Chapter 4 Reactive Systems

When reactors are present in the flowsheet under consideration, there

are more choices that must be made regarding how variables are
counted and how the balances are formulated.

In previous chapters a pattern was established of beginning with a section

on tools and techniques and then presenting collections of sample
flowsheets in which the application of the new tools is illustrated. That
pattern is continued here. Since the number of new tools and techniques
required to deal with reactive systems is extensive, section 4.1 consists of
five sub-sections.
Section 4.1.1 introduces the additional terms and tools necessary to
analyze flowsheets with reactors. Terms defined here include
stoichiometric coefficient, reaction rate, fractional conversion,
limiting reactant, stoichiometric air, and excess air fraction. The
formalism for dealing with more than one reaction is developed in
section 4.1.2. When multiple reactions occur a fractional yield may
be given as a specification. The issue of reaction independence also
arises.

Element balances find use under special circumstances explained in

section 4.1.3. The concept of the atom matrix is introduced along
with the question of the independence of the element balances.
In section 4.1.4 a technique is introduced which permits the
identification of an independent set of reactions given only the atom
matrix for a system. This technique is referred to as the
stoichiometric procedure. Lastly, the issue of dependence in
reactor specifications is presented in section 4.1.5 along with a
method for detecting combinations of reactor specifications which do
not represent independent information.

Section 4.2 is a collection of flowsheets in which a single reaction

occurs, with and without recycle. The flowsheets in section 4.3
contain a single reactor but multiple reactions occur. The focus is on
the formalism needed to deal with material balances for such systems.
Section 4.4 is a collection of problems in which element balances are
effectively used. Some issues dealing with fossil fuels are also presented.
 In section 4.5 several flowsheets are analyzed in which no reactions are
given for reactors with a complex set of products in the exit stream. The
stoichiometric procedure is applied in each of these systems to obtain an
independent set of reactions.

4.1 Fundamental Tools

4.2 Single Reactions

4.3 Multiple Reactions

4.4 Unspecified Stoichiometries

4.5 Formal Stoichiometries

4.1 Fundamental Tools

A generalized approach to stoichiometry makes available a set of tools

that are useful in dealing with reacting systems. This section summarizes
the techniques needed to count and to formulate meaningful species or
element balances, and to interpret specifications that are unique to
reacting systems.

4.1.1 Single Reactions

4.1.2 Multiple Reactions

4.1.3 Element Balances

4.1.4 Generating Independent Reactions

4.1.5 Dependence in Reactor Specifications

 4.1.1 Single Reactions

The terms stoichiometric coefficient, reaction rate, conversion, excess

fraction and limiting reactant are defined here. Topics related to the
combustion of fuels are discussed, in particular the concepts of an
ultimate analysis, stoichiometric air ratio and excess air fraction. The
counting of variables is reviewed for the special case of a single
reaction.

Stoichiometric The coefficients in a balanced chemical reaction are referred to as

Coefficient stoichiometric coefficients. Any chemical reaction among S
components can be symbolized as

S
0=
∑ σ s⋅ C
s
s =1

where Cs is the chemical symbol for species s. For example, the

reaction to form ammonia is N2 + 3H2 = 2NH3. It can be also be
written
0 = 2 ⋅ NH − N − 3 ⋅ H
3 2 2

You will note that the coefficients for products are positive, while
those for reactants are negative. For the above reaction: σNH3 = 2,
σN2 = -1, σH2 = -3.
Reaction Rate Let the molar flowrates of species s into and out of a reactor be ns,0 and
ns,1 respectively. Then the reaction rate r is independent of the species
selected and is defined as
n −n
s, 1 s, 0
r=
σs
For example:

We can compute the rate of the reaction using any species for which
information is known in both reactant and product streams.

Using hydrogen: r = [(28 - 40) mol/h] /(-3), nitrogen: r = [(8 -

12) mol/h]/(-1) or ammonia: r = [(8 - 0) mol/h] /(+2) the same
reaction rate r = 4 mol/h is obtained.

Limiting The reactant species for which the quantity [ ns,0 / (-σs) ] is a
Reactant minimum is called the limiting reactant.

For the reactor above, there are two reactants. [ nH2,0 /(-σH2)] =
40/3 while [ nN2,0 / (-σN2) ] = 12/1, so that for the feed shown, N2 is
the limiting reactant.

Note that the term "limiting reactant" is only defined for a fully specified
set of feed streams.
Fractional The term fractional conversion applies only to reactants. Using inlet
Conversion and exit streams numbered 0 and 1 respectively, the fractional
conversion of species s is defined as
n −n
s, 0 s, 1
X =
s n
s, 0

40 − 28
For example, the fractional conversion of H2 is XH2 := = 0.3
40
12 − 8 1
The fractional conversion of N2 is XN2 := =
12 3

Multiple Feed, If a reactor has more than one feed stream and several product streams
Product Streams the definitions presented above can be unambiguously extended.

For example, let the ammonia formation reaction occur N2 + 3 H2 = 2

NH3 in a reactor with I feed streams and J product streams numbered as
in the figure above. Then the definitions of reaction rate, fractional
conversion, and limiting reactant become:

I+ J I

∑ n
s, i
−
∑ n
s, i
i = I+ 1 i =1
r= s = H2, N2, NH3
σs
 I I+ J

∑ n
s, i
−
∑ n
s, i
i =1 i = I+ 1
X =
s
s = H2, N2
I

∑ n
s, i
i=1

The limiting reactant is the reactant species for which the following
quantity is a minimum.
I

∑ n
s, i
i =1
s = H2, N2
−σ s

Combustion of Many industrial processes involve the combustion of a fuel, usually with air.
Fuels Many different methods for designating the amount of air used have been
devised. These are reviewed here, first for simple fuels consisting of a
single, well-defined species and then for complex fuels such as the fossil
fuels: coal, oil.

CH4 Combustion The stoichiometric amount of oxygen required to burn a fuel is defined to
be that amount necessary to convert all carbon in the fuel to CO2, all
hydrogen to H2O and all sulfur to SO2. If the fuel contains any nitrogen, it
is assumed to leave as N2 in the product stream. Note, that this definition
says nothing about the reactions that actually occur in the combustion
process. The "combustion" reaction is that defined above. For
example, the combustion reaction for methane is
CH4 + σ O2 = CO2 + 2 H2O

Notice that the coefficients for CO2 and H2O are determined by the
molecular formula for methane. The stoichiometric coefficient for oxygen is
determined by an O-balance. In this case σ = 2.
 Stoichiometric Now consider a fuel with an elemental composition given by a set of atomic
Oxygen coefficients α. Let the elements occur in the order C, H, S, O, N. So
that the fuel has the formula Cα0 Hα1 Sα2 Oα3 Nα4.

The balanced combustion reaction for one "mol" of fuel is then

Fuel + σ O2 = α0 CO2 + ( α1/2 ) H2O + α2 SO2 + (α4/2) N2

The coefficient for oxygen is obtained by an O-balance.

α1
(
2⋅ α 0 + α 2 +) 2
− α3
σ =
2

Stoichiometric The molar amount of air required per mol of fuel is simply the
Air Ratio stoichiometric coefficient of oxygen in the combustion reaction divided by
the concentration of oxygen in air. The stoichiometric air/fuel ratio β is
usually placed on a mass basis

(nair)
stoich stoich
fair M air σ M air
β= = ⋅ = ⋅
ffuel n fuel M fuel x M fuel
O2 , air

Ultimate Many substances are used in combustion equipment including fossil fuels
Analysis such as coal, oil, and "natural" gas as well as many agricultural residues
such as waste wood, corn stover, bagasse, peanut shells etc. Municipal
solid waste is often burned to recover energy. These substances are
typically characterized by their elemental analyses on a mass basis.

Chemical To convert the elemental analysis to a chemical formula, simply divide the
Formula mass fraction of each element by its atomic weight. This yields the
relative molar amounts of each element. Only the ratios of these
quantities are significant. To obtain a chemical formula, it is customary to
select one of two bases: (1) a fixed molecular weight or (2) a specified
coefficient for carbon, often six. This simply means scaling all the
coefficients to give carbon a coefficient of exactly six. Both of these
bases are illustrated in the following examples.
Example 4.1.1.1 A fuel oil has the chemical composition: C: 84.00 wt%, H:11.40%, S:
3.20%, N: 1.40%. Determine a formula for the oil on a C-6 basis and
the stoichiometric amount of air to burn this oil.

