might be unduly involved
and not entirely necessary.
In any case, the temperature-sensitive
The
parameters considered here are of general interest to all reactors.
essential difference in approach taken for the treatment of nonhomogene
sec. 6.5]
ous systems
in
which
is
due primarily
these parameters
309
to the fundamentally
different
forms
configurations.
ability which involve integrals of cross sections over the fast (resonance)
Properly, the fast cross sections which are used in such functions
range.
must be computed from an integral relation of the type (4.210) which
accounts for the thermal motion of the nuclei.
The temperature depend
ence of the fast cross section appears, of course, through
function
2JJ defined
the distribution
the temperature
of the
REACTOR ANALYSIS
310
TV
material
[CHAP. G
2ft,
fast
cross-section
A change in
are easily denned in the case of bare homogeneous reactors.
the temperature produces, of course, a change in the various nuclear
densities which are involved in the macroscopic cross sections.
More
over, the thermal expansions which accompany increases in temperature
influence directly the physical size of the system and therefore affect the
neutron leakage.
dk
kdT
1 dri
3T
dplh
plk
dT
dt
dT
(6.81)
df
dgth
dT
glh
dT
dpNL
pNL
dT
W-lil>*
It
follows
k =
and
p-1
o>
(.
1\
dk
(6.143)*
coefficient
....
sec. 6.5]
311
Note that, in the event that the reactor was initially at steady state
(* = D,
=
af
af
(6145)
or
df
2""'
"I
2/ + 2"J
32(F>
2?> ~dT
Z
1
d2'0>l
~d
(6-146)
where the cross sections 2^' and 2^> refer to the thermal group properties
The temperature
of the fuel and nonfuel components, respectively.
derivative of these cross sections is obtained from the relation
32
,,T
(N)
da
dN
._.
(6.147)
where
NdT
(6-148)
'~ldT
31
is the
dimension
-\t--klr
The derivative of the microscopic cross sections a is obtained from the
definition of the thermal group (4.209) and depends, of course, on a
detailed knowledge of the energy dependence of <r in the vicinity of kT.
Unless otherwise noted, we will assume in this treatment that (1) all
absorption cross sections aa vary as l/v in the thermal range and (2)
scattering and transport cross sections a, and <rlr are energy (temperature)
In the case of a l/v dependence, it was shown previously
independent.
that the thermal-group cross section is also l/v [see (4.234)]; therefore, in
the case of the
aa,
REACTOR ANALYSIS
312
where
[CHAP.
If
system.
(6.148) through
(6.150) in (6.147), we
lw=-(*a
*r)
obtain
kw=-**
and
j|
-f)(a0
= 3(1
- aF)
(6.152)*
^-Wf
dU
-pL{vJT
dpNL
is
Probability.
probability
is
parable magnitude
d. Temperature
dB*\
W>w)
(6-153)
It
is
is
a
D ~ D0, then,
L2~L(1
where L% = D0/2fa0), the diffusion
follows from Eq. (6.155) that
-/)
(6.155)*
It
sec. 6.5]
313
(6.148), we
have
We substitute this result, along with (6.151), into (6.157) and obtain, for
the nonfuel components,
dLh =
dT
Jf
~j
L*0(6ao
(6.160)
dT
L.(6ao
^__L_^)
we define
(6.162)
B2
Mr
dJ*l
dT
lR
_L
dT
(trtt
r=7
- 6a ~
(6-164)
dp
pnl
Pl j-Rjf~
- - 3/o, - -L
/)
If
PL
215)
Pnl dT
d%
dpm.
3ao(2
(6.165)*
~B
Tt"
"aT
(6-166)
ar
B2 term
obtained from the relation (6.163).
The t,
This implies that the Fermi age
term we compute from Eq. (6.47).
model
applicable to the system in question and therefore all the require
ments for the validity of this model are satisfied.
Of immediate concern
is
As before, the
^
^
dT
"
ftn
_L
Probability. We
Coefficient
of Fast Nonleakage
e. Temperature
compute the temperature coefficient of the fast nonleakage probability to
Thus
thermal glb from the expression (6.79).
Tib
'
is
D(u) du
is,(u)
<3C
S,(u).
Thus,
if
is
?1
B2 =
where
If
of the reactor.
(fJ
acteristic dimension
B2
(6.161)
REACTOR ANALYSIS
314
[CHAP. 6
tions of practical
then be written
*i*
dT
a
dT
r,h
Jo
The temperature
importance.
d(m) du =
2.(u)
2.(uth)
fua
dT
J0
derivative of
pw l a
a r|j2,(u)J
aM
rth
may
ic
ifi7^
^a0'J
'
jip
We apply again the assumption that <r, and <rtr are temperature independ
ent; thus with the aid of Eqs. (6.151) and (6.159), (6.167) reduces to the
form
(6.168)
6a0Tth
is
if
alh
^-
6a0
j-
= SVth
is
d<7,h
-w
dln\
la
(6.169)*
t;
f.
v.
is
7j
is
- -rh
is
(3-21)
is
it
is
is
is
with ath denned by (6.58). Note that this estimate for the temperature
derivative of rth
based on the assumption that the essential effect of
variations in temperature
to modify the density of the reactor materials.
We observed above that this
true only
the fuel concentration
low.
Otherwise, the relation (6.48) for Tth will not apply, nor for that matter
(6.79) for the fast nonleakage probability.
The temperature derivative of gth obtained by substituting the results
(6.163) and (6.168) into (6.166).
sec. 6.5]
315
-}
are related by
a(T)
Note that
- a'(T)f.(T)
'(D
(D
[1
(6.170)
+ a'(T)]Fl+a(T)
1
1
(6.171)
+ oV.
+ '
(6.172)
Table
6.2
The Quantity
Temperature
Fuel
K
0
293.6
423
575
751
PuJS
J. A. Harvey and
lished work.
1
J.
+'
f l+a
ev
0
0.0253
0.0364
0.0496
0.0648
1.125
1.132
1.115
1.095
1.081
1.000
0.985
0.993
1.005
1.017
1 .207
1 .000
293.6
1.183
0.997
423
1 . 173
1 .001
575
1 . 163
1 .009
751
1.157
1 .018
1.407
1 .000
293.6
1 .416
1 .016
423
1 .039
")7")
1.418
1.423
751
1 .433
1 .093
1 .075
1 (1956),
and unpub
REACTOR ANALYSIS
316
[CHAP. 6
The substitution of the form (6.171) into (3.21) yields for the temper
ature coefficient of
idT
n:
"il7al;(1
a)-^ W
((U73)*
implies
temperature
not the
case, however,
significant
components;
it
is due
is required.
Probability.
of the Resonance-escape
of some of
only
which
follows
is
intended
to
an
idea
give
The treatment
the factors which enter into the calculation of the temperature coefficient
g. Temperature
Coefficient
Unfortunately,
in any calculation of the resonance-escape probability.
involves
more
nuclear
advanced concepts of
physics than
its evaluation
1 The
resonance
of
sec. 6.5]
317
have been considered so far; thus, the present analysis is limited to some
brief remarks about the nature of this effect, and the reader is referred to
other sources for more detailed studies.1 However, for several situations
of practical importance to reactor technology, the Doppler effect is a
secondary factor in its influence on the temperature coefficient of a reac
tor.
These situations will be discussed shortly; the formulation which
follows is best applied to such systems.
p*mexp[If
LiMm\
(6-174)
means of the usual notation, then the integral in (6.174), which we denote
/, may be written
[Nrf/KE)
+ Nq^(E)}
dE/E
(6.175)
Let us also separate the scattering cross section for the fuel
function.
Let
a(pF)
+ % ,
av m
(6.177)
"
r i+iw
H/>
+ Noa^/Nr]
dE/E
(6.178)
Fig.
10.17.
1 See,
pp.
for example,
398-411,
Neutron Research,"
Mass., 1953.
pp.
Company,
Reading,
REACTOR ANALYSIS
318
[CHAP. 6
Sec. 4.5c.
for estimating the resonance integral I, and we will consider such calcula
tions later in conjunction with the analysis of heterogeneous reactors.
For the present our interest in this result is focused on estimates of the
In this calculation it will
temperature coefficient of the integral proper.
suffice to examine an approximate form for the integral based on the
a(/>(E)
Fig.
6.8
(E)
Em,
<rlam
dE/E
(6.179)
<*p
It
should be recognized that the cross sections which appear in this rela
tion are the so-called effective cross sections which take into account the
thermal motion of the nuclei.
We observed in Sec. 4.7c that only in the
of very-high-energy
neutrons could one assume that the relative
speed (or energy) between neutron and nucleus was well approximated
by the absolute speed of the neutron in the laboratory system of coordi
case
At the lower end of the energy scale, the thermal motion of the
nuclei must be taken into account, and a suitable formula was given in
Eq. (4.210) for computing the appropriate reaction rates. In the present
nates.
/y
r?(|v
- V|)|v - V|9K(V;TA0 dV
(6.180)
we have retained
SEC. 6.5]
319
from Etb to Ei (some energy just below the first resonance) and a second
which begins at Ei and extends over the entire resonance range (see
Fig. 6.8). The energy Ei is so selected that in the interval (Etl,,Ei) the
In the resonance
cross section o\P{v) has approximately a 1/v form.1
<T(aF)(E)
'E
1
V/KE)
S2
= S
(6.181)
+ c'.P(E)]
(6.182)
Co/v<tv
WE)
^Si +
dE/E
(cq/v)
&i =
where
dE/E
dE/E
"
C'^{E)dE
A'
'
\<x?\E) + a'JKE)}
(6183)
If
coefficient of
Pth as
Pth
It
ol
dl Ifrp
Cpdl
iflpoL
follows from our definition of ap [see Eq. (6.177)] and the temperature
of the nuclear density JV [see (6.148) and (6.149)] that
derivatives
=
a ply p
<jp 0 1
$ 1 we
(a0
- aF) In
(6.185)
plE
By changing
["
>th
.
V{V
,
.
Co/<Tp)
2rJln(l
Pl \
<TpVth/
- ln(l +
\
-)]
(TpViJ]
(6.186)
1 See
1
af ~
1 /v,
then
so also is
(i>)
1 /v.
REACTOR ANALYSIS
320
For convenience
we let Vi >
[i.e.,
[CHAP. 6
as small compared
2ap In
Si
(6.187)
a'/ ''(vit>)
Moderator temperature
7\
T2>T,
Ej
(Neutron
Fig.
6.9
, resonance
energy
derivative of Si is
ST
\4T
(6.189)
It
Laboratory.
sec. 6.6]
321
6.6
Application
to a
Heavy-water-moderated Reactor
and nearly all the fissions which occur are due to thermal neutrons.
It
is to be expected, then, that the Fermi age model will yield a good descrip
tion of the essential features of its neutron physics.
A detailed report on the CP-5 reactor is given in the Atomic Energy
document1 "Research Reactors"; a few of its principal char
acteristics are summarized here.
The reactor core proper is an upright
cylinder of D20 2 ft high and 2 ft in diameter in which are immersed
Commission
configuration
is completely
the core and thereby reduce the critical fuel concentration of the system.
Clearly, if we were to ignore the reflector altogether, our estimate of the
1
1955.
REACTOR ANALYSIS
322
[chap. 6
tion.
These data, which were computed by H. Garabedian and are based on the
two-group theory, indicate that the reflector savings for a D20 reflector
In the CP-5 reactor the 62 cm of D20 on the
62 cm thick is about 23 cm.
sides and bottom of the core is supported by an additional 62 cm of graph
ite. This thickness of graphite reflector, if used alone, would correspond
is
is
is
is,
buckling to
"
be
494.
we estimate the
sec. 6.6]
= 2.08
Tth
323
use the
= 130 cm1!
p,h
following data:
= t =
1
be computed
is 330K.
(4.245)
2(d,o, = 0.00002311
cm-1
and
S^A"
= 0.0009531 cm"1
z,n
DB2 42(0)
-r*
>/epth0th
(6 191)
The
2J,0' refers to the nonfuel components D20 and aluminum and has the
value 0.0009762 cm-1.
Because of this relatively low nonfuel neutron
In that
be through changes
use the relation
D(49C)
It
= D(2Q?,loP^C)
= 0-890 cm
case we may
(6.192)
Mf
(f^
= 7183
REACTOR ANALYSIS
324
[CHAP. 6
Provided that our guess for the reflector savings is reasonably close, and
that the U236 is available in the nearly pure form, the above mass estimate
should be a reliable figure for the U2*6 required to make the CP-5 critical
at the full power condition before the reactor becomes contaminated with
fission-product
In order to deter
mine the fuel loading for the contaminated condition, we would require a
Suitable
knowledge of the fission-product buildup during a fuel cycle.
methods for computing these concentrations are developed in Sec. 9.6.
In the absence of such information we will estimate the critical loading for
the hot dirty condition from the published value of 1.15 kg for the critical
The corresponding value of the fission cross section is
mass.
s'
2'"fe), (nra)
"
0007681
cm"
This result may be used to compute the average thermal flux 4> in the
CP-5 when fully contaminated and at a full power. If P is the reactor
power and
-
BXfVm
=
(3.17
1013
neutrons/cm2/sec
The project data in the "Reactor Handbook" gives 2.3 X 1013 neutrons/
cm2/sec for the average flux.
c. Estimate of Temperature Coefficient.
The role of the temperature
coefficient in determining the stability of a reactor was discussed in
Sec. 6.5a, and some formulas for estimating the value of this coefficient
for the various factors which appear in the multiplication constant were
It should be recognized that, in
developed in subsequent sections.
calculating the temperature response of these quantities, the coefficients
per se are temperature-dependent functions and may vary in both magni
tude and algebraic sign over the entire operating range of the reactor.
Thus a complete study of the temperature coefficient for a given system
must necessarily consider all possible operational conditions to which the
reactor may be subjected.
A particular situation which is always of
interest in the design of a new system is the initial startup condition ; this
dition.
sec. 6.6]
325
at room temperature.
we compute the
Specifically,
temperature coefficient of the hot clean reactor from the temperature
derivatives of its thermal utilization, fast nonleakage probability, thermal
of the
reactor
These are
X 10-yC
X 10-yC
aM = 2.38
aF = 1.94
(for a 30.5%
U-Al alloy)
we
Its
value is
aDl0
-^pJf
= 9 876
we
Bg,b
= 0.002353 cm-2
= 0.7364
X 10-VC
will assume that the reactor
of temperature
For convenience in reference we list also the following results from the
for the CP-5 in Sec. 6.6b:
calculations
2 = 0.006751
pL = 0.2367
cm-1
L2 = 131.8 cm2
= 0.8554
pNL = 0.7633
A; = 1
By (6.152),
j^ -
3(1
- /)(oo - a) ~
- /)(a,0 - aF)
coefficient
1 This
p. 30.
= 3.443
X 10-yC
The temperature
obtained
3(1
is
Da
Engineering,"
REACTOR ANALYSIS
326
For
[CHAP. 6
relation
{2,
,2{" +
+ A12
^oS,'1"0'
we compute
U>2
The value for Dl0 we take as 0.527 (see Fig. 4.7) and for
Then, by (6.169),
these yield th = 3.974.
1
dg,h
D2
a'i>
^=BTa\T^~6ao
from the
2,
0.4252 cm"1;
_i_2
+
rKdf)
If we assume
that the reactor does not change size with temperature (i.e.,
the physical dimensions of the core tank do not change with temperature),
then
dT
X 10-VC, and T
= 0
322K, we obtain
-H.95 x io-yc
If
independent,
dpuL
-45.94 X 10-7C
of
we are concerned
and by (6.190)
= 3(a0
- aF) In
Pth
(6.195)
SEC. 6.6]
we have p,h =
1 ;
327
For the
and
for
(fff) (0.005775)
<r>(kTJNF
S;1"0' = 0.4252
cm"1,
= 0.007072 cm"1
= 322K,
Tm
= 293K,
and
= 0.527, we obtain
^ = 8.20 x io-yc
i~
1
The
excess
Hie
multiplication
8k
-49.25 X 10-VC
for 8T
k =
I +
29C.
^ ^\
8T
= 0.01428
is
therefore
PROBLEMS
6.1
Consider an infinite nonmultiplying medium.
Within a spherical region of
radius R in this medium, go neutrons per unit time are being emitted uniformly in
Show that the slowing-down density q(r,r) is given by
space at age zero.
Sinh
where r is the radial distance from the center of the source sphere,
erf (f ) =
exp
Jo
e-(r+B)/4r
t(u), and
6.2
Isotropic sources of zero lethargy neutrons are distributed uniformly over a
spherical surface of radius a.
a. What is the slowing-down density due to this source distribution in an infinite
medium?
b. Sketch the slowing-down density q(r,u) as a function of r for various values of u.
6.3
A point source in an infinite medium is emitting neutrons of zero lethargy
isotropically at a constant rate in time. Assume that the slowing-down process may
be described by means of the Fermi age model once the first collision has occurred.
Thus at points of first collision, the neutrons enter a continuous slowing-down process.
REACTOR ANALYSIS
328
Find
[CHAP. 6
for the mean square distance traveled from the source point
Complete the analysis up to the
final step of integration over the radial distance variable.
6. If one were to consider the last scattering as a "free ride," what would be the
Note that neutrons will,
appropriate expression for the slowing-down density q(r,u)?
in general, reach lethargy u at some point r' which is different from r, and the prob
ability of a successful last flight to r must be taken into account.
(Leave this result
in integral form.)
It is desired to increase the thermal flux in the moderator region of a large
6.4
reactor in order to accommodate a certain experiment which is to be performed in
that region. This increase is to be achieved by introducing a very thin foil (of
thickness <) of fissionable material in the vicinity of the experiment.
Use the follow
ing model to examine this situation: (1) The moderator region is an infinite medium.
(2) There are no fission sources anywhere except in the foil, which may be treated as
an infinite plane source.
(3) In the absence of the foil, there is a uniform source S0
of thermal neutrons in this medium.
(4) The foil is entirely transparent to fast
neutrons, but has thermal cross sections 2^F> and 2/. (5) Neutrons from fissions in
the foil appear with zero lethargy and can be assumed to slow down according to
Fermi age theory.
a. Derive an expression for the thermal flux as a function of the distance from the
For this calculation use the kernel method and recall that if the diffusion equa
foil.
tion for thermal neutrons is written in the form
a.
an expression
Zih^h(*)
= S(x)
(P6.1)
is
-ZWttte)
then the solution
(P6.2)
is,
The cross sections in Eq. (P6.1) refer to the diffusion medium, and the source function
S(x) may be defined so as to include all the remaining neutron sources and sinks in the
medium.
The function *>(x;i') is the plane-source kernel and gives the flux at x due to
a unit source at x'.
The final result for #u,(x) is to include only the nuclear constants
of the system, So, and t.
b. Check the solution by showing that it reduces to the proper value when (1) x, the
distance from the foil, is large and (2) the foil is removed.
6.6
Consider a bare homogeneous spherical reactor with moderator and fuel cross
sections as specified in Prob. 4.8 and Prob. 5.12.
a. Find the critical radius, using Nmi/Ntae\ = 10,000.
Include the fast-leakage
correction only in so far as it affects the thermal fissions; that
neglect fast leakage
in computing the fast fissions.
Suppose that a material which has negligible absorption cross section for neu
bare critical reactor. If
trons, except at thermal,
spread uniformly throughout
the amount of this material used
such as to cause
change J2^h in the total thermalabsorption cross section, find the change Sk/k due to the change J2ih/2i".
c. Suppose, further, that the introduction of the absorber specified in part
to
be compensated for by an increase in the size of the reactor; that
more fuel-modera
tor mixture
added to the outer dimensions of the system.
Find the fractional
change Sk/k in terms of the change SR/R.
Neglect the extrapolation distance and
the effect of SR on the fast effect.
Show that by neglecting products of L*Bl and BVth
the result may be written
b
is,
L) SR
+L'fi
2fl'(rth
1
is
Sk
is
is
is
b.
329
Z,(u)
CHAPTER
THEORY
TRANSPORT
7.1
Introduction
neutrons.
lie
in differential
solid
(velocity vectors)
angle
dO
(7.1)
about
direction Q
In
Thus (r,v,(l,t)
n{x,v,a,t)
(7.2)
for these media has sufficient generality to treat a large variety of practical
Thus again we avoid questions of anisotropy and
reactor problems.
inhomogeneity, arguing that these are highly specialized situations which
are not amenable to the general methods of reactor analysis and, therefore
lie outside the scope of the present work.
We will lead up to the calculation of the neutron flux defined by (7.2)
by introducing first a somewhat more limited description of the neutron
The appropriate function
population, namely, the one-velocity model.
By omitting, for a time, the problem of describing the dis
is 0(r,Q,<).
directional
330
sec. 7.2]
TRANSPORT THEORY
331
sphere.
Section 7.5 deals with the application
of the spherical-harmonics
method developed for the one-velocity model of Sec. 7.2 to two situations
of interest to reactor analysis.
The emphasis is placed on various
aspects of the computational procedure involved in such calculations and
on the selection of suitable boundary conditions in terms of the directed
flux.
The last section is devoted to the application of the transport model
of Sec. 7.3, which includes the energy dependence of the neutron flux, to
the calculation of the age in hydrogenous mixtures.
This is the so-called
Fermi-Marshak
metal-water
method.
Computations
are
performed
for various
neutron age.
7.2
a. Neutron Balance.
We begin our study of the transport model by
introducing the one-velocity approximation.
For the analysis of this
*(r,O,0
(T,v,il,l) the
n(r,a,t)
The direction
(7.3)
332
ANALYSIS
REACTOR
[CHAP. 7
is
\f>
is
It
is
v = vQ
one-velocity
flux
(5.2)
4>(T,a,t)
related
of the
through
the
*(r,0
fa
equation
da
neutrons
Fig.
it
Number of neutrons
moving in direction a
which appear per unit
time from sources in dr
II
tions a' to a
- Ill
number of neutrons of
direction Q which are
IV
time by leakage
through the bound
aries of dt
direction
number of neutrons of
direction a which are
VI
(7.5)
TRANSPORT THEORY
sec. 7.2]
333
S(r,Q,t) dr dQ
(7.6)
The second term of (7.5) gives the rate at which neutrons arrive in the
differential element dr dQ by being scattered into the direction about Q
To determine this quantity we compute first
from all other directions.
the number of neutrons of direction Q' that are scattered into the direc
tion Q by collisions which occur in dr at time t. The integral of this
quantity over all directions Q' yields the required term. Now,
S,0(r,O',O dr dQ'
(7.7)
gives the total number of scattering collisions in dr per unit time experi
Of this number a
enced by neutrons with directions in dSi' about Q'.
moving in
from
collisions
/(Q;Q')
dQ
will
these
emerge
fraction
certain
Formally, we define the function
directions that lie in dQ about Q.
/(Q;Q') as the frequency function for the change of direction from Q' to
Generated for wjivans (University of Florida) on 2015-09-22 17:20 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Q.
Note that
-Q') dQ
[cf.
Eq. (2.11)]
_ g.
Thus the product of (7.7) and (7.8) gives the number of neutrons initially
directed along Q' which are scattered into the direction about Q while
occupying the volume dr. The integral of this product over all Q' gives
For
the total number of such contributions to dr dQ from all directions.
the second term of (7.5) we have, then,
2>(r,Q',<)
(7.9)
BEACTOR ANALYSIS
334
[CHAP. 7
n(x,y,z,a,t) da(v k) dx dy
where k is the unit vector in the z direction.
(7.10)
x(V
Fig.
7.2
Volume element dx dy
If dz
is taken sufficiently
the
k) dx dy
to n(x,y,z + dz,Q,t)
is
n(x,y,z + dz,a,t)
the net loss from
through the dx dy faces is
Therefore
n(x,y,z,a,t) dx dy
~ n(x,y,z,a,t)
+ n(x,y,z,Cl,t)
dz
dz dSi (v
k) =
0(r,Q,/)(Q
k) dx da
(7.11)
Analogous
where we have used (7.4) and the usual definition for the flux.
considerations for the other two faces lead to the following result: The
sec. 7.2]
TRANSPORT THEORY
335
4>(r,a,t)(a
J k)
*(r,>0(o
Kr,o,0(o
j)
j-x
da
(7.12)
is
is
2a(r,a,t) dQdr
(7.13)
is
Term
gives the rate at which neutrons are scattered out of dr dQ into
new directions.
This term may be represented by the form (7.7) with
Q' replaced by Q.
The last term VI
the time-rate of change of the neutron density in
This may
jt
be
written
n(i,Q,t) dQdr
=-vjt (r,a,t)
(7.14)
V<Kr,O,0 + 2<*(r,0,/)
= -S(r,O,0
*(r,0,<) + Q
dQ dr
f(Q;Q') dQ'
(7.15)*
a
where we have combined the losses from scattering and absorption into
single term, using the total cross section 2(, and have canceled out the
This integrodifferential equation for the neutron
common factor di dQ.
flux
In the general
usually called the one-velocity transport equation.
is
is
is
it
spherical harmonics.
it A
it
b. Frequency
di dQ.
REACTOR ANALYSIS
336
[CHAP. 7
Thus if
face of a sphere drawn about the collision point (see Fig. 7.3).
/(Q;Q')
dl
is the
the
of
the
then
the collision occurs at
center
sphere,
probability that the vector v passes through the area dCX on its surface.
x
Direction of
incident neutron
Unit sphere
Fig.
7.3
a frequency
/o/(Q;Q')dQ
is homogeneous
/ can
cos
If
(7.16)
in question
we introduce
0o
no
il
Q'
a neutron
moving
in
direction^
Q', when scattered, emerges with a new direction I (7.17)
(probability
whose cosine lies in duo about /i0
/(Q;Q') -gjUGi.jQ')
(7.18)*
TRANSPORT THEORY
SEC. 7.2]
337
defined in (7.17).
Since this function is defined by means of the single
variable /io, a suitable series representation can be given in terms of the
Legendre polynomials1 P,(no). Thus
PG.0
;0')
>
(7. 19)
^Pi|(mo)
,-0
J*t
The coefficients l), are obtained in the usual way by applying the orthogo
nality properties of the Legendre polynomials.
It follows that
U, =
PnM
(mo;Q')
(7.20)
duo
6.
^.
\i
is
0'
P\.
i
functions
\f/'
= cos
denote the arguments of the direction Q', and let
completely defined by the quantities
and
The
n, \f/',
Legendre polynomial P(mo) may be written in terms of these variables
by the application of the addition theorem2 which involves the associated
Let
and
Clearly, ^0
JL
used in
(7-21)
P,(M)
(7.22)
<
/3
and
?7
The associated
are related to those with positive index
0
>
/3
rj
- sin"
f*()
P?(m'} e*(*~*
P?(/i)
j^TWi
< j;.
P$()
(01
ps
=0
23)
Note that
(7.24)
for
>
(7.25)
P(M)
1)0
/3
and
-q
/8
p-e =
3(i
PJ(m) = 3M(1
P(M) = *(3m2
PrH**)
m2)*
Pr'GO
=
=
PJ(m) =
-Ml
-
i(l
m2)
M2)
-2Pr'(M)
(1
1)
PJGO
(7.26)
m2)1
Refer to Eq. (4.21) and the discussion for definition of the lower-degree poly'
nomials.
A. Sommerfeld, "Partial Differential Equations in Physics," p. 134, Academic
Press, Inc., New York, 1949.
Ibid., pp. 126-129.
'
P,M
REACTOR ANALYSIS
338
[CHAP. 7
/(0;Q')
JL
(7.27)
cal harmonics.
be the
orthonormal set:
1,
and
rr(O)
The
- H;P;(j*)er**
(7.28)
(-l)-F--(H)
and negative
Ran +
[
lKn-
47r(n
conjugate of this
m.
(7.29)
The coefficients
//"
m)!!*
+ m)!
Note that
=
(n+jn)!
(n
to)!
fQ
Yl'(Ck)
Y:(Q) da =
rf/i
f* Yl'in,*)
is stated
YZ(nM df
5i5"n
(7.32)
The symbol 8 denotes the Kronecker delta and is zero if n ^ a and unity
if n = ex.
The flux and the source may be expanded in terms of the functions
Thus
(7.28).
K(T,t)YnW
(7.33)
^2
S?(r,0F:(Q)
(7.34)
*(r,Q,0 =
n =0 m = n
S(r,O,0
n 0 m n
TRANSPORT THEORT
SEC. 7.2]
339
(r,0
fa
*(r,Q,0
- /Q S(r,O,0
ST(r,0
F:(Q) da
(7.35)
F*(a) da
(7.36)
/(OjC)
,_o 0- -n
W)
^F?(Q)
(7-37)
expansions.
yields
f"(q)
n,m
be reduced to a simpler
(r,0
2,
2.j,n)
+Q
C(r,0
- ST(r,0
^(O)
The substitu
2.
form; thus
rS(Q)
(7-38)
tj
(J
F"(o,)
n,m
c(r'
ij
s"(T't}
n,m
n,m
2>
c(r' r"(Q)
v +
2')
(7. 39)
*2(r,<)
G jt
2,^(r,<)
2.t,a<tf
(r,0
- Si(r,l)
(7.40)
340
ANALYSIS
REACTOR
[CHAP. 7
be evaluated
dx
cos
dy
sin
sin
j
e'*(m_U sin
("
gO^iOO
*r(r'
(7l42)
- w)PT+i(f)
43)
is
g)"|*
^_
+1)
to
^
%+1
For
- - l)(m
(n
(2n+l)sine[(w
^-
3)(2a
these calculations
gK
(2a
!"(+!L
/.*<-*(=<)
n,m
it
If
("
mP(m)
)P^z{(m)]
(7.45)
it
if
are
ew
(7
Tz)
- ^^iM
n(fl)
YjH? \[(n+
n,m
m^-^
n,m
functions of
v(r,<)
fl)
r:(o) a
(7.41)
dz
4>.
\[<
d
V = ii + sin
\f>
Ibid., p.
132.
TRANSPORT THEORY
sec. 7.2]
341
'
*fcl(r,0
(7.46)*
...
neutrons
in situations
of interest
to reactor analysis
are presented in
Sec. 7.5.
Diffusion Approximation.
It will
assumption
\,
it
it
is
all
0,
be consistent, we
so that
<S
2i(r,<)
REACTOR ANALYSIS
342
[CHAP. 7
(?-48)
!)
(7-47)
<
(I
s<)
<5)
*
*?
-^(S-'5)* + ?i(
G
S<)
-(^ +
**
*s
s)
'
(7-50)
<*mo
=
=
^(";Q') P'(")
>0o;o0
^mo
(Mo;0')/odMo
(7.51)
|\
j(mo;0') PoCmo)
7o
/_\ /_\
These are
(7.20).
/-i
(7.52)
it
0,
where
the solutions
(7'54)
<7'55)
'--v^Ua^-'i)*8
=
3(2,
(7.56)*
Xnn)
is
it
the
We call this quantity the diffusion coefficient and recognize that
function previously given by Eq. (5.44). The substitution of the results
single second-order partial dif
(7.53) through (7.55) into (7.47) yields
we apply also the definition (7.56), then for
ferential equation for
If
tj
(7-49)
"
(I
s^> =
**
2l)
"
(i
-(?i + *) + **-:?ss*
TRANSPORT THEORY
sec. 7.2]
343
equation is found to be
-V-^
-lla-TyDTy*S-izD+=
<Sjj
(7.57)
may be replaced by the "total" quantities <(r)
and
inasmuch as each represents the isotropic component of their
We compute, therefore, the total functions <p(i)
respective functions.
For this computation we
and <S(r) from relations of the form (7.5).
require appropriate expansions for the steady-state functions; thus, the
relations (7.33) and (7.34) are replaced by the set
The functions
*(r,Q)
5(r,Q)
n ^
<S(r),
and
W
(7.58)
SSW y?(0)
speed
regardless of
S" (r)
S(t,Q) dQ
Y,
S(r) =
<ftJS|
(7. 59)
produced by sources
(1)
is
+(r,a) da
+(r)
(7-60)
-V DV(r)
+ 20(r)
S(t)
(7.61)*
which in the case of a uniform medium reduces to the usual form (5.22).
e. Net Current.
The net current relation (5.19) introduced in the
development of the diffusion theory may be derived from the first-order
solution to the transport equation.
following definition:
we introduce
is
the
(7.62)
is
J.(i)
be
computed
from
the neutron
density
n(r,il)
di,
is
v,
n,m
344
ANALYSIS
REACTOR
[CHAP. 7
pass through the area dA shown in Fig. 7.4 per unit time is given by
n(r,Q) da](a
[v
dA = <Kr,0)(ti
s)
s)
dA da
(7.63)
dA
dA
f
JO
{s
0(r,O)(O
s)
da
*(r,0)0
(7.64)
da]]
is
[fa
J,(r) dA
of the
J(r)
Neutron veloc
7.4
ity and the surface dA.
(7.65)
Fig.
>(r,Q)a da
(7.66)*
sin
(7.67)
sin
+
k
cos
sin
cos
expansion
for
4>(r,Q)
j(r)
fa
7.3)
given in (7.58)
a
(see Fig.
is
The integral
be
direction
J(r)
YZ(a)a da
(7.68)
FT.T71
f;(q)
[Y\W
If
sin
sin ecos^ =
sin
(7.69)
tzJjilYlM
YT\a)]
+
cos
We note that
(7.70)
Yl\a)]
(7.71)
J(r)
it
these relations
J. {!)
form
first-order
27/1
MM
estimate
*rx(r)]j
Mtt
be
+ *rl(r)]k
obtained
(7.72)
by using
sec. 7.2]
TRANSPORT THEORY
345
J(r)
- D V*
[i
yx *8(r) +
J(r)
yy
4>\,
*8(r) + k
fc1, and
*g(r)
The
<}.
If
-DV*(r)
.
M
-y
n-0
(7-?4)
(2-!L^)
jlx
compute
4>{x,n,t)
Pn{n) dn
(7.75)
These coefficients are related to the quantities %(x,t) used in the general
expansion of the function (i,Q,t) [see Eq. (7.35)].
spherical-harmonics
is
following manner:
REACTOR ANALYSIS
346
This
[CHAP. 7
relationship
demonstrated.
easily
(x,a,t) #
4>(x,n,t)
(*,) HnP? 00
n,m
e**
2t
fQ2'
*!(*.0 W-GO
(7.76)
n=0
it
is
if'
follows that
(7-77)
<Pn=i
set of coupled differential
is
neutrons
n-0
\p
2,
integration
("8)
of (7.39) yields
- 2.A tf(*,0 -
HIPM
0|
then
it
n-v-n
(7.79)
S(x,a,t)
2t
S(x,m,0 ^
is
j2J
sn(x,t)
ffJP.GO
n-0
and
.Sn
30
(2^j)
n-0
P"^}
An appropriate
and
<.
(7-80)
TRANSPORT THEORY
sec. 7.2]
347
n-0
(sir)
"Fn(M) P(M)
(7'81)
/"I
*2
(am)
I)
2<
*-)
(lit
0,
is
mP.(m)
nPn-M
1)Pn+lM
(7.82)
(2a
l)(2a +
(stt)
2g_,
ax
0a-1 ~
3)(2a+
ax
(asrr)
-^
1)
(2 +
HM
2(
1)
21
2-')
- -37
of
<fc,(x,0
~ 2,0o(x,<) + (*,<) +
2t)a</)a(x,0
....
is
is
1
;
is
ANALYSIS
REACTOR
348
The appropriate
[CHAP. 7
<S
= 0
I L *o(x)
+ (Z' ~
(7.85)
= 0
(7-86)
-D ^
>(*)
+ 2a*o(x)
S0(x)
(7.87)
*,(*)
-D^*o(i)
(7.88)
where we have used (7.56) and assumed that the medium is homogeneous.
Now, from (7.75),
=
i(x)
f1,,
- (*) - oftotal
*.
track length
al[ neutronsg
Mixdn
_ .
(7.89)
(7.90)
is
it
i)
has
in
a
with
x
direction
direction whose angle
tion)
cosine in dn about
is
(diffusion)
*(*,/) =
J(x)
[Ho(x)
(x,n)n
dn +
(x,ri(i
(7.92)
approximation
iMx)
Jt
J{x) = *,(*)
j*
if
fi
<b(x
'
|m*i(x)
(7.93)*
we obtain
i*i(x)]
(7.94)
i_(x) =
J*
/_\
*.(*)
4>{x,n)n dn =
j~l
*(x,p)MdM
\,{x)
=
\4*(x)
Ifaix)
- Mx)
(7.95)
(7.96)
sec. 7.2]
If
TRANSPORT THEORY
we note that
349
D',
D'0
could not be satisfied in detail when used in conjunction with the dif
fusion theory.
In addition, it was mentioned that these conditions could
=0
ix
< 0
(7.97)*
sion for the directed flux must be obtained from a truncated set of
equations based on (7.84), and this solution will necessarily involve a
finite number of arbitrary constants.
Thus it is not possible to satisfy
In
difficulty is circumvented by
actual
this
rigorously.
practice
(7.97)
introducing an alternate
5,
(7.98)
it
is
We note immediately
3,
m =
1,
dM
(0,riPm(n)
It
is
j_x
is
currents from certain directions are still possible, provided of course that
the gains from these be exactly balanced by flows outward from other
This condition on the inward current
precisely the state
directions.
is
I,
theory relation.
1
REACTOR ANALYSIS
350
[CHAP. 7
The rigorous statement for the interface condition between two dis
similar media is that the directed flux for each specified angle be contin
If
II,
and
0,(R,a) = <MR,G)
for all Q
(7.100)*
function of
a single space
<fr(x,/0 =
-1
u(x0!li)
<
<
(7.101)
A numerical example of
x0 denotes the position of the interface.
the application of this boundary condition for the P3 approximation1 is
given in Sec. 7.5.
h. Time-dependent Equation in First-order Approximation.
We
conclude the present section with a discussion of the approximations
where
involved in deriving
partial differential
equation
+ So(x)
*!(*,<)
(7.102)
It has become customary to refer to the order of an approximation in the sphericalharmonics technique by indicating the highest-order harmonic retained from the
infinite set of coupled equations.
1
TRANSPORT THEORY
sec. 7.2]
351
+ 3ZD\
(\
3D
6"
;**^ + ir^*(^
Za*(x,<)
Z>
0(1,0
(7.104)
= 4>{x,t)
where we have used (7.56) for D, along with the fact that <o(x, = 4>{x,t)
In treating specific problems
and 5o(x) = 5(x) [see, for example, (7.75)].
with this equation, it may be necessary to use Eqs. (7.102) and (7.103)
to evaluate the initial and/or boundary conditions.
It is of interest to compare Eq. (7.104) with the general form (5.21)
which
had been
which have resulted in the omission of the two features noted above.
Some of these are considered in the discussion which follows.
It should
be recognized, however, that the adequacy of the first-order formulation
(7.104) can be determined in the final analysis only by comparison with
experiment since the error involved in the omission of the various higherorder terms is not evident, a priori.
Before discussing the discrepancies between Eqs. (7.104) and (5.21)
be in order to comment further on the nature of Eq. (7.104).
This equation in its general form is the so-called telegraphist's equation.1
it would
[chap.
REACTOR ANALYSIS
352
[l
as
-v
32a(vD)
|2 +
(vD)
-(Mn
(vD)
(7.105)
reduces to'
Tr
dt
- Zwi + Dv
d.r2
a better one-
7.3
present model only the most elementary physical features of the processes
involved.
results of this analysis are the two simplifying assumptions listed below:
(1)
(2)
(7.106)
been
TRANSPORT THEORY
sec. 7.3]
(2)
is to
353
As
and spatial
- |v
ANALYSIS
REACTOR
354
at
(T,v,a,t) + a
For
V0(r,i,Q,O +
[CHAP. 7
these we can
3S,(i>)
write
- S{i,v,a,t) j dr
*(rf,O,0
dv
da
(7.107)
The first term gives the time rate of change of the neutrons which occupy
the space dr dv dCk, the second, the losses due to the net transport of
neutrons through the boundaries of dr, the third, the losses from this
due to neutrons heing absorbed or scattered ojrj^or_^slowed out"
(of the dv interval), and the fourth, the contributions from sources.
Now for the scattering-integral term. The total number of neutrons
scattered into dr dv da per unit time from all possible speeds v' and
space
a'
2.(y') 4>(r,v',a',t)
and multiplying
di
dv'
da'
scattering
collision
in
probability that
with an initial
of motion a', when
= scattered, emerges from the collision with
a speed in dv about v and direction of
speed
f(v,a;v',a')
dv
da
a neutron
motion in
(7.108)
da about a
The product of the total scatterings and this probability integrated over
all accessible v' and a' gives the total gain in the neutron population of
element dr dv da due to the neutrons scattered in from
and
different directions of motion :
higher speeds
/,<
the differential
Q>
*(r/>V)
2-(y')
f(v,a;v',a') di
dv'
da'
dv
da
(7.109)
is
is
is
4>(r,v',a',t)
S(i,v,a,i)
2,(t/) f(v,a;v',a')
dv'
da'
(7.110)*
where v' = v'a' and denotes the velocity vector of the incident neutron.
Equation (7.110)
the Boltzmann (transport) relation which describes
This relation
the distribution of neutrons in space, velocity, and time.
is
dr dv'
TRANSPORT THEORY
sec. 7.3]
355
is supplied.
(3)
(7.111)
be
where
-O
mo
= cos
mm'
VI
fl0
cos
M2
Vl -
m" cos
d8 d$ = \dn d\f/\.
(<A
(7.112)
^')
(7.113)
before
in dv about
= scattering
is
f(v,a;v',Q') dv dn d+ = f(vlfi,t;v') dv dp d+
probability that, given a neutron with velocity vector
v'
The velocity vector v' used here to represent the condition on the proba
and
bility, namely, that the initial coordinates of the neutron be
For the present calculation, scattering collisions are assumed to be
azimuthally symmetric about Si'. That
equally probable that
the final vector O will lie in any position on the conical surface which has
is
it
is,
p!
v',
is
REACTOR ANALYSIS
356
[CHAP. 7
Q' as its axis and a vertex angle 60 (see Fig. 7.5). Thus the final direc
tion of motion will depend only on the scattering angle between the initial
We define, therefore, the function [cf. Eq. (7.17)]
and final directions.
q(y,/i0;v')
dv duo = is
dv dfio about
and
/by
[cf.
This function
is related to
Eq. (7.18)]
f(v,n,f;v')
We will
(7.114)
fio
use the
function
q(p,Mo;v')
(7.115)
in the
henceforth
integral of (7.110).
d. Expansion in Spherical Harmonics.
The
= S(r,v,Q,t)
2,00
(T,v,a,t) + a v(r,v,a,t)
(i,v',a',l) Z.(i/)
qO,Mo;v')
dv' <V
#'
(7.116)
with the aid of (7.115). As this equation now stands, further calculation
is not possible until some additional information is supplied on the
functions
For convenience, we assume that all these func
S, and q.
tions can be expanded in infinite series of spherical harmonics.
On this
basis we may write for the frequency function q
q(i;,Mo;v')
- (- ^)
P'M
(7-117)
<ko
(7.118)
q,0>;v') =
/_1
6',
8,
is
it
is
it
is
~jt(r,v,Q,t) +
TRANSPORT THEORY
SEC. 7.3]
q(y,Mo;V)
357
2r
be shown
y$(fl)
q,(t>;v')
Ff(O')
0(r,t-,Q,O =
that
X ^(^'O F-(Q)
(7.119)
set
F"(ii)
take
*rOV.O
(7-120)
n 0 m n
oo
n
S(r,A0
(7-121)
n
n = 0 m
with
*r(r,,o =
*(r,A0 do
/0 rr(o)
(7.122)
S(r,v,a,t) da
(7.123)
- /Q Y*(a)
<S|T(r,,0
,s':(r'M)
"
n,m
-"0
f?(q)
X
/
ij
q,(!,;v,) dv'
tj
*r(r'p,'
n,m
f:(q)
n,m
/S
c(r,u'0 y"(n)
2,(0
2,w]
(7-124)
n,m
qn(v;v')
dv'
"
and
<S.
(r/,0
(r,,0
- S5(r,ffl
Z.(i>') q(o;v')
(7.125)
The procedure
is
Z,(iO
v
+
-t
is
m n
Y*0h4>)
dt>j
[iwt
358
ANALYSIS
REACTOR
[CHAP. 7
(2a +
+ l+j8)
(2a +
1)
1)
|
.
^+^r'V^|
1)1*
1)
/3
^l(r'M)
+
2)(a +
(2a +
0
1)
3)
/3
(7.126)*
1)
1)
(a
<fc\(T,v,t)
*)(<
1^(^,0
+ l)1*
3)(2a+
j3)(a T(a
(2 + 1)(2,
(2a
D(2
(2a
.
/3
1,
0,
< a.
and a <
e. Isotropic Scattering in Center-of-mass System.
Again, in order to
make further progress in this development, we require additional infor
where a =
q.
it
tion
Before introducing this information, however,
to transform the relevant functions from the speed variable
is
is
is
convenient
(7.127)
./
%(u,n0y)J
where the symbol
to the leth
We make this transformation in the case of the func
argy variable u.
tion
new function
such that
by denning
du
(7.128)
dv
it
dv duo
is
if
(,Mo,V)
function
(m,mo;m',Q')
(7.126) we require
In order to
v,
2)(-0
(2a +
I"
d\
_ -(-0
/3l)(2a +
l)(2a
/})(a
(a
(2a
3*)
{dy
(2a +
/3)
(a-/3)(a
+ +
1)
(a
(r/,0
Jp'
(a
at
|S)
3)
n-0
^--^(u;<Q')
PM
(7.129)
TRANSPORT THEORY
sec. 7.3]
359
/_\
where we have replaced the statement of the initial velocity v' by the
initial direction of motion Q' and the initial lethargy u'. The coefficients
8 are computed from the relation
=
(;',')
(m,mo;m',Q')
Pn(no) duo
(7.130)
is
- u(mo;u,u',0')
(u,mo;u',o')
fl(w;u',a')
(7.131)
where
given a neutron with
coordinates u' and Q' before collision and
lethargy u after collision, the neutron has
been scattered through angle cos-1 no in (L)
(7.132)
neutron with
Q', the neutron
scattered into lethargy interval du about u
(7.133)
= probability
that,
It
that,
u'
given
and
is
X)
/n0
can
is
f)
neutron,
occur.
E'
(4.18)
1)
+ 2A +
Ay
is
where
tj
=
(1
'
is
we transform
2A,
fio
*[(A +
l)e<'-"
is
(4-27)
A*)*
to u and eliminate
(1
(A
77
If
'M
"
tj
The cosine
scattering angle in the center-of-mass (C) system.
related
cosine no for the (L) system through the equation
to the corresponding
is
from (4.27) by
- l)e-*]
(7.134)
t)
g.
be
written
is
B(m;u',Q') du = probability
coordinates
is
dm,
,U')
Xf(p0;u,u'
a delta
REACTOR ANALYSIS
360
[CHAP. 7
- f)
(7.135)
then the substitution of this result, along with the application of (7.131),
into (7.130) yields
.(;u',0')
fl(;u',0')
8(Mo
- f) P-(mo)
d^o
= 9(;m',0')
(7.136)
This result
(4)
Eq. (4.50)]
fl(u;u',Q') ^
fT-
u><u + \n (? 13g)
= 0
+ \n
(? 13g)
elsewhere
P CfVu'-u
1
(7-139)
Thus these functions relate the initial and final lethargies of the scattered
neutron with the cosine of the scattering angle in the (L) system.
We will
use these results in the analysis which follows.
ing.
\-
TRANSPORT THEORY
SEC. 7.3]
361
(T,v,t)
(7.140)
\dv\
*J(r,u,0
Jui
2,()]
il)
^*?(r'U'
(7-143)
(I
(7-142)
- ~ + 2,()1 ^(r.M.i)
(?-141)
4>Tl(T,u',t)
5)
*!(r-"'<)
li)
^(r^t)
--^( +
*(r'M'
(7-144)
i =
r
a
(7-145)
ra
o =
is is
is
theory,
can be obtained
we impose an
is
+ S,(u)
\jt
^3^0;(r'M'-^(^-^)*rI(r'M'
(fy
0(T,u,t) + S0(T,u,t) +
S,(m)
\jt
ANALYSIS
362
REACTOR
[CHAP. 7
(5) Collision densities 2,(u')%(r,u',t) are slowly varying functions over the lethargy range of the scattering integral
In this
2,(')2(r,M',i) in
point u.
S.(w') *8(r,tt',0
=a
S.(m') tf(t,u',t)
~ Z.(u)
...
- ) ^ [S.(u)
*g(r,,0]
*?(r,u,t)
(7.147)
(7.148)
If these approximations
through
elementary
=-i /
*8(r,u',f)2.(w')e'-du'
_ S.(M) dj(r,,Q
I
+ r^
|-
eu,_du,
Ju+\na
['
[S.(u) 8(r,,0]
- )e-'-" du'
('
(7.149)
^Mu)tf(T,U,t)
/^/u
lna
V=~a
f"+i
r^~
Ju+In
e"'-du'
(u'
- )e' du'
a >+lna
lna
dA
afu
it
-{
(7.151)
(7.152)
TRANSPORT THEORY
sec. 7.3]
363
where we have used the definition (4.42) for and (4.30) for no- The
f which appears in (7.153) is the cosine of the scattering angle
in (C) which corresponds to the lethargy change from u' to u.
function
I)
(7-157)
{ry
(7'154)
(7-155)
**
S')
it
^ =~^
*!
-4nk-
(7160)
')*'
- 8[*fr0 -
(7-161)
be
-V
DV
4>{t,u),
(7.162)
which
is
SJ
-2
1-
This
0 it
REACTOR ANALYSIS
364
by the equation
related to
*(r,)
The substitution of
2 is independent
[cf.
and also
these
[CHAP. 7
functions into
^fi-
(7.163)
of r,
-D()V*(r,) + Z.()
4>{t,u)
S(t,u)
I^
[2.(u) *(r,u)]
(7.164)*
As will
be demonstrated
In order to combine
fact the slowing-down density at thermal energy.
the description of the fast-flux distribution given by (7.164) with an
equation for the thermal flux, we require, therefore, an expression which
couples the fast flux with the slowing-down density.
We derive here an expression for the slowing-down density which is
based on the assumption that the lethargy values in question are far
Let
g(r,i<) =
(7.165)
point r
This number
q(T,u)
f"
S.(u')
4>(r,u')
- -)
du'
(7. 166)
^
If
S.(u')
(T,u') ~ 2.(u)
4>(t,u)
{u'
-u)^-
[2.() *(r,u)]
TRANSPORT THEORY
sec. 7.3]
3f)5
j(r,tt)
?2.(u) *(r,) + a
where
A [2.()
- a) -
0(r,)]
density
(7.167)*
ln* "
2(1
(7.168)
- D{u) VV(r)U)
+ 2.(u) *(r,)
S(i,u)
- AhV2tb(r)
A [2.(w)
^ [2.(u)
+ 2^lh(r)
4>(t,u)]
(r,w)]
(7.164)
(7.167)
= ?(r,Mth)
(7.169)
{r,u) +
7[20(w)
{t,u)
y =
where
- D(u)^(t,u) - S(r,u)]
(7.170)
(7.171)*
The first term on the right-hand side of (7.170) is the predominant one
For very small reactors, the second term
for reactors of reasonable size.
can be of comparable magnitude because of the large leakage (represented
here by the Laplacian term).
The number y is of the order of and
ranges from zero to unity (see Fig. 4.18).
2,(m) 4>(r,u)
A first-order
(7.172)
REACTOR ANALYSIS
366
[chap.
-D{u)V*4>{i,u)
+ 2(w) *(r,)
S(t,u)
qit'u)
which is Eq. (6.5). The source is included in this relation since it may
very well be of comparable magnitude to the other terms. It could be
argued that its omission from (7.170) was permissible only because it
represents a correction on a correction (the y term).
A second-order approximation to (7.170) for large reactors is obtained
by neglecting the leakage term only:
q(r,u)
=a
if
q(r,u) =*teZ,(tt)
which is to
we use the
- yS(T,u)
approximation y
(7.173)
*(r,tt)V- &S(r,ti)
then
(7.174)
with (7f164).
A set of equations which has proved to be very useful in reactor cal
culations, wherein the Fermi age inodel is applicable, is
be used in conjunction^
q(T,u)
=\J(w)
(r,u)J
=
5(r,u)
- ^ g(r,)
(6.6)
(6.5)
which are the equations used in Chap. 6. These relations hold if 2(?i) <3C
2,(u), and u is much larger than the values corresponding to the bulk of
Ke"fission neutrons.
be developed by
of integral equations.
These methods are generally quite powerful, and
formal solutions are readily derived even for nonhomogeneous systems
with complex source distributions. Unfortunately, the mathematical
techniques required for these applications are of a somewhat sophisticated
TRANSPORT THEORY
sec. 7.4]
nature;
moreover,
simple geometry,
367
reasonable effort.
presented in Chap.
8.
As an
Neutron Distribution in Media with Zero Cross Section.
introductory calculation we derive the neutron-flux distribution in a
system which has zero cross section; media with more complex neutron
a.
supplied from
that
then,
follows,
[cf. Eqs. (7.5) and (7.15)]
balanced
be
It
V*(r,a)
S(t,Q)
(7.175)
(7.176)
= 8(xja)
(7.177)
V =
*(r,Q)
Integration over the variable
*(r,Q)
The function
4>0(r,Q)
ft
it will
be
S(r,Q)
ds
Q
and
yields
v>(r,Q)
arbitrarily taken
as zero
(r,Q)
(7.178)
= 0
(7.179)
REACTOR ANALYSIS
368
more convenient
[CHAP. 7
Jj S(t -
4>(r,Q) =
RQ,Q) dR
(7.180)*
The solution (7.180) states that the total directed flux 4>(r,a) is the sum
total of contributions from all sources which lie on
the segment of a line through r parallel to Q for which
0 < R < <* .
Clearly, sources off this line cannot
contribute to the flux at r (in direction a), nor can
< 0.
It
a')
a'
= o
(7.181)
g(a)
(7.182)
ja ff(O')
Note that the function
5{a
a'
') da'
**(Q
Eq. (7.135)]
&{a-a'
1)
S(no
(7.183)
1)
a'
1)
dQ =
1^
8(mo
1) dn0
i(q a') =
^-
equation
Zir
s(a a'
= 2ir
(7.184)
l)
8(0
1)
fa
|q2'
We have used here our earlier notation Q a' = cos 0O mo- If (7.183)
is integrated over all directions (i.e., over all possible orientations of a'
to a, see Fig. 7.1), we obtain
(7.185)
TRANSPORT THEORY
SEC. 7.4]
the identity
S(r,Q) =
jj
dR
S(t
=
I"
RQ',Q')
la
Ra'
be
a') da'
dp =
R*dRda'
(7.188)
R =
|r
is
extends over
This result may be used to compute the Mai track length (r), which
defined by Eq. (7.59).
The substitution of (7.190) into (7.59) yields
da
Jv~
|r
- r'|
(7.191)
r0)
(7.192)
is
a
<S'0located
V|r-r,y
r0|-
(?
*(r,)
flux
|r
-JTs(< -'i)><(ai)
(7.189)
iww
(7187)
by the relations
and
52(Q
s(r-e>i)8ia-i)dRda'
r' = r
(7.186)
written
be
a') da'
|p|
(r,Q)
2(Q
r'|
S(r,a')
369
m)
REACTOR ANALYSIS
370
[CHAP. 7
In
|r
+(r)
rl|)
(|r
by S0A/\r
(7.194)
r,|
at
is
precisely
r,
statement
+w
of
the
neutron-balance
a V*(r,Q) +
S(r) *(r.O)
S(r.O)
(7.195)
is
which states that the only loss mechanism for neutrons from unit volume
and unit solid angle about Q
by transport (Q V<) and
around
is
is
it
"
/3,
2o(r0
- wO) dw
(7.196)
0,
(7.197)
s
r|
|r.
The variable
in Eq. (7.176).
The
condition.
30
=
dd 3/3
(7.198)
TRANSPORT THEORY
fe
Z.(r
|r
SEC. 7.4]
371
- r|Q) - 2a(r)
(7.199)
e?
op
+ 2a*
|t
If
we note
that
/J""^'
'
y0
S(r0
=
f"
with
r'
S(r
R.
If
gives
- wQ)
- wQ) dw
S,,(ro
2(r0
u>a) du>
we use
= /3(r
- wO)
JqR
/"'r-r|
y|ro
0'
=
r|
= /3(r';r0)
definition for
use the
0
We
0'
- R;r)
2(r
rfw
(7.202)
- wO) dw
= /3(r';r)
(7.203)
(7.204)
dfi' = 2a de' = S dR
we obtain,
finally
[cf.
Eq. (7.180)]
(r,Q)
fQ"
S(t
- R,Q)e-*fr-R^ dfl
(7.205)*
(7.180).
The expression for the total flux <(r)
the result (7.205).
We first transform
This
is
Thus the expression for the directed flux in a purely absorbing medium
may be obtained from the result for the medium with zero cross section
by merely introducing the attenuation factor e~& into the integrand of
is
where
Ta
Integration over
is
which
p"
{+eB)
bp
372
ANALYSIS
HEACTOB
[CHAP. 7
*>
f^T-T"
<7-207)
It must be remembered that the Q' which appears in (7.206) and (7.207)
is in fact the direction given by R/R.
We demonstrate the use of these results by computing 4>(t,Q) and
#(r) for the case of a point source. As an example let us take an isotropic
source of strength So at the origin; then,
S(r,a) = - 505(r)
(7.208)
<Kr,0) =
~-t
52
(fl
e-WV
(7.209)
*(r)
to be expected.
The special case of
- ^b*-
is
(7-210)
is
homogeneous mgdkras
Za|r'
(7.211)*
r|
tance.
is
as
S(,
- R,Q)e-*. <r -
s-^i h^isgp^
/;
For the
(r)
and
'
(7.212)
r>|a
(7.213)
r/r)rv
^T-
(7-215)
is
is
sec. 7.4]
TRANSPORT THEORY
373
average escape
neutron sources in V
(7.216)*
of volume
Fio.
7.8
subtended by the area dA from the point r, then rftl/4ir is the fraction of
neutrons emitted by the source which is headed in the direction of dA.
Now suppose V contains a homogeneous material of absorption cross
section 2, and let the distance of dA from r be s, then exp ( Zs) is the
a neutron
(7.217)
In
sources throughout V,
(7.218)
REACTOR ANALYSIS
374
,iA:
be
To compute
gration.
ment dr we picture the volume V as
consisting of a series of tubes of cross
section (Q n<) dA and length s0, as shown
[CHAP. 7
=
n;
that
n). Then the volume element may be written
,,
dr =
(a-
n.) dA ds
(7.219)
or
P'"
47F fA
ja
4irVza
fo
L (1
e~*"
"
'
n<) dS
'
e~z'^a
da dA
*)
da dA
(?-22)
probability from
a volume
the average flux in the
volume were it to be immersed in an infinite medium supporting a uniform
This relationship has considerable practical importance;
isotropic flux.
escape
To demon
moreover, the two problems are analytically equivalent.
strate this equivalence, consider a body V of the shape and material
assumed in the calculation of the escape probability to be located in an
isotropic bath of neutrons such that at great distances from V the total
track length is 4>x. Then the number of neutrons incident upon the
body per unit time per unit solid angle from direction Qo through the
surface element dA is
ndA
(7.221)
Now if >(r,Q) is the directed flux at r in V, then the total track length at
r is given by (7.59). Further let >(r,Q;r0,Qo) be the directed flux at r
due to a unit source at r0 radiating neutrons in direction Q0, and let
Then
(r0,Qo) be the source strength.
