Академический Документы
Профессиональный Документы
Культура Документы
• ANALYSIS
11111111111111111jlIIIUII~111I11 -~
#90408#
A THESIS BY
SEPTEMBER, 1996
•••• Jo"'t
••••r
o -~
,••
A DESIGN RATIONALE FOR MAT FOUNDATION BASED ON
FINITE ELEMENT ANALYSIS
A THESIS BY
A. S. M. MONZURUL MORSHED
4~~
Dr. Sohrabuddin Ahmad Chairman
Professor (supervisor)
Department of Civil Engineering
BUET, Dhaka.
-~Q+=1L--
Dr. M. Azadur Rahman Member
Professor and Head
Department of Civil Engineering
BUET,~
Dr. Ah5~-Kh-.-an----
Member
Managing Director (external)
Construction and Development Co.
44, Dhanmondi, Road No. 4A, Dhaka.
ii
DECLARAnON
Declared that, except where specified references are made to other investigators, the
work embodied in this thesis is the result of investigation carried out by the author
under the supervision of Dr. Sohrabuddin Ahmad, Professor of Civil Engineering,
BUET.
Neither the thesis nor any part thereof has been submitted or is being concurrently
submitted in candidature for any degree at any other institution.
Author
iii
ACKNOWLEDGEMENT
The author wishes to express his indebtedness to Dr. Sohrabuddin AJunad, Professor
of Civil Engineering, BUET for his continuous guidance, invaluable suggestions and
affectionate encouragement at every stage of this study.
The author expresses his profound gratitude to Dr. Abdul Muqtadir, Professor of
Civil Engineering, BUET for his constant guidance, suggestions and encouragement
at all phases of this work.
Heartiest thanks are expressed to Dr. M. Azadur Rahman, Professor and Head,
Department of Civil Engineering, BUET for his co-operation.
iv
CONTENTS
Declaration
111
Acknowledgment
IV
Abstract
Xl
List of Symbols
Xlll
Chapter 1: INTRODUCTION 1
1.1 General
1
1.2 Background of the Study
2
1.3 Objective of the Present Research 4
1.4 Methodology
5
1.5 Organization of the Thesis 6
v
2.2.10 Calculation of Punching Shear 19
2.3 Limitations of the Present Finite Element Model 21
2.4 Presentation of Results 21
I
4.8 Summary of the Parametric Analysis of Mat 40
1,
Chapter 4: DESIGN OF MAT FOUNDATION 41
t
I 4.1 General 41
!
4.2 Difficulty with the Formulation of Mat Behavior 41
\ 4.3 General Design Requirements for Mat Foundation 42
vi
I,
4.3.1 Deflection Requirements for Mat Foundation 42
4.3.2 Design of Mat for Bending Moment 43
4.3.3 Design of Mat for Shear Force 44
4.4 Design Example 45
4.4.1 Some Frequently Used Design Values 46
4.4.2 Deflection of Columns 48
4.4.3 Design for Punching Shear 48
4.4.4 Design for Flexural Shear 50
4.4.5 Design for Bending Moment 52
4.4.6 A Different Interpretation of Flexural Shear 53
4.4.7 Average Bending Moments Across the Widths
of Column Strips 54
4.4.8 Effect of Column Base Moments Due to Gravity
Loads 55
4.4.9 Discussion on Design by Different Methods 56
4.5 Mat Thickness Requirement : A More Economic
Approach 58
vii
Chapter 5 : DISCUSSION AND CONCLUSIONS 89
5.1 General 89
5.2 Relative Perfonnance of Different Analysis Methods
forMat 89
5.2.1 Comparison of Bending Moments and Shear
Forces Based on Strips of Unit Width 90.
5.2.2 Comparison of Bending Moments and Shear
Forces Based the Entire Column Strips 92
5.2.3 Economic Evaluation of AnalysisMethods 93
5.3 Results of the Parametric Study 95
5.4 Calculation of Shear Forces 96
5.5 Prospect of Mat with Greater Thickness Around
Columns and SmallerThicknessElsewhere 97
5.6 Conclusions 99
5.6.1 General Conclusions 99
5.6.2 Design Approach for Mat with Non-uniform 101
Thickness
5.7 Recommendations for Future Study 102
REFERENCES 103
viii
B-1
Appendix B : MODULUS OF SUBGRADE REACTION
B-1
B.l General
B.2 Methods to Determine Modulus of Subgrade Reaction B-2
B.3 Modification of Modulus of Subgrade Reaction to be
Used with Mat Foundation B-4
B.4 Approximate Values of Modulus of Subgrade
B-6
Reaction
C-l
METHODS OF MAT
Appendix C : AVAILABLE
C-l
FOUNDATION ANALYSIS
C-l
C.l General
C.2 Conventional Method of Analysis
C-l
C-3
C.3 ACI Approximate Flexible Method
C-7
C.4 Baker's Method
C-8
C.4.1 System of Forces
C.4.2 Division ofthe Force System C-8
C-IO
C.4.4 Beam Equation
C-IO
C.4.5 Soil Equation
C-ll
C.4.6 Beam Bending Moments
C-12
C. 5 Finite Difference Method
C-14
C.6 Finite Grid Method
'.
E-l
Appendix E : ADDITIONAL FIGURES
ix
Appendix F : MISCELLANEOUS F-l
F.l List of Special Terms F-l
F.2 Unit Conversion Table F-2
x
ABSTRACT
Recently, mat has become quite popular among foundation engineers. Although there
are several methods available for the analysis of mat, none of these is accurate
enough to picture the actual behavior of mat. In the current study, two popular
methods for the analysis of mat, namely Conventional method and ACI Approximate
Flexible method, have been compared with the Finite Element solution of mat using
Ahmad's thick shell element. A general computer program, capable of solving any
shell type structural problem, has been used for this purpose. Various data generation
and result interpretation modules have been added to the original program to convert
it into a versatile mat analysis software. The software is capable of analyzing mats
with any kind of column and shear wall arrangement as frequently encountered in
engineering practice. Also, separate computer programs have been written for
analyzing mat using Conventional and ACI methods for the purpose of comparison.
In performing the design of mat, the USD method of ACI code of practice has been
followed in this study. It has been found that Finite Element method gives the most
economic solution and its economy increases sharply with the increase of column
loads. It has also been observed that frequently minimum requirements of ACI code
govern the design of mat and because of this, Conventional method results in only
slightly uneconomic design in comparison to Finite Element method despite its
inaccuracy and various drawbacks. For heavy column loads, ACI method has been
found to give considerable overdesign in comparison to Finite Element method
despite its rigorous nature of analysis.
A parametric study has been made in order to identifY the parameters which play the
most important roles in defining the behavior of mat. It has been found that mat •
response is not much sensitive to most of its parameters. The most significant role
played in this regard has been identified to be that of mat thickness. Finally, the
economic prospect of a reshaping scheme for mat has been examined. Relative
performance of mat with non-uniform thickness with respect to mat with uniform
xi
thickness, which is the most popular practice so far, has been evaluated. Effects of
the new geometric parameters associated with mat with non-uniform thickness have
also been investigated in order to find a safe guide line for this redimensioning
proposal. It has been found that mat with non-uniform thickness offers about 20 to
30% saving of concrete and reinforcement and its design is not much complicated
compared to the design of mat with uniform thickness. In the end, scope for future
research has been indicated.
xii
LIST OF SYMBOLS
xiv
r Radial distance of the point under investigation from loads
t Thickness of mat
tavg Weighted average thickness of mat
Greater thickness of mat with non-uniform thickness
t, Smaller thickness of mat with non-uniform thickness
t'enta';v, Tentative thickness for the first trial of mat with non-uniform
thickness
V Acting shear force
Ve End correction to shear force in ACI method
. Vd Number of y directional divisions of finite element mesh
V, End-correction shear in ACI method
Vru Ultimate flexural shear strength
Vp Acting punching shear
vpu Ultimate punching shear strength
w Mat displacement
WL Wind load
w, , Wz & W3 Displacements due to the first, second and third force system
respectively in Baker's method
We End correction to displacement in ACI method
{X} External nodal displacement vector in Finite Grid method
Z's Functions first introduced by ScWeicher (1926)
a Angle of orientation of finite grid element
. Lit Change in mat thickness
4> Angle of the radial line passing through the point under
consideration w.r.t. the positive x axis in a clockwise direction
v Poisson's ratio of mat material
P Steel ratio
xv
Bottom and top normal stress respectively
xvi
CHAPTER 1
INTRODUCTION
1.1 GENERAL
Mat foundation has gained widespread popularity among the engineering community
of Bangladesh. Bangladesh is running its race for prosperity and urbanization has
become an integral part of this endeavor. Now-a-days high rise buildings are common
features of major cities in Bangladesh. Many have been constructed, many are under
construction and many more are on the planning table. As their foundation, engineers
often opt for mat.
Going high is not an easy task in Bangladesh. As engmeers come to the soil
eventually, seeking support for their super-structures, frequently they face the
unfavorable reality. The soft soil of Bangladesh has relatively low bearing capacity.
Heavy column loads may demand for large footing areas sumining upto more than
half of the total foundation area available. The most formidable hurdle foundation
engineers often have to encounter is excessive settlement,. especially excessive
differential settlement. Although pile foundation offers solution to these problems,
sharp jump in expense becomes a serious point to ponder in the poverty lashed
economy of Bangladesh, not to mention the disturbance to city life and the possibility
of damage to underground structures associated with pile foundation construction.
These are the situations when mat comes out to be a better option.
Mat lowers soil contact pressure, displaces mass volume of soil giving floatation
effect, combines its own rigidity and that of the superstructure to activate bridging
effect and can tolerate larger settlement. In addition, mat has other attractive
advantages. It provides basement floor which has considerable commercial values in
urban areas. In case of basement at or below ground water table, mat acts as water
barrier. No doubt, in terms of ease of construction, utility and economy, mat is going
to prevail as a superior choice to foundation engineers. Apart from buildings, mat has
. application beneath silo clusters, chimneys, storage tanks and various tower
structures.
Although mat, a large concrete slab used to interface one or more columns in several
lines with the base soil, usually covers the entire foundation area, its definition does
not necessarily require so. There are several configurations for mat and some of
them, as used with buildings, are shown in Fig. 1.1. But a mat with uniform thickness
has been the most common one so far.
Although there are several methods for the analysis of mat, virtually none is accurate
and convenient enough. There are some approximate methods which are considerably
crude and in use for a long time. Recently, there are some computer based methods
available. However, these computer based methods idealize mat unrealistically.
In the most common and simplified method, known as Conventional method, column
loads are distributed under the mat; then the mat is divided into strips midway
between the column lines and the force system acting on each strip is adjusted to
establish equilibrium. Finally each strip is designed as a combined footing. This is
repeated for the other direction. This method is recommended by ACI when adjacent
spans and column loads do not vary by more than 20%. However, in this method mat
is considered to be fully rigid in determining the soil pressure and also the divided
strips lose their plate characteristics .i.e. two way bending.
One of the most popular methods of mat analysis is the ACI Approximate Flexible
method (Committee 336) which is essentially based on the analysis of Schleicher
2
EI:fE A-A
.'
B.3
r.,,
A
t
• • •
• • •
•
•
•
•
A
+
8
•
- ~I
I.'
'
r.
.) (~l
,
.(.'
,,~
"
Iii'
-'
./
" '
,
\
8
+
• • • • •
,.
.j 'ii'
'C_ I ii' ,ii' 'jI
-' -/
.'
"
• • • • • Ii' Ii' Ii' ~
(a)
(0)
-
c-c
~ EEf;El D-D
c D
t +
• 00 [!J [!J
• 00 00 00
(c) (d)
~
c-E
E r--l
,
r- --, r---, r-
, I, I ,
--I
, E
, I I I r I I I
•• L_.:......J L
'---I
ll. __ -J L __ -'
r---, r---, r--, +
I I, I I I' I
I-Ii I, '[-I
L __ J L __ ..J L __ ...J L __ --'
r- -, r- -., r---, r--l
I r
I I I I J
I I I I I " I
L __ J L __ J L __ J L __ J
(e)
Fig. I. I. Common types of mat foundation: (a) flat plate; (b) plate thickened
under columns; (c) waffle slab; (d) plate with pedestals and (e)
basement wall as part of mat.
(1926). In this method, as the name implies, the mat is considered to be a flexible
plate acted upon by concentrated column loads and resting on a Winkler medium.
Effect of each load is calculated as if the plate were infinite. Forces for each
individual column are summed up. To have the edges free from forces, the mat is
divided into strips and the forces obtained by this approach on the edges are applied
in opposite direction on the respective edges considering each strip to be supported
on semi-infinite elastic foundation [Hetenyi (1946)]. Finally this new set of forces are
added to the previous ones. Shukla (1984) presented design aids for using this
method. Later, this method was further modified by Mician (1985). However, this
complicated and lengthy approximate method fails to take into account the boundary
conditions and moments from columns.
Baker (1948) proposed a method of mat foundation analysis in which the mat is
divided into column strips resting on Winkler medium and each strip is analyzed
separately. A particular soil pressure distribution of unknown magnitude is assumed
and applied on both the column strip and the supporting. soil. The magnitude of soil
pressure is determined by taking the maximum differential displacements of the
column strip and the supporting soil as the matching criteria. However, the detailed
deflected shapes of the column strip and the supporting soil do not. match in this
method. Also, this method can not take into account column base moments and fails
to represent the plate characteristics of mat.
Bowles (1974) has proposed Finite Grid method in which the mat is reduced into a
grid system consisting of beam-column elements resting on Winkler medium. Gupta
and Patel (1996) have used this method in an extensive study of mat foundation.
Solution to the problem is obtained following the stiffuess matrix approach.
According to Bowles, there is a little computational improvement if the soil is
modeled using its modulus of elasticity and Poison's ratio instead of its modulus of
subgrade reaction. However, the plate characteristics of mat is lost in this method.
Also, the program requires extensive data in. generating the grid geometry, sectional
and material properties.
3
There is also Finite Difference method [Bowles (1974), Deryck and Severn (1960,
1961)] where the mat is modeled as large flat plate on elastic medium. The fourth
order differential equation as given by plate theory is then numerically solved using
finite difference technique. Handling column base moments is beyond the scope of
this method. Also, interpretation of the boundary points is cumbersome in this
method. ,. ry..
In an effort to make the infinite plate solution phase of the ACI Approximate Flexible
method more realistic, Homsley (1988) has derived equations for bending moments,
shear forces and deflections of infinite plate on Winkler medium acted upon by single
distributed force and moment. These derivations have also been compared to half-
space model [Homsley (1988)]. Homsley (1987) has also given solution to circular
mat acted upon by peripheral line load and resting on Winkler medium or half-space. _! __
Attempts have also been made [Horvath (1983)] to develop more realistic model for
the supporting soil, based on theory of elasticity solution of elastic layers under
surface loads. Horvath (1993) has applied this new subgrade model to mat
foundation analysis. Chilton and Wekezer (1990) have performed finite element
analysis of plates on elastic foundation using 12-noded plate element after modifYing
its stiffiJess matrix. Underlying principal of the modification of the stiffiJess matrix in
their work has been that the elastic foundation deflects in a manner consistent with
the structural element it supports. However, all this developments are still at
experimental level and yet to be applied to any real mat. ,
4
analysis of mat foundation by Finite Element method using Ahmad's thick shell
element with those by Finite Grid and Finite Difference methods.
1.4 METHODOLOGY
For analyzing mat foundation by finite element technique soil is modeled as Winkler
medium and Ahmad's general thick shell finite element (1969,1970) is used. Selection
of this particular finite element is justified on the following grounds:
(ii) Ahmad's general thick shell finite element is a powerful and versatile
element in the analysis of shell type structural problems.
(iii) There exists a general computer program [Ahmad (1969)] for carrying
out finite element analysis using the chosen element.
5
The first step of this research project is to develop supporting computer programs to
feed the general program with the huge amount of data associated with the analysis
of such problems and also for converting the output produced by the general
program into more convenient form for parametric study and design. For comparing
the results of finite element analysis with those obtained from conventional methods,
separate computer programs are also written. Finally, an extensive study is made to
suggest a more economic design approach.
The main body of this thesis is structured into four chapters. Chapter 2 describes the
finite element model of mat foundation used in the study. Comparison of finite
element analysis with different available mat analysis methods and parametric study of
mat foundation are discussed in chapter 3. In chapter 4, relative economy in mat
foundation design associated with different analysis methods is discussed. A
reshaping scheme for mat foundation in order to achieve more economy in mat
design and its effect on mat behavior and design economy are also given in this
chapter. Chapter 5 contains a summery of the study and the conclusions drawn from
the research.
In addition, there are five appendices in this thesis. Appendix A describes Ahmad's
thick shell element and the general computer program for analyzing structures using
this element. Appendix B deals with various methods of determining modulus of
subgrade reaction. Different available methods for mat foundation analysis are
discussed in appendix C. User instructions of the mat analysis software developed in
the present study is given in appendix D. Appendix E contains some additional graphs
which are relevant to the study. Finally, some special terms used in this thesis are
listed in appendix F. The present study is conducted in FPS unit system. A unit
conversion table showing the relationship between FPS and SI units is also included
in appendix F.
•
6
CHAPTER 2
2.1 GENERAL
The basic purpose of this thesis is to analyze mat foundation by finite element
method. The ultimate goal of proposing economic design approach will be based
upon this finite element analysis. Finite element technique is the most powerful and
versatile of all the available numerical analysis techniques. In this method, the
structure to be analyzed is modeled as an assemblage of a finite number of discrete
interconnected elements and the displacements of the connecting points, called the
nodal points, of these elements are taken as the basic unknowns. The applied loads
are transformed into equivalent nodal loads and for each element, the relationship
between the nodal loads and nodal displacements is established. Then with a suitable
assembling of the element load-displacement relationship, sufficient number of load-
displacement relations for the whole structure can be found which, upon solution,
give the nodal displacements. Once known, these displacements are used to calculate
stresses at the nodes or at points within the elements.
There are many types of finite elements. The one to be used with a particular
structure must be such that its properties match the overall behavior of the structure
as perfectly as possible. Otherwise analysis will give spurious results. Mat foundation
is a three dimensional thick plate structure. In case of mat, transverse flexural shears
, 7
"
.f).
and bending moments are the most important internal forces produced in response to
the loads it usually encounters e.g. axial column forces and column base moments.
Thick shell element is one of the most suitable elements which matches all of these
above mentioned criteria. It can represent the thickness of the mat by its very own
geometry. Also, in thick shell elements the nodal variables are taken to be the
displacements of nodal normals which, though approximately normal to the
undeformed middle surfaces of the elements, are not necessarily normal to the
deformed middle surface, leaving scope for transverse shear stress and shear strain
calculation. So thick shell element is a good choice for mat foundation analysis.
Apart from its suitability, another reason for selecting thick shell element for mat
foundation is that a general computer program for that element is available at BUET.
This program was written by Ahmad (1969), who also devised the element. There are
two types of Ahmad's element, namely 8 noded element and 12 noded element. For
the present study the first one is used since it gives results accurate enough.
The general program is in FORTRAN77. It can analyze any plate or shell type
structure using 8 noded or 12 noded Ahmad's thick shell element. The shell elements
. may be curved or plane on both or anyone of its faces and may be of uniform or
variable thickness. The program takes the general information about the finite
element mesh such as total number of elements, total number of nodes, number of
variables per elements etc. ; coordinates and coefficients of element integration
points; nodal global coordinates, and pressure and temperature data; and element
data such as material properties, number of variables per element, element topology
as described by nodes defining the element etc. The program can handle various
inplane and out of plane loads such as gravity load, uniformly distributed surface
load, any kind of varying load, concentrated load etc. As output, the program gives
the nodal global displacements and stresses. However, the material of a particular
8
I•
".
element must be homogeneous and isotropic though material properties may vary
from element to element.
2.2.3 Modification of the General Program for Building a Software for the
Analysis of Mat Foundation
Since the original program is very general in nature, it requires a huge amount of
data, volume of which is proportional to the number of elements. Initially, the original
program used to give only nodal stresses. Though nodal normal stresses are
excellent, out of plane shear stresses at the nodes are not very good. In fact,
transverse shear is not very important in ordinary curved shell structures where out of
plane shear is low. But in case of mat this shear is a very important feature. Now for
this element, element stiffness matrix is evaluated by Gauss integration and it has
been observed that Gauss points within an element are the best stress sampling
points. So use of Gauss point stresses ensures excellent performance of the element
in terms of any kind of stress in general and for this, necessary option has already
been created in the program by Amanat (1993) under the guidance of Ahmad.
A parametric study to reveal the behavior of mat requires analysis of a large number
of examples. So a data generation program is necessary to make the analysis more
efficient and to lower the chance of error in data input. Also, stresses are not a very
familiar way of understanding the behavior of a structure. Engineers often feel more
comfortable with bending moments and shear forces and design formulations of
almost all the codes of practice utilize bending moment and shear force instead of the
stresses responsible for them. So transformation of the stresses, as output by the
general program, to bending moments and shear forces adds to the utility value of the
finite element analysis.
9
The general program, with the desired modifications and development of supporting
modules for data generation and result interpretation, has been turned into a versatile
mat analysis software. Attributes of the software are as follows :
(i) The program can run in the original mode requiring original form of data input,
in interactive mode or in file input mode.
(ii) The program takes mat dimensions, material properties, element mesh features,
information about loads, items of output and locations and types of result
interpretation etc. as its input.
(iii) It gives nodal displacements, stresses, moments and shears as output.
Displacements are always at nodes but moments and shears may be at the nodal
points or at the Gauss points upon request.
(iv) The program can analyze a full mat or, in case of symmetric problems, half or
quarter of a mat.
(v) Column axial loads and column base moments can be applied as concentrated
or distributed loads.
(vi) In case of distributed column axial loads and column base moments, an element
is taken under the concerned column and that element can be made more rigid
than the other portion of the mat.
(vii) Elements near the loads can be taken of smaller dimensions than the other
elements away from the columns.
(viii) Thickness of mat under the columns may be higher than that elsewhere.
(ix) Strips of thickness greater than the thickness of mat elsewhere can be
incorporated along the column lines.
(x) Punching shear for each column can be calculated.
(xi) As for boundary condition, the program can attach horizontal soil springs along
the edges, can frx the edges horizontally or can fix only the comer nodes
horizontally.
(xii) Self weight of mat may be neglected or considered.
(xiii) Detail or specified output items may be obtained upon request.
(xiv) Modulus of subgrade reaction may be uniform or zoned.
10
(xv) Element dimensions may be generated by the program itself or may be given by
the user.
(xvi) Due to the limitation of the operating system, only sixty output files can be
opened by a program. So the program has the option of being specifically
instructed about the locations for which graph data are required. The graph
files for displacements, moments and shears are automatically generated for
those locations.
(xvii) Bending moments and shear forces along any column line may be given for a
strip of width equal to unity, the width of the corresponding column strip or
any specified width after averaging across the width requested.
Besides these, the software has many other features for fine-tuning. A limitation of
the software is that it can handle only mats with rectangular plan area.
The simplest element division scheme is adopted for mat. It makes both data
generation and result interpretation straight forward. The mesh used here is a
rectangular grid with the option of finer elements near column loads. The mesh is
characterized by the number of X and Y directional divisions of the grid and at
present, each of these can be increased lipto fifty. Details of the mesh are shown in
Fig 2.1.
2.2.5 Element and Node Numbering and Determination of the Front Width
11
enter line
Column
.........
tL~. y: :
••••••••••••
Fig. 2.1. Typical finite element mesh for mat foundation.
Front can be narrowed by numbering the elements serially from one comer of the mat
to the other and if the numbering of elements proceeds row wise along that direction
of the mesh which has the smaller number of divisions, the minimal front width for a
particular mesh can be obtained. As the Gauss point stresses are calculated element
wise, so the numbering of Gauss points is done element-wise. Since two point Gauss
integration is used (though three point integration is possible within the scope of the
program), each element contains four Gauss points. The node numbering scheme is
such that it makes nodal coordinate generation simple. Nodes are numbered serially
from one comer of the mat to the other comer. Details of the node and element
numbering scheme are given in Fig. 2.2 and Fig. 2.3.
Under the scheme adopted, total number of elements Ne , total number of nodes Nn ,
total number of Gauss points Ng and the front width Fw can be calculated as follows:
(2.1)
12
57
- 58 59 - 60 61 - 62 63 - 64 65
@ @ @ @
52 53 54 55 56
43 44 45 46 47 48 49 50 51
@ @ 4 @ @
38 39 40 41 42
29 30 31 32 33 34 35 36 37
@ @ @ @ t
24 25 26 27 28
15 16 17 18 19 20 21 22 23
(J) @ @ @
-
10 II 12 13 14 Y
I 2 3 4 5
- 6 7 8 9
Fig. 2.2. A typical finite element mesh for mat with element and node numbering.
LocaI y
Global Y t
• 52 • 51 • 56 • 55 • 60
• 59
• 64 63 Local X
• 49
• 50
• 53
• 54
• 57 • 58 • 61 • 62
• 36
• 35
• 40 • 39 • 44 • 43 • 48 • 47
• 33 • 34 • 37 • 38 • 41 • 42 • 45 • 46
• 20 • 19
• 24
• 23 • 28 • 27 • 32 • 31
• 17 • 18
• 21
• 22 • 25 • 26 • 29 • 30
• 4
• 3 • 8
• 7 • 12 • II • 16
• 15
• • • • •
Fig. 2.3.
I 2 5 6 •9
A typical finite element mesh for mat with Gauss point numbering and
10 • 13 • 14
-
Global X
global and local axes (Z axes are perpendicular to the XY plane at the
intersection of X and Y axes of the corresponding system).
13
No = Hv x ( 3 x lLJ + 2 ) + 2 X Hd + 1 (22)
Ng=4 x N, (2.3)
Fw=40+ 16 x (lLJ-I) (2.4)
where
lLJ = number of X directional divisions,
Vd = number ofY directional divisions.
As mat is a flat block type structure, selection of the global and local axes is very
straight forward. The global axis system is so selected that the bottom face of the
mat away from the colunms is the XY plane ( i.e. in case of greater thickness of mat
under the colunms, the extra thickness goes below the XY plane ), the leftmost face
of the mat is the YZ plane and the front face is the ZX plane. Details are shown in
Fig. 2.4.
/ YG1ob,1
., .
Mat
ZGlobal
• Xalob,1
Fig. 2.4. Global and local coordinate systems for mat foundation.