Define: Element indices (C, H, S, O, N) e := 0 .. 4

Coefficient of element e in chemical α e

formula for oil

Data: Ultimate analysis Atomic weights

 0.8400   12.011 
   
 0.1140   1.008  gm
η :=  0.0320  A :=  32.064  ⋅
 0.0000   16.00  mol
   
 0.0140   14.007 

Desired coefficient of carbon α C := 6

Oxygen concentration in air xO2 := 0.21

gm
Molecular weight of air M air := 28.97⋅
mol
Example 4.1.1.1 Chemical formula for oil:
 6 
(continued)  
η e A0  9.703 
α e := ⋅ ⋅αC α =  0.086 
η0 A e  0 
 
 0.086 
That is, the chemical formula for the oil is C6 H9.703 S.086 N.086.

Since the coefficient of C was set, the molecular weight of the oil is also
determined by this choice.
gm
M oil :=
∑ αe⋅Ae M oil = 85.793
mol
e
α1
(
2⋅ α 0 + α 2 + ) 2
− α3
Stoichiometric oxygen: σ := σ = 8.511
2

σ M air
Stoichiometric air/fuel ratio:β := ⋅ β = 13.686
xO2 M oil

Thus 13.686 g-air is required to burn one g of fuel.

Example 4.1.1.2 Repeat the above calculations but use a basis of a specified molecular
weight for the oil.
gm
Data: M'oil := 100 ⋅
mol

The elemental coefficients calculated need only be scaled to match the

desired formula weight.
 6.994 
 
M'oil  11.31 
α e := α e⋅ α =  0.1 
M oil
 0 
 
 0.1 
gm
M oil :=
∑ αe⋅Ae M oil = 100
mol
e
 The stoichiometric coefficient of oxygen in the combustion reaction has
changed because the formula for the oil has changed, but the air fuel ratio is
the same.
α1
Stoichiometric oxygen:
(
2⋅ α 0 + α 2 + ) 2
− α3
σ := σ = 9.921
2

σ M air
Stoichiometric air/fuel ratio:β := ⋅ β = 13.686
xO2 M oil

As before, 13.686 g-air is required to burn one g of fuel.

Excess Air When burning fossil fuels or agricultural waste materials to generate
Fractions energy, it is usual practice to use more than the stoichiometric amount of
air. The excess oxidant promotes a more intense combustion reaction
and makes it possible to achieve complete conversion of the fuel.
A common specification for combustion processes is the "percent excess
air". We will use the term excess air fraction defined as
n air
Φ= −1
stoich
n air

Note that an excess air fraction of 0.0 means that a stoichiometric

amount of air was used, while an excess air fraction of 0.1 means "10%
excess air", and a fraction 1.2 means "120% excess air".

Given an excess air fraction Φ, the actual amount of air used is then
computed using the stoichiometric amount of air for that fuel:

σ
⋅(1 + Φ ) = ⋅(1 + Φ )
stoich
n air = n air
xO2
Counting Each independent reaction adds one reaction rate variable to the total
Variables number of variables for the flowsheet. If a single reaction is occurring
then of course, only one reaction rate is needed.
In the reactor shown below, the conversion of H2 is 0.25. Compute all
the flow rates.

Item Reactor
__________________________
Variables
Stream 5
Rxn rate 1
--------------------------
Sp. Balances 3
Flowrates 2
Conversion 1
__________________________
DF 0

The DF analysis indicates that there is sufficient information to solve this

problem. The definitions of reaction rate and conversion are combined to
give
XH2⋅ n
0 , H2 10 mol
r := r := ⋅
−σ H2 3 hr

Now all the product flowrates can be computed from the species balances.

Let the species be indexed in the order H2, N2, NH3: s := 0 .. 2

mol mol
n := 40⋅ n := 0 ⋅  −3 
0, 0 hr 2, 0 hr
σ :=  −1 
n := 12⋅
mol  
1, 0 hr 2 
Species balances: n := n + σ s⋅ r  40 30 
n =  12 8.667 
s, 1 s, 0 mol
  hr
 0 6.667 
mol ≡ mole
 4.1.2 Multiple Reactions

When multiple reactions occur there is a need for a formal approach to

keep track of the stoichiometry. The stoichiometric matrix is defined
for this purpose. When counting variables, a number of reaction rates
equal to the number of independent reactions must be included. The
number of independent reactions is simply the rank of the stoichiometric
matrix. These issues are explored in this section. When multiple
reactions occur, an additional type of specification arises: fractional
yield. The definition of this quantity

Stoichiometric The stoichiometric coefficient of species s in reaction k is σs,k. This

Matrix two-index quantity is the stoichiometric matrix for the reacting system.
The extension of the species balances to multiple reactions is
straightforward. Each species is augmented by its weighted
participation in each reaction that occurs. For input stream 0, exit
stream 1 of a reactor in which R reactions occur with rates rk, we have
R
n
s, 1
=n
s, 0
+
∑ σ s , k⋅ r
k
k =1

The use of the stoichiometric matrix in performing material balances is

illustrated in the next example.
Example 4.1.2.1 One hundred mols/hr of an equimolar mixture of A and B reacts with
20% conversion of A to produce a mixture of A,B,C,E in which the
concentration of C is 1.5 times that of E. There are only two reactions
occurring:

A + B = 2C and B + C = E.

Stream Vars: 6
Equipment Vars: 2
---------------------
Balances: 4
Flows: 2
Conversion 1
Comp. ratio 1
_____________________
DF 0
It is clear that we have sufficient information to proceed. Try doing the
material balances on your own. Click on the text below to check your
work.
View the material balances
Independent In many systems, particularly at high temperatures, the precise pathway
Reactions by which reactants are converted to products is unknown or extremely
complex. These are vital issues for reactor design. But for material
balance purposes it is usually satisfactory to represent the overall
chemistry by some list of simple reactions. This list must be complete
enough to account for all products.
For the purposes of counting variables, the list of reactions must also be
independent. That is, no reaction can be a sum or difference of any other
subset of reactions. The precise number of such reactions needed to
represent the stoichiometry of any system can be determined from the
stoichiometric matrix. Let the number of nonzero columns obtained after
exhaustive column reduction of σ be symbolized by ρ(σ). We call this the
rank of σ. In performing a DF analysis for a multi-reactive system , we
count exactly ρ(σ) reaction rate variables.

Example 4.1.2.2 The steam reforming of methane results in the following reactions.

CH4 + CO2 = 2CO + 2H2

CO + H20 = CO2 + H2
CH4 + H20 = CO + 3H2
CH4 + 2H2O = CO2 + 4H2

There are five species (CH4,CO2,CO,H2O,H2) and four reactions, so the

stoichiometric matrix is 5 x 4.

 −1 0 −1 −1  1 0 0 0
   
 −1 1 0 1
 0 1 0 0

σ :=  2 −1 1 0  σ :=  −1 −1 0 0 
0 −1 −1 −2  1 −1 0 0 
   
2 1 3 4   −3 1 0 0

The second form is obtained from the first by elementary column

operations. These include multiplying by a constant, or adding a multiple
of one column to another. The final form shows clearly that there are
only two independent reactions for this system, that is, r(s) = 2.
 The reactions represented by the second form of the stoichiometric
matrix are: CO + 3H2 = CH4 + H2O and C0 + H2O = CO2 +
H2.

For many systems it is easy to determine by inspection that a set of

reactions is independent. If each reaction contains one unique species
then each reaction is necessarily independent. In more complex cases,
the matrix reduction procedure may be required.

When a feed substance is converted to multiple products by more than one

reaction, it is common to specify the relative amount of some desired
product obtained. This new type of specification is called the yield.