*(r)
If we
j[n4>(x,a)da
a
f
jIda
a
J u
0(r,Q;r,ao)
S(r0,Oo)
dQ0
(7.222)
*(r)
This result may
=
be
t5
/ dQ Ja,
/ Qo-n 0(r,Q^o,Qo) dQtdA
"wr
Ja
(7.223)
TRANSPORT THEORY
sec. 7.4]
375
Q.
Therefore
=
0(r,a;ro,Qo)
*(r0,-O0^,-Q)
(7.224)
QB'n (T0,-a0;r,
da
t2
-a)
da0d A
(7.225)
Q and
Go,
space.
then
If
(r)/4>M
escape through A
p(r)
identified
; thus,
as the
fraction of
(7.226)
(7.227)*
taining
dX
ds is
f
1
dA
n,
da
ANALYSIS
REACTOR
376
[CHAP. 7
where the directions Q over which the integration is performed are selected
If this relation is divided by the total
such that s < s' < s + ds.
number of chords in V, using the cosine weighting mentioned above for
each direction
and s + ds in
h(s) ds =
ds of
dA f
. Ja=ow
/ dA
da
n,
la- n, da
(7.228)
The denominator
JQa-
dA
f*
dn
h(s) ds
JA
(7.229)
(7.230)
Ja-ait)
'
dA
J,I
a
itA
rA
Thus, for
n-
da
(7.231)
ttA
Jq
da
sa-n,dA
JQ
da
wA
JI
(7.232)*
3"W
e_Z"*)
("
"
dA f'm"
(1
irPs; fAdA
PK =
fa(1
computing
n>)
- e-z-*)(Q-n,)dO
(7.233)
it
s,
4tV
h(s) ds =
dA
a nida
The application
fA
through
s
da
n,
dA
j^s
fA
f**
d\p
da = dn
sec.
TRANSPORT THEORY
7.4]
377
P.-
(1
and
(1
Q-QW
Q-TLidQ
. <r*.')A() ds
(7.234)*
*w*.-^
f
]AdA
fn+d*
swjQ
his) ds =
or
,dty'
~ 2fi
ds
8*
dii
(7.235)
r2,(r) + S.(r)
2,(r)
(7.236)*
where the various cross sections and v have the usual definitions.
Accord
c is the average number of neutrons produced per collision.
The neutron-balance condition may be written in terms of this number.
ing to (7.236),
2,(r) c(r)
Eq. (7.15)]
jQi
[cf.
S(r,Q)
/*(tl;tl')
dCk
12'
suffers
secondary (product)
tion in dQ about Q
neutron with
collision,
yields a
neutron with direc
it
where
(7.237)*
C2t
(7.238)
REACTOR ANALYSIS
378
[CHAP. 7
Note that this function describes the general collision process consisting
The average yield per collision corresponding
of scatterings and fissions.
It follows that the integral of (7.237)
to f* is the quantity c of (7.236).
multiplied by c2, gives the number of neutrons which emerge from all
In the case of a
collisions with direction in unit solid angle about Q.
nonmultiplying medium c = 2/2( and f*(Q;a') = /(0;Q'), as given by
(7.8) . Finally, it should be mentioned that the source function S which
appears in (7.237) does not include the contributions from fission and
therefore differs from the function S given in Eq. (7.15).
In the present calculation we assume that all collisions are isotropic in
(L)
therefore,
/*(Q;Q')
(7.239)
The integral which appears in (7.237) now reduces to the expression for
the total flux <(r) [cf. Eq. (7.59)], and (7.237) may be written
a V0(r,Q) +
2(r)
*(r,fl)
(7.240)
If
4ir
Jo
2,(r
(7.241)*
The expression for the total flux is obtained from the general result (7.207).
Again noting that the source consists of the two functions on the righthand side of (7.240) and that /? is defined in terms of 2,, we have
^
'P
4>(t)
(?-242)
Both
where Q! = R/R [as used previously in Eqs. (7.206) and (7.207)].
of these relations are integral equations and may be solved by means of a
The
Neumann series1 which is developed by successive approximations.
'See, for example, H. Margenau and G. M. Murphy, "The Mathematics and
Physics of Chemistry," pp. 504-505, D. Van Nostrand Company, Inc., Princeton,
N.J.,
1943.
sec. 7.4]
TRANSPORT THEORY
379
is
<S,
a V*(W)
is
transport equation
+
i
^
(r,O,0
easily developed
stationary one-velocity
system
is
it
2(r) 4>(r,Q,<)
2,(r) c(r)
4>(r,t)
(7.243)
we have omitted the external source term
to avoid undue
The solution to (7.243)
obtained by seeking
number
complication.
a such that (T,Q,t) may be given in the form
(T,a,t)
(7.244)
4>{T,a)e<
is
is
V(r,tl)
Q,
obtain immediately
2((r) c(r)
(7.245)*
<f,{r)
= 2, +
(7.246)*
be called an
is
2<
"time-absorption"
in
reactor
V0(r,O)
2,(r) (r,Q)
= 2(r) c(r)
jQi
is
wherein
Kr,a')
/*(";')
da'
For
(7.247)
REACTOR ANALYSIS
380
[CHAP. 7
If
*(r,Q)
if
4tt
Jo
"
2(r
(7.248)*
If
*(r,U)
*(r)
*(r
jf
= c2(
4?
- R)e"*.
- rT
0(r>-^-'"l
JK
|r
its magnitude
>(r) = e,Br
(7.251)*
these
by substituting
Thus
into (7.251).
cX,
If
we define
be
written
= r
g-S'lr-r'l+i'B
|r-rt
4ir
Jv
dp
(7.253)
To evaluate the
P2
(see
r' Wr'
functions which
Eigenvector
and coordinate system.
7.12
with
(7.250)
is determined
~-%}r
Fig. 7.12).
e*8"'
is
a
Thus
"4r
4t
!H(H
(7.254)*
General Solution for <p(r) for Infinite Media. The general solution
obtained by superimposing the elementary solutions
to Eq. (7.251)
is
2.
pip.i,^)
Fio.
dff
r',
Elementary
rfr'
/S
written
"
then
TRANSPORT THEORY
sec. 7.4]
eiBT
if
381
B = Bu
where u is a unit vector (a direction), then the general solution may be
obtained by integration over all directions u; thus
*(r)
|u
a(u)e'Bu
du
(7.255)*
(r)
Is Solution
Laplacian of
(r)
V*4,(r)
Stationary-wave Equation.
using (7.251) and (7.255), we obtain
=
to the
WBu
Ja a(u)
du =
-B2
If
o(u)e'Bu
we compute the
du
(7.256)
Thus the solution (7.255) for the infinite medium satisfies the differential
Generated for wjivans (University of Florida) on 2015-09-22 17:20 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
equation
V2<Kr)
#2<Kr) = 0
'
4ir
['
Jo
dR
[ a(n)e--**
Jn
dn =
ff
4ir
Ja
W
1
f*
+ iBu
O/S,
(7.257)*
REACTOR ANALYSIS
382
[CHAP. 7
slab reactor.
medium.
(x,6)
form
cos
etB*
cos
1
1
iB
-
2^
Bx + cos
sin Bx
2<i
cos
.)
l +
Y2
cos2
(7.258)
TRANSPORT THEORY
sec. 7.4]
383
However, an approximate
the boundary condition completely.
can be obtained which can be useful for actual applications.
(a,e) =0
<
<
solution
Eq. (7.97)]
(7.259)
If
(7.260)*
(7.258), the following equation
is obtained:
2(o
This equation
(7.254).
slab.
It
cot Ba =
(7.261)
only at
other
a-,
it
is of interest
directions.
dA
Jin
0(a,a)(O
s) da = 0
(7.263)*
REACTOR ANALYSIS
384
[chap.
/a
(t>(a,0) cos 0
sin
= 0
0 dO
dQ = sin
(7.264)
dd d\j/.
This condition
1.2 1
-V-
1.0
0.8
0.6
ai
u
TO
No return current
c = 4/ir
% 0.4
M
X
"S
(a,v)=0
=
" c 4/t
fl/2,
=1
Ba=1.017
B/2,
=1
Ba=jr/4
V
s v.
0.2
-0.2
-0.4
20
40
60
80
100
120
140
160
180
Fig.
7.14
will yield 4>(a,6) < 0 for some values of 6; but the effect of this on the
estimate of the inward current will be compensated for by the contribu
tions from the remaining directions.
The substitution of the infinite-medium
yields, after integration,
cot Ba
B
2,
(1
(
(Vg)
In
[1
solution
tan-1
(B/Z,)]
+ (B/S,)*]
(7.265)*
A comparison of this result with (7.261) reveals that the right-hand side
of (7.265) is always less than B/Xt; thus these two conditions give different
This discrepancy is well
values of a for a given set of cross sections.
demonstrated by the sample computation for c = A/ir and B/S( = 1. In
the
Ba
case
(7.261)
yields
(7.265) yields
TRANSPORT THEORY
SEC. 7.4]
385
= 1.017.
noticeably
0(r,O)
*(r
- R)e-*' dR
(7.250)
*r)
Ajel*:
(7.266)
This function will also satisfy the integral equation (7.250) ; thus, we may
use this relation for the total flux 4>(i) under the integral sign.
The result
is
*(r O)
-cS'A f"
4*
(B|r-
dR
-R\)e~^
R|
sin [B(r* + R* f"
(rf + R* - 2rR)l
Jo
4r Jo
cL,A
sin
2rRM)*]e~1-*
dR
(7.267)
where n is the cosine of the angle subtended by the two vectors r and R
Note that in the present system 4>(r,Q) 4>{r,0).
(i.e., i R = rRy).
All the statements which have so far been made apply to the infinite
medium.
We now assume that the solution (7.267) is
valid in the interior of the specified sphere. If (7.267) were an exact
solution for this finite system, then we could satisfy the boundary con
dition 4>(a,0) = 0 for tt/2 < 0 < t. Since this is not the case we attempt
to satisfy this requirement in part.
As in the slab problem, if we ask
that
> 0 everywhere within the sphere, then the appropriate condition
homogeneous
to be satisfied is
*(<*,) = 0
ANALYSIS
REACTOR
386
[CHAP. 7
cX,A
ir
f"
e~z' sin
a
Jo
B(a + R) dR
+ R
cZtAez*
4*
t"
e~z- sin
Bs
ds
(7.268)
f"
e-*"sin Bsds
^.
Zl
[EiCSfi
J.
2i
[-
,_.a<_iB)
- iBa) - Ei(X,a
e_.(2l+iB)]
+ iBa)]
ds
- Im #i(2,a + iBa)
(7.269)
where E\{z) is the exponential integral [see Eq. (5.306)] and Im 2?i(z)
denotes the imaginary part of 2?i(z). The exponential integrals of com
plex argument have been tabulated and will be discussed in detail in
connection with the Serber-Wilson method.
From the above results we find, then, that the condition
yields the requirement
-Im
E^Xfi + iBa)
= 0
4>{a,ir) =
(7.270)*
putational purposes.
lIbid.
TRANSPORT THEORY
sec. 7.5]
7.5
387
of Spherical-harmonics Method
Applications
will
value of
use is independent
re), we
(i.e., the
set (7.271) at n = 3.
This yields the
which is the next more accurate model after
so-called P3 approximation
It is of interest to note in this connection that only
the diffusion theory.
yield higher degrees of accuracy.
the odd-order approximations
For
=
=
(Pi), re
1 is the first-order approximation
3 is the second
example, if re
The approximation corresponding to re = 2 yields the same order
order.
equation (second) as does the re = 1 approximation.
Thus
calculation yields no new information about the angular
the re
distribution of the flux, and so we must go to re = 3 for a more accurate
In the same way n = 4 is of the same order as re = 3, and so the
model.
differential
= 2
(a)
- - Tx 3 +
-2(*2 + 2.th*2 +
S2
S3 =
2(03
+ 2,t)33 +
'=
<4 term
-J-
<2
in (d).
,
^
(7.272)
(<0
REACTOR ANALYSIS
388
[CHAP. 7
Consider
two
x = 0 as a common
form.
dfa
^ + Zo =
So
(a)
dfa
, ,
dfa
dx
= 0
(d)
2,
2,
2,4>3
2.D
(7.273)
(7.274)
and have recognized that for an isotropic source Sn = 0 for all n > 0.
Since the above equations describe the directed flux in a homogeneous
medium which contains a constant source, the application of these
equations to a two-medium system requires that a separate solution be
obtained for each region and the two results matched at the common
surface x = 0.
We solve the four simultaneous equations in (7.273) by eliminating the
three harmonics fa,fa, and fa and thereby obtaining a single fourth-order
differential equation in fa.
The solution of the fourth-order equation
may then be substituted into (7.273) and corresponding expressions
The procedure is straightforward
obtained for the higher harmonics.
Suffice it to say that the calcula
and will not be discussed in detail here.
tion may be carried out by first eliminating fa from (7.273c) by means of
(7.273d), then fa and finally fa by performing several differentiations of
the Eqs. (7.273a), (7.2736), and (7.273c).
TRANSPORT THEORY
sec. 7.5]
o may
389
be
d*o
d2a
dx
dx2
7<o
= SiZjiSo
(7.275)
where
Z
P ~
3
35S3
92qSi
42
Its solution
coefficients.
o(x)
y = SoSiSj
equation
(7.276)
with constant
is easily shown to be
= ae
+ be-"" +
ce"z
+ aV
So
(7.277)
6, c,
given by
m
2a
(7.278)
4n(x)
2{x)
e~ - - e + - e~ - - c
(7.279)
(n
n
m
\m
J +-/
iMae-* + ^Mbe-"" + ^Nce"* + iNde-"1
f/2ix - |2i0
-So
*^
(7.280)
.(*)
3
(
142;
amMemx
6mMe_mT
cnNeni + dnNe~
+ 32!/)
(7.281)
where
/ and
and
(7.282)
If
the solutions (7.279) through (7.281) are used in Eq. (7.273c), then a
comparison of the coefficients of like functions of x reveals that both
medium by requiring that the flux remain finite for all x. This immedi
ately identifies several of the integration constants, and it may be seen
that for the medium which occupies the space x < 0 (which we denote
REACTOR ANALYSIS
390
[CHAP. 7
4$*\x)
4>{m(x)
<W\x)
+ ce>" + =^
(7.283)
e-* +
f
nM
\mM
iM^ae"-* + ^Nttce""*
/"(ac) =
set
-2^'
e"* J
{flmuMMe*
(7.284)'
v
(7.285)
+ cnMNMe'*')
(7.286)
If
we denote by the index F the medium which occupies the space x > 0,
then the appropriate solutions are
=
=
4>[n(x)
<&"(*) =
*<3"(x)
be" +
de-"
2i,">
"
+
+
(7.287)
(7.288)
iMr6e-"** + iNwde-"'*
(7.289)
i4cF
(6^f* "
+ dnFNre-'*)
(7.290)
The parameters
M0,m)
-1
= 4>r(fl,v)
<
J) (^r^)
<
(7.291)
is obtained by
*-(:r) p(,l)
terminat
(7-292)
i-O
at
s =
<Hx,M)
If
we
of
**o(x) +
*2(*)0V
1)
+ i*,(x)(5/i*
3M)
(7.293)*
apply this approximation to each side of (7.291), then for all values
we must have that
#"'(0)
#"(0)
(7.294)
TRANSPORT THEORY
SEC. 7.5]
a-b
[
mM
mj?
and d:
b, c,
c- d =
S0
1^
[
- A/F6
Um
A/ji,a
391
n'
J
+ NMc - NFd
L
= 0
(7.295)
The solution of these equations for the four unknowns when substituted
into the harmonic relations (7.283) through (7.290) gives a complete
description of the directed flux through the two regions M and F.
We illustrate the above method by means of a sample computation.
Let
If
we assume
2<,M)
2'*"
that
X)n
0.0022^>
S^>
0.99821*"
= 0
for n
20 = 2( 2,l)o = 2a
2JP>
= 0.5
> 1, then
2i
W>
jtf0 = 0
(7.296)
it follows that
= 2, 2,1)! = 2,
/i0
(7.297)
^1
22
23 2|
2<
M, and N.
mM = 1.9742jM>
A/.w = -0.9921
=
2.4412
A//, =
-0.7481
= 0.98762<f,)
JVF = 0.5382
The substitution of these values into Eqs. (7.295) yields for the coefficients
a =
c =
-0.09324
-8.634
[M>(x)
= 10
are then
- 0.09324e1"42''"'i -
S O.SloSe0
^(x)
= 0.2517
d = 0.5212
b
8.634e0
06477S'1J"*
064772'""1
(7.298)
= 0.04625eI974J;<(,"
0^>(x)
0.039 13e1974S'(J"*
and
4>\P(x)
\">(x)
^)(x)
<tiF)(x)
= 0.5 + 0.2517e-2-4412'<r)l
+ 0.5212e--987s'<"*
= O.OSlSOe-2-4411'"'' + 0.2635e-0S,87<s'(')l
= -0.09414e-2-441-"'^
= -0.09849e"2 "I*."1*
+ 0.1402e--987,z'c')*
+ 0.05936e- M7eS'",J!
(7.299)
REACTOR ANALYSIS
392
[CHAP. 7
and
i(x)
-4xDe"
(7.300)
+ CxDe~"
Ac-
(7.301)
(7.302)
-^D^e"""
(7.303)
tiPW
Ce- +
(7.304)
#"(*)
CDrxFe-"'
(7.305)
.4
F,
4>[M,(x)
and
This yields
- (i/^)]
and
io(x) +
(7>306)
given by
(7.93)
%4>x(x)n
tf(x)
4>\M){x)
and
Table
oFKx)
\F){x)
7.1
= 10
8.728e0-""z'<J"*
= 0.3153e "87S'""*
= 0.5 + 0.7722e-1-2262'""*
= 0.3153e-1-2252'1"1
lists (x,p)/tf>0(x)
approximations
given
(7.307)
(7.308)
to the P3
The numbers in the
is
according
parenthesis are the corresponding values obtained from the diffusiontheory relations (7.307) and (7.308).
These results have been used to compute the total flux 4>o(x) throughout
the vicinity of the interface.
Figure 7.15 shows this distribution at
various stations measured from the interface x =
in units of mean free
0
(7.88)
are
Ae + Ce~" +
-D'0(x)
*o(x) =
4T>(x) =
condition (7.294).
(7.87)
2//).
*,(x)
is
x2
= S0
=
Za4>o(*)
where
f-
-Dtfix)
paths
TRANSPORT THEORY
sec. 7.5]
393
(scattering)
a moderator
It
1
P3 approximation
Pj
approximation
****
*"
****
Interf ice
0>
>
Scat erer
ol
Absorber
65432101
Distance
Fig. 7.15 Total neutron flux in the vicinity of the interface between
scattering materials.
23
absorbing and
Distance
from
7.1
-1
-1
interface
\,
0.3708
0.2724
0.2(100
0.3352
(0. 1B54)
0.3268
JV"
0.3379
x,c>
0.3725
0.4956
(0.3716)
(0.2431)
(0.3477)
(0.3141)
0.4022
(0.4044)
(0.6523)
0.6154
(0.8718)
0.9859
(0.6395)
(0.5000)
0.5593
(0.5000)
(0.8046)
0.8548
(0.6859)
(0.5000)
0.4231
(0.7568)
0.7890
0 7215
0.4190
(0.3605)
(0.2209)
(0.3088)
(0.5000)
0.4546
0.3441
0.7453
(0.6284)
0.6579
0.4862
0.2558
(0.1282)
0.6269
(0.5000)
(0.7791)
0.9348
(0.5956)
(0.6911)
aoi.ovaa sisaivnv
jvho]
/,
"Oidoj^osi
'fyl'L
(I)
()
uoipajiQ auisoD jo ucuinau 'uo|)Oui n
1
aaueisiQ
oij
0
wojj oDcpajui u; ueaoi osjj sqied
IJ9AV
()
jujnSjuy uopnquisip jo
jjjv
uoa-jnau
fjjyj
ijJIV
fjjjv
sec. 7.5]
TRANSPORT THEORY
395
Figure 7.16gr is a composite sketch of the directed flux for several values
It allows a comparison of the relative density of neutrons moving
of ji.
in a particular direction to the total neutron population at each station.
b. Infinite-slab Reactor.
Consider an infinite slab of multiplying
material with width 2a operating at steady state.
The neutron sources
in this system are the fission reactions, and on the basis of the one-
vZ,
jlx
velocity model we find that the number of neutrons produced per unit
volume per unit time is given by the expression
4>(x,n)
dn =
= S(x)
vXf<f,o(x)
(7.309)
yzT^
2i
dx
(7310)
(7-310)
S.po
vZt = 2
- vZ,
is
22
IA
_
Z'Q
dx
d0
djn
dfa
^+^
(7.311)
is
analogous
5,
3,
n =
1,
0(O,M) dn =
0}:
P.GO
(7.312)*
This
is
(x
is
symbol 20
Because of the symmetry of the present
replaced by Z.
some
of
the
arbitrary constants which appear in the general
problem,
solution can be eliminated.
For this purpose we introduce the following
Pn(M) d" =
*-(0)
(^r) *,(0)
/-i
yields
(7313)
ANALYSIS
REACTOR
396
[chap. 7
<^
0,(0)
If
we
= 0
= 0
0,(0)
(7.314)
Mm(b
which
a)
+ Nn(d
constitute
c) = 0
(6
o{x) may
a = 0
it follows immediately
be
c) = 0
(d
Since
i -
a)
(6
and
a)
(7.315)
(d
c).
the only
(7.316)
for
[see (7.277)]
(7.317)
0i(x)
fa(x)
i(x)
sinh mx
sinh nx
(7.318)
$ AM cosh mx + CN cosh nx
1423
(7.319)
(mA/ A sinh mx +
(7.320)
*o(o)
"
2l{a)
*'2(a) =
**(a)
X C^^)
j-0
/-i
- 803(a)
-o
The application of
(7.322)
conditions
l+-:N+
_ tanh ma
/23
12niV
iZii
.r
ma
(52V/8) cosh na
+ +
tanh na
22o
on
.
,r
-N
tanh ma
m
s
-z-M
\
+
A C
C,
two
3,
(7.321)
and for n
tanh na
(7.323)
(7.324)
sec. 7.5]
TRANSPORT THEORY
397
where (7.324) may be identified as the criticality equation for the infinite"
"
slab reactor in the P% approximation.
Consider as a numerical example of the slab problem the system defined
by the following
set of parameters:
2 = 0.22,
2J
-0.0462,
Zi
= 0.5
JTo
= 0.62,
22 = 23 =
M, and
2/ = 0.12,
= 0.82,
2.
= 2.46
AT
(7.325)
2,
the values
n = 0.2965t'2,
-0.05820
This solution
(7.325).
o2, = 4.196
is
found to
be
P3 approximation
(7.326)
The value of A/C is computed from (7.323), and the result yielded
A/C = -0.00002898. Thus the total flux according to (7.317) is
o(x)
= C[cos (0.29652,*)
is
(7.327)
(7.328)
using D =
we have
(D'-K1-5!) (-!-!)
1/3(2, The substitution
of the data
jl^2.).
(7.325) into
^
for the
Pi
= 0.2878
approximation.
0o(x)
Ba
= 1.261
o2, = 4.382
is
cos (0.28782,x)
REACTOR
398
condition (7.260)
ANALYSIS
[CHAP. 7
>(<!,*)
(7.262).
In the
present problem
e
= 1.046
If
we apply instead
= 3.198
(7.330)
(7.331)
Pi
Pi
4.382
4.196
*(o,x)
= 0
3.198
No return
current
3.552
It
Fig.
7.17
mations.
Pi
and
Pt approxi
TRANSPORT THEORY
SEC. 7.6]
399
Computational
It is convenient
of the physical system.
to discuss these at this point since they serve well to demonstrate the
application of the Fermi age model and the transport-theory methods
some rather gross idealizations
is a rationale
system.
a. Fermi-Marshak
Method for Hydrogenous Mixtures. A very
instructive and useful application of the general Boltzmann equation
of the neutron
age in hydrogenous
systems.
This problem was first considered by Fermi,1 and later by Ornstein and
Uhlenbeck2 and by Marshak.3 The present treatment, which draws
principally from the work of Marshak, is based on several simplifying
about the physical situation that render the method
assumptions
especially attractive for computing the neutron age in water-metal
systems.
Other more accurate models for describing the neutron popu
lation in general hydrogenous systems are developed and discussed in
considerable detail in Chap. 12.
Special attention is given there to
provide a complete mathematical treatment along with a comparison of
The analysis which
the analytical and physical features of each model.
follows, on the other hand, is intended primarily as a demonstration of
1 E.
'L.
7 (2),
S. Ornstein and G.
E. Uhlenbeck, Physica,
13 (1936).
R. E. Marshak,
and 209-211.
400
ANALYSIS
REACTOR
[CHAP. 7
We know from our earlier study of the Fermi age model that the age
is proportional to the quantity r2(w), the mean radial distance
squared at which fast neutrons from a specified source reach lethargy u;
the constant of proportionality in this relation is dependent upon the
nature of the source [see, for example, Eq. (6.52)].
Our ultimate objec
For this purpose we introduce the
tive, then, is to compute r2(w).
following model:
t(m)
(1)
isotropic
\
1
(7.332)
terns of coordinates
1
(5) The absorption cross section of the medium is small com- I
pared to the scattering cross section
/
a V*(r,ttfQ) +
+
*(r,u,Q) = S(r,ufi)
hfu-
S,(M,) du'
L *(r,u''a,)
(M-";M'.Q')
da'
(7-333)
(x,u,Q)
*(*,w,fi) =
S(x,u,n) =
%[**
{x,u,Q)
jo
tive
S(x,u,Q)
d+
S(x,u,n)
0,
(3)
5(x) S(u)
(7.334)
monoenergetic
(l/4x)
S(x) S(u).
(7.335)
TRANSPORT THEORY
SEC. 7.6]
If
401
^,
fa
hL
$ S(x) S(u)
s,(w'} du'
fa-
*(x'M'-'1') ("mo;u',o')
da'
(7. 336)
Let us define
the relationship
is
6.
f(u;x,y,z) dx dy
dz
(7.337)
are independent
Since x, y, and
in the form
S(u;x,y,z)
(7.338)
fi
fr*f(uyt,y/)dxdydz
x2
y2
r2
J2m
x*(u) =
/i(u;x)(s2 +
y*
oo
(7.339)
z2)
dx
Since
z2.
x2fi(u;x) dx
(7.340)
obvious that
Thus the
F(u) = x*(u)
x2(w) in
turn
x\u)
(7.341)
iffy)
ix*(u)
(7.342)*
The quantity
is
t(u)
T\u) = Zx\u)
y*(u) +
x2dx
J-=^
J_m
dx
Ja ff
is
f"
Mu?)dz[m oo h(u;y)dy
+
ac
?()
/""
72(u) m
it
(x,u,n) da
4>(x,u,n)
da
(7.343)*
402
ANALYSIS
REACTOR
[CHAP. 7
The calculation of the age requires therefore the solution of Eq. (7.336)
for the flux 4>(x,u,n). In carrying out the calculation (7.343), it is actually
more convenient to work with the Fourier transform of the collision
This technique is generally useful for
density than with the flux proper.
computations of this type, and we will have occasion to apply it to other
problems later, specifically to the calculation of the age using the ChristyWheeler kernel and in the analysis of hydrogenous systems (Chap. 12).
We develop the appropriate relations required for this technique in the
Since our primary interest here is to demonstrate
analysis which follows.
an application of the Boltzmann equation, we will not examine all the
details of the mathematical analysis but will instead concentrate atten
Thus the present
tion on the reactor-physics aspects of the problem.
analysis is to serve as an outline, and the reader is referred to the report
by Marshak and to Chap. 12 for further analytical details.
in Eq. (7.336) by
Uu)
-\-
(7.345)
(5), Eq. (7.332).
The
\ Hu)
h jo
du'
fa
S(x)
F{x'u',tl') (u>"o;',a')
F(z,H,M) + F(x,M,M)
!T{
Uu)n ^
da'
(7.346)
of this equation.
If
we
lX(()fM]<pU,W,M)
F(x,u,n)e't* dx
(7.347)
-^L
v 2jt
fa-
jo
du'
*(*'W''M,) (".w.-u'.O')
da>
(7-348)
The procedure
is
is
Our purpose
2^
of (7.346)
is
<p(?,m,m)
!'
5{F(x,u,n)\ =
^=
call
[1
We
(7.344)
similar to
TRANSPORT THEORY
SEC. 7.G]
*(,u,M)
(W,Mo;u',a')
403
- (^T^) **(*,) PW
m-0
(7-349)
(7.350)
-(u;u',0')
-a
a-0
y(fl)
rr(O')
27
(^)
X
m-0
p.W
*.(,)
*-a'M,)
*a(Uy'Q,)
(^^j
m-0
/o"
- *,()*,]
=0
K(Q)
3- -a
rr(a')
Pm{n')
da'
(7.351)
if
is
m-0
*-
n.
in $m
we
Thus, for
if
/;
(^) "- w
(7.352)
e*i(f,M)
*o(,u') o(u;u',Q')
^"
#0({,u)
0,
[i
d'
(7.353)
REACTOR ANALYSIS
404
[CHAP.
.(,)
[(n +
-5
*-(*,') (;'
"
dM'
(7-354)
(7.355)
v(//)
then
Generated for wjivans (University of Florida) on 2015-09-22 17:20 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
2 Z
A m =a-
Let
- A = ^-
= A(w>
and
But,
with S. =
+ ^^r
for isotropic
and (7.138)]
scattering
In the
case of hydrogen,
(7.357)
A)<0)
system
[see Eqs.
(7.131),
= *(mo~ f)c"'~"
1 a
this yields
(1
in the (C)
S(,Mo;u',Q')
(7.356)
= 5(Mo
- f//)e'-
(7.135),
(7.358)
(7.359)
in
By assumption
= e("'-u)
(7.360)
(0)
$ g(M'
u)
(7.361)
Note that the integral of this function over all u and no is unity. If this
function and (7.359) arc used in 3, the scattering function for a hydrog
enous mixture may be written in the form
(u)MoV,Q')
A(u')
(mo
<;)"-"
HI
- A(u')] S(u' - u)
(7.362)
TRANSPORT THEORY
SEC. 7.G]
The coefficients
For example,
8 may be computed
o =
A(w')e'- +
, =
M(')e|
405
- A(u')l
[1
('
(7.130).
- u)
These functions may now be substituted into Eqs. (7.353) and (7.354)
and the various harmonics
obtained from a simultaneous solution of
a finite set of these relations which yields an acceptable approximation
function
to the transform
<.
f
^
x2dx
}-
F(x,u,n) da
(7-364)
F(x,u,ri
ff-M*,u,M)}
-=
('
HZ,u,n)
i.e.,
d$
(7.365)
Moreover, 4> is given by the expansion (7.349) ; it follows, then, that the
integral in the denominator of (7.364) reduces to
^F(W)
dU
-~
30
d{
m=0
-^mrsf'.MM'-",dl
(7-366>
FTKx,u)
_ (-D
V2ir
f"
J-
{-*o(C,w)e-*
d{
(7.367)
Note that
F$>(x,u)
m F0(x,u)
-j= f"
V2ir J
JL
F(x,iM)
*0({,u)-***
<p0(,w).
^oF^u)
(7.368)
Evidently,
(7.369)
REACTOK ANALYSIS
406
In an analogous fashion
we introduce
*iT"(,w) =
[CHAP. 7
also
F^x,u)e^ dx
-=
/
2ir J _
V
(7.370)
-L=
V 2ir
Fo(x,M)e'^dx
(7.371)
S>0(,M)
/""
*i,0'(i,w)
is
x*dx
F(x,u,n) da
dx
^M0,u)
(7.372)
~B
-w*mu)
. F"(x'u)
x2
F0(x,u) dx
--m [S*a'tt)L.
The expression (7.364) may then
?()
1K4
be
written
[^f^l
- -
<PoU,)
J{_0
(7-373)
and substitute
l-K
Jo
^-
/""*o(f,M')o(tt;',Q') du'
Jo
- i\t(u) *i(,w)
+
V 2t
- *X.() *o(,u)
d5
*o(f,M)
(7.375)
"*i(,m')i( ;u',a')du'
$>o
into (7.374).
is
3>o
the age of
J~
277
/_
F{x'u',l) da =
f-.dx
J/
which
neutron
of lethargy
TRANSPORT THEORY
sec. 7.6]
+ h(u;a) + h{a)U{u;a) +
This equation applies when u
defined
a.
407
76(M;a)
I<(u;a)h{a)
(7.376)*
7,(a) = X<(a)e-*
Ii{a)
X,(a)e*
(7.377)
and
(7.378)
to reactor
mixtures
out using the Fermiare given in Fig. 7.18 for
carried
relative magnitude
1 L.
H,0
REACTOR
408
ANALYSIS
[chap. 7
Alumini im
- water
a t68'F
Stainless
steel
at 4 503F
- wat er
^\
DO
0.2
0.4
0.6
Metal
FiO. 7.18
- water
Zir conium
v/ater
at 68 F
0.8
1.0
1.2
1.4
volume ratio
method.
TRANSPORT THEORY
sec. 7.6]
409
for
2*'], it is clear that there exists some volume fraction of water
> w*', such that the ages in the two systems are
system a, where
identical.
If one could somehow determine this value of vl\ then a
knowledge
system a.
A clue
We observe
age.
that this result is completely specified by the mean free path in the
mixture.
Thus, if as argued previously, the details of the energy struc
ture of the metal cross section are not a primary factor in the determination
of the age, then any two systems with the same mixture mean free path
would exhibit the same age.
One might argue, then, that the comparison
is obtained from the
value of
X(.
us defer
Xi0
X(
~ X
^y^]"
and
Pi m density f of metal i
p m density t of water in system containing
y, = volume fraction of metal i
v\p m
(7.379)
metal
i
metal
+ v?
i
The
(7.380)
For the
(7.381)
XJ we use expression
>,
-aRP- +
Aa
^
A
ib
v)
-~a'
Jib
Aw
- *)
(7-382)
Suppose that the age has been determined accurately by some method for
system a and that these data are available as a function of the volume
fractions of metal and water as shown in Fig. 7.18.
These results may
t These densities refer
ANALYSIS
REACTOR
410
[CHAP. 7
then be used to estimate the age in system b, since given a volume fraction
vi, of metal for system b, we can compute from (7.382) the volume frac
tion va for system a which would have roughly the same age. Using
this value of va we could read off immediately from the age-versusvolume fraction curve for system a the age for system b which contains
a volume fraction vb of metal.
va
in terms of
in the form
Va
vb
- x) + (x -
9a
1)
(7 383)
-g
be
and
- r^yi
(7.385)
- + (x v(ea
elY+W^ld
1)
Vb(6b
Va ~
1)
(7-386)
This relation, or (7.383), may be used to relate one metal- water system
We observe that the comparison is based on some suitable
to another.
value of the microscopic scattering cross sections of the water and metal
in each system and on the densities of the various constituents.
Evi
dently this technique may be used to extrapolate data from one metalwater system to another as well as from one temperature to another.
">
=_
El
J
tb;-
E
(IE
=
Ml
'
Mo
Ju0
(r.(u) du
(7.387)*
Ei
TRANSPORT THEORY
sec. 7.6]
411
Table
7.2
at
Volume ratio of
stainless to water
t, cm*
Fermi-Marshak
0.3
0.6
1.0
were
(2)
(3)
(4)
(5)
(6)
40.0
46.5
53.0
61.5
0.332
0.500
0.652
0.799
41.2
49.0
54.5
60.0
0.472
0.779
1.038
1.346
41.5
48.5
54.5
62.0
computed
from
densities:
Table
SS-HA
450F
0.834 g/cm'
Pu = 7.75
= 39.3 barns
= 6.8
c. Age
Calculations
Steel-Water Mixtures
A1-H20 ratio
(1)
0
in Stainless
450F
7.3
AI-HjO,
Zr-H20,
68F
68F
pIa,) =
1 0
2 .7
PAl
a\kn
42 .7
1 .5
by Christy-Wheeler
P<?r)
io
6.43
P7.T =
-(Zr) =42.7
= 9.0
Kernel.
The calculation
of
the age on the basis of the continuous slowing-down model has two
serious shortcomings.
The first arises in connection with the scattering
properties of materials at high energies, and the second because of the
use of the slowing-down density in place of the flux in the calculation of r2.
The first of these considerations is important when the scattering cross
energies and/or exhibits forward direction
In both
tendency [see, for example, Eq. (4.31) and related discussion].
instances the neutron will travel large distances from its point of origin
before the scatterings become sufficiently frequent and isotropic to
section
is small
at high
justify the
collisions
412
REACTOR
ANALYSIS
[CHAP. 7
4>(t,u)
is required.
point r' different from r, a relation is required between >(r,w) and q(r',u)
which accounts for local spatial variations in the neutron distribution.
As in our earlier work, we compute the age from the value of r2(u) for
For this purpose
a point source of fast neutrons in an infinite medium.
we select a point source of unit strength at the origin of a suitable coordi
The calculation of r2(w) is based on the
nate system in the medium.
continuous slowing-down model which has been modified to include the
first and last flight corrections proposed above.
Consider, therefore, a
source for the continuous slowing-down process which consists of the
first-scattering centers of neutrons emitted from a point source at the
If
origin.
S(r0) dr0
G(|r
f0|)
dl
(7-388)
4Tr2
"
process.
%rr()II
(?-389)
slow past lethargy u per unit time in volume dr' around r'.
The first
scattering sources then give rise to a slowing-down density at r':
q(T',u) =
G(|r'
4>(t,u)
- r,|)S(ro) du
product gives the number of neutrons actually occupying the interval du.
To obtain the flux at r we integrate this source with the transport kernel ;
TRANSPORT THEORY
sec. 7.6]
413
thus,
*(r,u) du
where
T^
- r'|)
7\|r
^j
[q(r',u)
dx'
- r'D - hr=?\>
(7-390)
4>(r,u)
T(\x
- r'|) j
G(|r'
- r0|)S(r0) du
(7.391)
4>(x,u)
- f"
\
T(\x
- x'\) dx'
Jj
oo
G{\x'
- xu)S(xa)
dx0
(7.392)
4>{x,u)
be obtained,
/
*2() = J-=^
/
x2(x,u) dx
(7.393)
${x,u) dx
As in the
SiKz,u)|
1
(Z,u)
~j=
4>(x,u)e*t*
dx
p. 4 in the reference
(7 394)
paper.
See
414
It
ANALYSIS
REACTOR
(,)
[CHAP. 7
f*
i"
({,) =
x'4>(x,u)e*** dx
^=
(7.395)
/_",
J
Thus (7.393) may
be
(x,u)
dx = -s/2^9(0,m)
- ~ V2^(2,(0,w)
***(*)
written
[cf.
(7.396)
Eq. (7.374)]
a*
e(f,)
(7.397)
x2(w)
The calculation
it
T(\x
- x'\) dx'
T,
S,
j"
J"
T*G+S =
G(\x'
(7.398)
follows that
= 6(,W) =
~ 0(,u)
= d"gs
0()
be computed
+ g"s8 + s"g8
5{(x,u)\
and
ff{7VG*,S} = 0(*)ffU)s()
(7.399)
g(l)
(7.400)
s($)
For
(7.401)
B(*>")
T-
it
that
1
seen
\Si
is
0,
S,
T,
G,
{.
g"(0) s(0)
6(0)
+ s"(0) g(0)
9(0)
jHu)W - _
^(0)
p(0)
i"(0)
s(0)
_ r(0)
8(0)
I.
is
G,
If
pp. 23-27,
{7-WZ)
McGraw-Hill
TRANSPORT THEORY
sec. 7.6]
415
The terms at the right are the x2(u) values for the kernels G, S, and T,
respectively; for example,
g"(0)
Thus
But,
z2(w)
r2(w) = 3x2(m)
_9(0)
+ xj(u)
(7.403)
(7.404)
a(ii) + x%(u)
therefore,
=
7*(u)
The first term at the right is immediately recognized as the r2(u) com
puted for a point source on the basis of the Fermi age model [see Eqs.
(6.49) through
(6.52)], since1
*oW = jGdT
= 6t0(m)
(7.405)
SSdi
r()
'
[2.(0)]'
pe-z.Wrdr
(7.406)
[Z.()]s
The substitution of these results into (7.404) yields an expression for r2(u).
If we define the age t(u) for this system by the relation t(m) = %r2(u),
then we obtain
t(w) = tg{u) +
7.4
Age
Material
Beryllium
Graphite
3[Z.(0)]2
+'
(7.407)*
3[Z.()]
and Graphite
Fermi age
Christy-Wheeler
Experiment0
79 . 7 em'
84.0
97.2
364
364
352
Physics,"
p.
485,
tq(u) obtained from the Fermi model and should yield reasonably good
results for nonhydrogenous materials.
Some sample results are given
in Table
7.4.
1 The symbol tq(u) denotes the age based on the Fermi continuous
model.
slowing-down
REACTOR ANALYSIS
416
of Fission
d. Age
neutrons
Neutrons.
Inasmuch
[CHAP.
as
primary source of
of particular interest
the
Table
to the spectrum
corresponding
7.5
=0
(u
corresponds to
Mev)
10
An
it
Am
0 0.5
0.5 1
1 1.5
0.021
0.104
0.207
0.231
0.183
0. 117
0.067
3.5 4
4 4.5
0.035
4.55
0.0093
0.0056
0.0034
0.0014
0.0002
1.52
Generated for wjivans (University of Florida) on 2015-09-22 17:21 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
f (w)
2 2.5
2.5 3
3 3.5
Consider,
therefore,
i*
0 0151
5 5.5
5.5
6.5
6.5 7
6
a normalized
function
$(?*)
which
describes the
If t(m;m') is the
f
If
f\(u;u')
j(u') du'
(7.408)
}-^j-u
approximated
by
N
Y
"-'
^
t(u;u) j
A?<
(7.409)
^ J-AWn
n-1
-y
Awn
n >
At/
UlS-y
sec. 7.6]
TRANSPORT THEORY
be convenient
417
1
II
All 2
Fig.
7.19
^11 .
4
1
>
lethargy interval.
Thus the
Si = i Aw
age
is given
by
f(u)
r(u;un)it
(7.410)
CHAPTER
REFLECTED REACTORS
8.1
Introduction
Reflector Savings.
reflect back some of the neutrons escaping from the core and thereby
allows them further opportunity to cause fissions.
The combination of
in
a reflector depends, of course, upon the usual
materials to be used
of materials compatibility, economy, strength, shielding,
On the basis of neutron conservation alone, there are two major
benefits to be obtained from the addition of reflectors:
requirements
etc.
(1)
(2)
In the models which have been used for a bare reactor, the vacuum
has been treated as a perfect absorber for neutrons; thus, the addition
of any material whatsoever, no matter how thin or how highly absorbing,
to the outer surface of the core will be of some advantage, since it will
always return some fraction (albeit small) of all neutrons incident upon
it from the core. In our work with the diffusion-theory model we found
it useful to describe the reflective properties of a medium by means of
This concept is especially appropriate in the consideration
of multiregion reactor systems.
Since the albedo is the fraction of
neutrons incident on the surface of a medium which are reflected back, its
the albedo.
value depends upon the thickness of the medium as well as upon its
It is clear that, although a thin reflector will
absorption cross section.
to the core, it will also allow a certain
and thereby be lost to the system.
As the
thickness is increased, the fraction of neutrons lost by escape through
It is neither profitable
the outer surface of the reflector is decreased.
nor practical, however, to employ reflectors of very large thickness.
always
return
some neutrons
Although the albedo increases with increasing thickness (Fig. 8.1), little
is gained by extending the thickness beyond a few diffusion lengths.
The second advantage offered by a reflector is derived from its moder
This characteristic is especially important in dealing
ating properties.
418
419
REFLECTED REACTORS
SEC. 8.1]
neutrons appear at all energies, and if the nuclear properties of the system
are designed so as to exploit the thermal characteristics of the fuel, it is
Depend
necessary to introduce a moderating material into the system.
ing upon the engineering requirements of the reactor, it may not always
fuel.
01
4L
3L
ZL
Reflector thickness
Fig.
8.1
thickness.
Flo. 8.2
Hectors.
escape probability.
(or desired),
the principal concern is to
There
considerations.
is given by
vZ, = 2 + DB2
where
5r/2o.
(5.126)
420
REACTOR ANALYSIS
[CHAP. 8
is,
for the dimension a. Now assume that this slab is reflected by two semiinfinite media of graphite as shown in Fig. 8.2. The albedo for a semiinfinite medium is given by (5.116), and for the properties of graphite
listed in Table 5.1, this relation gives approximately 0.92; that
92
per cent of all neutrons leaving the faces of the slab are reflected back.
Thus the leakage losses expressed by the DB2 term in (5.126) are con
siderably reduced, and the multiplying slab with the proposed reflector
could be maintained at criticality with
markedly lower
of fuel.
On the other hand,
we wish to retain the original
(bare-slab) fuel concentration, the width of the core in the reflected system
could be reduced.
It this difference in critical size which we call the
reflector savings; thus
the critical size of
bare system and or
a<>
a
configuration
that of
if
is
is
if
concentration
Reflector savings =
This number
the addition of
relative
(8.2)*
is
saved by
a reflector.
The reflector savings depends, of course, on the
nuclear properties of the core and reflector as well as on the
a measure
of the amount
of core material
is
the number
commonly
extremely
it
is
is
the diffusion
In
may be extended,
is
a0
then we define
with suitable
modifications,
to
REFLECTED
SEC. 8.1]
REACTORS
421
diffusion
model.
involves the
Generated for wjivans (University of Florida) on 2015-09-22 17:21 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
down
differential
models which
are amenable
to hand calculation.
We begin
particular reflector.
REACTOR ANALYSIS
422
[CHAP. 8
two-group model developed here differs markedly from the general multimodel in one essential feature.
The two-group treatment is an
exact analytical solution of a particular set of differential equations which
describe an idealized reactor system, whereas the solution from the multigroup
a. General Equations.
We have noted that in reflected reactors,
which are dependent at least in part on thermal fissions, the reflector
It is clear that, on
serves as an important source of thermal neutrons.
the basis of this fact alone, the spatial distribution of the slowing-down
density at thermal cannot be proportional, everywhere, to the distribu
multigroup methods.
Throughout the analysis which follows we assume that the reflected
reactor consists of a central core of multiplying material which is inter
nally homogeneous, and surrounding the core, although not necessarily
covering all its surface, is another internally homogeneous medium which
Thus all the reactors
is nonmultiplying and which we call the reflector.
This does not limit our analysis,
considered here are two-region systems.
however, since all the fundamental ideas and techniques involved are well
The
demonstrated by these systems and are more generally applicable.
analysis of systems having three or more different regions can be carried
These calculations are merely more laborious.
out by the same methods.
The basic relations which give the distribution of neutrons in the core
(C) and reflector (R) of a one-velocity
- DcV*Mr)
2<c'= rZ/*c(r)
rZ/*c(r)
(a)
(8.3)
-DmW+m{t)
2<fi(r)
= 0
(b)
REFLECTED REACTORS
SEC. 8.2]
423
functions
be satisfied by the
4>c
and
4>R
are
assumed to be:
(l) Symmetry
core center
reflector interface
The flux must vanish on all extrapolated
(3)
reflector interface
Continuity of the net current at all points of the core-
^)
outer surfaces
of the assembly
Wc(r)
V^(r)
BJ*c(r) =0
B2
- . [2/ - Z]
(ft)
=04-7)~
if
is
0)
b. Spherical Geometry.
The case of a spherical core (radius
and
particularly simple
spherical shell reflector (outer radius Ri)
spherical
In this case the general solutions of Eqs. (8.5) are
assumed.
symmetry
is
cosh xRr
n
CR
sinh xnr
K,
r<
sin
/ts
sinh xR(Ri
(8.8)
MRo)
MR*)
9)
<P(r)
c,
Br
r)v
, .
<Pc(r)
(8.7)
is
0,
= ft! + 0.71X'*'
with
Also,
(8
If
4>k(Ri)
is
=
Since the function cos Br/r
we take Cc = 0.
singular at
condition
in
for
the
by
(8.4), we have
present case,
(4)
it
6)
AK
cos
Br
Cc
(8
.
sin Br
Ac
= 4>u(r)
. .
(r)
c(r)
,
>c(r) =
,
ANALYSIS
REACTOR
424
sinh
[CHAP. 8
these relations, an equation
sin BR0
hr(Ri
(8.10)
Ro)
If Eqs.
(8.8) are used in (8.9), and the flux relation is divided by the cur
rent relation, the constants A and C cancel out and the following equation
results :
- [xrRo
coth x(#i
Ro)
1]
(8,11)*
In order that the system operate at steady state, the above relation
is
is,
This
ship must be satisfied by the parameters of the core and reflector.
in fact, the criticality condition for the one-velocity two-region
relation
spherical reactor.
The computational procedure involved in the application of this relation is straightforward and follows closely the methods
used in bare-reactor calculations.
that
The only difference, of course,
the size and material
it
is
is
is
is
is
is
is
the sphere, but similar relations can be obtained for the cylinder and the
rectangular-block
reactors.
We begin with
study of the reflectedsphere system in which the extrapolated
thickness (Ri Ro) of the
reflector
transcendental
properties are here related by
may be used, consider the following situ
ation: A spherically symmetric reactor ol core radius Ro and outer reflec
tor radius /?i
to be analyzed.
The reflector material has been selected,
To illustrate how
equation.
REFLECTED
SEC. 8.2]
REACTORS
425
above does not apply well in this case, but the qualitative behavior is
correctly presented.)
Now, for small values of the argument,
the hyperbolic
is
cotangent
coth x =
(8.12)
we assume that Dc ^ DR, then the application of the first term in the
above expression to the criticality equation (8.11) yields
If
8.1
Comparison
(8.13)
and Reflected-sphere
of Bare-
Flux in core
Definition of B*
- (,S/ -
Geometry
A' sin Br
This
Ro
transcendental
Ro
Ri
0.
Ri
s,
(8.14)
in (8.13) gives
BRo
when
Ro) + 1]
is
BRo
cot
The
- S.]
- 1 - - UkRo coth
and
Ro
1.
therefore,
e)
[X/
(ir
We require,
S.)
- -r
state
sphere
^4sin Br
where
Reflected
Reactors
~
R\
Ro
(8.15)
relation
is
Bare sphere
BRo^t.
the
bare spherical
REACTOR ANALYSIS
426
[CHAP. 8
reactor.
definition
be
Rao',
then by
Ro
(8.16)
where Ro is the core radius of the reflected reactor having the same core
If we apply the requirement that xr{Ri Ro) <SC 1, then
composition.
the criticality relation takes the approximate form~~
- (-
an
Dc
Ro
\Ri
Ro
l)}
(8.17)
-r~-^
Dc\f
BRo
BRo
If we
that T /Ro
l)
<C
1,
with
f - R,-
Ro
(8.18)
for Ro yields
We know from our relations for the bare spherical reactor that Roo =
therefore, the substitution of (8.19) into (8.16) gives
Reflector savings
In many
thin reflector
r/B;
(8.20)*
Dc
DR
Ro) >s> 1 and furthermore that xrRq ^> 1 ; i.e., both the reflector
xh{Ri
and the core size are to be much larger than the diffusion
of the reflector material.
Because of the former requirement,
thickness
length
coth
xr(Ri
Ro)
1,
- ^ (XsRo +
1)
by
(8.21)
RKFLECTED REACTORS
sec. 8.2]
427
is large and
negative, and therefore the approximation (8.14) applies for the left-hand
side. It follows that
T
DCLR
and by (8.16)
Reflector savings
DCLR
thick, reflector
(8.22)*
Reflector
Fig.
thickness
Reflector savings.
8.3
Thus let
by means of the albedo concept.
us denote the albedo of the spherical shell reflector by a, where
ot
This number
(8.23)
function of the inner and outer radii of the shell (R\ and
transport and absorption cross sections of
In general, a is also dependent upon the angular
the material of the shell.
distribution of the incident neutrons; in the present analysis we select
the distribution which is consistent with diffusion theory.
The one-velocity flux in the core which satisfies the differential equation
,
(8.5a) must now meet the boundary conditions
is a
(1)
(2)
4>c(r)
j^(Ro)
for r
a??{Ro)
< Ro
(8.24)
The solution of the core equation which satisfies condition (1) in (8.24) is
given in (8.8).
The substitution of this expression into condition (2) in
(8.24) yields the relation
Dc
Ro
(1
- BR0 cot BR
~*
(8.25)*
>-K^-:)
+ 2DC
BR0 cot BR0 =
R0 and
for
Ra
Dc
428
If
REACTOR
the concentration
ANALYSIS
[CHAP. 8
Critical core
radius Jf
Fio.
Critical
Fio.
8.5
Fio.
8.6
Cylindrical
reflector.
reactor.
Sec. 8.6 by means of an approximate
procedure.
B2<pc(p'z)
Mp'z)
*c(p'z)
tp
\tp
(a)
(8.26)
c. Cylindrical Geometry.
The general problem of a finite cylinder of
multiplying material completely surrounded by a reflector cannot be
For the present we will consider
easily solved analytically (see Fig. 8.5).
two less general cases.
The completely reflected reactor is treated in
that
dependent component;
4>c{p,z)
*aM
The substitution of
4-
dp
[pF'c(p)}
-B2
(a)
\\
Gc(z) =
AJ0(\cp) +
(8.30)
A2Y<,{\cp)
CJo(\ap) + C2K0(\ap)
cos p-cz + Ai sin p.cz
Fa(p)
as follows
Fc(p) =
Ga{z)
= 2
(8.29)
G^W ~
^
_ Mc
GcW ~
=
s^Wlwr
pF*(p)
d sin
(8.31)
nRz
is
0.
cos^^=
a(po,z)
C3 cos
be satisfied
(u^Ci/o(Xpo)
AiJo(\cpb)A3
Mfi
If
CiK0(\apo)]
(8.33)
AJ o(Xcp)
We now satisfy
condition
cos nz
c(p,z)
3,
C2K0(\rp)]
(4) in (8.4)
32.
if
4>c(po,z)
interface.
core-reflector
and
(8.31a)
(8.316) yields
tion
is
0,
, .
is
"c
-xi
In
CrC(z)
=
~
Gc(z)
-ApF'cW
pFc(p)""cv"J
It
GR{t)
fied only
if
and
,R o7)
Fab)
these relations
Vrr
fit c(p)
Since
= Fc(p) Gc(z)
M|
429
X2,
REACTORS
is,
REFLECTED
SEC. 8.2]
cos hz
single constant
,g
REACTOR ANALYSIS
430
[CHAP. 8
Thus, over
4,R{pfi)
cos y.K
In any
"
cos nK = 0
(8.36)
that the difference in extrapolation distance for the core and reflector
would be difficult to take into account, since we have already required
Equation (8.36) can be satisfied only if
that the p's be the same.
so
(8.37,
Cl
If
we call
^
L0(p)
<pc(p,z)
<P(p,z)
(8-38)
~7o(AW~
/o(XfiPi) Xo(Xsp)
/o(Xp) X0(XRp1)
o(Xcp) cos
= CLo(p) cos
curved
(8.39)
TTZ
(8.40)
DcAJ'0(\cpo)
= CLo(po)
= DnCL'0(po)
CR4n
lS'"'
of
DcJ o (XcPo) _
o(Xcpo)
DrL'q
(pq)
Lo{po)
sec. 8.2]
REFLECTED REACTORS
431
y()
" 4-
(0
(8.43)
be computed from
(8.44)
required to maintain
(8.45)
case may be
over here through the separation of
variables since Eqs. (8.26) through (8.28) are
valid for the present calculation.
However,
it is convenient in fitting the present bound
taken
reac
ary conditions to make a different choice for the algebraic sign of the
constants.
Instead of using relations (8.29), let
pFc(p)
[PF'C
()]' =
goto
Gc(z)
-\%
-4
i
(8.46)
Gr{z)
REACTOR ANALYSIS
432
It
-\\ +
= B2
n%
Gc(z)
n%
G(z) =
Aa
If the
m 0.
(8.47)
AJofrcp) + AtY0(\cP)
C,Jo(XP) + C2Ko(Xp)
A3 cos ncz + Ai sin mc2
C3 cosh hhZ + Ct sinh p.z
Fc(p) =
F(p)
[CHAP. 8
(a)
(6)
(8.48)
(c)
(d)
\c
= Aft = X
(8.49)
C3 =
If we use
C4
tanh Ma
(8.50)
the results so far derived in the expressions for the flux, we have
AJ0(\p)
c(p,z)
<Pr(p,z)
= (Vo(Xp)
cos ncz
sinh p.(a
(g
51>
2)
The flux is also to vanish all along the curved surfaces ; thus,
(pi,2)
and it follows that
= <MP1,2)
= 0
(8.52)
c^
M^v
AcoB
sinh pft(a
Mcft
h)
h)
nch = Dp. coth p(a
'cL
(g 53)
(8.54)*
REFLECTED REACTORS
sec. 8.2]
433
Fig.
rec-
8.9
Block
reactor with
slab-reflected
ends.
dx2
c(x,y,z) +
^ c{x,y,z)
are
4>c(x,y,z)
+ B2Mx,y,z)
d2
dx2
d2
dz2
= 0
(a)
= 0
(b)
x\R{x,y,z)
4>c(x,y,z)
4>R{x,y,z)
(8.56)
4>R{x,y,z)
Assume
(8.57)
in
=
=
-B2
(8.58)
V2
Choose
F'cix)
Fc(x)
F'R\x)
Fr(x)
-xf
M?
G'tiy)
Gc(y)
GR(y)
-M
"Ml
X2
-Ml
Hc(z)
HR(.z)
(8.59)
REACTOR ANALYSIS
434
[CHAP. 8
x!
xl
xr
B:
(8.60)
The general solutions for the various functions in (8.59) may now be
written (taking account of the symmetry requirement to eliminate the
sine functions)
Fc{x) = A \ cos \\X
Gc(ij) = A2cos\2y
Hc(z) = A3 cos Xjz
Fr(x)
Ci sinh mx +
GR(y)
cosh mx
(8.61)
HR(z)
be continuous
across the
is,
Ei
= C2 cos my
=
cos n&
c(a,y,z)
R(a,y,z)
(8.62)
where Co =
\iy
A'i/Ci,
cos Xjz) =
C'(sinh
set_
we
X3 = Ms
X2 = M2
(8.63)
In order to meet the requirement that the flux vanish at all points of the
outer boundary of the system, we must have
X2 =
m =
7T
7T
X3 = 113 =
26
26
"%
= tanh /xid
(8.64)
extrapolated
(Again
is
the assumption
cos
cos
(8'65)
^Z
sinh m(3,
^-
cos
^|
B(x,y,z)
x)
fa{x,y,z)
These solutions
ix\{3,
a)
sinh
tan
Xi<z
= Drmi coth
m(cl
and
cos Xia
a)
^>DcXi
be continuous
L j
cos Xia(cos
or
(b)
(8.66)*
REFLECTED REACTORS
sec. 8.2]
435
(8.606)
(8.666)
Gi)2
^/
+
(868)
()'
Fig.