The program takes data and gives output with respect to the global axes but to
calculate shears and moments, stresses in local coordinate system are needed. The
advantage of the simple shape of mat under consideration is taken. by choosing a
14
local axis system parallel to the global one having its origin at the node of interest
(Fig. 2.4).
Supporting soil is modeled as Winkler medium . Soil springs may be uniform or may
be zoned to incorporate the coupling effect of soil springs.
Uniformly distributed soil springs are first concentrated at the nodes. This is done
element wise and by lumping the total spring stiffness under an element at its nodes.
There are two approaches to accomplish this purpose. These are :
(a) Equivalence between the nodal spnng stiffuesses and the distributed
spring stiffuess under the concerned element is made by adopting a
model which assumes that middle nodes takes twice as much stiffness as
the comer nodes.
(b) Middle nodes takes thrice as much stiffuess as the comer nodes
according to the contributing area concept as shown in Fig. 2.5.
Ae/l6 3Ae/16
Fig. 2.5. Distribution of nodal springs by contributing area (Ae = area of the element).
Nodal soil springs may be zoned by softening or hardening them in proportion to the
ratio of vertical soil pressure under the concerned node at some depth, in response to
15
a uniform surface pressure of unit magnitude, to the average soil pressure at the same
depth level due to the same surface load.
Both column axial loads and column base moments can be applied. Column axial load
may be applied as concentrated load or it may be distributed over an area equal to the
cross-section of the column by taking an element, designated as column element,
there and applying the load as a uniform surface pressure (Fig. 2.6). Similarly,
column base moments may be concentrated or distributed. The later is accomplished
by resolving column base moment into nodal loads acting at the nodes of the column
element along the sides parallel to the axis of the moment (Fig. 2.6).1n this case,
middle nodes take four times load than comer nodes as justified by the shape
functions.
To simulate the higher rigidity of column elements with respect to other portions of
the mat, modulus of elasticity of column elements is increased.
For calculating bending moments and shear forces, global stresses are transformed
into local stresses first. This is done as follows:
(25)
where
16
p
p
'11("
(a)
(b) I
Column Column
Element Element
Cx
p, =p / CxCy
(c)
Fig. 2.6. Different modes of application ofload : (a) Concentrated column axial load;
(b) Concentrated column base moment; (c) Distributed column axial load
'and (d) Distributed column base moment ('11(" = column base moment and
P = column axial load).
(D,] = direction cosign matrix of the local axes W.r.t. the global axes.
Here
17
0" xx 'zx
[0"]= 'xy 'yz (2.7)
O"zz
where
In Equation 2.6 and Equation 2.7, the first subscript of any stress tenn denotes the
nonnal to the surface on which the stress acts and the second one stands for its own
direction. In Equation 2.8, the first subscript of any angle tenn represents the local
axis and the second one denotes the global axis.
Now, as the selected local and global axis systems are parallel, [Dc] is a unit matrix.
So local stresses are the same as the global stresses. Sign of any stress is obtained by
multiplying the sign of the nonnal of the plane on which it acts by that of the
direction of the stress itself
Having found the local stresses, bending moments and shear forces per unit width can
be calculated at the Gauss or nodal points, using Gauss point or nodal stresses
respectively, according to the following fonnulae for any desired direction (Fig. 2.7).
18
where
crt M
t ~)
crb
(a) ~
(b)
r
Fig. 2.7. Calculation of (a) bending moment and (b) shear force from normal and
shear stresses respectively.
Bending moments are positive if they cause compression at the top fibers. Shear
forces are positive if they act downward on the right faces of z directional sections
through the mat at the corresponding locations.
Punching shear around a column is calculated by taking elements of width Ep,n around
the column and integrating the flexural shear forces per unit width at the nodes or
Gauss points along or near the outer peripheries of these elements (Fig. 2.8).
19
Here
where
c
C = distance of the reinforcement centroid from the nearest mat surface.
Line through
........•......•.. _ .•............••.
Gauss points
: Column.:
...... , . . . .. .... - ,.. Line through
: ................
Element: ~
Nodal points
...............;.-
....:
....... _.~ •.......•. ,-_ .................•...
................ ;...
8
Fig. 2. . Adjustment of element dimensions around columns for punching shear
calculation ..
To perform the numerical integration of the nodal or Gauss point shear forces in
order to obtain punching shear, Gauss integration scheme is used in case of Gauss
point stress analysis and Simpson's method is followed for nodal stress analysis.
20
2.3 LIMITATIONS OF THE PRESENT FINITE ELEMENT MODEL
The finite element model of mat foundation used in this study is subject to the
following limitations:
(iii) The value of modulus of subgrade reaction is taken to be uniform under the
entire mat.
(iv) Lumping of soil springs at the nodes is not consistent.
Displacements, bending moments and shear forces of lines along the element
boundaries or through the Gauss points and parallel to the mat edges i.e. X and Y
axes are presented graphically. For symmetric problems only half of the diagrams are
presented. Corresponding diagrams for any column strip as a whole are also
presented.
It is understood that portions of the mat under the columns are highly indeterminate.
In the present FE analysis, these portions are idealized by taking an element under
each column and by magnifYing the modulus of elasticity of that element w.r.t. that of
the rest of the mat under consideration. No more refinement of these portions is done
(though possible under the scope of the program used) as it is evident that due to the
high rigidity of the columns, bending moments and shear forces under the columns
will always be of negligible magnitude and hence will never come to play in the
design process of mat. It is also clear that because of this simplified modeling of the
underccolumn portions of mat, bending moments and shear forces calculated there
will be rather approximate. So in order to avoid ambiguity and misinterpretation, in
21
the graphical presentation of bending moments and shear forces, diagrams will be
made discontinuous at the under-column locations.
22
CHAPTER 3
3.1 GENERAL
A comparison of the finite element solution of mat foundation with that by other
methods is necessary only to delineate the versatility of finite element approach over
the conventional ones. Although there are various methods to analyze mat, the most
commonly practiced solution techniques seem to be appropriate targets for this
purpose. Considered from this point of view, FE solution is compared with
Conventional Method and ACI Approximate Flexible Method in this chapter. Also, in
order to take a close look deep into the problem, the traits of mat must be examined
which is why a parametric study, revealing its sensitivity to various parameters, is
necessary. This sensitivity analysis will help to understand which parameters play
important roles in defining the behavior of mat.
Although mat has a relatively small number of variables associated with it, its loading
pattern is nearly unpredictable, depending entirely on the purpose of the structure
that the mat supports and the idea of architects involved. In order to concentrate on
the purpose of this chapter, a certain case must be chosen and diagnosed. So a
particular problem is devised and the comparison and parametric study are made
using that example. But first, an infinite plate resting on elastic foundation is solved
using Ahmad's Thick Shell finite element and the results are compared with the
theoretical solution [Schleicher (1926)] to gain confidence.
23
3.2 VERIFICATION OF FINITE ELEMENT CODE
Schleicher (1926) solved the problem of an infinite plate of uniform thickness acted
upon by a concentrated vertical force and resting on elastic foundation. This method
was later adopted by ACI (1966), along with other modifications, under the title of
ACI Approximate Flexible Method. As ACI method is described in detail in appendix
C, the infinite plate solution is not discussed here.
Modulus of subgrade
reaction : 100.00 kef'
Type ofloading : single concentrated column load
Location of the load : center of the mat
Value of the load : 100.00 k
Location of bending
moment and shear
force calculation . : face of the load
24
Deflection, bending moment and shear force diagrams of the above mat, as obtained
form FE and elastic solutions, are shown in Fig. 3.1 through Fig. 3.3. Fig. 3.1
through Fig. 3.3 show that agreement between these two solutions is excellent.
In this example, all typical values are selected. Calculation of dead load and live load
is in accordance with the USD method of ACI (1983), the most widely accepted code
of practice. Proper live load reduction is employed in calculating column loads.
However, before arriving at the final loading condition, following loading cases are
where
LL = live load,
DL = dead load,
WL = wind load.
25
0.0002
0.0000
-0.0002
-0.0004
g
c:: -0.0006
o
13
Cll
0=
Cll
o
-0.0008
-0.0010
-0.0012
-0.0014
o 20 40 60 80 100 120
X (fl)
25
~ 20
6
"E 15
G>
E
o 10
:2
o
-5
o 20 40 60 80 100 120
X (ft)
Fig. 3.2. Comparison of bending moment calculation by FE method and
elastic solution.
20
10
~
~ 0
~
01
G>
.c -10
Cf)
-20
-30
-40
o 20 40 60 80 100 120
X (ft)
Conventional and ACI method can not take into account column base moments
which are essential products of loads on frames. But finite element analysis has the
capability of taking care of virtually any type ofload. Now, column base moments are
not significant in case of gravity loading, but they are substantial in magnitude for
lateral loading. But to keep the situation comparable for all the analysis methods,
column base moments must be avoided. So for the example mat (Fig. 3.4),
conventional practice of load calculation is adopted. Column base moments are
neglected and the effect of wind load is materialized by reducing the column loads on
the leeward side of the building and increasing those on the windward side. However,
due to 25% reduction of loads in case II, load case I governs in this particular
example. Since this load case produces symmetric column loads, advantage of
analyzing quarter of the mat is taken.
For the calculated loads and the bearing capacity of soil considered, total required
bearing area is more than 50% of the available foundation area i.e. the plan area of
the mat. This justifies the use of mat in transferring the calculated loads to soil.
Column sizes are designed to be adequate to transfer the calculated loads. From
experience, it is expected that mat thickness is governed by shear and through a trial
and error approach 3 ft thickness is tentatively adopted for the example mat.
No. of columns 16
Column loads,
comer columns 268.00 k
26
edge columns 435.50 k
central columns 697.00 k
Column size and shape,
Center line
10'
20'
Width of overhanging portion
C2(1.5'xl') LINE 1
o 435.5 k
2.5'
-+--f----f--f-
2.5' 20' 10'
27
Before taking the job on, some critical aspects of finite element analysis of mat must
be examined and justified. The next article is devoted to this purpose.
Versatility and power of finite element method enable us to reflect reality in structural
analysis more than any other numerical approach. The goal of this chapter is to show
the versatility of finite element technique over other available methods. So efforts are
made to optimize structural idealization of mat in finite element analysis.
One of the advantages of FE analysis of mat foundation is its ability of treating the
column loads as distributed load over the respective column cross-sections. This is
straight forward and adopted right away. However, some other factors associated
with FE solution of mat foundation need be investigated in detail. These are
discussed below.
28
0.0310
0.0305
0.0300
:2
~
<:
0
""Ql
U
0:: 0.0295
Ql
1::>
E
::J
E
'x
'"
:2
0.0290
0.0285
0.0280
o 100 200 300 400
No of elements
Columns are monolithically built with mat and they act integrally with it. Height of
columns highly increases the rigidity of mat at these locations. This effect is realized
by increasing the value of the modulus of elasticity of the portions of mat under the
columris. This results in a shift of maximum moment value from column center to
column face (Fig. 3.7) and a slight reduction of deflection under columns (Fig. 3.6).
No significant change in shear force diagrams, except for a slight reduction of column
face shears (Fig. 3.8), occurs.
To quantifY this effect, column deflections, column face moments and shears are
plotted against the ratio of the modulus of elasticity of the under -column portions and
that of the rest of the mat in Fig. 3.9 through Fig. 3.1 L It appears that effect of
column rigidity dies out with the increase of this modulus of elasticity ratio.
Examining these, a modulus of elasticity ratio of 8 is found to be justifiable.
In the FE analysis of mat, soil under mat is replaced by a fictitious support consisting
of an infinite number oflinearly elastic uncoupled springs (Winkler medium). Material
of mat is taken to be concrete and treated as linearly elastic. Column loads are
applied after distributing them over the respective column cross-sectional areas.
Modulus of elasticity of mat material is taken to be eight times higher under the
columns than that at other locations.
29
-0.018
-0.020
~
~
~
c
0 -0.022
UQ)
""0
Q)
-0.024
o 10 20 30
X (ft)
(a)
-0.016
g -0.018
c:
o
U
Q)
""ClQ)
-0.020
o 10 20 30
X (It)
(b)
Fig. 3.6. Deflection diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4) for different
column rigidities (Mat thickness 3 ft).
o
- Stiffness ratio under column = 1
100 - Stiffness ratio under column = 3.375 CF CF
--..-.. Stiffness ratio under column = 8
~ Stiffness ratio under column = 27
50
~ CF CF
-
6
c
Ql
E
0
o
~
-50
-100
o 5 10 15 20 25 30
X (ft)
(a)
150
- Stiffness ratio under column = 1
- Stiffness ratio under column = 3.375
CF CF
--..-.. Stiffness ratio under column 8=
100 ~ Stiffness ratio under column = 27
e- 50
~, CF CF
.:<:
~
C
Ql
E 0
0
~
-50
-100
o 5 10 15 20 25 30
X (ft)
(b)
Fig. 3.7. Bending moment diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4) for
different column rigidities (Mat thickness 3 ft, CF = column face).
0.030
o
o 5 10 15 20 25 30
Fig. 3.10. Effect of column rigidity on column face bending moment (Mat
thickness 3 ft).
LINE 2 (Fig. 3.4). These are the most critical of all locations and will play the
governing role in the design phase. Also, due to the symmetry of the problem these
column lines, which can in other words be denoted as C I C2 and C3C4 lines (Fig.
3.4), also represent column lines CIC3 and C2C4 (Fig. 3.4). Since presentation is
graphical and only for a typical case, and hence only qualitative, numerical
comparison is not of much practical worth. So response of mat to column loads, as
found by different methods, is compared mainly qualitatively. However, when felt
necessary quantitative comparison is also made.
Maximum negative moments in-between the columns and column face positive
moments along the column lines are used for flexural design of mat. Fig. 3.13 shows
that Conventional method fails to predict any positive moment at the column faces
whereas ACI method gives substantially lower column face positive moments
compared to FE method. Conventional method predicts maximum negative moments
with reasonable accuracy. But ACI method overestimates maximum negative
moments. However, this overestimation is less pronounced near the edges of mat
30
0.00
-0.01
-0.02
~
~
c
0 -0.03
U
(])
-=
0
(])
-0.04
-0.05
0 10 20 30
X (ft)
(a)
0.00
-0.02
-
~
~
c
-0.04
0
U
(])
-0.06
-=
(])
0
-0.08
o 10 20 30
X (ft)
(b)
Fig. 3.12. Comparison of deflection calculation of (a) LINE 1and (2) LINE 2
(Fig. 3.4) by different methods (Mat thickness 3 ft).
- Finite Element method
100 __ ACI approximate flexible method CF CF
~ Conventional method
50
-50
-100
0 5 10 15 20 25 30
X (ft)
(a)
200
- Finite Element method
150 -- ACI approximate flexible method
~ Conventional method
CF CF
100
~ 50
<:=
~
-
~
~
C
<Il
E
0
0 -50
:2
-100
-150
-200
0 5 10 15 20 25 30
X (ft)
(b)
Fig. 3.13. Comparison of bending moment calculation of (a) LINE 1 and (b) LINE 2
(Fig. 3.4) by different methods (Mat thickness 3 ft, CF = column face).
than near the mat center. For the mat under consideration, a list of all these moments
as estimated by different analysis methods in comparison to those from FE analysis
. are listed in Table 3. I.
Fig. 3.14 shows that conventional method gives much low shear compared to the
other methods. This method do not picture the variational pattern of shear because of
treating individual column strips as a whole throughout the entire analysis. Prediction
of shear force by ACI method appears to be satisfactory. As a matter of fact, shear
force away from the columns as calculated by ACI method is in excellent agreement
with that from FE analysis. However, shear forces near columns are important for
mat design and AsI method overestimates their values. A comparison of shear force
per unit widVar the face of column C4 shows that ACI shear force is 24% in
excess of FE value whereas Conventional shear force is only 17.5% of the FE shear.
However, near column C2 an exception to this trend is observed. Here FE shear is
somewhat higher than ACI shear.
31
I'" •
I"..
•••
100
- Finite Element solution CF
75 - ACI approximate flexible method
----A--- Conventional method
50 CF
25
~ 0
~
'"
OJ
.c
(f) -25
-50
-75 CF
CF
-100
0 5 10 15 20 25 30
X (ft)
(a)
50
~
~ 0
~
'"
OJ
.c
(f)
-50
-100
CF
-150 CF
0 5 10 15 20 25 30
X (ft)
(b)
Fig. 3.14. Comparison of shear force calculation of (a) LINE 1 and (b) LINE 2
(Fig. 3.4) by different methods (Mat thickness 3 ft, CF = column face).
...
•
\...:
3.7 PARAMETRIC STUDY OF MAT FOUNDATION
To investigate the sensitivity of mat to the variation of its deferent properties, the
same example (Fig. 3.4) is used except that the property under investigation is varied
within certain practical range and the magnitude and trend of variation of deflections,
shears and moments are studied carefully. Some important locations are selected for
this purpose. The locations should be so selected that parametric effects are
magnified there. Also, these positions must be important from design point of view.
These two criteria, though not always satisfied together, has led to the positions
presented in Table 3.2.
Table 3.2. Locations of the example mat (Fig. 3.4) for the calculation of various
items for parametric study.
Once column locations are selected, loads calculated and column sizes determined
from structural and architectural points of view; there are three independent
parameters associated with a mat. These are : modulus of elasticity of mat material
(which in case of concrete is determined from ultimate concrete strength), modulus of
subgrade reaction of soil and mat thickness. Poison's ratio of concrete varies within a
very narrow range and has negligible effect on mat behavior. For the present study
32
the following ranges of these parameters are selected which cover all practically
typical values.
Apart from the parameters mentioned above, there are three more variables
associated with mat foundation which have important influence on mat behavior.
These are : column spacing, width of overhanging portion and column dimension.
Effect of these three variables on the structural response of mat is also investigated.
Usually column spacing is determined from architectural and utility point of view.
Structural feasibility of providing the proposed column spacing plays an important
role in this context. Extent of the overhanging portion is determined from soil .
condition, loading and property line considerations. Column dimension is a function
ofload and column spacing. Column dimensions are designed irrespective of the mat
the columns may rest on later. However, effect of column dimension is of interest to
investigate.
It should be noted that due to the symmetry of the example mat, it is sufficient to
concentrate on moments and shears in only one direction. So values along LINE I
and LINE2 (Fig. 3.4) are used in the following discussion.
ultimate concrete strength (f,) are presented in Fig. 3.15 through Fig. 3.17. Also
33
percent changes in the values of these items for a change in f e from 2 ksi to 6 ksi are
given in Table 3.3.
Table 3.3. Variation of displacement, moment and shear w.r.t. concrete strength
fe', when £,' is increased from 2.00 ksi to 4.00 ksi.
Deflection decreases as concrete strength increases (Fig. 3.15). The decrease is less
pronounced near the center of the mat. Over the gamut of f e considered, deflection
of column C4 (Fig. 3.4) is reduced 1.9% whereas reduction of column C1 (Fig. 3.4)
deflection is 7.6%.
Table 3.3 and Fig. 3.16 indicate that as the strength of concrete increases there is a
reduction in C2 and C4 (Fig. 3.4) column face moments but changes in C1 and C3
(Fig. 3.4) column face moments are negligible. Negative moment increases with the
increase in concrete strength and this increase is the greatest at the center of LINE 2
(Fig. 3.4) .i.e. at the mat center.
Shear is not virtually affected by the change in concrete strength (Fig. 3.17). Only C3
(Fig. 3.4) column face shear shows an increase of 1.3% and all other column face
shears undergo insignificant changes when fe is increased from 2.00 ksi to 6.00 ksi.
34
0.026
0.025
0.024
0.023 - Column 1
- Column 2
---4r- Column 3
2'
~ 0.022 -T- Column 4
c
o
U
Jg
Q)
0.021
"0
C
E
:J 0.020
o
~.:
U
0.019
0.018
0.017
0.016
2 3 4 5 6
Ultimate concrete strength (ksi)
Fig. 3.15. Effect of ultimate concrete strenght on column deflection (Mat thick-
ness 3 ft).
120
•••
••• ••• ••• •••
~ 100 • •
-=~, • • •
-
~
-'"
c
Q)
E
80 -
-
Column 1
Column 2
0 --A-- Column 3
E
Q) ---.- Column 4
0 60
.•...
<U
c
E
::J
<5 40
U
••• ••• ••• ••• •••
20 ~ • • • •
2 3 4 5 6
Ultimate concrete strength (ksi)
(a)
75
• • • •
:;:;-
70 •
~, 65 ••••
•••• ••••
-
6
E
c
Q) 60
••••
••••
0
E - In-betweencolumn C1 and C2 of LINE 1
55
Q) - Center of LINE 1
>
--A-- In-betweencolumn C3 and C4 of LINE 2
""OJ
<U 50
Q)
---.- Center of LINE 2
C
E 45
::J • • •
E
•• •
'x<U 40 •••
•••
:2 •••
35 .•... •••
2 3 4 5 6
Ultimate concrete strength (ksi)
(b)
Fig. 3.16. Effect of concrete strength on (a) positive and (b) negative moment (Mat
thickness 3 ft).
.~ l..,:.
3.7.2 .Effect of Modulus of Subgrade Reaction
Effect of modulus of subgrade reaction is presented in Fig. 3.18 through Fig. 3.20
and in Table 3.4. As expected, deflection is substantially affected by modulus of
subgrade reaction. As the soil becomes stiffer j. e. as modulus of subgrade reaction of
soil increases, deflection reduces and this reduction is at a very fast rate for a
modulus of subgrade reaction value up to 225 kef (Fig. 3.18). Table 3.4 shows that
for a change in the value of modulus of subgrade reaction from 100 kcf to 500 kcf, all
the column deflections decrease by up to about 75%.
Column face positive moments increase with the increase of modulus of subgrade
reaction and maximum negative moments decrease. It may be noticed from Table 3.4
that effect of modulus of subgrade reaction on positive moment at column face C4
(Fig. 3.4) is negligible but that on negative moment at the center ofLINE2 (Fig. 3.4)
is considerable. Fig. 3. 19 indicates that the effect of modulus of subgrade reaction on
moments gradually decreases beyond a value of modulus of subgrade reaction equal
to 225 kcf However, CI-C2 and C3-C4 (Table 3.2) negative moments are
exceptions to this conclusion.
35
- Column 1
0.025 - Column 2
-A- Column 3
-.- Column 4
0.020
g
c
0
U
1Il
<;: 0.015
1Il
"0
C
E
:J
(5
U
0.010 "•
0.005
E
o
~ 50
>
:;:; - In-between column C1 and C2 of LINE 1
'"
Cl - Center of LINE 1
Q) ---A- In-between column C3 and C4 of LINE 2
c 40
j 30 ~ C'"~"f UN" : :
(b)
Fig. 3.19. Effect of modulus of subgrade reaction on (a) positive and (b) negative
bending moment (Mat thickness 3ft).
Shear forces at the faces of columns C2 and C4 (Fig. 3.4) are not practically affected
by modulus of subgrade reaction (Fig. 3.20) but those at the faces of columns C I and
C3 (Fig. 3.4) are decreased by 2.8% and 3.6% respectively over the entire range of
modulus of subgrade reaction considered.
Fig. 3.21 through Fig. 3.23 depict the effect of mat thickness on deflection, moment
and shear force. Also, the overall response of the example mat to the variation of mat
thickness is summarized in Table 3.5. It is evident that all the items are very sensitive
to mat thickness.
Table 3.5.
Variation of displacement, moment and shear W.r.t mat thickness t,
when t is increased from 2.00 ft to 4.00 ft.
Deflection under the comer column C I (Fig. 3.4) is more sensitive to mat thickness
than that under the central column C4 (Fig. 3.21). Also the rate of change of column
deflections decreases after a thickness of2.75 ft (Fig. 3.21).
Mat thickness affects C2 and C4 (Fig. 3.4) column face moments substantially (Fig.
3.22). Except column C3 (Fig. 3.4), all other column face positive moments decrease
36
0.030 - Column 1
-- Column 2
~ Column 3
----...- Column 4
0.028
0.026
2
~0.024
o
U<Il
~
"0
~ 0022
OJ
ou
0.020
0.018
0.016
,
120
~, 100
-'~"
C
Q) -l.c1234_c1
E 80 -l.c1234_c2
0
E ----A-- l.c1234_c3
Q)
---y- l.c1234_c4
-'"
u
c
E
::l
60
0
0 40
•••• •••• •••• •••• ••••
20 ~ ~ • • •
2.0 2.5 3.0 3.5 4.0
Mat thickness (ft)
(a)
80
£ 70
~,
-
6
c
Q)
E
60
0 - negetive_c1.2
E - negelive_c2.cenler
Q)
> 50 ----A-- negetive_c3.4
""Cl negetive_c4.cenler
'"
Q)
---y-
c
E 40
::l
E
'x
'"
:2 30
(b)
Fig. 3.22. Effect of mat thickness on (a) positive and (b) negative bending moment.
,
90
• • • • •
85
- Column 1
~ 80 - Column 2
::: -A- Column 3
:>2
~
~ ---.- Column 4
III 75
CD
.£:
'"
CD
u 70
-
III
c
E 65
::J
0
060
55
• • • • •
2.0 2.5 3.0 3.5 4.0
Mat thickness (ft)
(a)
700
• • • • •
200
2.0 2.5 3.0 3.5 4.0
Mat thickness (ft)
(b)
Fig. 3.23. Effect of mat thickness on (a) flexural and (b) punching shear.
as mat thickness is increased. Negative moments are very sensitive to mat thickness.
As mat thickness increases, negative moments increase too and this increase ranges
from 20.9% for CI-C2 location (Table 3.2) to 74.7% for LlNE2 CL location (Table
3.2). As mat thickness is increased, the rate of change of negative moments gradually
decreases near the ends of any column line whereas that increases near the center.
While Cl and C4 (Fig. 3.4) column face shear forces are virtually unaffected by the
variation of mat thickness, C2 and C3 (Fig. 3.4) column face shear forces are quite
sensitive to this parameter (Fig. 3.23). However, C2 column face shear decreases and
C3 column face shear increases with the increase of mat thickness with their rate of
change being gradually decreased as mat gets thicker.
To investigate the effect of column spacing on mat behavior, two more problems
have been designed in addition to the 20' x 20' column grid problem of Fig. 3.4, one
mat with 12' x 20' column grid and another with 15' x 20' column grid. These are
shown in Fig. 3.24 and Fig 3.25 respectively. Since the purpose is to show the effect
of column spacing, total load carried my all these three mats and all other properties
of them are kept the same.