Yield This ratio applies only to reactors with at least two competing reactions.
Let reactant A form desired product C by one reaction and one or more
other products (but no C) by competing reactions. The yield of C is
simply the actual amount of C formed divided by the maximum amount
of C that would have been formed if all the A converted went to
producing C. Again using streams 0, 1 as inlet and exit streams:

Y =
( C1 − nC0)
σ A⋅ n
C σ C⋅ ( n −n )
A1 A0

The s's are the stoichiometric coefficients in the balanced reaction that
produces C. Recall that such coefficients are defined to be negative
for reactants, so that the ratio above is always positive.
 4.1.3 Element Balances

For most systems, the species balances give rise to algebraically

simpler equation sets than do the element balances. The material
balances for a special class of problems, however, can be simpler
when phrased in terms of element balances. This section explores the
issues involved in setting up and solving element balances.

Elements are In reacting systems, species are not conserved and the balance
Conserved equations must reflect this. Recall that the species balances for reacting
systems have terms involving the stoichiometric matrix and reaction
rates.
In contrast, elements are conserved. But the balances do not reduce to the
unrestricted form of the non-reactive species balances. The elements are
"packaged" into distinct species, and our material balance variables are the
flowrates of these species. The balances must be expressed in terms of
species flowrates but reflect element conservation.

Atom Matrix The formal method of dealing with this is to define an atom matrix. The
element αes of this matrix is simply the coefficient of element e in the
molecular formula for species s. The element balances merely tally the
amount of each element accompanying each species entering and leaving
the system. If there are I input streams, and J exit streams in a system
containing S species consisting of E elements, then the balances are
J S I S

∑ ∑ α e , s⋅ n
s, j
=
∑ ∑ α e , s⋅ n
s, i
e = 1 ... E
j=1 s =1 i =1 s =1

There is one such equation for each element. The formidable double-sum
form of these balances hides an underlying simplicity: each element is
conserved. In practice, the equations are very simple to formulate, as
shown in the next example.
Example 4.1.3.1 Ethane is burned with oxygen to produce carbon dioxide, carbon
monoxide, methane, and water. When a 3:1 ratio of oxygen to ethane is
used, the ratio CO2/CO in the product is 3:1.

DF = 6 - (3 + 2 + 1) = 0 where we have three element balances, two

flowrates, and one compostion ratio. The problem is well posed. Note
that we do not count any reaction rate variables when doing element
balances. The element balances are
Carbon n CO2 + n CO + n CH4 = 2 ⋅ n C2H6

Hydrogen n H2O + 2 ⋅ nCH4 = 3 ⋅ nC2H6

Oxygen 2 ⋅ n CO2 + nCO + nH2O = 2 ⋅ nO2

Substituting nCO2 = 3 nCO and solving the resulting system of three

unknowns by elementary methods we obtain the solution:
7 mol 21 mol 2 mol 41 mol
n CO := ⋅ n CO2 := ⋅ n CH4 := ⋅ n H2O := ⋅
15 hr 15 hr 15 hr 15 hr

Things to First, notice how many terms are involved in the element balances even
Note About for a unit with only one inlet and one exit stream. The balances are very
Ex4.1.3.1 easy to write down, because element conservation is a simple concept.
But the equations are usually algebraically less tractable than the
corresponding species balances because typically many terms appear in
each equation.
Independent If there are E elements in a system, there will be E element balances that
Element can be written. Ordinarily, these will be independent. But for some
Balances systems this isn't true. This occurs frequently enough that a method of
counting independent balances is needed.
The number of independent element balances for a system with atom
matrix α is simply ρ(α), the rank of the atom matrix. The rank of the
atom matrix is the number of nonzero rows remaining after exhaustive
row reduction. The technique is illustrated in the next example.

Example 4.1.3.2 Urea is produced by reacting ammonia with carbon dioxide. When a
feed consisting of 1/3 CO2 is reacted with 99% conversion of the carbon
dioxide, the product stream contains the species shown in the figure. Is
this problem well-posed?

The problem can be well-posed only if there are four independent

element balances. DF = 7 - [ ρ(α) + 1 + 1 + 1] Numbering the
species in the order listed for the product stream, and numbering
elements in the order (C,O,H,N) the atom matrix is

 1 0 1 1 0

1 1 2 2 0
α := 
4 2 6 0 3
2 
 0 2 0 1

Elementary row operations can be used to reduce the atom matrix to a

form in which there is a zero row. Then ρ(α) = 3 and the problem is
underspecified from this viewpoint.

 1 
0 1 1 0
0 1 1 1 0
α :=  
0 0 0 2 −1 
0 
 0 0 0 0 
Summary In general, a good rule of thumb is to always use species balances for
non-reactive systems. The number of independent species balances is
always S, the number of species, and S >= ρ(α) for any system. Also,
the species balances are usually easier to solve.
For reactive systems, when ρ(σ) = S - ρ(α), it is feasible to use either
species or element balances. When this condition does not hold, use
species balances. In other words, use element balances only when
they offer the possibility of an obvious simplification.
Element balances are only essential when a reactant is a complex
mixture such as a fossil fuel or agricultural product such as wood, corn
stover, or bagasse.
 4.1.4 Generating Independent Reactions

Species balances are preferable to element balances in most cases. But

what if there are no reactions given? Suppose that all we know about
the system is the set of species present. The species balances require
that we have a set of independent reactions in order to formulate
meaningful expressions for the product flow rates. This section
discusses how to generate such a set of reactions knowing only the atom
matrix. The resulting technique is called the stoichiometric procedure.
Stoichiometric Procedure: α ==> σ
Stoichiometric With knowledge only of the species present in the system we can write
Procedure the atom matrix α for the system. This is an array with E rows and S
columns, where E is the number of elements present and S is the number
of species.
To determine a set of independent reactions for this system, use
elementary row operations to put α in the form
I β
α = 
0 0 
I is an identity matrix of size ρ(α), and β has dimensions ρ(α) x [S -
ρ(α)]. The stoichiometric matrix is then an S x [S - ρ(α)] matrix of the
form
 −β 
σ = 
 I 

The next example illustrates the procedure.

Example 4.1.4.1 This "alphabet soup" system consists of five species A,B,C,D,E and four
elements w,x,y,z. The molecular formulas of the species are reflected in
the atom matrix shown below.
A = w x y3
B= y2 z
The system consists of the species: C= w y
D = w2 y4 z
E = x y2

 1 0 1 2 0

1 0 0 0 1
It is represented by the atom matrix: α =
3 2 1 4 2
0 
 1 0 1 0

 1 0 0 0 1 
0 1 0 1 0
The atom matrix can be row-reduced to α =  
0 0 1 2 −1 
the form: 0 
 0 0 0 0 

0 1 
We identify a 3x3 identity matrix in the
β = 1 0 
upper left, and the matrix β is:  
 2 −1 
0 −1 
 
The stoichiometric matrix is obtained by  −1 0

taking the negative of b and appending anσ =  −2 1 
identity matrix of size [S - ρ(α)] or 2. 1 0 
 
0 1 

This matrix corresponds to the two reactions: B + 2C = D and A = C +

E.
Additional examples are given by the ammonia oxidation system of
section 4.5.1 and the alcohols from syngas system of section 4.5.2.
Summary When a system is characterized by knowledge only of the species
present, a set of independent reactions can easily be found by the
stoichiometric procedure.
4.1.5 Dependence in Reactor Specifications

Once a set of R independent reactions has been specified for a reactor,

information given in the specifications is then used to fix the reaction
rates. Fixing a reaction rate requires that at least one species be
completely specified: that is, both its input and output flow rates must be
known. Some combinations of completely specified species do not
represent independent information. In this section a test is presented to
detect when this situation occurs.

