You are on page 1of 3

PAPER

Conventional calculations
Geotechnical evaluation of the ground properties is necessary to
roc rariIriIinc provide the information for the structural design. This is done as an
independent exercise. The following parameters are normally required
in this type of geotechnical analysis:

'or: >ree- ~ Safe ground bearing pressure, which is based on the allowable bearing
capacity assessment, Veisc (1975).
~ Total and differential settlements, which are often very difficult to
estimate due to the uncertainty of the extent of the loaded area. The
current methods are inaccurate since they do not take into account the
c igaensiona rigidity of the base, see Mindlin (1936)or Boussinesq (1885).
~ Modulus of subgrade reaction for ground bearing slabs, which is
difficult to obtain using hand calculations owing to the flexible

'ini:e e egaenI: behaviour of the base. More sophisticated techniques such as finite
difference methods would be required.
~ If piles are necessary to either resist uplift or to accommodate
compressive loads the raft-ground beam-pile method must be applied.

ana ysis These piles are normally designed to bear compressive load even if
designed for uplift only. An estimate of the percentage of load taken by
the raft and pile is very important in order to make an economical
deslgll.
~ Lateral earth pressure envelope, which is relatively straightforward
o'nc

erc lrounc to calculate, although some doubts may exist regarding an appropriate

" uic re:ainincl value for the coefficient of horizontal pressure. These doubts are more
evident where the wall has locally increased stiffness, eg being adjacent
to cover slabs, internal walls etc. The use of the at-rest coefficient or a
value between active and at-rest state is normally considered
appropriate.
The calculation of the deflection of the walls, which is essential to

sI:rue'ures
AN Sainak and D Ajenkol, Haswell Consulting Engineers
assess the serviceability of the design, is difficult using simple hand
calculation. Judgment based on permissible strains and maximum
movement required to develop the active pressure are normally applied.
Structural design is carried out using simple statics, by hand or with the
aid of a suitable computer package. In the case of the conical tanks a
method such as the one suggested by Tong (1978) can be implemented.
All structural designs conform with BS8110and BS8007.
Synopsis Where necessary, the bearing capacity of the soils forming the
Interface programs have been developed to assist engineers with little foundations is made sufficient by compression piles and problems of
knowledge of the finite element method in carrying out three- floatation are resolved by the installation of tension piles. Sometimes,
dimensional finite element analysis of underground fluid retaining the problems of bearing capacity and/or floatation are underestimated
structures. The programs are aimed at reducing the gap between the which can result in premature failure of the slabs, or large differential
specialists and the ordinary engineers, carrying out repetitive tasks and settlements, which in turn lead to large cracks in the structure.
producing robust and economical designs. Also, the global and local Alternatively problems of bearing capacity and/or floating can be
results such as stresses, strains, bending moments and displacements exaggerated, which may result in an uneconomical design.
can give a better understanding of many unknown factors in the design Two and three dimensional finite difference methods may be
of these structures. employed in calculating the deformation of the structures, but is not
well established practice for these type of problems. The method
Introduction incorporates input data obtained from geotechnical analyses, which are
It is now commonplace for engineers to find themselves in situations not always accurate or easy to interpolate.
having to choose between powerful computing techniques such as finite Various problems associated with the design procedures may
element analysis, which they do not fully understand, and the necessitate a complete reassessment of the design process. The need for
conventional hand calculation methods, which take too long or are an integrated program, that can deal with both soil and structure
inadequate. If they choose the former they feel at the mercy of the analysis, is apparent. Ground bearing capacity, for example, is quite
computer and the distant figure who developed the software. If they straightforward to calculate but difficulty may occur when piles are
choose the latter they find themselves using somewhat outdated introduced. Assumptions made at the design stage are not necessarily
methodology and face the many inadequately explained material representative of the actual conditions, especially regarding the load
behavi ours. distribution between the raft and piles. Computer programs have been
The design of underground water treatment tanks has traditionally developed which gave the user a degree of flexibility in analysing the
been based on conventional hand calculation procedures which take structural member but would normally require input information. This
into account the behaviour of the soil and of the structures information is not always easy to calculate accurately, eg modulus of
independently. The simplified assumptions necessary in all available subgrade reaction.
methods, and the uncertainty in the behaviour of soil/structure
interaction remain as the main obstacles in producing economical Finite element analysis and parametric programming
designs. Finite element analysis, on the other hand, can combine the of this type of complex problems should
3-D finite element analysis
analysis of the soil and the structure in one model producing clear include both the structure and the soils in one model to allow for
pictures of the behaviour of all parts of the soil and the structure. The soil/structure interaction. It should include all possible loading regimes
main difficulty in this technique is the need for specialists to model the which may vary in intensity and direction. It should also accommodate a
problem and interpret the vast amount of output generated. range of variations in soil properties and a number of constitutive
This research is aimed at exploring the development of interface models to govern their behaviour. Many commercial finite element
programs "Parametric Programs" to carry out the specialist tasks analysis programs are available in the market. Some of these programs
without the need for a sound knowledge of the finite element method are for general engineering problems, others are for specific use. LUSAS
theory. It covers four programs developed to model the physical problem, finite element program, which is a general purpose program, is used in
analyse the data and present the relevant results in standard format for this research along with the graphical program MYSTRO.
rectangular and conical buried fluid retaining structures. The programs To model the physical problem into a finite element analysis problem
use LUSAS finite element program for the analysis and the MYSTRO realistically, specialist expertise is essential. The need for finite element
graphical program for the pre and post-processing operations. The study method expertise becomes more apparent when modifications and/or
also contains one case study for each parametric program to illustrate alternative designs are proposed. This led to a distinctive gap between
their application and accuracy of results. The hand calculations and the the works of ordinary engineers and the analysts. The problem can be
3-D finite element analysis for these case studies are then compared. minimised by developing interface software to assist engineers in