Bending moment, shear force and deflection diagrams of LINE 1 & 2 (Fig. 3.4, Fig.
3.24 and Fig. 3.25) are presented in Fig. 3.26 through Fig. 3.28. It is observed that
column spacing has remarkable effect on mat response. As expected, more columns
to transfer the same load means a more uniform load distribution, which in tum
means a reduction in deflections, bending moments and shear forces. This fact is
reflected in Fig. 3.26 through Fig. 3.28. From Fig. 3.26, it is clear that reduction in
deflections due a closer column distribution is particularly significant for LINE 1.
Attenuation of the maximum values of both positive and negative moments is
37
noticeable. Shear forces undergo significant reduction in magnitude when column
spacing is reduced.
Center line
10'
295 k 417 k 417 k LINE 1
o o o
2.5
o 185kl 0
160 k
0
160 k
LINE 2
+i-
2.5' 12'
i- 12'
i- 6'
t--
Fig. 3.24. Plan of the mat with 12' x 20' column grid.
Overhanging portion is the part of any mat projected outside the exterior most
column line on any side of the mat. For the present study, its width is defined as the
distance between the center of the exterior most column line and the nearest edge of
the mat (Fig. 3.4). Effect of the width of overhanging portion on mat behavior is
depicted in Fig. 3.29 through Fig. 3.31. The graphs are obtained by analyzing the
example mat of Fig. 3.4 by varying its width of overhanging portion from 0.00 ft to
7.50 ft. It is observed that as the width of overhanging portion is increased, deflection
pattern changes with the maximum deflection of any column line being shifted from
38
the end towards the center of the column line. Also, deflection pattern becomes more
uniform with a reduction in magnitude (Fig. 3.29). With the increase of the width of
the overhanging portion, a decrease in the magnitude of negative moment and an
increase in the magnitude of positive moment is observed (Fig. 3.30). Particularly,
reduction of negative moment is significant. Shear force is not much affected by this
parameter. However, from Fig. 3.3 I it can be observed that when any overhanging
portion is absent i.e. exterior most column faces are flash with the mat edges, LINE I
shear undergoes a considerable magnification.
Center line
10'
20'
L Width of overhanging portion
2.5'
0 0 --
+4-
2.5' IS'
.~
15' ."
Fig. 3.25. Plan of the mat with 15' x 20' column grid.
Since column loads are distributed over the respective column cross sectional areas,
smaller cross sections of columns mean greater intensity of column loads. This
increase in force concentration is expected to increase deflection, bending moment
and shear force magnitudes in the vicinity of the columns. The example mat is
39
-0.012
-0.014
-0.016
isc -0.018
o
15
Ql -0.020
;;::
Ql
o
-0.022
-0.024
-- X directional column spacing = 20 It
-0.026
-- X directional column spacing =15 It
---6- X directional column spacing = 12 It
o 10 20 30
X (ft)
(a)
-0.016
isc -0.018
o
UQl
~
o -0.020
o 10 20 30
X (ft)
(b)
Fig. 3.26. Deflection diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4, 3.24 & 3.25)
for different column spacings (Mat thickness 3 ft).
- X directional column spacing = 20 ft
100 _ X directional column spacing =
15 ft CF CF
---- X directional column spacing = 12 ft
50
2
~,
-'~"
c Q) 0
E
0
:2
-50
-100
o 5 10 15 20 25 30
X (ft)
(a)
50
~
<I:'
,
e
""
'E
Q) 0
E
0
:2
-50
-100
o 5 10 15 20 25 30
X (ft)
(b)
Fig. 3.27. Bending moment diagrams of (a) LINE 1 and (b) LINE 2 ( Fig. 3.4,
3.24 & Fig. 3.25) for different column spacings (Mat thickness 3 ft,
eF = column face).
80 -- X directional column spacing = 20 ft CF
---.- X directional column spacing = 15 ft
60 -A-- X directional column spacing = 2ft CF
CF CF
40 CF
20
2'
~ 0
~
<lJ
(])
.<::: -20
en
-40
-60 CF
-80 CF
0 5 10 15 20 25 30
X (ft)
(a)
2'
~
~
~ 0
<lJ
(])
.<:::
en
-50
CF
-100 CF CF
0 5 10 15 20 25 30
X (ft)
(b)
Fig. 3.28. Shear force diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4,
3.24 & Fig. 3.25) for different column spacings (Mat thickness
3 ft, CF = column face).
0.000
- Width of overhanging portion = 0.0 It
-0.005 ---- Width of overhanging portion = 2.5 It
~ Width of overhanging portion = 7.5 It
-0.010
-0.015
~
~
~
c -0.020
0
U Q)
-0.025
-=
Q)
0
-0.030
-0.035
-0.040
0 10 20 30
X (ft)
(a)
0.000
-0.010
~
::s
c
0
-0.015
ti
Q)
-=
Q)
0
-0.020
-0.025
o 10 20 30
X (ft)
(b)
Fig. 3.29. Deflection diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4) for different
widths of overhanging portion (Mat thickness 3 ft).
150
- Widht of overhanging portion = 0.0 ft
- Widht of overhanging portion = 2.5 ft CF
CF
100 --A- Widht of overhanging portion = 7.5 ft
50
~
~,
-
-'~
"
c:
Q)
E
a
0
:2 -50
-100
-150
-5 a 5 10 15 20 25 30
X (ft)
(a)
150
- Widht of overhanging portion = 0.0 ft
- Widht of overhanging portion = 2.5 ft CF CF
--A- Widht of overhanging portion = 7.5 ft
100
e-, 50
-
-'~
Q)
E
"
c:
a
0
:2
-50
-100
-5 a 5 10 15 20 25 30
X (ft)
(b)
Fig. 3.30. Bending moment diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4)
for different widhts of overhanging portion ( Mat thickness 3 ft,
eF = column face).
150 - Width of overhanging portion = 0.0 ft
- Width of overhanging portion = 2.5 ft CF
--A-- Width of overhanging portion = 7.5 ft
100
50
~
~
~ 0
Ol
Q)
.<:::
(jJ
-50
. -100
-150 CF
-5 a 5 10 15 20 25 30
X (ft)
(a)
~ 0
~
Ol
Q)
.<:::
(jJ
-50
CF
-100 CF
-5 0 5 10 15
, 20 25 30
X (ft)
(b)
Fig. 3.31. Shear force diagrams of (a) LINE 1 and (b) LINE 2 ( Fig. 3.4 ) for
different widths of overhanging portion (Mat thickness 3 ft, CF =
column face).
analyzed with all the columns being square in cross-section and ranging in dimension
from 1.00 ft to 2.00 ft. Results are presented in Fig. 3.32 through Fig. 3.34. It is clear
that predictions just made resemble the results. The plots show that effect of column
size is almost solely concentrated near the localities of the columns except for LINE 1
deflections (Fig. 3.32), which are equally affected throughout. However, it is also
seen form Fig. 3.33 and Fig. 3.34 that bending moment and shear force diagrams for
columns with larger sizes remain more or less aligned with those for narrower column
cross-sections, except for a reduction in the peak values of moments and shears at the
faces of the columns. This is a good indication in the sense that analysis with one
size of columns may well be applicable if column sizes are slightly changed later for
. any reason.
From the above parametric study, it becomes clear that behavior of mat is principally
determined by modulus of subgrade reaction and mat thickness. It has been found
that mat thickness is the most important parameter of all. It has also been observed
that flexural shear of mat is the least sensitive item to most of the parameters. Study
of the effect of width of overhanging portion and column spacing reveals that these
two variables substantially affect the magnitudes of deflections, bending moments and
shear forces of mat. Small column spacing and wide overhanging portion result in a
more uniform distribution of column loads and thereby reduce deflections, bending
moments and shear forces.
40
-0.020
-0.022
-0.024
isc -0026
o
U
QJ
-0.028
0:::
QJ
o -0.030
-0.032
- 1.0 It square column
- 1.5 It square column
-0.034
--A- 2.0 It square column
o 10 20 30
X (ft)
(a)
-0.014
-0.016
-0.018
~
.:=
~
c -0.020
0
t5QJ
0:::
QJ -0.022
0
-0.024
o 10 20 30
X (ft)
(b)
Fig. 3.32. Deflection diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4) for different
column sizes (Mat thickness 3 ft).
-- 1.0 It square column CF CF
100 -- 1.5 It square column
----A-- 2.0 It square column
50
2
~ CF CF
-
~
~
c:
a>
10 0
0
:2
-50
o 5 10 15 20 25 30
X (ft)
(a)
CF CF
-- 1.0 It square column
-- 1.5 It square column
----A-- 2.0 It square column
100
50
~,
-
6
c:
a>
10
0 0
:2
-50
o 5 10 15 20 25 30
X (ft)
(b)
Fig. 3.33. Bending moment diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4) for
different column sizes (Mat thickness 3 ft, CF = column face).
CF
-- 1.0 ft square column
-- 1.5 ft square column
~ 2.0 ft square column
50
CF
2'
~ 0
~
Ol
Ql
.<=
(j)
-50
CF
CF
0 5 10 15 20 25 30
X (ft)
(a)
CF
50
~
~
~ 0
III
Ql
.<=
(j)
-50
CF
-100
CF
0 5 10 15 20 25 30
X (ft)
(b)
Fig. 3.34. Shear force diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4) for different
column sizes (Mat thickness 3 ft, CF = column face).
CHAPTER 4
4.1 GENERAL
This chapter will focus on the relative economy in design achieved by finite element
analysis. Also, confidence will be gained in more economic shaping and dimensioning
of mat foundation. In the ctirrent study, Ultimate Strength Design (USD) method of
American Concrete Institute (ACI) is followed.
Some structures are complicated to analyze but simple to categorize. For example, a
caisson foundation. Analysis of caisson is difficult but its loading pattern and
geometry do not vary much. So an approximate analysis based on parametric study is
simple to develop and simpler and efficient to use. But structures like mat differ from
this type by a great deal.
Mat looks simple. Once everything is set by engineers and architects for a particular
project, it has a relatively smaIl number of parameters to be dealt with; namely elastic
properties of mat material, modulus of subgrade reaction of the supporting soil and
mat thickness. Difficulty with the formulation of the characteristics of mat arises from
its loading pattern.
Generalization of loading for mat is a very difficult task. Structurally, mat acts
integrally with the building frame it supports and resists loads which do not act on it
directly but comes through the frame. To analyze mat separately, column base
41
moments and axial forces produced in response to the external loads acting on the
frame is calculated first and applied on mat later. Locations of these columns i.e. of
these loads depend on architectural planing of the building and column spacing.
These two variables, which are selected from a large number of options and hence
almost unpredictable, determine the magnitudes of column base moments and axial
forces, which in turn determine the size of the columns or in other words the
distribution of these loads on mat. In addition, two other very important factors
involved in load calculation are the number of stories and the number of bays of the
building frame supported by the mat. It is clear that loads on mat are the outcomes
of a series of interconnected processes.
Floatation effect and plate behavior enable most mat settlements to be limited to
tolerable magnitudes. Also it is easier to conceal any reasonable amount of overall
deflection by providing higher plinth level initially. A problem of more considerable
42
concern is the differential settlement, which may induce detrimental stresses in the
superstructure. This is why a strict control of differential settlement is essential.
Mc Donald and Skempton (1955) made a study of 98 buildings which was later
substantiated by Grant et. al. (1974) from a study of 95 additional buildings of more
recent construction. Also, Fled (1965) investigated a large number of specific
structures. All these sources combinedly give rise to a conclusion that for mat on
clay, recommended design differential settlement is 2.5 to 4 inches and for
satisfactory structural performance deferential settlement should never be greater
than 3 to 5 inches. For mat on sand, the respective ranges are 1.5 to 2.5 inches and 2
to 3 inches. These are, however, safe ranges determined from only statistical surveys.
According to USSR building code, acceptable slope between two points of mat
supporting concrete frame structures is 0.002, with an average maximum settlement
of 4 inches.
Design of mat for bending moment is straight forward. Since mat suffers two
directional bending, strips of mat is selected in two perpendicular directions and the
strips are designed as beams for critical positive. (upside compression) and negative
(downside compression) moments. ACI (1983) restricts that a minimum steel ratio be
maintained for any section, both near the top and near the bottom of mat. Also,
design positive moments should be those at the faces of the columns according to
ACI (1983). Formulae used for calculation are given below.
(4.3)
43
f.' -4000
131=a.85-a.a5x-c-_ and
1000 a.65~131~o.85 (4.4)
Here, qJ is the strength reduction factor which is equal to 0..9 for moment and
A. = steel area,
b =
width of the mat strip to be designed,
fc' = ultimate concrete strength,
fy = yield strength of steel.
Perhaps the most important design criterion for mat foundation is shear force. Since
mat is a plate structure, shear reinforcement is avoided from practical considerations.
In stead, mat is made sufficiently thick to encounter shear. Presence of heavy column
loads concentrated on very small areas makes consideration of punching shear, in
addition to flexural shear, essential. As per ACI code (1983), punching shear is
calculated as the total shear force acting on the surface of a peripheral section around
the column under consideration, taken at a distance d/2 from each face of that
column, where d is the effective depth of mat. For the convenience of description, the
area encompassed by this section will subsequently be referred to as punching area,
its perimeter as punching perimeter and its width on any side as punching width.
Physically, punching shear is the column load less the total soil pressure under the
punching area of that column. Both the flexural and punching shear stresses must be
lower than the respective ACI (1983) shear capacities of the mat section. According
to ACI (1983), flexural shears to be considered are those at a distance d away from
the column faces. ACr design formulae for shear capacity calculation are :
44
Ultimate punching shear strength, Vpu =
4eNf[ psi (4.7)
Ultimate flexural shear strength, Vfu = 2cpf£[ psi (4.8)
Where <pis the strength reduction factor and equal to 0.85 for shear.
The most commonly used design approach is to divide mat into column strips and
design the column strips individually. For the present discussion, the example mat
shown in Fig. 3.4 is used. This particular example mat is symmetric. So design of
column strips centered on LINE I and 2 (Fig. 3.4) is demonstrated here. These strips
will subsequently be referred to as strip 1 and strip 2 respectively. For this example,
and all other examples considered in this chapter, the following material properties
and design data are selected.
The value of modulus of subgrade reaction is selected for weak clayey soil to be
representative of Dhaka clay. Since mat thickness is governed by shear and often the
flexural design of mat requires minimum steel, it is felt convenient in the present
study to calculate some critical values for some probable thickness of the mat under
consideration in advance. In calculating these values, it is assumed that a single layer
of steel will be sufficient for any bending moment that may act in the example mat
(Fig. 3.4).
45
r,
\J'
4.4.1 Some Frequently Used Design Values
For the problem at hand, the maximum and minimum steel ratio (Prna, and Pmin),
ultimate punching and flexural shear capacity per unit width (Vpu/copand Veu/cop),
maximum and minimum steel area per unit width (Ai"",x and Aimin) and maximum
and minimum ultimate bending moment capacity per unit width (Mrn""copand M )
rninlcop
are calculated for mat thickness ranging from I. 00 ft to 3.50 ft and are presented in
Table 4.1. Also, readily solvable formulae to calculate many other items are derived
for the material properties selected. In the following discussion, these values and
formulae will be directly used which will save computational effort. Interpolation will
be done for intermediate values, when it will be necessary.
Table 4.1.
Values of Vplcop,Vflcop, M,"""cop, MnUnicop
, Ainmx , Aiminfor different
thicknesses of mat (Fig. 3.4).
46
.'
Pmax = 0.0214
(4.9)
Pmin = 00033
(4.10)
Punching shear capacity per unit length of punching perimeter,
Vp/cap = 30.965 x d k (d in inches) (4.11)
Flexural shear capacity per unit width,
V£,,,,p = 15.483 x d k (din inches) (4.12)
d = t - 3.5 (all in inches) (4.13)
Mmaxlcap = 2
093674 X d k-ft/ft (d in inches) (4.14)
Mminlcap = 2
0.17469 X d k-ft/ft (d in inches) (415)
Awmax 3.08160 x d in 2
/ ft (d in ft) (4.16)
Awmin 0.48000 x d in 2
/ ft (d in ft) (417)
Effective depth from punching shear requirement,
dpunch =
~(-c+Jc2 +4.65x Vp) in
drequiredlmoment =
J 54p x (I ~8.85 x p)
drequiredlshear =
0.775 x V (Vin klftand din inches) (4.23)
V = Flexural shear
47
4.4.2 Deflection of Columns
Only differential column deflections are important for the purpose of design.
Conventional method has no provision for deflection calculation. Column deflections
of the example mat, as obtained by ACI and FE methods, are given in Table 4.2.
Viewed in light of the deflection requirements for mat, it is found that deflection does
not impose any problem in any method for this particular mat. Even the maximum
differential deflection is well below the allowable limit. However, it is worth noticing
that due to overestimation of deflection near the edge and underestimation of that
near the center of mat by ACI method, for larger mats with heavier column loads
. differential deflection may emerge as a problem of serious concern in case of ACI
analysis, whereas the present FE analysis may reveal the solution to be acceptable for
the same case.
Table 4.2. Column deflections of the example mat (Fig. 3.4) by different methods.
Punching shear can be calculated either by integrating the flexural shear along the
punching perimeter or by deducting the total soil pressure under the punching area
from the axial column load. The second method is convenient with ACI and
Conventional method while the first one is more suitable for FE method. However, to
show the accuracy of the first approach with FE method and to highlight the
excellence of Ahmad's thick shell element in calculating shear, both the methods are
applied to calculate punching shear around column C4 (Fig. 3.4) and punching shears
48
obtained from flexural shear and mat displacement are found to be 665.60 k and
671.50 k respectively. These two values are in excellent agreement and this
agreement serves as another evidence of the superiority of the present element.
Values of punching shears around columns C I, C2, C3 and C4 (Fig. 3.4) are
presented in Table 4.3. It may be noted that punching shears around columns C2 and
C3 are the same due to the symmetry of the problem and hence are not given
separately.
Table 4.3. Punching shears of the example mat (Fig. 3.4) by different methods.
Table 4.3 shows that conventional method is excellent in predicting punching shear
despite its simplified assumptions whereas the. lengthy solution process of ACI
method gives reasonable punching shear values w.r.t. FE method only around the
central columns of the mat. Since punching shear is expected to be the governing
criterion for mat thickness design, this discrepancy of ACI method may result in
inaccurate design if heavy loads occur near the edges of mat. However, this trend in
ACI prediction of punching shear is quite understandable from the pattern of
deflections as obtained by the same method (Fig. 3.12), which shows that for this
method, mat deflection is much higher near the edges than near the center. As soil is
modeled as spring, this deflection pattern gives rise to higher total soil pressure under
the punching area if a column occurs near the edges of mat. Since this soil reaction is
deducted from the column axial force to obtain punching shear, overestimation of
49
deflection results in underestimation of punching shear near the edges of mat by ACI
method.
Thicknesses required for punching shear by Conventional, ACI and FE methods are
23.75 inches, 24.06 inches and 23.74 inches respectively. From practical
considerations, 24.00 inches thickness is adequate for all the methods to encounter
punching shears. This may seem a bit queer, but as has been pointed out in the
previous paragraph, this happens in this particular example, and in all mats with
heavy central column loads, because all the methods are almost equally good in
predicting punching shear near the center of the mat. However, thickness
requirement from the corner column punching shear (this is not a governing design
criterion for the current example but serves the purpose of showing the difference
between various method) are 11.88 inches by Conventional method, 10.49 inches by
ACI method and 11.85 inches by FE method. Underestimation of this required
thickness by ACI method is 11.5% for this example. This implies that when heavily
loaded corner columns are present, punching shears of the corner columns may
govern the design and this amount of underestimation by ACI method will be a
matter of concern then.
Flexural shear to be considered is the shear at a distance d (2.71 ft for the current
example) from the face of column C4 (Fig. 3.4). The most highly stressed strip of
unit width is considered for this purpose. Shear force diagrams for LINE I and 2
(Fig. 3.4) have already been presented in the previous chapter (Fig. 3.14). Flexural
shears per unit width at that location as found by Conventional, ACI and FE methods
are 11.40 klft, 37.30 klft and 29.10 klft respectively. Thickness required for these
shears are 12.34 inches, 32.41 inches and 26.05 inches respectively. It is interesting
to note that if flexural shear of the most highly stressed strip of unit width is
considered, both ACI and FE methods require greater thickness for flexural shear
50
than for punching shear. If this criteria is to govern, then a second trial thickness
should be chosen and verified. Now, it has been revealed by parametric study in the
previous chapter (Article 3.7.3 and Fig. 3.23) that shear is much insensitive to mat
thickness in the central region of mat, which happens to be the critical location for
shear calculation for the present problem. So the shear force diagrams for the 3 ft
thick mat (Fig. 3.14) may be used to select a new thickness. For a thickness equal to
33 inches or 2.75 ft, d is equal to 2.46 ft and from Fig. 3.14, design flexural shear for
2.75 ft thick mat is approximately estimated to be 11.80 klft, 39.90 klft and 32.80
klft for Conventional, ACI and FE methods respectively. Table 4.1 shows that for
ACI method 2.75 ft thickness does not satisfY flexural shear requirement. Table 4.1
also shows that flexural shear capacity of a 2.75 ft thick mat is in excess of the acting
FE flexural shear (32.80 klft), indicating that further reduction of mat thickness may
be possible for FE analysis. However, if a third trial of 2.50 ft thick mat is made for
FE method, approximate FE design shear, as estimated from Fig. 3.14, becomes
36.80 klft, which exceeds the flexural shear capacity of 2.50 ft thick mat (Table 4.1).
As a result, a mat thickness equal to 2.75 ft seems acceptable for FE method. An
analysis of the example mat (Fig. 3.4) with 2.75 ft thickness gives the shear force
diagrams of LINE 1 and 2 (Fig. 3.4) as presented in Fig. 4.3. Conventional method
does not take into account the effect of mat thickness, so the same solution remains
valid no mater what the mat thickness is. From Fig. 4.3 it is found that design flexural
shear for 2.75 ft mat thickness by ACI method is 40.00 klft and that by FE method is
32.80 klft, both of which are in excellent agreement with the estimated values.
The most economic solution comes from the Conventional method despite its crude
assumptions. FE method offers 8% economy compared to ACI method. Change in
mat thickness as necessary for FE method brings back the question of limiting
51
-0.02
-0.03
~
S
c
0
""Ql
(j
0;::
Ql -0.04
0
o 10 20 30
X (ft)
(a)
-0.01
-0.02
g
c
-0.03
o
U
Ql
Qj -0.04
o
-0.05
-- FinITeElement solution
-- ACI approximate flexible method
-0.06
o 10 20 30
X (ft)
(b)
Fig. 4.1. Comparison of deflection calculation of (a) LINE 1and (2) LINE 2
(Fig. 3.4) by different methods (Mat thickness 2.75 ft).
\/
150
50
2
<=,
6
C
Q)
0
E
0
::;;
-50
100
- Finite Element solution
75 - ACI approximateflexible method
-- Conventional method
50
25
~
-•...
:~
>2 0
'"
Q)
.<::
if) -25
-50
-75 CF
CF
-100
0 5 10 15 20 25 30
X (ft)
(a)
-100
CF
-150 CF
0 5 10 15 20 25 30
X (ft)
(b)
Fig. 4.3. Comparison of shear force calculation of (a) LINE 1 and (b) LINE 2
(Fig. 3.4) by different methods ( Mat thickness 2.75 ft, CF = column
face).
displacements and a check on it is made by using Fig. 4.1 ( deflection diagrams of
LINE I and 2 for 2.75 ft thick mat) and found to be satisfactory.
However, before drawing any final conclusion, the criterion of flexural shear requires
to be investigated with special attention and from more realistic point of view. This
will be done in the article next to the following.
Column face positive and in-between column negative bending moments by all of the
three methods considered are obtained from Fig. 3.13 and found to be less than the
moment capacity of the mat with 3 ft thickness when reinforced with minimum steel
(Table 4.1). Even change of mat thickness from 3.00 ft to 2.75 ft does not change
this feature. The implication is that for the entire mat, a total of four layers of steel (at
top and bottom and in two perpendicular directions), each with a steel area of 1.296
2
in /ft for 3.00 ft thickness and 1.181 in2/ft for 2.75 ft thickness, are required (Table
4.1). However, if mat thickness is lowered to 2.00 ft, as per the requirement by
Conventional method, negative moments of LINE I and LINE 2 (79.30 and 81.50 k- r .
ft/ft respectively) become higher than the capacity with minimum steel i.e. 73.40 k-
ft/ft (Table 4.1). Steel requirements for these moments are 0.888 in2/ft and 0.914
2
in /ft (at the bottom and in-between the columns) respectively and elsewhere steel
2
requirement is 0.821 in /ft, the minimum steel required for 2.00 ft thickness (Table
4.1).
Once again, Conventional method comes out to be the least costly solution. Since
minimum steel requirement governs in case of ACI and FE solutions, comparative
economy of these two methods is almost directly proportional to the thickness
requirement and FE method outweighs ACI method in terms of economy in the cost
of steel by 10%.
52
4.4.6 A Different Interpretation of Flexural Shear
Conventional method estimates quite low values of flexural shears presumably due to
the averaging of shear force over the entire widths of column strips. To examine the
validity of this reasoning, flexural shear averaging over the entire widths of column
strip 1 and 2 is done for the 3.00 ft thick mat (Fig. 3.4) and the result is shown in Fig.
4.4. It is evident that once this averaging is done, results by all the three methods
move towards closer agreement. In fact, estimation by FE method remarkably
matches that by Conventional method. Fig. 4.4 also shows that the difference
between FE method and ACI method in calculating average column strip shear
diminishes away from the column faces.
If these average column strip shears are considered in stead of maximum unit-width
shears; for all the methods, except ACI method, punching shear governs the thickness
design. Design flexural shear for Conventional method is 13.10 klft and those by ACI
and FE method become 27.10 klft and 14.50 klft respectively. Required thickness by
ACI method is still governed by flexural shear and is 24.50 inches. It is important to
note that although ACI method gives average strip shear slightly above the capacity
of the thickness required form punching shear, resulting in only 2% increase in the
cost of concrete for this particular example, this trend of ACI method will surely
demand greater thickness for heavier column loads and lead to an uneconomic
solution.