8.10
all the boundary conditions with functions in which the variables have
been separated.
An approximate technique for treating these problems
is discussed in Sec. 8.6.
e. Criticality Relation.
is,
it
is
is
during slowing down, since the flux preserves its spatial form at all
reflected reactor the spatial
energies.
However, as we have noted, in
dependence of the flux will in general be different at different lethargies.
In
is
ANALYSIS
REACTOR
436
[CHAP. 8
relation1 is valid
-DcV^dr) + 2r*c(r)
*Zsp&2+c(r)
(8.69)
In the usual
way, we define
neutron production
total neutron
where
losses
di
fc cfr)
2/Pggg
_ Dc
denotes integration
^ v>c(r)
dr
dl + s,c,
^ 0c(r)
The volume
integral of the Laplacian term, which gives the neutron losses from the
We have from
core, may be computed from the differential equation.
(g
0c
71)
with
fc
is
vc
(8.72)
Z*>]
v;
the system
critical,
in the usual way; i.e.,
[*e2/PS0g
If
B2(f>c,
then VJce =
fc
B* m
If
|^2gg*
2gj
v^c +
(8.69)
system
1.
vc
fc
where
is
Vc
required to maintain
the
sec. 8.2]
REFLECTED REACTORS
437
Sphere:
2.
-Dh[hrR0
Cylinder:
1)
coth xh(Ri
1]
DrL'0(pq)
La(po)
o(XcPo)
/o(Xp)i;Co(XPi)
u)
\*c
'
X2
end-reflected:
4-4 +
B-
fx = 2.405
-*
(*)' h)
Cylinder:
3.
"
Io(\rPi)K0(\kp)
B2 =
Ro)
side-reflected:
DcJ'o&cpo)
L0(p) =
[CHAP. 8
REACTOR ANALYSIS
438
It
Let
of thermal neutrons:
-Z)VV(r) +
2a(r)
G(r;r') dt'
fc 2^(r')
(8.74)
(8.75)
correct (within the limitations of the thermalif the proper kernel G(r;r') is chosen for the
diffusion model)
In general, G(r;r') depends upon the highsystem under consideration.
energy cross sections of the reactor materials and the geometry of the
group
medium.
infinite-medium
kernel of Eq. (6.38) :
k(r)
G{t;t')
Reflector
Reflector
It
slowing-down
(8.76)
distribution
<fc
large
in
sec. 8.2]
REFLECTED REACTORS
439
will
be made in extending the integral of Eq. (8.74) over all space when
describing the flux distribution in the core; therefore, take
-DcV^dt)
G{r;r') dr'
fm
(8.77)
-Z)W(r) + 2>(r)
*(r) + (r
- r')
- r')
v>(r) + *[(r
V]V(r) +
(8.78)
The substitution of this expression into (8.77), along with (8.76) for Gx,
yields the following results for the three terms:
vZ,
vZ,
jm
fm
(r)
GK(r;r') dr'
fm
(r
r2/Ptll(r)
= 0
=
(8.79)
^/rthptllW(r)
Since the point r is in the core, the fluxes on the right of these equations
The application of these results to the core equation
are core fluxes.
yields
(Dc + v2/P,hrlh)V^c(r)
If
we define
&
- n+
where
fc
"
[vZ/Plb
-^
33
2<C)]c(r)
= 0
rfer- |s
(8.80)
(8-81)
(8.82)
equations
for the flux in both the core and reflector are now identical to the set
(8.5), it follows that the multiplication constant is given by the expression
fc
=
2[1
Jfc.ro)]
(8'83)
440
[CHAP. 8
REACTOR ANALYSIS
seen
gfc.
ga
e~B^
If
*2fP*
2<(1
L^J)(1 +
JSVu.)
I +
(8.84)
Tth
been used
>Z/P*
2ic'[l + B\L%+
rth)]
directly
(g 85)J
^
L%
8.3
To, as
dA =
-D frV2dr
(8.86)
It
is evident from this expression that so long as the computed flux >
gives a reasonably good picture of the spatial distribution of neutrons
over most of the region, fairly large errors can be tolerated over small
portions of the region, for example, near boundaries (see Fig. 8.12).
The present argument, then, is based on the fact that the net-current
condition is an integral-type relation and requires simply that a certain
volume integral of the estimated flux approximate closely the integral for
The continuity-of-flux condition, on the other
the exact calculation.
hand, although physically acceptable, is not always the best condition
sec. 8.3]
REFLECTED REACTORS
441
to
Of course, this
Calculated
flux
Interface
/-Actual flux
large
at an
we discard
Fig.
r
8.12
Flux distribution
around
interface.
conditions
(1)
(8.87)
REACTOR ANALYSIS
442
[CHAP. 8
r-r'
Coordinate system.
of coordinates.
Sec. 7.4c
a Vtf(r,G) + 2(r)
Generated for wjivans (University of Florida) on 2015-09-22 17:21 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
It
viously in
where
*(r)
fv
4>(T,a)
c(r') S,(r')
S(r) c(r)
(i)
(7.245)
(8.88)
The quantity
is
given by
c(r) =
and the effective absorption
cross section by
2(r)
- S,(r) + ?
/3
is given by (7.196)
(7-236)
>Z/(r^Z'(r)
with
(7.246)
(t)
is defined
- rT
e-/S(r';r)
47r|r
(8.89)
2 replaced by 2.
is in the
4>{r)
an
*(r)
= ccZ((C)
G(r;r') di' +
fc c(r')
c2
G(r;r')
fR (r')
dv'
(8.90)
REFLECTED REACTORS
sec. 8.3]
443
This relation gives the flux at any general point r in the system in terms
of integrals involving the flux in both the core and the reflector.
The
notation ( )c refers to properties of the core and ( )n to properties of
the reflector;
Jc
jR denotes integra
of
c,
we have
Cc '
vXf +
2(C>
'
_
=
CR
2(ft)
(8-91^
Sp
/3(r';r) = 2|r
r'|
(8.92)
is
c*2>
w,
>
p(r')e-fc-5*(''
dr'
*R(r
)e
MXr,2
]r f
<br(i')e-fcr' dr'
ccZP
jc f
MO)
(8.93)
We are dealing
(i)
(r),
with
here
a spherically
With
symmetric
system,
these substitutions,
so
that
(8.93) reduces
Jo.
J"
(8.94)
Thus
dr
[fQ" c(r)e-~zc
c2^'
4,c(r)C-^ dr
]
= cc2<
<f,c(0)
1=1Jo
Jo
<Mr)e-fc-2K(r-a) dr
("
form
be convenient
c2JR>
f~
will
It
Mr)e~fcr dr
= cc2ic
c(0)
to
0(r)e--^f<'-) dr
(8.95)
For the functions <c(r) and ^s(r) which appear in (8.95) we use firstorder approximations of the solutions to the transport equation.
The
basis for this choice
derived from the following argument.
It was
shown previously in Sec. 7.4d that the elementary solution to the oneis
c.
infinite multiplying
444
REACTOR
e*8-',
[chap. 8
ANALYSIS
B is given by the
it is seen that the
B2
f}
in
B real
= 0
(5.161)
Thus the
in Eq. (7.256) and the subsequent discussion.
general solution of the integral equation (8.88), which can be constructed
from the elementary solution [see, for example, Eq. (7.255)], will also
as demonstrated
(r)
(8.97)
be a satisfactory
^(IH
wherein we have introduced
the quantity
(898)
defined by
+ Z'
(8.99)
Note that
c2 = c2,
(8.100)
It
is seen from Eq. (8.98) that the requirement that B be real implies
that c > 1, which corresponds to a multiplying medium [see (7.236)].
(In the present context, multiplying means capable of supporting a selfsustaining
plying,
1
<
REFLECTED REACTORS
SEC. 8.3]
445
tenh-K
w.th
k m
2/
The corresponding expression for the flux is obtained from (5.161) and
is seen to be
V2
- x2
c and
= 0
x real
(8.102)
4>R,
with B
(8.101).
It will
= c(r)
<p(i)
Thus, physically,
<p is
(8.103)
then
(8.104)
(8.105)
it
Material
-1
<fi
given below:
<p are
and
0 1
<p
is
1,
only
has the customary definition.
and
Note that
>
>
the average number of neutrons produced per absorption (tj)
greater
than unity. These parameters may also be defined in terms of the effec
where
if
It
is
g/
<p
was
tive absorption cross section 2. The appropriate definition for
given previously in (8.99). The corresponding expression for the net
neutron production per collision
then
(8.106)
+2S-
?2
= ^2,
-"
(8.107)
the solution
The asymptotic form to be used in the core region
The
solution
for
the
reflector
Eq.
of
corresponding
region
(5.161).
(8.97)
is
446
REACTOR
[chap. 8
ANALYSIS
is given by Eq. (8.102) ; the only solution to this equation which is spheri
Thus for the present
cally symmetric is the negative exponential e~'T/r.
analysis we select the following asymptotic expressions, denoted by the
symbol
*~c
where
- =^
(8-108)
These are
is to be computed
conveniently
summarized
=
with tan =
l + fc
tSE=r--1 + -
(8.109)
(8-110)
**'mm
Note that, if the material composition of the core and reflector is specified,
the parameters B and x may be obtained directly from the above tran
scendental relations.
The quantities w and k have been extensively
tabulated1 as functions of <pc and <?r.
The expressions (8.108) are now applied to Eq. (8.95). First we note
that
must satisfy the integral equation (8.88) ; therefore
*?(r)e-Scr dr
j'
=
c*S{S)eJ
0?(r)e-J<^ dr
(8.
% into
1)
cc2Sc>
1
1
(8.95) along
,
*g(0)
jQ"
0J(r)-!' dr
(8.112)
is
is
[Note that the use of (8.111) allows us to cancel the term at the left of
The equation (8.112)
the
(8.95) with the first term at the right.]
Serber-Wilson condition.
It based on the requirement that the solution
be satisfied
the system
the reactor =
is
if
is
is
it
for the flux satisfy the transport equation at the center of the reactor, and
replaces the condition of continuity of flux at the core-reflector interface.
The second boundary condition to be used
the requirement that the
net current be continuous at the interface.
For the present calculation,
this condition
more conveniently stated in integral form; thus,
(8.113)
to be at steady state.
5, a
I,
- ^~
REFLECTED REACTORS
SEC. 8.3]
447
However, it applies with equal validity to noncritical systems (nonstationary systems) since these systems can always be described by
means of an equivalent
steady-state
problem
through
fr p(r)
This integral may
be separated
= 0
(8.114)
fc *cW
dr =
-?R2fi
Jr flj(r)
dr
(8.115)*
This integral equation along with (8.112) provides the boundary con
ditions to be satisfied by the solutions 0.
d. Criticality Equation.
The criticality equation for the spherical
reactor with infinite reflector is obtained by dividing the flux condition
Before performing this opera
(8.112) by the current condition (8.115).
In
tion, we first evaluate the integrals involved in these expressions.
for
and
will
be
used.
every case, the asymptotic solutions (8.108)
<pc
4>r
The first integral to be computed is that from the left-hand side of
r>
sin Br dr
Bl
(sin aB
- aB
cos
aB)
(8.116)
we obtain
re~"r
= 4*
Jo
f (r) dr
Jc
*j[
(8.115):
dr
(ax
+ l)e-M
(8.117)
f"
A\
e-
is
e~lcr sin
Br dr
AEi(alR
ax)
(8.118)
p-Za'-*r Ar
T
-Im
Ei(2ca
where Im
+ iBa)
/n
E\
f"
(8.119)
Pc2c
ANALYSIS
REACTOR
448
[CHAP. 8
cality condition
ccZSe-c
<Pc^c
Im Ei(Sca + iaB)
"y^, (sin aB
aB
cos
cifS<*(ax)21(a2
<pHlR{ax
aB)
+
+
ax)e<-*+"
1)
(8.120)
This result was obtained by both Wilson and by R. Serber and gives the
critical radius of the core for a specified core and reflector composition
We will discuss this relation in more detail shortly.
(or vice versa).
1.5
Fig.
8.14
2.5
3.5
4.5
and
aB = a2c tan
we introduce
to
= x tan
o>
the quantities
= y
(8.121)
Some of the functions on the left side of (8.120) have been tabulated.1
It will be convenient, therefore, to rewrite the criticality equation in
terms of the tabulated functions (see also Fig. 8.14):
S(V) =
Q(y,o>)
1
Ibid.
yl (sin
-y
cos y)
- - e-"-" Im El (2ty
\tan
(sh + *)
o}
co
(8.122)
REFLECTED REACTORS
sec. 8.3]
If
449
S(y)
We have used
wn
<Pr2iR
for
(8,123)
to obtain
ax =
= nf
<c2a
(8.124)
In every
only the 2 terms which appear in the preceding analysis.
instant 2 becomes 2(, and the results derived above take the forms sum
marized below.
The criticality equation (8.123) becomes
<pc
S(y)
<pH
Xpa
Equation (8.125)
also,
ta^
fc
with tan
+ ^
are now
u>
with" =
= 2iR,a
and
and
*~an
J^r.
(8.126)
(8-127)
spr)
= x tan u =
aB
(8.128)
is
it
is
f,
so
Tc(x,<pc)
= TR([,<pK)
(8.129)
x
is
is
is
f,
Thus, given
to find
value of
known.
The problem, then,
TK
such that Tc = TR. If the core composition 2p
specified, this value
a
terms of our present notation as the special case when the reactor period
This corresponds to a > 0 and the effective absorption
1/a is infinite.
cross section 2 2, [see Eq. (7.246)].
Formally, this change affects
450
REACTOR ANALYSIS
[chap. 8
Figure 8.15
tions for a range of reactor compositions of practical interest.
is a plot of Tc(x) for various values of <pc, and Fig. 8.16 is a plot of T(f)
for various values of <p. It should be noted that x and f are related by
Fig.
8.15
2 P a)
Core properties.
the ratio of the total cross section in the core to the total cross section
in the reflector; thus
X
*f
=
2p">
(8'130)
REFLECTED REACTORS
sec. 8.3]
451
The critical core radius x (in mean free paths) has been computed for
some combinations of <pc and <pr which may be encountered in reactor
These data are presented as a function of d, the core-reflector
design.
For these cases, the critical size can
cross section ratio, in Fig. 8.17.
be obtained
are given.
above can always be applied.
Fig.
8.16
(()
Reflector properties.
P2C
= 3^c
and
- - Prx
0
0%
= -3<pR
(8.131)
(8.132)
The flux distributions for the case <pc = 0.2, ipR = 0, and 6 = 0.8 are
shown in Fig. 8.19.
It is of interest to note the discontinuity in the flux
due to the application of the Serber-Wilson condition.
Also, note that
the critical mass computed by the continuity-of-flux condition exceeds
the Serber-Wilson estimate by as much as 40 per cent for small reactors.
aoiovas sisaivnv
Z9f
SIO
ZO
VP
90
80
O'T
<ivho]
']
OE
UI)L Xjir
ni >-
B
3
a
o
UOS|l/\
91
ZQ=
0="
o
20
so
'olA gl"8 uosueduiog jo sijnsoj uiojj uos|i;\Y-joqjog uoiiipuoo pun uiojj -X^mupnoo
xng-jo uonipuoa
sec. 8.3]
f.
453
REFLECTED REACTORS
Noncritical Systems.
We return
0.5
2.0
1.5
1.0
Radial coordinate
Fig.
form (7.244).
by integrating
8.19
2.5
3.5
3.0
4.0
[rZ|]
Flux distributions.
tf>0M)
thus
=
(8.133)
4>{i)e'
is yet to be determined.
In dealing with the nonstationary system we found it convenient to define an equivalent steadystate system by introducing the effective cross section 2.
The relation
where the period
l/a
which determines this cross section, and consequently the reactor period,
is Eq. (8.123).
Our purpose in the present calculation is to obtain a
procedure for computing a from (8.123) and, subsequently, the multipli
cation constant k of the reactor.
It will be convenient at this point to work, not with a, but with another
parameter y which is related to a through the equation
y =
;J
(8.134)
Sf>(l
+7)
or
I=
+ 7
(8.135)
KEACTOB ANALYSIS
454
r =
i +
70)
[CHAP.
0,
= s,(S,(i
+ ye)
(8.136)
- 7)
-
2f >(*>c
= 2<'(^
(8.137)
yd)
1+*?
^'
and
1+^L
(8.138)
+
= -j
1
1
<pc
and
1+7
w^; "^r
be
k
r
j
1
tanh
written
<pr
+:
+ yd
1
z
1
- VtT^;
,Q 1QQv
(8.139)
(8-140)*
It
x =
x(l +
7) =
r1
tan w
(8.141)
These values of y and f are then inserted in (8.140), and, if they satisfy
the selected value of 7 is the solution for the system in
the equation,
SEC.
REFLECTED REACTORS
8.3]
455
[cf. (7.245)].
V*c(r,a) + 2c*c(r,0)
dr
jaa- Vc(r,a)
dQ + 2e*c = cc2,(C)*c
L c(r)
*c =
where
-L 2pcc*c(r)
(8.142)
Jc
dr
The
(8.143)
(8.144)
is denned by
production
leakage + absorption
(8.145)
A; j
1,
form
=
If
ccZjC)3>c
leakage +
2*c
ccZf^g
leakage + 2c^c
- yZf^c
cc2c
k =
cc
of this expression is
-y
(8.146)*
i=
<Pc
2ca
= 3.0
= 0.3
<pR
2o
-0.03
= 3.3
= 0.91
REACTOR ANALYSIS
456
[CHAP.
8.2.
Table
8.2
Function
to
t
y
Q(v,*>)
S(y)
f
E.lfd +)]
Tr
Tc(x,fc)
Tc/Tr
8.4
The corre
Constant
y = 0.130
y = 0.142
y = 0.150
0.617
0.599
3.39
2.41
0.0674
0.424
3.69
4.04 X 10~J
1.468
0.748
0.510
0.596
0.615
3.43
2.32
0.1066
0.425
3.73
0.579
0.620
3.45
2.25
0.1375
0.433
3.75
3.29 X 10-'
1.360
1.590
1.168
3.5 X
10-*
1.405
1.230
0.895
a. Reactor
The analytical model of a reactor which
Equations.
describes the neutron population as two separate groups of diffusing
neutrons, one fast and one thermal, has a great deal of usefulness in
reactor calculations.
To begin with, it is sufficiently simple from an
be analyzed
The solutions
REFLECTED REACTORS
sec. 8.4]
457
-D,V2^,(r) +
Fast group:
+ 24>2(r)
v2/*s(r)
pc2^i(r)
(a)
(8.147)*
(b)
Di and
2^' which
tion cross section in the core, respectively, computed in the usual way.
The source term for the thermal neutron equation (6) is taken to be the
number of neutrons slowing down out of the fast group, reduced by the
number lost in actual absorption during slowing down.
We define
is
is,
Thus, we
following
Fast group:
Thermal group:
-D,V24(r) + 2*(r)
Z>3V23(r)
+ 2J?*(r) =
set of equations:
quantity 2^'.
For the reflector
(c)
(8.149)*
= pA2g>0,(r)
(d)
The parameter
The quantity
2(i'<#i(r)
REACTOR ANALYSIS
458
[CHAP.
2>*2 = pc2#>*i
vXffr
If
(8-150)
and
(8-151)
*i(r)
- *iF(r)
'
*,(r)
and
*F(r)
(8.152)
have required that both the fast and thermal fluxes vanish on the extrap
olated boundary surfaces of the reactor; the extrapolation distance Is
assumed to be the same for the fast and thermal groups.
The application of the above relations to Eqs. (8.147) yields the set
Dtftfr
As before
SJ}>0,
we take the
the constants
and
ratio of
D2B* +
be
these
2?>*, = pc2j>>*i
equations
(8.153)
tf>2 obtain
vZ,
This may
D2B^ +
v2f*t
_ Drf* +
pcZ
2f
(8.154)
2(1
+ L\B%{\ + L\B*)
With L? ~
(8-155)
REFLECTED
sec. 8.4]
459
REACTORS
If
we compare this result with the Fermi age analysis of the bare reactor
(6.78) for the case wherein there are no fast fissions ( = 1), then we can
immediately identify the factor (1 + L\B2)~l as the fast nonleakage
probability; thus,
+ LIB1)-1
(1
For
(1
(8.156)
and
L?~rth
m e~B^
Zg>
that
(8.157)
This is one way in which one could define tfieTemoval cross sections 2.
As experience
would certainly be an adequate first approximation.
is gained from the comparison of theory and experiment, better estimates
can be made for this parameter.
The fast-diffusion coefficients may be computed from a suitable expres
Thus we take
sion for the neutron leakage.
<0
It
Tth
D(u)du
Jo
Wth
MthTth
JO
(8.159)
D{u) du
(8.160)*
as
2g>
(8.161)
r^-
Thus the "mean free path to removal" from the fast group
*
_1_
2$
X.Mih
{
_ / mean
free
\for scattering
^collisions to thermal
is seen to be
\
/
The parameter
is usually calculated by choosing a
[see Eq. (4.47)].
fast-diffusion coefficient for the entire group and dividing by the age to
thermal as in (8.160).
Note that (8.147a) may be written directly in
terms of the slowing-down
density
thus,
=
vXffa
(8.163)
REACTOR
460
ANALYSIS
[CHAP.
General Solution.
as
described
In selecting these
as a linear combination
of two known functions.
functions we will be guided by our previous experience with the onevelocity model. The proper weighting factors (coefficients) for the two
functions in each combination will be determined by the requirement that
each general solution satisfy the appropriate differential equation from the
set (8.147) and (8.149) as well as the various boundary and symmetry
conditions imposed on the system.
We begin with the differential equations for the core.
Let Z(r) and
W(i) be two functions such that the following linear combinations give
the fast and thermal fluxes in the core:
as the solutions
V(r)
V2Z(r) + B\Z{r) = 0
(^
l0*)
to the equations
+ B\W(t)
= 0
(8.165)
and
B2 are to be chosen so
Z{t)[AiDxB\ + ZgAx
Z{r)[A2D2B\ + 2?A2
- vSfAi]
-+ W(t)[C1D1B*
pc2g>Ax]
+ W(t)[C,D,B\ +
Ci2<
C22
- C,w2,]
- pcC&tf]
=0
^'i0O;
= 0
S/
Ax = A
\DzB\ +
Zffl ~
_ Cl f pcS<
,2,
J
[d2BI +
\DlBl
f [DiB\ +
[[DiBl
+
+ 2g>][DsS| +
2>][D2Z??
2ff>"
(8.167)
Ai,
2]
2]
.pcS/S*"
= vpcZ/Zff
^,1D8;
SEC.
REFLECTED REACTORS
8.4]
Thus B\ and B\
are solutions
(B* +
461
of the quadratic
4)(B* +
k\)
equation
1!ff
(8.169)*
^.-^p
(8.170)
We find, then, that if the expressions (8.164) are to be solutions of the core
equations (8.147), with Z and W given by (8.165), the quantities B\ and
Bi must be taken as the roots of the B1 equation (8.169). These are
seen to be
82
--
^4-^
\ [w
k')2
4K?Ki(fc-
"
f /fV
^
(8-i7i)
4f"
It
B\ =
B\=
4kWK
!)]
>
(8-172)
K\
X2
K\
n1
X2
= 0
V*W(r)
- \*W(t)
= 0
(8.174)
and the
functions.
. _
_
PC*?
_- PcW 2
>2X2
iM2 +
^
-DiX2 +
*,(r)
=
=
(8.175)*
gjj>
vZf
*(r)
73
2%>
AiZ(r) + C,W(t)
0lAiZ(r) +aiClW(t)
'
>
^1'0J*
^>
REACTOR ANALYSIS
462
^jjj
gjS-
Vc''<7
[CHAP.
constants.
We consider next the reflector equations (8.149), which may be written
- S*.(r)
- S*(r)
V2*3(r)
V^(D
= 0
(8-177)
V(4)
V<3)
with
K's^
K'sfe
(8'178)
f^0
vjy pP6
*,(ft')
*(ft')
= 0
(8.179)
equations
V2C/(r)
- K|f/(r)
= 0
=
C7(ft')
and
- k\V(i)
V2F(r)
V(ft')
= 0
= 0
(8.180)
(8.181)
Now, the form of the fa equation is seen to be the same as the reflector
relation used in the one-velocity model [cf. Eq. (8.5b)], and it was found
that the solution to that equation could be given in terms of a single
function.
It
*,(r) = Sitf(r)
(8.182)
*(r)
- TiV(r)
The substitution of
Tt
Thus
<0
we have
-o,Si
+ 7W(r)
with a,
and
*,(r)
Sitf(r)
*(r) =
(8.183)
into the
<fn
rnrf2^
TiV(r)
on the
a3S1C7(r)
equation of
(8-184)
(8.185)*
d. Criticality Condition.
The general solutions (8.176) and (8.185)
give the thermal and fast flux distributions in both core and reflector in
terms of the functions Z(i), W(r), f/(r), and F(r). These functions, in
turn, are defined by Eqs. (8.174) and (8.180) and by the symmetry and
REFLECTED REACTORS
sec. 8.4]
463
(1)
(2)
(3)
(4)
i(R)
*,(R)
The application
<*A(R)
(8-186)*
dA(R)
(D,E/')Si
{-DiZ')Ax
(aJT)C,
+
{-V)Tx
(a3U)Sl
(oiZ)Ai
(-a,Z)2Z')A, + (-o,7),W')Ci + (-o,Dl/')Si
=
=
(a)
(b)
(c)
(d)
(8.187)
atW
aiU
-a3DtU'
-a&iZ'
This expression may
-atDiW
-V
-DxW
OiZ
-U
-D,Z'
is,
JJ
(8.188)*
DAV
as
it
be
way, the specification of the fuel concentration or the core size allows the
determination of the other through the use of the above relation.
Unlike
the simpler models and applications previously considered, however, the
somewhat more involved.
The details of an
procedure
procedure will be discussed later in connection with a sample
and some computational approximations which have been
computational
appropriate
calculation,
is
+ (D7')r,
=
=
(-W)C,
of these conditions
VaJ
j>
HEACTOK
464
ANALYSIS
[CHAP.
requirements
determined
above.
It
fuel concentration,
tation may be repeated, using the corrected values of the less sensitive
parameters.
Once a value has been assumed for the fuel concentration,
nuclear
of the
That value of k, call it kj,, which satisfies (8.188) gives the actual multi
plication constant k. Thus if kx corresponds to the experimental value
of v , then the multiplication of a noncritical system is given by
(8.189)
It is
Fermi
REFLECTED REACTORS
sec. 8.4]
for
as B
465
In the two-group
t/R
Pc
L\ =
where
+LV)U +LW)
(1
L\
and
-2
(8.191)
-2
Reactors.
however, the function cos n2y cos nsz appears as a factor in every term,
and since this factor is not identically zero, it may be factored out of the
determinant, which, then, becomes independent of y and z.
f. Comparison with One-group Method. Some idea of the errors in
the criticality requirements which can be caused by the omission of the
effects of the fast neutrons in a reactor can be obtained from the compari
son of the results based on the two-group model with those of the one-
Table
Summary
8.3.
As
model.
group (velocity)
a representative
Z'(R)
w(*)
W'(R)
cos
= M
iiR
sin /xfl
sinh Xi?
=
=
X cosh
XB
U(R)
W
sinh \R
R*
sinh
=
_
V'(R)
V'CR)
Kt cosh
sinh k,(R'
- R)
ks(R'
R
R)
sinh k}(R'
TP
R)
R)
<i cosh
R~
sinh Kl(R'
R)
- R)
REACTOR ANALYSIS
466
Table
Summary or Functions
8.3.
[CHAP.
(Continued)
Cylinder
=
Z(R)
Z'(R)
WB) =
W'kX)
J(liR)
(Side Reflector)
cos l,z
IJAUR)
cos Uz
h(l,R) cos hz
/,/,(i,fi) cos l,z
l/(R)
IT'(R)
[/o^K) KoM')
/o(i.S') #<,(/.)] cos /2z
U[li(ltR) KSaR') + h(Uft') Ki(UR)\ cos i2z
[IoihR) K0(hR')
/,(*,#') K0(lR)] cos In
hUiVJi) Kt(hR') + IoihR')
Ki(ltR)] cos 1*
=
=
Z(R)
Z'(R)
W(R)
2.405
m|
t/(R)
U'(R)
V(R)
V'(R)
X' +
A)
ml
h)
mi
m}
+ m|
h)
toi(<2
TO?
Block Geometry
cos nia cos nty cos n>z
Z(R)
Z'(R)
;
W(R)
R"(R)
H*
- nl - n\
n%z
ni
l/(R)
i/'(R)
V(R)
K'(R)
+ n + n\
n, =
f.
2C
11
25
= sinh ns(3
a) cos n2t/ cos n2z
= ns cosh nt(<2 o) cos n2i/ cos njz
= sinh n( a) cos n2j/ cos njz
= ne cosh 7i(3 a) cos n%y cos njz
J +
n\ + n\
= k\ + n\ + n\
Di
2}J>
= Z)2 = 1 cm
= 0.01 cm-1
pc = 0.90
2
2/
= 0.05 cm-1
= 0.0229 cm-1
Reflector: D3 = Dt = 1 cm
2J," = 0.01 cm"1
pR = 0.90
2< =
0.0049 cm"1
= 22 cm
REFLECTED REACTOKS
sec. 8.4]
467
Spherical
(^Critical ^ADrl of
pth(1 + L'B'>"'e"<',r", =
&' extrapolated
(^f,) Ml
reflector
1)
DC(BR cot BR
-D{*XR
>V')-'(1 +
76.8
(6.83)
187
(8.155)
1S7
66.5
(8.73)
(8.11)
(8.85)
coth
L.V')-'
.('
=
R) +
186
(8.11)
(8.188)
(8.171)
180
(8.190)
.(S'
- R)
D[*R coth
1)
two-group
BR
"
1]
oore, one-
oore,
L',B*)"'
1]
Pe(1 + Lc'B')~'
(cm)
0. Reflected
L''B'yH1
(^)
DC(BR cot
th
velocity (using
migration area)
pth(1
4. Reflected core,'1
one-velocity
Tw)
Critical
radius
(5.183)
outer boundary.
4,
is 2
is
identical results because of the large size of the bare core. It interesting
to note the comparative accuracy of the "migration-area" calculation,
which predicts, in this case, very nearly the same result as the two-group
method.
The form of the flux predicted
two-group
calculation,
yields a thermal-flux
is
lation,
4,
leakage)
+ L'B')
2^)
P'h(1
"
z!r)
5. Refleoted
Reference
equations
Critioal equation
Method
8.4
Table
where
distribution, which,
as
ANALYSIS
REACTOR
468
[CHAP.
shown by Eqs. (8.176) and (8.185) and Table 8.3, is a linear combination
of two functions in both the core and the reflector:
For the thermal flux in a spherical core,
Mr)
Mr) *(r)
c,
t(r)
(sin
r
a spherical-shell
[sinh Kt(R'
y.r
sinh Xr)
(8.192)
reflector,
(8.193)
where
4>r, Co,
however,
of the core).
This feature
determinant and rearranging
may
be
demonstrated
by expanding
the
by the func
REFLECTED REACTORS
sec. 8.4]
469
Z'
w 33
^"
a3
= 0
(8.194)
Z)2a2co
If
U'
^
77
(8 .195)
is,
where
-Dtf
a,
-DlU
A =
In this
D1Z)2co(a2
ox)
DJhufHfli
Z)!Z)/3(a2
gi) + D3DtW(ai
at)
a,) + ^{a^D^ + asDiZ)*)
+ a2D2P3)
a^(q3Di4
In many
is
is
is
<c2and
kK.
The computationaj_procedure
to be followed in using (8.196) consists
in selecting various values of the fuel concentration, computing the cor
X)
it
responding values of
of
For example,
particular value of
(8.197)
for
cot pR
is
in the case of
function of
y.
is
intersection
simple calculation
f.
is
(8.196)*
REACTOR ANALYSIS
470
Sample Computation
8.5
[CHAP.
Description of
which serves both as coolant and moderator, enters the pressure shell
through a set of pipes, as shown in Fig. 8.21a, and first flows downward
Control
rod
Pressure
shell
55.8
(a)
Fig.
8.21
(6)
idealized model.
The inlet water which first passes through the annular region surround
ing the core serves as a side reflector, and the water below and above the
The individual fuel elements, which are assem
core, as an end reflector.
bled vertically, consist of a box structure of parallel plates similar to the
The structural material of these elements is
design shown in Fig. 5.28.
stainless steel, and the fuel-bearing plates are fabricated from an iron
clad on both sides with stainless steel.
alloy of
U2S6
!>B8
of its constituents
= 0.110
vA
= 0.093
are
(8.198)
REFLECTED REACTORS
sec. 8.5]
471
where vw refers to the water, i'SB to the stainless steel, and vFA to the fuelbearing alloy.
Our primary interest in this calculation is the determination of the
critical mass of the hot clean reactor and the radial distribution of the
fast and thermal flux throughout the core and reflector.
An accurate
analysis of this system must necessarily take into account the completely
reflected cylindrical geometry shown in Fig. 8.216.
However, since this
would entail a somewhat involved calculation, we will approximate the
actual configuration by an equivalent reflected sphere of the same compo
sition.
This will reduce our computation appreciably and yet not obscure
A
any of the essential steps in the application of the two-group model.
study of the effect of the corners in the completely reflected cylinder will
until the next section.
For the purposes of the present calculation
be deferred
we assume
lent spherical system1 is one which has the same core volume as the
actual geometry and a spherical-shell reflector 17.8 cm thick.
On this
=
basis the radius of the spherical core is R
31.93 cm, and for con
venience we take the extrapolated outer radius of the reflector to be
R'
= 50 cm.
It
following
core we take:
Material
Water (450F)
Stainless
Alloy
Fuel
"alloy"
Density
0 . 832 g/cm
7.75
7.85
10.9
where vF is the volume fraction of the core occupied by the actual fission
able material and vA the fraction occupied by the diluent.
This notation
will be useful when selecting various fuel concentrations (i.e., values of
vF) for evaluating the criticality determinant.
1 Alternative procedures for
determining equivalent systems is given by F. G. Prohammer, "A Comparison of One-dimensional Critical Mass Computations for
Completely Reflected Reactors," ORNL-2007, Mar. 1, 1956.
REACTOR ANALYSIS
472
Nw
of the components
in the above
listed
These are
NA = 0.08443
N, = 0.02459
nuclei/cm3
1024
[CHAP.
X 10"
X 102*va
X 10"ty
(8.200)
The group constants for this reactor have been computed from the
methods described in Sec. 8.4b, using the data (8.200).
These are listed
in Table 8.5.
Table
Two-group Constants
8.5
Flux group
Core
Reflector
TC = 45.3 cm1
Fast
Dt =
1 .44
= 1.0
PC
y(l)
Di
Thermal
TK = 40O cm1
cm
- D,/tC
= 0.03179 cm"1
Dz
1.85 cm
PR
1.0
D,
= 0.229 cm
L\
- 0.04625 cm"1
- 0.204 cm
- 16.64 cm'
0.01226
*
The corresponding
concentrations
macroscopic
(8.200) are
2^'
2<a8)
s(1>
A1
TC
2)
#4
cross sections
Zf
= 0.152i>A
= 10.92y
(8.201)
k,
= 0.02208 cm-2
'"-=-=
2>
= 1.80 barns
= 444.1 barns
<r(aF)
= 0.01002 cm-1
= 0.01705
from the
of
= 0.4518 barn
= 1.86 barns
a?8'
be computed
cm-'
0.0250 cm-2
= 0.06010 cm"2
id = 0.1486
cm-'
k3
= 0.1581
cm-1
k4
= 0.2452 cm-1
(8.202)
REFLECTED REACTORS
sec. 8.5]
473
tion,
will not
For v,
Zf
2"'
= 0.01322 cm-1
2,
thus
= 0.1058
=
k\
= 0.05496 cm-1
2(2)
fc
(8.171).
= 0.4620
^%f^
cm"'
cm"!
= 1.278
equation
(8.169), using
= 0.07606
X =
0.6999
(8.204)
(8.203)
= ijY =
=
0 =
4f
V
=
=
be computed.
- 4 0.6686
-0.1905
-k3 coth Kf - 4
K
- K -0.2766
-** coth
X
coth
\R
kaT
(8.205)
Dlfi* +
at =
'
-4.983
(8.206)
- oi) -
+u(a3DiDt +
ft
>1D2o,(a2
a2D2D3)
DiD^jai
DiDtPia* +
a*)
+
+
o3)
D3D^<3(ai
iK^D,/), +
-0.1598
The function
the
- oj
a3Z)1)4)
(8.207)
we compute from
= ~ = n cot tiR
-i
-0.1193
(8.208)
REACTOR ANALYSIS
474
[chap.
-0.3,
Figure
volume
fractions
and
9 over
The solution
a range of
fuel-
to the present
problem is seen to be
(vF)ianl = 0.006425
It
f and
8.
Evidently,
M,
p,vFV
vF.
Thus
= (10.9)(0.006425)(136,400)
= 9.55 kg
SEC.
REFLECTED REACTORS
8.5]
2/'
= 0.01316 cm-1
oi
475
= 0.07016 cm"1
kK = 1.316
4 = 0.4820 cm-2
=
= 0.2842
-4.859
a2
o3 =
=
=
0.08057
X
0.7146
/x
fa
The constants A
i,
cm-'
= 0.1104
(8.209)
-6.461
fa = AXZ +
=
be
dW
S,U
fa
Ci, Th and S\
= a^AyZ
+
TyV - a^U
fa
=
a^CyW
(8.210)*
have already
we
procedure is followed in the case of Eqs. (8.187a) and (8.1876), the fol
lowing relationship is obtained between C\ and A\\
c.=f(|^:)^
The functions which appear in this expression are given in the notation of
Eqs. (8.195), and in the usual way, these are to be evaluated at the coreThe result (8.211) may be used to determine Si
reflector interface.
from the relation
St =
+ WC1)
jj{ZAl
(8.212)*
Ti
+
^ [AxZ{ai
a3)
Cjr(a2 +
a,)]
(8.213)*
These three expressions, then, give Ci, Si, and Ti in terms of A\.
We now apply these general results to the present problem.
For this
This value of Ai
purpose we use the data (8.209) and let A\ = 12.41.
ie.,
Ai
= 12.41
Si = 0.6986
Ci = -1.548 X
Ti
lO"10
= -0.8135
we obtain
A\
by
(8.214)
REACTOR ANALYSIS
476
40
30
20
10
[CHAP.
Radial distance r, cm
Fig.
8.23
Radial distribution
sphere
reactor.
The substitution of these results into the flux relations (8.210) yields, for
the radial distributions,
*<r) =
12.41
S-^-r
r
Mr)
= 3.527
4>t(r)
= 0.6986
*(r)
1.548
X 10-
7.552
sinh k3(R'
-0.8135
10(8.215)
r)
sinhK<(^-r)
r
4.513
sinh
*(/*'
r
- r)
It is of interest to note
These functions have been plotted in Fig. 8.23.
the peak in the thermal flux near the core-reflector interface.
8.6
Completely
Reflected Systems
a. One-group Model.
A general analytical solution of the completely
This creates no particular
reflected reactor is presently not available.
In these
hardship when only a rough reactor calculation is required.
which is
cases the effect of the "corners" in reflected configurations,
generally small, can be entirely neglected since only a very small portion
Thus to ignore
of the total neutron population occupies these regions.
the corners introduces small errors into the critical-mass, calculations
even though the estimated local flux distributions may be in considerable
error.
When more accurate critical-mass estimates are required, these
sec.
REFLECTED REACTORS
8.6]
477
In the dis
concept mentioned previously.
cussion which follows we describe two such methods which have been
applied with some success to the one-velocity model and to the two-group
model.
As a first example, consider the analysis of a completely reflected
"equivalent-bare-reactor"
Fio.
8.24
model.
A typical
pi and height
2a.
III
REACTOR ANALYSIS
478
[CHAP.
the actual system) and having the dimensions shown in Fig. 8.25, II. Thus
the value of I is determined by requiring that the two systems shown in
Fig.
8.25 be
In this approxi
for the com
Fig.
1.
8.25
B,1,
B"
<8-217)
^-
value of
(8.218)
of system
I.
sion for
be
I:
where
is
b;"
will
Gi)'
thus, compute
III
()'
3.
With the
where Xc
is
mation
given by (8.42).
of the actual
SEC. 8.6]
REFLECTED REACTORS
I,
479
we set
(8.220)
i-[2,-
Z<]
Bl
(8.221)
It should be noted in using this procedure that there does indeed exist
an I for which Eqs. (8.219) and (8.220) hold accurately; however, the
value obtained by steps 1 and 2 is only an approximation, and one whose
validity could be ascertained only with great difficulty.
b. Two-group Model.
The general approach used in the case of the
one-velocity model is easily extended for use with the standard two-group
method described in Sec. 8.4.
As in that calculation, the primary objec
is
tive here
the estimation of the reflector savings in the direction of each
axis of symmetry of the reactor.
One scheme which has been devised
carried out with the aid of the two-group model, and each calculation
yields an estimate of the reflector savings for the face(s) which has been
reflected.
By progressing down the sequence of reactors, a complete
cycle is eventually made around the actual reactor and a reflector savings
obtained for every face.
It should be evident from these introductory remarks that the crux of
the entire matter is the selection of the dimensions (or bucklings) of the
bare faces for each of the elementary reactors in the sequence.
this point that we Introduce the iterative procedure mentioned
It
is at
above.
In carrying out the calculations for the first complete cycle around the
reactor, it is necessary to make a first rough estimate of the reflector
savings (i.e., bucklings) for all three faces.
These data may then be
used to obtain an improved estimate for the savings.
For example, in
analyzing reactor 1, the rough estimates for the bucklings in the x-i and
x% directions may be used to compute a better value for the savings in the
The same procedure may be followed in the case of the
Xi direction.
other two axes x2 and x%. This first cycle yields, then, an improved value
REACTOR
480
ANALYSIS
[CHAP.
It has been
bucklings (reflector savings) for each of the principal axes.
found in practice that this procedure does indeed converge very rapidly
and the results appear to be independent of the initial guess which may
Unfortunately it has not yet been
be used in computing reactor 1.
shown precisely what reactor is analyzed by this rationale since an exact
solution of a completely reflected system is not presently available.
How
ever, criticality estimates obtained by this procedure have been found to
agree surprisingly well with experimental measurements on the ORNL
Swimming Pool Reactor.
The application of the two-group model to each of the elementary
reactors in the iterative sequence described above requires the solution
For this calculation it is convenient
of the criticality equation (8.169).
to write this relation in the form
1
k'
=
(1
^4 p
=
where
(8-222)
+ L|B)(1 + tcB*)
- I ^ - rc
=
L\
(8.223)
k =
(1
(8-224)
+ L\B*){\ + tcB*)
we obtain [cf.
Eq. (8.189)]
k
K" ~ 2
"M2
M*
(8
2^Ll j
OD
(if
Bi
[Mt
225)
^^.(T-Ot'-t^T"-')]
X!-rF
i(-')
and
Taylor series.
<8m)*
sec. 8.6]
REFLECTED REACTORS
481
Core
Multiplication k
Fig.
8.27
Buckling as function
flected on one pair of faces only.
of multiplication
for rectangular-block
reactor re
various values for the bucklings (i.e., bare widths) of the other two pairs
The computational procedures outlined in Sec. 8.4 are directly
to graph these results
applicable here. The next step in the calculation
The notation which has been
in a suitable form, as shown in Fig. 8.27.
of sides.
is
a ~"~d
(b)
(a)
Fig.
"
REACTOR
482
ANALYSIS
[CHAP.
and the
B,
B,
= x/26
= t/2c
y
2
Bl
B\ + B*
Buckling
B,
with
B%, m
of core
Half-width
B\ + B\
r/2a
B*
B\ + B\
(8.228)
"total" buckling
B* =
Bl + Bl + B]
(8.229)
4. Substitute these three values for k into Eq. (8.226) and compute
the corresponding values of B2 (here called fi1). Note that the kx which
appears in (8.226) is a fixed number throughout this calculation and is
[>]' =
[B<]
- B*[ky\
[B?]*
m B*[k?>]
B*[k's]
- [fi]'
- [B
- [B\*
(8.230)
fcj1'
determined
direction.
and each
sec. 8.6]
REFLECTED REACTORS
483
example.
Fig.
8.28
for
array of MTR-type fuel elements (cf. Fig. 5.27). The half-width in the
x direction is o = 27.7 cm, in the y direction b = 16.15 cm, and in the z
(vertical) direction c = 30.0 cm.
(2) This central region is surrounded
on four sides by a 7.75-cm-thick beryllium reflector.
(3) The beryllium
in turn is contained in a 1.905-cm-thick aluminum tank.
(4) Both the
top and bottom of the core are reflected by a double layer of metal and
484
REACTOR
ANALYSIS
[CHAP.
The principal nuclear data for the two-group model are listed below:
kK = 1.677
= 0.07992 cm-1
Vc = 107,000 cm'
= 64 cm2
=
0.269 cm
Z)2
=
3.366 cm1
L\
tc
tp =
D,
= 1.32 cm
= 67.37 cm1
14 1
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
Multiplication constant k
Via. 8.29
Buckling-versus-multiplication
(standard core loading).
in
three-
of the buckling
discussed above.
is,
sec. 8.6]
REFLECTED REACTORS
485
dimensions
Bl
0.001670
0.002934
0.001570
Bl.
Bl.
Bl,
=
=
0.001503
Bl.
Bl.
Bl,
002767
0.001542
0.002754
0.001805
Bl.
0.001818
Bl,
Bl.
Bl.
Bl,
=
=
0.01220
0.005958
0.01268
k.
k,
: 0 8570
=
1 037
k.
0.8350
0.004504
0.003240
0.004604
k.
k,
k.
0.004585
0.003321
0.004270
k, -.
0.004559
0.003347
0.004296
k,
=
=
=
B'
0.01387
B = 0.008963
B'
1 187
0.01425
0.006007
1 187
fla = 0.006007
1 163
B*
1 180
1
0.003216
0.009460
0.002742
183
183
B] B',
Bl
182
182
182
B*
0.006422
0.006127
B1 = 0.006075
B*
fls
/?J
=
=
0.006075
0.006092
0.006092
0.006092
It seen from the results of Table 8.6 that to four-figure accuracy the
= 1.182; this corresponds to
multiplication constant for the reactor
=
multiplication
1.677.
To find the k (i.e., the
an infinite-medium
kx
fuel loading) for criticality, several other guesses of kx need to be made,
computed, and the criticality value obtained from
the corresponding
graph or by interpolation.
As remarked previously,
not known to what solution the proposed
There
some reason to believe, however, that the
rationale converges.
Since an "exact" solution
end result corresponds to the actual reactor.
it
is is
is
k
is
a two-group
Bl
to the bare-core
Bl
8.6
B\
B\
correspond
KB\
B\
k,
cycle
k.
Table
1
under
ky
a = 27.7,
k.
values
ANALYSIS
REACTOR
486
[CHAP.
Swimming Pool reactor, along with the principal dimensions of the system,
are given below:
Table
8.7
Two-group Constants
Flux group
for ORNL
Swimming
Core
Reflector
TC = 64.0 cm*
Fast
Pc
D,
Thermal
TR
= 1.320 cm
D,
PR
= 0.02062
Pool Reactor
cm"1
= 0.269 cm
Dt
2<o
= 32.7 cm'
= 1.19 cm
=
1.
= 0.0363
cm-'
0.1602 cm
= 0.01916 cm"1
Reactor geometry:
assume;
The multiplication constant for the system described above was com
puted by the iteration program for several values of the fuel loading (kx).
By this procedure a critical mass Mc = 3.050 kg was obtained. The
known experimental value is Afp = 3.190 kg.
The two-group tworegion model using the iterative procedure gives, then, a critical-mass
estimate 4.4 per cent low.
8.7
Feynman-Welton Method
however,
effectively
distribution of each
REFLECTED REACTORS
sec. 8.7]
487
reliable
models,
especially
in applications
to
water-moderated
systems.
shown1
The
1 B. Davison,
"Transport Theory of Neutrons," National Research Council of
Canada, LT-18, Chalk River, Ontario, January, 1947.
'See, for example, U.S. Atomic Energy Commission, "Reactor Handbook:
Physics," p. 450, McGraw-Hill Book Company, Inc., New York, 1955; and R. P.
Feynman and T. A. Welton, "The Calculation of Critical Masses Including the
Effects of the Distribution of Neutron Energies," Los Alamos Scientific Laboratory,
Report LA-524, Jan. 21, 1947, declassified.
REACTOR ANALYSIS
488
[CHAP.
in the fast group. We write the neutron-balance relation for this system
in terms of the removal density (number of neutrons of group i removed
per unit time per unit volume around r): Ai(r) = S<?^,(r), where 0,(r)
is the neutron flux for group i and 2< denotes the removal cross section
For the thermal group (t = 2, 4), 2< is the usual absorption
cross section, and for the fast group (t = 1, 3), it may be approximated by
for group i.
of energy group i
collisions in both core
a neutron
r',
where
is
r'
is a
is
born at
after any number of eZash'c-scattering
eventually removed as group neutron in dr about
and reflector,
in
the core. The function Pi2(r;r')dr
the probability that a group
in the core escapes removal in the core,
neutron born at
slowed down
in the reflector, and finally returns to the core to be removed as
group
group
r'
neutron in dr about r.
S2(r) = ^i(r)
(8.233)
If
2}2)
is
is
(8.157).
neutrons slowing down to thermal in the core, and the second, to those
which have slowed down in the reflector and then diffused into the core.
but
represent these by rough approximations
are complicated
functions,
model.
In
the
489
BEFLECTED REACTORS
sec. 8.7]
A.(r)
- AWo(r)
(8.234)
where A?1 is some constant and ^0(r) is an elementary function (sin B,r/r
for a spherically symmetric system). These functions are so selected
that they provide a good representation of the spatial distribution of the
neutrons in each group throughout
central region of the core (see
Fig. 8.30). It is to be expected, then,
the
that
Actual flux
Core-reflector
Fig.
It is shown in a subse
quent analysis that in first approxi
mations the kernel Pn is independent of r and r' and may be given by
Pi2(r;r') cm w/Vc, where Vo is the volume of the core and w is a constant.
Eq. (8.232).
Then
[s1(t')
PhM
dr'
y-c
dx' =
jcS1(r')
jc At(f)
dr'
(8.235)
<Si
Mi,0
Pi(r;r') dr'
fc Mi')
(8.236)
is
r,
is
produced
r' r',
interface
490
REACTOR
integrated
ANALYSIS
[chap.
removals per unit volume per unit time at r, namely, A^oir), as speci
This equation is evidently a direct
fied by the left-hand side of (8.236).
application of the kernel method discussed in Sec. 5. 2d.
If we use the expression (8.234), the integral relation (8.236), and the
source function (8.233) in (8.231), the result is an equation in Ap and
Jc
\po(r) dr
implied in the
term (8.235).
The resulting pair of equations in the unknowns A'J> and
A^ is homogeneous, and a functional relationship must be established
between the coefficients in order that there exist a nontrivial solution.
This is easily obtained by eliminating the Atf; the result is the criticality
condition:
" =
2>*>[i
(8"237)
MM8']
To solve this relation for the critical concentration implied in 2'*' and
Generated for wjivans (University of Florida) on 2015-09-22 17:21 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
2J?>
we require
system.
a relation
of the
-A-Wo(r)
2<'Vo(r)
/TOYoto
(8.238)
A|?Yo(r)
Z*(r)
(8.239)
wrim
(8-240)
where, in the case of a spherical reactor with infinite reflector, the bucklings Bi are obtained from the familiar one- velocity relations:
Dz
j
((c3a
=1
D\
{Kta
+1)
1)
with
k\ =
2(3)
2<>
1
~
(8241>
k] = -Jr-
It
sec. 8.7]
REFLECTED REACTORS
491
[l
^]
>
symmetric system,
1_
M(.nJ
32<<>
(,
aggVfKj
+ OK,)(l +
<M)
(8.240) to obtain the ntf. The substitution of these results into (8.237)
The details of this model are presented
yields the fuel concentration.
at the end of the present section, and suitable formulas are developed for
a direct computation of the critical concentration.
c. One-velocity Model.
Since the one-velocity integral theory plays
an essential role in the development of the Feynman-Welton method, we
will direct attention first to a study of some of the important properties
>V2<Kr)
2<Kr)
= 2!/*(r)
We showed previously in our study of the kernel method (see Sec. 5. 2d)
that the solution to this equation could be written in the integral form
+(r) = vL,
K(t;t')
Jc *(r')
dr'
(8.242)
where the integration extends over the core and K(i;t') denotes the flux
at r due to a unit point source at r'.
It follows from this definition that
K(t;t') is the solution to the differential equation
-DV2K(i;i') +
In general K(i;i') will
S0/C(r;r')
8(r
- r')
(8.243)*
REACTOR ANALYSIS
492
[CHAP.
*(r)
wjif
Then
Jc *(r')
K(rf)
dr'
(8.244)*
r'
r',
if
is
and the
vn are the eigenvalues of this equation
We now demonstrate that the
corresponding eigenfunctions.
is orthogonal on the assumption that the kernel K is symmetric3
K (r;r') the flux at due to unit
variables r and r'. That
at
then this must be identical to K(i';i), the flux at
due to
is,
The quantities
source at r.
the
\f/
set
\f/n
in the
source
a
unit
is
K(r?)
K(r';r)
(8.245)*
(8.244) by
dr = 2,
*(r) dr
fQ
Mr) Mr)
fc
fc
(8.246)
I,
is
Chemistry," pp. 509-512, D. Van Nostrand Company, Inc., Princeton, N.J., 1943.
The symmetry property of the one-velocity kernel K
easily demonstrated in the
case of both the bare and reflected slab and sphere geometries.
For example, note
that +(x;s) = _(s;x) in Eq. (5.266).
*
SEC.
REFLECTED REACTORS
8.7]
Mr)
dx =
n,
= VmXf
m we have also
Jc f(r)
fc
493
dr
MO
fc MO
K(t;t') dr'
K{r;x')
fc Mr)
dr'
dr
(8.247)
where the last step follows from a change in the order of integration.
Since we have required symmetry in K, (8.247) may also be written
fc *(r) Mr)
dr = ymX,
fc
MO
dr'
fc *(r)
Jf (r'jr)
dr
(8.248)
But, the right-hand side of this equation, except for the factor
identical to (8.246) ; therefore,
fc
Mr) Mr)
vm,
is
m^n
dr = 0
In this
fc Mr) Mr)
where
5mn
dr =
(8.249)*
Smn
The orthonormal
(r)
+(r)
- 2 ^"(r)
(8-250)
The
are the coefficients of the expansion and can
relation
4.
fcM') (r)dr
(8.251)
Kn(r')
f(r)
Kn{r')
fcMr)K(r;r')dr
(8.252)
that
is,
K(r;r')
be
(8.253)
494
ANALYSIS
REACTOR
[CHAP.
jT
*B(r)
K(t;i')
dx =
K(i';i)
j^n{x)
(8.253)
dx =
yields
Kn(t')
(8.254)
(8.255)
/_/
vnLf
tfB(r') =
and compute the quantities
(8.257)
An
K{x;x>) =
(8.258)
(fly
- S.) T
-8(r-r')
If we multiply
W-to
Am
- h Kit)
-+m(r)
Am
(8.259)
But we know that the function {x) satisfies both Eq. (8.242) and
V*(r) +
In the
y[/m
B*4>(x)
= 0
VYm(r)
BWm{x)
= 0
(8.260)*
sec. 8.7]
REFLECTED REACTORS
495
= 20 +
DBl
(8.261)
*
(8.262)
Tt
^fm
we find
that
1.
the next fewest number, and so on. In the case of the bare-slab reactor
in diffusion theory we found, for example, that
Mx)
= cos
^|
4>i(x)
= cos
+2(x) = cos
Note that
has no nodes,
has two, and ^2 has four.
3. The eigenfunctions ^(r) represent the spatial distributions of the
flux according to these modes, and the coefficients n of Eq. (8.250) give
the weighting function for each mode, i.e., the fraction of the neutron
population distributed in each space form
4. The eigenvalues vn may be ordered; thus
vo
< vi <
v2
<
(8.264)
[chap.
REACTOR ANALYSIS
496
Thus if
neutron yield per fission for this system we denote by v<>. The distribution
on the other hand, may exhibit several nodes, and with each there
will be a region in which
We may think of these regions as
< 0.
occupied by certain particles, the presence of which can cancel an equal
yp.
space mode
v0.
\pn
given by
where
t/
(8.265)
The above
and Sa denotes the absorption cross section of the mixture.
relation follows immediately from the interpretation given for v in
statement
above.
Since
(8.266)
Vn
dr ~
X,
Jc
fadr
dx
Z'
Then
Jc
is
independent.
dependent
Studies
reflectors
energyin
are considered
subsequent sections.
In the present study of multivelocity reactors
will be convenient to
To the first cate
separate all scattering collisions into two categories.
gory we assign all nuclear scattering phenomena which cause substantial
it
Leakage =
the mode
is
5.
REFLECTED REACTORS
sec. 8.7]
497
(8.267)
(8.268)
2()
be
been included
in
ate these nuclear events from other processes, such as fission reactions,
which can yield neutrons at markedly different energies.
The particular merit of this formulation of the scattering phenomena is
that it allows one to treat a large variety of reactor problems by means
of the energy (velocity) group methods developed for the present model.
Thus by a suitable choice of the scattering-cross section assignments it
is possible to study thermal, epithermal, or fast reactors.
The cross sections introduced above may be combined with the cross
sections for the absorption phenomena to obtain a new definition for the
total cross section; thus,
2t 27
+ 2/ + 2* + 2.
so
that
2r = 2 + 2*
2, = 2r + 2e
(8.269)
It
is convenient
to
(8.270)
(8.271)
REACTOR ANALYSIS
498
neutrons
"removal"
[CHAP.
be useful
an energy group,
processes.
Homogeneous Reactor. We begin our treatment of multia study of the bare reactor in which all cross sec
tions are energy independent.
Neutrons emerging from fission and 2*
1.
Bare
will, in general,
distributed according
be
to some specified
(t,E)
S(t,E)
we introduce
dE about E
a neutron produced at
(8.272)
r' is
is
if
r',
The function /,fl(r;r') defined here will be the kernel in the integralequation formulation of the neutron-balance equation; note that it is
independent of energy in this particular model in which the cross sections
are energy independent.
It should be noted that this definition differs
from that given the function K(r;x') used in the preceding section.