Example Let the combustion of methane with pure oxygen be modeled with the
following two reactions:
CH4 + 2 O2 = CO2 + 2 H2O
2CH4 + 3 O2 = 2CO + 4 H2O

The methane is completely burned to produce an oxygen-free stack gas

consisting of CO2, CO, and H2O.
The two reactions are clearly independent since each involves a unique
species. There are five balances and a basis can be selected. We will
use the methane flow rate. This means that one additional specification
can be made. The same conclusion is drawn from a DF analysis.
DF = [5 + 2] - [5 + 1] = 1

Let this additional specification be the water flow rate. The two species
that are completely specified are methane and water. Let the methane
flow rate be 1.0 mol/h and that of water be 2.0 mol/h. Their balances are

CH4: 0 = 1 - r1 - 2 r2

H2O: 2 = 2 r1 + 4 r2
By an obvious rearrangement, the water balance is seen to be a multiple
of the methane balance, and is therefore not an independent specification.
Criterion for For species balances, a specification is independent if the rows of the
Independence stoichiometric matrix corresponding to the completely specified species are
linearly independent.
In the above example, the stoichiometric matrix with species ordering
methane, oxygen, water, carbon dioxide, carbon monoxide is:

 −1 −2 
 
 −2 −3

σ = 2 4 
1 0 
 
0 2 

The rows corresponding to the completely specified species ( methane and

water ) are clearly not linearly independent. The row for water is just (-2)
times the row for methane.
 −1 −2 
σ cs :=  
2 4 

Note that the original problem can be solved if another specification is

selected. For example, a flowrate of CO2 or CO, or a CO2/CO ratio
would be valid specifications.
4.2 Single Reactions

Each flowsheet in this section involves a single reaction. The sample

problems are intended to give you practice in setting up a
stoichiometric matrix and using an assortment of specifications to find
the product composition.
4.2.1 Chlorine Dioxide Synthesis

4.2.2 Hydration of Ethylene

4.2.3 Acetic Acid Synthesis

4.2.1 Chlorine Dioxide Synthesis

Problem Chlorine dioxide is a bleaching agent produced by the following

Statement reaction. 6 NaClO3 + 6 H2SO4 + CH3OH = 6 NaHSO4 + 6 ClO2
+ CO2 + 5 H2O

If 14 mols of the chlorate stream are fed for every mol of methanol and a
conversion of 90% is achieved, what is the composition of the product?

Click on the icon at the left to see the DF analysis for this problem.

Click on the underlined text below to see the material balances.

View the material balances for the chlorine dioxide synthesis.

4.2.2 Hydration of Ethylene

Problem The catalytic hydration of ethylene is carried out with a single-pass

Statement conversion of only 4.5%. The separator removes all the alcohol and
the recycle gas contains 6.5 mol% steam. The molar ratio of steam to
ethylene in the mixed feed to the reactor is 0.55. Calculate the
compositions of all streams, and the ratio of steam to ethylene fed to the
process.

On Your Own Click on the underlined text below to check your DF analysis.

View the DF analysis for the hydration of ethylene.

Complete the material balances for this problem, then check your
work by clicking on the underlined text below.

View the material balances for the hydration of ethylene.

4.2.3 Acetic Acid Synthesis

Problem The recycle reactor shown below oxidizes ethanol to acetic acid with a
Statement mixture of dichromate and sulfuric acid. A 90% overall conversion is
achieved with a recycle to ethanol feed ratio of 1:1, a 10% excess of
sulfuric acid, and a 20% excess of sodium dichromate. The recycle
stream contains only ethanol and sulfuric acid. The concentration of
acid in the recycle is 94 mol%. A perfect separation of acetic is
achieved. The reaction occurring is
3C2H5OH + 2NaCr2O7+ 8H2SO4 =
3CH3COOH + 2Cr2(SO4)3 + 2Na2SO4 +
11H2O

On Your Own Perform a DF analysis on the acetic acid synthesis flowsheet. Check
your work by clicking on the underlined text below.

View the DF analysis for the acetic acid synthesis

Now do the material balances in your worksheet. Check your work

by clicking on the underlined text below.

View the material balances for the acetic acid synthesis.

4.3 Multiple Reactions

The flowsheets in this section all involve multiple reactions. These

problems focus on calculating the reaction rates from given specifications
and then using those rates to compute final product compositions.

4.3.1 Acetaldehyde Synthesis

4.3.2 Carbon Disulfide Synthesis

4.3.3 Reformate Hydrodealkylation

4.3.4 Toluene Dealkylation

4.3.1 Acetaldehyde Synthesis

Problem Acetaldehyde can be produced by the catalytic dehydrogenation of

Statement ethanol. There is however, a parallel reaction that produces ethyl
acetate.
C2H5OH = CH3CHO + H2
2C2H5OH = CH3COOC2H5 + 2H2
In a particular reactor, an ethanol conversion of 95% is obtained with
an 80% yield of acetaldehyde. Calculate the product composition
assuming the feed consists of pure ethanol.

On Your Own Try performing the DF analysis for this problem. Check your work by
clicking on the exercise icon at the left.

Now compute the product composition. Click on the underlined text to

check your work.

View the Acetaldehyde Synthesis material balances.

4.3.2 Carbon Disulfide Synthesis

Problem Vaporized sulfur is reacted with methane to produce CS2. There are
Statement three reactions known to occur.
CH4 + 4S = CS 2 + 2H2S
CH4 + 2S = CS 2 + 2H2
CH4 + 2H2S = CS2 + 4H2
A feed containing four mols of sulfur per mol of methane results in a
product mix characterized by 90% methane conversion and 70% sulfur
conversion. Calculate the product composition.

On Your Own Complete a DF analysis for this problem. Click on the underlined text
below to check your work.
View the DF analysis for the Carbon Disulfide synthesis.

Do the material balances for this problem. Click on the underlined text
below to check your work.

View the material balances for the Carbon Disulfide synthesis.

4.3.3 Reformate Hydrodealkylation

Problem The process of removing alkyl groups from aromatic compounds with
Statement hydrogen is called hydrodealkylation. In the process shown below, a
refinery reformate stream consisting of 5 mol% benzene (B), 20%
toluene (T), 35% xylene (X), and 40% mesitylene (Y) reacts with
hydrogen. When 5.0 mols of hydrogen is used per mole of
hydrocarbon feed, the following conversions are obtained: T:80%,
X:74%, Y:13%. The product stream also contains 0.1% biphenyl (P).
The reactions occurring are:

C6H5CH3 + H2 = C6H6 + CH4 (1)

C6H4(CH3)2 + H2 = C6H5CH3 + CH4 (2)

C6H3(CH3)3 + H2 = C6H4(CH3)2 + CH4 (3)

2C6H6 = (C6H5)2 + H2 (4)

What is the composition of the product stream?

On Your Own Determine if the problem is well-posed.

Click here to check your DF analysis for the hydrodealkylation

Compute the composition of the product stream.

Click here to see the material balances for the

hydrodealkylation
4.3.4 Toluene Dealkylation

Problem The dealkylation of toluene to produce benzene is accompanied by a

Statement side reaction which produces biphenyl.

C6H5CH3 + H2 = C6H6 + CH4 2C6H5CH3 + H2 = (C6H5)2 +

2CH4

The side reaction requires that conversions be kept low in the reactor,
and this complicates the separation sequence. The first stage of
separation recovers pure benzene and produces two recycle streams.
The overhead consists of hydrogen and methane, part of which must
be purged to remove the methane. The bottom product of the first
separation stage is a liquid stream which undergoes further separation
into a pure toluene recycle stream, and a pure diphenyl byproduct.
When the reactor feed has a 5:1 ratio of hydrogen to toluene, a 75%
toluene conversion is achieved.
a. If the reactor outlet stream is 5% benzene, and 2% toluene,
calculate the fractional yield of benzene and the make-up (stream 2)
hydrogen required per mol of toluene fed to the process.
b. Repeat for a reactor outlet specification of 2% toluene and 58%
methane.
On Your Own In your worksheet, perform a degree of freedom analysis for part a.
Click on this text to check your DF analysis.

Continuing in your worksheet, compute all the material flows.

Click on this text to check your material balances.

Now repeat the DF analysis using the specifications for part b.

Click on this text to check your DF analysis for part b.

4.4 Unspecified Stoichiometries

The problems in this section focus on the application of element

balances to process flowsheets. Species balances are usually preferred
over element balances. Element balances can be useful, however, in
processes involving natural products or in cases where the reaction
stoichiometry is unspecified.
Natural products such as fossil fuels or agricultural residues are
complex mixtures of materials with a wide range of molecular weights.
Their composition as a mixture of distinct species is too complex to deal
with effectively for material balance purposes. Instead, balances are
done at the element rather than the species level.
The reactions involved in some processes, even those involving only
simple, distinct species, are often unknown or very complex. Element
balances are sometimes useful for these systems.