In this context it may be argued that it is unlikely that the most highly stressed strip of
unit width will act individually to tackle the flexural shear. As a matter of fact, for the
3.00 ft thick mat (Fig. 3.4), a plot of lateral distribution of flexural shear along the
column faces (Fig. 4.5 and Fig. 4.6) shows that distribution of shear is remarkably
uneven for mat. High concentration of flexural shear occurs near the columns and the
rate of decay of shear force magnitude away from the columns is very high. It is
expected that a redistribution of shear will occur when material will be stressed
beyond the elastic limit. Even within the elastic limit, a wide portion of any column
53
- Conventional Method
30
- ACI Approximate Flexible Method
---A- Finite Element Method
20
CF
10
-6
~
~ 0
ro
Cl
.c:
(/J -10
-20
CF
-30
CF
0 10 20 30
X (ft)
(a)
40 - Conventional Method CF
- ACI Approximate Flexible Method
---A- Finite Element Method
20
CF
~
'~
:;2
~ 0
ro
Cl
.c:
(/J
-20
CF
-40 CF
o 10 20 30
X (ft)
(b)
Fig. 4.4. Comparison of average strip shear calculation of (a) column strip 1
and (b) column strip 2 by different methods ( Mat thickness 3 ft, CF =
column face).
100
- Left face of columns C1 and C3
80
- Right face of columns C1 and C3
-A- Left face of columns C2 and C4
60
---T- Right face of columns C2 and C4
40
~ 20
~
~ 0
~x
Ol
Cll -20
.<::
(/)
-40
-60
-80
-100
0 5 10 15 20 25 30
Y (ft)
Fig. 4.5. Lateral distribution of shear as calculated by Finite Element
method (Mat thickness 3 ft).
150
- Left face of columns C1 and C3
- Right face of columns C1 and C3
100 -A- Left face of columns C2 and C4
---T- Right face of columns C2 and C4
50
~
~
~ 0
~
~x
Ol
Cll
.<:: -50
(/)
-100
-150
0 5 10 15 20 25 30
Y (ft)
If average column strip flexural shears are taken to be the design criteria, the same
reasoning applies to bending moments. Fig. 4.11 shows the average column strip
54
45
40
~ 35
~
~
Cll
Q)
30
.c
<Jl
Q)
0> 25
~
Q)
>
« 20
8.75ft
15
o 5 10 15 20
Averaging widht (ft)
Fig. 4.7. Relationship between average flexural shear and averaging width (Mat
thickness 2 ft, column spacing 20 ft).
30
2
32
~
~x
Cll 20
Q)
.c
(f)
10
o
10 15 20 25 30
Y (ft)
Fig. 4.8. Averaging widht for which average flexural shear does not exceed mat
capacity (Mat thickness 2 ft, column spacing 20 ft).
50
~ 40
~
III
Q)
.r::
en
:g, 30
e
~
«
20 9ft
o 5 10 15 20 25
Averaging widht (ft)
Fig. 4.9. Relationship between average flexural shear and averaging widht (Mat
thickness 2.5 ft, column spacing 25 ft).
50
40
£ 34.2 klft
:>2
~ 30
~x
III
Q)
.r::
en 20
10
9ft
o
o 10 20 30 40
Y (ft)
Fig. 4.10. Averaging widht for which average flexural shear does not exceed mat
capacity (Mat thickness 2.5 ft, column spacing 25 ft).
40 - Conventional Method CF
CF
- ACI Approximate Flexible Method
----.t.-- Finite Element Method
20
~ 0
--
0:=
:;:,
-"~c -20
Q)
E
0 -40
:2
-60
-80
0 10 20 30
X (ft)
(a)
40 - Conventional Method CF CF
- ACI Approximate Flexible Method
20 ----.t.-- Finite Element Method
CF
CF
0
-
~
~ -20
-
6
c
Q)
E
-40
0
:2 -60
-80
-100
0 . 10 20 30
X (ft)
(b)
Fig. 4.11. Comparison of average strip bending moment calculation of (a) column
strip 1 and (b) column strip 2 by different methods (Mat thickness 3 ft,
CF = column face).
.b"''''''g mom", ill""""" 11" "rip I ~d 'rip 2 of 'h" m""P', mOlby dl"",,,,
""'ood, M"",Iud~ of mom"", d"""" OO"&d"""ydu, 10"'" "_"S
'" Ih, fu"''''''g -"", "'" d,'", II fu" """ fOUodtim, wl"" 10m!_bm,,;oo
do~ "'" gwom "" "'"go fm ili, ""'1_ "'" 00""'''01 (FIg 3.4) Fm ili,
g"".;ly ,_ OO".d<red,"'hmm b"" "'O"'ffil, h,", """ 1""""" do< '0 ~,
limit,,;"" of C~ffill_ ~d ACI ""'hod, 10",""", lhom. Co",,", """ mom"",
du, '0 -~ 10"", " ""'Y ""'" "'" '"""" ""' 1"",,_ "'" """'YJI, """n 10
l1oi" O"'ffil ~."'" ~~, omm"",,~ '" I"ffi ;"10"ooom. To d''''''''''''''h,
"" "'" fu, g"",1y 10'<1,001""" "", mo"""" """ ",gllwbl, <If",. """" of"",
-"", of "" m""'pl, ""I (FIg 3A) wl~ ~d with""l 001_ b"" mo"""" '"
presented in Fig. 4.12 through Fig. 4.14. It is clear from this preSentation that there is
""--"'" dtif_" I" "" d"'","o" b"'dlog "'0_ ~d"'= """ ill,,,,,,,,,,
b"_,, 1'-' Iwo "", ThI,""'" u , pmof"," 'h,""'"'d"igo I, ""'''''',
55
.bending moment diagrams for strip I and strip 2 of the example mat by different
methods. Magnitudes of moments decrease considerably due to the averaging.
When averagmg is done across the widths of colullIn strips, mat thickness is
governed by punching shear in Conventional and FE method. For the ACI method,
flexural shear still may control the design in this regard. However, in any method mat
thickness is reduced and hence the bending moment capacity. For the present
example (Fig. 3.4) 2 ft thickness seems adequate under the changed course of design
(24.50 inches for ACI analysis) and'Table 4.1 shows that average colullIn strip
. moments still stay below the minimum-steel moment capacity of the mat for FE and
ACI methods. The situation is different for Conventional method and has already
been discussed in Article 4.4.5.
In the forgoing analysis and design, it has been found that wind load combination
does not govern the design for the particular mat considered (Fig. 3.4). For the
gravity loads considered, colullIn base moments have been ignored due to the
limitation of Conventional and ACI methods to handle them. ColullIn base moments
due to gravity loads are very small and should not influence the analysis much. In
finite element analysis, these moments can be taken into account. To demonstrate the
fact that for gravity loads colullIn base moments have negligible effect, results of the
analysis of the example mat (Fig. 3.4) with and without colullIn base moments are
presented in Fig. 4.12 through Fig. 4.14. It is clear from this presentation that there is
no significant difference in the deflection, bending moment and shear force diagrams
between these two cases. This serves as a proof that the above design is acceptable.
55
-0.020
-0.022
-0.024
g
c -0.026
0
U
Q)
<;::
Q) -0.028
0
-0.030
-0.032
- Column axial loads with column base moments
- Column axial loads only
-0.034
0 10 20 30
X (ft)
(a)
-0.014
-0.016
-0.018
g
c -0.020
0
U
Q)
<;:: -0.022
Q)
0
-0.024
-0.026
- Column axial loads WIThcolumn base moments
- Column axial loads only
-0.028
0 10 20 30
X (ft)
(b)
Fig. 4.12. Effect of column base moment on deflection diagrams of (a) LINE 1
and (b) LINE 2 (Fig. 3.4, mat thickness 3 ft).
-- Column axial loads with column base moments
40 _ Column axial loads only CF
20
~ o
.,,:,
~
-
c:
~ -20
o
:2:
-40
-60
o 5 10 15 20 25 30
X (ft)
(a)
20
CF
2
-
:;:,
•
-"~
0
c
Q)
E -20
0
:2:
-40
-60
o 5 10 15 20 25 30
X (ft)
(b)
When average column strip bending moments and shear forces are used, Finite
Element analysis gives the most economic solution. It saves concrete volume
compared to ACI method and cuts the cost of steel in comparison to both
Conventional and ACI methods. If the thickness required by ACI method is much
higher than that by FE method, it is likely that minimum flexural reinforcement
criteria will govern reinforcement design, which in tum implies that FE method will
result in substantial economy in terms of concrete and reinforcement costs compared
to ACI method in that case. This implies that although economy achieved by FE
method is not significant for relatively light column loads, it becomes substantial as
column loads increase. To demonstrate this fact, a mat similar to the example mat
discussed previously (Fig. 3.4) but having column dimensions 24"x24", thickness 2.5
ft and CI, C2, C3 and C4 column (Fig. 3.4) loads 460 k, 735 k, 735 k and 1176 k
respectively (this problem will subsequently be referred to as test mat and has been
devised from a typical 12 story building) is analyzed and designed by the three
methods. Displacement, bending moment and shear force diagrams of column strip 1
and 2 of this mat is shown in Fig. 4.15 through Fig. 4.17. Design in the first trial is
summarized in Table 4.4. Subsequent trials, whether necessary or not, are avoided
since a single trial is sufficient to serve the purpose of demonstration. In calculating
total steel requirement a special term, weighted average steel area per unit width
(A:vg), is used. This is calculated using Equation 4.24 and serves as an index of the
(4.24)
where
= calculated steel area per unit width,
= plan area of mat covered by A. .
56
-0.03
-0.04
-0.05
-
~
~
c -0.06
0
t5Q)
0:: -0.07
Q)
0
-0.08
-0.09
- Finite Element method
-0.10 - ACI approximate flexible method
0 10 20 30
X (tt)
(a)
-0.02
-0.04
2
~
c
0
:;::;
u -0.06
Q)
0::
Q)
0
-0.08
Fig. 4.15. Comparison of deflection calculation of (a) LINE 1 and (b) LINE 2 ( Fig. 3.4)
by different methods for a 12 story building (Test mat, mat thickness 2.5 tt).
Fig. 4.16. Comparison of moment calculation of (a) column strip 1 and (b) column
strip 2 by different methods for a 12 story buliding (Test mat, mat thick-
ness 2.5 ft, CF = column face).
- Conventional method CF
40 - ACI approximateflexible method
---A- Finite Eiement method
CF
20
E"
6 •... 0
'.<::"
Q)
(/)
-20
-40
CF
0 10 20 30
X (ft)
(a)
60 - Conventional method
- ACI approximateflexible method
---A- Finite Element method
40
CF
20
Z'
~
~
•... 0
'"
Q)
.<::
(/)
-20
-40
CF
-60
CF
0 10 20 30
X (ft)
(b)
Fig. 4.17. Comparison of shear force calculation of (a) column strip 1and (b) column
strip 2 by different methods for a 12 story building ( Test mat, mat thick-
ness 2.5 ft, CF = column face).
Table 4.4. Design values and their comparison for different methods for the test mat.
It is evident from Table 4.4 that higher cost associated with Conventiot)al method is
due to the over estimation of negative moments and that of ACI method arises from
the overestimation of flexural shears. Conventional method seems to have faced no
serious problem so far despite its limitation in predicting column face positive
moment since positive moment is not a likely problem for mat and is expected to be
below the minimum-steel moment capacity. ACI approximate flexible method, in
spite of its rigorous nature of analysis, gives overdesign compared to FE method,
particularly for heavily loaded mats. It appears that it is always better and more
economic to use FE method. As the superiority of FE method is established, the
following discussion will be based on finite element analysis only.
57
4.5 MAT THICKNESS REQUIREMENT A MORE ECONOMIC
APPROACH
It has already been established that flexural shears are not likely to govern the design
of mat thickness. Moreover, flexural shear decreases rapidly, away from the column
face and hence becomes less significant in regions not in the vicinity of columns. It is
therefore possible to reduce mat thickness in those areas. However, the problem with
this plan seems to be the negative moments, which occur in-between columns in a
column strip where this thickness reduction is to be implemented. now, it has been
shown earlier (Fig. 3.22) that negative moments decrease with the reduction of mat
thickness whereas positive moments increase. Consequently variation of thickness is
expected to result in reduction in negative moments and increase of positive
moments. The first trend gives a positive indication in favor of the thickness
reduction scheme and the second one is not of much concern since it has already been
observed that positive moment is not an important design criterion. Also, variation of
thickness will result in greater differential settlement. But this should not be a matter
of concern since mat is allowed to be subject t? higher differential settlement.
To examine the effect of non-uniform mat thickness, the previous mat of Fig. 3.4 is
modified to design two new problems with non-uniform thickness. Greater thickness
is maintained around column peripheries. Typical cross-sections along LINE I and 2
(Fig. 3.4) of mats with uniform and non-uniform thickness are shown in Fig. 4.18.
Every other aspect of the previous mat is kept the same. Results of the analysis are
given in Fig. 4.19 through Fig. 4.21. It is important to note that when checking
flexural shear, in addition to the usual locations (d distance away from column faces)
sections A, B and C, which will be referred to as neck sections later on, must be given
due attention for mats with non-uniform thickness.
Important design data are presented in Table 4.6 through Table 4.9. Based on these
data, MAT I, II and III are designed separately. In the comparison of the design of
58
MAT I, II and III, a special term, weighted average thickness of mat (1"-,,,,),is used
which is calculated as follows:
(4.25)
,
where
t = mat thickness,
At = plan area of mat having thickness t.
2 ft thick throughout
,
,,
~CL
'1
20' 10'
MAT I
12.5'
•
'J
1.5'
'1
~
~. ~' ~.
'CL
Fig. 4.18. Mats with uniform and varying thickness (CL = center line).
59
-A-- MATI
-0.01
-MATII
-- MATIII
-0.02
~
;::
~ -0.03
o
UQ)
"" -0.04
~
-0.05
-0.06
0 10 20 30
X (tt) .1"').
, )
(a) \,,'
-0.02
2
.,
~
c:
0
u -0.03
Q)
""0
Q)
-0.04
-0.05
o 10 20 30
X (tt)
(b)
Fig. 4.19. Comparison of deflection diagrams of (a) LINE 1 and (b) LINE 2 (Fig. 3.4)
for MATI, MATil and MAT "' (Fig. 4.18).
60
---4-- MATI
CF CF
-MATII
40
- MATIII
20
-a=
Z'
0
-
6
c
Q)
E -20
a
:2
-40
-60
0 5 10 15 20 25 30
X (ft)
(a)
60
---4-- MATI
-MATII CF CF
40
- MATIII
20
2
a=, CF CF
6 0
'E
Q)
E
a -20
:2
-40
-60
0 5 10 15 20 25 30
X (ft)
(b)
Fig. 4.20. Comparison of bending moment diagrams of (a) column strip 1 and
(b) column strip 2 for MAT I, MAT II and MAT III (Fig. 4.18, CF =
column face).
CF
-20
o 5 10 15 20 25 30
X (tl)
(b)
Fig. 4.21. Comparison of shear force diagrams of (a) column strip 1and (b) column
strip 2 for MAT I, MAT II and MAT III (Fig. 4.18, CF = column face).
/'
U
Table 4.5. Column displacements of MAT I, II and III (Fig. 4.18).
Table 4.7. Column face positive moments of MAT I, II and III (Fig. 4.18).
Table 4.8. Maximum column strip negative moments of MAT I, II and III (Fig.
4.18).
60
Table 4.9. Critical flexural shears of MAT I, II and III (Fig. 4.18).
(Vd (VB)
I 16.0 k/ft 12.2 k/ft
II 14.8 k/ft 12.3 k/ft
As expected, decrease of mat thickness away from the columns gives greater column
deflections as wen as higher differential settlements. But this does not create any
problem since mat is capable of tolerating higher differential settlement. As a mater of
fact, Table 4.5 shows that column deflections found by FE analysis are below the
allowable limits by a substantial margin for all of the three mats considered even
though weak clayey soil has been used in an of these examples. It can be concluded
confidently that differential settlement will seldom cause any problem when thickness
is non-uniform.
Punching shear does not change significantly (Table 4.6) due to the reduction of mat
thickness away from the columns. Due to the increase in column deflections, a
decrease in punching shear is observed. However, variation of punching shear is not
practically significant, especiany near the central portions of the mats. This implies
that thickness requirement in the vicinity of columns win not change significantly and
the same thicknesses at those locations, as obtained from uniform thickness solution,
can be maintained without further revision.
Flexural shears change mainly near the columns. At the neck sections A, B or C (Fig.
4.18 and Table 4.9), there is practically no change in flexural shear which means that
the magnitude of thickness reduction can be estimated from the uniform thickness
solution. This is because at the locations where change of thickness occurs, punching
shear will be of no concern for thickness design. Rather flexural shear win control
61
thickness requirement at the neck sections. And as neck flexural shear can be
estimated from the uniform thickness solution, thickness requirement at the neck
sections can be calculated directly. So the magnitude of thickness reduction may be
estimated. However, in regions where transition from higher to lower thickness takes
place (these regions will be referred to as transition zones subsequently), shear
magnification is observed in the central regions of the column strips (Fig. 4.21). This
phenomenon should be given due attention since this magnification occurs in contrast
to the flexural shear capacity reduction of mat associated with thickness reduction at
those locations.
~
t;:, and Equation 4.2) and when thickness, hence effective depth, is small, required steel
::J- area will be small too.
€I
()-.....
Increase in positive moment due to the reduction of mat thickness away from the
columns is also significant (Fig. 4.20 and Table 4.7) and is found to be upto 31 %.
However, thickness is high at the locations where positive moments occur. As a
result, positive moments remain below the minimum moment capacity despite
amplification. Evidently, making thickness non-uniform allows a more efficient and
economic use of material strength near the columns since minimum steel must always
be provided there.
62
Table 4.10. Design values of MAT I, II and III (Fig. 4.18).
Table 4.11. Economic evaluation and comparison of the design of MAT I and II.
Table 4.10 shows that MAT III is not an acceptable solution. This is because
thickness required for flexural shear at section B (Fig. 4.18) is higher than thickness
provided at that location. With the help of Equation 4.23 it can also be calculated that
13.00 inches thickness is sufficient for section B of MAT III. But this will lead to
higher negative reinforcement requirement which can be understood from Table 4.1.
However, this inadequacy of MAT III implies that because of the current absence of
any design guideline for mat with non-uniform thickness, the arduous trial and error
63
design process may. has to be iterated several times. So the necessity of the
development of such a guideline and its utility value become apparent.
MAT II comes out to be an acceptable design for the time being, in terms of both
concrete and steel requirements. A comparative evaluation of the design of MATI
and II is made in Table 4.11.
In finding the steel area, calculated reinforcement requirements are used. It must be
remembered that while placing reinforcing bars, exact steel ratio as per calculation
may not be maintained. When it comes to the choice of bar sizes to provide
reinforcements, engineers invariably differ in decision since there are many different
paths to the same destination. As a result, the deviation from theoretical requirements
in placing steel is not easy to predict. As the concern here is to perform a relative
economic analysis this particular problem can be avoided for the sake of simplicity.
, In calculating reinforcement for column face positive moments for MAT II, it is
found that minimum reinforcement for 2.00 ft thickness is sufficient. But these
moments are found by taking average across the widths of the column strips and
thickness is not the same across the width of any column strip near the column faces.
2
If 0.821 in /ft steel, the minimum for 2.00 ft thickness (Table 4.1), is provided over
the entire widths of the column strips than total moment capacities of column strip 1
and 2, at the faces of columns C2 and C4 (Fig. 3.4) respectively, become 663.50 k-ft
and 1031.20 k-ft respectively. Actual total moment acting at these locations are
563.40 k-ft and 900.60 k-ft respectively, which are less than the respective capacities.
Since it is being assumed that individual column strips act as a whole, provision of
2
0.821 in /ft steel is adequate for MAT II. Flexural shear capacities of strip I and strip
2 of MAT II, at a distance d (1. 71 ft for 2.00 ft thickness) away from the columns C2
and C4 (Fig. 3.4) respectively, are 241.30 k and 376.10 k respectively against the
acting total flexural shear of 183.90 k and 299.70 k respectively. Although
considerable flexural shear magnification has occurred in the central transition zones
of MAT II, average mat thickness in those zones (17.50 inches for column strip I
64
"1,
and 16.90 inches for column strip 2) are found to be adequate. For strip I and 2,
acting maximum shears in the transition zones are 15.40 k/ft and 15.50 k/ft
respectively whereas average flexural shear capacities of these strips are 18.10 k/ft
and 17.30 k/ft respectively in the transition zones. It is clear that despite variation of
mat thickness, shear capacity is well above the requirement for MAT II.
In the previous article, mat with different thicknesses at different locations has been
found to be a better design than mat with uniform thickness. The saving, which is a
function of the greater thickness tg and the smaller thickness t, (Fig. 4.22), is
substantial, both in the cost of concrete and in the cost of reinforcement (Table 4.11).
Difficulty in reinforcement placement is a consequent disadvantage of this proposition
which will increase construction expenditure. But even then, overall cost will surely . ,
decrease because of the considerable saving attained in material cost. This is
especially true for regions like Bangladesh, where labor is available at a very cheap
rate. However, for the non-uniform thickness solution, some critically important ,
factors must be given due attention. These, for instance, are : the distance over which
the change in thickness should be implemented (d, ), the lateral extent of greater
thickness around the columns (dg ) and the amount of reduction in mat thickness
away from the columns (tJ.t ) (Fig. 4.22). Of the above items, tJ.t can be realized in
terms of tg and t, . For the convenience of further discussion these items are named as
.".
follows:
d, = Slope width,
dg = Width of greater thickness,
tg = Greater thickness,
t, = Smaller thickness,
tJ.t Change in thickness.
65
Slope angle
Transition zone
Neck section
Section A
..",
Tilt
.,j'--,j' ~
It
Ai
d, dg
(a)
Region of
greater
thickness
Region of
smaller
thickness
.
t
(b)
66
i
4.6.1 Effect of Slope Width (d, )
To examine the effect of slope width, MAT II of Fig. 4.18 is analyzed by varying its
d, from 0.75 ft to 2.00 ft. Variation of column displacements, column face positive
bending moments, maximum negative bending moments, column face flexural shears
and kink shears (those in the transition zones) are shown in Fig. 4.23 through Fig.
4.26. Variation of punching shear is listed in Table 4.12. An inspection of these plots
and Table 4.12 reveals that punching shears, deflections and bending moments are
not significantly affected by slope width. For example, a survey of the most
significantly affected items shows that over the entire range of slope width
considered, column C4 (Fig. 3.4) deflection changes by only +5.0% and column face
positive moment at the same location undergoes a change of +6.7%. The most
noticeably affected negative moment is that at the center of strip 2 which changes
from 16.90 k-ft to 14.50 k-ft, a decrease of 14.3%. Although this particular negative
moment shows significant sensitivity to slope width, its magnitude indicates that this
is not high enough for being considered to be critical to design.
Table 4.12. Effect of slope width on punching shear (MAT II of Fig. 4.18).
Effect of slope width on flexural shear is noticeable. Fig 4.25 shows that high shear
concentration occurs in th~ transition zones (Fig. 4.22) near the center of the column
67
0.038
• • • • • • •
0.036 - Column 1
- Column 2
---A- Column 3
~ Column 4
0.034
0.032
2
~
c 0.030
0
:;:;
<J
OJ • • •
"""0
OJ
• • • •
c 0.028 ••• •••
E
••• ••• •••
:J ••• •••
0
U
0.026
0.024
0.022
0.020
------.,. 't'
't' ..,. ..,. ..,.
0.4 0.8 1.2 1.6 2.0
Slope width (ft)
Fig. 4.23. Effect of slope width on column deflection (MAT II of Fig. 4.18).
50
•••• •••
•• •
••• •••
• •••
• •••• :
40
--- Column 1 of strip 1
-
2
:;:,
,
-'"
~
C 30
---
-A-
~
Column 2 of strip 1
Column 1 of strip 2
Column 2 of strip 2
Ql
E
0
E
-
2J
<1l
c
E
~
20
810
0
" ill ill ill ill ill
•
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
Slope widht (ft)
(a)
40
• • • • • •
~ 35 •
~,
~
-
-'"
c
Ql
E
30
••• ••• ••• ••• ••• ••• •••
E
~ 20 - • • •
E
.x
<1l
:2 ~
••• •••
• • -.
15 ••• ••• •••
."
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
Slope width (ft)
(b)
Fig. 4.24.
Effect of slope width on (a) positive and (b) negative bending
moment (MAT" of Fig. 4.18).
25
---- Slope width = 1.50 It
20 -- Slope width = 1.25 It
----6- Slope width = 1.00 It
15
----....- Slope width = 0.75 It
---+- Slope width = 0.50 It
10
CF
5
~ 0
~
III
(])
.r::: -5
en
-10
CF
-15
-20
-25
0 5 10 15 20 25 30
X (ft)
(a)
~ 0
~
III
(])
.r::: -5
en
-10
CF
-15
-20
0 5 10 15 20 25 30
X (ft)
(b)
Fig. 4.25. Shear force diagrams of (a) column strip 1and (b) column strip 2
for different slope widths (MAT II of Fig. 4.18, CF = column face).
22
-- Around column 1
20
- Around column 2
---y- Around column 3
18
- Around column 4
16
Fig. 4.26. Effect of slope widht on kink shear as a function of (a) slope width
and (b) slope angle (MAT II of Fig. 4.18).
strips and local shear failure is likely to occur if proper attention is not given to this
matter. On the contrary, near the ends of the column strips, flexural shears decrease
in the transition zones. A plot of the ratio of the magnified shear to the shear at the
corresponding points in case of uniform thickness, against slope width (Fig. 4.26)
shows that the ratio, which will be designated as kink shear magnification later on,
tends to unity as slope width is increased. When considered in terms of the angle of
the slope (Fig. 4.26), the ratio comes close to unity as the slope flattens out.
Final selection of the slope depends on the optimization between the flexural shear
capacity of the transition zone and the degree of kink shear magnification. But sharp
change in thickness always creates bad locations from structural point of view and
should be avoided when possible.
For the same problem (MAT II of Fig. 4.18) change in thickness (8t) is varied from
0.00 to 1.00 ft, keeping the greater thickness tg (Fig. 4.22) fixed at 2 ft. Variation of
different items thus observed is depicted in Fig. 4.27 through Fig. 4.30. It is evident
that the response of mat with non-uniform thickness is higWy sensitive to this
particular parameter. This is not new in light of the parametric study made in the
previous chapter where mat with uniform thickness has shown the greatest sensitivity
to thickness (Article 3.7.3).