Whereas the kernel K represented the flux at r due to a unit source at
function of the type K multiplied by 2r. This
the kernel Pb implies
we consider, for example, the
relationship may be seen more clearly
quantity 2rX(r;r') dr, a physical interpretation of which
about
r'
time from
= probability that neutron released at
removed in dr about
r'
is
is
(8.273)
The last statement follows from the fact that the flux
due to a unit
it
source.
will
(collision) density
be convenient
(8.274)*
then given by
A{r,E)
J/
is
to use,
A(r,E), where
A(r,E) = 2r<Kr,)
reactor
S(r',E) PB(r;r')dr'
(8.275)
is
removed in dr about r
sec.
REFLECTED REACTORS
8.7]
499
form
S(r,E)
J" (r,E')
CE)Z,
fj A(r,E') (,')
dE' + 2*
fj
4>{t,E')
h(E;E') dE'
dE'
(8.276)
VZ^'
+ **W>E')
probability that
(8.277)
a neutron of energy
A(r,E)
["
JO
&(E,E')dE'
J\reactor
E' when
(s) represented by
2*
(8.278)
The substitution
A(T',E')PB(r?')dr'
(8.279)
A(r,E) = A(E)*(t)
(8.280)
complete
I"
J.
= m
JL
A(E)
(a)
Pb^ dI'
reactor
(8.281)
mX =
These results are entirely consistent with those of our previous analysis
In this
of the bare reactor using the Fermi age model (refer to Sec. 6.3).
is
h(E;E')
- a)E'
Sec. 4.6 are directly applicable for obtaining solutions in the asymptotic
ANALYSIS
REACTOR
500
[CHAP.
range and lead to the critical equation in the usual way [see, for example,
Eq. (4.162)].
Equation (8.2816) gives the spatial distribution of the neutrons
throughout the reactor at all energies. The solutions to this equation
are given by a set of eigenfunctions ^(r) with eigenvalues X [cf. Eq.
(8.244)]:
*(r)
= X
*(r') Pb(t;t')
J[reactor
dx'
(8.283)
X0
Ms')
mode.
{x;x')
is
PB(.x;x')
where
Pb(x;x') dx'
(8.284)
of Eq. (5.266),
2t(x;x')
(8.285)
Substitution into
= X02,
Mx')
4>{x;x')
dx' = vXf
Ja
Mx)
f'a
Mx') (x;x')dx'
(8.286)
If
the last equality follows from the general form1 (8.242).
easily
apply the solutions (5.266) for the kernel (x;x'), then
shown that the \po{x) given by (8.286) also satisfies the equation
where B% = {vV,
i'0'(x) + Bl4>0(x) =
2)/D. Thus in this simple
case ^o(x) has the usual trigonometric
form, and the condition on the
the usual criticality relation for the one-velocity model.
eigenvalue
The critical size
this reactor
obtained by computing the lowest
of
Eq. (8.281a) and adjusting the size of the reactor until
eigenvalue m0
it
we
is
where
0,
is
is
Xo
Xo =
*o0c)
|_o
For an actual system having the nuclear and geometric properties assumed
This
above, only one of the solutions (8.283) is physically acceptable.
It is clear that, for n > 0, the
solution is the fundamental mode ^o(r).
steady-state spatial distribution would require that A{r,E) < 0 in cer
tain regions of the reactor.
The determination of the neutron-flux
spatial distribution in a finite system by means of the integral-equation
formulation was demonstrated in Sec. 5.7c for the case of an infinite slab.
Those results may be applied directly to the multiplying medium problem.
In the case of the infinite-slab reactor, of width 2a, Eq. (8.283) takes the
form
(8.287)
B\ equation.
K
and
(x;x')
REFLECTED REACTORS
sec. 8.7]
501
As in the pre
2. Reflected Reactor with No Moderation in Reflector.
ceding case, all cross sections in both core and reflector are independent
All
of energy although the spectrums z{E) and h{E;E') need not be.
cross sections in the core 2/, 2*,, 2JP, and 2<C) are in general different
In the reflector
and 2^S) are different from zero, but
from zero.
2/
= 2i*> = 0.
di
P(r-r')
'
is removed in dr about
reflector.
=
~~
gn(E,E')
m 0
+ 2lC)HE;E')
(8.289)
where
A(t,E)
$C(E,E') dE'
denotes integration
fc
fo"
The neutron-balance
tion
accomplished by the separation of variables,
calculation. The results are
yf
"
m m
Jc
A(E)
%C(E,E')
(8.290)
is
as in the bare-reactor
A(E') dE'
(a)
(8.291)
(b)
Equation
The
ypn
1,
on reactors which are either (1) bare, or (2) have reflectors which do not
moderate neutrons.
For all reactors in class
the cross sections of the
core 2/, 2T, 2#, and 2 will be, in general, some specified functions of
energy.
2,
REACTOR ANALYSIS
502
[CHAP.
but that
be energy dependent,
2<ft)
P(t;r',E)
dr =
a neutron born at
(8.292)
Thus this new kernel includes the possibility that the neutron of energy
dr about r.
A(t,E)
is clear
S(t',E) = ,*()
f' ME')
be defined
(T',E') dE'
f0"
A
(8.294)
A(t,E)
W) h(E;E')
ZiC){E')
these expressions
ME')
fj
The substitution of
= vz(E)
$(E,E') dE'
(8.29o)*
fc
>(E,K
It
[cf. (8.276)]
fQ"
by
(8.293)*
JcS(T',E)P(r;x',E)dT'
(8.296)*
clear from the above expressions that, since there are no "sources"
of neutrons in the reflector, the spatial integration extends over the core
volume only.
the determination of
The immediate objective of the present section
is
is
sec.
REFLECTED REACTORS
8.7]
503
2i ^
2;
(2)
(3)
Z'. 2'e
(1)
= 2T
= 2,
= 27
the core:
2;
the reflector:
= 2,
= 2.
(8"297)
In
In
~L'y
point of view.
2*
is
We note immediately from these expressions that the removal cross sec
identical to that of the actual system,
tion of the fictitious system (2^)
=
2;
2;
2;
2;
+
2;
2;
since
= 2, + 27 + 2* = 2r
(8.298)
is
that
The only significant difference between the two systems, then,
" slowing-down " cross section of the fictitious system
zero.
(2*)
quantity to define the neutron yield for
Now, let us introduce also
the fictitious system; thus,
a
is
the
(8.299)*
clear from these definitions that the source S(t,E) in the fictitious
due only to the removal reactions.
We can write, therefore,
system
is
is
It
S(t,E) =
(T,E)
ul{E) 2'T(E)
p(E) A(t,E)
(8.300)
A(i,E)
n(E)
(8.301)*
fc
tJjJB)
where the
It
present system.
unctions 4>{t,E)
is
is
is
is
4,n(r',E)
P(t;t',E) di'
(8.302)
it
m(2?)
is
is
is
it
orthogonality properties.
From physical requirements
clear that
the kernel P(t;t',E) must be symmetric in and r'. If we bear in mind
that
to be treated as
parameter, then all the derivation leading to
the orthogonality relation (8.249)
directly applicable to the functions
REACTOR ANALYSIS
504
\f/n(i,E) and
P(t;t',E), and
we obtain
jc Mr',E)
[CHAP.
*m(r',E) di'
(8.303)*
Smn
The series expansion for the kernel P(t;t',E) in terms of the eigenAs in the one-velocity
ipn{r,E) is also easily demonstrated.
derivation we postulate that the ^/{r,E) form a complete set. It follows,
therefore, that we can construct an expression for A{t,E) in the form
functions
A(t,E)
An(E)
Mt,E)
(8.304)
An(E)
K,
we define,
PW t^'M
are given by
Thus,
(8.305)
P.W
P(r?',E)
(8-30(i)
*(r>)
(8-307)
(8.308)*
The two basic results obtained above, (8.303) and (8.308), are gener
ally valid and may be applied directly to the physical system described
Our only purpose in introducing the fictitious system was to
by (8.296).
offer some aid in grasping the essential ideas and analytical tools which
Let us now apply
may be used to solve the neutron-balance equation.
The substitution of (8.304) into (8.296)
these results to Eq. (8.296).
M',E)
g(E,E')
dE>
P(t;t',E)
An(E)
fc
f0~
yields
An(E')
W,E')
dv'
dE>
An(E') fc*m(r,E) di
fc
g(E,E')
Am{E) =
fQ~
P(t;t',E)
With
dr1
ME')
Mr',E') di'
jQ
(,') dE'
fc
J*
dx
Pn(E)
A(t,E)
P(t;t',E)
If
fcMr,E)
P{t;t',E) +m(r,E) di
(8.309)
REFLECTED
SEC. 8.7]
is symmetric,
REACTORS
505
we may
the form
Am(E)
^ JJ
An{E') jjn{r',E')
%{E,E') dE'
+m(r',E) dt'
(8.310)
If
we call
Anm(E,E')
fc
Mr',E') Mi',E)
dr'
(8.311)
Am(E)
It
V /
^ /_/
Jo
Hm{&)
(8.312)*
Anm(E,E)
= 5
(8.313)
for
use
as
will
is well adapted
be demonstrated
in subsequent
analyses.
f. Upper and Lower Approximations.
Various approximations have
been suggested for solving the neutron-balance equation (8.296) which
we have reduced to the set (8.302) and (8.312).
These involve the use
expressions for either of the two functions A{t,E) or
we apply the approximation to the
removal density A{t,E). Specifically, we assume that
of approximate
P(t;t',E).
A(i,E)
An(E)
0
Mr,E)
for n >
0.
=a
A(E)
Mr,E)
(8.314)
50G
REACTOR
ANALYSIS
_-
[CHAP.
*(r')
If
A(t,E)
MET
Ap(E) Mt,E)
(8'315)
ME)
equation
Ao(E)
-J
MA) Jo["
(8.316)*
yields an overestimate
the fact that in estimating
use of
the
is,
*o(r,)
^ 4">(r,E')
= *0(r)
(8.317)
(8.318)
Ao(E)
This relation
-j
Mo(.a)
J0
and
it
Aoo(E,E')
That
ME')
g{E,E') dE'
(8.319)*
is
is
it
is
is
Since both estimates appear to be fairly reliable, we will use the lower
approximation in all subsequent analyses in view of the fact that the
relation (8.319)
more easily applied to reactor calculations.
We con
is
state.
REFLECTED REACTORS
sec. 8.7]
507
tinue with the calculation on this basis and introduce the source function
(8.295)
into (8.319).
=
Ho{h) A{E)
vz(E)
')
+
This equation may also
dE'
zlC)(E')
be
'
Let
Since A{t,E) is directly related to the flux and has been defined by the
series (8.304), it is easy to show that
An(E)
ME) n(E)
(8.322)
= z{E)
f* ME') h(E;E')
ME') ME')
dE'
(') dE'
(8.323)*
fo~
is
A(t,E)
S(i',E) =
Pb(t;i',E)
P(t;t',E)
[vz(E)
ME')
(8.325)
ME')
h{E;E')] dE'
(8.326)
where
[" [
lr(E)
to
REACTOR ANALYSIS
508
[chap.
In
i=l
. .
. .
2u
dr about r
average number of scattering collisions per unit path of
= travel by neutron of energy group 1 which reduce it to an
energy group 2 neutron
,g
,g ^2g\
(8.329)
(8.330)
Note that the group cross section 2i2 is in fact the " slo wing-down " cross
With these
section (analogous to 2*) from the fast group to the slow.
definitions Eqs. (8.325) and (8.326) may be written
with
S,(t)
(8.331)
(8.332)
where
z<
y<
WW
produced per
(8.334)
vfe
'r
(8.333)
^W(r)
to this calcu
(8.335)
= fl,2'(r) = *o(r)
(8.336)
WW
If
the kernel
Ml*
nXr;r')
fc WW
dr1
(8.337)
would yield the spatial distributions for the neutrons in group i. These
results in turn could be used in Eqs. (8.331) through (8.333), and a
simultaneous solution would give the criticality condition for the reactor.
As
of the
P$ will
be somewhat
involved, like
REFLECTED REACTORS
BEC. 8.7]
509
diffusion-theory or infinite-medium
(7.254)].
transport equation
The use of
tational procedure based on the diffusion-theory solutions.
the transport-theory results in the calculation of the eigenvalues is easily
shown and involves some minor changes in the procedure developed for
the diffusion theory.
The application of the diffusion theory to the calculation of the eigenfunctions and eigenvalues for the Feynman-Welton method requires a
restatement of the neutron-balance equation in terms of the removal
cross section 2^ and the neutron yield per removal ntf. It is obvious
that the appropriate form is [cf. equation preceding (8.242) and Eq.
(8.238)]
-DtV4p(t) +
2<= Mi20p(r)=
(8.338)
Mi20p(r)
(8.338)
In
where ^(r) is the fundamental mode for the flux in the ith group.
Eq.
(8.242)], the neutron balance would be
the kernel formulation [cf.
given by [with P$ = 2?K$]
m*)
fW fc m*1)
KSW)
dr'
(8.339)
It
follows directly from the results of Sec. 8.7c [cf. Eq. (8.263)] that the
eigenvalues p must satisfy the relation
=
(8-240)
VW(r) + BW(r)
Note that the mode index n = 0 has
Thus, as
to simplify the notation.
(8.337), we use the solution
= 0
(8.340)
approximation
buckling B{
to the Eq.
the eigenvalues /4'1 from (8.240), keeping in mind that the B, are given
by the usual relations for the bare reactor (e.g., in the case of the sphere,
Bi
=
1.
B = t/R).
Application
to
Thermal Reactor.
= 0
=
2/
Zx
z2 = 0
2?>
= 0
2,2
5*
= 2'
+ 2/
(8.341)
ANALYSIS
REACTOR
510
[CHAP.
Si(r)
= uy2A2{i)
4,(r)
St(r)
(8.342)
and by (8.331)
A,(r)
/cAi(r')
(8.343)
Pf{r;t')dT>
A.(r)
so
(8.344)
Wo(r)
AWa{r)
by means of (8.337).
^
V
A (2)
(8.345)
The resulting
A (1)
(8-346)
The ratio of these two equations gives the criticality condition for the
jS}
reactor; thus,
= MiM2'
(8.347)*
DiB*
2+
D2B2
~^^>
(8.348)
if
where we have used the fact that the B< = B, a constant independent of
the energy dependence of the extrapolation dis
the neutron energy,
is
ignored.
If
we obtain
[2i2
+ L>,2][2 + D2Bl\
(8.349)
tance
dition,
is
ard two-group method of Sec. 8.4b [cf. Eqs. (8.155)], since (8.349) may
also be written
2S,2(1
We
have
notation,
+ L22)(1
+LW)
used here the identity 2^2) = 2> 4and L\ = D,/212 and LI = D2/Z.
(8.350)
A}fV0(r)
2/
2,
in the usual
REFLECTED REACTORS
sec. 8.7]
511
72
dr'j
A,(r') P$(r;t')
(8
= *,
jf
At(r)
Tti
fc
As(r')
P(r;r')
ic
Ai"
dr'
[7iAi + 7SAi]
(8.352)
^[71A0
7.Ai]+|^
gives
the simpler
between
relationship
and A0:
Al)
353
^,7Ag
"Zi7i
Mo"
And, the simultaneous solution of the pair (8.352) yields the criticality
=
2i[7.Mo2)
+ TW2<"]
"
relation
(8-354)*
2*72/4"
is
most general reactor problem so far considered has been the reflected
multivelocity reactor with energy-dependent cross sections but with no
slowing down in the reflector.
general method for treating this sys
tem, and other more elementary ones, has been discussed in some detail.
general
is
group model.
It has already
it
Ai =
Ap
P(r;r')
^-
The application
Ax(r')
dr'j
Pg>(r;r') aY + y,
i.(r')
= wt,
jc
Ai(t)
[71
(8.351)
diffi
512
KEACTOH
ANALYSIS
[CHAP.
introduce
The analysis of the reactor with moderating reflector is carried out for
As in the bare-reactor calculation we
the case of the infinite reflector.
in
the core by the index i = 1 and the slow group
denote the fast group
=
In the reflector, the fast group is denoted by * = 3 and the
2.
by i
slow by
= 4, so as to be consistent
notation of
Sec. 8.4.
equations,
form
Ai(r)
A2(r)
(8.231)
fcS1(r')P1(i;T')di'
fc
fc Stf) P(t^)
dr'
(8.232)
where the S.(r) are given by (8.332) and (8.333) and the kernels
are defined by
a neutron of energy group i born
at
number
of elastic collisions in both
a^er an^
dr =
(r-r'1
'
core and reflector, is eventually removed as a
group t neutron in dr about r
Pi(r;r')
probability that
P
It
probability that
P 12 (rr'f-r') drf
a fast neutron
to the
P(r?',E) of
This
is the only new quantity which appears in the set (8.231) and
In determining this function it is convenient to introduce the
(8.232).
slow-neutron removal density in the reflector, At(r') and a new kernel
for slow neutrons, Q2(r';r), which we define:
Q2(r';r)
(8.357)
BEC.
REFLECTED
8.7]
REACTORS
513
Note that this function is symmetric in r and r7; i.e., Q2(r;r') is the track
length at r in the core produced by a unit source at r' in the reflector.
The removal density At(i') may be written in terms of Q2. Thus if
<S2(r) is the distribution of slow-neutron sources in the core, we can write
A4(0
=Q2(r';r)
Z< dt (8.358)
2<"= Z< fc 5,(r)
fc
eigenf unctions
5,(r) Q2(r';r) dt
be represented
#,2>(r) ;
S,(r)
(8.358)
thus,
n <SW(r)
(8.359)
It is convenient to
set Afir'); thus,
Z<<>
*(r) Qtf?)
dr
A*W
Note that At(r')
S>2<
/c
tf<r) Q2(r';r) dr
as some
(8.360)
A4(rO
- J A<(r') - J
(8-361)
be
AJfVW)
[cf.
Eq. (8.359)].
With this
written
=
fc #(r)
dr
(8.362)
These results may be used to obtain suitable expansions for the kernel
If
we define
Q2(r';r)
Gp^) #?(r)
Q2.
(8.363)
",(r,)
Now,
Qj,2)(r')
Sgfew
(8"364)
neutron-conservation
requirements, the total number of
per unit time of slow neutrons in mode n in the reflector must
be precisely the number of these neutrons escaping per unit time from
from
removals
REACTOR ANALYSIS
514
[CHAP.
<P
L A?(i)
M<2*
M<2)
dr
[
&
dr
Jc A'?(i)
Therefore,
^<41
A(r) dr
^4,(r) dr =
Tl
S<
^ j jf
- -il
"
#(r) dr
(8.365)
(8.366)
Q,(r';r)
V [l-ilJ JC[W(r)dr)
L
Mn
/H^4,(r')*'
W(r'W(r)
(8.367)
= ty,(4>(r)
for all n
(8.368)
is some constant.
Then, we can factor out the quantities
which
on
these
depend
functions, and
(8.367)
where
~
R
*>(r)
- -L]
Q2(r';r)
|~1
**(r)
in
dr
The sums which appear in this expression are integrals over the core
volume (Vc) of
^(r')
*S?W
- r')
(8-369>
P,(r';r)
(8.370)
(r
'
'
WHr)
itwCr/)
Wn
Mn
If
2?' JR^t)(r')dr'
Ubid.
Jc
P2(r";r) dr"]
Q2(r';r)=-
the integrations are carried out as indicated and the results applied to
the above approximation to Q2, we obtain
i \l
removal
(8.371)
REFLECTED REACTORS
sec. 8.7]
515
in (8.356),
= 2<
P12(r;r')
fRS<(j"?)
Q2(r;r")
di"
(8.372)
Pi2(r;r')
2342<2'
(8.373)
2 jR
Q2(r;r")
2<4'
jR
^<3)(r")
dr"
dr"
Ptfrtfdt]
P,(r';r)dr']
(8.374)
(8.375)
REACTOR ANALYSIS
516
[CHAP.
JL
P2(r';r) dr']
(g
f P,(r;r')
12
\l
fl
2i4<P
Pi(r^)
di]
yields
The sub
the slow neutrons and may be derived by similar arguments.
for
and
into
Eq.
of
above
stitution
the
approximations
Qi
Q2
(8.373)
376)
Mo
(P42
J 4sl
WVo
2<<>
7*
|fl^(r)
W)
^(4)(r") dr"
dr
(8.377)
yB^(r)dr
kernels
(8.369) we obtain
P^rjr')
=^ (r
(TT>>
PP i2(r,r
Z34g42[l
(1M)1 _
=
2VC
Vc
^.^i-^J^^,
with
It
(8.378)
,0,7CA
(-379)
(8.380)
is
is
m(<)
fc
is
m'o
(8.381)
fc
Mi')
Pi(rjr') di'
(8.236)
The substitution of these forms into Eq. (8.231), along with the source
=
Mo
t^o1' + "Mi2)]
is
- r')
obtained
(8.382)
by using (8.333)
SEC.
REFLECTED
8.7]
for
Pn
REACTORS
517
over all r.
+ iw
[7^i,l>
+ 7iAp] +
fg&
The result
(8.383)
-~
(KRa
1)
(8.385)
This equation is to
In
Bid
COt
Bid
=1
yr-
Dn-
(/c3a
1)
(8-241)
(x4a
1)
and
(8.386)
REACTOR ANALYSIS
518
[CHAP.
Bi
With the
*(4)(r) =
Note that
these would
(8.387)
case since
of
It
(P42.
can be shown that the substitution of (8.387) into the integrals of (8.377)
^(,V(4> dr
fR
fR
<
dr
fR
<
dr
(4r)2
j"
I.'
yields
g-r(,+ ,)
we use also
Vc
rer dr
k)(1
+ o3)(l +
am)
(8.388)
= ^ira',
and
Feynman-Welton
to
m'o2'
Thermal Reactor
reflector.
symmetric system
cross sections listed in Table 8.8.
with
OK3)(l
OK4)
(8.389)*
Moderating Reflector.
An
z2<y"
2<
- 2
2/
- 2
2<4
i =
2<" 2<
2-
2(3,
H =
Model of
Reflector
Fast
0 0
Energy group
Thermal
<c4)(l
a spherically
8.8
useful
Table
(C3
is
Application
interesting
48
reflector,
i.
4tt(k3
If
Jj
re' dr
fa
2,
sec.
REFLECTED REACTORS
8.7]
be obtained by
519
L\B\
Here,
M<*
L\B\
= Z>,/2 and
L\
(8.391)
=
L\
Z),/2.
Then
+L\B\){\ + L\B\)
w(l + L\B\){\ + L\B\)]
Zj(l
2,(1 +
K*-6VZ)
relation.
the B, given
by
(8.241) are also independent of the fuel concentration and the values
computed from these equations may be substituted immediately into the
criticality condition.
Now, for the present system [from Eq. (8.380)],
"
[1-i']^
It
(8,393)
(rrTbi)(P42
32<<>(k3
4)(1
+ )
(8.394)
Here C denotes all the various factors in (8.389) which are independent
of the fuel concentration.
If we use these two expressions in (8.392) and
solve for the critical fuel concentration
'
vo,[{\ +
NF, we obtain
CD2L\B\Bl)/(l
+ L\B\)\
<r'a2F)
^MWv*
The symbol 2^20) denotes the thermal absorption cross sections of the
nonfuel components of the core and 2'aF\ the thermal absorption cross
section of the fuel.
Thus the present method permits a direct solution
for Nf and avoids entirely the trial-and-error computations required for
the standard two-group method of Sec. 8.4.
It should be evident from
the procedure outlined above that the computational effort involved in
the Feynman-Welton two-group model is only a small fraction of that
required in the standard method.
This fact may be more clearly demon
strated by considering a numerical example.
As an example, let us compute the critical mass for the nuclear package
power plant analyzed in Sec. 8.5.
Since the group constants selected
for the present method are defined in the exact manner as those of the
method, the data listed in Table 8.5 are directly applicable to
The only information we lack is the values for the
and the constant C. The direct substitution of the appropriate
standard
this calculation.
Bi
REACTOR ANALYSIS
520
B\
= 0.007444
cm-J
B\
[CHAP.
= 0.007878 cm-2
= 61.9 cm
(8.396)
If
8.8
two-group
system.
neutrons
Generated for wjivans (University of Florida) on 2015-09-22 17:21 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
It
involving many groups of diffusing neutrons such that the one- and twoHow
group models would be special cases of the more general theory.
ever, it is to be noted that the one- and two-group models are really
based on fairly crude assumptions and are expected to be applicable to
only a limited class of reactors; even so, the use of either model is justified
only by previous successful application in reactor calculations.
Much
more confidence in a many-group model would result if it could be derived
from a more basic treatment of neutron problems.
At least one such
approach has been sufficiently successful to justify detailed consideration;
this approach, known as the multigroup method, is based on the age-
equations.
sec. 8.8]
REFLECTED
REACTORS
521
and has
a multiregion
reac
-D()V>(r,w)
+ Z.(u)
4>(t,u)
- du q(T,u)
S(r,u)
A < u <
0 <
wlh
(a)
< U*
(b)
u = utk
(6.54)
(c)
with
S(r,u)
This
[|o"'h
du
2}kMr)]
(6.90)
set of equations
has already been applied (in Sec. 6.3) to a singleregion reactor, and it was possible to obtain a solution there by the
method of separation of space and lethargy variables; as has been dis
cussed previously (Sec. 8.1b) this is not possible, in general, in the
case.
Thus we are led to consider techniques of numerical
of
the
The method under consideration
integration
equations above.
here is intended to be a natural one for calculations using (6.54) and
(6.90) in the multiregion case performed on large-capacity fast computing
multiregion
machines.
the zero of lethargy such that there are (for all practical
purposes) no neutrons in the reactor whose energies are greater than
that energy corresponding to lethargy zero (e.g., 10 Mev). Consider
the lethargy range from u = 0 to u = uth as divided up into AT intervals
Consider
AUi =
Mo
"v
i
= 0
=
= 1, 2,
(8.397)
interval; thus
D(u) V20(r,u) du +
=
j"
\
2(u)
4>(t,u)
du
q(r,u) du +
'
S(t,u) du
(8.398)
original form of this method (not exactly the one given here) was developed by
the Physics Division of KAPL [see R. Ehrlich and H. Hurwitz, Jr., Nucleonics, 12
. i
(2), 23 (1954)].
1 The
REACTOR ANALYSIS
522
[CHAP.
In order to progress from Eq. (8.398) when confronted with the unknown
variation of the neutron cross sections and the flux with lethargy, the
following definitions are made :
Dt
y~
fu'
*.(r) =
(r,u)
I
Jfi-i
'
du
(6)
(r,u)
S(u)
'
4>{t,u)
(8.399)
du
du
S/(m)
J
JUi-l
u =
;
du
4>(t,u)
{r,u) du
J/ Ui-l
- 9(r,u.)
i(u) du
[
^
S(t,u) du
gfou.-,)
+
if
+ 2~AU,:(r)
/"
-^Aw.V^r)
f"t
The definition of 2/, is intended for later use in connection with the treat
ment of the term <S(r,u).
If we use Eq. (8.399) in (8.398) and perform
the integration in the final term on the right-hand side of (8.398), we
obtain
j~t
S(r,u) du
we apply (6.90),
*(r,u) du +
k-i
(8.400)
2}"th(r)]
(8.401)
^note
SAui
itAu* =
lj,
Now define
AT
j(u) du
(8.402)
IJZ,
2?
been
= xj.AUi
2/.**(r)Au*
(8.403)
^7x(t)
0,h(r)
AwlV+1
S(t,u) du
JV+l
J*i
(a)
4>(i,u) du
/
V
(r,u)
D(u)
du
(8.404)
sec.
REFLECTED REACTORS
8.8]
523
-TMw.V^r)
+ 2^AM,^(r)
=
g(r,u,_i)
q(T,Ui)
be regarded
K&Atti
hi
ZA4>*(r)Aw*
as a restatement
of Eqs.
(8.405)*
(6.54a),
(6.54c), and (6.90), since the thermal equation can be formally included
of Eq. (8.404) and the following limits:
g(r,w.v+i)
= 0
(8.406)
9(r,M.)
Z**(r)Aiii
(8.407)
-AV^,(r)
N+l
+ (S + Z,)*(r)
with
<fc(r)
2ft._,*.-i(r) +
ifcAWi
z7**(r)
(8.408)
(8.409)
<fc(r)Au (8.409)
is
it
set of N
equations for N
unknowns, the <fc(r)
then,
since the equations are homogeneous
to be expected that there will
determinantal equation [cf.
be a condition for solubility in the form of
This
is,
lM+i = 0
Eq. (8.188) in the two-group case] which would serve as a criticality con
dition. The average cross sections needed for Eq. (8.408) present no
more problem than in any other scheme based on Eq. (8.405) and accom
panying definitions; since these calculations will be discussed later, let us
assume here that they are known.
However, the removal cross sections
ANALYSIS
REACTOR
524
[CHAP.
medium for which the materials and lethargy groups are so chosen that
fissions are caused only by thermal neutrons and neutrons from fission
appear only in the highest lethargy group, thus
jiAwi =1
2,+1 =
For this
case, Eqs.
Xf
it
= 0
for
k >
2A
= 0
for
k <
N+
-ThV'frii)
-DlV2Mr)
(2^+
+ (2^+ 2,)2(r)
-IT^Mi)
2,)*i(r)
(2a*
-D^V^wii)
2Rk)<fc(r)
2.+1^+1(r)
(8.410)
1
0]
*Z>h.v+i(r)
= 2,<Ai(r)
(8.411)
= 2i_,<fc_l(r)
=
2fi>iV(r)
Mr)
thus
the
Eqs.
+ B*F(i)
= 0
(8.412)
N +
(with
with
ous geometries
the
2ih =
2^, A,,
D^[,
and L*. m
DJX*)
1
+ LlB>)->
(l
"? (1
[[]
(8-413)
we compare this result with the accurate equation (6.78) for a bare
= 1), we find that the
reactor in the age-diffusion approximation (with
removal cross sections should be chosen so that
If
[no+fe+tF)]"1
i-1
be carried
step further
(8-4i4)
if
REFLECTED REACTORS
sec. 8.8]
525
[nO+S+^r~[n(i+)n(i+^)r
i
i=1
If
JV
^exp(-2fc)exp(-*22S)
~
<
Vs.
_-/Z*
i-l
va
Z/2.
i-l
be chosen to be consistent
with
(u,h S.(tO du
&,{u)
Jo
fu"D(u)du _
jo
*2,(u)
V / xA
Z/V?2<A
=
YTjjT
Z/VV<
i=l
D(tt)
C?M
(8.417)
Further, if the lethargy intervals are sufficiently small and the cross sec
tions sufficiently slowly varying over the intervals that an average over a
product or a quotient of cross sections is approximately equal to the
product or quotient of the separate averages, both statements of Eq.
(8.416) are consistent with the choice for each interval:
2*
(8.418)
interval.
pjA(r)
(8.419)
which seems to be consistent with Eq. (6.546), at least for very small
intervals Aw,; however, Eq. (8.419) is by no means a necessary conse
quence of (6.546). Indeed, the assumption of Eq. (8.407) is only an
526
REACTOR
possible
ANALYSIS
relationships
[CHAP.
which
may be used.
As
(8.420)
where the Aki may be chosen for consistency with an appropriate slowingdown kernel.
It is even possible that a set of Aki might be so chosen
that the resulting multigroup equations gave better representation of the
slowing-down process than the original age-diffusion theory which basi
^'(r)
are the core and reflector fluxes, respectively, we might take the set
(1)
(2)
(3)
(4)
<#(r) is nonsingular
=
*f>(R)
and nonnegative
#fi)(R)
-Dfv^'(R)
#>(R') = 0
-Z)v#*>(R)
(8.421)
functions,
The functions
<,(r)
may be expanded in a
over the range includ
(8.408) and
sec. 8.8]
REFLECTED REACTORS
527
appropriate
the functions Z and U are given in Table 8.3 for a few geometries.
The
parameters /i* are obtained by assuming that the core fluxes are propor
tional to a space function, as in (8.412). Then the solutions for B2 are
(>
L\
where
'
fe)
found from
- g(i
fum
LI
and
Ri
(8.424)
^a
The
(S-
Zb<)r
x\
be complex.
(8.425)
if
satisfy
the reflector (with the proper constants) except that there would be
zero source term on the right-hand side of the equation for group one.
In this special case the solutions for the reflector would be of the form:
0i(r)
= a07(xi,r)
*.(r)
aiU(xhT)
N+l
(4^F)
\*2
*1/
-UM,
etc.
(fU26)
REACTOR ANALYSIS
528
[CHAP.
unknown
The 2(N +
coefficients.
homogeneous equations
obtained by matching flux and current in each group at the core-reflector
interface allow for a complete solution except for a scale factor in the
The condition for solubility of this set of 2(N + 1) equa
flux, as usual.
1)
1)
A = 0
(8.427)
.
.
2,
- 9(r,uf)
-AAM<V2^(r) + Z^Au^r)
1,
(t*
c.
xi,Aw,F(r)
(8.428)
is
Here the total number of fissions per unit volume per unit time caused
by neutrons of all energies at space point has been designated by F(i),
This
replacing the summation over lethargy groups of Eq. (8.405).
technique.
it
Then,
(8.428) give the complete history of one generation of neutrons.
necessary to apply the condition that the spatial
for the reactor case,
distribution of fissions calculated from the results of applying Eq. (8.428)
must be the same as the source distribution with which the generation
is
was started.
In reactor calculation, then, certain source distribution
which we shall call F0(i)
chosen (somewhat arbitrarily) to start the
calculation supposing that all the constants in all regions and groups are
This
known, then the spatial distribution of fissions may be calculated.
as
distribution
be
called
and
used
source
distribution may then
/*\(r)
This process may be repeated until the calcu
to start new generation.
a
go
SEC.
REFLECTED REACTORS
8.8]
529
(r,u)
du
(r,u)
Ui-l
(r)
({2^
(r)
(8.431)
(a)
tt(i,Ui)
+ +
o(r,M,) = 2((w,)^(r)
<fc(r)
we call
(8.430)
du
/*g(r,uO
,(r) = i[q(i,Ui)
(8.429)
2((m) <p(x,u) du
J/
(iZ().
(6)
?(!,,_,))
(c)
(i,Ui-i)}
with
q(T,Ui)
= 2,(w.)
(r,u,)
(8-432)
(d)
is
it
fm
is
If
q(t,u) du =
j"'
3<(r)
Ibid.
530
REACTOR
ANALYSIS
[CHAP.
Equation (8.432d) may be used in Eq. (8.428) to eliminate the slowingdown density out of the ith group, g(r,Mi) ; thus
- A-A^V^r)
[I^Aw,-
+ 2(2,)u,]&(r)
1
(^rj (r>u<-i)
+ "iAtaF(T)
(8.433)
q(T,u0)
The calculation
lethargy
(8.434)
A differential
group.
-D'1'"Aw1VJ*r(r)
[S'.I'Am,
+ 2({2()l^'(r)
^"'Au^to
(8.435)
is written for each region of the reactor, where the superscript (n) desig
Proper boundary conditions are chosen to
nates the region considered.
connect the 0in)(r) and their derivatives at interfaces between regions,
qM(r,Ul)
(SBjfffWfr.uO
= (Z,)i?[2ft-(r)
0]
(8.436)
-D2">AM2V2&">(r)
u0
[2>Au2
=
[l
2({2(r)
jjjH!]
8(">(r,0
xS^Fnr)
(8.437)
sec. 8.8]
REFLECTED
REACTORS
531
The 0"(r) are then found for each region, again by numerical integration
The slowing-down den
and application of proper boundary conditions.
sity from the second group may be calculated from
(pyS^r.ud
g(">(r,) =
= (*2,)S>
|W>(r)
^ff]
(8-438)
The process thus proceeds from group to group until the (N + l)st group,
For the thermal group Eq. (8.433)
i.e., the thermal group, is reached.
does not really apply and we must return to Eq. (8.428), along with
Eqs. (8.404) and (8.406), to write
-S&iV'fl&xW
with qM(r,un) to
g(r,uw)
As
be calculated
(*2()(r,Ujv)
(8.439)
from
= (2,)
qw(r,uN)
2^'(r)
- ^g=^j
(8.440)
region, 4>^(t), are known at every space point [the additional subscript
zero indicates that these fluxes are derived from the assumed source dis
tribution F{,n)(r)].
The comparison of the derived source distribution
N+l
i
- 1
2>#;>(r)Au,~n")(r)
y+i
v
L4
j.Aw,
f
J reactor nn)(r)
dr =
di
J[reactor F^(i)
(8.441)
where the appropriate form for F'0n)(r) is to be used in each region when
performing the integration.
Equation (8.441) states that one neutron is
introduced into the reactor to start the generation.
At the start of the
next generation, we have n0 neutrons, given by
N+l
n0 =
y J[reactor
2V>W>(t)Atti
11
0
dr
(8.442)
>= 1
By definition,
REACTOR ANALYSIS
532
[CHAP.
fco
is
that this
use of the
As
'
(8.445)
dr =
(8.444)
reactor
S^'^'WAm,-
t-i
FP(i)
1
J[
FV>(r)
AT-+-1
are obtained,
fc,
w+i
Wl
t-l
/reactor
(8.446)
W*< *
(8-44?)
general,
is
constant
it
JUi-l
2(u)
h~rir.
JUi-i
du
(8-448)
du
is
2(u)(r,u) du
s hh^i
4>M(r,u) du
["'
%H)
is
is
it
is
is
is
neglected.
REFLECTED REACTORS
sec. 8.8]
533
If
mr) s
q^r.Ui-x)
f
Eq. (8.433) may
be
(>(r)
Em
T
T/(n)
m
(8.449)
S_r ^
*L
jpj>(r)
oJ"W5>(r)
af s
(r\
frj^ [22'
CJ>(r) =
V(SHr)
+ 2({21)]
-CJ->(r)
(8.450)
been used:
and
<>
- |g
REACTOR ANALYSIS
534
[CHAP.
#(m Ar)
[W<p]m
be denoted
[#]
(8.452)
f[W-+
[WT'Wi
2
where
be approximated
be
kFWPU
a(Ar)s
- 2[lF!n)]-
by (for example)
(8-453)
- [TPM - [SP)m
(8.454)
(Ar)'[C],
(8.455)
W'"){r)
(aV
[^">]m
Note that, from the definitions above and the assumption of (8.432d),
In (8.454) the fc$"> and [<S;n)]m are known from the calculated constants,
the assumed source distribution and the results of the calculation of pre
may proceed, for example,
assumed that
$"(0)
[Wp]t
[W?U
W]t
~mh
k?WPh
- [Slh
(8.457)
etc.
is
is
is
[inn>]m
it
is
is
it
higher index m.
Z)f>
jfl.Ar
(8.458)
[WT>]*
[WT'Ib*
';2,]^Ar
[W?>]**J
(8.459)
Ibid.
is
is
it
vious groups.
REFLECTED REACTORS
sec. 8.8]
535
(Ar)2
+ 6*,
r+(r)dr
m Mj Ar
~2
i
/ra
miAr
*.)}
(8.460)
where
This approximation
may be used
by
recognizing that
V#1!(r)-dS
Vs#(r) dx
-D^Aw.4*
p-^r"
rfr
[wrk-o
ark+i
may be
[wf]B.
WW. - [WTk-i)
(8-462)
|
[HT>U
A"
[I
(8.461)
(8.463)
total neutrons
total neutrons
lost by
degraded into
next group
[absorption
neutrons degraded
from previous group
[Total
is,
is
If
it
volume
-D^Aiti
J/
aurf ace
-D^Au,
JI
J<il)(R)4irR2 =
REACTOR ANALYSIS
536
[CHAP.
= 0
[Wp].
(8.464)
where index
denotes the final, outermost region.
The resultant value
of [Jiy]/)0 may be seen to depend on the choice for
of Eq. (8.457)
which is otherwise unspecified.
Any general choice of [Hr!"]i will not
result in a zero value for [JT/'Ib,.
One simple way around this difficulty
is to define two functions
Thus
and
(B
such that
- [an
=
= o
[],
[}]. = o
W%
(o SRz,
{*Abd)
(8-466)
of
as a linear combination
a and
[Wf]
(B;
thus
[af]m + cfffii-'l,
(8.467)
be calculated
c
at
is determined
condition
c[<&f)B, = 0
(8.468)
is,
While the scheme just presented for handling the outside boundary
condition has been used successfully, it is not the most convenient or
the most accurate method.
The presentation of improved methods
however, deemed outside the scope of the present discussion; the
KAPL
should be
Further,
values are not regarded as having physical meaning.
noted that one of these
never really calculated; the quantity [WJ"]b,-i >s related to
through the recursion relationship of Eq. (8.454) for n
[W'J2)]fl. and to
and so does not need to be computed explicitly.
it
these
is
[av>]+i
Ct
sec.
REFLECTED REACTORS
8.8]
537
about the reactor system. Besides the multiplication constant, the dis
tribution of flux in energy and space is given, from which may be com
puted, for example, the distribution of neutron absorptions or fissions in
energy and space or the current of neutrons of various energies emerging
10"1
8
6
r~
/ 2,
i
fE
rJ
c
o
r"
i
3
<nirT2
10
w
M
u
Q.
lJ
pJ
o
O
1, i
r--J
102
L_J
r-
(0
'
10
r1
+-2 a
10"
10
12
14
16
18
20
Lethargy
Fia.
8.31
for
a fast computing
is the collection of nuclear data and the computation of the average cross
sections for the various groups.
Even this chore has been eliminated at
many of the major computing facilities wherein "libraries" of subroutines
Mev
These
cross-section
538
REACTOR
[chap.
ANALYSIS
10"
io
\"
"
:,-ic
.
1
[
r1
-2.
10"
10"
10
12
14
16
18
20
Lethargy
Fig.
8.32
Table
8.9
and
Reflector
Core
0.0003946
0.0000835
0.0004274
0.0005713
0.05827
Fluorine
Oxygen
0 05742
Uranium
Inconel
(235)
Reflector
0.0002365
0.0004096
0.000256
0.0002365
0.06577
0.06526
0.00004418
0.01546
0.00127
sec. 8.8]
REFLECTED KEACTORS
539
2.2
2.01
1.8
1.6
E
..jr"
IQ
D
c
1.
1.0
S 08
3
n
5
re
- -i
0.6
Reflec to
0.4
0.2
10
11
12
13
14
15
16
17
18
Fig.
8.33
1.4
1.2
1.0
0.8
0.6
0.4
0.2
Fio.
8.34
10
12
14
considered; three of these are: (1) the extension of the method to cylindri
cal geometries, (2) the modification to give improved results for hydrog
enous systems, and (3) the improvement of the equations to include
more information about the angular distribution of the neutron flux so as
to gain an improvement
in the treatment
of the slowing-down
process.
0f9
aoxovaH
sisaivnv
dVHO]
8*0
1
! I
L.
so
L_
L.
T
-1
|
fr'O
1
j
isej xnu le
_
.18)1183JO 8J03
i L.|
90
'0
*1
....
'T
U
0
'oij
"
L.
xn|| 03 8
38|}<JO UlSJJ81 83
ro
r J
Z"0
E"0
c,{vg
uoissij
fr
01
21
81
91
r
r
n
'oij
gg's
01
iod
%\un
ZT
'S2iv\\%o\
frt
9t
anp 0} suoa-jnau
8t
jo
U3A13
REFLECTED REACTORS
sec. 8.8]
541
The average cross sections and results for a typical multigroup calcu
The particular assembly
lation are shown in Figs. 8.31 through 8.39.
chosen is a spherically symmetric system consisting of core and reflector.
The fuel is U2S6 and the principal moderating material in both the core
0.5
1
nl
S3 0.4
0.3
0.2
if
0.1
-44-1
,
8
10
12
14
16
18
Lethargy
Fio.
8.37
0.6 r
10
12
14
16
18
Lethargy
Fio.
8.38
unit lethargy.
REACTOR ANALYSIS
542
[CHAP.
1
1
1
1
1
1
= 1 Mevflux
(0.821-1.35
u=
1/10 Mevflux
(0.009-0.183 Mev)
100 ev flux (33.7-112 ev)
m=
0 E = 1 / 10
\a
v
102-0. 125
ev f lux (0.
_ (' ottc
\sar ne sc ale)
F efle<:tor
\\
<i
\
l
Cor e-
12
Radial coordinate
Fig.
8.39
ev)
'V
I'm
1
1
Mev)
'
\
V
~T
16
20
24
\
28
Spatial distribution of average flux in four fast groups and in thermal group.
The originally chosen form was based on previous experience with similar
Figure 8.35 shows the lethargy distribution assumed for the
assemblies.
source neutrons [the j, of Eq. (8.402)] along with the average neutron flux
in each group at two special space points, the center of the reactor, and
the core-reflector interface.
The thermal flux is not shown in Fig. 8.35
since the fast flux and the thermal flux are not dimensionally equivalent
quantities.
For each neutron introduced into the assembly (in the distribution
described in Figs. 8.34 and 8.35) the following final results are obtained:
sec. 8.8]
REFLECTED REACTORS
543
Oooo
and
neutron enter into fission processes
0.28139 neutron enter into fission processes
0.08837
as fast neutrons
as thermal neutrons
Thus,
76.1 per cent of all fissions are caused by thermal neutrons; 75.7 per
cent of all neutrons absorbed are absorbed as thermal neutrons; 90.0 per
cent of all neutrons escaping from the assembly escape as thermal neu
Figures 8.36, 8.37, and 8.38 show the distributions in lethargy
trons.
of the fissions, absorptions, and escapes, respectively (all per unit leth
argy)) m the fast range. Figure 8.39 shows the spatial distribution
through the core and reflector of the average fluxes in four particular
fast groups plus the thermal flux distribution.
The assembly was found to have a multiplication constant k = 0.924.
PROBLEMS
8.1
Determine the critical equation for a one-velocity spherical reflected reactor
which has the following specifications: (1) a homogeneous active core of radius a;
a; the neutron properties of which are,
(2) a spherical shell reflector of thickness b
in general, different from those of the core; (3) an infinitesimally thin spherical shell
of absorbing material at the core-reflector interface; this absorber has the property
that it transmits a fraction or of all neutrons incident on either face.
Show that the critical equation for this system reduces to the usual relation for the
one-velocity reflected reactor when the transmittancy a 1. Obtain an expression
for k(a) and compute
Sk
*()
"
k(l)
k(a)
0.95
1 cm, and
544
ANALYSIS
REACTOR
[CHAP.
where <P(t,u) is the neutron flux in the core and a some constant independent of
lethargy (which can be determined from the albedo of the shell material).
(3) The
Fermi age model applies.
(4) The fissionable material has a zero fission cross section
at all energies except thermal energy. (5) One of the components of the circulating
fuel has an (n,2n) reaction cross section which is different from zero for all u < uu.
(6) All neutrons produced from fission and (n,2n) reactions appear at u = 0.
o. Obtain expressions for the slowing-down density g(r,u), the fast flux (r,u), and
Use the relations
the thermal flux 4>u,(r).
- 2(u) + 2.(u)
S.(u) - 2/(u) + 27(u)
2,(u)
=
q(T,u)
Z2t(u)(i,u)
with
+ 2(u)
d.
is
is
6.
is,
where 2/ and 2T have the usual definitions and 2 denotes the (n,2n) reaction cross
section of the circulating fuel.
b. Find the critical equation for this system.
c. Derive an expression for the multiplication k in this reactor.
d. Evaluate any arbitrary
constant which remains in the results of part (a) in
terms of P, the total power output of the reactor. Note that the energy release during
a fission reaction will be, in general, different from that released from the (n,2n) reac
tion. Also, the energy released by the (n,2n) reaction may be dependent on the
energy of the incident neutron.
bare spherical reactor consisting of the materials
8.4 a. Consider a homogeneous,
Compute the critical radius Rc of this reactor for Nu/Ny =
described in Prob. 4.8.
10,000 on the assumption that the resonance escape and fast nonleakage probabilities
are unity, that
the one-velocity model applies.
On the basis of the information given in part (a), compute and plot Nu/Ny for
criticality as function of reflector thickness
by using the above sphere of radius Rc
For these calculations assume that the same material
as the active core.
used in the
reflector as in the moderator of the core and that Dc Dr.
c. What fraction of the decrease in fuel mass obtainable with an infinite reflector
realized with
reflector of one diffusion-length thickness?
Plot the ratio of the decrease in fuel mass to the reflector mass as a function of
reflector thickness.
e. Plot, as
function of reflector thickness, the fraction of neutrons produced per
second in the core which are (1) absorbed in the core, (2) absorbed in the reflector, and
(3) escape from the reactor.
for
reflected spherical reactor which has an active core radius
8.5 Compute
37.2 cm and a spherical shell reflector of thickness 20 cm.
For this calculation
use: (1) the material composition described for the reactor in Prob. 4.8, in the case
Nu/Ny = 15,500; (2) nuclear cross sections for fuel and moderator given in Prob. 4.8:
(3) the two-group method.
Assume that the moderator and fuel are uniformly distributed in the reactor core
and that the reflector consists of pure moderator.
Compare the value of
obtained
from this calculation with the value obtained for this reactor on the basis of a onevelocity model (as computed in Prob. 8.4).
8.6
right cylinder of beryllium containing some Um has been left
homogeneous
The cylinder
standing upright on
exposed in an unattended area of a laboratory.
is
REFLECTED REACTORS
545
Beryllium:
= 23.6 cm
= 0.70 cm
Za
= 0.0013
p = 1.84
Water:
L
D
I/Employee
cm"1
g/cm
= 2.85 cm
= 0.16 cm
Beryllium
cylinder
and the following idealized model: (1) One-velocity diffusion theory applies in all
regions.
(2) All surfaces and diffusion media, which may be near the cylinder (such
as floor, etc.) may be ignored, except the presence of the employee.
(3) Employee is
composed of pure water.
CHAPTER
REACTOR KINETICS
models.
the problems.
describing
them.
to the analysis
duction and decay of the principal fission fragments that appear as poisons
The effect of these substances upon the reactivity of the
in the reactor.
This treat
system is examined and their practical significance noted.
ment is concluded with a study of the use of burnout poisons for con
trolling these long-term changes in reactivity.
Nearly all the physical situations considered in these analyses refer to
This restriction allows solutions to be obtained by the
bare reactors.
Thus the kinetic features of a specific
method of separation of variables.
'J.
M. Stein, "Simplified
Equations,
1951.
546
SEC.
REACTOR
9.1]
KINETICS
547
A fur
problem can be examined independently of the system geometry.
ther restriction which is made is that the transient problems of the first
category arise from nuclear changes which are introduced uniformly over
the entire reactor.
Transients in the flux (and power and temperature)
due to local changes are best treated by numerical methods or by the
application of perturbation
theory.
9.1
a. Statement
is
is
if
Fermi
is
preliminary studies and for gaining insight into the essential features of
many reactor problems.
The Fermi age or continuous-slowing-down
diffusion theory, then, offers some generality without introducing exces
sive computational
difficulties.
The appropriate time-dependent differential equations for the combined-slowing-down diffusion theory may be obtained by recognizing
that the time rate of change of the neutron density
given by the differ
is
is
n(t,u,t)
S(r,w,<)
Slh(r,<)
+ D(u,t) V20(r,)t,O
D.(0
V->lh(r,<)
S0(w,<)
- S^O
*(r,w,0
-~
for u >
q(x,u,t)
ulh
(9.1)
u = ulh
(9.2)
<MM)
for
548
REACTOR
ANALYSIS
JL 0(r,u,i)
to be
n,h(r,t) =
g(r,w,<) = $2,(ti,t)
and
[CHAP.
J-
*,(r,<)
(9.3)
(T,u,t)
(9.4)
= thermal-neutron
flux
<th(r, = thermal-neutron
flux around r at time
/
The set of equations (9.1) through (9.4) define the distribution of neu
trons in space, time, and lethargy for a homogeneous system (note that
all neutron constants are independent of r, but are, in general, functions of
u and
t).
(2)
set
The flux (or neutron density) must be symmetric about the cen
tral axes of the reactor; or the solutions must have no singularities
at points not occupied by localized sources.
The flux must be continuous through interfaces between dissimilar
media.
(3)
(9.5)
time-dependent
to steady-state
operation
in
multiplying slab
differential
(Sec.
equation
is
approach
around r at time /
Sth(r,0 = number of thermal neutrons produced by sources per unit vol
ume per unit time around r at time t
5.4a).
(9.2),
SEC.
KINETICS
REACTOR
9.1]
549
rzy*tt(r,0
(9.6)
cos
<fch(x, = (?=
y (Ane-D^"
cos anx + Cne-D^" sin
yKx)anx
(9.7)
(?
(Ane-D^"
+ Cne-D^" sin
yKx)
(9.7)
where the a and y are given by (5.137) and the X and to by (5.139).
The coefficients An and C are obtained from (5.143).
The second special one-velocity problem previously considered dealt
with the spreading out of neutrons through a nonmultiplying infinite
medium after an initial burst at t = 0 from a plane source at x = 0
In this case, Eq. (9.2) is used everywhere except at
(Sees. 6.2a and b).
x = 0, and the source condition is given by [cf. Eq. (6.7)]
"
with lim f
S*(x,t) m 0
(9.8)
nlx.t) dx = q0
nih{x>t)
As
(9.9)
(4rZW)
a somewhat
p^,
e_B'T'h
- J *h(r,0
(9.10)
is the resonance-escape
nonleakage probability
0,h
= y2fptigth4>*(T,t)
is given by
with
B2 =
al
fr
where
+ S?*,h(r,0
(j^j
- AhV'^frQ
slaD reactor
(9.11)
Note that Eq. (9.10) implies that all neutrons are prompt.
The solution th(T,t) to (9.10) must satisfy the conditions
(1)
*th(r,0)
(2)
nonsingular
<th(r, is nonsingular
(9.12)
(3)
Mft.t)
*0(r)
is
(9.12)
REACTOR ANALYSIS
550
[CHAP.
The
V2(r) + 52*o(r)
(9.13)
where B is given in (9.11) for the slab (see also Table 5.3).
The solution to (9.10) may be obtained by the method of separation
of variables used in the case of the slab problem of Sec. 5.4a [cf. Eq.
(5.128)]; therefore, take
*th(r,<)
0(0
*0(r)
(9.14)
Clearly, all conditions of (9.12) are met by this relation since <o(r) is
chosen to meet conditions (2) and (3). However, we require also
0(0)
(9.15)
The substitution of (9.14) into (9.10), using also (9.13), yields the
equation
d
with
L2 = Dth/Zla
that
is,
v^{\
+ L2fi2) di
^"Ua+w)-1]^0
(916)
Z?(l+L2fl2)
= JoPnl
(9.18)
(9.16) reduces to
G(t) =
6k
G(t)
(9.20)
Note that these results depend upon the assumption that no change
made in the dimensions of the bare reactor.
The solution to (9.20)
is is
the life of
Further,
and pNL
+ LIB2)
where
lo
^"(1
1.
Note that the cross sections which appear in this relation were derived
from the noncritical system of Eq. (9.10); therefore, by definition
^
The factor on the left-hand side of Eq. (9.16) may be recognized as
if is is
.
The first term at the right may be recognized as the
effective multiplication constant of the system after the change is made;
seen to be
0(0
= e**'"
(9.21)
SEC.
HEACTOR
9.1]
KINETICS
551
<fch(M)
is given
by
*o(r)e"'
(9.22)*
Thus we find on the basis of this simple model that the flux rises and falls
exponentially in time, depending on whether the change made at t = 0
results in a k which is greater or less than unity, respectively.
The period of the reactor is defined as the time required for the flux to
times its value at I = 0 (in the case of an increase in k).
the above model
rise to a value
By
The period = T =
(9.23)*
OK
It
a complete
the
which
The differential
equations
S(r,u,t)
source is
vZf 4,^,1)
(u)
(9.24)
(3)
The thermal sources are due entirely to the slowing down of fast
neutrons; thus,
Sth(r,0 = g(r,W)
(9-25)
(4)
and
0(R,m,/) and
<ti,(r,0
are
are everywhere nonsingular
ih(&,t)
are zero
(9.26)
(9.27)
REACTOR
552
(5)
ANALYSIS
[CHAP.
o(r,u) and
<o(r)
o(j,u)
(9.28)
*.(r)
(9.29)
= 0.
(6) At t = 0, a change is made in the thermal cross sections so that the
reactor is no longer critical, and these new cross sections remain
t
> 0.
IT ^ **(r>0
*2,,(r,0
- 9(r,uth,0
(9.30)
= ~ O,hV2,h(r,0
+ Zf*fcfr,fl
We know from the analysis of the one-velocity bare reactor that the
solutions may be written as the product of a space function and an expo
nential in time.
Let us look, then, for solutions to the above equations
in the form [see also (6.59) and (6.60)]
*lh(r,<)
- F(r) ***
nr)
Wliere
h
V
Q(0)
^*
= (AB2 +
(9.27).
Q(uth)
[(')B2
/0"
afu p j -
(y"^;
The substitu
(9.32)
(6)
'W
(c)
to (9.32a), which
^yk(y)]
dM')
(9.33)
(D*B* + S*)**
(9.34)
exp
[(')*'
(') +
fr(u,;Z|(M>)]
rf
If
written
case
9(r,0,0
~
(1) may be
SEC. 9.1]
KINETICS
REACTOR
lr
be simplified
we introduce the
du
Uth
M)
Jo
if
553
Ub
2.()
following identifications1
(9-35)*
+ A52)
and recognize that the first term of the exponential is simply gth, the fast
nonleakage probability to thermal, and that the second term is pth, the
Thus, (9.34) may be written
resonance-escape probability to thermal.
her*1'
1
+ Jit*
"_
(9.36)*
k =
(9.37)
2* + DlhB*
is,
Equation (9.36) may be regarded as the defining relation for n and must
be satisfied if the forms (9.31) are to be solutions to the set (9.30).
It should be borne in mind that the solution obtained in this way
at times long after the initial pertur
applies only asymptotically, that
a
is
0.
if
0)
exp
density prior
du'\
,2jrW(r)
|^
<z(r,u,0)
{^(u')
(9'38)
We may conclude, therefore, that the solutions (9.31) describe only the
long-term behavior of the system and omit entirely the transients which
arise immediately following the introduction of the changes in the cross
sections.
is
lF
is
Eq.
- lF) .h(r,< -
lF)plhe-Blrth
only an approximation.
function
of time
given in
is
for
is
The expression
(9.40).
vZfit
lr
Solutions
g(r,^h,)
0;
=
bation has occurred at
however, the result given may be
good
=
It evident that complete solution
approximation very near
has not been developed since (9.33) does not satisfy the initial condition
Thus,
the nuclear properties of the reactor at steady
on the reactor.
are denoted by the superscript (0), then we find that
state (for
<
554
ANALYSIS
REACTOR
[CHAP.
(1) through
(6) of
density at time
by repeated application of the form
q(v,t) = q(u
if
Am, t
- At) -
q{u
and lethargy
Am,
u is to be obtained
- At) ^~^r
Am
A<
Am =
2,(m,0 v(u) At
a thermal
neutron
ith
previously
is
If
is
obtained
is
given in (9.35)
in (9.18).
The function lF,
on the other hand,
an entirely new quantity and needs to be examined
in detail.
It evident from the form of this function that
may be
precisely the relation
is
(4.47)]
Average time
to slow down
through
average number of
collisions needed to
slow down through An
Am
Am
X,(w)
(9.39)
lF
if
lv{uyz.(u)
As
approximately
if
is
which
SEC. 9.1]
so that,
<K 1.