4.4.1 Fuel Oil Combustion

4.4.2 Coal Gasification

4.4.3 Fuel Oil Desulfurization

4.4.4 Formaldehyde Synthesis

4.4.1 Fuel Oil Combustion

Problem A fuel oil with analysis C:84.0%, H:11.4%, N:1.4%, S:3.2% is burned
Statement with air. The stack is found to contain 2130 ppm by wt (dry basis) of
sulfur dioxide. Determine a chemical formula for the oil on a C-6 basis,
and compute the percent excess air that was used in the burner.

On Your Own Perform a DF analysis for the oil burner.

Click on the this text to see the DF analysis for the oil burner.

On your worksheet, solve for the percent excess air.

Click on this text to view the material balances for the oil
burner.
4.4.2 Coal Gasification

Problem A dry coal consisting of ash:3%, N:1.5%, S:0.6% and unknown

Statement amounts of C, H, and O is gasified with steam and air to produce a fuel
gas. The dry-basis composition of the fuel gas is H2:23%, CH4:3.2%,
CO:16.2%, CO2:12%, H2S:1%, and N2:44.6%. In addition to the fuel
gas, a solid residue is produced consisting of 10% C and the rest ash.
If a 1:1mass ratio of steam/coal is fed, calculate the air rate and
moisture content of the fuel gas.

On Your Own Perform the DF analysis for this system.

Click on this text to check your DF analysis for the coal gasifier.

Now complete the material balances.

Click on this text to see the material balances for the coal
gasifier.
4.4.3 Fuel Oil Desulfurization

Problem A fuel oil with analysis C:84.0%, H:11.4%, N:1.4%, S:3.2% is

Statement desulfurized by contact with a hydrogen-rich gas ( 44 mol% H2, 56%
CH4) in a catalytic reactor. The desulfurized oil contains only 1 ppm N,
9 ppm S. The gas stream exiting the reactor contains H2, CH4, NH3,
H2S. The methane to hydrogen ratio in the off-gas is 7:3, while the
ammonia:hydrogen ratio is 1:4.

The reactor off-gas is subjected to two stages of purification. The first

stage removes 90% of the hydrogen sulfide and all of the ammonia. The
second stage removes all the remaining H2S. The recycle stream is 40%
H2. It is mixed with fresh gas containing 48% H2. All gas compositions
are in mol %, while the oil compositions are in wt%.
If 60 lb oil are fed to the reactor per mol of gas, calculate the product oil
composition.
On Your Own Perform a DF analysis for this system.
Click here to check your DF analysis of the oil desulfurization

Now complete all the material balances.

Click here to check your material balances for this process.

4.4.4 Formaldehyde Synthesis

Problem Formaldehyde can be produced by partial oxidation of methanol over a

Statement silver catalyst. Formic acid, hydrogen, carbon monoxide, carbon
dioxide and water are also produced. An overall yield of formaldehyde
of 2/3 is achieved in the process shown below. The reactor product is
scrubbed to separate the light gases, and the residue is distilled to
produce pure methanol for recycle. The scrubber and distillation units
are combined into a single separator in the flowsheet. The reactor inlet
stream is 35 mol% methanol and the off-gas composition is H2: 20.2%,
CO2:4.8%, CO:0.2%, N2:74.5%, O2:0.3%. Calculate the
compositions of all streams.

On Your Own Perform a DF analysis for this flowsheet.

Click here to see the DF analysis for the formaldehyde process.

Complete the material balances for the formaldehyde process.

Click here to view the material balances for this process.

4.5 Formal Stoichiometries

This section deals with problems in which species balances are used in
preference to element balances even if no reactions are given.

We have two reasons for preferring species balances to element

balances. There are always as many or more species balances as there
are element balances. Also, the species balances are usually
algebraically simpler. If the species present in the system are known,
then the atom matrix for the system can be constructed. The
technique of generating a feasible stoichiometric matrix from the atom
matrix, and thus an independent set of reactions, was discussed in
section 4.1.4. The problems in this section illustrate the use of this
technique.

4.5.1 Ammonia Oxidation

4.5.2 Alcohols from Syngas

4.5.3 Formaldehyde Synthesis

4.5.1 Oxidation of Ammonia

Problem A feed stream formed from air and ammonia (20 mol air/mol NH3)
Statement reacts to produce a complex mixture consisting of H2O, O2, N2, NH3,
NO, and NO2. When an 80% conversion of ammonia is achieved, the
product stream contains 3% O2 and 6% NO. Calculate the
compositions of the other species.

On Your Own Perform a DF analysis on the ammonia oxidation reactor.

Click here to check your DF analysis.

Continuing with your worksheet, do the material balances.

Click here to check your material balances.

4.5.2 Alcohols from Syngas

Problem When a mixture of CO and H2 (syngas) with a H2:CO ratio of 3.2 is

Statement reacted over a nickel catalyst, a mixture is produced that consists of
hydrogen, carbon monoxide, water, methane, ethane, propane,
methanol, ethanol, and carbon dioxide. When the conversion of CO is
97.9% and that of hydrogen is 73.8%, the product gas is characterized
as follows: CO2/CO = 2.0, C2H5OH/CH3OH = 0.05, C3H8/CH4 =
0.01, C2H6/CH4 = 0.05. Find the composition of the product gas using
only element balances. Repeat using only species balances.

On Your Own Perform a DF analysis for this process.

Click here to see a DF analysis of the alcohols from syngas
process.

In your worksheet, solve for the composition of the product gas using
only the element balances.
Click here to see the element balances for this process.

In your worksheet, solve for the composition of the product gas using
only species balances.

Click here to see the species balances for this process.

4.5.3 Formaldehyde Synthesis

Problem Formaldehyde can be produced by partial oxidation of methanol over a

Statement silver catalyst. Formic acid, hydrogen, carbon monoxide, carbon
dioxide and water are also produced. An overall yield of formaldehyde
of 2/3 is achieved in the process shown below. The reactor product is
scrubbed to separate the light gases, and the residue is distilled to
produce pure methanol for recycle. The scrubber and distillation units
are combined into a single separator in the flowsheet. The reactor inlet
stream is 35 mol% methanol and the off-gas composition is H2: 20.2%,
CO2:4.8%, CO:0.2%, N2:74.5%, O2:0.3%. Calculate the
compositions of all streams using species balances.

On Your Own Perform a DF analysis for this flowsheet.

Click here to check your DF analysis for the formaldehyde

process.

Complete the material balances for the formaldehyde process.

Click here to view the material balances for this process.

Chapter 5 Sequencing with Partial Solution

In all the flowsheets considered thus far we have implemented a

sequential-unit solution strategy. We were able to solve all the balances
for a selected process unit, then we moved on to the next unit. This
chapter deals with flowsheets for which this procedure breaks down.
Rather than abandoning the sequential-unit approach, we will augment
our strategy by proceeding with partial solution of one or more sets of
balances.
We follow the pattern established in previous chapters. The first
section defines relevant terms and explores the special issues associated
with this new technique. Subsequent sections apply the technique to
selected process flowsheets.

5.1 Extending the Sequential-Unit Approach

5.2 Simple Recycle

5.3 Perchloric Acid Synthesis

5.4 Methyl Iodide Synthesis

5.5 Acid-Gas Removal

5.6 Solving Non-Linear Equations

5.6.1 Numerical Techniques for Single Equations

5.6.2 Multiple Non-Linear Equations

5.1 Extending the Sequential-Unit Approach

For all the flowsheets considered thus far, our calculation sequencing
procedure has resulted in a partitioning of the material balances into
ordered subsets that were completely and separately solvable. We
simply identified a process unit with a degree-of-freedom of zero,
solved the balances for that unit, and then used the new information thus
obtained to reduce the DF for subsequent units to zero. This process
was repeated until all the streams were fully characterized.

Sequencing This chapter deals with flowsheets whose specifications do not permit
with Partial the balances for some unit(s) to be solved completely. Rather than
Solution abandoning the sequential-unit approach, we augment it with the
following technique.