Changes suffered by different items due to the change in LIt from 0.0 ft to 1.00 ft is
compiled in Table 4.13. It is worth mentioning here that punching shear is the least
affected item of all. As a matter of fact, punching shears for central columns show no
noticeable change as 8t is varied.
Reduction of negative moments (Fig. 4.28), which occur in regions away from the
columns, is a good indication in favor of thickness reduction scheme, since thickness
68
0.045
- Column 1
- Column 2
--A-- Column 3
---y- Column 4
0.040
0.035
g
c
0
:;::;
u
Q)
0::: 0.030
Q)
"lJ
C
E
:J
(5
()
0.025
0.020
0.015
0.0 0.2 0.4 0.6 0.8 1.0
M (tt)
2
40
• - : :
-
:;:,
,
~ -
-
Column 1 of strip 1
Column 2 of strip 1
-
-'"
~
c 30
Q)
E
-A-
---T-
Column 3 of strip 2
Column 4 of strip 2
a
E
-
Q)
u
ro 20
c
E
::J
0 10
U
0
• • • I •
0.0 0.2 0.4 0.6 0.8 1.0
At (ft)
(a)
55
- In-between column C1
=
~,
50 -
and C2 of strip 1
Center line of strip 1
In-between column C3
-
645 -A-
c and C4 of strip 2
Q)
'"
Ol
Q)
30
c
E 25
::J
E
'xro 20
:2:
15
10
0.0 0.2 0.4 0.6 0.8 1.0
At (ft)
(b)
Fig. 4.28. Efect of change of thickness, At on (a) positive and (b) negative
bending moment (MAT II of Fig. 4.18).
17
16
15
2
:><
::-14
(tJ
I]) - Column 1
.s=
en 13 - Column 2
I])
u ~ Column 3
-
(tJ
<::
E
12 -- Column 4
OJ
0
U 11
10
9
0.0 0.2 0.4 0.6 0.8 1.0
!1t(ft)
(a)
600
- Column 1
550 - Column 2 and 3
~ Column 4
g
~
500
(tJ
~ 450
en
Cl
.1; 400
.s= • • • •
u
<:: •
a. 350
OJ
300
250
• • • •
200 •
0.0 0.2 0.4 0.6 0.8 1.0
!1t (ft)
(b)
Fig. 4.29. Effect of change of thickness, !1t on (a) column face and (b) punching
shear (MAT II of Fig. 4.18).
16
~ 12
14
: • I
-
--..-
Around column 1
Around column 2
~
•... - Around column 3
l1l
QJ ---y---,- Around column 4
.J:::: 10
-'"'"c
:;;: 8
E
::l
E 6
'xl1l
::2'
4
(a)
1.25
- Around column 2
- Around column 4
1.20
c
.Q
0;
(J 1.15
u::
'c
Cl
l1l
E
•... 1.10
l1l
QJ
.J::::
-'"'"c
1.05
~
1.00
o 10 20 30 40
Angle of slope (degree)
(b)
I
j
'\.._)
is reduced at those locations. This too is justifiable, because increase in LIt means .
consequent reduction of mat thickness in the negative moment regions of mat and it
has been found earlier (Fig. 3.22) that negative moment decreases with the increase
of mat thickness. The indication is that the resulting decrease in moment capacity will
not be accompanied by any alarming problem, at least not of the degree that may
impose any serious restriction on thickness reduction. But high magnification of
column face moments gives a warning in the sense that minimum steel requirement
might no longer be satisfactory in the thicker part of the mat. However, steel ratio is
always adjustable within a wide range to match the required moment capacity.
Though individual column deflections increase in the range of 25% to 50% (Table
4.13), an inspection of this figures reveals that differential settlement increases to a
much lower extent. Besides, it has already been established that differential settlement".
is not a potential problem for mat design. ~I
Table 4.13. Variation of mat response with the change of LIt from 0.00 ft to 1.00 ft.
\
t
I
Flexural shears to be careful about are central kink shears. Kink shear increases as the
slope of the transition zones gets steeper (with greater change in thickness, slope of
the transition zones gets steeper). This is expected, as has been found in the study of
69
the effect of slope width on mat behavior (Article 4.6.1). A plot of kink shear W.r.t.
to the zero slope an[le (Fig. 4.22) solution verses slope angle (Fig. 4.30) shows as
before that kink shear magnification becomes less prominent as the slope flattens out.
Five different problems, each derived from MAT II (Fig. 4.18) by varying only d in
g
the range of 15.00 inches to 27.00 inches, are analyzed. Final results are graphically
compared in an effort to investigate the effect of dg on mat behavior. The graphs are
shown in Fig. 4.31 through Fig. 4.34. It must be noted here that since punching shear
is calculated by integrating Gauss point shear stresses around the columns, they are t
not comparable in this case because of the changes in locations of sampling points i
Fig. 4.31 through Fig. 4.33 show that width of greater thickness does not have
significant effect on mat response. The most noticeably affected column deflection is
that of column C4 (Fig. 4.31), which undergoes a change of -5.5% when the width of
greater thickness is changed from 15.00 inches to 27.00 inches. Column face positive
moments (Fig. 4.32) practically do not change with the width of greater thickness.
Only significant changes that occur in bending moments are those of negative
moments at the center of strip 1 and 2. At these locations, a maximum change of
-12.2% and -15.3% respectively are observed. However, usually central negative
moments are below the minimum moment capacity of the mat thickness there, which
is why their reduction carries little practical significance. Flexural shear is lowered
when width of greater thickness is increased (Fig. 4.33). With the widening of thicker
regions, transition zones move away from the faces of the columns towards lower
flexural shear regions. Consequently kink shear reduction is observed which, due to
the change in their locations, are not comparable. However, Fig. 4.33 and Fig. 4.34\
are presented to visualize this effect. Fig. 4.33 shows that Cl through C4 (Fig. 3.4)
column face shears are decreased by 8.1%, 3.5%, 8.2% and 3.5% respectively. But.
70
0.040,--- -,
~ Column 1
- Column 2
0.035 - - Column 3
-- Column 4
2
~
c 0.030 -
0
:;::;
u
.,. .,. .,.
Ql
0::
.,. .,.
Ql
-0 •• •• ••
c
E
•• ••
::J
(5
0
0.025 -
0.020 - • • • ••
I I I • I • I I •
1.2 1.4 1.6 1.8 2.0 2.2 .2.4
Width of greater thickness (ft)
16
14
16
: : •
:
~
~ 14
15
:
~ - Column 1
'"
Ql
J::: - Column 2
<Jl
~
.J!l
. 13 ----.to-
-y-
Column 3
Column 4
c
E 12
::J
.0
()
11
10
1.2 1.4 1.6 1.8 2.0 2.2 2.4
Width of greater thickness (ft)
(b)
Fig. 4.33. Effect of width of greater thickness on (a) kink shear and (b) column
face shear (MAT" of Fig. 4.18).
20
10
CF
•
:2
6
.~ 0
(Il
Q)
.r:;
(f)
-10
Width of greater thickness = 1.25 It
CF - Width of greater thickness = 1.50 It
----A- Width of greater thickness = 1.75 It
-T- Width of greater thickness = 2.00 It
-20 =
o 5 10 15 20 25 30
X (ft)
(a)
20
10
CF
~
~
~ 0
(Il
Q)
.r:;
(f)
Fig. 4.34. Shear force diagrams of (a) column strip 1 and (b) column strip 2
for different widths of gretaer thickness (MAT" of Fig. 4.18, CF =
column face).
since flexural shears do not govern the thickness design near the columns, this
reduction in the magnitude of column face shear forces do not offer any additional
advantage.
With the reduction of thickness in the low shear and high negative moment regions,
column face positive moments increases. But since greater thickness is provided in
the positive moment and high shear regions, this amplification does not pose any
threat. As a matter of fact, column face positive moments, after being averaged
across column strip widths, come out to be much lower than the moment capacities
of mat at those sections (reinforced throughout with the minimum steel required for
the greater thickness).
71
Width over which change of thickness should be made .i.e. slope width is not very
critical as long as the slope of the transition zones is kept as flat as practicable.
Increase in shear within the transition zones near the center of mat emerges as a
matter demanding careful consideration. Further study will be made to examine this
particular aspect.
Width of greater thickness is not a potential problem. But since greater thickness is
required for punching shear purpose and since in the calculation of the punching
shear capacity, flexural shear cracks are assumed to propagate downward from the
column face at an angle of 45°, a width equal to the effective depth corresponding to
the greater thickness seems to be rational. Of course experimental validation of this
proposition is yet to be established.
72
LOADING I = 1.4 x Dead Load + 1.7 x Live Load (4.26)
LOADINGII = 0.75 x (1.4 x Dead Load + 1.7 x Live Load +
12.5'
C5 C6 C7 C8
0 0 0 0
y
~x
25'
Cl C2 C3 C4
0 D 0 0
3.25'
-H 'k 'k 'k4-
3.25' 25' 25' 25' 3.25'
Fig. 4.35. Plan of the mats of CASE I and II.
e 310'
--rr =rr
~,
?S
*
3.25' 25'
'k
25'
CASE I
25'
'kr
3.25'
(a)
3.75' 2' 11'
<4
~ Neck section Transition zone
~......----,;'oo, ,~~, ~~
3.25' 25' 25' 25' 3.25'
CASE II
(b)
Fig. 4.36. Cross sections of the mats of (a) CASE I (uniform thickness) and (b)
CASE II (non-uniform thickness).
73
Table 4.14. Column loads for LOADING I and II.
Column axial forces and base moments due to dead load, live load and wind load
combination i.e. LOADING IT are calculated using MICROFEAP-II (1988), not by
simplified methods in which only column axial forces are calculated by reducing
column forces due to gravity loads on the windward side and increasing those on the
leeward side to account for lateral loads. Also, column base moments are considered
in the analysis of the mats. These make loading condition of the present mats more
realistic and also complicated. It is difficult to decide in advance whether LOADING
I or II will govern the design of mat. This is because LOADING I gives higher
column axial forces and lower column base moments whereas LOADING II gives
lower axial forces and much higher base moments. So analysis is performed using
both the loading cases. Results of column strips CI-C4, C5-C8, CI-C5, C2-C6, C3-
C7 and C4-C8 (Fig. 4.35) are presented in Fig. 4.37 through Fig. 4.45. Design values
of column deflections, punching shears, flexural shears, column face positive
moments and negative moments are listed in Table 4.15 through Table 4.23. Design
thicknesses, reinforcements and relative economic evaluation of CASE I and II are
presented in Table 4.24 and Table 4.25.
74
-- CASE I and LOADING I
- CASE I and LOADING II
-----A-- CASE II and LOADING I
~ CASE II and LOADING II
20 40 60 80
X (ft)
(a)
-0.02
-0.03
c
o -0.04
fJ
Q)
ii=
Q)
o -0.05
o 20 40 60 80
X (ft)
(b)
Fig. 4.37. Deflection diagrams of X directional Column lines (a) C1-C4 and (b) C5-C8
for CASE I and CASE II (Fig. 4.36) under LOADING I and LOADING II (Eqn.
4.26 and 4.27).
-0.015
-0.020
-0.025
-0.030
is -0.035
c
0 -0.040
t5
Ql
-0.045
""Ql
0
-0.050
-0.055
- CASE I and LOADING I
-0.060 - CASE I and LOADING II
~ CASE II and LOADING I
-0.065 ----y--- CASE II and LOADING II
0 10 20 30 40
Y (ft)
(a)
-0.020
-0.025
-0.030
£
~ -0.035
0
t5
Ql -0.040
""Ql
0
-0.045
0 10 20 30 40
Y (ft)
(b)
Fig. 4.38. Deflection diagrams of Y directional Column lines (a) C1-C5 and (b) C2-C6
for CASE I and CASE II (Fig. 4.36) under LOADING I and LOADING II (Eqn.
4.26 and 4.27).
-0.020
-0.025
-0.030
~
~
~
c -0.035
0
15
Ql
0:: -0.040
Ql
0
-0.045
0 10 20 30 40
Y (ft)
(a)
-0.035
-0.040
-0.045
z:-
~
c -0.050
0
U
Ql
0::
Ql
-0.055
0
-0.060
- CASE I and LOADING I
- CASE I and LOADING II
-0.065 --A- CASE II and LOADING I
-T- CASE II and LOADING II
0 10 20 30 40
Y (ft)
(b)
Fig. 4.39. Deflection diagrams of Y directional Column lines (a) C3-C7 and (b) C4-Ca
for CASE I and CASE II (Fig. 4.36) under LOADING I and LOADING II (Eqn.
4.26 and 4.27).
150 CF CF CFCF
100
50
~ CF
0:::
~, 0
6 CF
C
Ql -50
E
0
:2:-100
0 20 40 60 80
X (ft)
(a)
CF .CF CF. CF
150
100
50
C
~
-'::0
-
~
c:
~ -50
o
:2: -100
-- CASE I and LOADING I
-150 ----- CASE I and LOADING II
-- CASE II and LOADING I
--...--- CASE II and LOADING II
-200
o 20 40 60 80
X (ft)
(b)
Fig. 4.40. Bending moment diagrams of X directional Column strips (a) C1-C4 and
(b) C5-C8 for CASE I and CASE II (Fig. 4.36) under LOADING I and LOAD-
ING II (Eqn. 4.26 and 4.27, CF = column face).
Fig. 4.41. Bending moment diagrams of Y directional Column strips (a) C 1-C5 and
(b) C2-C6 for CASE I and CASE II ( Fig. 4.36 ) under LOADING I and LOAD-
ING " (Eqn. 4.26 and 4.27, CF = column face).
..
• ,
. ,.
\
, {
.,
CF CF
150
100
50
,
6-
i2
:;:,
C
a
<ll
E
0 -50
::i'
-100
- CASE I and LOADING I
- CASE I and LOADING II
-150
-- CASE II and LOADING I
-.- CASE II and LOADING II
a 10 20 30 40
~
Y (ft)
(a)
150 CF CF
100
50
i2
~, CF (::)
a
-
-'"
~
c
<ll
E -50
r--;
~
\-/
/
0
::i'
-100
- CASE I and LOADING I
- CASE I and LOADING II
-150
-- CASE II and LOADING I
-.- CASE II and LOADING II
a 10 20 30 40
Y (ft)
(b)
Fig. 4.42. Bending moment diagrams of Y directional Column strips (a) C3-C7 and
(b) C4-C8 for CASE I and CASE II (Fig. 4.36) under LOADING I and LOAD-
ING II (Eqn. 4.26 and 4.27, CF = column face).
40 -- CASE I and LOADING I CF
-- CASE I and LOADING II
--4- CASE II and LOADING I
--- CASE II and LOADING II
20
CF
2
::i2
~
~ 0 d
.r:'"
Q)
(/)
-20
CF
-40 CF
0 10 20 30 40
Y (ft)
(a)
-- CASE I an LOADING I CF
40 -- CASE I and LOADING II
--4- CASE II and LOADING I
--- CASE II and LOADING II
20
CF
"
~
~
~ 0
'"
Q)
.r:
(/)
-20
CF
-40
CF
0 10 20 30 40
Y (ft)
(b)
Fig. 4.44. Shear force diagrams of Y directional column strips (a) C1-C5 and
(b) C2-C6 for CASE I and CASE II ( Fig. 4.36) under LOADING I and
LOADING II (Eqn. 4.26 and 4.27, CF = column face).
- CASE I and LOADING I CF
40 - CASE I and LOADING"
---4- CASE" and LOADING I
----or-- CASE" and LOADING"
20
~ CF
<:::
:>2
~
~
'"
r:
(f)
<Il
0 ••
-20
CF
-40
0 10 20 30 40
Y (tt)
(a)
~ CF
<:::
:>2
~ eM
~ 0
'"
<Il
r:
(f)
-20
CF
-40 CF
0 10 20 30 40
Y (tt)
(b)
Fig. 4.45. Shear force diagrams of Y directional column strips (a) C3-C7 and
(b) C4-C8 for CASE I and CASE II ( Fig. 4.36) under LOADING' and
LOADING II (Eqn. 4.26 and 4.27, CF = column face).
In finalizing the design, it must be noted that wind may hit the building from any
direction. This implies that the possibility of LOADING II being applied from each
side of the building, one side at a time, must be considered. Also, mat geometry and
LOADING I is symmetric. As a result, only the most severely stressed quarter of the
mat needs be designed and this can be any quarter in practice. The quarter with the
.colunms CI, C2, C5 and C6 (Fig. 4.35) is selected for the current presentation. It
should also be noticed that colunm base moments for gravity loads are very small
and, as has already been shown, their effect is negligible. As a result, LOADING I
gives virtually equal moments and shears for X and Y directional strips for any point
on any line of symmetry. However, to be more precise a comparison is made between
X and Y directional moments and shears and the greater ones are selected for design.
Following the above reasoning, design and comparison of CASE I and II (Fig. 4.36)
can be limited to any two colunm strips, for instance strip C I -C2 and C5-C6 (Fig.
4.35). Design of strip C1-C5 and C2-C6 (Fig. 4.35) will be that ofC1-C2 and C5-C6
respectively. Design of the rest of the mat involves only copying the design of this
quarter as applicable. However, colunm deflections and punching shears remain as
exceptions to this conclusion and these are considered for halves of the mats.
Table 4.15. Colunm deflections of CASE I and II (Fig. 4.36) for LOADING I and II.
CASE I CASE II
Column LOADING I LOADING II LOADING I LOADING II
C1 0.58 in 0.24 in 0.69 in 0.31 in
C2 0.46 in 0.34 in 0.53 in 0.40 in
C3 0.46 in 0.35 in 0.53 in 0.39 in
C4 0.53 in 0.55 in 0.69 in 0.72 in
C5 0.48 in 0.19 in 0.56 in 0.21 in
C6 0.42 in 0.30 in 0.42 in 0.32 in
C7 0.42 in 0.32 in 0.42 in 0.30 in
C8 0.48 in 0.52 in 0.56 in 0.63 in
75
Table 4.16. Column punching shears of CASE I and II (Fig. 4.36) for LOADING
I and II.
CASE I CASE II
Co1wnn LOADING I (k) LOADING II (k) LOADING I (k) LOADING II (k)
CI 620 325 581 308
C2 I 170 878 1I51 861
C3 II 70 877 1I51 866
C4 620 605 581 564
C5 II75 585 II56 589
C6 2210 1656 2212 1654
C7 2210 1658 2212 1664
C8 II 75 II91 II 56 II67
Table 4. I 7. Column face positive moments of CASE I and II (Fig. 4.36) for
LOADING I and II.
76
Table 4.18. Design positive moments for CASE I and II (Fig. 4.36).
Table 4.20. Design negative moments for CASE I and II (Fig. 4.36).
77
Table 4.21. Column strip flexural shears of CASE I and II for LOADING I and II
(Fig. 4.36).
Table 4.23. Neck flexural shears for CASE I and II (Fig. 4.36).
78
Table 4.24. Design requirements for CASE I and II (Fig. 4.36).
CASE tavg +ve A,'''' -ve ~avg Concrete Saving Steel Saving
(in) (in'/ft) (in'/ft) W.r.t. CASE I w.r.t. CASE I
I 42 1.52 1.52 Not applicable
II 29 1.23 1.24 31% 19%
Once again, economy of mat with non-uniform thickness is significant (Table 4.25). It
is of interest to note that despite application of heavy loads, differential deflections
are well below the allowable limits, even when wind load is considered (Table 4.15).
One other important thing to be noted is the column punching shear. Presence of
column base moments does not affect column punching shears much. Mainly axial
column loads determine punching shears. In this regard, a list of punching shears
W.r.t. respective column axial loads is presented in Table 4.26 for both LOADING I
and II. Table 4.26 shows that the percent column axial load encountered as punching
shear remarkably matches for both LOADING I and LOADING II.
79
Table 4.26. Punching shears as a percentage of respective column axial loads for
CASE I and CASE II (Fig. 4.36).
CASE I CASE II
Columns LOADING I LOADING II LOADING I LOADING II
Cl 83.5% 84.9% 78.3% 80.4%
C2 90.8% 90.9% 89.3% 89.2%
C3 90.8%
.
90.6% 89.3% 89.4%
C4 83.5% 82.9% 78.3% 77.2%
C5 90.7% 93.6% 89.3% 94.3%
C6 94.4% 94.5% 94.4% 94.4%
C7 94.4% 94.3% 94.4% 94.6%
C8 90.7% 90.4% 89.3% 88.6%
LOADING I governs all shear criteria while moment criteria are mainly governed by
LOADING II. Once again, whether thickness is lowered away from the columns or
not, positive moments are found to be low enough to be covered by minimum
reinforcement. Near the column faces, 1.52 in2/ft steel (minimum requirement for
3.50 ft thickness) is provided across the entire column strip widths. This gives total
moment capacities of 3173.00 k-ft and 4706.00 k-ft for strips CI-C4 and C5-C8
respectively, against acting maximum moments of 2360.00 k-ft and 4053.00 k-ft
respectively. Flexural shear capacity of strip CI-C2 and C5-C6 at a distance d, for
3.50 ft thickness, away from the faces of column C2 and C6 respectively are 621.20 k
and 923.80 k respectively, against a maximum acting total shear of 480.40 k and
815.00 k respectively, which are considerably higher than requirements. In the
transition zones, average mat thickness of column strip 1 and 2 are 30.90 inches and
29.70 inches respectively, giving flexural shear capacities equal to 35.30 klft and
33.80 klft respectively thereby. The respective maximum acting kink shears are 28.40
80
kift and 33.90 kift and it is clear that no further reduction of slope width can be
performed.
Reduction of the thickness of mat away from the columns produces shear
magnification in the transition zones in the central region of mat as before. This
seems to be an important factor to be cautious about when mat thickness is to be
reduced. This has also been verified by the preceding example of Article 4.5 (MAT
II). For MAT II (Fig. 4.18) kink shear magnification has been found to be dependent
on slope of the transition zone. So a study is made by varying the slope width of
CASE II (Fig. 4.36) from 1.00 ft to 3.00 ft. A plot of kink shear magnification versus
slope angle is presented in Fig. 4.46. Since shear criteria are governed by gravity
loading i.e. LOADING I, only magnification for this load case is considered. Fig.
4.46 shows that once again finding resembles that of the previous study (Fig. 4.26
and Fig. 4.30). Flattening of the slope reduces kink shear magnification. From both
the examples, namely MAT II of Fig. 4.18 and CASE II of Fig. 4.36, it seems that a
slope angle near 20° is practically sufficient against any alarming shear concentration
in the transition zone (Fig. 4.47).
However in the transition zones, average thickness, and hence average shear
capacity, of mat will be higher than the acting shears if slope angle is not too high
there. As a matter of fact, slope angle much higher than 20° will be acceptable
because of this excess shear capacity in the transition zones, which will take care of
the resulting shear magnification. In the previous examples, namely MAT II of Fig.
4.18 and CASE II of Fig. 4.36, slope angle equal to 26.6° and 36.8° respectively has
been used without any possibility of shear failure in the transition zones. In this
context, it is of interest to note that although MAT III of Fig. 4.18 has been found to
be unacceptable due to inadequate neck thickness, it is sufficiently strong against kink
shears. With a slope angle equal to 33.7°, MAT III has a minimum kink shear
capacity equal to 16 kift which is just equal to the maximum acting kink shear.
81
1.5 - Left face of column C2
--- Right face of column C2
---*- Left face of column C6
1.4 -y- Right face of column C6
c
.2
iii 1.3
u
<;;:
.c
OJ
'E~" 1.2
QJ
.r:'" 1.1
(f)
-'c"
S2
1.0
0.9
(a)
'E" 1.2
1.1
1.0
0.9
25 30 35 40 45 50 55
Slope angle (degree)
(b)
Fig. 4.46. Effect of slope width on kink shear as a function of (a) slope width
and (b) slope angle (CASE" of Fig. 4.36 under LOADING I of Eqn.
4.26).
-.- Column C2 of MAT II (Fig. 4.18)
- Column C4 of MAT II (Fig..4.18)
1.5 --A- Column C2 of CASE II (Fig. 4.35)
-..y- Column C6 of CASE II (Fig. 4.35)
1.4
c
.Q
iii
()
~ 1.3
c
Cl
ro
E
~
ro
OJ
.<:
en
-'C"
~ 1.2
1.1
1.0
20 30 40 50 60
Angle of slope (degree)
Fig. 4.47. Effect of slope angle on kink shear magnification (MAT II of Fig. 4.18
and CASE II of Fig. 4.36 under LOADING I of Eqn. 4.26).
4.8 VERIFICATION OF THE FINDINGS ABOUT MAT WITH NON-
UNIFORM THICKNESS
In order to verify the validity of the findings about mat with non-uniform thickness
(Article 4.6.4) another mat is designed. Plan of this new mat, which will subsequently
be referred to as MA T IV, is similar to that shown in Fig. 4.3 5. Various features of
MAT IV are as follows:
Loads on this mat is calculated for a 20 story building using MICROFEAP-II (1988)
and are listed in Table 4.27.
82
Table 4.27. Column loads of MAT IV for LOADING III and IV.
Now, punching shear and neck shear, which determine the overall shape of mat, are
not much sensitive to mat thickness. So shears found for any reasonable mat
thickness can be used for the subsequent trials. Also, as punching shears are mainly
determined by column axial forces, design punching shears are expected to come
from LOADING III. So a uniform thickness of mat equal to 50.00 inches is selected
for the first trial, which has been calculated by taking the maximum column axial
force i.e. 3096 k (Table 4.27), as the acting maximum punching shear.
FE solution of MAT IV with this trial uniform thickness gives the shear force
diagrams of column strip 1 and 2 as shown in Fig. 4.48. Punching shear around
column C6 or C7, as obtained form this trial solution, is 2905.50 k. The greater
thickness for MAT IV is calculated form this punching shear and is found to be 4.00
ft. Width of greater thickness is taken to be equal to the greater thickness itself i.e.
4.00 ft.
83
-- Column strip C1-C4
60 -- Column strip C5-C8
CF
40
20
CF
2
:>2
~
~
III
;
Q)
J:: 0
(J)
-20
-40
CF
CF
-60
o 10 20 30 40
X (ft)
Fig. 4.48. Average shear force diagrams of column strip (a) C1-C4 and (b) C5-C8
(Fig. 4.35) of MAT IV under LOADING III for a tentative thickness of 50
=
inches (CF column face).