REACTOR
KINETICS
Equation
(9.36)
+ n(U + h)
=1
555
where
y.
(9.41)
y-
+ If
I* ^
(rA^j-)
(9.42)
Reflector
Diffusion coefficients:
Cross sections: 2f,z<0)
Buckling: B\ =
Z)'
2?
- Z>]
l>
Diffusion length:
L\
- - - -JL
1
/)(0)
Core:
-D>Wc(r,0
[sc<o,
J*,(r,0
Reflector:
-2>jfW(r,0
(r,t)
[Zf>
J*,(r,0
_IA0c(M)
2<0c(r,O
(9.43)
- \jt4>n(j,t)
conditions
is nonsingular
= >k0,VMR,<)
DV*C(R,<)
(2)
(4)
*c(R,0
*(R',0
s(R,0
= 0
(VM)
where R denotes the core-reflector interface and R' the extrapolated outer
boundary of the reflector.
REACTOR ANALYSIS
556
[CHAP.
- F(i)e-'S1-'
M*,t)
= G(r)e-*2-'
(9.45)
V2G(r)
- x?G(r)
(9.46)
where
= 0
V*F(r) + *F(r)
= 0
and
(1)
(3)
DffVFQL)
(2)
D$VG(R)
(4)
F(R)
G(R')
G(R)
((
= 0
.
}
Thus all the conditions (9.44) are met, and 4>c(r,<) and >(r,<) reduce to
the proper space forms at t = 0.
Direct substitution of (9.45) into (9.43)
shows that the assumed solutions for the fluxes satisfy the required differ
ential equations.
(1)
DV*C(R)
(3)
(4)
*zy0c(r)
= 0
*,(R)
**(ft')
- MR)
^*yj
,Q ,Q>.
Equations (9.48) and (9.49) lead to the criticality condition which will be
of the type given in (8.11) for a spherical reflected reactor
BR cot BR
R)
1]
(9.50)
62al
(9-51)
where
B* =
4?
["c:S}0>
2aCl<"
5Zo]
*'
"
^ [2afi
Thus, for known reactor dimensions and original core and reflector cross
sections, and a given value of 6S0, a value of vc may be computed from
Then 52 may be related to Sk, which is given by
(9.50) using (9.51).
Sk =
--
(9.52)
Vc
9.2
a.
operation
(for
< 0).
Reactor Parameters
The Experiment.
reactor
length, as
sec. 9.2]
REACTOR
KINETICS
557
well as the determination of the critical size and the fast nonleakage
probabilities for a specific reactor geometry and internal arrangement of
nuclear components, is an essential step in the design and development
Unfortunately this phase of a project is
of every basically new system.
This is especially
of the critical size esti
REACTOR
558
ANALYSIS
[CHAP.
Neutron
burst
-Triton target
+ ,H2
2He3
2He<
+
+
on1
0n'
(9.53)
sec. 9.2]
REACTOR
KINETICS
559
by means of suitable counting devices, and from these one obtains the
The
decay rates of the spatial modes of the neutron-flux distribution.
determination of the various reactor parameters from a knowledge of the
decay constant for the fundamental
which follows.
mode is demonstrated
in the work
on the study of the space and time distribution of the thermal neutrons
in the specimen. For this purpose we introduce the following model of
the neutron processes in the system: (1) A short burst of fast neutrons
occurs either outside or within the test specimen at t = 0.
(2) These
Eq. (9.10)]
-Z>V0(r,O
+ 2*(r,()
= vXfPtbg^(T,t)
- ~ *(r,t)
where
4>(R,t)
= 0
(r,0) *o(r)
(9.55)
V**(r,0 + B'*(M)
with
2 = ~ (vS/Pthff,
- S.)
0(r,O
"^g^
(9.54)
It
is
(9.56)
*
(9.57)
REACTOR ANALYSIS
560
(i,t)
written
be
G'
is some constant.
p2
Thus take
= F{i) G(t).
where
[CHAP.
(7(0 = G(0)e-^'<
V2F(r) + (B2 + p2)F(r) = 0
(a)
^oy;
(6)
+ ain,Fmn.
= 0
with Fm.(ft)
= 0
(9.60)
where the Fmn, are the various eigenf unctions (modes) of (9.596), and the
am, are the corresponding eigenvalues.
If the results of Sec. 5.4a are
extended to the case of the parallelepiped, it is easy to show that
fimy cos
y,z
where
(2n
+1)
pm
(2m
- i.
"
dimensions,
(2s
(9.62)
1)
of the
respectively,
(9.63)
and in general,
p2
i.
/32
7. =
1)
(9.61)
7.2
an =
-* Pi..
(9.64)
If
is
- G(0) Fmn.(T)e-Dna'~"-B,)'
(r,0
(^..^^(r.O
.)
0,
1,
(9.65)
s
then
is
4>n,n.(i,t)
4>
(9.66)
These
where G(0) has been absorbed into the arbitrary coefficients amn,.
Thus, from (9.55)
are of course obtained from the initial condition.
*(r,0)
- *o(r)
aFmn.(r)
(9.67)
sec. 9.2]
KINETICS
REACTOR
561
fo(r)
F(r) di
(9.68)
exp [
Dv
*2
m,n,
- ^g^)
^
X F.(r)
t]
*(r,<)
the
(L
to extend
fy
^o(r')
FM(r') dr'
(9.69)
aooo
exp [-(t>20
+ vDa\w
mode F0oo(r).
i>v2/p,h^)<]F0oo(r)
(9.70)
is
X000
vZa
+ vDalo
multiplying medium
- wE/pag^
will
It
i'2
+ vDalM
vl</\\
Then,
in the
(9.72)
be convenient
is
exponential factor.
case of the
()'
decay constant
In the
(I)'
it
The
()'
- -
to separate
we write, in the
- vPlhgJ
(9.73)*
is
is
is
is
if
various
placed
if
is
as function of time.
When
density, or some quantity proportional to
plotted on semilog graph paper, the curve reduces to
this function
is
is
the detector
REACTOR ANALYSIS
562
[chap.
straight line over that portion of the time scale wherein the approxi
An example of such a curve is shown in Fig.
mation (9.70) is satisfied.
These data were obtained by Campbell and Stelson from measure
9.2.
ments performed on a 5-in. cylinder of pure water.1 The decay of the
2,000 1
10'
0
1
100
200
300
400
I
500
Time, Msec
Fig.
higher spatial modes mentioned above has also been studied, and a typi
cal set of thermal-flux distributions at various time instances is given in
Fig. 9.3. These measurements were made on a rectangular block of
beryllium. The curves in the figure show the shapes of the flux distri
The presence of the higher
bution at successive 100-jusec time intervals.
modes, which decay more rapidly than the fundamental, can be seen from
From these
the gradual shift to the symmetric fundamental mode.
within
mode
is
established
about
that
the
fundamental
it
appears
results
one millisecond after the initial burst.
It is evident from the relation (9.70) that the decay constant of the
1 Oak Ridge National Laboratory, Physics Division, Semi-Annual
for Period Ending Mar. 10, 1956, ORNL Report 2076, p. 156.
Progress
Report
sec. 9.2]
REACTOR
KINETICS
563
fundamental mode is simply the slope of the decay curve, and measure
ments of Xqoo may be readily obtained from data of the type given in
Fig. 9.2. The second and final step in the evaluation of the reactor
parameters, then, is the measurement of Xqoo for various alw, and this
10
13
Distance
9.3
(9
Fig.
block
at,
16
19
22
in.
(1)
[2<0)
material
(9-74)
Dal00]
REACTOR ANALYSIS
564
[CHAP.
(multiplying
media)
p[x+i*4.]
(nonmultiplying
E
to
T3
media)
Buckling a2000
(increasing
size)
"Critical size
Fio.
9.4
medium.
= 0
(2)
(9.75)
case
a00->
is the diffusion
oo
(9.76)
ORNL
BEACTOR KINETICS
sec. 9.2]
565
mixture,
With
0.1
0.2
Buckling,
Fig.
0.3
0.4
0.5
0.6
ao00(cm-2)
of
9.5
a0oo is
the intermediate
(4)
(a0Oo)erili<ml
Xooo-
Thus for
range,
<
"000 <
00
pthg+
If
that
(5)
For small
a0oo
ff*
It
g,
= e-wu,
(9.78)
is
45
566
ANALYSIS
REACTOR
[CHAP.
is,
9.3
is
a. Delayed-neutron Groups.
Most of the neutrons produced by the
fission process appear, effectively, at the instant of fission.
However,
small fraction of the neutrons produced in fission
emitted by daughter
nuclei from the primary fission fragments when these nuclei subsequently
Neutrons produced by this mechanism do not
undergo beta decay.
appear at the instant of fission but at some later time.
The delay period
is
is
being
is
X,C<(r)
(9.79)
XiC.W
is,
Xi
where C,(r) = concentration of beta emitters of the tth type which are
precursors of delayed-neutron emitters
= decay constant for the neutron emitter of the tth type
N = total number of delayed groups
In the steady state each emitter which decays must, on the average, be
f.Z^r)
(9.80)
is
f,
1,
I,
is
CO C
a o
'i
3 ^
cc
O
?
U
CU
r*
5?-
iO 00 Ci M
N
M 5
-h in
8-;o
o o
d d
p o o
5
5 o S
o ci o
CU CQ.
r.
ii
CO O
Cf &
PL, 2
n
o
o
o
f
o n
a)
2C
SO M
3 -a
_,
^
in
s a
OU o
o o as
o o
o
1S
lO 'f Tj< H
iO CN
t
m
J5 oi
cu
cu II
n
n qo oo e
CI N O N N
O O "i o o
o o o o o
o o o o
d d d
CD
t5
53
co 3
c
.2
c o
S 3
o
0
IO iO
bo
s~ S
n 'bo
&
&
03 Ci
HOO
_ w
n h
o o o o o o
o o o o o
eo N oo
CO CO O CO CO
53
n n
h
(i
t
h N
N
O
od
8
3
U5 to co
M IO
in io
o o o o
co
53
>> o
a
3
ii
n,
-S
h N
CO
667
(S. -a
WQ
^ o a
JJ
Sod.
cu b
d T3
cu .3
be
"3
ca q
0,
03 o
X
If S
a"
CO
XI Q
CO
c3 -= a
53
to O
cu 2 -i
S pi
J*
-o
c
i 13^
iO
-ill
o o o
Si '
s
o
- >
9
2 =3 -a
m co co on o>
N iO ^ co *f
Q S
"H CM eg m
m m
m in *
m in
BC 22 o
O O
O
O O O O
O O O O O O
ss
5 c
e
s
o3 cn
Q Ml
t~ 00 CD CO
oo -h
3
.2
eg m
o o o o o o
1 S.
'6q
iO
O
N
t-
J3
<
I
"
2 (
53
REACTOR ANALYSIS
568
[chap. 9
unit time is
Total neutrons produced
per unit volume per
unit time
where
/3
1\
Wr)
(1
- P)2/+(t)
(9.81)
is
and
j8,
i= 1
Table
9.2
ThJ
Group
54.0
22.0
6.2
2.1
0.52
0.18
0.00208
0.00838
0.01083
0.02890
0.01068
0.00221
53.0
22.0
5.30
2.00
0.51
0.18
'
1
2
3
4
5
0
Total...
Neutrons
per fission
0.00035
0.00445
0.00695
0.01796
0.01048
0.00383
0.044
0.063
4,
Uses
of
ful
1
ith precursor,
f,-2,0(r) +
(1
steady state,
/3)"2/<Kr)
(9.82)
"2/cKr)
[ifc+a-]
*2/*(r)
The last step in the equation follows from the fact that each precursor
decay yields one neutron, so that
N
=
(9.83)
'-
We may conclude from this calculation that in the steady state the
total number of neutrons produced from fission
not altered by the dis-
is
Neutrons
per fission
Half-life,
sec
Half-life,
REACTOR
sec. 9.3]
KINETICS
569
some difference
delays (see
point of the calculation
in a thermal
-DV'*(r,<) + Z.*(r,t)
~ *(r,J)
(1
of precursors are
- P)y2fPg*+(T,t)
AT
jtCi{T,t)
-X,Ci(r,0 + U2,(T,t)
P.h0h
2, WM)
(9.84)*
(9.85)*
= 1, 2,
where all the neutron constants which appear refer to the thermal neu
trons.
The second and third terms on the right side of (9.84) account
for, respectively, the source neutrons which are prompt and those which
are released from the delayed-neutron
represented by the concentrations:
n.fft)
'
concentration
precursors.
of delayed-neutron
precursors
The quantities pih and gtk in (9.84) denote the resonance-escape and fast
nonleakage probabilities to thermal in the usual way [(4.260) and (6.79)].
Finally, note that since the reactor described by (9.84) is not critical the
multiplication constant will be different from unity and may be obtained
from the relation
k =
(f
with B*
(i)2for
a sPhere- etc"
<9-87)
570
ANALYSIS
REACTOR
Thus
- F(r) *(Q
(Dfl +
- i (0
so)4>(0
we assume
- F(r)
C,(r,0
[CHAP. 9
- P)2fP*g*4>(.l) +
(1
(9.88)
C(0
V X.^i(0
p,hff,h
i= i
<
(9.89)
JV
1, 2
be
Gi(t) =
- ff)k -
-[(1
and
= yfr
(9.90)
thus
-W(0
1]*(Q
fc
^(0
(9-91)
t=i
t =
(9.92)
0(0 =
y-i
N'
and
^-e-'-'
G.(0 =
y-i
IV*
(9.93)
is
The number
N'
We attempt
differential
JV
1, 2
given by (9.35).
solution of Eqs. (9.91) and (9.92) in the form
is
Z,h
where
-XA(<) + ft0(t)
(9.92) yields
e"'' are
nfc*
linearly independent,
y-i
2,
1,
X,)e"><
y-i
I\,(ay +
JV'
AT'
TV
(9.94)
r<y(y
requiring
X.) =
vfr
(9.95)
-1]
- La,
-k
- 0)k
j=l,2,
...
,N'
(9.96)
[(1
0,(0
Thus, Eqs. (9.93) are acceptable solutions of (9.91) and (9.92) provided
SEC. 9.3]
REACTOR
KINETICS
571
that the quantities a, are chosen to be the solutions (N' in number) of the
following characteristic equation for a:
[(1
This equation
- fi)k - - La
We have
/tha
i-l
Ztllar
0 =
9.6
o>
previ
a +
X,
Equation (9.98) is
ft-
a con-
if
the indicated
-X2
X3
functions
are to
a=-\i
X,
ltb
satisfy the relations (9.91) and (9.92), then the a, to be used are the
solutions to Eq. (9.98).
Thus given the neutron lifetime
and the
and j8,, (9.98) yields the set a,.
delayed-neutron properties
Figure 9.6
The reactivity
is
is
is
shows a sketch of the function F(a), given by the right-hand side of (9.98).
The situation under discussion here
one where, over the period of
time in which Eqs. (9.93) apply [and therefore (9.98)], the multiplication
is
is
it
is,
Fig.
(9.97)
is conveniently
-k
1]
ai
doosec"
REACTOK ANALYSIS
572
It
[CHAP. 9
is easy to verify that the function F(a) has the following properties:
F(0)
lim F(a)
= 0
=
a*
-* o
lim
lim
e-0
F(-\i
F(
()
t
Hh
= +oo
= +
(9.99)
00
From these facts it may be seen that for a given w (positive or negative)
Here N is the number
there are (N + 1) solutions for a of Eq. (9.98).
If we denote the roots ay in the order shown
of delayed precursor types.
in Fig. 9.6 we find that <*i is positive if co is positive; if w is negative, <*i is
also negative, but it is also, then, the smallest root in absolute magnitude ;
i.e.,
< || <
Evidently in this
for
|a*fi|
w < 0
(9.100)
.
N +
1) are
all negative.
In
oi(0) =
G.(0) =
(9.101)
Giu
(t)
y-i
with
JV
AT+ l
ipj
Tij(aj +
X.)
<pfii
(9.102)
'
'
t =
1, 2,
1, 2,
N
N, N +
is,
(t)~e'*t
Gi(0~e
1)
(9.103)
sec. 9.3]
KINETICS
REACTOR
in a Taylor
for small
~F(0) + F'(0)a, +
It
573
a>,
(9.104)
oi =
(9.105)
1= 1
We find, then, that after sufficiently long time the effective reactor period
would be
i-l
+ Pi
(l*
(1
(l*
fii
)+"
+'*(i*+)-,*+2
(9-io?)
is,
(1
that
reactor period;
is,
is
(9.108)
c.
c*i
is
Time Dependence
It
picture of the time-dependent behavior of the reactor.
possible, however, to display the essential features of the system by intro
simplified model based on the concept of single group of delayed
ducing
a
is
physical
REACTOR ANALYSIS
574
[chap. 9
by setting
choose
/3
for
U236 (see
= 0.00755
(9.109)
/3
i- it
<9i"
and
X,
where
/3,
i= 1
delay groups.
The appropriate
rin^
X)
ri(i
<pi
IV"'
IV"'
(9.111)
(9.112)
drx)
T2 = Go
=
r2(<*2 +
Ti
*>,/3
(9.113)
<ptP
now a
^)-0
i,h("
^V
4X.
'th(w
li
IM
J
[l
i{Ml
1)|
is
where m
1)
2U" /,x)
co(l
(g
1)
The roots c*i and a2 are readily computed from (9.112), which
quadratic equation in a; these are
is
If
G(<)
X)
"
and
= Vie"1'
<K0
results
always limited to systems in which the separation of time and
space functions, as in Eqs. (9.88), may be assumed.]
For the special case
wherein the reactor was at steady state with flux <o prior to =
we
may obtain
relation between <o and Go by noting the balance for pre
cursors in the steady-state condition as given by (9.80).
In the case of
a
0,
is
relations
REACTOR KINETICS
SEC. 9.3]
575
yields
0o
(9.115)
a2(ai +
X(a2
\)o
r-
fi
ai)
= i(a2
X(ai
\)o
r-
(9.116)
a2)
(9.117)
is
/3.
"
Ul "
reactor period calculated for the system in which the delayed neutrons
were neglected, (9.23),
(l^)
\L
7-^ co G(t)
(9.118)
W(D~CW
(9.119)
coy
()
co
/3,
W(0
When
delay-group
/3,
is
is
CO
is is
is
is
as prompt critical.
= 0.00755
co
= 0.001
1^
involved
/3
/3,
= 0.076 sec"1
= 0.001 sec
ANALYSIS
REACTOR
576
[CHAP. 9
at
tii
be approximated
(9.120)
= 0.0117 sec-1
<*i
= 6.55 sec-1
<p2 = 0. 132#o
ar2
ip\ = 1.132(^o
= 1.132e0
to
0117'
(t)
is
0.132e-8-65'
(9.121)
Corresponds to stable
1.16
reactor period
PPT
1.12
X
3.0
1.08
132eo.on7c
1.041
Using:
dh = 0.001 sec
/8=0.00755
X = 0.076 sec'1 u= 0.001
1.00,
0.4
0.8
1.2
1.0,
20
_i
i
2.0
1.6
40
Time, sec
i i i i
2.4
i_
2.8
Time (, sec
Fig.
9.7
curve in figure),
1.132e0
0117<
for
t>
0.5 sec
The insert in the figure shows the long-term behavior of the flux.
sec. 9.4]
KINETICS
REACTOR
577
Another approximation
of the delayed
d. Infinite Delay Time.
neutrons, which is sometimes used, is that of the "infinite delay time."
In this approximation it is assumed that the delayed neutrons, which
appear after the time when the reactivity was introduced, appear at a
constant rate taken to be the rate at which they appeared in the reactor
t = 0.
The differential equation for the time-dependent
behavior of the flux based on the one delay-group model and infinite
delay time is obtained from (9.118) by setting the precursor term equal
to its value prior to t 0. Since the present approach to the delayed-
prior to time
W(0
(FL)*(0+t^L- w)
\1
(<)
= >o
a
0) p
exp
Wi - )
03
(9-122)
CO
/3
p~
(9.123)
This model would have some application in the early stages of the flux
rise after I = 0, and before the concentration of the delayed-neutron pre
cursors has been built up significantly over its steady-state value.
For
example, Eq. (9.122) is a good approximation to (9.121) when the tran
sient effects, represented by the second term, are significantly larger than
those due to the stable reactor period, i.e., when
<pt
vHorjf
[cf.
ipse0"'
(9.124)
Eq. (9.121)]
for ait
<. 1
9.4
Dependence
578
ANALYSIS
REACTOR
[CHAP.
For the purposes of the present study these effects are described by
means of a single parameter,
the temperature
coefficient
is due to the coupling between the flux level and the power generation.
It is clear that if for some reason there appears an increase in the flux
level this will result in a proportionate increase in the fission rate.
There
fore, an increase in flux gives rise to an increase in the power generation,
and this in turn disturbs the energy balance of the system.
As a conse
adjustment
there
must
an
in
be
the temperature level of the
quence
Since the nuclear characteristics of the reactor are temperature
Thus,
sensitive, a change in the flux level leads to a change in reactivity.
a realistic treatment of reactor kinetics must necessarily include this
coupling between the flux and the energy content of the system.
Generated for wjivans (University of Florida) on 2015-09-22 17:22 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
reactor.
The temperature-dependent
kinetics of stationary-fuel reactors is
treated on the basis of the one-velocity model.
The appropriate rela
tions for the time behavior of the thermal flux are given by (9.84) and
jt
H(T,t)
2^(r,0
- (P(r,0
where
(9.125)
(9.126)
conduction)
It
is convenient
temperature
T by introducing
is assumed:
H(t,t) = c[T(i,l)
1
ro(r,0]
- c0(r,i)
(9.127)
sec. 9.4]
where
KINETICS
REACTOR
c =
579
m some
- F(r) 0(0
- F(r)
e(r,<)
<P(r,0
CP(<)
In addi
(a)
(J-lzo>
(b)
The substitution of these relations, along with (9.88), into (9.84), (9.85),
and (9.125), yields
W(<)
= [(1
/3)fc(t)
+
C,(0
p]4>(t)
STW.
I/'
-X,C,(0 + wfiiZftffi
c0(<) = -<P(t)
+ 2,<K0
(0)
(9.129)
(&)
(c)
The first two equations are essentially the set (9.89), the only difference
being that we have included here the time dependence of the change in
multiplication &k(t). This dependence has been introduced since in the
present treatment the reactivity will be a temperature-sensitive function,
and the temperature in turn is coupled to the time-dependent neutron
relations.
= 2,0^,0)
If
2/4.(0)
(9.130)
<^(r,0),
(9.131)
which states that the steady-state value of the cooling (heat removal)
rate is proportional to the flux level.
It is convenient to introduce a
ANALYSIS
REACTOR
580
[CHAP. 9
in same
way
new function P(t) which is related to <( in the
theway
same
that (P(0) is
that (P(0)
is
The function P(t) will describe, then, the time behavior
related to <(0).
when the system is disturbed from its
equilibrium.
Therefore let
P(t) = 62/0(0
(9.132)
Note that this function has the units of energy per unit time per unit
volume [see Eqs. (9.88) and (9.128)].
The transients in the reactor power production will be of special inter
est in the present study; therefore we write the kinetic equations (9.129)
in terms of the function P(t).
[(1
- f)]P(t)
f})5k{t)
-X.C,(0
C\(0
P{0 -
C0(O =
X WO
(a)
(9.133)*
+|ftP(0
(&)
<K0
(c)
(and
is
5Ar0)
(9.135)
8fc(0
Sk0
y[T(t)
1.
An approximate expression
the product y6ko, which
is
if
reasonable
is
assumption
6T/T
be
obtained
from
this
equation
can
for 4fc(0
neglected
small in comparison to unity,
is
a
r,(0H
0
constant
- y[T(t) -
step change in
assumed to be
if
Experience
5fc(0
For
(9.134)
(1
Sk0
with
dependence
fc(0 =
k{T) = k[T(t)]
1
1
jfc
r,(0] =
Sk0
- 70(0
(9.136)
is
It evident
given in (9.127).
where we have used the definition of
that
coefficient
of
the reactor
above
relation
the
temperature
from the
dependent on the sign of the constant y. The temperature coefficient
is
In this
at
case
if
negative.
the temperature coefficient
>
step change in multiplication, Sko >
given
0,
(9.137)
+ 8k(T))*
0,
~~
8k{T)Y
dT
Sk0)y
[1
k2
dT
(1
dk
[1
du
system
is 7
Thus
if
of reactivity
using (9.136),
is
is
2a
U>(t)
then
REACTOR KINETICS
sec. 9.4]
581
is,
the reactivity is initially positive and the system will "accelerate"; that
the flux level will rise. However, as the flux rises, the power output
and temperature will also rise. But because of the negative temperature
coefficient [see (9.136)], the reactivity will decrease from its initial value
a
Clearly,
and thereby retard the acceleration so as to stabilize the system.
negative temperature coefficient implies an inherently stable chainreacting system.
>
0
0.
=
are those which were produced at the flux level existing before
This approximation has some practical value for analyzing situations
is
X<C,-(0)
& P(0)
(9.138)
c0(t)
=
=
Sk(l) =
- W;(0 - P]P(t)
P(t)
- 70(0
Id
[1
l*Ht)
is
This expression
used as
first estimate for the precursor concentra
tions in (9.133a); then, the set (9.133) reduces to
+ Bk(t)]pP(P)
(a)
(6)
<P(l)
Sk0
(9.139)*
(c)
This
is
it
In order to
there are three unknown functions P(t), 0(<), and <P(t).
obtain a complete solution,
necessary to specify one of these func
tions.
It
most convenient to give the cooling rate (P(0 (the power
With this information and
removed per unit volume from the reactor).
the following set of initial conditions,
is
is
<P(0)
(a)
(b)
T(0)
= To
dk
(c)
is
(9.140)*
dk
dk
. ..
it
P(0)
0(0)
0(0)
complete:
[CHAP. 9
ANALYSIS
REACTOR
582
Neutrons.
As a first
will
be those which
c0(<)
P(t)
of the
(9.143)
<Po
It
P(0
(9.144)
<?oewW
and look for a function w(i) such that Eqs. (9.142) and (9.143) are satis
fied along with the requirement that w(0) = 0 [cf. assumption (1)].
The
substitution of (9.144) into (9.142) and (9.143) yields the nonlinear set
c<5)(0
= (P0[e"'("
w(t) =
[fc0
- 7(0Jr
(9.145)
= 0
(9.147)
1]
(9. 146)
6(0)
be presented here,
w(t) =
- f 0)
- ^
[e<
1]
\ W)]*
[e<
w(t)}
+ C
sec. 9.4]
REACTOR
KINETICS
583
7(Po
Thus
WW
/5ko\*
2\llh J
lihc
=
khC
[e-(
w(t)
_ ij
i^y
(9.148)
Fig.
9.8
These functions represent the solutions for the case y > 0, which, as
previously noted, implies a negative temperature coefficient for the sys
tem.
This case describes a stable system that oscillates about the
steady-state condition existing prior to t = 0. The case y < 0 yields
a divergent
behavior
monotonically.
0(<) increasing
function &(t) may be computed from the condition 0(0 = 0. This con
dition requires that w(t) = 0 [see (9.145)]; thus we must determine the
roots of this equation.
The slope of the function w(t) at its zeros is
obtained directly from (9.148); these are
w
It
the maximum temperature values occur, then the magnitude of the tem
perature rise is obtained from (9.146):
em.
= 0(o)
(9.149)
REACTOR
584
ANALYSIS
[CHAP.
This result is generally valid, its only limitations are that the power
be constant and that the delayed-neutron
contributions be
removal
negligible.
In this
P(i) ~
+ w(0]
($>o[l
the
(9.150)
~ **E
w(t)
~^
[Sk0
- 70(0]
(9.151)
...
WW
sW(s)
[tt.
73(s)]
(9.1538)
U(s* +
7(Po/i,hC)
3(s) =
cUs(s*
Tt^o/^c)
(9-153)
0.
is,
REACTOR KINETICS
sec. 9.4]
585
Fio.
9.9
At
= 0.
t = 0
(2)
<P(t)
(3)
Sk(t)
<P
5fc
-P{0)
7(0
For
(9.155)
The appropriate
P(t)
[ifc(0
6(0
~[P(t)
- ^+
(9.156)
0]
(9.157)
<Po]
(P0
0(0)
0(0)
= 0
(9.158)
REACTOR ANALYSIS
586
[CHAP. 9
0(0
- 70(0
^[e<
r
1
[8k0
-P]+
fl#>->
(9.159)
(9.160)
1]
~ jtth
[6k0
70(<)
- 0w(fl]
0(0~pw(0
where we have introduced
(9.162)
the parameter
V
-y
(9.163)
r
tth
*(0
F w(0
tth
= 0
(9.164)
w(0) =
Me
(9.165)
temperature-coefficient
it
is a measure of the
(3.
(9.161)
York,
1940.
sec. 9.4]
KINETICS
REACTOR
587
a disturbance.
where
(|
sin
py
2Z,h
cos
- v!
(9.167)
for
case
Since the
0.
>
~ e~*'
yf/2
0)
sin
(9.166)
e~xl
= {5ka\
w(0
12r
"\
= 400
yr,
2.3
0=0. 00755
(P0/c
0.8
1.0
t,
Time
1.2
pc = 02/4/,7
I
0.6
.001
0.4
. r
= 0,
i*
50^-^
&
0.2
1.4
1.6
1.8
2.0
sec
Fig. 9.10 Temperature for constant power removal and one of group delayed neu
trons with infinite delay time.
of
0,
>
may take on all values
parameter
may be less than, equal to,
or greater than zero.
For each of these cases the solutions (9.166) have
a particular form.
The general results are well demonstrated by means
5fc0
>
lo-yc
Consider
This corresponds to
(1)
= 0.001
=
Vc
(9.168)
^2
discussion
= 0.00755
= o.ooi sec
u0
numerical example.
Figures 9.10 and 9.11 show the temperature
difference
and the puwer function w(t) for various values of
obtained
for the case
10
> 0.
REACTOR ANALYSIS
588
where
pc
\p
[chap. 9
= 0, i.e.,
(9.169)
The solutions for this case have the form of damped oscillations.
Thus the initial disturbance due to the introduction of Sko causes a
0.14,
-0.02,
0.6
0.8
1.0
1.2
1.8
2.0
Time (, sec
Fig. 9.11 Power function for constant power removal and one of group delayed
neutrons with infinite delay time.
rapid change in both the temperature and power output of the reactor.
The system stabilizes rapidly, however, and the oscillations decay to
It is obvious from (9.166)
negligible magnitudes within a few cycles.
stabilizes
at
eventually
that the reactor power,
(P0, as is to be expected,
and the temperature difference 0 at Sk0/y; these arguments follow from
the fact that
lim w(t) = 0
(-
lim 0() =
Sk0
(9.170)
stabilizes at a
is a consequence of the negative temperature coefficient of the system.
Note that the introduction of positive reactivity causes the power pro
duction to increase, this causes a rise in the temperature since the rate
REACTOR KINETICS
sec. 9.4]
589
(2)
In
V<
0)
0 [1
7
- tr{xt +
(9.171)
1)]
The limiting values of these functions are seen to be the same as those
This case corresponds to a system with "critical
given in (9.170).
damping." As the value of p decreases toward pc, the magnitude of the
oscillations in both temperature and power output diminishes until finally
at p = pc the oscillations disappear altogether (see Figs. 9.10 and 9.11).
It is helpful in understanding this behavior if one imagines the reduction
in the parameter p to be due to an increase in the heat capacity of the
large to accommodate
"overshooting."
P < Pc
(3)
^2
written
in the form
+
iS
with
S1
= x- ~
(9.172)
REACTOR ANALYSIS
590
case are
w(t)
e(<)
[CHAP.
S
=
r
1
sinh
S<
+ cosh
\i
(9ll73)
As in the preceding case, these solutions are nonoscillatory and the sys
This corresponds to very large values of the heat
tem is "overdamped."
In these cases the temperature and power are relatively loosely
capacity.
coupled functions and respond very slowly to reactivity changes (compare
curves labeled p = 50 to those labeled p = 142.3 in Figs. 9.10 and 9.11).
p = 0
(4)
This
!*! (i
,(()
0(0
= 0
e-*/th)
(9.174)
the result given above reaches a finite value it is clear that this elementary
model breaks down when applied to this case; however, note that for
short times after the addition of 5k0 (i.e., fit/h* <K 1) the power function
has the correct form.
ip(<)
In this
(j^jt
case
and
P(O~Poexp0^
(9.175)
sec. 9.5]
REACTOR
KINETICS
591
of the system.
In the present treatment the effect of the delayed neutrons and the
effect of temperature-flux coupling on the reactor kinetics are handled
separately along the general lines previously developed for the stationaryAlthough in general these two factors must be considered
fuel systems.
jointly, there are many practical situations wherein the two effects can be
The intent here is primarily to simplify the analysis so as
separated.
to focus attention on one feature at a time.
a. Low-power Kinetics with Delayed Neutrons.
The effect of delayed
neutrons on the kinetic behavior of circulating-fuel reactors is studied
with the aid of an analytical model which describes a low-power thermal
reactor.
is used to
denote systems wherein the power level is sufficiently low that small
variations in it produce negligible changes in the temperature of the
nuclear constituents of the reactor.
Thus the energy balance of the
system is not considered, and no coupling is allowed between the tempera
ture and the neutron flux.
For the present, then, we ignore entirely
temperature coefficients of reactivity and focus attention on the role of
The specific objective is to derive the inhour
the delayed neutrons.
1 The expression
flow"
is used to denote a one-dimensional flow field.
"slug
REACTOR ANALYSIS
592
[CHAP.
by
J. A.
Fleck.1
is
For the purposes of this analysis we assume that the reactor core is a
bare right circular cylinder of radius p0 and height h. The circulatingfuel circuit is a closed loop with the same cross-sectional area as the core.
The length of the circuit external to the core is h and the length of the
By definition, fissioning takes place
complete circuit (h + he) is A*.
only in the core region of the circuit. Although the assumption of a
uniform cross-sectional area for the entire circuit may introduce some
conceptual difficulties in establishing which portion of the circuit should
properly be termed the "core," it leads to no loss in the generality of the
results that are obtained since only the transit time 0e through the
It is clear that if a different
external loop is essential to the calculation.
cross-sectional
area
were
to
the external circuit so as
assigned
average
better to represent the piping and heat exchangers, etc., in the loop, this
would merely result in an average flow velocity through that portion
different from that through the core. This effect is easily accounted for
by adjusting the transit time
In this model we further assume that the hydrodynamics of the
system is adequately represented by the relations for slug flow and that
the neutron physics may be described by means of the one-velocity
Finally, we assume also that the fuel in the core is homogeneous
model.
and that the neutron macroscopic cross sections are time independent.
On the basis of these assumptions we can write the following pair of
differential equations to describe the circulating-fuel reactor:
-DV**(r,0 + S.0(r,O
- ~ *(r,0 + (1
jt
C,(r,l) + V
| CM
+ V
Ci(t,t)
^ C&fi
= W3.2,*(M)
=
-XA(r,<)
0) vZ/Pthgtl,(r,<)
Pt<7.h
X,C,(r,<)
V X,C,(r,0
(9.84)
0 < z < h
(9.176)
h<z(9.177)
(9.177)
sec. 9.5]
KINETICS
REACTOR
593
where V is the velocity of the circulating fuel and the coordinate z is the
distance around the circuit measured from the inlet to the core. All
the remaining
The differential
Sec.
addition of
term.
Note that Eqs. (9.176)
forms
of
and (9.177) are merely special
the usual continuity relation of
fluid mechanics.1
a
with
where
m R(P)
\pl
(II
\p
dp
dp
(9.84)
R(p)
<p(z,t)
R(P)
dt
S,(z,)
(9.178)
The substitution of
yields
+ R(p)
<r(z,t)
dz
(1
<p(z,t) \
+ ZaR(p)
- p)vX,plbgthR(p)
+ p*g,bR(p)
)
^
i
<p(z,t)
<p{z,t)
\iSi(z,t)
KSi(z,t
(9.179)
= 0
(9-180)
<p{z,t)
B^(2,oJ
= vXfpth/Xa,
+ [fc.0.(l
+
L2B2<p
where
M2
B2 =
m2
we call
B\
+
If
(9.179) may be
- -
1]
R(p~o)
P)
hft<f{z,0
fcM
*(*,<)
A.Si(z,0
= D/2, and
along with l0 = l/V2a,
further reduced to the form
(9.181)
^jY
B*R{p)
Jp [p Jp R{p)] +
~P
[ip
(9.182)
'See, for example, L. Prandtl and (). G. Tietjens, "Fundamentals of HydroAeromechanics," pp. 100-104, McGraw-Hill Book Company, Inc., New York, 1934;
S. Goldstein, "Modern Developments in Fluid Mechanics," vol.
chap. Ill, Oxford
University Press, New York, 1938.
I,
C,(r,0 = R(p)
<p{z,t)
_D
a reduction
ANALYSIS
REACTOR
594
divide through by
l*! *(s,f)
(1
ZAB2),
*>(*,<)
[CHAP.
mV(2,o]
-/*)*-
+ [(i
fl*(,o
Sfat)
Iat
+ V%- Si(z,t)
oz
I- Si(z,t)
+ V
|- S.(z,0
= v2fPi<p(z,t)
- \iSi(z,t)
-X,S,(z,0
0<z(9.184)
(9.184)
h<z(9.185)
(9.185)
<
<
(a)
dff
The
v2/^(z,<).
x.(0,s)
<
*(y,8)e"+x->"'
P(z,t) ^
(b)
of P(z,t) and
< h*
<
e-<"+*.'"-
(9.186)
(9.187)
is
the transform
where *(z,s)
solutions to these equations are
x,(z,8) =
Xi(z,s)
=0
<
*(*,)
*(*>*)
f1^)
(S
(*,)
y^)
jz *(,)
+
+
(*,)
Si(z,t)e dt
jQ"
{Si(z,t)\ =
transform.
where
<
< h*
(9.189)
(9.190)
= x<(fc,s)e-(*+Xi)(''F-*)
(9.188)
X.(2,s)
is
=
= fc*.
must equal Eq. (9.189) evaluated at
(9.188) evaluated at
The resulting expressions for the constants when used in (9.188) and
(9.189) yield the solutions
z
REACTOR
^SSSi
for
dy
*(^,*)e('+x-)"
<
< /i*
(9.191)
(9.192)
(9.193)
t>b
<
(9.194)
<
when
b)
=/()-**
0
\W)\
Fb(t) =
= F(t
where
is
obtained by the
These are easily determined by
the appropriate
(9.191)
71=
in the denominator
In this way
it
0.
less than
may be shown that
is
is
S.(z,0
- y)e-^dy
71=0
Vy,
- ntf* for
<z(9.195)
i/)e^."
<
^_
e~M*
&(*,) = v2,ft
forO
so
(9.195)
fy
< h*
(9.196)
0,
m=
^)Z(Z)
4>m(
with
Zm(z)
{
V
>
,(!,*)3,0
is
is
<
The symbol
where
<
for
*{Vy,)eP***' dy
Z^SS^
(Vy,s)e<'+^ dy
&
X.(z,)
595
J*
KINETICS
= /3,-e-M-WF
^"
x,(z,a)
J*/V
SEC. 9.5]
lsin^z
= xm/h
(9.197)
REACTOR ANALYSIS
596
[CHAP. 9
cMi(t)
c,fc)
[(1
P)&k
m 1
BO
- tf)
Ox*
<
e-V*
Zy() dz
- Vy)er^ dy\
Vdi,
n&* + Vdi,y)Zm(z
<M - n&* - y)Zm(z
- Vy)er^ dy\ (9.198)
J'/v+a.
(9.198)
If
= m
(9.199)
MUV)
^ m
|o*
Zy(z) Zm(z) dz ^
c,-
|Q*
Z,(z) Zn(z) dz =
Vy)
dz
(9.200)
jo'
/s]J
y^1^6
^" dy
c-^**-
('
m(t
nt>*
- ndi, -
Mjm((/)e"x."
dj/
Njm(y)e-*-" dy
(9.201)
J#
n-0
i/)
- fi)sk -
+ [(i
y)
mj)
(m2
,)
fc
hMt)
(gf
the subscripts
on
is
Note that,
M{S) =
{|o'
L(y)e->"* rfy
f*+'
AT
>*
<
<
the integrations
(9.203)
ir
- sin irf
(9.202)
tffo)e-**(to]}
;
*(
irf
cos
M, and
f)
?)"(7-r)
(f
US)
(l
The kernels
yield
L,
n-0
*(t
2/)
0,X,
fc
- P)5k -
= *(<)[(!
<K0
0]
mi
597
/*)
(note that
KINETICS
REACTOR
2/)
sec. 9.5]
;*'('-$)
+
??) ('"t)
+
<
<
\\
\\
0.4
0.6
\\
0.2
0.4
0.6
0.8
1.0
and
(9.206)
- MiS) NiS)
US)
0.2
kernels
\\
equation
Thus
(9.205)
through
(9.203)
in
(9.202) to
(9.205) may be used
obtain an expression for the inhour
calculation.
f.
approximation
results
The
\\\\
is
7C
'
sin ir
iV(f)
(9.204)
REACTOR ANALYSIS
598
[CHAP.
= oeat
(9.207)
If
we use the set (9.206), then it may be shown, after some algebra, that
(9.202) reduces to the form
u.
- - m - * + *Z * [. <i
ZP-^]
(1
7. = Xi
where
"
**
l + Ua
(9.209)
be
+ a
TTUa
fta
h Tt
\^
L1
4+
X,(l
**>
o>,
- e-*'*Q(l - e-M)1
~
*7(1
"
e-M*)
(9-210)
*(*.0
-(HSi)
7,.(f-t^l)
11
- e-y'"r
r^*^""-*'
1])
0 < z < h
~ l>ry'"V\
h<z(9.211)
(9.211)
of the first of these over the interval {0,h) and the inte
gration of the second over {h,h*) give the total number of precursors in
The fraction a, of precursors of type i in
core and loop, respectively.
the core is computed from
The integration
-'-868
then
a.() =
* i-***^
(
^-(1
- U)
WW
0-213)
KINETICS
REACTOR
SEC. 9.5]
599
CI
and when
X,t?*
<3C 1,
a<(0)
p-M,) (i
e-W)
(9.215)
Thus for transit times appreciably less than the precursor mean life (see
Table 9.1), the fraction in the core is given approximately by the fraction
0.4
FlO. 9.13
0.6 0.81
8 10
20
40
60 80 100
I?*so
(9.216)
It
co
IthOC
=
1
hwa
used
TTLraZ^[yi-XMa)]
to reduce the
(9-217)
This result has several interesting features. Consider first the reactivity
For
required to maintain the circulating-fuel system at steady state.
this condition a > 0, and (9.217) takes the form
codrc
(9.218)
600
ANALYSIS
REACTOR
[chap. 9
where a,(0) may be obtained from (9.214) or Fig. 9.13. Note that, when
o
# and de
This agrees with the
(no circulation), u > 0 by (9.218).
result for the stationary-fuel reactor (9.98).
Also, in the case of infinite
= f = constant and
circulation velocity V, d and 0e > 0. If we let
0, (9.218) yields
lim
Thus
o,olrc
= 0
(^
(9.219)
w toeiro =
^*h
obtain
[7,(1 +
Uha)ai(0)
\&i(a)
l*ay,]
(9.220)
It
is
If
eM
h^j
\idt
+ eM, _
-^_ X,i?*
X,t?
hi
where
~ o,(0)
l
a,()
(l
is
is
reactor
we
o>,,
a>
h
,
is
d/d*) both this quantity and /ttXi are much less than
^t?*X-(l
unity. Thus to zeroth-order approximation, the expression within the
bracket of the sum (9.220)
The result (9.220) then reduces to
aOi(O).
h(
(9.221)*
REACTOR KINETICS
SEC. 9.5]
601
>
is
it
is
/3,
multiplication constant of
Consider the multiplication fcoiro
is
Finally,
circu
*-
<w
a
The effect
is
is
it
is
of flux-temperature
J.
is
ORNL CF
is
An alternative derivation
given by W. K. Ergen, "The Inhour Formula for a
Circulating-fuel Nuclear Reactor with Slug Flow," Oak Ridge National Laboratory,
53-12-108,
J.
of
- - r-i? - >
co
follows that
and
it
(9.222)
REACTOR ANALYSIS
602
[chap. 9
for the flux in a Fourier series the resulting expressions for the
time-dependent Fourier coefficients are in fact the relations used by the
authors mentioned above.
Thus the present method offers a somewhat
more general approach and has the added advantage in that it contains
solution
and
conditions.
V,
All fluid particles and heat are transported around the circuit at
Po-
uniform velocity
is
a
is,
a
it
and
assumed that thermal equilibrium and
changes occur instantaneously with variations in temperature.
assumed that no lateral exchange of heat occurs between
Moreover,
taken to be a line from the
adjacent fluid streamlines; i.e., the reactor
is
is
it
density
differential
be1
H(T,t) +
--
kZ/Pu.^M
*(r,f)
2o<Kr,0
-DV2<Kr,0
jz
to
at
= S^(r,<)
H(r,t)
(9.224)
(9.225)
mV(z,<)
1M2,<)
(9.226)*
the definitions
H(T,t)
vol.
H{z,t) = c[T(z,t)
is
to be represented by
is
(k
<p(z,t)
is
7'(0)]
- cV(z,t)
see
(9.227)
sec. 9.5]
KINETICS
REACTOR
603
where To(0) denotes the temperature at the inlet to the core and c is the
heat capacity of the fuel.
Equation (9.225) may now be written
fat
0(z,t)
|az
0(z,O + V
' <p(z,t)
0<z(9.228)
(9.228)
a suitable
- yQ(z,t)
(9.229)
@(z,t)
ev
where
&
is the
Vt) +
fz/r
Jo
0(2,0
and
ip(z
gV
c
Jo
Vs, I
fz/V
<p(z
Vs,t-
<p(z
Jo
s) ds
>Vt
0 <
<
<Vt
0 <
< d
- s)ds
- Vs, -
0(z) = To(z)
(9.230)*
t>d
s) ds
(9.231)*
core, and
T0(0)
(9.232)
is
<
0.
C-)
it
is
&,
if
The results (9.230) and (9.231) are also easily deduced by physical
Note that
arguments.
<
i.e., before all the original fuel particles
in the core at =
are swept out, the temperature (rise) at
and
is
is
0;
>
the core to point z. Clearly, when
all the original fuel in the core
has been swept out.
In this case, the initial condition to be used with
the entrance condition 0(0,0 =
thus the temperature rise
(9.228)
experienced by
slug which has reached at time
due entirely to the
See, for example, F. B. Hildebrand,
378,
2,
= 0(z
0(z,i)
be shown
REACTOR ANALYSIS
604
[CHAP. 9
Vs (where it was
and
the
observation
z
point
present
s)
(its
location).
The solutions (9.230) and (9.231) may now be applied to (9.226).
For this purpose we first express the flux by means of a Fourier series.
heat
located at time
If
written
0(2,<) =
'
V f
3*
^,(1
s)z(z
t>d
vs) ds
(9.233)
71= 1
The substitution of this expression into (9.229) and thence into (9.226),
followed by multiplication by Z,(z) and integration over the interval (0,h) ,
yields [cf. Eq. (9.198)]
_ l&l V
Generated for wjivans (University of Florida) on 2015-09-22 17:22 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
cCj
L4
m,n 1
^m(<)
z,(z)
Jo
Zm{z) dz
[
Jo
'
n(t
(9.234)
s) Zn(z
i J[k
i
Cj
- Vs) ds
(9.199).
z) dy
If
we
(9.235)
l*fc(0
*i(t)
fe
- tf)
(m2
-^ V
fc
<M0
m,n = 1
>
&.
*(t)
fe
m-
(m2
*.(
- s)Kjmn(Vs)
ds
(9.236)
be shown
that when
relation is
**<mo =
Jo
/;)
skQ
m,n 1
^W
Z,(z) Z.()
rf2
(<
Z(2
Vs)
da
as indicated in (9.230).
REACTOR KINETICS
sec. 9.5]
G05
If
*f(0
0*2
m-
m,2)
*o]
UM)
[|l
we invert the order of the s and z integrations and note that @o(z) = 0
for z < 0, the above equation reduces to
BO
-^
- s)Kjmn{Vs)
(<
*m)
ds
(9.237)
m,R =
This result
is
i>
<
<
<
(a)
(9.238)
K(Vs)
ds
>
~(\
=
w) dz
for
Z\(z) Z,{z
appropriate
for
K(w)
4>(t
Vt) Z\(z) dz
s)
Sk0
J"
where
K(Vs)
-^
J,h<W) =
Jo
90(z
ds]
Jvt
s)
j*
--
[sfco
4>(t)
(b)
cos
(9.239)
flux.
(9.240)
Km(Vs)
t?
<
s)
<
>
P(t
ds
(a)
dsj
KAVs)
P{t)
[sk0
Jo
LP{t)
j*
P(t
--
~ Km(w)
"1
this relation
P(t), the result
is is
K(w)
we introduce
Thus
s)
in (9.206).
If
U4>(t) =
pair of equations
0.
is
tions
(b)
(9.241)
G06
ANALYSIS
REACTOR
[CHAP. 9
by introducing an
l
-
+ y
Jo
P(t-
s)
K(Vs)
- P{t -
K{0)
K(h) +
P(t
J*
-c
6(t) =
0)
that
|V(0
where
fa
8(t) m
ds
(9.242)
by (9.240).
s)
effective temperature
K(Vs)
Note
ds
(9.243)
(9.244)
ds]
P{t
s)
-c
jT"
\p(t)
6(t)
K by
we approximate
If
<
<
dz
(j^J
P(*
Vt) sin2
--ejo
dsj
90(z
^1
P(i)
we have
J*
summarizing,
|^Sfc
(9.245)*
and
(o)
(9.246)*
f*
i>m
*(0=-|P(0-5
jo f*
8(t)
1
a/rt
-yP(t)
1
/>-s)dsJ
(6)
is
0)
is
is
It of interest
Note that the function K* has been used throughout.
to mention that the set (9.246) constitute the relations used by Ergen,
Weinberg, and Welton in their studies of the kinetics of circulating-fuel
reactors.
These studies place emphasis on the stability aspects of the
nonlinear set (9.246) and include the extension of the present theory
to account for the effects of alternate fuel paths through the core with
different transit times as well as the effects of delayed neutrons.
Since Eqs. (9.246) apply only for values of
greater than the core
transit time, the set may be conveniently called the "asymptotic" solu
In the general study of
tions for the temperature-dependent
kinetics.
this problem one should properly use first Eq. (9.245), which describes
the reactor power transients during the time interval before all the original
=
fuel in the core (at
then
The solution for >
swept out.
=
constructed from (9.246), using the results from (9.245) for
as
initial conditions.
In order to obtain
solution for the power transients during the time
a
UP(t)
8{t)
temperature
This relation and the definition (9.242) may be used to replace Eq.
(9.2416). The set (9.241) may then be written in terms of the effective
KINETICS
REACTOR
SEC. 9.5]
607
V2*0(p,z)
For the
case of a sinusoidal
= 0
(9.247)
axial flux
Z(z) =
The steady-state
= R(p) Z(z)
0(p,z)
4>o sin
(9.248)
^0o(z)=^sin^
If
0O(2)
cos
(9.250)
-j
ds\
|)
- - s sin - - g- sin
j
*)
P(t)
(^1
UP{t)
|fe
<t (9.251)
(9.251)
is
is
This equation
to be solved simultaneously with the pair (9.246).
The expressions (9.251) and (9.246) form
nonlinear set of integrodifferential equations in P(<) and 6(t). An exact solution to this set
set (9.246) in
\ (l -j
Jo['
In P0
-___
V/
w{t).
8ln
With this
0<t (9.253)
ds
-70(0
(9.253)
(a)
<
Po
(9.252)
turn reduces to
,,A
w(t
f*
l*i*(0
-^
C
~]
sin
The
OWz
*.-
becomes
(9.251)
&
ltMt)
In P(t)
so that
s)
approximation
with P(0) = Po
w(t)\
^1
that w
and require
(t)
Pit) = P0[l
linearized equations.
(9.249)
yields
from
>
(9.254)
608
ANALYSIS
REACTOR
[CHAP.
u>(0)
-f2
tf(0) =
**!
tf(0) =
f2
where
- *C
(9-255)
(9.256)
In obtaining
W) - ^ w(t)
/(0 =
with
= f2/(<)
for
0 <
<
(9.257)
_^__sm___8in_j
(9.258)
- w(t)]
=0
for
>
t>
[w(t
tf)
w(t) + f2u)() +
(9.259)
The initial conditions for this equation are obtained from the solution
= d.
to Eq. (9.257) evaluated at
For this purpose we require the
function w(d) and the derivatives w(d) and w(d) as computed from
(9.257).
We will not concern ourselves further with the detailed solution to
this problem but will instead examine briefly some of the physical
features of the system described by the pair of equations (9.257) and
(9.259).
Consider first the form of the power oscillations.
studied by introducing the elementary solution
w(t) = e"*'
be
(9.260)
ai +
thus
f2a*
(e-**
t?;
1)
and
+
^
of the parameters
(9.261)
= x* +
iy-k
sec. 9.5]
REACTOR KINETICS
609
xi
- 3x*yl
-Vl
+ f2x* +
-j
f2
+ 3xly* + r-V*
e~*x*
and y*
x+
1)
= 0
sin dy* = 0
(9.262)
It
In these
is .).
pure
cases, a*
an undamped steady sinusoidal
is
a
(n = 1, 2, . .
imaginary, and the solution to (9.259)
occur when
oscillation.
is
Jo
~ - ["y P(t
c
0(z,<)
s)
Vs) ds
(9.263)
w(t-
sin
J ds]
+
Jo
f^
[i
s)
[*
= (<) =
0(M)
[\
ature
(9.264)
it
is
is
it
source of power.
little danger
Not only does indicate that there
to the reactor through excessive temperature variations, but since the
as
In
REACTOR
610
ANALYSIS
[chap. 9
(9.265)
which states that if some positive reactivity is added to the system, then
because of the negative temperature coefficient of the fuel, the reactor
must operate at some higher temperature
9.6
In the time-dependent
been taken
is,
is
As the fuel
is
realized, then,
The primary function of these poisons
during the initial stages of the reactor fuel cycle.
to examine some of the problems
The purpose of the present section
associated with the estimation of the concentration of poisons in the
function of time for periods of operation which are compareactor as
is
is
in the reactor.
sec. 9.6]
REACTOR KINETICS
611
We will consider first the concentration of 1 136 and determine its value
Iodine 135 is a first-stage decay product of Te186,
as a function of time.
which is a direct fission product.
The tellurium which appears from
fission reactions decays by negative beta emission to form I136, which in
turn decays to Xe136, again by negative beta emission. The decay proc
ess is indicated below, along with the appropriate half -lives:
62Te136
<2 mm
53I136
-4-*
6.7 hr
54Xe136
(9.266)
The yield of Te135 is 0.056 atom per fission. We note from the above
relation that the decay rate of the Te136 is relatively fast; so a reasonable
approximation to this reaction would be to assume that I136 appears as a
direct fission product of yield 0.056 and to ignore entirely the presence
This approximation has some merit since the half-life
of the tellurium.
is short, relative to the time periods of interest here.
On the basis of this assumption we can write the following balance
equation for I(<), the I136 concentration:
of
Te136
1(0
~ -X,I(t)
- e?l(t)<Kt)
+ MiZ/(fl*(0
(9.267)
612
REACTOR
ANALYSIS
[chap. 9
where
m=e-f>)dt[lfo'Me)mJ"de
L(<) = X,
m]
+ <r(?(t)
0.268)*
(9.269)
Thus, given the time behavior of the flux, (9.268) yields an explicit rela
tion for the iodine concentration.
The balance relation for the Xe186 may now be given in terms of the
I11*
Xe135
concentration,
X,I(<)
+ MxS/O
then the
appropriate
X(t)
(9.270)
and fix are the decay constant, the thermalwhere the quantities X
cross
and
the fission yield, respectively, for xenon.
section,
absorption
0.003 and \z = 0.0754 hr"1.
Note that
The value of the yield is /**
the first two terms on the right side of this equation represent the reduc
of radioactive decay and
neutron capture and that the last two terms give the rate of production
from the decay of I135 [the parent nucleus, cf Eq. (9.266)] and as a direct
product of fission. The general solution to (9.270) is
tion
XW
with
in the
Xe135
e-l>)d
concentration
[uw
as a result
(9)\Jem")d"
M(t)
m X,
(t)
de
X(0)}
(9.271)*
(9.272)
47 hr
REACTOR KINETICS
sec. 9.6]
per fission.
methium
613
Pm(<)
and samarium
S(i)
\PPm(t)
nuclei,
pro-
are given by
Sm(<),
[Xp
o(t)Wm(t)
(t)
<7Sm(<)
(9.274)
0-275)
M.2,(<) *()
e-/(9)
"
MB)
dB
|>P
P.(0
/'
P(0)]
with
G(t) =
10
is
(9.277)
[X,P(9)
d0
S.(0)}
(9.278)*
The results (9.268), (9.271), and (9.278) are the general expressions
Detailed
for the time dependence of the principal fission products.
calculations for these functions can be carried out once the (t) and 2/(<)
is
are specified.
Some simple cases are now considered.
b. Steady-state
The concentrations of the
Poison Concentrations.
reactor which
operating at steady
poisons I136, Xe136, and Sm149 in
is,
is
At
state are easily obtained from the original differential equation.
stabilized to some value o- As
steady state we assume that the flux
the rate of fuel
simplification we also take 2/ to be a constant; that
is
lo
is
is
iJ%r
Al +
0
0-279)
Xo
Sm149
X'Io
Ax
Xe136
^y
o
aa
The steady-state
computed from (9.279).
derived from (9.274) and (9.275)
(9.280)
concentration
of
where
of
is
aa
is I0
is
+ M.2/(0) *(9)]e'0/^<9')'i9'
Sm(<)
\j0
8,(0
XP
(0)
S/
M<rj,p'fnoail
fnoail
REACTOR
(514
ANALYSIS
[chap. 9
I.
&
lim
(T
X0 =
lim
S%>
&
(9.282)
Poison Concentrations
following Reactor
Shutdown.
The time-
In the
Sm(<)
case of samarium,
= S(0)
+ P(0)(1
this procedure
- rV)
easily
(9.283)
= 0.01472 hi-1.
Although the postshutdown concentration of samarium can be
a serious
tration increases rapidly after shutdown and can reach appreciable mag
This buildup is important from the
nitudes before decaying away.
standpoint of control, since a large positive reactivity change must be
available if it is to be required that there be a restart of the reactor
Thus special "xenon-override"
around the time the maximum occurs.
control rods would be required in a mobile nuclear power plant if shorttime startups were to be an essential operational feature.
At the time of shutdown, there exists a particular combination of
in the reactor. This combination
iodine and xenon concentrations
depends, of course, upon the operating history (i.e., flux program) as
In any case, these can be deter
indicated by Eqs. (9.208) and (9.271).
1(0 = I.e-X
X,
X;2.)=
IX'
X,
(X,
X*)
e-x-'
(9.284)
e-M'J
instant of shutdown.