30 GROI IND ENGINEERING MAY 1998


PAPER

Geo met
in the stiffness of soils with depth. Three types of soil are considered in the
relevant
models: around the tanks, below the tanks and the backfill.
~ Sliding or coupling actions between the concrete structures and the
soil may be modelled.
~ The rectangular tank is meshed using (8-noded) linear elements and
the circular tank is meshed using (20-noded) parabolic elements.
~ Any number of rows and columns of piles can be incorporated in the
rectangular tank model. Up to four circles of piles can be included in the
conical tank model. The piles can be designed to accommodate tension
and/or compression and may also transfer bending moments from the
slab. The piles can be modelled as springs or 2-noded beam elements.
~ The following loading regimes may be applied to the structures:-
a) uplift due to pore water pressure
b) lateral loading from backfill
c) weight of water in the tank, different water levels in the bays of the
rectangular tank
Surrounding soil and backfill Foundation d) tension and compression piles
Made ground Marcia mudstone grade III/IV e) surcharge
y=18 kN/ms y = 20 kN/ms f) self-weight of tank
= 0.2 v = 0.4
v In the post-processing programs the user replies to a series of prompts
E = 20 x 109 kN/mz E = 38 x 10S kN/m2
to narrow the choice of output. The user must choose the output
required such as stresses, strains, bending moments or displacements
Geometry of conical tank and 11m and their relevant direction. He/she must then choose the area of
relevant soil properties interest such as walls, slabs, soil layers, backfill or global view. The user
may request a cross section through the domain, print, save or plot the
results. The execution of the programs takes between five and 15
minutes and the analysis takes between 40 and 80 minutes using a
0.95m Personal Computer Pentium 200MMX with 32 Mb SDRAM.