If a slope width equal to 2.125 ft is selected for MAT IV, maximum neck shear is
found to be 31.40 k (Fig. 4.48). This amount of neck shear requires a smaller
thickness equal to 2.375 ft (Equation 4.23), giving a slope angle equal to 37.40°
thereby. Slope angle has been 26.60° for MAT II of Fig. 4.18 and 36.80° for CASE
II of Fig. 4.36. In case of MAT II, kink shear capacity has been well above maximum
acting kink shear whereas kink shear capacity has been just sufficient for maximum
acting kink shear for CASE II. So this high value of slope angle i.e. 37.40°, is being
tried for MAT N with caution.
It is also decided in advance that minimum steel required for the greater thickness,
which is equal to 1.76 in2/ft for MAT IV, will be provided as bottom reinforcement
under the columns.
Analysis of MAT IV with the new dimensions gives the deflection, bending moment
and shear force diagrams as shown in Fig. 4.49 through Fig. 4.54. Column punching
shears as obtained from this solution are presented in Table 4.28. In Table 4.28,
column punching shears, expressed as a percentage of respective column axial forces,
are also given.
Comparison of different acting moments and shears and the corresponding moment
and shear capacities for MAT IV with non-uniform thickness is compiled in Table
4.29. Finally the estimated and the designed values of various items of MAT IV with
non-uniform thickness are presented in Table 4.30.
Table 4.28 reveals that column punching shears are mainly determined from column
axial forces. Table 4.29 confirms that the adopted dimensions for MAT N are
acceptable. However, average kink shear capacity is found to be just sufficient for the
loadings considered (Table 4.29), which indicates that no further increase of the slope
angle of the underside of the transition zones can be implemented. Table 4.30 serves
as a strong evidence of the reliability of the findings obtained from the present study
about mat with non-uniform thickness. As a final check on the design, Fig. 4.49 and
84 I
~
':.\f:
Fig. 4.49. Deflection diagrams of the column lines of MAT IV under LOADING III
of Eqn. 4.28.
r
,
. .
,
..-....----'
0.00
-0.02
~ -0.04
~
c
o
U
Q)
""Q) -0.06
o
-0.08
(a)
-0.02
-0.04
~
~
c
0
:;:; /
u
Q)
""0
Q) -0.06
o 10 20 30 40
Y (ft)
(b)
Fig. 4.50. Deflection diagrams of the column lines of MAT IV under LOADING IV
of Eqn. 4.29.
o
-- Column strip Cl-C4 CF
CF
200 - Column strip C5-C8
100
-6
~
-
c
ill
E
0
CF
0
:2
-100
-200
0 10 20 30 40
X (ft)
(a)
100 r~
~
•
-
~
c
ill
E
0
CF
0
:2
-100
-200
o 10 20 30 40
Y (ft)
(b)
Fig. 4.51. Bending moment diagrams of the column strips of MAT IV under
LOADING III of Eqn. 4.28 (CF = column face).
ov
300
- Column strip C1-C4
CF
- Column strip C5-C8
200
CF
CF
100 CF
2
2, C
-
~
C
Ql
E
0
0 CF
:2 -100
-200
-300
0 20 40 60 80
X (ft)
(a)
250
- Column strip C1-C5 CF
CF
200 - Column strip C2-C6
~ Column strip C3-C7
150 -- Column strip C4-C8
100
--
~
6
C
50
0
Ql
E
0
-50
:2
-100
-150
-200
0 10 20 30 40
Y (ft)
(b)
40
20
2 CF
6
~
wi
l1l 0
Ql
.r:
(/)
-20
-40 CF
CF
-60
0 10 20 30 40
X (ft)
(a)
40
20
2 CF
:>2
~ wi
~ 0
l1l
Ql
.r:
(/)
-20
-40 CF
-60 CF
0 10 20 30 40
Y (ft)
(b)
Fig. 4.53. Shear force diagrams of the column strips of MAT IV under LOADING III
of Eqn. 4.28 (eF = column face).
40 CF CF
CF
20
CF
~ 0
~
~
ro
Ql
.c
(fJ -20
CF
-40
(a)
20
~
.:= CF
:>2
~
~ 0
ro
Ql
.c
(fJ
-20
-40
CF
-60 CF
0 10 20 30 40
Y (ft)
(b)
Fig. 4.54. Shear force diagrams for the columnstrips of MAT IV under LOADING IV
of Eqn. 4.29 (CF = Column face).
Fig. 4.50 show that differential column settlements are far below the allowable limits.
In comparison to mat with uniform thickness required for LOADING III and IV (48
inches thick with 1.76 in2/f! top and bottom reinforcements), MAT IV saves 28%
concrete and 14% reinforcement.
Table 4.28. Column punching shears of MAT IV for LOADING III and IV.
Table 4.29. Comparison of acting moments and shears with respective moment
and shear capacities for MAT IV with non-uniform thickness.
85
Table 4.30. Comparison of the estimated and designed values for MAT IV with
non-uniform thickness.
In Article 4.4.8, it has already been established that since column base moments are
small for gravity load combination, their effect can be neglected in the analysis of mat
without causing any noticeable change in results. Now, column base moments are
high for lateral load combination. But there is no provision for calculating column
base moments in the conventional practice of load calculation, where effect of lateral
loads is simulated by simply increasing the column axial forces due to gravity loads
on the windward side and reducing those on the leeward side. It is felt necessary in
the present study to investigate the effect of neglecting column base moments in case
of lateral load combination. CASE I of Fig. 4.36 is selected for this purpose. Exact
column loads for LOADING II (Equation 4.27) of this mat have already been listed
in Table 4.14. Approximate column loads for LOADING II are given in Table 4.31.
86
Table 4.31. Approximate column loads of CASE II (Fig. 4.36) for LOADING II
(Equation 4.27).
Column C1 C2 C3 C4 C5 C6 C7 C8
Axial 369.8 904.2 1028.7 743.3 597.9 1631.3 1880.3 1344.9
force (k)
Deflection, bending moment and shear force diagrams of CASE I (Fig. 4.36) under
exact and approximate loading by different methods are presented in Fig. 4.55
through Fig. 4.63. Punching shears corresponding to exact and approximate loads are
compiled in Table 4.32. Table 4.32 shows that punching shear is not much dependent
on column base moments.
Table 4.32: Column punching shears for exact and approximate loads (LOADING
II) of CASE II (Fig. 4.36)
87
-0.02
-0.03
p
:=-c
0
:;:;
(J
Q)
;:
Q) -0.04
0
-0.05
- Finite Element method with exact loads
- Finite Element method with approximate loads
o 20 40 60 80
X (ft)
(a)
-0.01
-0.02
p
:=-c -0.03
o
t5
Q)
;:
Q)
o -0.04
o 20 40 60 80
X (ft)
(b)
Fig. 4.55. Comparison of deflection calculation of line (a) C1-C4 and (b) C5-C8
(Fig. 4.36) under exact and approximate loadings.
-0.0150
-0.0175
~
.:=
~
c
0 -0.0200
15
Q)
;;::
Q)
0
-0.0225
(a)
-0.020
g -0.025
c
o
15
Q)
;;::
Q)
o
-0.030
(b)
Fig. 4.56. Comparison of deflection calculation of line (a) C1-C5 and (b) C2-C6
(Fig. 4.36) under exact and approximate loadings.
-0.025
c
o
t5 Ql
'iii -0030
a
(a)
-0.035
~ -0.040
<:=
~
c
o
U
Ql
-=
Ql
a
-0.045
o 10 20 30 40
X (ft)
(b)
Fig. 4.57. Comparison of deflection calculation of line (a) C3-C7 and (b) C4-C8
(Fig. 4.28) under exact and approximate loadings.
150
CF
100 CF
CF
50 CF
~
CF
~, 0
-
-'"
~
c
Ql
E
-50
CF
0
:2 -100
-150
200
CF
CF
100 CF
CF
~
;:::
~, CF
0
-'"
~ CF
"E
Ql
E
0
:2 -100
-200
- Finije Element method with exact loads
- Finite Element method wijh approximate loads
o 20 40 60 80
X (ft)
(b)
Fig. 4.58. Comparison of bending moment calculation of line (a) C1-C4 and (b) C5-C8
(Fig. 4.36) under exact and approximate loadings (CF = column face).
60 - Finite Element method with exact loads
- Finite Element method with approximate loads
CF
40
20
CF
~,
a
6
C -20
Ql
E
o -40
:2
-60
-80
-100
a 10 20 30 40
Y (ft)
(a)
150
- Finite Element method with exact loads
- Finite Element method with approximate loads
100 CF
CF
50
---
Z'
:;:,
, CF
6
c
a
Ql
E
0
:2 -50
-100
a 10 20 30 40
Y (ft)
(b)
Fig. 4.59. Comparison of bending moment calculation of line (a) C1-C5 and (b) C2-C6
(Fig. 4.36) under exact and approximate loadings (CF = column face).
150
-- Finite Element method with exact loads
- Finite Element method with approximate loads
100 CF
CF
~ 50
~
2,
-
~
c
Q)
E
0
CF
CF
0
:2
-50
-100
0 10 20 30 40
Y (ft)
(a)
150
-- Finite Ele ent method with ex ct loads
- Finite Element method with approximate loads CF
100 CF
50
2' CF
2, 0
-
.:<:
~
c
Q)
E -50
0
:2
-100
-150
0 10 20 30 40
Y (ft)
(b)
Fig. 4.60. Comparison of bending moment calculation of line (a) C3-C7 and (b) C4-C8
(Fig. 4.36) under exact and approximate loadings (CF = column face).
- Finite Element method with exact loads
-- Finite Element method with approximate loads
CF CF
CF
CF
CF
CF
o 20 40 60 80
X (tt)
(a)
CF
CF
CF
o 20 40 60 80
X (tt)
(b)
Fig. 4.61. Comparison of shear force calculation of line (a) C1-C4 and (b) C5-C8
(Fig. 4.36) under exact and approximate loadings (CF = column face).
-- Finite Element method with exact loads
--- Finite Element method with approximate loads
20 . CF
10
~
il:::
CF
6 L-
III
Q)
0 ••
.r:
(j)
-10
CF
CF
-20
0 10 20 30 40
Y (ft)
(a)
50
-- Finite Element method with exact loads
--- Finite Element method with approximate loads
CF
~
il:::
CF
6 L-
III
Q)
.r:
0 ••
(j)
CF
CF
o 10 20 30 40
Y (ft)
(b)
Fig. 4.62. Comparison of shear force calculation of line (a) C1-C5 and (b) C2-C6
(Fig. 4.36) under exact and approximate loadings (CF = column face).
50
-- Finite Element method with exact loads
-- Finite Element method with approximate loads
CF
~ CF
~o"
w
.t::
(f)
CF
CF
o 10 20 30 40
Y (ft)
(a)
-6
Z'
~ CF
ro
w
.t:: 0
••
(f)
CF
CF.
o 10 20 30 40
Y (ft)
(b)
Fig. 4.63. Comparison of shear force calculation of line (a) C3-C7 and (b) C4-C8
(Fig. 4.36) under exact and approximate loadings (CF = column face).
It is evident that column base moments do not affect mat response to the extent as
expected usually. Analysis of mat with approximate loads is almost as accurate as the
calculation of approximate column axial forces.
88
CHAPTERS
5.1 GENERAL
Two examples of mat foundation have been analyzed by Finite Element method using
Ahmad's thick shell element, Conventional method and ACI Approximate Flexible
method. Results obtained from these analyses have been compared for evaluating
relative performance of these analysis methods. One of the examples has further been
analyzed by varying its material and geometric properties in an effort to investigate
the parametric effects on mat behavior. Two mats have also been designed using the
results found from the above three methods in order to delineate the relative
economy associated with them. Finally, a reshaping scheme for mat has been
examined with a view to finding more economy in mat design. Three mats have been
analyzed in this context using both uniform and non-uniform thickness in order to
investigate relative merits and demerits of the non-uniform thickness scheme. Effects
of the new geometric parameters introduced by this proposition have also been
investigated. Findings of the study made so far are discussed and conclusions drawn
from them are listed in this chapter.
Comparison of Finite Element analysis with Conventional and ACI methods has been
made in two different ways. The first approach has been based on the bending
89
moments and shear forces of the most highly stressed strip of unit width and the
second one has been based on average column strip bending moments and shear
forces. Unless otherwise specified, all discussions will be with respect to Finite
Element method.
Since there is no practically conceivable term such as average column strip deflection,
comparison of deflections has been unique regardless of the widths of the strips
considered. Comparison of column line deflections has revealed that ACI method
substantially overestimates edge deflections and underestimates central deflections in
comparison to FE method. This feature of ACI method results in considerably large
differential column deflections. However, since limits of tolerable differential
deflection are relatively high for mat, overestimation of differential deflections is
seldom expected to cause any difficulty.
Conventional method can not depict the variational pattern of mat bending moments
and shear forces. This extensively used, simple and popular method considers each
column strip of mat as a whole throughout the analysis and thus always gives average
column strip moments and shears.
Now, intensity of bending moments and shear forces is relatively high in the vicinity
of columns and reduces substantially away from the columns. Also, columns of mat
are usually spaced at considerable distances and positive moments (upside
compression) occur only near the columns. This leaves most of the parts of mat under
the rein of negative moments. If a cross-section in-between the columns of any
column strip is considered, then bending moments across the width of the column
strip are all negative and their magnitudes do not vary much. But at a cross-section
near the column faces of any column strip, there are large positive bending moments
90
adjacent to the columns whereas bending moments near the edges of the column strip
are very small and may be positive or negative . As a result, average column strip
positive bending moments will be very low while magnitudes of negative bending
moments will not change much after averaging is performed across the column strip
width.
The above explanation has been found to be true in the current study. Comparison
has shown that Conventional method gives much lower positive bending moments
and relatively accurate estimates of negative bending moments in comparison to FE
method.
Due to the averaging of shear forces across the widths of the column strips,flexural
shear forces calculated by Conventional method are very low. However, this method
is excellent in calculating punching shears.
All the methods are almost equally good at predicting punching shears around central
columns. However, ACI method considerably underestimates punching shears around
91
Q
columns near mat edges and thus becomes unreliable when heavy column loads occur
in these regions.
When design is performed on the basis of bending moments and shear forces of the
most highly stressed strips of unit width, thickness of mat is governed by punching
shears for Conventional method and by flexural shears for ACI and FE methods. If
this approach of design is followed, Conventional method gives the thinnest mat and
ACI method requires the thickest mat. As a consequence of the greater thickness
requirement in case of FE and ACI methods, minimum steel as per the requirements
of ACI code governs at every point of mat. Conventional method requires slightly
higher reinforcement in the negative moment regions of mat. Elsewhere minimum
reinforcement governs for this method too. As a result, Conventional method gives
the least costly solution and ACI method gives the most uneconomic design.
5.2.2 Comparison of Bending Moments and Shear Forces Based on the Entire
Column Strips
It has been felt unrealistic in the present study that strips of unit width act individually
in resisting bending moments and shear forces. It is more likely that a wider portion
of any column strip will structurally act together. But no experimental data is
available about the actual width of this portion. However, a study based on FE
analysis has been conducted with a view to finding the minimum averaging width for
which punching shear governs the thickness design of mat. It has been observed that
this minimum averaging width is less than 50% of the column spacing. Since
conventional practice of considering the entire width of a column strip for design
purpose has not been reported to result in any significant functional discrepancy so
far, average column strip bending moments and shear forces have finally been
adopted as the design criteria for mat.
92
Prediction of average column strip flexural shears by Conventional method matches
FE solution very well. But ACI method overestimates average column strip flexural
shears near the column faces which are the locations important for design. As a
result, although punching shears govern thickness design for FE and Conventional
methods; flexural shears, in case of ACI method, continue to be the deciding factor in
this regard, giving higher required thickness. Since the difference in punching shears
as calculated form FE and Conventional methods is not significant, thickness
requirements by these two methods are almost the same.
(i) Design is performed column strip-wise, making the actual variational pattern of
internal forces less significant thereby.
93
(ii) Mat thickness is governed by punching shears, which can be predicted with
excellent accuracy by Conventional method.
It has also been found that extra cost associated with Conventional method arises
from its overestimation of negative moments. This overestimation results in steel
ratio for negative moments in excess of the minimum limit.
ACI method gives uneconomic design compared to FE method. The amount of extra
cost associated with ACI method varies with the magnitude of column loads. In the
present study, this extra cost has been found to be only about 2% compared to FE
method for a 7 story building whereas the extra cost has risen to about 35% for a 12
story building. Evidently, difference of costs between ACI and FE methods will be
higher for taller buildings. Uneconomic performance of ACI method is principally
due to the overestimation of flexural shears by this method in the vicinity of columns
which requires higher mat thickness than FE method. ACI method can broadly be
divided into two phases, namely infinite plate solution phase and end correction
phase. It has been found that infinite plate solution matches FE solution excellently
well. So it can be inferred that uneconomic solution of ACI method results from the
end corrections.
94
•
The material parameters associated with mat foundation are modulus of subgrade
reaction and concrete strength. The main geometric parameter of mat is its thickness.
However, other geometric parameters include column spacing, column dimension and
width of overhanging portion.
Mat deflection has been found to be the most sensitive item of all to its parameters.
However, mat deflection has also been found to be the least significant item of all
from design point of view. So its parametric sensitivity is not of much practical
significance. This item decreases with the increase of concrete strength, modulus of
subgrade reaction and mat thickness. However, effects of the last two parameters on
mat deflection are quite significant.
Mat shear has been found to be insensitive to most of its parameters. The only
noticeable variation of flexural shear is that near mat edges with respect to mat
thickness. However, since critical flexural shears usually occur near central columns,
this variation will seldom be of any practical significance.
Positive moments are less sensitive to mat parameters than negative moments and this
sensitivity is confined in the central portion of mat. In this region, positive moment
decreases with the increase of concrete strength and mat thickness, but increases with
the increase of modulus of subgrade reaction. Since positive moments are small
95
enough to be usually covered by minimum steel, its variation with mat parameters is
expected to be of no practical significance. Response of negative moments to
parametric variation is opposite to that of positive moments but of much higher
degree. Modulus of subgrade reaction and mat thickness have the most significant
effects on negative moments.
It has been observed in the current study that punching shears and flexural shears of
mat are principally determined by column axial forces. Column base moments have
much less influence on the shear behavior of mat. Now, of the two loading cases to
be considered in connection with the calculation of column loads, namely the dead
load and live load combination and the dead load, live load and lateral load
combination; the second combination is expected to give much lower column axial
forces due to 25% reduction of loads associated with this loading case. As a result,
lateral load combination will seldom govern design punching and flexural shear
forces. For preliminary dimensioning of mat it will always be sufficient to consider
the case with gravity loads only since mat thickness is governed by shear forces.
96
t
•
Plan area of mat can be divided into two distinguished zones. First of these includes
the neighborhoods of the columns where high flexural shears, punching shears and
positive moments occur. The second region consists of the areas in-between the
columns which are subject to high negative moments but low shear forces. Since
thickness of mat is determined from shear criteria, it is obvious that thickness
requirement in the second zone is much lower than that in the first zone. So it is
possible to provide lower thickness in the second zone making mat thickness non-
uniform.
However, available methods have no provision for analyzing mat with non-uniform
thickness. As a result, conventional practice has been to provide uniform thickness
for mat as found from the greatest punching shear. But the present Finite Element
software has the great potential of analyzing mats with both uniform and non-uniform
thicknesses. Finite Element investigation of three design examples made in this regard
has shown that mat with non-uniform thickness offers substantial economy in terms
of both concrete and reinforcement costs, which has been found to be of the order of
20 to 30% compared to mat with uniform thickness.
This approach of mat dimensioning creates transition zones around the columns
across which the change in thickness is implemented gradually. Neck sections i.e. the
peripheral sections around columns where the region of smaller thickness begins, are
also created as a result of such thickness reduction. Effects of the new geometric
parameters evolved as a result of this non-uniformity in thickness have been
investigated. It has been found that width of the transition zones should be such that
the slope angle of the transition zones be around 200 in order to avoid any sharp
amplification of flexural shear in these zones. Width of greater thickness around the
columns equal to the greater thickness itself, or at least equal to the effective depth
corresponding to the greater thickness, has been found to be satisfactory. It has been
97
found that flexural shears beyond the transition zones remain practically unaltered by
any parametric variation. It is also evident that punching shears are not critical at
these locations. Consequently, it can be inferred that neck thickness can be designed
from the flexural shears at probable neck locations obtained form the solution of the
problem at hand with uniform thickness equal to that found tentatively by treating the
maximum column axial force as the critical punching shear. Despite significant
magnification of column face positive moments, provision of the minimum
reinforcement, as required for the greater thickness, across the entire column strip
widths at these locations is expected to furnish average column strip positive moment
capacity with adequate margin above the respective acting moments. This kind of
positive moment reinforcement provision will also result in more efficient use of
material strength which is often sacrificed in mat of uniform thickness in order to
comply with the minimum requirements set forth by ACI code of practice. Thickness
reduction in the negative moment zones will reduce negative moments which implies
that smaller thickness as found by maximum neck shear will be able to handle
negative moments with reasonably low steel ratio.
From the present study it has also been found that although slope angle around 20°
results in negligible shear magnification in the transition zones, some degree of shear
magnification is actually affordable in these regions. This is because if slope angle is
around 20°, average flexural shear capacity of the transition zones is higher than the
flexural shears at these locations. Also, steeper slope gives neck sections closer to
column faces .i.e. closer to the peaks of shear force diagrams. As a result neck shear
will be higher. So neck thickness, and hence the thickness of the transition zones, will
increase with the increase of shear capacity in the transition zones. Evidently, steeper
slopes can be used in the transition zones to make use of this excess flexural shear
capacity of these regions. In the present study, slope angle upto around 35° has been
found to be safe and it has appeared that slope angle upto around 30° can safely be
recommended.
98
5.6 CONCLUSIONS
General conclusions drawn form the present study are summarized below.
(ii) Conventional method gives no column face positive bending moments. This
method gives relatively close estimates of negative bending moments in
comparison to Finite Element method when strips of unit width are considered.
However, if average column strip moments are considered, Conventional
method gives much higher negative moments than Finite Element method.
(iii) Whether strips of unit width or the entire column strips are considered, ACI
method gives substantially low positive moments at the faces of columns and
considerably high negative 'bending moments in-between the columns in
comparison to Finite Element method.
(iv) When strips of unit width are considered, flexural shear forces calculated by
Conventional method are very low compared to Finite Element method. But
calculation of average column strip flexural shears by Conventional method
matches Finite Element solution very well.
(v) ACI method gives higher flexural shears near column faces than Finite Element
method for both strips of unit width and column strips.
(vi) Conventional method is excellent in calculating punching shears compared to
Finite Element method. Although ACI method gives close estimate of punching
shears around central columns in comparison to Finite Element method, this
method considerably underestimates punching shears around columns near mat
edges.
99
column spacing, revealing that design of mat based on average column strip
forces will be acceptable if at least 50% width of any column strip acts
integrally.
(ix) Mat shear has been found to be insensitive to most of its parameters.
(x) Response of negative moments to parametric variation is opposite to that of
positive moments but of higher degree.
(xi) Deflection is seldom expected to cause any difficulty in mat design since limits
of tolerable differential deflection are relatively high for mat.
(xii) Positive moments are small enough to be covered by minimum steel.
(xiii) Shear behavior of mat is principally determined by column axial forces.
(xiv) Parametric sensitivity of deflection and positive moment are of little practical
significance.
(xv) When design is performed on the basis of forces of strips of unit width,
Conventional method gives the most economic solution and ACI method
results in the most costly design. But if mat is designed for average column
strip forces, which is the most widely practiced approach, Finite Element
method gives the most economic solution. Conventional method requires
slightly higher reinforcement cost than Finite Element method and ACI method
results in much higher material cost than Finite Element method. ••
(xvi) Since minimum requirements. govern mat design frequently, Conventional
method shows quite satisfactory performance. Extra cost associated with
Conventional method is due to its overestimation of negative moments.
(xvii) The amount of extra cost associated with ACI method sharply increases with
the increase of column loads and is due to the overestimation of flexural shears
by this method near column faces.
(xix) Design of mat with non-uniform thickness is not much complicated compared
to that of mat with uniform thickness. •
100
5.6.2 Design Approach for Mat with Non-uniform Thickness
The present study has revealed that the design approach for mat with non-uniform
thickness should be as follows:
(i) The problem at hand should be solved first with a uniform thickness equal to
that found tentatively by treating the maximum column axial force, due to the
dead load and live load combination only, as the critical punching shear (this
solution will be referred to as tentative solution in the following discussion).
(ii) Greater thickness under the columns should be designed from column
punching shears calculated from the tentative solution.
(iii) This greater thickness should be provided around the column peripheries over
a distance equal to the greater thickness itself or the effective depth
corresponding to the greater thickness on any side of the columns.
(iv) Smaller thickness should be calculated from the maximum flexural shear,
found from the tentative solution, at the probable neck sections.
(v) Width of the transition zones (for the variation of thickness) should be such
that slope angle of the underside of mat in these zones will be between 20° to
30°. A number of simple trials will be sufficient to find a compromise between
steps (iii) and (iv).
(vi) The problem should be analyzed again with the new geometry just designed in
order to calculate reinforcements. No further trial for geometric design will be
necessary. But it is always better to make a check on the adopted geometry
using the shear force diagrams and punching shears found from the new
solution.
(vii) Minimum reinforcement required for the greater thickness should be provided
as bottom reinforcement under the columns across the entire widths of
column strips. Once again, the new solution may be used to check the
adequacy of these reinforcements.
101
(ix) Bottom reinforcements in-between the columns and top reinforcements under
the columns should be calculated as per minimum requirements specified by
ACI code.
The following recommendations are made for future investigation on mat foundation.
(i) - Square mats with regularly spaced column grids have been investigated in the
current study. Behavior of irregularly shaped mat, mat with punches, mat with
irregular column arrangements and mat with shear walls may be investigated.
(ii) Both mat and supporting soil have been modeled as linearly elastic material in
the current study. A non-linear analysis may be performed.
(iii) Soil has been modeled as Winkler medium in the current investigation because
of the simplicity of this model. A half space modeling of the supporting soil
with FE analysis may be attempted in future.
(iv) Dynamic mat-soil interaction can be analyzed using FE method.
(v) The integrated structure consisting of mat monolithically built with the
building frame with and without shear walls may be analyzed for both static
and dynamic response.
(vi) Rectangular and circular mats subject to peripheral line lodes, as are found ••
under silos and oil tanks, may be analyzed.