(9.285)
SEC. 9.6]
where
and
I,
KINETICS
REACTOR
and
X,
z, =
of
615
I136
(9.286)
i*
We have introduced the parameter z, in the symbol for the xenon concen
tration since the absolute magnitude of the concentration depends upon
this ratio, as well as upon the value of I..
The expression (9.285) is generally applicable and gives the xenon
For
concentration after shutdown for a specified value of the ratio z,.
each value of z,, a separate calculation for X(t;z,) is required, and each
will yield a different time behavior.
This complete system of solutions,
The appro
however, can be combined into a single generalized relation.
priate form is obtained by applying a displacement transformation on the
Thus we define a new timelike variable 8 which is
time coordinate.
related to I by
=
6*
(9.287)
(X,
Xx)]
<rM-*)
e-x,(.-.*>j
The result
0.
is
(9.288)
lnd-Kz./x.Kx.-x,)]
Xi
i =
28Q)
Xx
Ai fti
be
Ai
(Xi
(e-M
e~)
we evaluate
in terms of its
"1m/<m-x,)
if
- X.)
f*
r-^V
X,
^
X(0;z.)
6a =
and
is
it
(9.290)
seen to occur at
ln (X[/Xj)
(9.291)
Xi Xx
follows that
(0;z.)
X(0m;z,)
- X.
XxeXl-
(c-x,
c-x,)
3(0)
(9.292)*
'A. G. Ward, "A Universal Curve for the Prediction of Xenon Poison after a
Reactor Shutdown," Atomic Energy of Canada Limited, CRRP-685, Jan. 25, 1957.
ANALYSIS
HEACTOR
616
[CHAP. 9
after shutdown
for the
28
32
36
Timefl, hr
Fig.
9.14
A further obser
to a more simple form which passes through the origin.
vation to be made is that the X and X functions are related by
X;e)
X(t;0)
(9.293)
at time
0 0*
after
o for a long
Xl(Ml
XiMi
+ Migq'^O
+ Mici^o
Ml)
Then,
(9.294)
sec. 9.6]
KINETICS
REACTOR
617
12
s
o
0)
>
n
a
Using:
A/
Xi
0.1037
hr" 1
0.0753 hr"1
0
Ratio of xenon concentration
to iodine concentration
Fig.
9.15
at shutdown
z.
The system
is not at steady state, however.
For convenience we assume that the
reactor is being started up for the first time, so that the initial concen
trations 1(0) = X(0) = Pm(0) = Sm(0) 0. Equation (9.268) gives the
I136 concentration
in this case to be (again taking 2/ constant)
case of a reactor which is operating at a fixed flux level
I,
1(0 = 1.(1
- e-*-')
L.
(9.295)
m X,
we
where
AW
with
A.
X.
L^
and
_
and
LJ
j
Af
- X. +
<r
identify
(9.296)
<,
L.v,(M,
with
<<>
- L,)|
(9.297)
(9.298)
ANALYSIS
REACTOR
G18
[CHAP. 9
2.0
1.5
1.0
Using:
0.80
X,
0.1037
X, = 0.0753
0.60
hr"1
hr"1
0.20
0.15
0.10
0.04
0.03
0.02
0.01
12
(j> X
Fig.
Ratio of equilibrium
steady-state flux.
9.16
"l
Sm149
10""
20
16
concentrations of xenon
[from (9.278)],
*>-.">*.<
24
28
(neutrons/cm 2 /sec)
/>-.< 1
to iodine as function
of
2,
(9.299)
where
b- =
~ar
and
(9.300)
These results give the buildup of the fission products for a reactor
Note that
starting at t = 0 and operating thereafter at flux (t) =
for t ao these solutions reduce to the steady-state values derived in
Sec. 9.6b if we take + = 0.
sec. 9.6]
REACTOR
KINETICS
619
P(t) = P0
2/(0 *(0
&T/Nr(0)
(9.301)
4>(0)
where P(t) is the power generated per unit volume by the reactor, Po is
some specified power level, NfU) is the concentration of fuel (nuclei per
unit volume), and is the energy release per fission.
As noted previously,
a statement
NF(t)
where
-*N
constant-power
is seen to be
relation
If the
the solution
(9.301) is applied
N,(t)
(9.302)
NP(0)[1
to this equation,
a<"(9.303)
(9.303)
(9"304)
above.
The substitution of this expression for (t) into
for
solutions
the fission-product poisons yields the appropri
the general
reactor.
These
ate time-dependent functions for the constant-power
results can be expressed in terms of the incomplete gamma function.
as mentioned
The symbol
(9.305)
ANALYSIS
REACTOR
620
[chap. 9
so
system
alF)(0)t
<F)0)t
It
In
solid-fuel
U,
SSfs
(9"306)
that Pok/ is the total fissions per unit volume which occur in
time k; but this may also be given by
is clear
If
we call
/3/-
[NF(0)
- N,(k)] %
Pr
- NF(U,)
NF(0)
(9-307)
o'aF)(0)t = &Fz
then
(9.308)
NF(0)(1
- Pfz)
4>(z)
j-z^
(9.309)*
a fission reaction
in the literature.1
It will suffice for the present
to consider an elementary approach for estimating the con
The fundamental assumption
centration of these secondary fragments.
will be that all the secondary poison nuclei which appear can be regarded
a fission is available
treatment
as a single
If
we
Nrp{t)
= c,NF{t)
(t)
(9.310)
sec. 9.6]
REACTOR
KINETICS
621
For the
considered.
case
of constant power,
we use
Nr(t)
iV,(0)e-'.1"^'
(9.312)
g.
case of constant
is given
by the
equation
Nj(t)
where
-o*Nffl (t)
(9.314)
iV>(0)(l
JVy(z)
Ar;(0)(l
where
$,zy<
/8jp)'
sj =
(9.315)
(9.316)
(9-317)
(6.81)
In
these calculations
we are concerned
(9.318)
622
ANALYSIS
REACTOR
[chap. 9
is
1,
if
is,
phenomena, and, strictly speaking, the relations (6.81) and (9.318) are
for steady-state operation.
meaningful only when k = 1, that
Thus
then the flux
we speak of
^
experiencing rapid variations in
it
is
is
shutting down or
time, so that the reactor either
supercritical.
will be useful in this analysis to think in terms of a timeNevertheless,
dependent multiplication k(t) which will describe the influence of the
is
is
various poisons and the fuel on the reactivity during the operating life
of the system.
Actually, k(t) =
all through this period, but this
accomplished only by the continual action of control system; fc(<) repre
sents, then, the available multiplication at any given instant and
the
amount which the control system must counter in order to maintain the
This concept of a
^
was previously treated
=
so
that
by introducing
Thus
a fictitious number of
vc
vc
v/vc.
neutrons which must be produced per fission in order to maintain the
In the present application we may think of a
system at steady state.
a
is
is
corresponding
If
*(0
we call
0(0
W*/(0 Pnl(0
(9.319)
- 2f^y
(9-320)
It
will
t!>
WW)
.Fwrai
l9-321)
m
pM'>
where
be convenient
Z(0 = Z> +
(9.322)
2?>(0
is
o() =
y^Wz
sMl
pendent of time)
i-1
^2)"-1
(9"323)
sec. 9.6]
KINETICS
REACTOR
623
...
of poison
j:
ArX0)
Uj =
be used
(9.324)
N,{0)
in (9.319):
-
wth
o()
iTT^
fc(z)
[[
then,
1
^1
The reactivity
is,
(9.325)
=^ (0)(1
(9.326)*
is
is
is
Fia.
9.17
Reactivity
change
poisons
rij and their types, represented by the cross section ratios 8j, to control
the reactivity to variations which lie within the capabilities of the con
trol system.
If no poison were added to the constant-power reactor, the
reactivity would vary as shown in Fig. 9.17.
The initial reactivity oj(0),
here, may be
0,
controls.
reactivity can be made to take the form shown by the solid-line curve in
the figure.
Note that the maximum excursion in oi{z) has been dimin
ished and might be brought within the range of the controls.
In some
situations the maximum rise in reactivity may still be beyond the capa
of the controls; such
is
it
is
bilities
(e.g.,
is
possible to use
combination of B10 and some other poison
cross section.
Now the com
Li6) which does not have so large
bined initial loadings of these two poisons will suppress the large cu(0),
and yet a sufficient amount of the lower cross section poison will always
cases
it
- Without
ANALYSIS
REACTOR
624
[CHAP.
= 0
(1) = 0
the
it is
(9.327)
PROBLEMS
Consider a bare thermal reactor [described by the constants D, B*, 2a(0),
9.1
2/(0), pth, and to,] which is operating at steady state. At some time instant, say
I = 0, the physical state of the system is caused to vary in time, so that the macro
Such behavior can be achieved, for
scopic cross sections become functions of time.
example, by changing the pressure on a compressible fluid fuel.
a. Derive the expression for the thermal flux (p(i,t) for this system on the basis of
the following model: (1) 2(<) = 2(0)(1 + o sin wt), 2/(0 = 2/(0)(l + 6 sin wt), and
D(t) m D; (2) neglect delayed neutrons; (3) neglect temperature effects on system;
(4) include oth e~B'Tth and pthj (5) assume that both gth and pth are time-independent
quantities.
6. Determine the function Sk(t) = k(t) 1 in terms of the steady-state constants
of the system. At what values of wt does 6k(t) attain its maximum value?
c. Sketch the functions &k(t) and (r,l) in terms of the variable wt.
d. Find an expression for {t,t) in terms of the mean lifetime I = \v\DB* + 2.(0)]
and the maximum positive value of Sk(t).
e. Determine the conditions on the frequency w such that ^m, > c^min, where c is
an arbitrary constant.
9.2
Determine the one-velocity flux distribution (x,t) in a nonmultiplying slab of
width 2a for which the diffusion medium is characterized by the one-velocity constants
Apply the condition that the flux vanish at the geometric boundaries of
2 and D.
the medium and that the initial condition is
4>(x,0+)
At
t = 0, a source
= 0
4>ocos
(1
+ s'n wi)
(t)
Ae"T
Compare these
applies, and find values of T which allow a solution to each case.
values for T, using the form (where a and b refer to the assumptions given above)
Tb = 7\,(1 + a)
and give an explicit form for a. In each case only one time-dependent equation is
In model 6 assume
needed; do not consider the precursors of the delayed neutrons.
that t/Tb <5C1 in order to solve the equation for 1\.
REACTOR KINETICS
625
= 0.001 sec
/S)
- T(t) -
7'(0)
given by
forM >
+ /*e(i)l
<P( = <P[1
<P[1
+ /*e(i)l
forM > 0
1.
T0
is
&(t)
is
- yS(t)
gXfth
sin
ft
A./th cos
ft
r/
Wth
~|
Skq
+')]
7^[1--,(fsin
ew
the specific
heat of
is
is
and l,h
the neutron lifetime in
finite reactor and
nized sample of the reactor material.
where
homoge
0.
k(t) = tko
CHAPTER
HETEROGENEOUS
10.1
10
REACTORS
Merits of Heterogeneous
Fuel-moderator Configurations.
Some
remarks were made in Sec. 1.2 about the advantages to be gained in
using heterogeneous arrangements and the basis for the choice of this
Aside from the physical
configuration for the earliest reactor designs.
separation of fuel and moderator, the most important feature of such
a.
systems is the fact that they are nearly completely "thermal" (i.e., at
least 95 per cent of the fissions are caused by thermal neutrons).
These
unique characteristics lead to significant improvements in some of the
By establishing a
which affect the multiplication constant.
thermal system it is possible to exploit the relatively large fission cross
sections of the nuclear fuels in the low-energy range, and by separating
the fuel and moderator, such quantities as the resonance-escape probabil
ity and the fast effect can be measurably increased.
factors
A
system with an optimum ratio of moderator to fuel is still subcritical.
heterogeneous arrangement of these materials, however, can be made
into a critical assembly. This is achieved in main part through the
626
HETEROGENEOUS REACTORS
SEC. 10.1]
627
The properties
of the graphite
arrangements.
heterogeneous configuration
40
Fig.
10.1
60
80 100
200
400
600 800
1,000
2,000
REACTOR
628
ANALYSIS
[CHAP.
10
exceeds
these factors have been determined to the desired accuracy, the kK for
the system is known and suitable expressions can be obtained for the
effective diffusion length and migration area of the lattice and, thence,
estimates of the over-all size of the core and reflector configuration.
The present treatment of the subject of heterogeneity begins with an
This portion
elementary discussion organized in the form of a summary.
Attention is directed
of the treatment is given in the present section.
here to the delineation of the essential features of heterogeneous systems,
the over-all effects of "lumping," and to simplified methods for estimating
Whenever possible, experimental
the characteristics of such systems.
results and formulas are presented which can be used directly in computa
The results and discussion given in this section should suffice
tions.
for a survey-type treatment of the general subject.
homogeneous system.
HETEHOOEN'EOUS
SEC. 10.1]
KEACTORS
029
A set of
collection of discrete line (or point) neutron sources and sinks.
written
this
be
for
and
may
the
solu
equations
system,
neutron-balance
tion of the set leads to the criticality condition for the reactor. This
general approach to the over-all problem is both elegant and precise;
it offers great flexibility in treating complex heterogeneous
arrays involving various lattice geometries, several degrees of periodicity,
The method is also useful for
and irregularities in rod size and spacing.
analyzing control-rod effectiveness, the influence of voids and cavities,
moreover,
It should be
as well as some aspects of the problem of fuel storage.
mentioned, too, that the mathematical treatment is somewhat more
complicated than that involved in the unit-cell method, and that even
the many attributes mentioned above, its
greatest usefulness is realized when applied to small heterogeneous arrays
The analysis of large regular geometries
and irregular configurations.
possesses
advantage
model.
which follows
calculating/,
p,h,
and
REACTOR ANALYSIS
630
[CHAP.
and
10
M, the
geneous systems.
One factor which
is,
1.
is
extremely
the thermal
if
is
distribution
-t
(10.1)
/..
If
the fuel and moderator regions are separately homogeneous and uni
form, then the 2<, = N&F are constants.
(Here AT,- denotes the actual
The above expression may then
nuclear density of the pure material i.)
written
>,
-w
be
fc(r) dt
(10.3)
<PF
1.
s
Z(f>
2f(Fv/7,)
/,
pth,
(1-5)*
and
km
is
/"
L,
if
if
is
is
at "disadvantage"
which we call the disadvantage factor, since the fuel
less than that throughout the
the average flux throughout its region
With this definition, (10.2) reduces to
>
moderator, i.e.,
still highly
HETEROGENEOUS REACTORS
SEC. 10.1]
631
as the heterogeneous
clearly
N?"VC^ =
where
NiVi
is the actual
(10.6)
7V
cross section of material
microscopic
i.
The
A.. <
/ta
(10.9)
a homogeneous system to a
heterogeneous, keeping the mass of each material constant, the thermal
utilization is reduced.
Another way of thinking about this behavior is to assign "effective"
cross sections to the materials in the heterogeneous system such that
Win
and
[*<>]. m
<r
(10.10)
Then, /h has
REACTOR ANALYSIS
632
10
[CHAP.
2.
Diffusion Length.
only approximately.
However, it seems clear that, in the case in which the unit-cell dimen
sions are small in comparison with the diffusion length, the form (1 +
L2g2)-1 Would be a good approximation to the thermal nonleakage prob
ability, and it would also be acceptable to use the general form L2 = D/2a.
Then, if
we use
"
D
along with (where
Fcen is
3[2{* +
(10-U)
Zf>]
throughout Fdi)
Zu#^-n
- VPirV, + VpiuVu
(10.12)
r,
(4>V^Y
2VM!
(1
\4>mVm)
where
LL
(10
V
131'
i +%^vTmW7t
* Ll,{l
" /hJ
(1<U4)*
1 Some
Resonance-escape
case of a
/het)^'f
F at VM,
3.
model.
HETEROGENEOUS REACTORS
SEC. 10.1]
633
Pth
exp
f"
/_
2a0
() dE
For
these
(6.174)
ty.^jj.
where
If the source of highto the number which enter the resonance range.
energy neutrons supplies to the system one neutron per unit volume per
unit time, then the appropriate statement for p(E) is
- -
n(F> =
p(E)
(10.15)
%{W)
F{E) = Z(()
where
(10.16)
4>(E)
in the system.
The corresponding
exP["
f" F(E)
Xa(E) dE~\
L~-ME)
is
of practical interest,
to unity so that
(10.17)
^-77^
1,
[~
F(E)Xa(E)dE
(10-18)*
approximated
close
plh
is
Pth
is,
is
first-order
useful
4,
J.
1958.
2,
J.
J.
J.
*J.
I,
1i
REACTOR
634
ANALYSIS
10
[CHAP.
approximations
ME)
-jj;
(10.19)
is
it
Nrfo'KE)
Nm<t(E)]
dE/E
2Q)
'
d
<rp
is
where
integral
ioI
by
(10.21)*
*" +
7T
(6'177)
is
*M'\E)
<>{P
<r(amN
m/N F.
i>2)
good approximation
+ NMe{E)/Nr\ dE/E
+ <r(E)] +cp + Nm<jM{E)/Nf
is
- W/\E)
(10.22) reduces to
o<T{E)
Eth
[N,*<(E)
(10.23)*
+ op{E)]
Nuclear Physics,"
pp. 398-411,
HETEROGENEOUS REACTORS
SEC. 10.1]
635
which is the usual form in which the effective resonance integral is denned.
It is important to point out that this result applies only to a system with
narrow resonances; as noted above, the low-energy resonances of some
absorbers are not well represented by this model.
When the fuel resonances are broad, it is better to use the NRIA
fEt
/**
The corresponding
approximation.
geneous system is
1
-l
+
ajF\E) dE/E
v.:
v..
<ri(E)Nr/
-if<i
(10.24)*
.>/>
The approximations
are
be
dK
8a(E) -=r
(10.25)
The functions sa and have several properties worth pointing out. These
may be illustrated with the aid of the expression for s developed for the
NR approximation.
(Similar arguments may be applied for the case of
Using then the form given in (10.23),
we make
o
that for
Op
(10.26)
+ aiT + <Pl
a very dilute system
(N?
>
0)
m
1
Sa
approximation.)
is
NRIA
and
(10-27)
increases, av
,f-:t>
resonance integral
a{p, and
(io-28)
sa/
it
<
Sf
)F
<
< Sx where
refers to
the fuel material alone, and
refers to the infinitely dilute fuelWe find, therefore, that as the fuel concentration
moderator mixture.
increases, the effective resonance integral decreases from its greatest
is
the
REACTOR ANALYSIS
636
[CHAP.
10
highly absorbing.
ME),(E)
'
mnE)
ME)
[ME)
MEWJ
- Wr
<n
+ e,]SE
When the number of fuel nuclei per scattering nucleus is large, o'f/tj, is
varies roughly as l/2aE; however, when the fuel concen
large, and
varies roughly as 1/2?.
tration is low, o^/op is relatively small and
Thus the higher fuel concentrations result in more pronounced dips in
the spectrum of the resonance flux and thereby fewer absorptions per
fuel nucleus per unit energy interval, hence, the smaller integral.
Another interesting observation involves the physical interpretation of
According to (10.21)
(10 29)
I/Np
Now
Conf. Peaceful Uses of Atomic Energy, Geneva, P/833, vol. 5, pp. 96-101, August, 1955.
' L. Dresner, The Evaluation of Resonance Integrals, Nuclear Energy, 2, 118-127
(1955).
HETEROGENEOUS REACTORS
SEC. 10.1]
637
U238
and Th232
have been studied in greatest detail, and empirical formulas have been
obtained for computing these integrals as functions of fuel concentration.
The best available experimental data are plotted in Fig. 10.2. Dresner1
has computed from resonance parameters, by methods to be discussed
later, the appropriate functional behavior and gives, as the best powerlaw mathematical fit to his computed data, the formula
g(Eih) =
This function
3.06V"-472
for uranium
(10.30)
Figure
10.2
8.33a"-263
for thorium
232.
(10.31)
of Thorium," Argonne
REACTOR ANALYSIS
G38
[chap. 10
Table
10.1
Th"
U9
93 1
269 X
282
69. 8
86. 2%
291
442
400
467 %
Pu"
Reference
Chernick- Vernon"
:)
520
Macklin-Pomerance'
Weinberg- Noderer*
Trubey-Lessig''
Calculated value.
Measured value.
J. Chernick and R. Vernon, Nuclear Set. Eng., 4, 649-672 (1958).
6 R. L. Macklin and H. S. Pomerance, Resonance Capture Integrals, Proc. Intern.
Conf. Peaceful Uses of Atomic Energy, Geneva, P/833, vol. 5, pp. 96-101, August, 1955.
c A. M. Weinberg and L. C. Noderer, "Theory of Neutron Chain
Reactions," vol. I,
chap. IV, Oak Ridge National Laboratory, CF 51-5-98, May 15, 1951.
* D. K. Trubey, personal communication; see also D. K.
Trubey and R. H. Lessig,
Neutron Physics Division Annual Progress Report, Oak Ridge National Laboratory,
ORNL-2609, Oct. 29, 1958.
Most recently,
upon measured values of the effective resonance integral.
more and better cross section data have become available, and it has
become profitable again to apply theoretical methods to this problem.
As a result, some new analytical models have evolved which have proved
quite helpful in further understanding and describing the resonance
SEC. 10.1]
HETEROGENEOUS REACTORS
639
Phet
If
>
improved.
As
a rule, then,
for
(10.32)
Phom
for both systems, then the corresponding value of the resonance integral
is about 9 barns, a factor of about 26 reduction.
Advantages of this
order of magnitude
are
generally
achieved
in going to heterogeneous
configurations.
The earliest
a linear
More
1 Ibid.
1 S.
J.
REACTOR ANALYSIS
640
[CHAP.
10
recently it has been shown1 that for many situations of practical interest
a given heterogeneous system can be described in terms of an equivalent
homogeneous system; thus many of the results developed for the homo
geneous systems may be carried over directly to applications for hetero
geneous arrangements.
standard
The United States, and generally Western, usage of the standard for
mula was based for some time on the following definition of the effective
:
s(w) = o(u)
4rl
cross section
J\l
absorption
K)
(10.33)
Let Q(u)
is
(10.34)
is
Q(u) = ImVmNm^4>m(u)
is
seen
(10.35)
it
If
that
Q{u)
Q(0) p{u)
Absorption
6,
is
denotes the surface area of the fuel lump and M its massThe appropriate expression for the effective resonance integral
then
developed by means of the following arguments:
where
of Neutrons in Nuclear
Chernick and Vernon,
1958; also,
I"
dw
641
= exp
So(u
,1
VFNF
_____
exp
p(M)
where
/"
HETEROGENEOUS REACTORS
SEC. 10.1]
VrNrSiu)'
^^^.^
(10.36)*
is
a{u) du
0
JC
U,h
b(u) du
Mr Jo
(10.37)
'"-(fsf;)8'
(ia39)
a(u) du =
/."*
fr4-'
Jo/
is
it
Thus
follows that the effective
the resonance-escape probability.
simply the integral of sa as in the homogeneous
resonance integral
systems [cf. Eq. (10.25)]. If we use (10.33), then,
fr
p'F
1,
f,
we take
= 8.6 barns
now,
if
Ar
and
Br = 27-5
is
then
= 8.6
27.5
uranium metal
(10.41)
is
where AF
given in square centimeters and MF in grams.
We observe
from this relation that, as the lump size increases, AF/MF decreases and
the effective resonance integral decreases.
In the limit for very large
1
fr
where
denotes the ratio of the "average resonance flux" in the mod
the
erator to the average resonance flux in the fuel lump; thus we call
resonance disadvantage factor in analogy to the
given by Eq. (10.4)
for the thermal flux. The Nf denotes the nuclear density of the fuel
REACTOR ANALYSIS
642
[chap.
10
0, and
lump (AF/MF)
-g/dE
a
= 8.6 barns
24.9
S = 20.4
3.16
24.5
27.4
(10.42)
(10.43)
U02
(10.44)
4- 4r
24.9
U308
thorium metal
0.1 <
ff
= 10.4
d
3
case of thorium,
= 8.4
Th02
0.2 <
< 1.3
(10.45)
is
(10.47)*
L;
(10.40)]
relation,
the constants
in standard
American
that the
Ibid.
I.
3,
(1958).
of
5,
I.
I.
of
In the
= 11.0
4^
4r-
&
4^
= 12.8
which is approximately -fa of the value for the infinitely dilute homo
geneous mixture of natural uranium and moderator.
Expressions of the
have
for
been developed also
the various compounds of
type (10.41)
uranium:1,2
HETEROGENEOUS REACTORS
SEC. 10.1]
643
24.7
6 = 4.15
26.G
hj^y
(jfj
uranium metal
(10.48)
UOj
(10.49)
thorium metal
(10.50)
Th02
(10.51)
and3
= 4.1
21.1
g = 0.2
25.2 (
(Jf\f
4^ J
d. Effects of Lumping.
The distribution of the fissionable material
in heterogeneous arrays leads to three important advantages and one
The advan
disadvantage from the standpoint of the reactor physics.
tages to be gained by utilizing a heterogeneous configuration, however,
generally outweigh the principal disadvantage, namely, the reduction
in thermal utilization. This effect has already been discussed with the
aid of an elementary analytical model, and it was observed that by
separating the fuel and moderator materials the fuel is placed at a dis
advantage in competing for the available supply of thermal neutrons.
Without entering into a detailed calculation, it is easy to establish this
The basis of the argument rests with
argument on physical grounds.
the fact that the slowing-down process, which is the primary source of
thermal neutrons, occurs in a region which is physically separated from
the fuel.
Thus the moderator region surrounding each fuel lump is a
source of thermals, whereas the fuel lumps are essentially sinks.
This
large source in the moderator provides a steady stream of thermal
neutrons to the fuel lumps; the "driving force" or potential is achieved
Thus
by maintaining a higher neutron density in the moderator regions.
the moderator is exposed to average flux levels higher than those in the
fuel.
G44
REACTOR
ANALYSIS
[chap. 10
Because
nance flux exhibits local depressions around each resonance.
of the lumping, the majority of the fuel nuclei in the lump are exposed
to a depressed flux much like that in a homogeneous mixture of high
As argued previously, this corresponds to a rela
fuel concentration.
tively small value of <rp and of the effective absorption cross section of the
fuel 8a [cf. Eq. (10.28)]; consequently the effective resonance integral is
reduced and the resonance-escape probability increased.
There is another, somewhat secondary, effect which serves further to
Because of the physical
enhance the resonance-escape probability.
separation of the two materials, the neutrons involved in the slowingdown process have less opportunity to encounter a fuel nucleus while
traversing the resonance range; the number of resonance absorptions is
Clearly, this effect is due primarily to geometric
reduced.
thereby
considerations.
difficulty stems, of course, from the fact that there are local variations
In the
in the neutron flux at all energies in each unit cell of the array.
central regions of the reactor core this does not lead to any particular
hardship since the flux distributions will be relatively flat and variations
This is not the case however in cells
from cell to cell will be gradual.
Here, variations across the cell can
near the boundaries of the lattice.
be substantial and will influence markedly such factors as the thermal
utilization and resonance-escape probabilities by distorting the neutron
These difficulties are by no means unique to
distribution in the cells.
the method based on the unit-cell concept; similar complications arise
In the latter appli
in the application of the Feinberg-Galanin method.
cation the difficulty appears in selecting suitable kernels to describe the
HETEROGENEOUS REACTORS
sec. 10.2]
645
determination
10.2
Disadvantage
Analytical Model for Unit Cell. A practical method for the calcu
lation of the disadvantage factor and the thermal utilization is based
on the Wigner-Seitz unit-cell model.
In order that this model be appli
cable to a particular situation, it is
necessary that certain simplifying
a.
Fuel
These
Moderator
may be illustrated by
means of an example.
the reactor-core
Suppose that
of in
Cell -
configuration
fuel
1.
All characteristic
cell half-width
with the
of the thermal utilization may
reactor dimensions
be large compared
b.
Then the calculation
proceed on the basis of an infinite medium of this structure, since the
regions near the edges of the reactor core (for which the disadvantage
factor will differ somewhat from the fairly uniform value over the central
region) will not play an important role in the criticality calculation.
The
disadvantage factor f we compute from a knowledge of the thermal-flux
646
REACTOR ANALYSIS
[chap. 10
distribution throughout the cell, and this we obtain with the aid of the
diffusion theory.
We further require, therefore, that:
2. Diffusion theory be valid in both fuel and moderator regions of the
unit cell. It must be pointed out that this assumption may be poor
on two counts: (a) The fuel region is likely to be highly absorbing for
thermal neutrons; thus the angular distribution of neutrons in the fuel
region and near the fuel-moderator interface in the moderator will very
probably deviate considerably from the distribution appropriate to
diffusion theory [see Eq. (7.93)].
(b) The dimensions of both regions
may be of the order of a transport mean free path or smaller, that is,
much smaller than a diffusion length in either medium.
The nonisotropy
of the flux in the vicinity of an absorber-moderator interface was demon
Figure 7.16<7 shows the typical strong forward bias
strated in Sec. 7.5a.
toward the absorber region which is to be expected in the angular dis
Finally, it should be mentioned that there
tribution at such boundaries.
is also a third difficulty which is encountered in the use of the diffusion
theory, namely, the neutron-energy spectrum around thermal varies
Thus there is some
appreciably through the moderator and fuel slug.
uncertainty in selecting the "one velocity" for the model.
In spite of all these shortcomings we will assume that diffusion theory
So the results obtained below will not be useful for treat
is applicable.
absorbing fuel plates (or pins) but could be well applied
thin,
highly
ing
to large natural uranium lumps in a good moderator such as graphite.
A suitable model, based on the "first-flight-transport" concept developed
in Sec. 5.8a, is applied in a later analysis to handle these cases.
It should
be mentioned, however, that disadvantage factors calculated by diffusion
theory can be low by about 20 per cent, giving an error in the multiplication
constant of about 2 per cent in practical cases for natural uranium with
graphite or heavy water.1 The success of the diffusion-theory treatment
during the Manhattan Project was achieved by adjusting the value of jj.
When more accurate estimates of the thermal utilization are required
than possible with the methods discussed here, it is necessary to resort
to the direct use of the spherical harmonics or the so-called Sn methods.2
Another essential question which must be considered in selecting an
analytical model is the specification of the thermal-neutron sources in the
cell. In the actual situation, neutrons are born with high energy in the
fuel plates; during the slowing-down
1
HETEROGENEOUS REACTORS
sec. 10.2]
647
original plate and may finally become thermal at a distance many slab
thicknesses removed from the source point at which they appeared as
fast neutrons.
is,
down
VniO,
is
10.4
Slowing-down
Fermi age theory.
is
is
This
of age theory when first-flight corrections are taken into account.
leads to
greater spreading out of the neutrons in most moderators
Thus for
because of the low scattering cross section at high energy.
is
is
3.
is
fuel plate
we may concentrate on a typical region which includes half
This region
moderator slab, as shown in Fig. 10.3.
adjacent to half
sufficient for the present purpose since we are interested only in the
ratio of the average fluxes in the fuel and moderator regions and since the
is
is
is
is
the entire core and (6) the net transport of neutrons across the surface
of the cell be zero.
of interest to mention that
In concluding the present discussion
it
Fig.
ANALYSIS
648
REACTOR
and approximations
[CHAP.
10
We demonstrate the
calculation of the disadvantage factor on the basis of the diffusiontheory model outlined above for the special case of the slab configuration
This derivation is followed by a summary of the
shown in Fig. 10.3.
general results for the elementary lump geometries.
The mathematical problem may be stated in terms of the general
For the present application these
relations (5.349) through (5.354).
may be written
In
the
-DF^(x)
In
2/V,(x)
= 0
(10.52)
the moderator
-DM^(x)
Generated for wjivans (University of Florida) on 2015-09-22 17:23 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
+
+
2<Mx)
= So
Ma)
(3)
*',(())
The solutions
= 0
(4)
=
DF4>'F(a)
= 0
4>'M(b)
distributions
Du'M(a)
4>f(x) and
postu
UUiWJ
4>m (x)
are
easily shown to be
.
cosh xf
- ID,*,*
,7a
- x)
cosh
- jTtfj f - rf/Wsinh
So
*'(X)
, v
.So
<M*)
where
xM{b
c"th
= a)
and
.\f(x) dx
r
J--ra
(b-
a)
F(x)
dx
r
aS^F)
**
in
- m ~- aW*
6V>
,.,
*(&-a)J
6 =
(10-4)
-sinh
)-
(10-55)
(10.56)
(10-57)
have
JSt
aZ<"
2(&
- a)
sec. 10.2]
HETEROGENEOUS REACTORS
649
aZ
volume
ratio constant
This result
of course, to
= constant
to
is
ing
is,
lim
o,60
It
b;
a)
(10.59)*
Vmx.m
coth
x.v/(6
a)
and
The functions
written
be
(10.58)
so that
- -
(b
coth x,v(b
be taken as
a)
xM(b
Vu
10.5
Disadvantage factor as
function of cell size.
- a)Z
xfQ. coth Kpa
+
(6
- a)2
Fio.
(10.60)
is
is
is
and E.
The two cases are:
geometries, with suitable expressions for
fuel
rods
in
lattice
and
spherical
lumps of fuel
cylindrical
square
(2)
(1)
in cubic lattice.
The appropriate expressions for the unit-cell dimensions
by requiring that the cell volume be equal to the volume
"element"
of the lattice; thus,
a
the
of an
the lattice spacing,
is
are computed
if
against size.
For convenience in presenting the cor
responding results in other geometries,
we rewrite Eq. (10.57) in the form
ANALYSIS
REACTOR
650
[CHAP.
10
lump radius, and Rm0 the radius of the unit cell, then the cell charac
teristics are given by:
V,
Lumps
Cylindrical
*R'
Spherical
frfi'
VM
a1
a3
Ke.ll
- t*
- |t3
ay
a'
The calculation
Table
10.2
The Functions
F
E
and
a)\M coth
x.w(6
a)
Cylindrical
F
p
PfxfI o(xfpf)
21
i(xfpf)
., ,
*.v(p5#
pV>
, ,
2pr
Spherical
4) r
1
x.vr.w
.
xjjr.urF
coth xu(ru
xm{tm
rr)
ESlFuel
I
| Moderator
case of the
10.6.
HETEROGENEOUS REACTORS
sec. 10.2]
definition
"{y'Z)
2y
y = x.MpF
with
[h(z) K,{y)
and
651
- K,{z) h{y)\
= xmPm
c. Disadvantage
theory
0.02
0.03
0.04
0.05
Fig.
10.6
0.06
Graph of function
0.07
0.08
0.09
0.10
*mPf
the fuel is present in the form of thin, highly absorbing plates or in the
form of fine pins.
For these situations it is better to apply the results
of the first-flight transport model developed in Sec. 5.8a.
The applica
tion of this model is demonstrated for the case of the thin plates, followed
by a summary of the corresponding results for the pins.
Consider, then, the slab geometry of Fig. 10.3 and suppose that the
fuel plate is very thin, so that we may assume that the plate either
absorbs a neutron incident
collisions
This formulation is in
are considered to have negligible probability.
accord with the definition preceding Eq. (5.275).
Let us assume further
ANALYSIS
HEACTOR
652
[chap. 10
that the diffusion theory is valid in the moderator and that neutrons are
being thermalized uniformly throughout the moderator region (with no
The boundary condition at the edge of the
thermal sources in the fuel).
unit cell is as before. The fuel-moderator interface condition on the
other hand is now conveniently given in terms of the transmission coef
ficient a of the plate [see Eq. (5.275)], since the plate is considered too
thin to allow the use of diffusion theory.
The quantity a may be com
puted in terms of the plate dimensions, its absorption cross section, and
the angular distribution of incident neutrons.
The calculation of the thermal-flux distribution in the moderator is
carried out in the usual way by applying (10.52) and (10.53), condition
For the interface condition we use the partial current relation
(4).
j+(a) = aj_(a)
cosh XM(b
a)
+ 2Dmxm
x)
tanhxAf(6
is
In this application
is
ZWM(a)
it
(10.62)
(10.63)
coth xM{b
1
1
xM(b
a)
<A^[(^:)'H
o)
- a)V>
(6
since the net current into the plate must be the rate at which neutrons
It follows that
are lost by absorption.
(10.64)
Note that the only properties of the fuel plate involved in this result
cross section and the quantity a. Thus, so long as
valid in the moderator (particularly in the vicinity of
the plate), the only new information we require in using this approach
is
is
is
it
is
if
cosh xm{o
a)
vm(x) =
is easily found to be
^jj
(10.61)
HETEROGENEOUS REACTORS
SEC. 10.2]
653
= 2a2;
(10.66)
The analysis for the system with thin fuel pins proceeds in the
Without further comment we list the results below:
way.
Hp)
S,
4>m(p)
2D,
where
- L(pf)
same
(10.67)
(10.68)
and
Prvr
(10-69)
Fio.
10.7
Thin fuel
Thin fuel
-plates:
-pins:
1
1
The results
- 4aSi"'
-
2PjP2/>
(10.70)
It
is of some interest to compare the results for a using the diffusiontheory and first-flight transport models for a thin fuel plate.
These are
found to be, with 2a2^>
1,
First-flight transport:
Diffusion
theory:
These functions
- 4a2 +
+ 2[2a2]2
[2a2^]2/!;1[2a2^]
(10.71)
AaVaF)
REACTOR ANALYSIS
654
[CHAP.
10
Thermal Utilization.
Fig.
10.8
for
/het
AIt
is understood
2<"
(10.72)
Z{VK/Vr)t
that the cross sections indicated here are for the pure
materials.
The general shape of the
ratio of values is shown in Fig. 10.9
as a function of a characteristic cell
dimension.
e. Engineering Considerations in
Unit-cell Analysis. The unit-cell
configurations discussed so far have
consisted simply of a region of
moderator
practice,
variety
sec. 10.2]
HETEROGENEOUS REACTORS
655
pf ) is clad with
a dissimilar
of thickness t.
the rod is an
Concentric
metal
with
annular passage
(outside radius p) for the flow of
coolant gas.
Beyond this is the
moderator
of ceil
Fl- J0-10
Unit
cel1
for cylindrical
fuel
assume that (1) diffusion theory applies in both the fuel and moderator
regions, (2) the cladding material has a transmission coefficient a, and
(3) there is a uniform source of thermals S0 in the moderator region.
On
this basis the pair of equations (5.349) and
are
applicable.
These
(5.350)
equations are to be solved with the boundary conditions
= 0
(1)
4>'F(0)
(3)
aprjr(Pr) + 0pjl"'(pa)
(4)
PFjLF>(pF)
(2)
(1
4>'m(pm)
= 0
f>j?>(pt)
- H)PaV'\pQ)
(10.73)
where
0
It
55
(10.74)
that relations (3) and (4) account for both the absorption
losses in the cladding material and for the fact that with a gap there is
always a nonzero probability that some neutrons will pass from moder
ator surface to moderator surface without encountering the fuel rod.
is seen then
10
[CHAP.
CL(P) +
and
The con
and the
be
vrvr*,
is
defined by (10.68).
function L(p)
The thermal utilization for this cell may
(10.75)
4>m{p)
A/,(x,p)
4>f(p)
/3;
/3
REACTOR ANALYSIS
656
nn
7(V.
is
is
ratio of
the average flux in the cladding to the average flux in the fuel lump.
For
the moderator and fuel regions we obtain the following expressions for
the average flux:
V^Z
FXF
2
So
^a
2ttpL'(pii)C
V~7*
ttxM
*M ~
as well as the
'
2irpFh{xFpF)A
*'
(10.78)
Pf(1
If Eq.
(10.79)
is
7,
where
- 0)(
P,(1
and
are evaluated next.
From the boundary con
The constants
ditions (3) and (4) we obtain two equations involving these quantities.
Equation (10.73), condition (4), yields
7/0
lT^[CLM+w]
yM =
(10-79)
(10.80)
D. A. Newmarch,
with
J.
-fi)(G
+ yF)
T+w+w+wr,
(10-81)
\Zt
(10.82)
Nuclear Energy,
2,
y* _
(1
it
it
52-58 (1955).
HETEROGENEOUS REACTORS
SEC. 10.2]
657
If
and (10.80).
"
- gXg
(1
2DM[l +PG +
- - PKG + T'Wa)
PhrWM
(G
+ y,)S9/V>
(1
fJ)L'(p,)DS0
2ap,(l +(?)(!
PFZ^h{xFpF){2DM[l + 0G + (G + fihrWM
- - PKG + yF)L{p,)\
(1
(10.84)
GF
P,2<"j
l\ i
it
p
(10-85)
/in0M
(10-86)
(1-87)
P = pFXFIg(xFpF)
7rGp
/3G
i+
E* ~
2Jl(*,p,)
~ VMxlfL(pg)
(10-88)
arp,L'(p.)
If
Trp^l
0,
(10.89)
0)
'-?(rb)'
>
>
>
0,
/3
pP, then
/3
P0
/(r) =
--
sin-1
IT
- IT
(1
where
+ W(r)
+ 1M
that
(10.90)
r2)*
/Wd.
H. Tait has performed similar computations using transport theory.
The
Calculation of the Fine Structure of the Thermal Neutron Flux in
Pile, by the
Spherical Harmonics Method, Proc. Intern. Conf. Peaceful Uses
Atomic Energy,
Geneva, 1955, P/433, pp. 417-422.
of
J.
vMx<m
T^y-wJ
and
Po
The functions
vFz<p)
i^g [tw^J
where
4>,
ANALYSIS
REACTOR
r>.-)8
[CHAP.
10
Newmarch performed his analysis for the rod without cladding, i.e.,
a = 1 and G = 0. The above expression for /3 can be taken over directly
for the more general case with cladding (a < 1), however, by simply
applying the appropriate expression for yM, namely, (10.81), Then, /S
takes the form
0 =
It
- r)(l + JrG)
(1
+ yrG) +
(1
(G
+ (G + yr)Kr)
+ 7*0/(r) + r(G + yr)
(10.91)
+ yfG)5
+G)(1 + yF)
(1
(1
(10.92)
a =
For this
case
G^a,,
PF
where a
Pi
+
1
El
/3a
- 0
I + _F
E +
(g
aJF^'
-
(10.93)
*P.(1
p7
W(i
P,(1
0)
xp^jj
p^t \E
0)J
+ a*
Pa
10-94)
J)
PF
+ E
p,
pF
ggggjg
irp(l
Trpr(l /3)J
(10.95)
0)
Po
+ _
V.,,2
irp(l
-0))
ipS
P.
Vm2?F
F,Z/>(1
0)
I [
hollow-tube
It
been
HETEROGENEOUS REACTORS
sec. 10.2]
659
that the above result agrees term by term with the earlier treatments.12
The only apparent difference between this work and the previous calcu
lations is in the handling of the boundary condition at the gap and the
The
description of the flux distribution through the cladding material.
of
is
the
more
subtle
consideration
the
two.
latter
The dis
perhaps
tinction between the two approaches on this point is easily demonstrated.
In the earlier work it is assumed that the flux distribution at the outer
surface of the cladding is given by the linear function
Mpf
t)~ c(pr)
U'cM
Mpf)
^(Py)
This
The last equality follows by virtue of continuity requirements.
quantity may also be obtained from the condition on the partial current
through the cladding,
appj^M
If
we use a =
Pc??(pp +
t)
pc,
a* in this relation, and the approximations pF
as small compared to unity, then it
c(pf
t) = <Pf(pf)
+ 2aifDF'F{pF)
For
a very
If
the relation
(10-96)
2lTc
be appropriate
to use
2Sf'i
previously mentioned.
There are several approximations which may be generally applied
in the analysis of composite unit cell structures of the type considered
here: (1) Narrow gaps and channels containing air or other gases are
most easily treated by requiring that the flux and the net current be
continuous
DrVF(PF)
1 See,
M{p)
= DMV<t,(pg)
nno/n
UU-y/;
REACTOR ANALYSIS
660
[chap. 10
If a
with continuity conditions to be satisfied at the interfaces.
region is very thin, then (4) it is convenient to use the transmission
For composite regions of
coefficient to describe its neutron properties.
different moderating materials it is necessary (5) to take the thermalneutron source in each as proportional to its value of 2, ; however, (6) if
the region is thin (and especially if it is "metallic"), then the thermal
(3)
the
producing)
properties of the material.
(thermal-neutron
In the case of water the situation is further aggravated because of the
small diffusion coefficient which creates a large gradient in the neutron
density.
moderating
10.3
Resonance-escape
Probability
An important exception
is aluminum.
HETEROGENEOUS REACTORS
SEC. 10.3]
661
Q(u) =
=
q(x,u +
- Jf
cell
Am)
dx
q(x,u) dx
2a(u)(x,u) Audx
(10.98)
<Kx,E3)
(.x,Et)
*(*.,)
i,h s
Fio.
10.11
Flux
E4
Ea
E3
depression
E,
As
heterogeneous system.
probability will
be higher in the
REACTOK ANALYSIS
062
[CHAP.
10
attributed
to two effects, namely, the self-shielding of the fuel lump and the phys
ical separation of the two materials (refer to Sees. 10.1c and d).
The
in
study
of
section
is
to
these
effects
detail.
the
present
The
purpose
treatment begins with an elementary analysis of the latter effect, which
This study is followed by
arises from purely geometric considerations.
an outline of an analytical procedure (based on nuclear-resonance param
eters) for computing the integrals which appear in the expression for the
resonance-escape probability [see, for example, (10.25) and (10.37)].
Application is made first to homogeneous systems, and the results are
later extended to heterogeneous systems.
b. Geometric Effect.
The influence of the geometric effect on the
resonance-escape probability is demonstrated for the case of slab geom
Consider therefore an infinite heterogeneous array of the type
etry.
Previously, we considered the diffusion of thermal
shown in Fig. 10.3.
in
with an assumed uniform source of thermal
this
system
neutrons
been
-DM
, |
4>u{x,u)
q-x
4>m(x,u)
2<
vytf(-K,M)
- ^ qtt{x,u)
9Q)
conditions
= 0
(2)
j^\a,u)
af^(a,u)
(10.100)
F(x) qM(u)
(10.101)
Sec. 0.3a for discussion of this point in connection with the extrapolated
boundary condition.
HETEROGENEOUS REACTORS
sec. 10.3]
663
F"{x) + \*F(x)
and the solutions
F(x)
qu(u)
= cos
=
g0
\(b
exp
x)
(10.102)
(10.103)
u\
qM(xfl)
F(x)
= F(x)q0
qM(0)
(10.104)
The quantity
is determined
= 2
tan X(6
(y^T^)
(10.105)
a)
In the case of thick fuel plates, a may be computed from diffusion theory,
and is given by (5.308); for thin plates the result (5.331) may be used.
The calculation of the resonance-escape probability proceeds from the
If we compute the total number of neutrons
expression for qm{x,u).
reaching thermal throughout the entire cell, and divide this by the total
number of neutrons entering the slowing-down process, we obtain
Pw(tO
[%M(x,uth)dx
- ^s
qM(x,0)dx
Ja
This result is to
= exp
|-)
X2
UmK
S(.vn
Mf'~
fjfZJ
uJ
(10.106)
- n ik?
DM{b/a
10.107)
1)
= exp
"^(.w)^
- a/b)
""'}
(10-108)
REACTOR ANALYSIS
064
[chap. 10
A comparison of this result with (10.106) shows the effect of the scale of
the heterogeneity on the resonance-escape probability (see Fig. 10.12).
The principal conclusions to be drawn are: (1) as the cell size becomes
larger the resonance-escape probability increases monotonically from its
to that
in the equivalent homogeneous systern; and (2) for large cells the
'
Piwt(",h)
resonance-escape probability
proaches a limiting value.
It
Phom("",)
is
the elementary
present
8
Fio.
ap-
^_
calculation
was
selected
c.
F(u)
fu
F(u')
r-^ /
+
where
is
2(F)(V)
TJ
is denned by (10.16).
e'- du'
/
If Eq.
W^r^-V-d*'
(10.109) is solved for
(10. 109)
F(u), then
formula (10.18) to
This calculation may be
compute the resonance-escape probability.
carried out with whatever precision is consistent with the requirement
the resulting
HETEROGENEOUS REACTORS
sec. 10.3]
665
In this development
on p.
to F.
precision.
with
a moderator
F()
and it follows
(10.110)
^5 /
F{u>)
(see
Fig. 4.21).
(10.111)
e'- du' +
or
integral.
Similar approximations
we represent
heterogeneous systems.
the heterogeneous system by a unit cell of volume V consisting of a seg
ment V r which is occupied by the fuel material and a segment V m by
assumption,
when: (1) the fuel elements (represented by Vr) are closely packed com
pared to the mean free path \\m in the moderator material, or (2) when
the spacing of the elements is of the order of the slowing-down distance
(i-jy) in the moderator, or (3) when the characteristic dimension of the
fuel elements is much larger than
1 L. W. Nordheim,
"The Theory of Resonance Absorptions,"
GAMD-638, San Diego, Calif., Jan. 6, 1959.
General Atomics,
666
ANALYSIS
REACTOR
[CHAP.
10
the total number of neutrons slowing down in the cell Qo may be approxi
mated by the asymptotic form; thus we take
*s(U)isrr
where one neutron
We define
with
is introduced
Z7P =
N pa'-p,
and
Qo
(io.ii2)
i,2>V, + ZmV^Vm
V.
(10.113)
4>m
gpj ft [1
- e-*nHs)
ds
(10.1 14)
*,() Zj"()
\I
aF
/
Ju-\n
,
**(')
du' + 2<r\u)
(10.115)
P.
4>t,{u)
(10.116)
in a Close-
HETEROGENEOUS REACTORS
sec. 10.3]
667
and using
FF(u)
m 2<"()
(10.117)
"
lT~7
(ian8)
e""u du' +
If we take also
obtained by setting FF(u) = constant.
be consistent, then it is easily shown that
FV(u)~J^L
and
{2.7
4>F(u)
at
range is
^<r,^to to
=L
2.7
(10.119)
as is to be expected.
F(u)
(NR) approximation:
NR for homogeneous
(10.110)
The solution is
FF{u) =
JLL
{2.7
seen to be
Wlp,(u)
(1
NR for
heterogeneous
(10.120)
F{u) = -^-rT-^TWTT-^
[2,(u)
2/(u)]
we use
(10.121)
REACTOR ANALYSIS
668
similar approximation
=
F,{u)
Z.V
Pm
_____
NRIA for
aIFHu)
-(i-P._)^y
10
-r|i
F
[CHAP.
heterogeneous
(10.122)
resonance integral.
is
2a(w) du
F(u)
The effective resonance integral may
J_
Substitution into
(10.110).
given in (10.20); in the present
is
^>(M) du
fuih
&P J0
j +
_- KFKu)
Op
(10.124)
+ <#'()]
#>()
<P(u)
11
for homogeneous
,
,Tr>
/ml
k/>()
vfKu) du
a<f\u)]
(10.12o)*
is
with
is
-t:
- 2<(u)]
2(w) du
f[2((u)
Z_">
cri"(M)
_ 2i"'
jo
2<(w)
Jo
2/>(u)
2__u) du
*{u)NrMM,NM
(10.126)
HETEROGENEOUS REACTORS
sec. 10.3]
069
if
we neglect everywhere
= frmI =
Jo
2">
"
)Z
+ olP(y)/(rm
NRIA for
crm
<r
homogeneous
(10.127)*
(10.128)
[ FF(u) *<f\u)
du
M0
12QU
(10.129)*
*r(u)
"Jo
unit cell,
NRIA
tively.
P^
- Pe
(10.130)
P.
- ^~
9m
and
f[l- e-^r)]h(s)
ds
(10.131)
where we have used (10.114) for P,; note that the penetrability 9 is
related to the transmission coefficient a of Chap. 5 by [cf. Eq. (5.275)]
9 =
In general,
-a
(10.132)
as mentioned
- r+W1
K. M.
the definition is
(10-133)*
Neutron
REACTOR ANALYSIS
670
This formula is
[chap.
10
function when
Some values of
sX(tF) is either much greater or much less than unity.1
P*k are given in Table 10.3 for various elementary shapes, along with the
corresponding values of P. These results are also plotted in Fig. 10.13.
It is of interest to note that the dependence on geometry is not strong;
P*,. for spheres and cylinders are very close, with slabs differing only
The rational approximation is seen to be quite good.
slightly.
Table
10.3
a good
to the actual
0.04
0.1
0.2
0.3
0.5
1
2
3
5
approximation
10
Sphere
Cylinder
Slab
Rational
0.978
0.946
0.896
0.850
0.767
0.607
0.411
0.302
0.193
0.099
0.974
0.939
0.885
0.819
0.753
0.596
0.407
0.302
0.193
0.099
0.952
0.902
0.837
0.785
0.701
0.557
0.390
0.295
0.193
0.100
0.962
0.909
0.823
0.769
0.667
0.500
0.333
0.250
0.167
0.091
I=
UW
(1
- P-X'l
(10.134)
and
If
"
1 =
h [P^r
(1
"
(1-135)
Pe,c)<>1
=/Jo
+ &'t
'
NR for
NRIA
heterogeneous
approximation
(10.136)*
is seen to be
Nrf, f^P^Tdu
2.F jo
It
follows
1 The
op. cit.
/ +
PrfS"
by Dresner, Thesis,
10
Characteristic dimension
Fio. 10.13 The escape probability and the rational approximation as functions of the
characteristic dimension S2<1,,).
integral is [using the definition
f uth
* =
(F ) flu
Jo
of Homogeneous
e. Equivalence
and Heterogeneous
(10.138)*
Systems.
The
/u"
/
Jo
I
jo
If
<rjF)
du
[1
aP
[v!P +
quantity
homogeneous
(10.125)
heterogeneous
(10.136)
<" +
+ sVpyaF) du
1 + sZ'
we define a new
Consider
a* such
that
(10-139)
672
REACTOR
ANALYSIS
[CHAP.
10
a(F) du
"">
a/ok."
i +
Jo
air + <n
(10-14)*
in form to (10.125).
Thus we find that a hetero
geneous system is equivalent to a homogeneous system in the NR approx
imation if (tp for the homogeneous system is the same as the a* for the
which
is identical
It
heterogeneous system.
geneous systems with the same <r* have the same effective resonance
This equivalence relationship has been recognized by P. Bakintegral.
by L. Dresner.2 3
Similar relationships can
shi1 and
approximation.
3 =
be established
/""th
du
u
j _L_(f)/,
r/> du
1
we introduce
homogeneous
(10.127)
heterogeneous
(10.138)
<j%
(10.141)
then
(10.142)
uth
CT(^>du
is
approximation
(10.133)
was used in establishing the two equivalence relations (10.140) and
These relations are valid only so long as the width of
(10.143).
a
6.
L. Dresner, Thesis,
1,
If
+ 52/ >
NRIA
of the treat-
sec. 10.3]
ment
of energy
HETEROGENEOUS REACTORS
degradation
673
in neutron
relations are those arising from the geometric properties of the hetero
These become important when any of the conditions
geneous system.
imposed by the flat-flux assumption are not satisfied.
The equivalence relations may be summed up in terms of two very use
ful practical rules: (1) All fuel lumps containing the same fuel (absorber)
material, but with different amounts of other moderator materials, and
having the same value of <r*, have the same effective resonance integral
regardless of shape or size.
(2) The effective resonance integral for a
lump is the same as that for a homogeneous mixture of the moderator
and lump materials in which the potential-scattering
cross section per
fuel atom is equal to <r* [as given by (10.139)].
An interesting interpretation of rule 1 is that data on the resonance
integral for a series of fuel lumps of one material can be used to com
pute directly the value for a related series of lumps utilizing the same
"absorber" but in an entirely different geometry.
Thus, for example,
data on uranium metal rods can be used to predict the resonance integrals
for UO2 spheres. It is of considerable importance to note that these
rules do not depend on any knowledge whatever of the resonance param
eters. They can be used both for predictive purposes and as a check
on the theory.
f. Evaluation of Resonance Integrals Using Breit-Wigner Line Shape.
between neutron
and nucleus.
It
is shown
in a subsequent
analysis
674
ANALYSIS
REACTOR
that, in the
case of
stationary nuclei,
8 =
[CHAP.
8 is given
10
by
ny
(10.147)
* - Y^a
(1(U48)
with m and A denoting the mass of the neutron and nucleus, respectively.
and rn are the radiative capture and neutron widths, and
The symbols
I\
ff0
T m
r +
(10.149)
T7
Note that
plus scattering)
<ro is the peak value of the total (absorption
All the relations indicated here apply to the center-of-mass
resonance.
system of coordinates, and X, the reduced neutron wavelength, is given
by1
The symbol
equation
statistical
35
2i
=
2t(2m*S)
J'
(10.150)
h denotes
follows
9J =
where
2^
is the
2(21' +
(10-151)
1)
/'
Consider
1
It is
| (8 - Ei)
(10.152)
(10.153)
(Tpl
to the evaluation of
shown in Eq. (10.176) that in the case of stationary nuclei the relative energy=
E. There
^
fore, the relations (10.144) through (10.146) have the same form in (L) if the widths r
are also measured in (L).
* Note that
/'
-J for neutrons having zero orbital angular momentum
note
UiS8 and Th*", gj = I.
waves);
also
that
for
(s
E is
/'
sec. 10.3]
HETEROGENEOUS REACTORS
NR
If
approximation.
,' _
~
Jre, cp +
<r(l
675
+ s')]-Hr/2s) dx
+ x2)-1 + 2x(<r0b)Kl +
U.io*J
a2)"1
In the
case of narrow
resonances,
J-.
dx
0r2 +
2(0b)x
Sf
(1
N
0
for homogeneous
(10.155)*
(10.156)
CO
if
and
(10.157)*
is
Si^ST
(10.158)
b,
is
is
is
is
it
j3
is
This result
the so-called infinite-dilution effective resonance integral.
The sum of " over all resonances has been measured for many materials1
At the other extreme, when <3C
(see, for example, Table 10.1).
as often happens in the low-energy resonances of uranium and thorium
or other absorbers in fairly concentrated mixtures,
seen that Si
pro
roughly verified by experiment.2
portional to a\. This rule
We found previously that in the NR approximation the effective
resonance integral for heterogeneous systems could be written in the form
mathematically identical to (10.125) or (10.23); thus
(10.140) which
It
fol
=rr
NR for
+i - b)T
heterogeneous
(10.159)*
Si
is
op. cit.
J.
J.
A. C. G. Mitchell, L.
R. Pruett, and E. D. Nering,
Brown,
"Resonance Absorption of Uranium in Mixtures," Argonne Metallurgical Laboratory,
CP-1676, Mar. 31, 1944.
'
where
[l+J(l-b)J
It
2Ei
/S)
9*
REACTOR ANALYSIS
676
[chap.
where
(TO
or
|8
10
(10.160)
by
b*
(10.161)
When
/S*
<5C 1,
condition nearly
always
10.14 as a
function of
/3//3
t>= 1
4,
Fio.
10.14
Geometric dependence
56789
of resonance
10
integral.
for various b. It is seen that all the curves are asymptotic to (/3 //?)*.
Thus with increasing /3//8, the dependence of the effective resonance
integral
geneous systems.