1. We first solve the balances for all units that have a zero DF, if there
are any.

2. Then we consider the balances for that unit with the lowest DF. For
this discussion, let the DF for this unit be '1'. The positive value of DF
means that we will not be able to solve all the balances completely.
Instead, we will have to estimate the value of one carry-forward
variable, and express the solutions for all other computed quantities in
terms of this variable.

3. We then continue with the balances for the remaining process units.

4. Eventually, since the DF for the process is zero, we will have a

redundant balance that can serve as an "objective function". We can
vary the value of the estimated variable in order to satisfy this objective
function.

Before giving more details about implementing the procedure of

sequencing with partial solution, let us deal with some preliminary issues
that sometimes cause confusion.
An Alternative Instead of solving the material balances sequentially, unit by unit, we
Strategy could write down all the balance equations at one time and solve them
simultaneously. The fact that the DF for the process is zero means that
this procedure can be made to work. Modern calculators and tools like
Mathcad make the computational burden a manageable one. However,
there are some good reasons why this is not a good approach for hand
calculations.
1. There is the practical problem of selecting an independent set of
balances. Verifying that a set of three or four balances for one unit is an
independent set is much easier than trying to make this determination for a
much larger set of equations.
2. It is easier to locate an error in one equation (or in a small set of
equations) than it is to locate such an error in a large set of equations.
A complex flowsheet with a dozen process units and several species
may involve hundreds of variables. The sequential-unit approach is
predicated on the principle that a small set of equations is easier for a
human being to deal with than a large set of equations. This ability to
"debug" a solution is a compelling reason to favor the sequential-unit
strategy over the simultaneous-equation approach.
3. By dealing with the balances on a unit-by-unit basis, intuitive
approaches for a particular application are more likely to be employed.
These "obvious simplifications" are more likely to be recognized in the
sequential approach than in the simultaneous-equation approach.
Although it would be nice to believe that a solution to the general
material balance problem is available in some sense, it is still true that
leaps of intuition can often arrive at a solution with far greater efficiency
than a "general" method. Although we are trying to develop a
framework of tools to lessen the need for such intuitive approaches,
many flowsheets display unique features which can be profitably
exploited by an engineer with the skill to recognize them.
Hand The reader may have wondered why we continue to employ he term
Calculation "hand calculation" when we are clearly gearing our solution methods for
implementation by computing tools such as Mathcad, or similar
packages.
We define a hand calculation as one in which we formulate the balances
ourselves. Whether we use an abacus, slide rule, calculator, or
Mathcad to actually solve those balances is irrelevant. These tools only
handle the tedious mathematical manipulations required to solve the
equations. We still have to tell the tool what equations to solve.

There are software tools called process simulators which (with some
restrictions on the specifications) could formulate as well as solve all the
balances for us. All the user has to do is enter the data. Such a tool
permits fully automated solution of all the material balances. This
electronic book does not concern itself with such tools.

We return now to the issue of determining the value of a carry-forward

variable.

Non-Linear The final determination of the estimated variable is made by satisfying

Equations some objective function. The objective function is usually one of the
redundant balances that involves the process unit for which the stream
variable estimate was made. In one sense, the problem has been
reduced to solving f(x) = 0, where f(x) is some non-linear expression in
x.
Evaluating the Unlike the situation in which we have a single, equation to "plug into", it is
Objective usually the case that evaluating the objective function may involve
Function performing many intermediate balances. In the figure below, let x be the
carry-forward variable.

There is no distinction in a mathematical sense between a closed-form

expression, like f(x) = 3 x2 - sin(x) + 5, and a procedure that can
uniquely assign a value to our objective function, given a value of the
variable x. All that matters is that we can indeed evaluate the function.

Converging to Once we have estimated the value of the carry-forward variable, say x,
a Solution and selected an objective function f(x), we must determine the value of x
that satisfies f(x) = 0. Distinguish clearly the process of evaluating the
objective function and the process of solving the objective function.

One way to solve the equation is to simply guess values of x until we get
lucky enough to find one that makes f(x) =0. This method is called
solving by "trial and error". It is usually impractical to proceed by this
fashion.

A better method is to guess only once, and then use an updating rule to
successively refine the value of x, that is, converge to a solution. Many
such updating rules exist.

The Newton, Secant and Wegstein methods are examples of such

convergence techniques. They all employ some rule for updating the
current estimate of the solution. The Secant method is illustrated below.
Other iterative methods are discussed in section 5.6.
Secant Method To illustrate the technique, let us solve f(x) = x3 - 2 = 0. The answer is
the cube root of two. But let us converge to that value by starting with
an initial estimate of 1.0.
3
f ( x) := x − 2
x := 1
0 ( 0) = −1
f x

x := 1.5 f ( x ) = 1.375
1 1

Currently, we have two argument values x1 = 1, x2 = 1.5 and two

function values: f1 = -1, f2 = 1.375. Let us generate some new
estimates for x. We will use index k to keep track of them. The
updating rule for the secant method is

k := 1 .. 5 x :=
x
k− 1 ( k) − xk⋅ f (xk−1)
⋅f x
k+ 1 f (x ) − f (x )
k k− 1

The sequence of values generated is:  1 

 
 1.5 
 1.21053 
x =  1.25141 
 
 1.26027 
 1.25992 
 
 1.25992 
The last value of x is very close to the exact solution.
1
3 − 10
Err := x − 2 Err = −6.381 × 10
6

Summary The essence of the technique of carry-forward variables is simple.

When the DF for a unit is positive, introduce one or more carry-forward
variables. Values for these variables must be estimated.

To determine the actual values of the variables, complete sufficient

additional balances to permit evaluation of some objective function. If the
objective function is not satisfied within some tolerance, then vary the
assumed values of the variables by some convergence technique, and
re-evaluate the objective function. Repeat this procedure until the
objective function is satisfied to within the desired tolerance.
The sample flowsheet analyses in sections 5.2 - 5.5 illustrate the
technique with far greater clarity than any word description can afford.
5.2 Simple Recycle

Problem A feed of 1000 mol/h consisting of 1/3 A, 2/3 B is mixed with a recycle
Description stream and reacts with the stoichiometry A + B --> D. The conversion
of A per pass is 20%. The split fractions from the separator feed to the
recycle stream are A:0.80, B:0.90, D:0.10. Calculate the flow rates of
all streams.

Click on the diamond icon to copy the figure and the problem
statement to your worksheet.

On Your Own Show that the problem is specified correctly and determine a calculation
order. If any carry-forward variables are required, be sure to note
when they will be needed, and how they will be determined.

Click on this text to view the DF analysis for this problem.

Set up and solve all the material balances.

Click on this text to view the material balances for this problem.
 5.3 Perchloric Acid Synthesis

Problem Perchloric acid is produced by reacting barium chlorate with sulfuric acid.
Description
Ba(ClO4)2 + H2SO4 = BaSO4 + 2 HClO4

In the process shown below, an 80% conversion of barium chlorate is

achieved when the ratio of sulfuric acid to barium chlorate in the
combined reactor feed is 1:1.2. The feed (stream 1) contains 90 wt%
barium chlorate in perchloric acid. The recycle stream contains only
Ba(ClO4)2. Calculate all flowrates.

Click on the diamond icon to copy the figure and problem statement to your
worksheet.

On Your Own Show that the problem is specified correctly and determine a calculation
order. Note if any carry-forward variables are required.
Click on this text to view the DF analysis for this problem.

In your own worksheet, set up and solve all the material balances for this
problem.
Click on this text to check your material balances.
 5.4 Methyl Iodide Synthesis

Problem Two thousand lb/day of HI is added to an excess of methanol in the

Description reactor shown below. The compositions in the product stream are
given as mass fractions. The recycle stream is pure HI. What is the
recycle ratio (stream 5: stream 1) if a single-pass conversion of 30%
is achieved via the reaction

HI + CH3OH = CH3I + H2O

Click on the diamond icon to copy the problem statement and the figure to
your worksheet.

On Your Own Determine if the problem is specified correctly. Specify a calculation

sequence, and determine if any carry-forward variables are required.
Click on this text to see the DF analysis for this problem.

Solve for the required recycle ratio.