102
REFERENCES
(1) ACI Committee 436 (at present committee 336), "Suggested Design Procedure
for Combined Footing and Mats" , ACI Journal Proceedings, Vol. 63, No. 10,
October, 1966, pp. 1041-1057.
(2) Ahmad, S., Curved Finite Elements in the Analysis of Solid, Shell and Plate
Structures, Ph.D. Thesis, University College of Swansea, 1969.
(3) Ahmad, S., "General Thick Shell Element Program - Part A and Part B ",
Computer Program Report, No. 22, University of Wales, Swansea, October,
1969.
(4) Ahmad, S., Irons, B.M. and Zienkiewicz, O.C., "Analysis of Thick and Thin
Shell Structures by Curved Finite Elements" , International Journal of
Numerical Methods in Engineering, Vol. 2, 1970, pp. 419-451.
(5) Amanat, KM., A Design Rationale for Free Standing Stair Slab Based on
Finite Element Analysis, M.Sc. Thesis, BUET, Dhaka, September, 1993.
(6) Arora, KR., Soil Mechanics and Foundation Engineering, 1st Ed., Standard
Publishers and Distributors, New Delhi, 1992, pp. 948-975.
(7) Bathe, KJ., Numerical Methods in Finite Element Analysis, 1st Ed., Prentice
Hall Inc., New Jersey, 1976.
(8) Bowels, J.E., Foundation Analysis and Design, 2nd Ed., McGraw Hill Book
Co., New York, 1968, pp. 294-320.
103
(9) Bowels, I.E., Analytical and Computer Methods in Foundation Engineering,
1st Ed., McGraw Hill Book Co., New York, 1974, pp. 209-254.
(10) Bowels, I.E., "Mat Design" , ACI Journal, Vol. 83, NO.6, November-
December, 1986, pp. 1010-1017.
(11) Bowels, I.E., Foundation Analysis and Design ,4th Ed., McGraw Hill Book
Co., New York, 1988, pp. 436-470.
(12) Chilton, D.S. and Wekezer, I.w., "Plates on Elastic Foundation", Journal of
Structural Engineering, ASCE, Vol. 116, No. 11, November, 1990, pp. 3236-
3241.
(13) Das, B.M., Advanced Soil Mechanics, 1st Ed., Hemisphere Publishing
Corporation, Washington, 1983, pp. 167-241 and 339-401.
(16) Gupta, A. and Patel, A.N., "Rational Analysis of Mat Foundations Using Finite
Element Technique", Proceedings of the International Seminar on Civil
Engineering Practice in the Twenty first Century, Roorkee, Vol. I, February,
1996, pp. 149-159.
(17) Hemsley, I.A., "Elastic Solutions for Axisymetrically Loaded Circular Raft with
,
Free or Clamped Edges Founded on Winkler Springs or a Half-space",
Proceedings of the Institute of Civil Engineers (London), Part 2, Vol. 83,
104
March, 1987, pp. 61-90.
(18) Hemsley, 1.A., "Plane Elastostatic Flexure of an Infinite Plate on a Half space" ,
Proceedings of the Institute of Civil Engineers (London), Part 2, Vol. 85,
December, 1988, pp. 629-664.
(20) Hetenyi, M., Beams on Elastic foundation, 1st Ed., University of Michigan
Press, En Arbor, 1946, pp. 245-255.
(21) Hooper, 1.A. and West, DJ., "Structural Analysis of Circular Raft on Yielding
Soil", Proceedings of the Institute of Civil Engineers (London), Part 2, Vol. 75,
June, 1983, pp. 205-241.
(22) Horvath, 1.S., "New Subgrade Model Applied to Mat Foundations", Journal of
Geotechnical Engineering, ASCE, Vol. 109, No. 12, December, 1983, pp
1567-1587.
(24) Horvath, J.S., "New Subgrade Model Applied to Mat Foundations", Journal of
Geotechnical Engineering, ASCE, Vol. 119, NO.2, February, 1993, pp. 1567-
1587.
105
(26) Irons, B. and Ahmad, S., Techniques of Finite Elements, 1st Ed., Ellis.
Horwood Ltd., Chichester, 1980.
(27) Krishnamoorthy, e.S., Finite Element Analysis: Theory and Programming, 1st
Ed., Tata McGraw Hill PublishingCo. Ltd., New Delhi, 1987.
(29) Molla, MAS., Finite Element Analysis of Footings and Mat foundations,
M.Sc. Thesis, BUET, Dhaka, February, 1995.
(30) Nilson, AH. and Winter, G., Design of Concrete Structures, 10th Ed.,
McGraw Hill Book Company, Singapore, 1986.
(31) Scott, R.F., Foundation Analysis, 1st Ed., Prentice Hall Inc., New Jersey,
1981, pp. 119-200.
(32) Shukla, S.N., "A Simplified Method for Design of Mats on Elastic
Foundation" , ACI Journal, Vol. 81, No. 5, September-October, 1984, pp. 469-
475.
(33) Singh, A, Modern Geotechnical Engineering, 3rd Ed., CBS Publishers and
Distributors Pvt. Ltd., New Delhi, 1992, pp. 532-554.
(35) Teng, W.e., Foundation Design, 13th Ed., Prentice Hall of India Private
Limited, New Delhi, 1992, pp. 159-191.
106
(36) Wotsak, K.N., Sompom, A. and Sarun, U.,"MICROFEAP II: Analysis of 2D
Truss, Frame and Wall", Module PI, Release 3.1, MICRO-ACE Club, Asian
Institute of Technology, 1988.
107
APPENDIX A
The general thick shell finite element was developed by Ahmad (1969). Typical thick
shell elements are shown in Fig. AI. In thick shells, bending effect can be expected
to be significant. The transverse shear deformation is also significant. In the element
formulation, two assumptions were made. Firstly, the original normal to the middle
surface are assumed to remain straight. Secondly, the distance of a point along the
normal from the middle surface remains unaffected.
The external faces of the elements are curved, while the sections across the thickness
are generated by straight lines. Pairs of points itop and ibottom, each with given Cartesian
coordinates, describe the shape of the element.
If ~ and 11 be the two curvilinear coordinates in the middle plane of the shell element
(Fig. A2) and ~ be a linear coordinate in the thickness direction and further, if it is
assumed that ~, 11 and ~ vary between + I and -Ion the respective faces of the
element, then a relationship can be written between Cartesian coordinates of any
point of the shell and the curvilinear coordinates in the form :
(AI)
A-I
-
(a)
Parent Generalised
(b)
Fig. A. I. Thick shell elements: (a) parabolic element and (b) cubic element.
Here, N; (~ , lj) are shape functions taking a value of unity at the nodes i and zero at
all other nodes. If the basic functions, N; are derived as shape functions of a parent
two dimensional element, square or even triangular in plan and are so designed that
compatibility is achieved at interfaces, then the curved shape elements will fit into
each other. Arbitrarily curved shapes of the element can be achieved by using shape
functions of different orders. Only parabolic and cubic types are shown in Fig. A. I .
For the purpose of present analysis, a parabolic element has been used. By placing a
larger number of nodes on the surfaces of the element, more elaborate shapes can be
x X'1
z' 1 z' 1
top bottom
li
Here v 3i is a vector whose length is equal to the shell thickness.
Since strains in the direction normal to the mid surface is assumed to be negligible,
the displacement throughout the element will be taken to be uniquely defined by the
three mid surface nodal displacements in the directions of the three Cartesian
A-2
(=+ 1
,/
7]=+1
/ .
/
;-'
'>
.•
z -T,
7]=-1
x
Fig. A.2. Geometry of thick shell element.
','J. ,Y'(VV'))
V'Cu')
.
----- ---
Z(w)
Fig, A.3. Local and global coordinate systems and nodal degrees of freedom for
thick shell element.
coordinates and two rotations of the nodal vector V 3i about orthogonal directions
normal to it. If two such orthogonal directions are given by vector V 2i and v Ii of
unit magnitude with corresponding scalar rotations u; and Pi respectively, the
displacement field can be expressed as follows:
u u.1
y = LNi y.
1 + L NiSj-[Vli
t. - V2iJti}Pi
(A4)
w w.1
The thick shell program is a FORTRAN code to implement the general thick shell
element [Ahmad (1969)). The geometry of a structure is defined in a global system
which is a rectangular Cartesian coordinate system. The loading and boundary
conditions must be given in the same unit as the nodal displacements of an element.
The stresses are usually calculated at the nodal points in the global system.
The top and bottom coordinates of each node with respect to Cartesian coordinate
system are fed into the program. Coordinates for non-comer nodes lying on straight
edges are not required to be given. If these coordinates of the nodes are fed into the
program, then the shape of the element is automatically defined in the program.
Therefore the thickness of the element can vary from node to node and the edges may
be curved parabolically and cubically depending on the type of element used. The
program as at present can handle isotropic elastic material. The material properties
are defined for every element, thus allowing the program to deal with materials
varying from element to element. Temperature and pressure can be varied from node
to node.
A-3
A.2.t Output from the Program
Displacements are calculated and printed against each node in the ascending order for
every loading case. Stresses are first calculated in the local orthogonal system and
then transformed into the global Cartesian system. For every node, the top surface
stresses are followed by the bottom surface stresses.
Global stresses are also stored separately for top and bottom surfaces against node
numbers and in the end, a simple averaging is performed on them. The average
stresses are then printed out in the ascending order of the node numbers. The top
surface stresses for all the loading cases are followed by the bottom surface stresses.
First of all, the structure is divided into suitable elements and the nodes are numbered
in any suitable manner as shown in the example of Fig. A4. The elements are also
suitably numbered in some sequence on which they are fed into the computer. Two
probable sequences are shown in Fig. AA (a) and A4 (b). Each element is
topographically defined by its nodal numbers in a consistent right hand screw system
as shown in Fig. AS (a) and AS (b).
The thick shell program uses the frontal solution technique. Here the assembly of an
element stiffuess and the corresponding right hand side is immediately followed by
the process of elimination of the variables corresponding to nodes which occur for
the last time. This is indicated to the program by inserting a negative sign before
these nodes. This can easily be put in the shell structures once the element sequence
has been selected.
A-4
7
(b)
22
top surface
top surface
BOUom 'urfac::~
(A5)
abscissa of the Gauss points, f(a;) is the ordinate i.e. the function value at a;, H; is the
weight coefficient and n is the number of sampling points. Values of a; and H; depend
on the value of n. Table Al lists these values for n ranging upto 3.
In Ahmad's program, 2-point and 3-point Gauss' quadratures are applied for the 8-
noded and 12-noded elements respectively. Integration is performed in the local
element coordinate system (~, 1'], ~). Coordinates of the points within the element
A-5
varies between +1 and -1. If 2-point (n=2) integration is performed on the middle
surface, there are four Gauss points since integration is carried out in both ~ and Tj
directions. The coordinates of these four points with respect to (~ , Tj) system will be
(:to.577350269, :to.577350269). These points are shown in Fig. A.6.
n :ta H
1 0.0 2.0
2 0.577350269 1.0
3 0.774596692 0.5555555555
0.0 0.8888888888
In finite element analysis using displacement methods, the stresses are discontinuous
between elements because of the nature of the assumed displacement variation. For
this reason, stress at a node is calculated by averaging the stresses obtained at that
node from the elements common to that node. Experience has shown that in case of
isoparametric elements the Gauss' integration points are the best stress sampling
points because shape function derivatives, and hence stresses, evaluated at the
interior of the elements are more accurate than those calculated at the element
boundary. The element nodes, which are the most useful points for output and
interpretation of stresses, appear to be not so good as stress sampling points.
The thick shell element developed by Ahmad (1969) is excellent for analyzing singly
or doubly curved shell structures where load is carried by bending as well as inplane
forces. In most shell structures, transverse shear is not a very important quantity. It
has been found from experience that out of plane shears in such ordinary structures
.+ are small in magnitude and they can safely be ignored in design. Consequently, while
formulating element characteristics, less importance was given to the evaluation of
A-6
transverse shear stresses. However, stresses at Gauss points are predicted with good
accuracy in shell or plate structures like mat, where transverse shear is important. For
this reason, evaluation of stresses at Gauss points becomes essential with Ahmad's
element for such structures.
7
6
5
.0 C.
4
A B. +
I
• "
<v-
"
<v-
2
7)=-1
3
Gauss points
A,B,C,D
A-7
APPENDIXB
B.t GENERAL
(B.I)
where
A typical plot of q versus 0 from load tests is qualitatively shown in Fig. B.I(a). If
z
this type of curve is used to obtain k, which is the slope of the cu;"e as given in
Equation (B. I), it is evident that the value depends on whether it is a tangent or
secant modulus and location of the coordinates of q and O. Obviously the q-o curve
could be divide into several regions so that k takes different values of the slope in the
different regions. However, since usually analysis proceeds on the basis of estimated
or at best of approximate load test, this refinement tends to make the problem of
foundation analysis too much complicated in contradiction to the intention of
modeling soil using the soil spring i.e. to simplifY analysis.
B-1
Fig. B.2(b) is the idealized stress-deformation curve in which k is constant upto a
deflection XmoxB
. eyond Xm,xthe soil pressure is a constant value, qcon,defined by
The value ofXmoxmay be directly estimated at some small value (say 12 to 25 mm) or
from inspection of load settlement curve. It might also be estimated from the peak
strain in a triaxial compression test.
The most popular method to determine modulus of subgrade reaction is to use the
plate load test shown in Fig. B.2.
Field load-test procedure commonly uses square plates and round plates from 12
inches to 30 inches in diameter, in increments of 6 inch. Plate thickness is commonly
1 inch to reduce bending. Also, in order to reduce bending the plates are machined so
that the smaller plates can be stacked on the larger ones.
.''j
Field load testing requires some means of applying a large load to the plates. This can
be accomplished by jacking against a crawler tractor or loaded flat-bed equipment
trailers or driving tension piles used with a cross beam to provide the load reaction.
From the test results, a q-IJ curve is drawn. Taking a suitable coordinate, the slope of
the curve i.e. the modulus of subgrade reaction can be found. Since the curve is
seldom straight over any appreciable range of deformation, coordinates must be
arbitrarily chosen. The Road Research Laboratory (1952) uses the pressure
corresponding to 0.13 cm (.05 in) deflection, obtaining:
B-2
\/
----
I
t
q I
I
I
u-
ta 1 (b)
p
Load block
Stacked plates
,.
q"'q aod b.,.O
"If plate nol rlgrd
qL
A
Fig. B.Z. Determination of modulus of subgrade reaction using plate load test,
k =~ kg/cm3
0.13 (B.3)
The U.S Corps of Engineers uses the deformation corresponding to 0.7 kg/cm2 (10
psi) pressure for the evaluation of subgrade for airfields, giving :
(BA)
The Navdocks Design Manual (1961) uses the pressure and corresponding deflection
at one-half the yield pressure. The data can be plotted to a log-log scale and where
tangents of two straight-line parts of the resulting curve intersect can be taken as the
yield point.
California Bearing Ratio (CBR) tests have also been used to obtain the subgrade
modulus. The penetration-load at 0.25 cm (0.1 in) displacement, converted to
pressure based on a piston diameter of 4.95 cm (qCBR),yields:
k= qCER
025 (B.5)
(B.6)
where
k', = kB (forcellength2),
B-3
v = Poisson's ratio,
EFIF = Flexural rigidity of footing.
Observation has shown that ill equation (B.6) .the term 0.65 12 EsB4 is
EFIF
approximately equal to 0.9 to 1.5, with an average value for footing of modest
proportion of about 1.2. Thus
k's."'12 Es 2 (B.7)
1- v
(i) Thickness, t of mat is calculated from critical punching shear requirements as per
ACI (1983).
B-4
(iii) Flexural rigidity, D of the mat is calculated from Equation C.2.
(B.8)
where
3
k = modulus of subgrade react{~n in kip/ft ,
For the case where the spacing of column lines in two orthogonal directions are
different, the average of the two spacing should be taken for Be.
(v) The radius of effective stiffuess, L is calculated from Equation C.3. The distance
4L is called radius of influence because the response due to a given load dampens
very quickly after this distance.
B-5
in Equation B.8 or R9.
(b) When column spacing in both directions are smaller than 4L, the formula
for rectangular plates is used, which is
k - k m+O.5
(RIO)
- vI l.5m
where m = Bo] I B02 and Bo] and Be2 are column spacing in orthogonal directions.
Shukla (1984) also presented a chart (Fig. B.3) giving values of modulus of subgrade
reaction based on average results obtained by applying load through a 1 ft2 bearing
plate. Using this chart, values of kv1 can be found if unconfined compressive strength
(qu) of clayey soil or relative density of sandy soil is known.
Bowles (1988) suggested the following equations for approximating k from the
allowable bearing capacity,
where
B-6
Unconfined compressive strength (tsf)
o 1 2 3
350 4
veryl Soft Stiff
soft Clay Very
clay
clay stiff
clay
300
250
--- Fine grained soil
-- Coarse grained soil
200
150
100
50
Very
Medium
loose Loose Very
dense
sand sand Dense dense
sand
sand sand
0
0 10 20 30 40 50 60 70 80 90 100
Relative density (%)
Bowles (1988) also presented a table (Table B.l) showing ranges of values of k for
different types of soil. Table B. I may be used to estimate a value of k to determine
the correct order of magnitude of the modulus of subgrade reaction obtained using
one of the above approximate equations.
B-7
APPENDIXC
,
AVAILABLE METHODS OF MAT FOUNDA nON ANALYSIS
C.l GENERAL
Conventional method is the crudest one. Here mat is treated as a rigid slab. All the
other methods are based on Winkler spring model. ACI method is based on
SeWeicher' s solution of infinite flexible slab on continuous spring support and
Baker's method is designed to emulate that solution. Finite Grid and Finite Difference
methods are numerical analysis methods and require extensive computer support
though they do not offer any substantial improvement in results over ScWeicher's
solution. These methods are discussed in some detail in this chapter.
In this method mat is treated as infinitely rigid giving planer soil pressure distribution.
The mat is divided in both directions into strips centered on the column lines and
having widths equal to half the spacing of column lines on both sides of the column
lines. Each strip is loaded by column loads and supported by soil pressure.
C-I
The procedure for the conventional analysis consists of the following steps:
(i) The resultant of all loads acting on the mat and its location is determined.
(ii) The slab is divided into strips in x and y directions, one direction at a time, half
way in-between the column lines. Each strip is assumed to act as an independent
beam supported by soil pressure and acted upon by column loads. Steps (iii) to (vi)
are applicable to each of these strips.
(iii) Soil pressure is calci.I1atedat the four comers of the strip by treating the mat as
fully rigid and their average is obtained. This time a check is made that the maximum
soil pressure is less than the allowable bearing capacity.
(iv) Total soil reaction under the column strip and the column strip loads are first
averaged to obtain the average force acting on the column strip. Each of the column
forces and the average soil pressure are then modified in proportion to this average
force in order to enforce static force equilibrium.
(v) To ensure moment equilibrium, the modified average soil pressure is then shaped
as a trapezoid so that the resultants of the soil pressure and the column loads meet at
the same point.
(vi) Shear forces and bending moments are calculated at various points of the strip
for the modified column loads and soil pressure.
Arora (1992) proposes that as the analysis is approximate, the actual reinforcement
provided should be twice the computed values.
According to the ACI Committee 336 (1966), the design of mats may be
accomplished using the Conventional method if:
C-2
(i) The average of the two adjacent spans in a continuous strip is less than I.75/'A
(Equation CII) and adjacent column loads and column spacings do not vary by more
than 20 percent of the respective greater value.
(ii) The relative stiffuess factor Kr is found to be greater than 0.5, where
K = EIB (CI)
r 3
EsBF
where
Though quite suitable for hand calculation, this method can not give the variational
pattern of the internal forces of mat. Also column base moments can not be
incorporated in the analysis. Besides, since mat is divided into column strips from the
very beginning of the analysis in Conventional method, its plate characteristics are not
reflected in the analysis.
ACI Committee 336 (1966) suggested this method for the general case of a flexible
mat supporting columns at random locations with varying intensities of loads. This
procedure is essentially based on Schleicher's solution of infinite plate on Winkler
medium. Shukla (1984) provided charts for prompt calculation of moments and
shears following this method.
C-3
therefore, possible to consider mat as a plate of infinite dimension and determine the
effect of a column load in a specific region, called the zone of influence, surrounding
the load. The total effect of all the column loads at any point can thus be determined
by superimposing the effect of all the column loads, within whose zones of influence
the point lies. If moments and shears are found along the edges then these are later
applied in opposite directions at the same locations and with a semi-infinite beam
analysis their effects inside the mat are calculated and superimposed on the previous
solution to have the final moments and shears. The later part of the analysis is called
end correction.
In this method a problem can systematically be solved through the following steps:
Et3
D=--- (C.2)
12(l-v2)
where
(ii) Radius of influence of any column load is four times the radius of effective
stiffness, L around the column. L is calculated as follows:
(C.3)
where
C-4
(iii) Radial and tangential moments, shear force and deflection at a point are
calculated using the following formulae:
P r Z;(~)
= - "4[Z4 CC) - (1- v) ~ ) (C.4)
L
P r Z3(~)
M, = --[vZ4 (-) + (1-v)--) (CS)
4 L r
L
(C6)
(C 7)
where
P = concentrated column load,
r = radial distance of the point under investigation from P,
= radial and tangential moments per unit width of mat,
Q = shear force per unit width of mat at a radial section,
w = mat displacement,
z's = functions first introduced by ScWeicher (1926).
The Z functions have the characteristic features of exponential waves and their
variation W.Lt. r/L is shown in Fig. C.l.
(iv) The radial and tangential moments are then converted into bending moments in
Cartesian coordinate system (Fig. C.2), whereas shear forces and displacements in the
Cartesian coordinate system remain the same.
(v) For the resultant effect of all the column loads, radii of influence of whose overlap
at points of interest, the moments, shears and displacements due to individual
columns are superimposed.
C-s
0 5 6
0 0
~ X
-0.1
-0.2
-0.3
-0.4 Z4{x)
-10.0 -0.5
Fig. C.l. Variation ofZ functions with respect to x = r/L [After Hetenti (1946)).
Mr
M" Oy
y
Lox
MT
My
i ..
~
x
The mat is divided into strips of unit width in both x and y directions.
Assuming the strips as semi-infinite beams, shears and moments equal and
opposite to those obtained in the previous analysis at the edges are applied at
the respective positions and their effects at various points are superimposed
on the respective values obtained earlier.
For moments, shears and deflections of a semi infinite beam, the following
relationships are used :
(C.8)
(C9)
2M /.,2
wc=- ke Ch+SX (ClO)
/.,=4[kB (Cll)
~rn::
where
C-6
" .. -.
r = distance of the point under consideration from the end
conditioning force i.e. Me or Ve .
Shukla (1984) has presented the infinite plate solution portion of the ACI method in a
more convenient form. This is given below.
M, = Crne P (CI2)
M, =Cm,P (C 13)
All the symbols in Equation C.12 through Equation C.13 have already been
explained. It is clear that with the graphs of C"" and Cm, given (Fig. C.4), M, and M,
can readily be calculated.
An approximate method of analysis which can take into account the variation of soil
properties across a building site without unduly complicating the computational
effort, in comparison to that needed for uniform soil property condition, was
proposed by Baker in 1948. It is simple and straight forward in application, giving
results which compare well with the Winkler analysis in comparable situations. In this
method, a specific shape of soil pressure distribution of unknown magnitude is
assumed. The soil pressure is applied to the strip of mat, consisting of a single
column line and cut separate along the middle lines of adjacent column spacing on its
C-7
1.0
-AA'
0.8 -BA,
-4- CAr
---?- DAr
0.6
t.
0 "
u
c 0.4
III
:;,
()
t.
m" 0.2
t.
"
<l;
0.0
-0.2
0 1 2 3 4 5 6
Ar
Fig. C.3. Variation of A)." B " CAr and DA, with Ar [After Hetenyi(1946)].
A
0.30
0.25 -Cm
-Cm
0.20
E 0.15
()
u
c
III 0.10
E
()
0.05
0.00
-0.05
0 1 2 3 4
x=r/L
Fig. CA. Variation of Cmr and Cmt with x = r / L [After Shukla (1984)].
both sides, and the maximum relative displacement of the beam is expressed as a
function of the pressure magnitude parameter, which is called the beam equation. The
same pressure distribution is applied to. the soil and the maximum differential
settlement is related to the pressure magnitude, which is designated as soil equation.
Simultaneous solution of these two equations gives the magnitude of the pressure and
the relative settlement. Values of bending moments can be calculated by adding the
moments developed by column loads and the differential settlement.
C.4.1 SystemofForces
Individual column strips, both in x and y directions, are first isolated from the mat.
These strips are analyzed separately treating them as single foundation beams. The
isolated beam is loaded by a central line of column loads, Pi where i ranges from 1 to
the number of columns in the column strip under consideration. The column forces
are modified by the reactions of the transverse beams at each of the intersections. The
column forces Pi are balanced by a varying distribution of soil pressure p acting on
the underside of the beam (Fig. c.5).
To solve the problem, Baker divided it up into several subsidiary problems, each
consisting of a force system in equilibrium.
In the first of these (Fig. C.6), the column bases are held at the same level and the
foundation beam is acted upon by a uniform upward soil pressure pe and supported
by the columns. Here, pe is the average soil pressure obtained by dividing the total
building load by the total bearing area. The column loads Fi of this system is
calculated as the reactions to pe. This force system gives rise to bending moments MI
and relative displacements WI .
C-8
P,
Average
soil pressure Pe Soil
reaction
p
+ve Moments
I
+ I Bending
mome.nt
-I M
I
Settlement
I
I - Maximum
r
_ settlement
Fig. C.5. Baker's method of mat foundation analysis : (a) typical load diagram, (b)
typical bending moment diagram and (c) typical deflection diagram of an
isolated column strip.
F,
Uniform
t soil reaction
p.
Bending
moment
M,
Fig. C.6. Force system I : (a) load diagram; (b) bending moment diagram and (c)
deflection diagram.
The second equilibrium force system (Fig. C.7) consists of the difference in forces Pi
minus F; acting at each column location on the beam. Displacements and bending
moments of this force system are designated as W2 and M2 respectively.
Finally, the remaining force system 3 (Fig. C.S) consists of the total soil reaction
pressure p less the uniform component of the pressure pe . This non-uniform pressure
causes relative displacements W3 and moments M,.
The total forces and moments acting on the beam are given by the sum of the
separate forces and moments of the force systems 1, 2 and 3.