The evaluation
procedure.
If
of
using the
the same
then
l<tm
HETEROGENEOUS REACTORS
SEC. 10.3]
677
formation
6i =
<r andand
trans
trans
X.
(i +
*2)
[i + VorX-r)(i + *')-]
(iai64)
Eq. (10.155).
Si =
=rrr
r[I+.
(<Tol y/(Tml
NRIA for
)]'
homogeneous
(10.165)*
Si = n j 7
NRIA for
'
7 m
heterogeneous
(10.166)*
The heterogeneous system formulas (10.159) and (10.166) for the con
tribution to S based on the NR and NRIA approximations may be used
to deduce the dependence of the effective resonance integral upon the
surface-to-mass ratio (AF/MF) of the fuel lump.
In the cases wherein
the NRIA formula applies, it frequently occurs that (<r0Ty/a*Y) ~ 1.
Then the form (10.166) is well approximated by
since by (10.141)
*^K^y~(^y
*
m Mfr,
=
(ioi6?)
(10-168)
r[i
is
If
is
p'P
+ *:/<>;
if
U238
and
2.
with
REACTOR ANALYSIS
678
r<7(F,i
ST
-M
<r*
by
(10.169)
thin,
s
is
f),(NR)
10
"i*
2> 1 ,
general oo/o'f'
[CHAP.
*(NR)~^Y
Thus for many situations
is
reaction-rate
relation
w(v) =
|ya-(|v
- V|)|v - V|2fl(V,0)
dV
(4.210)
for the cross sections in the resonance integrals in place of the natural
In the present treatment
line shape given by the Breit-Wigner formula.
and use the Breitwe base the calculation on the NR approximation
is
Wigner formulas for the a" which appears in (4.210) and the MaxwellBoltzmann distribution (4.213) to describe the motion of the nuclei.
This general problem has been examined by
number of investi
In
the
treatment
which
follows
we
draw
principally from the
gators.1
The calculation
analysis of L. Dresner.2
developed from an approxi
mate expression for the relative energy between the neutron and the
For this purpose we require the velocities w and W (for
nucleus.
neutron and nucleus, respectively) in the center-of-mass system of coordi
nates (C). These are [cf. Eqs. (4.6)]
w =
iTa(v_V)
rr~A(V~v)
(10171)
H. A. Bethe and G. Platzek, Phys. Rev., 61, 450 (1937); E. Creutz, H. Jupnik,
Appl. Phys., 26, 257-259 (1955); E. P. Wigner, E.
T. Snyder, and E. P. Wigner,
Creutz, H. Jupnik, and T. Snyder, J. Appl. Phys., 26, 260-270 (1955); E. Creutz, H.
Jupnik, T. Snyder, and E. P. Wigner, J. Appl. Phys., 26, 271-275 (1955); E. Creutz,
H. Jupnik, and E. P. Wigner, J. Appl. Phys., 26, 276-279 (1955); G. M. Roe, "The
Absorption of Neutrons in Doppler Broadened Resonances," KAPL-1241, 1954.
68-79 (1956); also, Thesis, op. tit.
L. Dresner, Nuclear Sci. Eng.,
1,
J.
sec. 10.3]
HETEROGENEOUS REACTORS
Thus the
and (10.147)]
iw! + pfW*
E, --
in,
(v
679
- V)2
Eqs. (4.9)
(10.172)
Now, if n0 denotes
where m* is the reduced mass defined in Eq. (10.148).
the cosine of the angle between the velocity vectors v and V, then the
expansion of (10.172) yields
Eni
j-p^- (E
+ imV*
- mvVno)
(10.173)
p2
Clearly,
Enl
fr.p-
l^V*
= 8
6 =
with
In.v*
(10.174)
2vVn0
- (2M8)U^o
E
(10.175)
(10.176)
Note that, if the nuclei were stationary, 6 would be the relative energy
This was mentioned previously in connection with
of the neutron.
Now, in most situations,
Eqs. (10.144) through (10.147).
(10.177)
(^)S~^1
Thus (10.175) may
be approximated
Eni
by
P =
[l
"
(10.178)
(2M8)FMo
We have also
~L)]1
=* v
Fmo
(10.179)
|v-V|~y
Then, Eq. (4.210) reduces to
r(i>)
Jv<r-(|v
- V|) 9JJ(V,0)
(10.180)
dV
(10.181)
Enl~&-
(2M.S)*F,
(10.182)
REACTOR ANALYSIS
680
[CHAP.
10
(i^rj
^Mf] *
-p
[
/:.
obtain
we
(10.184)
is
2,(2^)l
(1-185)
f
P
- Enly]
4/crM.s
Generally speaking,
(^i\
('
(&)
M(S
Only when
Ex
relation
dEnl
(10.186)
is
is
is
is
of an absorption resonance
near Ei
there
significant
contribution to the integrand of (10.186) therefore, as before, we assume
independent of energy and set its value to be that at the
that aorT/T
If we call
resonance.
appreciably
(8
'
- Et)
- Ei)
ym (Enl
(10.187)
y)t]
{x
exp
dy
- *
It
is
This function
rh
7$t
/_"
is
(10-188>*
the counterpart
HETEROGENEOUS REACTORS
sec. 10.3]
lira
681
*(,x)
=
1
-r
* 0
(10.190)
xz
ence is
8
^(#,x) dx =
t.
lim
is
which
(10.193)
x(*,i)
x(,x)
2r
^r
the generalization
be used to obtain
approximations
which follows
is
tion
(10.194)
NRIA
(10192)*
0,
= 2x*(tf,x)
dy
Note that
x(*,x)
"
(*
(10.191)*
/_".
&
^4=
rrp exp "
0.
see,
J.
Expressions of the form (10.188) can also be obtained for the resonance
These are easily shown to be
and interference scattering cross sections.
REACTOR
682
ANALYSIS
[CHAP.
10
Now
[We have used here also (10.149) for T and (10.156) for /3.]
=
r dx/2&, and if as before [refer to Eq. (10.155)] we consider
dE/8
narrow resonances and ignore the energy dependence of T and 8 over
the range of integration and simply evaluate them at Eit then (10.195)
may be approximated
S*-
by
2E, }-2Ei,r
&
E<
Jo P + +(0,x)
uulwo;
In this
case
=
<70l
(10.198)
WA- f;*4&
There
(10.199)*
then
Si = -1=--
J{d,fi)
Si = -TT- /(#,$)
in
NR
in
NRIA
approximation
(10.200)*
approximation
sec. 10.3]
HETEROGENEOUS REACTORS
683
The function J(d,&) has been tabulated and graphed by Dresner,1 and
analytical expressions are given by Roe.2-5 Dresner also lists some of
the more important properties of the function and various series expan
sions suitable for computation.
Of particular interest here is the limiting
>L. Dresner, Nuclear Sci. Eng., 1, 68-79 (1956); also Thesis, op. cit.
' Roe, op. cit.
' See also Sampson and Chernick, op. cit.
* VVigner, op. cit., BNL-433.
6 Sampson and Chernick, op. cit.
ANALYSIS
REACTOR
684
[CHAP.
10
portions
ioJ,
5
1
-< ==0.1
-0.2
,1 =
*.
1.0'
i\\X
ll
4
I
6
10
*(where(3
Fio.
10.15
I
12
I
14
I
16
I
18
= 10"6 X 2*)
materials
U238
in Hexagonal
HETEROGENEOUS REACTORS
sec. 10.3]
G85
will remain
and
and
approximation
Chernick show that the NR formula is applicable to all but the lowest
several resonances in the case of the pure metals U238 and Th232.
How
the
neutron
NRIA
Specifically,'-6
these are,
for
U238,
66.3, 103.5, and 192 ev, and for Th232, those at 22.0, 27.7, 70.1, and 174 ev.
In the
practical width the oxygen cross section cannot compete with the reso
nance cross section, and so the approximations appropriate to the pure
metal apply to the oxides as well.6
2.
REACTOR ANALYSIS
68(5
[CHAP.
10
it
is possible1
resonances
0<
(where
=i
material).
NR
and
NRIA
formulas
and various
resonances the more precise versions of the two models are found to yield
similar results.7 Such a comparison is shown in Fig. 10.16 for U238.
1 Chernick
*
and
2.
Mixtures," Proceedings
of the Brookhaven Conference on Resonance Absorption of Neutrons in Nuclear
Reactors, BNL-433, pp. 103-109, 1956.
3 Sampson and Chernick, op. cit., pp. 236-240.
4 K. T. Spinney, J. Nuclear Energy, 6, 53
(1957).
5 Chernick and Vernon, op. cit.
'A. M. Weinberg and E. P. Wigner, "A Second-order Correction to the Usual
Resonance Escape Formula," Proceedings of the Brookhaven Conference on Reso
nance Absorption of Neutrons in Nuclear Reactors, BNL-433, pp. 125-126, 1956.
W. Rothenstein, Nuclear Set. Eng., 7, 162-171
(1960).
in Homogeneous
HETEROGENEOUS REACTORS
sec. 10.3]
687
interference term (which is important for thick lumps1) ; when this term
is negligible, however, the extensive compilations of the function J(&,(i)
are generally and directly applicable.
If
!L JM T
'-w
JM 7
(ia201)
where Pn has the value of or/a0 at the resonance En] the averaged param
eters I) and
are discussed by Dresner.6 The approximation
sign
in the range of interest, T ~ 2F7, or greater.
applies when
I\
rI\;
Also,
that
1%,
'^s^ (\ ^ ^a^
En
FT is independent
(10 202)
of energy and
like E.
resonance range.
6.5g.
(1955).
(10.203)*
REACTOR ANALYSIS
688
With
[CHAP.
10
written
* = ay>(ifc)
* + ^ /" Jim f
(io.204)*
=i
where the function di is to be selected from the two forms given in (10.200)
according to the characteristics of the particular resonance represented
by the index t. This general form has been used by Dresner to compute
the resonance integrals for various homogeneous mixtures of both U238
and Th232 in moderator.
His results are summarized in Table 10.4 for
Table
Nucleus
Calculated Contributions
Integrals of U"8
10.4
T, K
TJHB
barns
Resolved
Unresolved
resonances
resonances
7.9
20
200
20
74.
8.1
80
17.
200
27.
86.
2,000
Th"
12.
2,000
nances.
the
Figure
3.0
6.6
9.4
12.
33.
2.2
6.3
14.
20
4.0
80
8.9
4.0
8.6
200
15.
12.
2,000
42.
20.
system temperatures
computation
11 .
3.7
20
200
300
1.7
4.8
24.
2,000
300
and
NR
10.17
of
0K and T
l/v
1.4
1.4
1.4
1.4
Effective resonance
integral, barns
11.0
30
86.
12.5
14
25.
1.4
1.4
38.
99.
3.5
3.5
3.5
3.5
3.5
3.5
3.5
22.
51.
11.5
21.0
31.
66.
= 300K.
In this particular
iAF/MF)*.
The models developed in this section are especially useful for under
the principal features of resonance absorption, and by careful
standing
choice of formulas,
J.
HETEROGENEOUS REACTORS
sec. 10.3]
689
60 1
0.4
0.8
(AFIMr)
Fig.
10.17
1.2
2, (cm2/gm)
i. The Standard
tematic
Formula.
and consistent
1.6
2.0
V2
development
from
a common
a sys
of
useful
basis
expressions
ment were the flat flux assumption and the application of the rational
This general treat
approximation for the fuel-lump escape probability.
ment represents the more recent approach to the problem of evaluating
resonance integrals.
It is expected that in time, as more precise measure
ments are obtained of the resonance parameters, there will be greater
incentive for further development of these general methods and models
in order to achieve wider applicability and accuracy.
Although
REACTOR ANALYSIS
690
[CHAP.
10
in understanding
For this
is given
by
(10.135)
If
we separate the integrand into its two terms as written above, then
comparison with Eq. (10.37) reveals that we can identify the first integral
as
(10.205)
and the second as
(10.206)
We have used here Eq. (10.131) for P.K, and MP = VFp'F.
When these
expressions are compared with (10.38) and (10.39), it is seen that the
NR approximation for $ leads rather easily to a statement in the standard
form.
Note, however, that because of the flat flux assumption the
present derivation does not contain the resonance disadvantage factor1'2
This quantity is customarily computed using a one-velocity model
fr.
to represent the entire fast-neutron population.
that this point of view is crude and somewhat
hand, when used with the
It
is well recognized
unclear.
On the one
NR
It
formula
is
A discussion of the
resonance
Theory
Chicago, 1958.
'See also S. Glaastone and M. C. Edlund, "The Elements of Nuclear Reactor
Theory," pp. 258-263, D. Van Nostrand Company, Inc., Princeton, N.J., 1952; and
Sampson and Chernick, op. cit.
3 Chernick and Vernon, op.
cit.
HETEROGENEOUS REACTORS
sec. 10.3]
691
the factor AF/MP and through <P; thus the surface term
proportional to Ay/ My, as implied in the construction of the
Although this dependency was known in the early
standard formula.
days of the technology, the (P factor fell into disuse; it was reintroduced
both through
is not really
resonance
l/E
lump.
1
E. P. Wigner,
J.
J.
Chernick,
Confer
Physics of Reactor Design, vol. 5,
; see also
the work of
the International
092
REACTOR ANALYSIS
[chap. 10
This point of view12 places emphasis on the volume term, and treats
the surface contributions as a secondary effect. Although this approach
is useful for handling most of the resonances in the important fuels, it is
inadequate when dealing with the low-lying broad resonances wherein the
volume term actually makes a relatively small contribution.
was recognized by the early investigators, but its importance
This point
was under
estimated
phenomenon
and treat
the
10.4
a. Significance
Fast Effect
effect
in
For further discussion of the arguments leading to the use of the standard formula,
Weinberg and Wigner, "The Physical Theory of Neutron Chain Reactors,"
pp. 681-689.
* Also, Sampson and Chernick, op. cit.
3 Gurevich and Pomeranehouk, op. oil.; see also Sampson and Chernick, op. cit.,
see
p. 247.
sec. 10.4]
U!",
HETEROGENEOUS REACTORS
693
and Pu239), and second there are the fast fissions which
U238 and Th232, which have fission thresholds
in the Mev range. Once the fast-fission threshold1 has been exceeded,
these sources can become an important factor in the over-all fission rate.
as
U236,
In the
effects.
b. Cascade Effects on the Fast-fission Factor.
Fast neutrons appear
in the fuel lump as a result of both fast and thermal fissions.
This supply
is continually depleted, however, by two processes which reduce the
energy of the fast neutrons to values below the fast-fission threshold.
Once so degraded, a neutron can no longer contribute to the fast effect.
The first removal process is the inelastic scattering in the fuel lump.
Since the inelastic-scattering cross section is a sizable fraction of the total
cross section in the Mev range, this process is a powerful loss mechanism.
Approximately
U*".
REACTOR
G94
ANALYSIS
[CHAP.
(P
It
10
(10.207)
is understood
initiated by
the function
(1)
(2)
a collision
i:
of type
<$*
2<Pn
2 = Zy +
2, +
2in
+ 2.
(10.208)
thus,
for
("2/
neutron
S-)
collision
each first-generation
<?i
(10.209)
cfP!
is
fast neutrons
y2' + 2'
(10.210)
The first generation also suffers some losses by escapes, inelastic collisions,
Of these,
certain fraction enter
and radiative captures, as noted above.
2in
process:
1
the slowing-down
1
inelastic
~~
process
sec. 10.4]
HETEROGENEOUS REACTORS
695
Fast-effect
Fast neutrons
available
Neutrons
participating
in cascade
c(P,
C(Pi
e3>i(P,
c-'(P,
. . . <P_,
Cascade
Neutrons entering
slowing-down process
C(Pi(Pa<P,
c"(P, .
Histories
Genera
tion
:'>,,
c-'(P,
(P.
. . .
- ft +
-
|\
>_,
a>,
|\
<ft
Y<Paj
(PB_,
Y <*]'+
- +y +
(5>t
(Pn
(Pnj
+ c"-'(Pi
+C(P1Jl
i-ff*!
[^1
t=
(10.212)
The collision probabilities (P given here vary from one generation to the
next since the spatial distribution of the fast-fission neutrons changes
The original, or source, distribution F0(r) of
with successive cascades.
is
depressed toward
the first generation varies as the thermal flux, which
the center of the fuel lump.
The spatial form of the next generation
is
Fi(r)
is
distribution becomes more peaked toward the center of the lump. The
asymptotic form corresponding to n does not differ much, however,
10.5
""-J
fast-fission
REACTOR
696
or 2.
ANALYSIS
Figure
[CHAP.
10.18 shows a
10
typical sketch of
It may be
the spatial forms Fn of the first few fast-fission generations.
from these remarks that (Pi < (P < 1, but because of the
small variation in the Fn for n > 0, a reasonable approximation is to take
=
Conceivably it might be
some fixed number independent of n.
3>n
concluded
(P0)
better to take the actual (Pi for the first generation and then some other
number for all n > 1 ; but this much care may not be entirely justified.
If
is
(Po
readily
summed, giving
2,(P0
2(1
c(P0)
(10.21.3)*
It
if
is clear
<
c(P0
1.
c. Computation
distribution of suc
of fast-fission neu
trons.
of Fast-neutron
(Po
= /(Po(m) 3(u) du
(10.214)
it is understood that the integration extends down to the fastA consistent definition for the various cross sections
fission threshold.
where
2( is then
If it
is assumed that
I -s/^ *
(P0
is energy independent,
f 2,(u)
2,-
then
, , ,
(10.216)
Values for
arct are
2t
(2/ +
27)
(10.217)
HETEROGENEOUS REACTORS
sec. 10.4]
697
For the total cross section 2 of these fast neutrons we take the transport
This choice is suggested by certain physical features
cross section S,r.
of the problem, the most important being the strong forward bias of
scattered fast neutrons.
Because of this effect, the fast neutrons have a
The net
greater tendency to move outward from their point of origin.
result is as though the supporting
what larger than its actual value. Moreover, in calculating the func
tions Fn and (P we will assume that the scattering process, as well as the
By combining this assumption with the use of the
others, is isotropic.
transport cross section based on a value of
appropriate to the forwardscattering tendency of the nucleus in question,1 it is expected that a
fairly good representation of the fast-neutron "diffusion" in the fuel lump
can be achieved.
Thus we take
S = 2tr = 2V + 2, +
2in
+ 2.
(10.218)
Table
Table
10.6
UM9
a barns
"Total"
0.04
0.29
2.47
1.5
4.3
<rtr
dr' about
r',
e-2t,|r-r'|
(10.219)
I1
Jv,
|reTr
dr
Jvr
<P,+1
r'
is
Since Fn
normalized, to compute (P+i we simply integrate the above
and
thus
expression over all
S.
J.
Note that if the relation (10.217) is used to obtain 2in then Eq. (10.218),
The approach
along with a computed 2tr, can be used to determine 2,.
outlined here for estimating the fast-neutron cross sections is found to
give reasonably good agreement with experiment in the case of U!M.
REACTOR ANALYSIS
098
It
[chap. 10
follows that
fWi(r)
The functions Fn may
- r'|*
2*
/"
FnCrQe-^lr-
4r<JWi
Jvr
[r
(10.220)
of F0.
Some of these
10.5
holes.
will
HETEROGENEOUS REACTORS
sec. 10.5]
699
sion length L of the reactor materials [computed from Eq. (10.14)] with
out the holes through the equation
L%
where
vH
L*(l + vy
(10.221)
\"s
(10.222)
and Vh denotes the total volume of holes in the reactor and V s the
volume occupied by the solid materials.
It has been demonstrated by D. J. Behrens1 that the estimate given
by (10.221) is the first-order isotropic correction term for small holes in
of the holes.
obtained in this way are used to compute the migration area and thence
the multiplication constant of the reactor.
a. Corrections to the Diffusion Length.
The general relations obtained
where
Vg,
LH, and
hydraulic radius of the holes (twice the volume of the hole divided by
the surface area), \m is the mean free path in the moderator,' and
G = s2/(s)2 for the holes; here denotes the average passage (chord)
The
length through a hole and s2 the average square passage length.
quantity G has been computed by Behrens for various shapes of holes,
It is of interest to note that
and the results are given in Table 10.7.
=
1 for the annular geometry corresponds to a twothe limiting case x
This is an unfavorable arrangement
dimensional gap in the reactor.
from the standpoint of neutron leakage and should in general be avoided.
Such a gap, however, has been incorporated in the design of the BrookThe effect
haven graphite reactor as a part of the gas cooling circuit.
of such gaps on criticality is discussed by Kaplan and Chernick in the
reference paper.
D.
'I.
May
be estimated from \m
REACTOR ANALYSIS
700
[chap.
10
When both the size and spacing of the holes are small compared
to the mean free path in the moderator, the expression (10.223) is well
approximated by (10.221); moreover, it may be shown that in any event
Table
The Quantity
10.7
Geometry
11 25
Sphere
Infinite cylindrical hole with:
Circular cross section
Hexagonal cross section
Square cross section
Equilateral-triangle cross section
Cylindrical annulus
Spherical annulus
x
1.333
1.397
1 .487
1 .648
1
.333f,(i)J
1 125F
,(x) t
F.(x)
0.1
0.2
0.3
0.4
0.5
1.001
the value of
F.(x)
F.(x)
Fc(x)
0.6
0.7
1.11
1 .0258
1 .005
1 .0536
0.8
1.2023
1.2604
1.3422
1 .016
1 .0839
0.9
1.18
1.27
1.43
1.04
1.07
1 .1178
1 .0
00
oo
is
at least
as large as
Thus, properly
jjf
1 .53
1 . 1567
>
(1+ vy
(10.224)
In the event that the holes are far apart, Eq. (10.223) may
(1
mated by
vHy +
be
approxi
(10.225)
Some representative values of L2/L2 are given in Table 10.8 for various
hole sizes and spacings so as to display the relative accuracy of Eq.
(10.221).
is
is
is
directionally non
When the distribution of cavities in the reactor
uniform, some modification must be made in the general result (10.223).
affected by the directional
The only term in this expression which
the last, and this dependence appears through
properties of the cavities
particular type of anisotropy which
frequently
the shape factor G.
in the form of parallel passages of cooling
encountered in reactor design
In this situation the corrected diffusion length in the direction
holes.
is
is
Shapes
HETEROGENEOUS REACTORS
SEC. 10.5]
701
of the passage axes will be different from that in the direction perpendic
ular to the passages.
Behrens has shown that in this case the appropriate
expressions for the diffusion lengths in the two directions are given by
=
2rt,/vH\M
v%
+ 2v +
v%
2oB
exp (2ri,/v\M)
2rH/vH\si
{2rH/v\M)
exp
3GrHvn
10.8
3Grv
2\M
shape.
(10.226)
(10.227)
The quantity G
\.
^\Lattice
spacing
Hole
diameter
1.016
1.000
1.066
1.000
1.200
1.000
1.548
1.000
1.019
1.003
1.079
1.012
1.234
1.028
1.609
1.039
1.026
1.010
1.108
1.039
1.319
1.099
1.815
1.172
1.047
1.194
1.066
1.574
1.312
2.466
1.593
10
^\
rB/\u
th/am = $
r///X. =
tbI\u
Lower
figures
1.031
computed from
(1
+ vh)*-
be extended
to account
to the diffusion
length
REACTOR ANALYSIS
702
neutron energy.
[chap.
10
age is to lower
the initial energy of a fast neutron entering the slo wing-down process;
this reduction results from inelastic-scattering collisions in the lump.
An estimate of the combined effect on the age of fuel lumps and cavities
flWCu^tO]
(PioWCUthjO)
(PiaTif(uu,;,.)]
(l
yA f1
(2
^f)
jf)
v"]
<Pi)Tjtf(wth;0)
(10-228)
v]
Tt,.. = [(1
(2
[(1
[l
(l
rw
may be obtained
passages
(10-229)
gives
.,)
(10.230)
^p?)
is
definitions
it
with
-P
^n
(1
it
is
quantity (Pin
the probability that
fission neutron makes an inelastic
collision before escaping from the fuel lump;
may be estimated by
is
for the fact that not all neutrons enter the slowing-down process at
u = 0. The first term in the bracket
a population-weighted
age for
=
The second term, on the other
those neutrons which do start at u
0.
hand, accounts for the age of fission neutrons which begin the slowingdown process in the moderator after they have first suffered an inelasticThese substantially lower-energy
scattering collision in the fuel lumps.
neutrons enter the slowing-down process at u = win and thereby acquire
Note that the age r,tf(uth;?*tn)
ages TM(uth',uiD) different from Tjtf(Mth;0).
is
weighted
inelastic-scattering
(Pin
which
have experienced
collisions.
a
first-order
The second bracket in Eqs. (10.228) and (10.229) gives
correction to the moderator density to account for the free ride acquired
This factor corresponds to
fuel lump.
by the neutron in traversing
the isotropic term (10.221) derived for cavities.
The last bracket in
the two expressions for t takes into account the presence of holes in the
form of parallel passages and includes the directional effects discussed
a
sec.
HETEROGENEOUS REACTORS
10.5]
703
These factors
are approximations
to expressions (10.226)
and (10.227) given in the form of (10.225).
c. Migration Area and Multiplication Constant.
Estimates for the
migration area which include streaming-effect corrections for cavities
previously.
and fuel lumps on the thermal-diffusion length and the age may be
obtained from the general relations (10.226) through (10.229).
The
appropriate expressions for A/2 are seen to be
M2axial =
-r
iwA
Taxial
1/2
iw
trana
= rTtrana
AI
i
I
T 2
^axial
7 2
/Ifi
OQ1"\
"traoe
It
An idea
of the reactor.
as to how this
anisotropy might
in the calculation
migration-area
relation for this system is
-DlhV2*th + 2?*th
V2</Hh + B2tb
or
be included
is suggested by the
The neutron-balance
fc.ZLWu,
(6.87)
= 0
with
M1
+ V-
Tlh
+ (kx
l)th
MlM
*J
M?_
(10.232)
system,
+ (*.
= 0
L)*
= 0
(10.233)*
is seen to be
A;
i + mLmbi+"mi.u.{BI + bd
(10-234)
where Bx, Bv, and Bt are the bucklings in the x, y, and z directions, with
x denoting the axis parallel to the gas passages and y and z the axes
perpendicular to the passages.
REACTOR ANALYSIS
704
10.6
[CHAP.
10
Such a model1 has been developed by Feinberg2 and Galanin, and in this
approach the general heterogeneous reactor is treated as a collection of
matrix of moderator
material.
The fast
(fission) neutrons produced by these sources slow down in the moderator,
and it is assumed that the resulting spatial distribution of thermal neu
a
application
is,
lattices of fuel lumps, each lattice having a different module and size and
nuclear property of the fuel lumps.
The practical interest in this
of course, the use of "seeded
cores" in heterogeneous
5,
of
First described by E. Teller, " Influence of the Lattice Structure on the Exponential
Pile Experiment," Axgonne Metallurgical Laboratory, CP-165, 1942.
S. M. Feinberg, "Heterogeneous Methods for Calculating Reactors: Survey of
Results and Comparison with Experiment," Proc. Intern. Conf. Peaceful Uses
Alomic
Energy, Geneva, vol.
1955, P/669, pp. 484-500.
A. D. Galanin, "Critical Size of Heterogeneous Reactor with Small Number of
Rods," Proc. Intern. Conf. Peaceful Uses
Alomic Energy, Geneva, vol.
1955, P/663,
pp. 462-465.
5,
of
trons may
sec. 10.6]
HETEROGENEOUS REACTORS
705
arrays.
as
b.
Neutron-balance Equation.
described
be favored
and the theory is applicable even in the case of quite large lumps.
The
second assumption requires that the absorption in the moderator be low
and that the diffusion length for the moderator material be small in
to the lattice spacing.
The third assumption is essentially
convenience, and its applicability is determined primar
ily by the accuracy of the model by which the thermal constant is com
A treat
The last assumption is made to simplify the analysis.
puted.
ment in which resonance absorption is included is given by Galanin.2
With the above assumptions as a model, consider an infinite hetero
comparison
a computational
(10.235)
*Ibid.
REACTOR ANALYSIS
706
[chap. 10
-OtfV'Mr)
20.w(r)
S(r)
Arod(r
rk)
flux at a
(10.236)
>
ik)
is ignored
-r,2TPFJ(ik)
S/(r) =
(10.237)
is the total number of fast neutrons produced per unit time per unit
But, by (10.235),
length of rod from fissions in rod rt.
7*
(10.238)*
a
(r \
<PM(Ik)
where \i(Tt) denotes the thermal flux at the surface of the rod.
Note
that this expression is for unit length of rod and that the proper units of
Substitution into (10.237) yields
7* are length squared.
Sf(tk) =
im<Mrt)
(10.239)
volume due to a fast source producing one neutron per unit time at r*.
The function S(r) is then simply the sum of such contributions from all
the rods; thus, we compute
S(r)
ss Vyk
y q(i;Tk)
(10.240)
<pM(rk)
tj
the two-dimensional
represented
sec. 10.6]
HETEROGENEOUS REACTORS
An appropriate relation
^rod(r- r*) =
is seen to be (cf.
- 2TPFJ(Tk)
(r
Prob.
r*) = 7*
6.4)
^ 5(r
707
r)
(Mr*)
(10.241)
2u(T)
= V7k
9(r;r*)
4>u(tk)
7a
(r
The substitution of
r)
*(r)
(10.242)*
9(r;r*) =
(10.243)
to
be
ff{<Mr)}
ff *u
*(s)
(')*** dr
(10.244)
ff exp[-sV
2 J J
V_
+ is (rt
s' + x2.,
- r)] ds _
"
["
2tJ0
se-'r ds
s' + x^Jo
2'
pf,.(rt_r)
(10.246)
where s = s(s,6) and 6 is the angle between the vectors s and z, with
z = r* r.
The 6 integration is recognized as the definition of the
REACTOR ANALYSIS
708
10
[CHAP.
&,""tde
00
2tJ0(sz)
similar result
"^T^iT
Jo
JJ
(10.247)
;
d8 =
J**
eW*-*)
J*'
zeroth-order
thus,
(10-248)
se~'',Jo(s\I
is
- n\)ds
<r
f"
yh
a.
is
is
the last step follows from an integral expression for the Ko function.8
If
substituted into (10.246) and the result, along with (10.248),
(10.247)
used in the inversion of (10.245), the solution to the flux
found to be
(10.249)*
It
will
be convenient
for
use in subsequent
2r
,-/<r
/3
v2
XmU
9r
It
txJ,)
^]*
(10.251)
K^)m |o-exp[-(S +
^)]|
(10.252)
G(r)-^/;p[-(.
where we have used the fact that
- r*|V4*L|,)]
(10.253)
[-
|r
exp
(a
/,"-/.")
"A
Press,
HETEROGENEOUS REACTORS
sec. 10.6]
709
with
co
H(\*
DM/Z(aM).
"unit cell."
Equation (10.253)
- r*D<Mr*)
(10.255)*
yiw
(10.256)
7*
Thermal Constant.
It was noted that the thermal constant could
be computed with various degrees of accuracy, ranging from an estimate
Since
based on diffusion theory to one based on the transport relations.
the diffusion-theory estimate will suffice in many instances, the appropri
The compu
ate relations will be developed on the basis of this model.
tation is straightforward and utilizes the unit-cell concept in the usual
In general the relations (5.349) through (5.354) will be required.
way.
By applying the continuity conditions for both the net current and the
flux at the surface of the fuel rod, it is necessary to consider only the
This is given by the usual differential
flux distribution on the inside.
In cylindrical geometry the solution to this equation
equation (5.350).
c.
is seen to be
Mp)
= AU{xFP)
(10.257)
wherein we have applied the requirement that the flux be finite within
The net current is obtained from (10.257); thus
the rod (5.354).
J(pf)
DF'F(pF)
ADFxFr0(xFPf.)
toefPUxrPr)
Io(xFpF)xF
(1Q 258)
It
F by (10.88).
(10.259)
d. Criticality Condition.
In the case of the infinite heterogeneous
lattice it is clear that there can be no gross spatial variations in the
These local vari
neutron flux, only the local variations between lumps.
REACTOR ANALYSIS
710
[CHAP.
10
relation
^ #(r*)
(10.260)
where r* now denotes the distance between some reference rod and any
This result is in fact the criticality condition
other rod in the lattice.
for the infinite lattice, since given any three of the four basic parameters tj,
the lattice spacing, the rod size, and the reactor materials, Eq. (10.260)
will yield the value of the fourth which will make the system critical.
The result (10.255) may also be used to obtain the criticality condition
for the reactor with finite fuel lattice but with infinite reflector (com
boundary
conditions
H(\X
- v|)*jf()
(10.261)
A solution to
the thermal fluxes at the surfaces of the fuel rods >a/(^.).
this system exists if the determinant is identically zero. By imposing
this requirement the determinant reduces to an algebraic equation, and
As in the
the roots of this equation yield the condition for criticality.
case of the infinite lattice, all four of the basic parameters will be involved
in this equation, and by specifying any three, the determinant will yield
the fourth. The calculational procedure involved here is demonstrated
in a subsequent analysis, and an example of a four-rod reactor is com
puted in detail.
In as much as the homogenie. Proof of the Homogenization Method.
zation method based on the unit-cell model is entirely adequate for com
puting most heterogeneous-reactor configurations, it is of interest to
determine the error acquired in using the method and to delineate the
To study this problem we postulate
conditions under which it is valid.
a reactor with a finite array of cylindrical fuel rods and an infinite reflec
tor composed of the same material as used for the moderator in the lattice.
We further specify that the lattice is a square array with spacing a.
sec. 10.6]
HETEKOC.EN'EOUS
REACTORS
711
The general solution for the thermal flux in this reactor may
as a linear combination
Thus
be given
Mr) =
elB-r
(10.262)
H(\i
- rt|)eiB-("-'>
(10.263)
identifiable
Fio.
of the latter.
Si
I 4r X
cxp
iBi{akl
~x) + iB*{akl
- [(at, - x)* +
- yY]
(afc2
(10.264)
The
ik = r*(afc,,afc,)
B = B(B,,B2)
summation
(10.265)
^ f(D
(2t)
go
SF{/} m
F(2irm)
(10-266)
- -
(afcl
xr~*
;
TS,(afc,
x)
= exp
L
/(*
fci
as given
10.19
rod.
4s
REACTOR ANALYSIS
712
[CHAP.
10
"
*)
e"P
(lr
/--
-5
follows that the double sum in (10.264) may be written in the form
+-(5-')-(5-),4]*
>
and
it
/(*0
">
The integration
with u = u(i,u,)
2L*(r) exp
(rt
- -
(rt
r)
^]
r)
exp
[iB
is
If
'^
-^X-K?)-'m
s,
The substitution of this result into Eq. (10.264), followed by the integra
tion over the range of the variable
yields the following double sum for
Si:
>[(*)"-(+)']
+
*r-L
if
(10.268)
us
(^i
where w = w(u>i,w2).
ing the new variable
is
(10.267)
the following
HETEROGENEOUS REACTORS
sec. 10.6]
713
4e-4"'"**l
^1
and
(10.271)
Si
is obtained
by taking the
(10-272)
- v|)e<-M
H(\X
(10.273)
(IkDc''8-
- -
(r*
- I)2t
(10-274)
(r>
r)
exp
eJir
[*
aS
s> =
/."
where y and X denote two rod positions and y X = ok; note that
r\ as X. The evaluation of (10.273) is a somewhat easier task than the
evaluation of (10.263).
The second sum 6'2 which appears in (10.263) is
0.
is
(10.270)
Xei(2x/o)mT
(2TWa)s[m +
m
(aB/2^)]2
As in the Si computation,
(10.275)
and
aB
s^rrxpr>
(10.276)
ANALYSIS
REACTOR
714
where
ci
with m = m(mi,m2)
(dfe)2
10
by
r+74F=
[CHAP.
"
(10.277)
coS[(2r/a)m.r]
(1027g)
m\ + ml
m2 =
0)
is omitted.
The comparison
YTTJTF*
and
co
=-
y~
Cl(0)
(10 279)
Finally, if
we call
+ c,(0)
(10.280)
(10.281)*
=
1
vfe-"*
+ L2B*
where L2 is defined by
(10.276);
then which
must
be
C^=44(lnS
where In 4x
1
t/3
= 1.48.
^-ln4T + l)
(1-282)
HETEROGENEOUS REACTORS
SEC. 10.6]
715
Other
model leads to the same result, but with $ in place of the 1.48.
lattice geometries have also been computed by Galanin, and the appropri
Another detail
ate results for Ci(0) are given in the reference paper.
of
should
take into
that should be mentioned is that the computation
c0
account the finite size of the fuel lumps; therefore, in applying Eq.
(10.256), one should use in place of Vu, V m = Fu irp%.
Finally, it is of interest to note that the quantity c defined by Eq.
(10.280) is essentially the disadvantage factor f used in unit-cell theory.
Since,
!-i [wJf
(10-283)
(10.284)
*i<
t)
t
1*
f;
- - Kt(xM\* -
(10.285)
dXx
= pf
whenX^M
rji'
2irL\,
j\xp(-s-
Then
H-fl-dx,
and
where
of function H
[*
- 2-4
3.
i>
is
(-
- Koid^u)
(102g6)
(10.287)
/3
REACTOR ANALYSIS
716
[CHAP.
10
j~ exp(-swhere
A(dx>)
d^fS)
pg
Ko(d^u)
A(dXll)
j\xp(-s - ^g^) ^
(10.288)
(10.289)
A(d) ~
|[(l
(^j
Next, the factor e~* is expanded in a Taylor series and the integration
carried out term by term; the first two terms yield the approximation
Ex
0e-w]
(10.290)
\.e~*
- xE^[x)\
n >
Finally
Kl*
ff(dx)
- ^4(dv)]
l)^o(rfx^A/)
(10.292)*
1,
H***
or
(10.293)
form [with
~
c0*x =
we write in the
.r
Co5x")*"
with
+ In
(o.5772
is
K0{x)
(10.294)
4>m(v):
is
is,
is
Here,
\H*
c05Xm|
(10.295)
if
is
(10.291)
-4rr
n 1
En(x)
sec. 10.6]
HETEROGENEOUS REACTORS
a (see
Fig.
717
The distances
10.20).
d\ between the
dii
^23
du
du
= a
(fu
d2i =
y/2
ggg)
An,
Hn
#12
Hu
(10.297)
o^rr [(^
iT^2a2
2^2"
- i?eMn]
l)Xo(axv) - ijeMu]
l)Ko(V2 ax*) - vefAu]
1)K0(pfxm)
(10.298)
Au
-4i2, and
The relation
Generated for wjivans (University of Florida) on 2015-09-22 17:23 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
coK
The appropriate
and
- MWn
+ bnY
46?2]
which is easily
= 0
(10.299)
a double root
bu + bu
Only one of
=
4>t,
since
an
they
must be identical by
Likewise the
symmetry requirements.
tfy
-(/)i
third root, corresponding to the negative sign in the second equation, must
be rejected since it would require that
= 2Only the root correspond-
these roots is
are rejected
<pt
(10.300)
2f>i2
= constant,
then
X/
/
/
4^
r*
FlG-
\s
10 '20
sw'
/ \s
//
~0~~
t
s\
\j>.
~
<
Four-rod
heterogeneous
REACTOR ANALYSIS
718
[CHAP.
10
#11
Hu + 2Hu
(10.301)
Co
which is precisely the second of equations (10.300), using the plus sign.
For computational convenience we write (10.301) in the form
ijW-u + *i8 +
2^12)
xii + Xn +
Hx = #x
where
2Xi2
Co
(10.302)
(10.303)
xxM
and
l^xji
a2gfl
2vli
[^o(d\xM)
x\n
A\\
a2
2^2 Ko(dixu)
(10.304)
Numerical Example.
cation
Lu
= 52 cm
PF = 3 cm
a28 = 2.75 barns
= 364 cm2
= 2.08
1726
Nv
= 0.0482
a?
1024
DF
nuclei/cm'
= 0.000355 cm"1
= 687 barns
= 0.702 cm
Ve
For
Co
we use [cf.
+ 2^)
^n+^,
+
Xll + Xl3
2X12 +
Co
= (a2
TpF)h{pFxr)'ZiaM'>
7-7
t^ttt
MnQfiR^
(10.306)
Equation (10.305) has been applied for the enrichment ratios Nn/Nu =
The
0.05, 0.10, 0.20 as well as for natural uranium N/Nv = 0.0071.
results are shown in Fig. 10.21.
Since the four-rod lattice is a near minimum
heterogeneous configu
which is incurred in
determine
the
error
it is of some interest to
To obtain a com
analyzing the system by the homogenization method.
ration,
sec. 10.6]
HETEROGENEOUS REACTORS
719
l-2i
30
40
Rod spacing a, cm
Fia.
10.21
Multiplication constant
reactor.
(v')'
The nuclear characteristics
of this
system are computed with the aid of the two-group model of Sec. 8.4.
The principal parameters required for this calculation are the infinitemedium multiplication constant A"'i", the roots of the characteristic equa
tion (8.171), and the constants ai and a2 (8.175), as well as the cylindrical
functions listed in Table 8.3. The infinite-medium multiplication con
stant is computed from
1
(10.308)
REACTOR ANALYSIS
720
present case
m2
[CHAP.
is given by
'i(-0[-SK->)]
where
the L2
10
is computed
L2 = L%(1
from
/).
(10'309)
The symbol
Arhom
(1-310)
2'*^
10.9
CoMPARisOiN
of the Feinberg-Galanin
HoMOGENIZATION METHOD
cm
cm
20
22.5
33.8
45.1
0.978
0.934
0.868
1,
30
40
*sr
&hotn
fchet
1 .890
1.198
1.220
1.060
0.954
1.005
0.978
1 .802
1.680
Error
+26
21
8.4
results given in Fig. 10.21 for the 3-cm fuel rods with 5 per cent enrich
The three cases correspond to a = 20, 30, and 40 cm. The
ment.
parameters common to the three systems are:
v = 1.935
^
IS u
= 0.05
Di
2tf>
D,
D,
= D4 = 0.909 cm
= 2g = 0.00249 cm"1
CHAPTER
CONTROL-ROD
11.1
11
THEORY
The model
is used here
REACTOR ANALYSIS
722
[CHAP.
11
As
a first
-H26
Fig.
11.1
Cylindrical
reactor with central control rod.
+ 2Lc'4>c(r)
DbV*Mt) + 2fl,B(r)
->cV2>c(r)
= v~2fP*g*4>c(r)
= 0
(11.1)
where C refers to the reactor and B to the control rod and the remaining
The neutron fluxes given by the
symbols have the usual definitions.
above equations are to satisfy the following boundary conditions:
(1)
c(R,z)
Mp,fi)
*(P.)
= 0
(2)
4>b
(3)
are continuous
at the rod-
(11.2)
Reactor Theory,"
H. L. Garabedian, "Boundary Conditions in Multi-region
Westinghouse Atomic Power Division, Pittsburgh, Pa. WAPD-TN-516, July, 1956.
1
THEORY
CONTROL-ROD
SEC. 11.1]
723
B(p,z)
A[Y0(\iR)
^o(Xip)
= CI0(X2p) cos
- Jo(Xi#)
ro(Xip)] cos
TTZ
where
*B
DB
(11.3)
(11.4)
of the interface
to
be
Dc\i[Y0(\iR)
J^b)
- JoixM)
F^X.b)]
F0(Xi6)
P^/t^b)
70(X2b)
calculations
is to deter
Ja(\iR)
= Y1{X1b)
Ji{\ib)
- DgXihiXj)) F0(X,b)/DcXi/o(X2b)
- DeXj^b) y0(Xib)/Z)cX,/o(X2b)
given
'
is,
= bk/k.
b.
Fo(X,) y0(X,6)
- Jo(X.tf)
REACTOR ANALYSIS
724
[chap.
11
permanently
concept
b, =
b >
(11.6)
As in
is the effective radius of the rod.
the case of the extrapolation distance
for the outer surface of the reactor, t is
6.
1>
reactor
If
K&,)~
/>(&)
we specify
- *'(&)
(11.7)
<Kh)
(11.8) *
Q>)
(11.9) *
we
will
use the
extrapolation-length
condition (11.8).
relation
(1 1 .9) ;
and Nordheim.2
Davison and S. Kushneriuk, "Linear Extrapolation Length for a Black Sphere
and a Black Cylinder," MT-214, Mar. 30, 1946.
R. Scalettar and L. W. Nordheim, "Theory of Pile Control Rods," MDDC-42,
1 B.
June
17,
1946.
SEC. 11.1]
THEORY
CONTROL-ROD
725
1-4 1 I
r-i
i i
i I i
0.7
0.6
2
Control-rod
Fig.
11.3
Extrapolation
condition.
3
radius
b/\u
(11.10)*
be solved
a procedure
for computing
ANALYSIS
REACTOR
726
[chap.
11
A[Y0(\,R)
- Jo(\iR)
70(Xp)
F(X3P)]
cos^|
XjsB!_C*Y
with
The
Y0(\3R)
use of
is
Mp'2)],-b
[fp
The flux distribution for this system is given by the differential equa
tion (11.1). In addition to the extrapolation boundary condition for the
outer surface, (1) in (11.2), we require for the rod-reactor interface1
(11.12)
(n.13)
condition (11.11) yields the
Y^Ub)
(11.14)
we have
z;f>
+ Dc{\\
(*M)2]
v;
then
thus,
2hy-\
\\
of the rod
is
worth,
The reactivity, or
Dc[\\ +
xi
s<C)
The reactor with rod inserted, and the same fuel concentration,
subcritical, and for steady-state operation we require some vc >
(1L15)
is
vlH^
just critical,
rod
is
is
is
+ {v/2Ky
(n.17)
The appropriate interior boundary condition for a finite cylindrical reactor with
central hole
discussed by E. L. Zimmerman, "Boundary Values for the Inner Radius
of a Cylindrical Annular Reactor," Oak Ridge National Laboratory, ORXL-2484,
June
5,
is
1958.
CONTROL-ROD
SEC. 11.2]
THEORY
727
It
11.2
(i)
*(r)
*,(r)
= Z(r)
=
+ W(r)
o,Z(r) + a,W(i)
*,(ft)
uli8J
*,(ft)
= 0
(11.19)
= 0
T-W(i)
- \W(t)
= 0
(8.174)
728
REACTOR
ANALYSIS
[CHAP.
11
and the parameters ps and X2 are the two roots given by Eq. (8.171).
The coupling constants Oi and a2 are defined by Eqs. (8.175). For the
reactor with a single central rod, the flux distributions are axially sym
metric and we may define
W(r)
H(z) 5(p)
(11.20)
If
these relations are used in Eqs. (8.174), it is readily shown that for the
bare cylindrical reactor the indicated functions are given by the expres
sions (refer to Fig. 11.1)
=
H(z)
and
fHKM
(11.22)
X- +
(11.23)
. - (gf , .
where
The neutron-flux
*1(p,2)
2(p,z)
(g
H(z){dJo(yp) + Fo(7p) +
H(z)\ai[aJ0(yp) +
(&Yo(yp)}
+ yiKoipp)]
+ a2[3n/0(/3P) + HK0{fip)]\
3Tt/o(/3P)
(11.24)
The coefficients a,
rod-reactor
(B, 3E,
and
31
are evaluated
interface condition.
the rod, and (2) the net current of fast neutrons is zero at the surface
of the rod.
Consider first case 1. Specifically, we will impose the conditions:
2
=
with
02(6,
6,
(2)
31
,z)
(1);
1
(11.25)
e2
0.
The first of these follows from the fact that the expression for the fast
=
flux
contains two functions F0 and K0, which are singular at
Since we have that
is
limp!g=_?
it
(1L21)
cos(i)
(11.26)
clear that the choice given in (1) of (11.25) provides that the fast
CONTROL-ROD
SEC. 11.2]
THEORY
729
of the rod 6* ,
cation of an extrapolation length correction based on the thermal-group
properties of the core [cf. Eq. (11.6)].
The combination of control-rod boundary conditions (11.25) is based
rod problem.
Ja/0(7P)
3HJoG3p)
H(z)
+ l^lY0(yp) +
joJo(7P)
Jo,
*,(p,z)
9lF0(7P)l
+ 9ltf
(11.24) yields
(&>)[
a2[3TC/o((8p)
(/3p)]J
(11.27)
obtained
if
is
3lff(2)
YojyR)
- M^
J0(yp)
Jo(yR)
+
a2
|-^|
(11.28)
yields
r(7P)
ZoOSp)
KoOSp)
(11.29)
j
<p2(p,z)
these results
j^a,
The substitution of
-aZtfrg
relation
ft
is
The application of condition (2) from (11.25) to this equation gives the
criticality condition for the reactor. This
found to be
2q2r
Y0(yR)
*,(p,) = H(z)
K{ffi)
11
(11.30)
REACTOR ANALYSIS
730
Yo(yR)
J0(yR)
_ Jo(yK)
1
<C 1;
[CHAP.
11
Ya(yK) +?^Ko(f)h)]
(11.31)*
J 0(ybt) ~ J
and so on.
The
use
of this
0{yb)
+ e2yj i(yb)
the alternative
expression
Y0(yb) +
wY^yb) +
Jn
J0(yR)
^ [Jf .056)
us-^-n
n
+ e2yJ i(yb)
+ crftf ,(/)]
(H-32)
o(yb)
Some authors
control rod as determined from the condition that the fast flux be finite
inside the rod.
Note that, although i(0,z) is finite as required, it dips
to large negative values inside the rod and produces a large gradient at
the rod-reactor
For
interface.
!"*i(p,)l
=0
(4)
2(6,,z)
= 0
(11.33)
(11.34)
SEC. 11.2]
CONTROL-ROD
THEORY
731
Fig.
Generated for wjivans (University of Florida) on 2015-09-22 17:23 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Flux distribution around a central rod using the condition that fast flux is
11.4
finite everywhere in the rod.
If
these results are used in Eqs. (11.24) and conditions (11.33) applied,
two equations relating, say, CI and 911. From condition (4)
we obtain
we have,
for example,
a \j0(yb.)
- ig. F(76,)l
311
(11.35)
If we
use also condition (3) and divide the resulting expression by (11.35),
the following criticality equation is obtained :
0
Y0(yR)
Jo(y&)
Y0(yb*)c3(b) + (a2/al)Y1{yb)c1(K)
I Jo{yK
(11.36)
)C3(6)
+ (0s/0l) J1(76)c1(6.
where
Ci(x) = i0(/3x)
The indicated
Ko(pR)
approximations
KoiPx)
apply for R
KoW)
>5> x.
(11.37)
732
ANALYSIS
REACTOR
[CHAP.
11
"<3
J*(yb.)
AA Kt(flb)
"
K,(7b)
,,M
YojyR)
(11.38)*
is
it
not fabricated from solid sections of high thermal cross section materials.
Since the mean free paths in such materials are very short,
clear that
very few thermal neutrons can ever
penetrate far inside
rod of several
centimeters width. Nearly all the
neutron captures occur close to the
surface of the rod, and the interior
serves no real purpose.
Thus the
one
is
In practice
I,
ft
use of cross-sectional
is
I.
it
is
is
hollow "control tube" and (2) the control tube containing moderator
material.
The first case
essentially the same as an absorber tube filled
In both situations the tube stops the
with high-atomic-mass material.
Fast neutrons, however, easily pass
passage of all thermal neutrons.
black to thermal neu
through the tube (since by assumption the tube
trons only) and, moreover, are little affected by the interior of the rod,
In either case the fast
whether
be empty or filled with heavy nuclei.
neutrons emerge with much the same angular distribution as that with
which they entered and without any significant degradation in energy.
From an analytical viewpoint both problems are equivalent to the solidcontrol-rod problem with zero fast-neutron net current at the surface.
Thus the results (11.36) and (11.38) apply directly to case
In case
the moderator material within the control element serves
2
is
CONTROL-ROD
SEC. 11.2]
THEORY
733
moderation
The appropriate
and finiteness:
(1)
fa(b,z) = fa(b,z)
(2)
^*3H,-=^*iH,
(3)
fa{p,z)
*4(W)
= 0
(11.39)
or
fa(b,z) =
-u
*4(p,z)l
(11.40)
+ u
(11.41)
REACTOR ANALYSIS
734
[CHAP.
11
The
(11.42)
eloiw)
y,3>
/ ir
(11.43)
\2
(11.44)
Note that the first term in this relation may be approximated by 1/r,
where r is the age to thermal for the central region of the rod.
These results may be used to determine the flux distribution within
the core and the criticality condition for the reactor.
The general
solutions for the fluxes in the core are given by Eqs. (11.24).
If we apply
boundary condition (11.19), i.e., the relations (11.34), the solutions may
be
written
- ff(2)
*,(p,*)
(a p,(w)
K(W)1
+
*,(,2)
;/(z)
L,a
- ['*>- 8 *HI
(11.45)
r,(Yp)J
The expression for the fast flux must satisfy the interface conditions
(1) and (2) in (11.39), and the thermal flux, condition (4) of (11.33).
The relations for the fast flux we write in the form
"3
_
where
dp
r
"
.
01(P'Z)
41
n i i
k<
(11.4b)
Jp
ti denotes the extrapolation distance for the fast group in the core.
Equation (11.46) together with (4) from (11.33) constitute the critical
ity condition for this system. From the latter, we obtain Eq. (11.35)
BEC. 11.2]
relating
CONTROL-ROD
and
Ct
THEORY
735
911.
0311
7,036)
Kxifib)
(11.47)
Yo(yR)
Jo(yR)
with
c\ and
c3
(a,/ai)ci(b.)c(fr) + F0(76t)[gf,c3(6)
(o/oi)ci(6,)c,(6) + y0(76*)[/3tlC3(6)
- c,(b)]
- ci(6)]
J0(76) + 7*1^1(76)
+ 7iF,(76)
(11.49)
c4(6) ^ F0(76)
When R
~2> 6,
a good
approximation to (11.48)
^0(76,) A7(6)
7o(7^)
e2,
(11.48)
at
is given by
c2(6)
rgilg1(|6M^oW]
(n
51)
where
*2(6,z) =
(H-52)
*[^*(p.*)]^
It
M(b)
~ j3ai6i7Li(/36)/a2/C0(i36)
c4 =^
Um^fg=-|
7*1(76)
fc-.o
1 R.
L. Murray and
J.
and
> 0, we may
Then, using
(11.53)
ANALYSIS
REACTOR
736
[CHAP.
11
_ Ya{yK)
/oA
{orb_^0
Ra(fiK)
(u
54)
The thermal-flux
control rod (tube)
-Z>V**(r)
ZS*>*4(r)
equation
p^Mr)
i-
a,T(p)]
(11-55)
Up,z)
= H{z)
[&U(P)
VV
x4V
The function
with x\ =
= 0
HU
(11.56)
denotes the solu
(11.57)
where k\ = x\ +
Zo(p)
Q^Y
(11.58)
We have used here the requirement that the solution remain finite inside
If we use (11.56) in (11.55),
the rod and that Eq. (11.19) be satisfied.
along with (11.42), we obtain
o,[Vs(ffD
- x]HT]
V\HT)
- x\HT
HT
= 0
[cf.
with x\ =
(11.59)
3,
we note that
(11.60)
Eq. (8.184)]
(11.61)
CONTROL-ROD
SEC. 11.2]
THEORY
737
If
fl/o(*ib)
we obtain
for the
It
n ?'e
^(Xj
*l)
V,
[{'{"H
L'o(*4t*)
/.(/)
/.(*)! H(z)
CI
and
911
(11.62)
through
(2) in (11.39).
Figure 11.7 shows the flux dis
tribution around
rod containing
d.
tubular control
moderator material.
Numerical Example.
sults
developed
The
re
in the preceding
Rodreactor interface
Table
8.5.
ample
we
obtain
cylinder with
= 36.7 cm.
REACTOR ANALYSIS
738
[chap.
11
down diffusion
Water filled
contrc Irod
equations,
are
Fio.
rod radius 6, cm
11.8
not well satisfied when the characteristic dimensions of the rod are in the
order of one diffusion length.
11.3
The calculation
of the effectiveness of a single eccentric control rod in a bare cylindrical
reactor on the basis of the two-group model requires the general solution
to Eqs. (8.147) which includes the azimuthal dependence.
The formal
solution for the fast and thermal fluxes
and d>2, respectively, we writ*
a.
General Solution
Z(x)mH{g)Q(p,9)
where
p,
z, and
>p
however, we define Z
(11.63)
If
these
SEC. 11.3]
CONTROL-ROD
THEORY
739
expressions are used in Eqs. (8.174), it is easily shown that the function
H{z) is given by Eq. (11.21), for the bare cylinder of height 2h, and the
functions G and F by
op'
(92F
+ JiS
1
p' dtpp op
dF
d2F
p op
p'
dtp'
op-
72G = 0
*(>)
We seek solutions
- (*) S(p)
*W)
(p)
(11.64)
to these
(11.66)
-*(<?)
n = 0, 1, 2,
e*"*
(11.67)
<&nYn(yp)
(11.68)
3lnXn(/3p)
(11.69)
f(p,v)
[-^.(7P) + ny(7P)]ein-
(11.70)
l3R"/s") + tt"^(/3p)]ein*
(11.71)
f| a 00
3C
n ao
The substitution of these results, along with (11.63), into (11.18) yields
for the fluxes in the core:
4>x(p,v,z)
= H{z)
|[-^-(7P) + ffinFn(7p)]
+
*(P,V0 = H(Z)
[3K/03p)
2_ aC |ai[n./n(7p)
+ HJCCftO]
}*-
(11.72)*
+ nF(7p)]
n ,
a2[sm/03P)
3iB/C(/JP)]}e<*
(11.73)*
1 See also
Chemistry,"
REACTOR ANALYSIS
740
[CHAP.
11
to
(1)
$2
(2)
= 0
at di = 6,
at*
= 0
It
is clear
the
Fig.
11.9
one containing
two functions which
vary strongly near the rod but are
SoFoMi) +
JFoKoGSdi)
(11.75)
But these functions are symmetric about the center of the rod; therefore,
they cannot satisfy the boundary condition (11.19) at p = R in the
1 F.
SEC. 11.3]
CONTROL-ROD
THEORY
741
We select, therefore,
= H(z)
|ai80F0(7rfi)
+ ajffoKoGSdi)
^ ao
[aia7n(7p)
a23Hn/n(i8p)]e<,",J
(11.76)
Note that the terms in ($>nY(yp) and yLK(f}p) have been omitted so as
to avoid the singularity at p = 0 [cf. Eq. (11.73)]. For the fast flux we
take the corresponding form
*i(p,>,c) =
H(z)
{SoFo(7<*i)
+ MToGMi)
[n^n(7p)
n = ao
In working with
+ SJIU/.OJp)]**"*}
(11.77)
it will be convenient to
With the aid of the
use series expansions for the singular functions.
addition theorem for Bessel functions1 we can write (refer to Fig. 11.9)
these expressions for the fluxes
F0(7<*i)
Koifidi)
Jm{ya)
"V"
2
Y^ei"'
(11-78)
Im(fia) Km(Mc""*
H(z)
fSoJm(7) Fm(7p)
+ So/m(0a) Km(fip)
+ dmJm(yp) +
3irm/m(0p)!e<m*
ae
2(p,<P,z)
H(z)
(11.79)
a2[ff0/m(/3a) /Cm03p)
3Tlm/m(/3p)]
je*-'
(11.80)
reactor (11.19), we must equate to zero the coefficients of eim* for all m,
and <2 evaluated at p = R. The simultaneous solution of the
resulting expressions yields the following set of relations between the
in both
chap.