Click on this text to see the material balances for this problem.
 5.5 Acid-Gas Removal

Problem The absorber-stripper system shown in the figure removes CO2 and H2S
Description from a feed consisting of 30% CO2 and 10% H2S in an inert gas. A
proprietary solvent is used to selectively absorb the acid gases. The
absorber bottoms are flashed. The ratio of the concentration of CO2 in
the exit vapor (stream 6) to that in the liquid leaving the flash unit is
13.33. A similar ratio of 6.9 is measured for H2S. The liquid leaving
the flash contains 5% CO2, and the overhead contains 20% solvent. The
stripper overhead contains 30% solvent. The bottoms stream leaving the
stripper is pure solvent.

The feed is cleaned to the extent that stream 2 contains only 1% CO2
and no H2S. The splitter beneath the flash unit operates to produce
equal flowrates in streams 3 and 8. Compute the compositions of all
streams and the amount of solvent per mole of feed required.
On Your Own Is there sufficient data supplied to solve this problem?

Click on this text to see the DF analysis for this problem.

On your own worksheet, determine the flowrates of all streams.

Click on this text to view the material balances for this problem.
5.6 Solving Non-Linear Equations

Whenever a carry-forward variable, or tear variable, is required to

proceed with the balances for a flowsheet, some technique must be
implemented to converge on the correct value of the assumed quantity.
If only a single tear variable is needed, then there are several techniques
which can reliably converge to a solution. The Newton, Secant,
Successive Substitution, and Wegstein methods are discussed in the first
section below.
The Wegstein convergence method is easily extended to deal with the
case of multiple tear variables. This technique is discussed in section
5.6.2.
5.6.1 Numerical Techniques for Single Equations

5.6.2 Multiple Non-Linear Equations

5.6.1 Numerical Techniques for Single Equations

The techniques in this section involve a very general class of equations.

Any problem that can be phrased in the form f(x) = 0 can be treated by
the methods discussed here. Whether a numerical method actually
converges to a root, of course, depends heavily on the nature of the
function whose root is sought.

Closed-Form We first consider the case when the function f(x) can be written in closed
Objective form. Such functions arise frequently when physical systems are
Function simulated.

An Example Suppose we are asked to find the value of the argument x when the
functions y(x) = exp(x) and z(x) = 5 - 4(x-3/2)2 intersect? First, let's
plot these functions to see what they look like.
i
M := 30 i := 0 .. M X i := 2 ⋅
M
2
 3
y ( x ) := exp ( x ) z ( x ) := 5 − 4 ⋅  x − 
 2
6

4
( )
y Xi

z( Xi )

0
0 0.5 1 1.5 2
Xi

The function z(x) is an inverted parabola centered on x = 1.5. There

are two roots, i.e. there are two values of x that satisfy the equation
y(x) = z(x). These two roots lie near x = 0.6 and x = 1.6.
General Form The problem can be restated in general as f(x) = 0, in which the function is
2
 3
f ( x ) := exp ( x ) + 4 ⋅  x −  − 5
 2

( )
f Xi

0
0

2
0 1 2
Xi

The function f(x) crosses the axis near x = 0.6 and x = 1.6. Suppose we
wish to find the exact value of the root near x = 1.6? We already have a
good estimate for the root. What we need is a technique to update and
improve this estimate. Below, we develop such a technique. Later we
will come back to this example and find the desired root.

Newton's
Method
 Illustrated above is an arbitrary function f of a single variable x. We seek
to find the root x* given the estimate of the root: x0. Newton first
suggested that an improved estimate of the root could be found by
following the tangent to the curve back to where it meets the axis at x1.
Two expressions can be written for the slope of the tangent: one is the
derivative f'(x), the other is the divided difference representing the slope of
the curve at x = x0.
( )
f x0 − 0
( )
f' x 0 =
x 1 − x0

This expression can be solved for the new estimate.

( )
f x0
x 1 = x0 −
f' ( x 0)

Note that both the function f and its derivative f' are evaluated at x0. That
is the entire right-hand side is a function only of x0. The calculated value x1
will in general be closer to x* than the starting value x0.

The updating expression, or iteration formula can be applied multiple times

to generate the sequence x2, x3, x4, .... which under certain conditions will
converge as closely as desired to the root. That is, after we have
calculated x1, we can use x1 on the right to generate x2, etc. We stop
when the new value of x differs from the last value by some desired
tolerance.
Iteration Scheme All iteration methods have the following characteristics:
Characteristics
0. An initial estimate must be supplied. This estimate is usually based
on intuition about a physical system, or analysis of the characteristics of the
mathematical expression. This is the place where "engineering problem
solving" skills are applied.
1. An iteration formula must be available to generate new values of the
argument. Newton's formula is an example. We will see others shortly.
2. Some convergence criterion must be selected. This is also an issue
which depends on the nature of the analysis being carried out. In essence
one must decide how many places of accuracy are reasonable for the
argument of interest.
Newton Method Newton's formula is applied below to the problem of finding the root of
Example f(x) near x = 1.6.
2
 3
f ( x ) := exp ( x ) + 4 ⋅  x −  − 5
 2

We need the derivative of f:

 3
fp ( x) := exp ( x ) + 8 ⋅  x − 
 2

The initial estimate of the root: x 0 := 1.6

Generate a sequence of values ... k := 0 .. 2

( )
f xk
... by applying Newton's iteration formula:x k+1 := x k −
fp ( xk)

 1.6 
 
 1.601211 
The sequence of values is: x=
 1.601209 
 
 1.601209 
It is clear that the third and fourth values are identical to six places. As a
check, the function value should be near zero ...
... and it is. ( )
f x3 = 0

Of course, we could just as easily have let Mathcad solve for this root
directly:
ξ := 1.6 x r := root ( f ( ξ ) , ξ ) x r = 1.601209
Extension to
Implicitly-Defined
Functions
Whenever an analytical expression for a function is available, it is clear that
the Mathcad root function will solve for the desired root. However,
Mathcad cannot deal directly with an implicitly defined function such as
those which arise in the analysis of process flowsheets. The "function" can
be visualized in the following context:

The argument "x" is some flowsheet variable, such as a species

composition, flow rate, or a dimensionless equivalent of such a quantity.
The objective function is some material balance, an expression involving
other flowsheet variables. Values for these other flowsheet variables are
computed from other balances, independent of the balance that is used as
the objective function. The values of the compositions and flow rates that
appear in the objective function depend on the value of the argument, even
if the argument itself does not appear explicitly in this balance.
A critical distinction must be made between the process of evaluating this
function and the method used to solve it, that is, to find the value of the
argument that satisfies the balance. The evaluation consists simply in
performing all the necessary intermediate balances. The "solving" consists
in the application of an iterative scheme that converges to the desired root.
The Newton method, which requires an explicit derivative, is not suitable
for this task. Instead, a related method in which the derivative is estimated
numerically can be employed. This method is termed the secant method.
 Secant The Newton method was based on following the tangent to the curve at x0
Method back to the x-axis. The tangent can be approximated by the secant to the
curve through two closely-spaced points, say x0 and x1.

We equate two expressions for the slope of the secant at x1.

( ) ( )
f x1 − f x0 ( )
f x1 − 0
=
x1 − x 0 x2 − x1

This expression can then be solved for the next value of x in the sequence.
( )
x0 ⋅ f x1 − x1 ⋅ f x0 ( )
x2 =
f ( x 1) − f ( x 0)

This is the secant updating rule.

Secant We can apply this technique to the function defined above.

Application
2
 3
f ( x ) := exp ( x ) + 4 ⋅  x −  −5
 2

Initial estimate of the root: x 0 := 1.6

Another value of the argument: x 1 := x0 + 0.05

 Let's generate a sequence of values ... k := 0 .. 3

( ) ( )
x k ⋅ f x k+1 − x k+ 1 ⋅ f xk
x k+2 :=
( ) ( )
f x k+ 1 − f x k

 1.6 
The method converged to the same  
root as before, albeit more slowly.  1.65 
 1.6011462 
x= 
 1.6012062 
 1.6012095 
 
 1.6012095 

In general, iteration is stopped when a convergence criterion like

x k − x k− 1 < δ

If the convergence tolerance δ = 0.001 then we could have stopped with

the fourth value (the second iteration) in the sequence. With δ = 0.000001
we needed six values (four iterations).