The forces, moments and deflections of systems 1 and 2 can all be obtained by the
normal methods of structural mechanics. A simplitying assumption made by Baker for
analyzing force system 3 is that the soil reaction can be represented by a triangular
distribution. In case of cohesive soil, the pressure acting at the slab edges is expected
to be higher than that in the interior and in sand, however, the edge pressure is found
to be lower than that at the center. Based on this fact, Baker proposed linearly
varying soil pressure distributions characterized by the magnitude parameter n, which
may be positive or negative, as shown in Fig. C.9.
From the triangular soil pressure distribution, the maximum displacement of system 3
IS
(C.14)
C-9
P, - F,
Resultant
column
loads
+ Bending
moment
M2 (--vel
Fig. C.7. Force system 2: (a) load diagram; (b) bending moment diagram and (c)
deflection diagram.
Resultant
soil
reaction
np,
Bending
moment
MJ
-------- --------- ------- Relative
displacement
wJ
Fig. e.8. Force system 3: (a) load diagram; (b) bending moment diagram and (c)
deflection diagram.
~-------
- Soil
(a)
--------
Beam
Soil
(b)
Fig.C.9. Triangular soil pressure distribution for Baker's method : (a) positive
distribution for cohesive soil and (b) negative distribution for
cohensionless soil.
The numerical coefficient in Equation C.14 is equal to 0.00365. Considering different
realistic forms of pressure distribution that can act on the beam under various
conditions, Baker concluded that a conservative value to use would be 0.0035.
I 2
M3 = -nBPe! (C.I5)
24
where! and B are the length and width of the beam respectively and EI is its flexural
rigidity.
The system I deflection may be neglected and so the maximum relative deflection of
the beam, w can be expressed as the sum ofw2 and W3 . This expression ofw is called
the beam equation.
If, for a simple case, the soil behavior is considered to correspond exactly to that of a
uniform Winkler material with subgrade reaction coefficient k, then the downward
deflections at the ends is equal to np/k and the upward deflection at the center is
equal to - np/k. The maximum relative soil deflection is therefore
w= (C.16)
C-1O
However, if the soil has a subgrade reaction coefficient k at the beam ends and sk at
the beam center, the total displacement at the beam center is P e (I - n) . At the
sk
Pe(l-n) Pe(l+n)
w=--------- (C.I?)
sk k
Equation C.16 or C.I?, whichever is applicable to the problem at hand, is called the
soil equation.
The beam and soil equations can be solved simultaneously to give values of the
pressure variation parameter n and the maximum relative settlement w at which the
loads and deflections of the beam and soil are compatible in the limited sense of this
analysis.
The principal influence on the distribution of soil pressure in system 3 is the relative
displacement of the beam caused by the balanced load system 2. Thus, with the
approximation adopted, Baker showed that the bending moment diagram developed
by the soil pressure of system 3 is similar in shape to M2 but is opposite in sign and of
smaller magnitude. Because of the small deflections induced by the first set of loads,
their bending moments remain substantially unchanged by the soil reaction.
If it is assumed that the two bending moment diagrams of M2 and M3 are exactly
similar and of opposite sign and that, moreover, the bending moments are
C-ll
proportional to the maximum relative deflection of the beam, then the total bending
moment M (= Mr + M2 + M3 ) can be expressed as follows:
(C.18)
In Baker's method, the detail shape of the deflected beam does not conform to the
undulations of the displaced soil since only the maximum differential deflection is
taken as the matching criterion. This method can be applied for axial column loads
only and accuracy of results obtained by this method is not rewarding enough
compared to the computational effort.
Deryck, Severn (1960,1961) and Bowles (1974) have used the finite difference
technique to analyze large thin flat slabs on elastic medium. The governing equation
of Finite Difference Method is :
(C.19)
where
w = deflection,
p = subgrade reaction per unit area of mat.
This equation can be transformed into a finite difference equation when r = 1 (Fig.
C.I0) and is given below.
C-12
20w o -Sew T
+w B
+w R
+w L
)+2(w TL
+w TR
+w BL
+w BR
)
h4 Ph2
+(w +w +w +w )=L+ __
.11 B8 LL RR D D
(C.20)
where
6 8 4 4 4 2
(-+-+6)w
4
+(---o)(w +w )+(---4)(w +w ) +-(w
r r2 0 r4 r.... L R r2 T B r2 TI-
l
+w TR
+w BL
+w BR
)+w IT
+w BB
+4(w LL
+w RR
)
r
prh4 Ph2
=--+--
rD rD
(C.2I)
p = -kwo (C.22)
All the notations used in Equation C.20 through Equation C.22 are explained in Fig.
C.IO.
For a given mat, one difference equation can be written for each point of intersection
of the finite difference grid. By solving these simultaneous equations, the deflections
at all points can be determined. Consequently, soil reactions, bending moments and
shear forces can be found.
This solution is applicable to mat when the plan dimensions of mat are reasonably
large in comparison to its thickness. Then the error due to higher plate thickness is
C-13
very small. Though more theoretical in nature, column base moments can not be
handled by this method and boundary conditions are difficult to represent. Also the
solution is inefficient in terms of time consumption.
In the Finite Grid method of analysis, the entire mat is transformed into a suitable grid
consisting of beam-column elements each of which is supported by soil springs at the
ends. Nodal displacements are considered to be the basic unknowns. Stiffness matrix
and load matrix for each element are built and assembled into a global stiffness matrix
and global load matrix respectively for the entire mat resting on Winkler medium.
When the global stiffness and the load matrices are known, the nodal deflection
matrix can readily be obtained by solution of simultaneous equations and hence the
member forces can be calculated.
where
{P} = external nodal force vector,
{F} = internal member force vector,
[A] = bridging constant matrix.
An equation relating internal member deformation "e" at any node to the external
nodal displacement "X" in matrix notation is given by
(C24)
C-14
The internal-member forces {F) are related to the internal member displacements as
Now that only unknowns in Equation C26 are the X's, with the values ofX's known,
the internal-member forces, which are necessary for design, can be obtained from:
The [A][S][A]T matrix is the global stiffuess matrix and it is built from one element
of the structure at a time and using superposition. The [A] and [S] matrices for an
element are as follows:
-smet 0 - COSet
COSet 0 0
(08)
l/Lg l/Lg 0
[A] =
0 -smet COSet
0 COSet smet
-l/Lg -l/Lg 0
4EI 2EI
- - 0
and Lg Lg
(C29)
2EI 4EI
[S] = - - 0
Lg Lg
QGJ
0 0 --
Lg
C-15
Here a is the angle of orientation of the grid element and G is the shear modulus of
mat material. The torsional factor J should be computed for a rectangular section.
The adjustment factor n in Equation C.29 is used to make the solution better fit the
theoretical solutions as found in plate theory analysis. At present the n factor for a
best fit is
(C.30)
where
The element stiffuess matrix is a 6x6 matrix. As has been discussed earlier, soil media
is considered as springs in this method. These springs should be included in. the
element stiffuess matrix. To accomplish this purpose, two spring stiffuesses, one for
each end of the element, , are inserted in the (3,3) and (6,6) position of the matrix.
Thus, in this approach the soil spring constants appear only in the diagonal terms and
when the mat at any point separates from the soil, the spring constant at that point is
just made zero.
Though the method is capable of handling column base moments and varying soil
properties, the plate characteristics of mat is lost when it is modeled as grid and this
makes the Finite Grid method unrealistic.
C-16
,
-+-- --
IW WT
----t-
,WTR
, TL
,
r" "
rh
I
I rh
IWL Wo ,wR
'" r"
-+-----
I "---r" --t-'",
IWSL W s ,WSR
" WSS
L
Nodal P- X
Element F-e
Fig. C.!l. External (nodal) and internal (member) finite grid forces.
.'.
APPENDIXD
D.I GENERAL
The mat analysis software developed in this study is very straight forward in the
interactive mode of data input. It actually leads the user in the right direction during
data input and asks for data only relevant to a particular type of problem. There is,
however, option for file mode input which uses the similar sequence of data as the
interactive mode. Only interactive data input mode is discussed here. Since the
program has various options and fine-tuning features only the most vital options are
covered by this discussion. For file mode data input, it is suggested that the software
be tentatively run once and the sequence of data items be noted to prepare data file.
Before using the software, the user should draw a plan of the mat to be analyzed with
the column arrangement. Then user must draw X and Y directional straight lines on
the mat plan to form a grid system. This will be -the finite element mesh of the mat.
The X and Y axes must be along the bottom and left edges of the mat respectively.
Columns must be represented by single elements. In making the finite element mesh,
finer elements may be taken around the columns. Elements under the columns may be
made rigid by increasing their modulus of elasticity. This can be done in two ways :
by specifYing the degree of magnification of their modulus of elasticity directly or by
specifYing the equivalent thickness of the column elements. It is also possible to
distribute the soil spring stiffuess under the elements to their nodes in two ways : by
allowing middle nodes to take twice as much stiffuess as comer nodes (one-twelfth
option) or by lumping thrice as much stiffuess at the middle nodes as at the comer
nodes (one-sixteenth option). It is also possible to analyze a problem using nodal or
Gauss point stresses.
D-I
D.2 USER INSTRUCTIONS
A typical sequence of data input for mat foundation analysis is described below.
I
I) Questions are typed in capital letters and the users' response in normal text style.
If the answer is Y then a file named MAT.DAT must be present with necessary data
otherwise the program will be terminated. If the answer is N the program will
proceed in the interactive mode.
(2) ANALYZEPARTOFTHEMAT?(YIN)
When the problem at hand is symmetric this question may be answered Y and the
next question will be 4. Otherwise the answer will be N and question 5 will be
asked.
(4) HALFPORTIONANALYSIS?(YIN).
The answer must be N for quarter portion analysis and questions 10.2.1 and 10.2.2
will be asked. For a "Y" answer question 10.1 will be asked.
i
(5) X DIMENSION (IN FT)
j
The user will enter X dimension of the portion (quarterlhalfi'full) of the mat to be
analyzed in response.
(6) YDIMENSION(INFT)
The user will enter Y dimension of the portion (quarterlhalfi'full) of the mat to be
analyzed in response.
D-2
(7) THICKNESS AWAY FROM THE COLUMNS (IN FT)
The user will enter the smaller thickness of mat in case of mat with non-uniform
thickness. For mat with uniform thickness, mat thickness should be entered.
Answer to this question is the number of the X directional divisions of the finite
, if
element mesh.
"
1\:
I ,
\ ,.,
II Answer to this question is the number of the Y directional divisions of the finite
element mesh. The following questions will be asked only if answer to question 2 is
., y.
User should specifY which boundary is the centerline of the mat, half of which is
being analyzed.
User specifies whether or not the top boundary represents the center line of the
quarter of the mat being analyzed.
User specifies whether or not the right boundary represents the center line of the
quarter of the mat being analyzed.
D-3
(II) MODULUS OF ELASTICITY DIRECTLY ? (YIN)
If the user wants to calculate modulus of elasticity from concrete strength, answer
should be N. If answer to question II is Y question 11.1 will be asked, otherwise
questions 11.2 will follow.
(11.1) MODULUS OF ELASTICITY (KSF)
The user should respond with the value of Poison's ratio of the mat material.
If element springs are to be lumped at the nodes by 16th part then answer will be N.
D-4
If nodal stresses are to be used for moment and shear calculation then the user should
respond with N.
Answer will be the number of columns. The following question will be repeated for
each column in the portion of the mat (quarterlhalfi'full) being analyzed.
The user should serially enter: no. of the X directional division of the finite element
mesh just left to the column, no. of the Y directional division of the finite element
mesh just below the column, column axial load, column base moment about X axis
and column base moment about Y axis.
If portions of mat under the columns are to be made more rigid than the rest of the
mat, then answer will be Y. Oilly for a Y answer question 22 will be asked.
D-5
(22) COLUMN ELEMENT THICKNESS DIRECTLY? (YIN)
If mat with non-uniform thickness with greater thickness only Under the columns is to
be analyzed then answer will be Y and only then questions 24 to 27 will be asked.
If thickness under different columns are different, answer will be N. Questions 25 and
27 will be repeated for each column.
D-6
Answer will be Y if width of greater thickness is the same for all columns.
(27) LEFT DIV, RIGHT DIV, BOTTOM DIV AND TOP DIV OF RANGES
User should input the number of left, right, bottom and top divisions of the finite
element mesh covering the width of greater thickness on the respective sides of
column.
User will input the total number of X directional divisions of the finite element mesh
with greater thickness.
User will input the serial numbers of X directional divisions of the finite element mesh
with greater thickness.
User will input the total number of Y directional divisions of the finite element mesh
with greater thickness.
D-7
(33) NUMBER OF Y DIVISIONS OF GREATER TIllCKNESS
User will input the serial numbers ofY directional divisions of the finite element mesh
with greater thickness.
The total number of X and Y directional column strips in the part of the mat
(quarter/hal£'full) should be entered.
D-8
Answer will be Y or N depending on whether detailed output is desired or not.
This question will be repeated for each Y directional division. Answer will be the X
coordinate of the right face of the Y directional division of the finite element mesh
W.r.t. to the bottom-left comer of the mat
This question will be repeated for each X directional division. Answer will be the Y
coordinate of the top face of the X directional division of the finite element mesh
W.r.t. to the bottom-left comer of the mat
The total number of X directional lines of the finite element mesh for which graph
files for displacement, moment and shear are desired should be entered. Question 44
will be repeated for the each such line.
User will input the serial no. of the X directional line of the finite element mesh for
which graph files are desired.
D-9
(45) GRAPHS FOR HOW MANY Y LINES?
The total number of Y directional lines of the finite element mesh for which graph
files for displacement, moment and shear are desired should be entered. Question 46
will be repeated for the each such line.
User will input the serial no. of the Y directional line of the finite element mesh for
which graph files are desired.
If there has been no error in the data input or there is no correction in data, answer
will be Y and the program will start execution. Otherwise answer will be N and data
input session will begin once again.
D-IO
APPENDIXE
ADDITIONAL FIGURES
50
•
~
:'£
-c
Q)
E
0
o
:2
-50
-100
o 5 10 15 20 25 30
X (ft)
(a)
50
CF CF
~
-
...:::
~
E
c
Q)
0
o
:2
-50
-100
o 5 10 15 20 25 30
X (ft)
(b)
Fig. E.1. Comparison of Gauss point and nodal stress analysis in bending
moment calculation (Example mat of Fig. 3.4, mat thickness 3 ft,
CF = column face).
E-l
80 CF
-- Gauss point stresses
60 -- Nodal stresses
CF
40
20
2'
6
~ 0
t1l
Q)
.<:
(f) -20
-40
-60 CF
-80
CF
0 5 10 15 20 25 30
X (ft)
(a) ,
CF
50
~
<:=
6
~
t1l 0
Q)
.<:
(f)
-50
CF
-100 CF
0 5 10 15 20 25 30
X (ft)
(b)
Fig. E.2. Comparision of Gauss point and nodal stress analysis in shear
force calculation (Example mat of Fig. 3.4, mat thickness 3 ft,
CF = column face).
-0.020
~ -0.022
:s
c
o
u
.2?
Q)
-0.024
o
-0.026
-0.016
g
c
-0.018
o
U
Q)
;;::
Q)
o -0.020
-0.022
---- Twelveth part distribution
- Sixteenth part distribution
o 10 20 30
X (ft)
(b)
Fig. E.3. Effect of nodal spring distribution on the deflection of (a) LINE 1
and (b) LINE 2 of the Example mat (Fig. 3.4, mat thickness 3 ft).
100
CF CF
50
CF
2
-.,;,
:;:,
-
~
c:
CD
E
0
0
:2
-50
(a)
CF CF
100
50 CF
2
~,
6
C 0
CD
E
0
:2
-50
Fig. EA. Effect of nodal spring distribution on bending moments of (a) LINE 1
and (b) LINE 2 of the example mat (Fig. 3A, mat thickness 3 ft, CF =
column face).
80
CF
60
40 F
20
2'
::>2
~
~ 0
ro
Ql
.c -20
(f)
-40
0 5 10 15 20 25 30
X (ft)
(a)
100 CF
50 CF
2'
::>2
~
~
ro 0
Ql
.c
(f)
Fig. E.5. Effect of nodal spring distribution on shear forces of (a) LINE 1
and (b) LINE 2 of the example mat (Fig. 3.4, mat thickness 3 ft,
eF =
column face).
0.005
0.000
-0.005
Only column 1 load
2
~ - Only column 2 load
c
0 -0.010 -.to- Only column 3 load
U(]J
---y- Only column 4 load
0;::
(]J - All column loads
0 -0.015 --+-- Superimposition of
indivisual column
-0.020
-0.025
0 10 20 30
X (tt)
(a)
0.005
0.000
-0.020
o 10 20 30
X (tt)
(b)
Fig. E.6. Indivisual and combined effect of all column loads on deflection
of (a) LINE 1 and (b) LINE 2 (fig. 3.4, mat thickness 3 tt).
200
- Only column 1 load
CF
- Only column 2 load CF
150 --Ao- Only column 3 load
--.- Only column 4 load
- All column loads
100 -+- Superimposition of indivisual
column loads
~
-
~
~
c
(])
E
50
CF
CF
0
:2 0
-50
-100
o 5 10 15 20 25 30
X (ft)
(a)
200
- Only column 1 load CF CF
- Only column 2 load
150 --Ao- Only column 3 load
--.- Only column 4 load
- All column loads
100 -+- Superimposition of indivisual
column loads
~
~
-
~
c
(])
E
50
o
:2 0
-50
-100
o 5 . 10 15 20 25 30
X (ft)
(b)
Fig. E.? Indivisual and combined effect of all column loads on bending
moments of (a) LINE 1 and (b) LINE 2 (Fig. 3.4, mat thickness 3
ft, CF = column face).
100
•... o
I1l
<ll
.c
(f) -20
-40
-60
CF
-80
CF
o 5 10
15 20
25 30
X (ft)
(a)
100
- Only column 1 load
- Only column 2 load CF
80
--A-- Only column 3 load
60 -- Only column 4 load
CF --+- All column loads
40
---+--- Superimposition of indivisual
20 column loads
•... o
I1l
J?
(f)
-20
-40
-60
-80 CF
-100 cr
o 5 10 15 20 25 30
X (ft)
(b)
Fig. E.8. Indivisual and combined effect of all column loads on shear forces
of (a) LINE 1 and (b) LINE 2 (Fig. 3.4, mat thickness
face).
3 ft, CF = column
-0.014
-0.016
-0.018
~ .0.020
o
U
Ql
;;:
Ql
o
-0.022
Fig. E.9. Deflection diagrams at different locations of the example mat of Fig. 3.4
(Mat thickness 3 ft).
- Edge of column strip 1
40 CF
CF
~
--
~
6
C
,
20
Ql
E.
0
~ 0
-20
-40
-60
-80
o 5 10 15 20 25 30
X (ft)
-
~
c
OJ
E 0
o
:2
-50
-100
o 5 10 15 20 25 30
Y (ft)
20
~
<:::
:;g 0
~
'"
Q)
.c
en
-20
-40
-60
CF
-80
o 5 10 15 20 25 30
X (ft)
Fig. E.12. Shear force diagrams at different locations of the example mat of Fig. 3.4
(Mat thickness 3ft, CF = column face).
100
20
-
Z'
:~
><
~ 0
III
<1l
..c:
en
-20
-40
-60
-80
-100
o 5 10 15 25 30
Y (ft)
50
CF CF
-50
o 5 10 15 20 25 30
X (ft)
Fig. E.14. Average bending moment diagrams of column strip 1 of the example
mat of Fig. 3.4 for different averaging width (Mat thickness = 3 ft, CF =
column face).
- Averaging width = 1.00 ft CF
CF
--- Averaging width = 4.00 ft
100 --- Averaging width = 8.00 ft
-- Averaging width = 12.00 ft
--- Averaging width = 16.00 ft
-+-- Averaging width = 20.00 ft
50
CF
2"
~, CF
6
C
(])
E
0
:2
0
0
t,,
-50
o 5 10 15 20 25 30
X (ft)
Fig. E.15. Average bending moment diagrams of column strip 2 of the example
mat of Fig. 3.4 for different averaging width (Mat thickness 3 ft, CF =
column face).
- Averaging width = 1.00 ft
80
- Averaging width = 3.50 ft
~ Averaging width = 6.00 ft CF
---,-y- Averaging width = 8.00 ft
40
CF
20
~
~
~ 0
III
Ql
.r::
en
-20
-40
-60 CF
-80 CF
o 5 10 15 20 25 30
X (ft)
Fig. E.16. Average shear force diagrams of column strip 1 of the example
mat of Fig. 3.4 for different averaging width (mat thickness = 3 ft,
eF = column face).
- Averaging width = 1.00 ft
-- Averaging width = 4.00 ft
100 -...- Averaging width = 8.00 ft
--T-- Averaging width = 12.00 ft
-- Averaging width = 16.00 ft CF
--+- Averaging width = 20.00 ft
50 CF
-50
CF
CF
-100
o 5 10 15 20 25 30
X (ft)
Fig. E.17. Average shear force diagrams of column strip 2 of the example-
mat of Fig. 3.4 for different averaging width (mat thickness 3 ft,
eF = column face).
- Column spacing = 15 ft
-- Column spacing = 20 ft
-A-- Column spacing = 25 ft
(a)
160
~
~.
140
.",~•
-
c:
Ql
E
120
- Column spacing = 15 ft
-- Column spacing = 20 ft
0
E -A-- Column spacing = 25 ft
Ql
> 100
:;::;
.in
0
a. 80
E
~
E
.x 60
III
:2:
40 • • •
• •
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
dt(tt)
(b)
Fig. E.18. Relationshipof (a) positive moment and (b) negative moment with
dt for 15 story-3bay building under gravity loading only (t = mat
thickness).
~
c 1.00
Ql
E
0 - Column spacing = 15 fl
E - Column spacing = 20 fl .
Ql 0.95
.~ ----.to-- Column spacing = 25 fl
iii
Ol
c
Ql 0.90
~
0
~~
E 0.85
.E
"E
..?
c
Ql 0.80
E
0
-
:2
C
Ql
E
0.75
0
:2 0.70
(a)
1.5
~
C
Ql
E
o
E 1.4
Ql
>
.0;
""
o
0. 1.3
~
.E~
E
.E
.E .1.2
..?
C
Ql
E
~ 1.1
-
~
C
Ql
-
-
Column spacing
Column spacing
----.to-- Column spacing
= 15 fl
= 20 fl
= 25 fl
E
o 1.0
:2
0.0 0.1 0.2 0.3 0.4 0.5
At / ~niform
(b)
Fig. E.19. Relationship of (a) positive moment and (b) negative moment with
At for 15 story-3bay building under gravity loading only in a non
-dimensional form (t " thickness of mat).
-E
c
Q)
1.08
1.10
0
E 1.06
Q)
>
:;::; 1.04
.00
0
C- 1.02
•...
.E
-E
oil
1.00
.", 0.98
...?
c 0.96
Q)
E
0
0.94
E = 15 It
-.-
:2
c
Q)
0.92
-
-
-A--
Column spacing
Column spacing
Column spacing
= 20 It
= 25 It
0.90
E
0
:2 0.88
0.0 0.1 0.2 0.3 0.4 0.5
M/ ~niform
(a)
C
Q)
1.00
E
0
E 0.95
Q)
.;::
0; 0.90
Ol
Q)
C
--
•...
0
E
0.85
~ 0.80
...?
c
Q)
E 0.75
0
-.-
:2
c 0.70
-
-
Column spacing
Column spacing
= 15 It
= 20 It
Q)
E
-A-- Column spacing = 25 It
0
:2 0.65
0.0 0.1 0.2 0.3 0.4 0.5
Llt I tuniform
(b)
Fig. E.21. Relationship of (a) positive moment and (b) negative moment with
M for 15 story-3bay building under gravity and wind loading in
a non-dimensional form (t = mat thickness).
1.05
30 40 50 60 70 80
Slope angle (degree)
(b)
Fig. E.22. Effect of slope angle on (a) kink shear and (b) kink shear magnification
for a 15 story 3 bay building under gravity loading only.
0.018 -
0.017 -
0.016 -
I •
I I • I •
0.15 I
0.16 0.17 0.18 0.19
Poison's ratio
Fig. E.23. Effect of Poison's ratio on column deflection (Mat thickness 3 ft).
110
y y y y y
100
90
• • • • •
~,
" 80 -- Column 1
-
"'~
c
Ql
E 70
--
----A-
Column 2
Column 3
0 ~ Column 4
E
Ql
60
- U
'"
c 50
E
::1
(5 40
U .6, .6, .6, .6,
30 •••
20 ~ • • • •
0.15 0.16 0.17 0.18 0.19
Poison's ratio
(a)
75
• • • • •
70
~
.6,
665
-
.6,
••• .6,
•••
c
Ql
E 60
0
-- In-between column Cl and C2 of LINE 1
E
-- Center of LINE 1
.~ 55 ----A- In-between column C3 and C4 of LINE 2
a; ~ Center of LINE 2
Ol
~ 50
E
::1
.S 45
x
'"
~
• • • • •
40
(b)
Fig. E.24.
Effect of Poison's ratio on (a) positive and (b) negative bending moment
(Mat thickness 3 ff).
20
- LIt = 1.00 It
CF
- M=0.751t
-- LIt = 0.50 It
----- LIt = 0.25 It
10
- LIt = 0.00 It
CF
~
~
•... 0
co
<Il
.r::.
(J)
-10
CF
-20
0 5 10 15 20 25 30
X (ft)
(a)
20
- = 1.00 It
LIt
CF
- M=0.751t
-- M=0.501t
10 ----- LIt = 0.25 It
- LIt = 0.00 It
CF
~
~
•... 0
co
<Il
.r::.
(J)
-10
CF
-20
0 5 10 15 20 25 30
X (ft)
(b)
Fig. E.26.
Shear force diagrams for different change in thickness, At of (a) column
strip 1 and (b) column strip 2 (MAT" of Fig. 4.18, CF = column face).
APPENDIXF
MISCELLANEOUS
F-I
Im!!. Used in Chapter Defined in Page
Punching Width 3 44
Slope Angle 4 70
Slope Width 4 65
Smaller Thickness 4 65
Strip I 4 45
Strip 2 4 45
Tentative Solution 5 101
Test Mat 4 56
Transition Zone 4 62
Weighted Average Steel Area 4 56
Weighted Average Thickness 4 59
Width of Greater Thickness 4 65
F-2