Case studies
Figure 1(a) and Figure 1(b) show the geometry and soil parameters for
two case studies considered in this research. The rectangular tank was
analysed as empty and under maximum uplift, lateral pressures and
surcharge of 10kN/m'. The conical tank was analysed as full with water
with no uplift or surcharge. Neither tension nor compression piles were
included in the problems. Initial analyses of the problems with
maximum possible loading conditions resulted in strain values of 0.2%.
Therefore it can be assumed that the behaviour of the soils is elastic,
0.5m Zienkiewicz (1977). As a result, the programs adopt piece-wise linear
constitutive laws in modelling the soils.
Surrounding soil and backfill Foundation
Made ground Mercia mudstone grade III/IV Results and conclusions
y= 18 kN/ms Y
= 20 kN/ms Three examples of the 3-D finite element analysis results are chosen to
v=02 v = 0.4
indicate the approach. Figure 2 shows a global distribution of the
E = 20 x 109 kN/mz E = 38 x 109 kN/mZ
bending moments (N.m) for the rectangular tank under external loads,
Flgure1 (a) and Figure 1(b) amount and the extent of ground movement in the case of the
rectangular tank. Figure 3 shows a section of the conical tank model
and the distribution of vertical displacements. It also indicates the zone
generating the data files accurately (pre-processing) and extracting the of soil influenced, resulting from the application of internal loads and
relevant results (post-processing). the self weight of the tank. Noticeable ground movements took place at
Four parametric programs have been developed, in this study to carry between 1.5 and 2 times the tank's dimension, in all relevant directions.
out all necessary tasks in the analysis of rectangular and conical Figure 4 shows a cross section of the bending moments of the conical
underground fluid retaining tanks. The programs run in association tank under the loading conditions described above.
with MYSTRO where the
user is prompted to either
make a choice or input a
value. The pre-processing STRESS
programs require the user CONTOURS OF Mx
to input the following data:
~ The geometry of the tank -0.9828E+05
~ The properties of
concrete, soils and backfill -0.7233K+05
~ Layout and dimensions of -0.4638K+05
piles, if the problem -0.2043E+05
includes piles 5524.
~ Depth of water in a tank, 0.3148E+05
depth of water table in the 0.5743E+05
ground and the amount of
surcharge where applicable.
0.833BE+05
A data file is generated at 0.1093E+06
the end of each run with the 0.1353E+06
following assumptions and
limitations:
~ The extent of the soil
mass model is limited to five
times the dimension of the
tank in the relevant
direction.
~ Piece-wise linear increase Figure 2: Sending momeuls (Lm) on rectangular tank due to external loads.
GROUND ENGINEERING MAY 1998
PAPER

@f 4 p/ lwhevtf 8! DISPLACEMENT
CONTOURS OF DY

-0.1705E-01
-0.1580K-01
-0.1454E-01
-0.1328E-01
-0.1203E-01
-0.1077E-01
-0.9518E —02
-0.8263E-02
-0.7008E-02
-0.5752E-02
-0.4497E —02
-0.3242E-02
-0.1986E-02
-0.7308E-03
0.5246E-03

Figure 3:Cross secgon of conical tank model showing vergcaf dlsfdacemonts (m).
A brief comparison between
STRESS
the hand calculation method and
CONTOURS OF tex
the 3-D finite element analysis,
with the aid of parametric
-0.1584E+06 programs, is summarised below
-0.1461E+06 left.
-0 '337E+06
A comprehensive comparison
-0.1213E+06 between 3-D finite
-0-1089E+06 element
analysis and conventional
-0.9657E+05 methods is beyond the scope of
-0.8419K+05 this study. Back analysis of
-0.7182K+05 more case studies with various
-0.5945E+05 loading conditions
-0.4708E+05 are being
carried out where the results
-0.3471E+05 will be compared and justified
-0.2234E+05 and made
-9964. available for
.'. Adl i!,g ,lit publication.
2408. The main advantage of these
0.1478E+05 programs is in allowing users
0.2715E+05 with very limited knowledge of
0.3952E+05 fuute element methods to carry
0 5190E+05
out the analysis using the
0.6427E+05 programs as a "black box". This
0.7664E+05 reduces the need for "specialist
expertise" and facilitates the
Figure ok Cross secgon of a conical tank, under Internal loading, with radial bending moments (Lm).
execution of repetitive analyses
by less experienced software
Conventional methods 3-D FEM with parametric programs users.
Practice Established and New and simplified
straightforward Veisc, AS (1975). Bearing capacity of
shallow foundations, Foundation
Modifications to geometry may require a can easily be carried out Engineering Handbook, p121-141, Van
and/or material properties Nostrand Rienhold Co, New York.
completely new design Mindlin, RD (1938). Force at a point in the
interior of a semi-infinite solid, p195-202,
Soil/structure interaction not included included Physics.
Soil properties Boussinesq, J (1885). Applications des
assumptions on isotropy variation of properties in all potenille a I'etude de I'equilibre et du
and homogeneity required directions can be accommodated mouvenment des solids elastiques,
Three dimensional effects not considered included Gauthier Villars, Paris.
Tong, AL (1978). Design of conical base
Strain assumptions slab for circular concrete tanks, p198-203.
plain strain assumed plain strain conditions not necessary ACIJournal.
Zienkiewicz OC (1977), The finite element
method, McGraw Hill.

32
GROUND ENGINEERING MAY 1998

You might also like