XI,
Cam
REACTOR ANALYSIS
742
[CHAP.
11
coefficients:
am =
-SoJ47a)^ll
Jm{yR)
<n-81)
-S0U(fia) Krm{f
(11.82)
3TCm
which
it will
be convenient
by the functions
BO
P(p,f) =
[mJm(7P) + mmIm(0p)]ei""'
(11.83)
(11.84)
-
X
Q(p)>)
[a!ttmym(7p)
a29Tlm/(/3p)]e<-"'
m=
In
condition
(1) from
(11.76)]
8=
-9.()^-QM
(11.85)
OC
where
Q(a,0)
[a^J^ya) +
6,=
wherein
we have neglected
(11.86)
as3ltm/m(/3a)]
-5o
(11.87)
the derivative of
P with
respect to di as
planned.
dm, and
ffo,
/o,\ 0K1(m
3Tlm
between
(11.81),
First we
Kt((ib,)
IKda)
V
,
F(7&.)
- ^(7a)
Vt^:
/i(|Sa)
,
>
f=
(0#)
-^(7a)
m 1
(11.88)
CONTROL-ROD
SEC. 11.4]
THEORY
743
I-n
'
l)"Jn
In
Yn
K_n
l)nYn
gg^
rtfrju
Note that
is
rFo(7feJ_/^jp(igx(^)l
further reduction
(n.90)*
Eq. (11.53)
may be applied.
11.4
condition that the fast flux remains finite inside the rods
Eqs.
to
[refer
(11.25)].
The problem of determining the criticality condition and the flux
distribution in the two-group model for a ring of gray rods has been
he applies the
considered
of the
Nordheim-Scalettar approach and may be applied when either the rods
or the reactor, or both, are multiplying media.
b. Application to Solid -absorber Rod.
The derivation of the criti
by Avery.3
is
a generalization
REACTOR ANALYSIS
744
[CHAP.
11
SoFoMm) +
If
we choose
(11.91)
VoK0(Pdm)
Rod 1
11.10
Fig.
11.11
of M rods.
to help meet all the boundary conditions, then the solutions for the fluxes
can be constructed in the form [cf. Eqs. (11.76) and (11.77)]
{8
m-
Y0(ydm)
fj
*i(p,*,) = H(z)
Ko(0dm)
1
m
[./(tp)
/C0(y3dM)
o,3RB/.C8p)]e*-'}
(11.93)
[a,aJ(7p)
a
is
it
a2ff
to use
series representation for each of the
Thus, as in (11.78), we write for rod m (refer to
convenient
functions.
(11.92)
m=
+
singular
Y
1
Y0(ydm)
m=
Again,
3TCn/n(fr>)]e<'}
t(.P,<p,z) = H(z)
|a,So
Y(ydm)
Fig. 11.11)
J.(ya) F.(7p)e-
(11.94)
>p
If
- jj (m -
1)
1,
9m
where
is
Fig.
(11.95)
CONTROL-ROD
SEC. 11.4]
THEORY
745
is,
Note that for a ring of M rods the angular interval between rods is 2t/M ;
therefore, <pm = 2r(m \)/M.
the interval between rod m and rod 1
If we use (11.95) in (11.94) and sum over all M rods as indicated in
m=
J.(ya)
Y.{yp)e<
Y0(ydm)
(11.92), then
=. oo
_-..<
u/*
(11.96)
m1
_t
1
1_-
j|f
e-'-"IM
if
g-Unfm-UlM
is
Now
an integer
(11.97)
_____ _
is
__
if
g-T.(m-i)/M
AT
not an integer
Thus all terms in Eq. (11.96) vanish except those in the form m Ym
and e"Mf, and these appear with factor M. Equation (11.96) therefore
reduces to
Yo(y<L)=
Jj..(7o) YM.(yp)eiM'*
(11.98)
KmMW'*
[SoJj_.(7cO F.v>(7p)
ffo/_f.03a)
[^n(Yp)
3Il,/.(|8p)]ei-}
a-a W
= H{z)
{M
',
Similar results can be obtained for the functions Ko(fidm) the substitution
of these, along with (11.98), into the expression for the fast flux yields
n o
(11.99)
is
reactor, we must equate to zero the coefficients of like functions e*B* (that
If this
all exponentials with n = Ms).
done for both <fi and fa,
is
is,
3Tl_f. =
All other
9TC
-M&,Ju.(ya) YT"jy
(11.100)
-M^lu.ifia)
(11.101)
JM.{yR)
KTM'^?1
REACTOR ANALYSIS
746
[CHAP.
11
(11.102)
(11.103)
and (11.93), respectively; these terms are all slowly varying functions in
The appli
the vicinity of rod 1 [cf. P and Q in (11.83) and (11.84)].
cation of condition (1) from (11.74) to (11.103) yields an equation analo
gous to (11.85); this is
\ai/
where
f^>--^
Y0(yb,)
U(a,0)
O!80
Y0(ydlm) +
a2ff
K0(pdlm)
m=2
m=2
+
Generated for wjivans (University of Florida) on 2015-09-22 17:23 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
(11.104)
aiF0(-yo*)
[aiJ(7o) +
a23Tl/n(/Sa)]
(11.105)
n =
and
dim
Fig.
In the general
11.12
Distance between
and rod m.
rod centers.
case, the distance between the centers of rod m' and rod m
is given by (refer to
Fig.
dm-m
11.12)
= 2a sin
(m'
m)
(11.106)
SEC. 11.4]
CONTROL-ROD
THEORY
747
tions involving
m==2
Y(yb.) +
"
Ya(ydlm)
- MJKya)
m=2
-1
- 2M V J2v.(7a) jf^fj
=1
(11.107)
written
R)
m=2
-fe)Sw[K,(*-)+SK'(H)
m =2
<1I108)*
CHAPTER
HYDROGENOUS
12.1
12
SYSTEMS
a. General Considerations.
Most of the discussion of reactor calcu
lations in this work has been based on the age-diffusion model of neutron
behavior.
This model is very convenient to use and has yielded valuable
results in innumerable cases.
However, there is an important class of
to
which
this
model
is
If a system includes
poorly applicable.
systems
a large proportion of hydrogen, the approximations involved in the agediffusion method are really not valid. As indicated in the discussion of
Sec.
method
would
collision,
relationship
(7.134) takes a
SEC. 12.1]
HYDROGENOUS
SYSTEMS
749
(b) An empirical
kernel may be used with the usual one-velocity diffusion
equations for the thermal neutrons.
In general, a particular method of the second class will give useful
results only for a limited range of types of reactors; thus, in undertaking
a new design analysis, considerable effort must be applied to find the
appropriate constants.
b. Utility of the Measured Neutron Age in Water.
One very impor
urements
slowing-down
tant quantity involved in all the methods indicated above is the neutron
A simple example
age to thermal (or to indium resonance) in water.
of the one approach (6) would be the application of the criticality condition
(6.80) to a bare water-moderated system, wherein for the fast leakage
losses one uses exp ( B2rtb) with the age to thermal as obtained from
More fundamentally, an experimental determination of the
experiment.
actually
in
water
gives the spatial distribution of neutrons of various
age
Thus an obvious
energies as they slow down from a fission source.
requirement for a proposed calculational method is that when it is applied
to the geometry of the experiment it should yield the observed distribu
In practice this is most conveniently approached by requiring
tion.
that the calculational method give accurately a finite number of the lower
If, in a spherical geometry,
moments of the slowing-down distribution.
the spatial distribution of neutrons reaching thermal energy at a distance
r from a fission source is Kth(r), then the nth moment r" is defined as
4tt
rnKtb(r)r2 dr
(12.1)*
The most important of these is the second, r"-, and in many cases, to
require that the second moment be adequately represented is regarded
In the age-diffusion approximation, the second moment is
as sufficient.
proportional to the
age, and
6t(w)
use
(6.52)
a generalized age
for
(12.2)*
Then, the simplest and most direct criterion for evaluating a proposed
calculational
method for water-moderated
systems is that it yield
correctly the measured age to thermal in water (i.e., the correct second
moment).
In the sections which follow,
REACTOR ANALYSIS
750
[CHAP.
12
12.2
case.
q(l,u) =
last scattering
collision with
a nonhydrogen
nucleus
of neutrons slowing down into lethargy
interval du about u per unit time per unit volume
r;(r,w) du =
about r which made their last scattering collision
number
(12.3)
with hydrogen
The slowing-down
2j0)
Zi>(M)*(r,W)
in the
(12.4)
except hydrogen.
1 G.
tf>(r,u)
sec. 12.2]
HYDROGENOUS
SYSTEMS
751
thus,
ij(r,M) du =
f* 2(u')
4>{T,u')e'-
du' du
(12.5)
where (H) refers to the hydrogen nuclei, and it is implied that there
are no sources above u = 0.
In order to determine the net gain in
neutron density in dr du from scattering collisions in and out of this
differential
2W)(u)
(j,u)
(12.6)
7(r,u)
be
du] dr
2(u) (r,u)
- [q(t,u + du)
written
du] dr
(12.7)
as a loss to be consistent
- D(u)
V2<Kr,w) du
+ 2(u)
2(u)
4>(r,u)
{t,u) du
du + q(r,u + du)
S(i,u) du + q(r,u) + ij(r,w) du
(12.8)
(r,u)
-v(t,u) +
-D(u)
V24>(t,u)
20(m) (r,w)
2(u)
*(r,u)
- ^-v(r,u)
- |-(r,w)
du
du
(12.9)*
+ S(r,u)
(12.10)*
A discussion of computational
752
RKACTOR
ANALYSIS
[CHAP.
in the calculation
12
of the
12.3
a mixture
of
nuclear species.
is
a-V(T,v,a)
+ St(y)
2^
S(r,v,0)
dv' dQ'
= 1
/*
= -S(z,u,ai)
X L dU'
=
Si'>(M,) *(x-u'>'1')*W(u''i;M'>Q')
dQ'
(l2-12)
2^
fa-
to the lethargy
variable.
We obtain the solution to Eq. (12.12) by expanding the flux and the
For the
frequency functions 8<0 in infinite series of Legendre functions.
frequency function we use the general form (7.129), and for the flux
[cf. Eq. (7.74)],
CO
KW)
(^P)
Mx,u)PM
(12.13)
similar
series
is
If
defined also for the source function S(x,u,p).
with
the
addition
used
in
theorem
(12.12),
along
Eq.
are
these expansions
(r,v,a)
sec. 12.3]
SYSTEMS
HYDROGENOUS
753
(7.21), and the orthogonality properties for the Legendre functions are
applied, the result is an infinite set of coupled equations in the coefficients
-2,(m)
>
Jf
2(u') (x,w')
*"+l(x'M) +
Ix
where
and
+ Sn(x,u) +
<fn(x,u)
a,
(> tt)
8'(m;m',Q') du'
*-l(:r'M)
(12-14)
(12.15)
|^
species.
range of integration for the lethargy variable extends over the usual
collision interval appropriate to the particular species.
It is of interest
to mention that the set (12.14) may also be derived directly from the
general form (7.125) by applying the procedure used in Sec. 7.2f. The
is
(^',0) = F"<(r')e""U
a,
(12.16)
[cf.
Eq. (7.134)]
^
f,(M,')
as demonstrated
i[(Ai +
l)e -
(At
- l)e-*]
If
with
(12.17)
(12.16) in (12.14) and approximate the infinite set by the first three
harmonics, we obtain, for
system with isotropic sources,
= S0(x,u)
i=
dx
+3dx
Z'fl(u') *(*,')
-f^ du'
(12.18)
-f1^
(12.19)
if
+ 2<M *i(x")
AT
is
\p,
du'
REACTOR ANALYSIS
754
2d
5 dx
3d
*1(X'M)
2(') *,(*,')
./u + lnoci
Z_/
=
It
these
[3f?-(",M,)
2(1
12
+ 2|(u)
*3(X'M)
A'
[CHAP.
11
e-'-
du'
*i)
(12.20)
is convenient
*(,) = t^Tj
1
So(s,w)
f"
/_
/"
*(x,u)e*
da;
(S0(a;,)e<
dx
of
(12.21)
= o(s,u)
A1
+y.r
/
Generated for wjivans (University of Florida) on 2015-09-22 17:24 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
-y
*o(s,m) +
2<(')
*o(,')
du'
(12.22)
du'
(12.23)
e"'-
<*w'
(12.24)
2>(m')
,(,') r.(,')
1= 1
^ *i(s,u)
^ (s,u)
2,(m) 2(s,u)
.V
y
Z_/
i=
./u+lna,
2i<1(w')
*(,') [3r](/?'"')
A1
x
ar<)
11
If
symmetric
*o and
go
s2
*o(,)
= *oo() + *o2(w)
2j + *o() Ti
So(,)
' - '
(12-25)
Soo(m)
+ Soj(w)
2]
S<(M)
4j +
'
Then, the form of Eqs. (12.22) through (12.24) shows that it is sufficient
to represent the higher-order coefficients in power series beginning with
SEC. 12.3]
HYDROGENOUS
755
SYSTEMS
*,(,) =
*(u) ^ + ,() +
*,(,) =
*()
+ (u)
|^
|* +
(12.26)
(12.27)
(12.28)
and so on.
tions
are
AT
2'(w') *oo(')
du'
(12.29)
*(')
du'
(12.30)
i=l
| So2(m)
iV
2>(u')
i=\
2,(u) *u(u)
*oo(")
(12.31)
If
the cross sections and the source distribution are known, these equa
thus, (12.29) yields *0o, which
tions may be solved simultaneously;
when applied to (12.31) yields
This in turn gives *<>2 from (12.30),
and so on.
In
the generalized definition of the age (12.2).
to
distance
the
mean
squared
age, we require
given by Eq. (7.343), which in t'ie
expression
is
/""
I 2,(u) *02(m)
tn(w) +
S-=
(12.32)
Fourier transforms
beginning with
discussion
to
previously
(12.21).
[refer
Eq. (7.364) and extending through Eq. (7.374)] that these are related
It
was shown
REACTOR ANALYSIS
756
[CHAP.
Eq. (7.374)]
PTC _
r(3Va')*.(,tt)1
*o(s,u)
[
J._o
..jaw
12
[cf.
(1M8)
*oo()
Pi
The accurate
set of
- 0(s,u)
V
-l
Z'(m') (,')
du'
(12.34)
du'
(12.35)
+ 2,(u) *i(s,u)
u+
lna.
As in
a,-
2(m)
where 2,(zt)
is
These approximations
is^i{s,u)
*o(s,u)]
(12.36)
AT
(12.37)
for
=* 2<0(m) i(s,m)
[2>(u)
1,
*i(s,u')
+ ('
2,
2(u')
=* 2i(w) *o(,)
*0(,')
2>(tt')
- A [J2.(u)
mo2.(m)
*.(8,u)]
*,(s,u)
(12.38)
(12.39)
^2i()
that
u)
IS
= So(s,w)
N
(12.40)
sec. 12.3]
HYDROGENOUS
SYSTEMS
757
If we define
{, and Mo" are related to A( in the usual way.
be
Eq.
to
then
reduced
to the form
may
according
(7.161),
(12.38)
D(u)
The quantities
= So(,)
- ^ IfS.(u)
*.(,)]
(12.41)
range In 1/a, for each species so that they may be represented by one or
These
two terms in a Taylor expansion [as in (12.36) and (12.37)].
in
Sees.
7.3f
and
The
7.3g.
conditions were established previously
gen
particular,
to
obtain
the
Goertzel-Selengut
equations:
(1)
Equation
*(,)
Z(m)
+ 2i'(w) *o(s,m)
J*
*(,)]
[So2,>(m)
= 8o(,)
i.
is
1,
1,
is,
(12.36) is applied
V\u') o(,u')e"'=
mo2.(w)
*i(s,u)
du'
(12.42)
(12.43)
H. Hurwitz and P. Zweifel, J. Appl. Phys., 26, 923 (1955), have considered the
expansion of the scattering function, rather than the flux, in spherical harmonics; by
dropping all terms beyond the nth, they obtain the so-called Bn approximation, which
has been shown to be more accurate than the corresponding P approximation devel
oped here.
it
is
is
ANALYSIS
REACTOR
[CHAP.
Here, the index 0 refers to all substances other than hydrogen, that
12
is,
758
N-l
2(u)
N-l
fcZ* ()
(12.45)
fo2i()
(12.44)
i=i
- ir [o2(w)*o(s,w)]
= So(s,m)
2(w)
2(m')
If
from
we eliminate
[using (7.161) for D(u)]
?0(s,u)
*0(,u')e-'-" *'
(12.46)
d2
-D(u)
= S0(x,u)
+ 2() *0(x,)
<^0(x,m)
- 2(u) <,(*,)
(*,) + (*,)
(12.47)
ij(i,)
JoU
g(x,u) = |02<0)(u)
(x,u)
(12.48)
>o(x,M')eu'-"
2Jff>(u')
dw'
(12.49)
is
In the consistent
Pi
is
iao(8,i0 +
[o210)(m)
2(m)
*o(,m)]
applied
Equa
+ 2i'(M) *(S,U)
Jo
2<*'(w')
*,(8ft*')-'- du'
(12.50)*
du'
(12.51)*
2<H(w')
*i(8,u')e*("'
>
- ir
du
= So(s,u)
f"
^""(w)
where
N-l
f?Z()
(12.52)
HYDROGENOUS
sec. 12.4]
SYSTEMS
759
Some
These equations are, in general, difficult to handle analytically.
for
the
consistent
set
to
the
has
been
Pi
given
treatment
analytical
Biedenharn.2
special case of pure hydrogen by Simon1 and by Welton and
In the present discussion we shall content ourselves with a comparison
of the various models on the basis of numerical calculations.
12.4
Numerical Results
includes a development for water-metal systems based on the FermiMarshak method. The comparison of the methods is drawn on the basis
Besides the relative ease of
of the computed age-to-indium resonance.
measurement (as contrasted with the age to thermal) it has the advantage
of easier interpretation because the thermal motion and molecular effects
are not generally important above the indium resonance (1.44 ev).
According to the general definition of the age, the quantity of interest
is the second moment of the neutron flux at the energy of the indium
resonance.3
= 26.7 cm2
Examples
and
REACTOR ANALYSIS
760
[CHAP.
12
"Medusa" is also available for the UNIVAC, and this is based entirely
12.1
Comparison
Model for
calculation
Age (cm1) to
indium resonance
Age-diffusion
17.5
Monte Carlo
25.6
Consistent
reactors
Pi
25.3
Reference
Comments
Oxygen mass
assumed infinite
Second moment of
slowing-down
density
R. R. Coveyou,
ORNL-
2081, p. 144.
H. Hurwitz and P.
Appl. Phys.,
Zweifel,
J.
Goertzel-Selengut
. . .
30.9
H. Hurwitz and P.
Zweifel,
Appl. Phys.,
J.
uranium)
Reflector: Spherical
For the analysis of this system the lethargy scale was divide into 29 fast
The spatial inte
groups plus a thermal group corresponding to 183F.
which
of
0.754
cm
yielded 16 space
intervals
gration was carried out using
The results of this compu
points in the core and 14 in the reflector.
tation are given in Table 12.2.
J. H. Alexander and N. D. Given, "A Machine Multigroup Calculation: The Eye
Wash Program for UNIVAC," Oak Ridge National Laboratory, ORNL-1925, Sept.
1
1955.
15,
" Multigroup
W. E. Kinney, R. R. Coveyou, J. G. Sullivan, and B. J. Garrick,
Calculations for Multiregion Reactor," Applied Nuclear Physics Division Annual
Report for Period Ending Sept. 10, 1956, ORNL-2081, pp. 133-139.
* Actually there are some slight differences because of different assumptions in the
averaging process for obtaining group constants from cross-section data.
1
sec. 12.4]
HYDROGENOUS
case
SYSTEMS
761
the
An examination
diffusion
Table
of the results given in Table 12.2 shows that the agegives a higher estimate of the multiplication
approximation
12.2
Multigboup
Model
Calculations fob
Wateb-modebated
Fraction of Fraction of
Total
neutrons
absorp
neutrons
absorbed
tions in
escaping
in reactor
core
Reactob
Fraction of Fraction of
neutrons
fissions
due to
escaping
as fast
thermal
neutrons
neutrons
Age-diffusion
(Medusa)
Goertzel-Selengut
(Eye Wash) ....
1.17
0.789
0.763
0.211
0.107
0.375
1.00
0.663
0.770
0.337
0.239
0.430
age
for water
G-S model
This
spatial distribution of the group fluxes does not differ appreciably.
point is illustrated in Fig. 12.1, which shows that the computed power
distributions for the two methods are nearly identical.
The multigroup calculational method based on the Goertzel-Selengut
model has also been tested in the calculation of critical assemblies.
E. G. Silver1 has compared the computed critical masses from the Eye
Wash code with the experimentally determined values for the case of
enriched uranium assemblies in water.2 Unfortunately the evaluation of
the analytical model was not clear cut since it was necessary to apply a
from the actual completely reflected parallelepiped geome
transformation
REACTOR ANALYSIS
762
12
[CHAP.
2.4
2.0
,-
1.6
a
s
0.8
0.4
10
12
14
16
Fig.
12.1
Comparison of the power distributions
Goertzel-Selengut models.
and
12.3
Comparison
by
Masses
61.3
45
61.3
38.5
31
40.5
Computed
Experimental
3.242
3.246
3.26
3.27
Computed
CHAPTER
13
PERTURBATION THEORY
13.1
The
in this chapter
are
ing situations of this sort utilize the fact that the changes are sufficiently
small that the characteristics of the perturbed system may be constructed
This approach is widely used in the
from the analysis of the original.
problems of physics and engineering science because it frequently leads
to substantial simplifications in the calculational procedure.
As an illustration of such problems, consider the situation wherein a
at some point within a
critical assembly.
Since the precise nuclear balance of the system has
been upset, it can no longer sustain itself in steady-state operation, and
it is to be expected that the over-all flux level in the reactor would fall in
small lump of absorbing
material
is introduced
the absorber.
1 The
The effect
REACTOR ANALYSIS
764
[CHAP.
13
a rigorous mathematical
O*0(r,w)
(13.1)
Here <o(r,M) denotes the neutron flux as a function of space and lethargy,
and O symbolizes an operator chosen appropriately for the system at
For example, in the one-velocity model the operator O would have
hand.
the form
O =
-V
DV +
(13.2)
vXf
0=0^-
vOi
(13.3)
Here the intent is to bring the quantity v explicitly into the calculation.
the one-velocity case, for example, we have from (13.2)
In
O! =
-V
DV +
Oi
S0
2/
(13.4)
PERTURBATION
SEC. 13.1]
THEORY
765
previously [cf. Eqs. (4.167) and (5.178)], the change in the effective
multiplication constant brought about by the nonuniform changes is
given by
T
If
" "
<13 5>
alterations,
O''0(T,u)
(01
v'
Oi)
4>'0(r,u)
= 0
(13.6)
Q of
Eq. (13.6)
Q'
Q + SQ
(13.7)
where the unprimed quantity refers to the unperturbed system, and the
change &Q in passing from the unperturbed to the perturbed system is
If we use the form (13.7) everywhere in (13.6), and
regarded as small.
apply also (13.1) and (13.3), and neglect all terms containing the product
of more than one small quantity, we obtain
SO 0o
+ O
54>o
(Oi
v02)
80 +
(BOi
v 8O2)4>0
&v Oi4>o
= 0
(13.8)
REACTOR ANALYSIS
766
solve for
v,
[CHAP.
13
we obtain
Jv
dr
Ju
^0(r,w) Oi^0(r,t<) du
(13.9)
O2*0(r,w) du
jydr Ju^0(r,M)
hv =
jydi Ju [&MOi
vO2)4>0
+ ^o(Oi
JY
dr
S02)o + MOi
\f/0 O2<0
- ?02) ] du
du
(13.10)
Because of Eq. (13.1) the first of the three terms under the integral in
the numerator of Eq. (13.10) vanishes.
We consider now the last of the
In order to evaluate this term we introduce two new oper
three terms.
ators
f
Jv
dr
fy
dr
f foOiF
Ju
fu
du =
toOtF du
f
jv
dr
fr
dr
f
Ju
fu
FOtyo du
(a)
FOfyv du
(6)
Jydr
JuM<>i
v02)8odu
Jvdr
O'Mr.M) ^ (OJ
0(Ol
- xO^o du
^o(r,w)>
- vO\) Mr,u)
(13.11)
(13.12)
= 0
(13.13)
With this choice the last integral of Eq. (13.10) vanishes, and the change
in the multiplication constant may be expressed, finally, as
5,
5k
r- = =
K
"
jYdxjuUiOi-v60i)^du
v \dx / Vo02<^odu
(13.14)
Equation (13.14)
of this chapter will
is of the general
PERTURBATION
SEC. 13.1]
THEORY
767
is,
calculation
generally
to a consideration
is
is
dr =
02
(13.15)
8Z(rWo(r) *o(r)] dr
(13.16)
is
jv
jrMr)
changed
Aa0{rB)
is
W(i)
Mr)
is
If
*o(r)
(13.17)
also possible.
of Eq. (13.16)
flux in the perturbed system (as
is
which
is
it
is
in this
removal of neutrons
weighted by the function ^o(r), which
For the simple one-velocity case
sense, called the importance function.
is
REACTOR ANALYSIS
768
[chap.
13
be shown
flux
Importance of neutrons at r
,
a n
model)
-x
(in one-velocity
l(i)
...
~fo(r)
, .
= 0o(r)
(13.18)
v
in one-velocity
model
that
is,
(as
turbed
(13.19)
placed in
very strong at the center, where the flux
thermal reactor
high.
In a slab reactor, for example, the statistical weight for absorption would
fall off as cos2 x. (The implications of this behavior are discussed fur
neutron
absorption
is
model
is
considered in
detail.)
Of course, such results
Still
it
sections are concerned, tends to vary spatially as the square of the flux.
c. The Adjoint Function and Operators.
We return now to the formal
outline of the perturbation method as started in Sec. 13.1a.
In the
usual mathematical
is
is
is
is
thermal
is
of extra
PERTURBATION
SEC. 13.1]
THEORY
769
system, allow an easy evaluation of the effects of many types and dis
tributions of small changes from the original.
In fact, in some cases,
as for the one-velocity model mentioned above, only one calculation is
required since the adjoint function is just equal to the flux itself.
This
situation arises when the operators in the reactor equation are self-adjoint,
Before turning to the calculation of
i.e., when Of and O are identical.
adjoint functions for special models, however, we shall examine one
general property of the adjoint function.
is
it
is
r',
introduced
it
For instance,
=
v
0<f,(i,u,t^',u')
at
>
0
is
down into the fundamental mode would be higher in the former case.
We now attempt to solve this problem in general terms and find the form
and magnitude of the ultimate flux.
taken to satisfy the equation
(r,u,t;r',u>) +
S(t)
h{u')
(13.20)
is
is
r,
at
Here >(r,w,<;r',w')
the neutron flux of lethargy u, at space point
time after the introduction of one neutron of lethargy u' at position r'.
denotes the reactor operator of Eq. (13.1). The approach
The symbol
to the solution of this problem
through expansion of the flux in space
2,
0,
= X,<fc(r,u)
1,
0f(r,)
it
is
(13.21)
is
is
X,
The eigenvalues
are assumed to be arranged in order of increasing
taken to be zero so that
magnitude with increasing index, and X0
just the reactor flux of Eq. (13.1). The basic assumption here
<o(r,u)
that the eigenvalues of Eq. (13.21) can be arranged in order of increas
The eigenvalues and eigenfunctions of
ing magnitude of the real part.
is
is
introduced
REACTOR ANALYSIS
770
[CHAP.
13
thus,
(T,u,t?',u')
a,;r',u') *(r,u)
(13.22)
Here
<,
dr
fu
fv
in the sense
fc(r,u) 0,(r,) du
(13.23)
5,7
is
is
is
If
du =
jydt
ju4>j04>idu
0+
dr
du = Xrf
frOti
fy
dr
X,
fv
of Eq. (13.13),
fydr jui[0^j\du
= X,50
(13.25)
The form of Eq. (13.25), along with the uniqueness of the functions
must satisfy the equation
shows that the
Otyy(r,tt) = X^,(r,)
(13.26)
Thus the orthogonal functions are just the adjoint functions, and ^0(r,w)
just the flux adjoint of Eq. (13.13).
We now introduce the expansion (13.22) into Eq. (13.20), multiply
The resultant
through by ^(r,w), and integrate over space and lethargy.
equations for the coefficients in the expansion are
- d,(')
X,ay(*;r',u') =
is
i(i,u)
5)
fc(r>')
(13.27)
sec. 13.2]
PERTURBATION
THEORY
771
This equation shows the expected result that the coefficients of all the
higher modes decay exponentially in time (t > 0) and that the coefficient
Thus the ulti
of the fundamental mode is a constant after the burst.
mate form of the flux
long after the burst is
x(t,u;i'u')
(13.28)
4>0(r,w)
is
it
if,
We find, then, that the relative values of the adjoint function to the
neutron flux from one space point to another gives directly the ratio
of the magnitudes of the ultimate fluxes from bursts of neutrons intro
Again it may be seen that the adjoint function
duced at these points.
measures the importance of the introduction (or removal) of neutrons at
a given point in a reactor assembly.
It should be pointed out that this rather interesting approach to the
adjoint function is not of direct importance for developing useful results
In fact, this line of thinking
which will be obtained in this chapter.
for example,
can be overextended
taken to imply that an extra
source of neutrons can be compensated by an appropriate change in the
multiplication constant.
In the sections which follow we develop with the aid of various reactor
models the application of the techniques discussed above (as well as
of the effects of parameter
changes on the
is
In many cases
possible to use Eqs. (13.11) to find explicit forms
for the adjoint operators so that (13.13) may be written as an explicit
set of equations for \po(T,u) (along with appropriate boundary conditions).
In the general case this particular approach leads to the consideration
of the adjoint to the Boltzmann equation in which all the generality
might be preserved, including angular dependence of the functions.
On
it
it
Eq. (13.14),
is
symmetrically.
Consequently,
customary and proper to base perturbation theory and the calculation
of the adjoint function on the model used for the calculation of the flux
it
vMr',W)
itself.
13.2
REACTOR ANALYSIS
772
[CHAP.
13
fj,
9() =
(13 29)
(') exp
/"
j(u') exp
y(u) du
fo"
Jou
Eq. (4.162)]
1
(13.31)
is
is
importance
du
/*o
9^
dH
^ Q^
rut
rue
Jo
9f(") J(u) du
rut
Jo
elaborate
dU
(13.32)
PERTURBATION
sec. 13.2]
THEORY
773
the treatment
The limit
to
Mo
thermal
/
hv =
rut
$(w)
gf(M) j(w) du
du /
du
Jo
~ "i(M)
S9(M) + ?t(w) ^(M) *9(m)
rut
rut
Q*(u) j(m) du
Tut
?t(M)
Jo
v
may be
rut
fw
Tut
Jo
one integral,
rw
Jo
y)
~i
*?*U') du'
dH
(13.33)
The first term on the right-hand side of (13.33), involving the crosssection changes 5/3 and 67, is in the desired form; the last term on the
right corresponds to the last term on the right of Eq. (13.10) and will
We first look for the operators adjoint to the
be treated analogously.
of
the
differential
The only one that pre
equation (13.29).
operators
sents any difficulty is the differential operator d/du; following the require
ment of Eq. (13.11a), consider
fut
0
qt (M)
=
*9(M) du
But,
C**
0
d
du
Sq^
f"'d
Jo du
t?t(M)
5?(M)1 du
fu
d
*?() du 9f(w) du
/
Jo
(13.34)
(13-35)
t-
The numerator
of the
REACTOR ANALYSIS
774
[CHAP.
13
5q(u)
q1{u)
joy{u']
&q(u,)
q*(u')
vy{u)
du'] du']
~
n{u)
"(tt) Sq{u)
5q{u)
[?t(M) Tu
fa"
du
(13.36)
qfM
P(u) q'(u)
seen to be
n(u)
f"
of 3t(u)
is
Thus, by proper choice of the function q*(u), the entire last term on the
From (13.36)
right-hand side of Eq. (13.33) may be made to vanish.
(13.37)*
is
is
take
and
if
which
may
be
compared
we
J*
y(u') exp
jj
(13.38)
f(w) du =
/0"
"
analogous to (13.31),
have exchanged roles.
j is
which
7
is
it
be
/3(u")
du"]
du'
(13.39)
obtain
PERTURBATION THEORY
sec. 13.2]
775
multiplication constant,
Sv
5k
In this result,
fun
/
qi(u) 5/S(w) q(u) du
fun
5y(u) q(u) du
(13.40)*
In fact,
as mentioned
standing
is the mathematical
In
v.
,,
,.v .. ,..,,
(13.41)*
The behavior of the slowing-down density and its adjoint are presented
in Fig. 13.1. The thermal range is represented as a small interval in
lethargy just below u = w0; throughout this interval the functions are
rapidly varying. Changes in the absorption cross section, as seen from
Eq. (13.40), are weighted by the function q*(u)q(u); from Fig. 13.1 this is
seen to be nearly a constant over the lethargy interval below the fission
REACTOR ANALYSIS
776
[CHAP.
13
Lethargy
Thermal range
-|
-u0
is
spectrum
>
q(u)
in the intermediate
(13.42)
(13.43)
and
qy(u) q(u)
= constant
in lethargy
"statistical weight"
(13.44)*
is
it
/3
13.3
is
Fig. 13.1 The slowing-down density q(u) and the importance function g'(u) for the
infinite multiplying medium.
-V DV
2tf>
vXfpfi
(13.45)
sec. 13.3]
PERTURBATION
THEORY
777
In this equation the parameters D, 20, 2/; and pfi are, in general, func
tions of position; the parameter p/( is taken to represent the losses in the
fast range, so that in the application of any of the results of this section
P/i may be replaced by whatever estimate of the fast losses, leakage and
In the undis
absorption, is regarded as satisfactory in a particular case.
turbed homogeneous system, the unperturbed flux o(i) satisfies the
equation
-Z)oVJ0o(r) +
S>*,(r)
v^p^iT)
The solution to
of position.
(13.46)
*0(r) = cFo(r)
(13.47)
V*F0(r) + BlFa(T)
(13.48)
F(ft)
= 0
(13.49)
It
will
be of interest
gl
in connection
P/'
Do
boundary
of the reactor.
(13.50)
be dis
cussed here to consider the complete set of eigenf unctions FH(t) of the
bare reactor, of which F0(r) is the lowest mode. These functions are
solutions to the equation [cf. Eq. (8.260)]
V!F(r) + BlFn(t)
It
is convenient to construct
thus [cf. Eq. (8.249)]
v
= 0
with F(ft)
= 0
(13.51)
F(r) Fm(r)
dx =
Smn
(13.52)
meet
1 For
physical
situations
ANALYSIS
REACTOR
778
[chap.
13
2,(r) =
X'f + 2(r)
2}>
+ SS^r)
(13.53)
(r)
<fo(r)
+ *(r)
(13.54)
we must allow a
hypothetical
change in v.
f-+
The substitution of
hv
(13.55.)
these
+ ^TH{r)
vvpp,M{r) + p2,(t) P/I0(r) + v2P//*o(r)
20(r)
>(r)
(13.56)
1.
Orthonormal
13.1
Set ok Eioenfunctions
for the
(A.
*".(*)
cos Bx
sin Bx
n even
n odd
A.
n = 0, 1, 2,
BU
with i,
j,
= <? +
m) +
k = 0, 1, 2, .
1)
n*
Aijk
"I
,
and
my
2c^
(j +
1>
= (0JC)"i
and
... .
Jf(Ux) 3
kx
'
ifcos
.
[sin kx
1)
.
,
g(m<y)
"v
"
fcos m<y
cos ntz
(sin ntjy
sin
i,j,k
even
odd
sec. 13.3]
PERTURBATION
THEORY
779
3.
with i,
li =
= 0, 1, 2
7
2a
(i +
and
A.,
t i
For
4.
reactor
and k
1, 2, 3,
. .
,*
1)
|^[-^V^,(^U;+1(vyi)]|-i
t i
/Bi
f coa j6
j there
each value of
are
lain j9
two functions Fiit,
Jj
i. i even
/cos i<2
?, j odd
(sin /;2
one with cos j$, the other with sin j$.
,
Spherical reactor
FijkirM)
- Aiiltfik(r)
Bl
g<i(9)
Ay(*)
()"
Atik
~{'[-\
?o() =
Pj
(cos 9)
fcos
(sin
jV>
(*
j,-,
= 0,
1,
+*!)/->)!'}
the perturbed
Physics," McGraw-Hill
n=0
= cF(r) +
(13-57)
n=l
Here the contribution from the original flux has been separated out so
Note
that the last sum on the right of Eq. (13.57) represents 5<(r).
=
also that we have taken a0
c for convenience since the magnitude of
REACTOR ANALYSIS
780
[chap.
13
&
- =^
Sk
5k
fv
*I
*o(r)[vp/,8S,(r)
- Z.(r)]*,(r)
di
(13.58)*
-.
Importance function
Statistical weight
~
~
4>o(r)
4>l{r)
The same result will be obtained in Sec. 13.4, somewhat more generally,
and will be discussed in more detail there.
For the present section, we shall concentrate on the extra information
If Eq. (13.57) is multi
to be obtained by the eigenfunction expansion.
plied through by/rm(r), where m ^ 0, and integrated over the volume V, an
expression may be obtained for the coefficients an of the expansion for
With the application of Eqs. (13.50) and (13.52) the result may
6>(r).
PERTURBATION
SEC. 13.4]
written
be
a =
-f
THEORY
- 2.(r)]*0(r)
DM - be
F,(r)[irp2,(r)
781
dx
(13-6)
Since the F(r) are known, the specification of the perturbing functions
52/ and 52a will allow explicit evaluation of the coefficients a from the
and, therefore, also an infinite series representation for
flux <(r) may be conveniently given the same
The
perturbed
50(r).
normalization as <o(r) by requiring that
above relation
f
By
*(r) di
= 0
(13.61)
(i)
13.4
General System
a. General Formulation.
The treatment of the general reactor prob
lem in the one-velocity model using the method of adjoint functions is
based on the neutron balance relation (13.45).
It is understood that in
the present application all the quantities appearing in that equation are,
in general, functions of the spatial coordinates so that this equation may
be used to describe any and all regions of the reactor.
For convenience
in writing, we combine the two factors 2/ and p/( into a single function
2/i, where
2/((r)
- 2/(r) p(r)
(13.62)
-V
= *2/((r)
*(r)
(13.63)
in v to compensate
for a general change in any of the nuclear parameters, we first multiply
through Eq. (13.63) by some, as yet unspecified, function >f(r) and
integrate over the volume V enclosed within the outer extrapolated sur
face of the reactor.
The resulting equation when solved for v yields
L*f(r)[-V.D(r)
= }
V + 20(r)] *(r) dx
-.
Jv
dx
(13.64)
&v
&D(r) V + 52n(r)
Jr *t(r)
+
**(r)[-V
/"
13
'
resulting expression is
L
^
[chap.
ANALYSIS
REACTOR
782
Z(r)] *(r) dr
2(r) *(r) dr
D(r) V + 2(r)
jv
- r2,,(r)]a+(r) dr
(13.65)
The first term on the right side of this equation is of the desired form.
The immediate problem is to choose ^(r) according to the method out
lined in Sec. 13.1 so that the second term vanishes.
b. The Adjoint Function.
The first step in the analysis is the determi
nation of the operators adjoint to those within the brackets in the inte
The only one of these which pre
grand of the last term of Eq. (13.65).
In order to treat this term
sents any difficulty is the operator V D(r) V.
we introduce the following identities:
DV8dr
D
Jv 6V
dr =
Jv
- Jy (V^)
(5 D V^) dr - jy
(W
jyV
V5tf>)
dr
(V</)
(D V) dr
(13.66)
(D V*f) dr
(13.67)
jv
(W VS) dr
(8D
Vtf) dr
(4>W V80)
Jg
(8 D V^)
dS
(13.68)
dS
(13.69)
S,
D(r)
V5<Kr) dr =
80(r)
Jy
tf{r)
Jv
is
D(r) V^(r) dr
(13.70)
f(r)
is
is
if
ft
V,
0,
if
Jv
sec. 13.4]
PERTURBATION
THEORY
783
-V
But this
D(r) V*t(r)
^(r)
(13.71)*
it
is
Since
precisely the equation satisfied by the flux (13.63).
the boundary conditions specified above for
are identical to those for
must be concluded that
**(r)
*(r)
- 82(r)
'
6S/'(f)
U-
*C(r)V]^(r) dr
(13.72)*
Sk
'ir
- 5, =
from (13.65) as
lv
be calculated
With the above result the change in the multiplication constant may
rF2y,(r)*(r)dr
Importance function
Statistical weight
~
~
(r)
(13.73)
4>\r)
(13.74)
is
it
if
is
ness
sk
it
is
/v
is
In so far as the changes in the absorption and fission cross sections are
seen to be formally identical to that of Eq.
concerned, this result
(13.58); however, Eq. (13.72) applies more generally to a multiregion
reactor.
It must be concluded, then, that (for absorption and fission
changes) in the one-velocity diffusion approximation [cf. Eqs. (13.59)],
Z(r) *2(r) dr
Statistical weight
~ (V)2
(13.75)*
coefficient,
(13.76)*
784
REACTOR ANALYSIS
[chap.
13
In the case of a bare slab reactor, the statistical weight for absorption
and fission cross-section changes would vary across the slab as cos1 Bx
[from Eq. (13.74)], whereas the statistical weight for changes in the
These forms
diffusion coefficient would go as sin2 Bx [from Eq. (13.76)].
in
that
the
changes
This figure reveals
absorption
are shown in Fig. 13.3.
and fission cross sections are most important if made near the center of
the reactor; changes made near the edge are not as important since extra
neutrons there are likely to leak out and therefore unlikely to contribute
Very nearly the opposite behavior is observed
to the next generation.
for changes in the diffusion coefficient; changes near the center are
unimportant while changes near the edge contribute heavily.
We may
approach the understanding of this result by considering the introduction
of a small void at various points in the reactor (neglecting the effects
of the displaced absorbing and fissionable material).
The presence of
the void increases the effective diffusion coefficient in the nearby region,
and this, according to (13.75), will result in a reduction in the reactivity
of the system.
This is seen to depend on the fact that neutrons are
streaming from regions of higher importance to regions of lower impor
In
tance, and an enhanced diffusion coefficient will aid this streaming.
regions where the gradient of the flux is large, the increased diffusion
coefficient will be most effective.
At the center, where the gradient
vanishes, no finite increase in D will make the net current anything but
zero, and thus will not affect the reactivity.
This increased net current
due to an increased diffusion coefficient accounts for one factor of V<
in Eq. (13.75); the other arises from the fact that the change in impor
tance function due to streaming is proportional to the change in the flux
itself since the flux is self-adjoint in this case.
PERTURBATION
sec. 13.5]
THEORY
13.5
-V
-V
785
V*,(r) +
D,(r) V*,(r) +
Dx(r)
Zg>fr)
2i2>(r)
*i(r)
*,(r)
=
=
rZ,(r) +.(r)
pc(r) 2^(r) ^(r)
"
where the various symbols and subscripts have the previous definitions.
2, DiVfa, and DtVfa
In this analysis we require that the functions
i,
Thus
In this construction,
be
written
If
O,*
vO&
(13.79)
This equation is in the form of the general relation (13.1), and may there
fore be treated by the method outlined in Sec. 13.1.
The analysis begins by introducing a yet unspecified row matrix 4f :
441
(13.80)
Jy *t02* dr
Next,
(13.81)
REACTOR ANALYSIS
786
jv 8&(Oi
13
[CHAP.
8.
- f02)*
dx
Jy
*+(50i
+
Sv =
*+02*
02)4 dt
[ *f(Oi
v02)*dr
(13.82)
dx
0\ and
fy *HOi
Thus
that
we require
that
v02)*
*f
dx =
**(Olt
- vO\)*l dt
(13.83)
0{*l
Jv
vOfrl
(13.84)
_Sk_
/F
-,*0,)*dr
*'(<>!
Jv *+02*
dx
which
0\ and 02
D{V\,
and
4>2(ft)
= 0.
Then the adjoint operators 0[ and 0\ are the transposes of the operators
Oi and 02, respectively (note that all quantities are real here), so that
Eq. (13.84) becomes
-V-
D,V +
0
2^'
-v.
d2v + 2i2,j
-V
-V
ZMr) V*I(r) +
>2(r) V^(r) +
u\\
- "is,
oj
uj
(i3-8b)
[{x)
*l(r)
= pc(x)
2^>(r)
= v2,(r)
*Kr)
t(x)
be
'
PERTURBATION
SEC. 13.6]
Finally
THEORY
787
constant
is calculated
from
(13.85):
fv
*J*[Pc2,]*i dl
- fv &D2Vt
- fy <^Z02 dr +
Jy
V$s dr
dr]
4\SZf^
(13.88)*
It
13.6
is
-V
where
"i(u)
PM q(r,u)
q(r,u)
4>(r)
[|fr^U
J~
D(r,u)
f~
-V
For the present application we use the following form for the agediffusion equations:
5(u'
du'
(13.89)
(13.90)
utb)]
*">-
US
All remaining symbols and notation, including the "th," have the usual
definition.
spec-
8,
J.
is
1951.
REACTOR ANALYSIS
788
[CHAP.
13
-VDvir+zS
*2'
- J~ ( - Of-
Oi =
(13.92)
- I* (V
du
Z),hV
+ 2*){-
du
- *)J
(13.95)
to be put inside the braces and integrated over lethargy, with g{u) as
(13
Jo
#()
is
The notation
weighting function.
Next we multiply through Eq. (13.92) by an as yet unspecified matrix
and over the entire lethargy
integrate over the extrapolated volume
range, and solve for v.
du
(13.96)
dl
lv
jo"
_ Io"du fv*?0*dt
{K*(01
"
dl
(13 97)
now defined by
LxWdu
is
xf
If
'
du
- ^(r.t*)
(13.98)
fO0x*
sk
vOid
--- - -
/o
Olst
&v
as usual.
du
fv
lo
J-L"
Then,
xt(Cl ~ "0j)5x dl
where
xf
is
du
lv
/o"
Um
d|
q{t,u)
2)"(r){-
^u)
/Q"
du
T(r,t)(0
J()
/"
(13.93)*
Jv Kt0*
,k)]
df
(13.99)*
(13.100)
THEORY
PERTURBATION
sec. 13.6]
789
-V
D(r>M)
5(u
^(r,u) + 0(r,)
gt(r|)
- uj^r) + ry(r,)
AhV^(r)
Zf*t(r)
rZjNr)
_A
u ?t(r )
jT"
may be shown to
(13.101)*
^"
" Wl^A
2<(r,u)
#f
(13.102)*
<f
4>f,
0.
and gf,
tinuous,
integral
Jo
in just the sense discussed for the infinite medium Eq. (13.37).
The
general procedure would be to choose a normalized space form for the
j(w') qf(T,u') du'
and
proceed
to
obtain
from Eq.
>f(r)
is
is
"up"
is
(in energy) to u =
[note that as in Eq. (13.37) there
in sign of the lethargy derivative in (13.101) as compared
a change
with the
is
is
above
is
gf
the same, and the same network in lethargy and space variables may be
used.
Thus the calculation of the importance functions
and tf, in
this case, add very little to the time and effort already invested in obtain
In fact, the second calcu
ing the slowing-down density and the flux.
as
Blatt,
Dwork,
J. M.,
I.
J.,
253
Edlund, M. C, 690
Eggler, C, 637
Egiazarov, V. B., 642
Ehrlich, R., 521
Ergen, W. K., 601
Case,
K. M.,
Campbell, E. C, 557
Carlson, B., 441, 446
Carlvik,
689
Carslaw, H. S., 141
Carter, R. S., 697
J.
2,
I.,
I.,
NAME INDEX
249, 367
REACTOR ANALYSIS
792
I.,
J.
724
699
Kaplan,
Karman, T. von, 124
Keepin, G. R., 566, 568
King, L. D. P., 620
Kinney, W. E., 141, 760
Klema, E., 260
Kushneriuk, S., 724
J. J.
109
I.,
Kahn, H.,
J.
I.
Jaeger, J. C, 141
Johnson, M., 646
Jupnik, H., 639, 678
J.
5,
Halliday, D.,
8-9, 13
Halperin, J., 620
Harvey, J. A., 143-144, 315
Hellstrand, E., 642
Hendrikson, J. K., 681
Hildebrand, F. B., 76, 184
Hinman, G. W., 681
Hoffmann, F. de, 249, 367
Holmes, D. K., 150
Hughes, D. J., 143, 317, 633, 637, 697
Hurwitz, H., Jr., 253, 521, 757
NAME INDEX
Simon, A., 759
Siostrand, N. O., 557
Sneddon, I. N., 140
Snyder, T., 639, 678
SokolnikofT, E. S., 65
Sokolnikoff, I. S., 65
Sommerfeld, A., 200, 337
Soodak, H., 407
Spinney, K. T., 686
Stein, J. M., 546
Stelson, P. H., 557
Stoughton, R. W., 620
Sullivan, J. G., 760
Sweeney, D., 446
J. H., 657
Teller, E., 145, 704
Thie, J. A., 688
Thomson, D. B., 557
Tietjens, O. G., 593
Tittle, C. W., 260
Trubey, D. K., 285, 638
Tsien, H. S., 609
Generated for wjivans (University of Florida) on 2015-09-22 17:24 GMT / http://hdl.handle.net/2027/uc1.b4118385
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Tait,
793
Weisskopf, V. F., 2, 6,
317
Yamauchi,
H., 317
Zemach, A. C, 145
Zimmerman, E. L., 726
Zweifel, P., 757
SUBJECT INDEX
Absorption cross section, values for
moderators, table, 165
Absorption probability, 101
Activation foil, 229
Adjoint function, 768
for age-diffusion model, 789
conditions on, 782, 786
for one-velocity model, 782
orthogonality of, 770
to slowing-down density, 774
Adjoint operator, 768, 770
for age-diffusion model, 787
for Boltzmann equation, 771
self-adjoint, 769, 782, 786
for two-group model, 786
Age-diffusion theory (see Slowing-down
diffusion theory)
Albedo, definition, 193-195
determination of, 195
general expression for, 196
parameters involved in, 196
reflector properties in terms of, 418420
Barn, 41
millibarn, 46
Barrier penetration, 6
Base temperature, 312
Beryllium and beryllium oxide, thermal
neutron properties of, table, 165
for, in case of
small argument, 735
modified, 184, 280, 708
regular, 184, 207, 708
zeroes of, 207
Beta rays, 9
Binding energy, average per nucleon, 3, 7
chemical, 78
total, 2-3, 7
Boltzmann equation, 22, 25, 28, 269, 400
application to hydrogenous systems,
Bessel function, expression
752
REACTOR
796
formula, 673
Brookhaven graphite reactor, 699
Buckling, approximation for, in twogroup theory, 480-481
in completely reflected reactor, 479483
621
476
ANALYSIS
Control rods, black, 722
effect on net neutron leakage, 727
effective radius of, 724
Feinberg-Galanin method for, 705
gray, 722
shim, 25
in two-group theory, boundary condi
tions in, 728-729, 732-733, 740
central black rod, 728-732
eccentric black rod, 738-743
flux distribution for, 730, 731 (Fig.),
736, 737 (Fig.)
general equations for, 727
hollow central rod, 732-737
ring of black rods, 743-747
worth of, 726
xenon-override, 614
Converter plate, 244
Converting, 71
Coordinate system, center-of-mass, 72,
359
laboratory, 13, 72
Core, 14
Coulomb interaction, 8
Coupling equation in age theory, 287,
364-365, 421
CP-5 Reactor, 321-323
Critical experiments, 23-24, 557
Critical fuel concentration, 58
in diffusion theory, 212
in Feynman-Welton method, 491, 519
in reflected reactor, 419
Critical mass, 22, 213
calculations, 299, 321
Critical size, 299
comparison of, from Serber- Wilson and
diffusion theory, 452 (Fig. 8.18)
Criticality condition, in age theory, 292
in diffusion theory, 209
table, 217
in infinite homogeneous medium, 55,
121
SUBJECT INDEX
Cross section, for compounds, 48
correction to, for non-l/v variation,
144-145 (Fig. )
temperature, 137
effective (see Effective cross section)
escape, 222
fission,
42
inelastic, 41
macroscopic, 36, 46-48
measured,
133, 135
measurement of, 44
microscopic, 40
radiative capture, 42
reaction, 696
scattering, 42
slowing-down, 508
thermal neutron values for, table, 46
total, 44, 72
transport, 164, 697
Crow-flight distance (see Root-meansquare radial distance)
797
364
REACTOR ANALYSIS
798
668
End
228-229, 305
Energy groups, 421
(See also Multigroup method)
Energy loss, 69
from stationary nuclei, 74
Energy space, 30
Energy spectrum, neutron, 19, 69-71
Equivalence of homogeneous and hetero
geneous systems, 640, 671
accuracy of, 672
based, on flat-flux assumption, 673
on rational approximation to escape
effects,
probability,
672
Error function,
304
373
"Eyewash"
code, 760
SUBJECT INDEX
Fast effect, produced by cascade, 693696
720
799
sources,
119,
199,
218
spontaneous, 6
in Th", 693
in U"8, 693
Fission energy, average, 19
distribution for U"s, 22
table, 9
Fission fragments, 3, 8
beta decay of, 566
secondary, 620
Fission products, buildup, 614
in constant-flux reactor, 617
in constant-power reactor, 618
decay of, 610
following reactor shutdown, 614
I1",
611, 613
Ndl4, 612
Sml",
611, 613
286, 295
REACTOR ANALYSIS
800
253
260
37
Gamma ray, 2
internal conversion process, 39
prompt, 9
Gaps and gas channels, effect of, on age,
701
on thermal diffusion length, 698701
tables,
700, 701
on thermal utilization,
654-660
647, 707
181
Half-life, 38
Hankel function of first kind, 184, 189
Harmonics (see Spherical harmonics)
Heat-conduction equation, 24-25, 141,
273-274
Heavy water (DaO), 19
thermal neutron properties of, table,
165
SUBJECT INDEX
Heterogenous method, advantages of,
704
H,0
(see Water)
Homogeneous fuel-moderator mixtures,
equivalent to heterogeneous, 640
infinitely dilute, 634, 637-638
table, 638
of natural uranium and graphite,
626-627 (Fig.)
resonance integrals for, 637 (Fig.)
table, 638
Homogenized system, 62
Hydrogen, scattering from, 146-147
scattering cross section in water, 147
Hydrogenous systems, 123
age calculation in, as check on theory,
749, 759
801
REACTOR ANALYSIS
802
fluxes, 785
788
transposed, 786
Matrix representation of operators, in
age-diffusion model, 788
in two-group model, 785
Maxwell-Boltzmann
distribution, 110,
130-135
speed for, 142
most probable speed for, 150
Maxwell-Boltzmann
flux, 136
Mean free path, 36
for removal reaction, 459
Mean square radial distance, first-flight
corrected, 239
for Gaussian slowing-down source, 240
relation of, to neutron slowing down,
average
749
to
401
for thermal neutrons, 224
"Medusa"
code, 760
146
211
SUBJECT INDEX
Narrow resonance approximation (NR),
633-635, 664, 684-687
system, col
use of, in heterogeneous
lision density, 667
effective resonance integral, 670,
675
550
system, collision
density, 667
effective resonance
integral, 669,
677, 682
270-271
in
803
delayed, 9, 566
tables, 567, 568
density function, 22-24, 28, 53, 64, 72,
161
definition, 167-168
in diffusion approximation, 174
Neutron sink, 23, 610
Neutron temperature (see Effective neu
tron temperature)
Neutron wave phenomena, 279, 351
No-return current condition, 176
application of, to slab, 190, 383
to sphere,
386
Nucleons,
3, 7
804
REACTOR
Perturbation
349
ORACLE, 484
ORNL Research Reactor, 483-485
ORNL Swimming Pool Reactor, 480,
485-486
ORNL
18
352
Pi
764
absorbers,
ANALYSIS
systems,
756
P,
Radiation, beta, 9
electromagnetic, 12
Radiation damage, 244
Radiative capture, 12-13
Radioactive decay, constant, 37
by emission, of alpha particle, 38
of beta particle, 38
of e~, internal conversion of gamma
ray, 39
of gamma ray, 38
of K, orbital electron capture, 38
half-life, 38
scheme,
38
Radioisotope, table, 20
Reaction, absorption (capture),
deflection, 13
deuteron-deuteron, 557
fission, 31
14, 31
SUBJECT INDEX
Reaction, between neutrons and station
ary nuclei, 134
proton-triton, 557
scattering, 31
Reaction rate, 134, 137, 678
for 1 /v absorber, 140
Reactivity, 310, 571
maximum excursion of, 632
Reactor, breeder, 21, 692
central station power, 20
circulating fuel, 24
CP-1, 18
CP-5, 321
epithermal, 19
fast, 15, 19
Hanford, 21
heterogeneous,
homogeneous,
18, 27
18
poisons,
22
reflected
(see
research,
20
Reflected reactors)
stability, 22
temperature effects on, 308
start-up, 325
thermal, 19, 57, 70, 155
(See also specific reactors)
Reactor heat capacity, 578-579
effect on temperature overshoot, 588589
805
426
153
REACTOR ANALYSIS
806
surface-absorption
683
integral)
Root-mean-square radial distance, 224
for fast neutrons, 283-286
for thermal neutrons, 225
(See also Effective resonance
447
356
13, 317
80
in lethargy, 84
Scattering frequency function, in energy,
521
79
general,
125
521
78, 342
333, 354
for hydrogen, 99
in lethargy, 84
119
function, 162
monoenergetic, 119
thermal neutron, 219-221
186
SUBJECT INDEX
Spatial modes, leakage by, 496
Spatial trajectory, 30
Spectrum, thermal neutron, hardening of,
132-133
784
medium,
776
Surface-tension energy,
Thermal energy, 2, 83
Thermal equilibrium between neutrons
and nuclei, 127
Thermal fissions,
Thermal fuel, 8
131
Speed-space,
807
152, 293
369
413
808
REACTOR
Unit-cell model
cell model)
UNIVAC,
(see
Wigner-Seitz unit-
760
U', 143
U", 684-685
ANALYSIS
Velocity space, 132-133
Volume-absorption term, 639-641, 683
Volume fraction, 59, 155
Water (H20), 19
thermal neutron properties of, table,
165
Yield curve,
10
(Fig.)
THIS
DATE
"JOW
22
1954
M '65
RET.
OCT 7
JAN
2 6 1968
JUN
14
1966
JUN 17 IS??
Book Slip-20m-3,'60(A9205s4)458
Call Number:
Meghreblian, R.V.
Reactor analysis.
20U850
PHYSICAL
SCIENCES
WARY
204850
TK9202