Successive A simpler technique for solving f(x) = 0 is to reformulate the objective

Substitution function (when possible) in the form x = g(x). The right-hand-side is then
an updating rule. For the function
2.2 2
f ( x ) := x − 2 ⋅ x + 2.8 ⋅ x − 1

2 2.2
1 + 2⋅x − x
One reformulation is g ( x) :=
2.8

Initial estimate: x 0 := 0.5 k := 0 .. 7

 Generate the sequence: x k+1 := g x k ( )  0.5 
 0.45799 
 0.44289 
 
 0.43773 
The sequence converges slowly.  
x = 0.43599
 
 0.43542 
 0.43523 
 
 0.43516 
 0.43514 
 
Convergence Although successive substitution is temptingly simple, it often fails to
Problem converge. We can try solving the function used in the Newton and Secant
examples by this technique to see this behavior.
2
 3
f ( x ) := exp ( x ) + 4 ⋅  x −  −5
 2

5 − exp ( x ) 3
g ( x) := +
4 2

Initial estimate: x 0 := 1.601

Generate the sequence: x k+1 := g x k ( )

 1.601 
 1.602 
 1.593

 
The sequence fails to  1.642 
converge even when supplied 
x = 1.5 + 0.205i

with an initial estimate that is  
very close to the root.  1.962 − 0.246i 
 1.79 + 0.748i 
 
 2.269 − 0.662i 
 2.196 + 1.068i 
 
Wegstein This step-limited version of successive substitution displays better
Iteration convergence properties. This technique also requires that the objective
function f(x) = 0 be reformuled as x = g(x). The algorithm is:
0. Estimate the argument value x0, and supply a tolerance δ.
1. Evaluate x1 = g(x0), g(x1), S = ∆g/∆x, and θ = 1/(1-S)
2. Update: x = (1 - θ) x1 + θ g(x1)

2
We apply the Wegstein method to: f ( x ) := exp ( x ) + 4 ⋅  x −
3
Example  −5
 2

1
5 − exp ( x ) 3 θ ( η ) :=
Auxiliary functions: g ( x) := + 1−η
4 2

 S0   0 
Initialization: x 0 := 1.601 k := 0 .. 7   :=  
 x1   g ( x 0) 
Generate a sequence:
 g ( x k+ 1) − g ( x k) 
 Sk+1   
  :=  x k+ 1 − xk 
 x k+ 2  x ⋅ ( 1 − θ ( S ) ) + θ ( S ) ⋅ g ( x ) 
 k+ 1 k k k+ 1 

 1.601 
 
 1.6024842 
 1.5930696 
 
 1.5997952 
 1.6013528 
The sequence converges. x= 
 1.6011789 
 1.6012104 
 
 1.6012095 
 1.6012095 
 
 1.6012095 
Back to The secant method is geared to solving f(x) = 0 while the successive
Reality substitution and Wegstein techniques are applied to problems that can be
phrased as x = g(x). Let's recall why we are interested in solving such
equations.
In the simple recycle system shown below, let us assume that the DF
analysis indicates that a carry-forward variable in stream 2 is required to
initiate the material balances on the reactor. Let this variable be the flow
rate of species A. That is, the argument of our objective function is x = nA2.
We proceed by performing the balances on the reactor, the separator, and
the overall subunit. The mixer balance is our objective function.

Secant The objective function, the mixer balance on A, can be written as an error
Formulation function:
f ( x ) = nA1 + nA5 − nA2 = 0

This formulation is tailor-made for the secant algorithm.

Wegstein In this case we formulate the mixer balance as the function g(x):

g ( x) = nA1 + nA5
Summary Whenever a tear variable is required to proceed with the material
balances in a process flowsheeet, it is necessary to implement some
technique for converging to the correct value of the assumed variable. In
essence, the problem has been reduced to the solution of a single equation
in a single unknown.

The Newton, Secant, Successive Substitution, and Wegstein techniques

were presented for the case of an analytic function of a single variable.

The only difficulty in applying these ideas to the solution of the material
balances for a process flowsheet is that the evaluation of the objective
function may involve the evaluation of many intermediate balances.
 5.6.2 Multiple Non-Linear Equations

If the degree-of-freedom analysis for a given flowsheet indicates that

more than one tear variable is required to proceed with the material
balances then one is faced with the task of converging simultaneously on
two or more arguments.
All the methods discussed in the previous section can be extended to the
case of finding roots to multiple, non-linear equations. However, these
methods differ markedly in the relative amount of computation involved in
the iterative algorithm. Each iteration may involve multiple evaluations of
the objective functions. Due to the implicit nature of the objective
functions involved in flowsheet analysis, and the subsequently heavy
computational load involved simply in function evaluation, only the two
simplest algorithms are discussed below. These are Successive
Substitution and the Wegstein method.
Sample System The system to be considered consists of two non-linear equations in two
unknowns: x, y.
2
x y
f1 ( x , y) := + + 0.8 − x
5 10
2 1
f2 ( x , y) := + −y
x + 7⋅y 2

The desired root is at (1.139,0.798).

Multidimensional As was the case for a single equation, the system must first be reformulated
Successive for updating. One such reformulation is:
Substitution
2
x y
x = g1 ( x , y) g1 ( x , y) := + + 0.8
5 10

2 1
y = g2 ( x , y) g2 ( x , y) := +
x + 7⋅y 2

Initial estimate of the roots: x 0 := 1 y0 := 0.5

  x k+ 1   g1 ( x k , yk) 
Generate the sequence: k := 0 .. 5   :=  
 yk+1   g2 ( x k , yk) 

The sequence converges to  1   0.5 

the desired multidimensional root.  1.05   0.944 
 1.115   0.761 
   
x =  1.125  y =  0.81 
 1.134   0.794 
   
 1.137   0.799 
 1.138   0.797 
   

Multidimensional The algorithm is similar to that used in the one-dimensional case:

Wegstein Iteration
0. Estimate the array of argument values x0
1. Calculate x1 = g(x0), the slopes Si = ∆gi/∆xi, and θi = 1/(1-Si)
2. Update with x2 = (1 - θ) x1 + θ g(x1)

This algorithm is applied to the same system used to illustrate the

multidimensional successive substitution method.

 Sx0   0 
Initialize:  
x 0 := 1 y0 := 0.5  x ,y 
 g1 ( 0 0) 
 x1 
  :=  
 Sy0  0
 
 g2 ( x 0 , y0) 
y 
 1 
1
Auxiliary function:θ ( s ) :=
1−s

 g1 ( x k+ 1 , yk) − g1 ( x k , yk) 
Iterate:  
 Sxk+1   x k+ 1 − x k 
  
 x k+2  ( 1 − θ ( Sxk) ) ⋅ x k+1 + θ ( Sxk) ⋅ g1 ( x k+1 , yk+1) 
  :=
 Syk+1   g2 ( xk+ 1 , yk+ 1) − g2 ( x k+ 1 , yk) 

y   yk+1 − yk
 k + 2  
( 1 − θ ( Sy ) ) ⋅ y + θ ( Sy ) ⋅ g ( x , y ) 
 k k+ 1 k 2 k+ 1 k+1 
  1   0.5 
   
 1.05   0.944 
The convergence is more  1.115   0.761 
   
rapid in this case.  1.132   0.796 
x= y=
 1.139   0.798 
   
 1.139   0.798 
 1.139   0.798 
   
 1.139   0.798 

Relation to Let us recall once more the context in which multiple non-linear equations
Flowsheet arise in process flowsheet analysis. It may occur that a degree-of-freedom
Analysis analysis indicates that several tear variables may be required to initiate
balances. For example, assume that the flow rates of all species in stream
two below must be estimated in order to proceed with the balances on the
reactor.

The unknowns then are the species flow rates in stream two ni2. One
possible scenario is that the reactor, separator, and overall subunit balances
can then be completed, leaving the mixer balances as an objective function.
The formulation for S species is:
ni2 = ni , 1 + ni , 5 i := 1 .. S

This form is precisely that needed to implement successive substitution or

Wegstein iteration x = g(x).
Summary The extension of the successive substitution and Wegstein methods to
multiple non-linear equations involves no new principles. The
convergence of systems of more than two equations is much slower than
for the case of a single equation.