第 23 卷 第 6 期 岩石力学与工程学报 23(6)：1001～1006

2004 年 3 月 Chinese Journal of Rock Mechanics and Engineering March，2004

THICK PLATE ANALYSIS WITH

APPLICATION TO PILE CAP DESIGN
，2
Cheng Y M1，Zhang Y H1
(1Department of Civil and Structural Engineering，The Hong Kong Polytechnic University，Hong Kong，China)
(2Institute of Rock and Soil Mechanics，The Chinese Academy of Sciencess， Wuhan 430071 China)

Abstract The thick plate is commonly used for the pile cap and the transfer plate, and a proper analysis of it can greatly
affect the cost of the thick plate. In actual practice，soft support condition is sometimes adopted by engineers for modeling
of thick plates. In this paper，the limitations of soft support in the determination of deflection and bending moment with
very thick plates are studied. The distribution of superstructure loadings to pile caps and piles is also carefully
investigated.
Key words thick plate，pile cap，soft support
CLC number TU 472.99 Document code A Article ID 1000-6915(2004)06-1001-06

l is relatively large (t / l > 0.1) ，shear deformation

1 INTRODUCTION will become noticeable and the classical thin plate
equation as given Eq.(1) will underestimate the
，6]
For simple thin slab ， it can be analyzed as deformation of thick plate[5 . For thick plate，the
one-way or two-way slab which will be good enough Mindlin thick plate formulation is commonly used
for general design. For transfer plate or pile cap where and there are many types of thick plate element
～16]
the thickness/span ratio is usually greater than 0.1，the developed [6 . For the classical Mindlin thick plate
simplified analysis will not be adequate and a refined element，there are various numerical difficulties and
analysis is required. The proper technique to analyze a some of the important problems are：
thin slab structure is the plate analysis and the (1) Shear locking can occur for 1 point
[1～4]
governing equation for a Kirchhoff thin plate is integration scheme in Q4 bilinear element and
，7]
given by stabilization matrix has been proposed by Hughes [6
q for this problem.
∇ 4ω = (1)
D (2) For the classical Q4 bilinear element as
where q is the uniformly distributed load and D is shown in Fig.1，the author discovered that if a point
given by Et / 12(1 − μ ) . In Eq.(1)，that plane section
3 2
load is applied at point A while points B，C，D，E are
remains plane is used and the shear deformation is restrained in vertical direction，the solution of the
neglected. The solution of Eq.(1) for simple geometry vertical displacement by Q4 element is acceptable
is given by Fourier series or Navier′s series solution. If while the solution of the rotations will experience
the thickness t of a plate relative to the span length shear locking phenomenon for 1 ， 4 or 9 points

Received 19 Sep. 2003.

Cheng Y M：Male，Born in 1959，PhD，Associate Professor，Department of Civil and Structural Engineering，The Hong Kong Polytechnic University，Hong
Kong，China. E-mail：[email protected].
• 1002 • 岩石力学与工程学报 2004 年

integration scheme. That is to say，only the vertical analysis，the only variable in the governing equation is
displacement can be determined for this situation. The the vertical deflection and rotations are given by
authors have also discovered that shear locking for this simple derivative of the deflection. For thick plate，the
case can be avoided by simply adding extra grid lines rotations are formulated as independent variables，
as shown by the dotted lines in Fig.1. hence the treatment of boundary condition has to
be considered carefully. In actual practice ， some
engineers adopt soft support condition in the analysis
of thick plate. Soft support condition is acceptable for
thin plate but is not applicable for thick plate analysis.
The author has compared this practice with some
published rigorous solution and the thin/thick plate
program PLATE developed by the author. There are
Fig.1 Shear locking in Q4 element
several major limitations of the soft support condition
as following.
(3) Q8 quadratic element was also used for
(1) The bending moment and deflection are
improved accuracy and the main problem of this
usually over-estimated and the situation becomes
element is shown in Fig.2. If point A and point C are
serious if t / l approaches 0.3.
restrained while point B is not restrained ， shear
(2) There are great errors in the determination of
locking will occur. That is to say，any restraint to Q8
shear force.
element can only be applied to a single corner or the
side constituted by ABC. More care is required in the
design of a suitable mesh for complicated real problem
2 STUDY ON THE BENDING
if Q8 element is used. Q8 element is hence more
MOMENT AND DEFLECTION
popular in research but is less popular in ordinary
OF THIN/THICK PLATE
analysis and design.
In the present study ， a square plate simply
A supported on four sides and loaded with UDL is
considered. Any plate element that is adopted for
analysis should be able to pass through such a simple
test with satisfaction. The author would like to
emphasize that for a line support along edge， the
rotation along the line should be restrained to 0 as well.
B
The numerical results by IB element with soft support
condition (using a commercial program)， PLATE and
the rigorous solutions are all based on this condition.
The maximum deflection and moment are located
at the center of the plate and the results of analysis are
C shown in Fig.3 and 4. The deflection is normalized
Fig.2 Shear locking in Q8 element with the term ql 4 / 100 D while the bending moment
is normalized with the term ql 2 / 10 and the
Since the classical Mindlin thick plate element is coefficients are used for comparisons. For thick plate
less satisfactory for thick plate analysis of complicated analysis，the deflection coefficient is a function of t / l
problems，new 3 nodes or 4 nodes elements which can while it is a constant for thin plate solution. From
provide higher accuracy in analysis are still under Fig.3，it is observed that the error by IB element is
[6，7，10，11]
active development at present . For thin plate generally small for t / l < 0.1 but increases rapidly
第 23 卷 第6期 Cheng Y M，et al. Thick Plate Analysis with Application to Pile Cap Design • 1003 •

beyond this ratio. The results by PLATE are very close requires differentiation of the deflection. For an
to the rigorous solution for very thin to very thick accurate determination of the bending moment，a finer
condition. It is also noticed that the results by 8× 8 mesh is required as compared with that for deflection
elements are very close to that by 128× 128 elements determination. The errors by soft support for a mesh of
for both IB element with soft support and PLATE， 8× 8 are relatively large and some improvements can
which indicates that a relatively coarse mesh is be achieved by using a finer mesh of 128×128
sufficient for analysis if only the deflection is elements. For t / l less than 0.1 ， the error in the
required. moment coefficient by soft support using a mesh of
128×128 is small and acceptable but this error
increases rapidly with t / l . In fact，for t / l = 0.3，soft
support over-estimates the bending moment by 12%
and this ratio cannot be further reduced even with
mesh refinement!
The performance of soft support in determining
the bending moment and deflection with mesh
refinement is another interesting phenomenon. As
shown in Fig.5 and 6，the performance of soft support
is not satisfactory and mesh refinement cannot reduce
the errors of the analysis if t / l is greater than 0.1.
There is also an interesting phenomenon in Fig.6. For
Fig.3 Deflection coefficient at center of plate vs. t/l for
a mesh of 2 × 2 at t / l ratio = 0.3 ， the bending
simply-supported plate subjected to UDL
moment coefficient from soft support is very good
even though the deflection coefficient is not good. For
such a coarse mesh，the technique in recovering the
nodal forces has a great influence on the results. The
author has tried several different nodal forces
recovering techniques on PLATE and an error ranging
from 7.6% to 67% can be obtained. The author
suspected that the error in the nodal forces recovering
technique has partly compensated the error in the IB
analysis for this case so that a very good result is
obtained. This explanation can further be supported by
the fact that the errors for the bending moment of a
4× 4 mesh is greater than that for a mesh of 2× 2
which contradicts the behavior of finite element. The
Fig.4 Moment coefficient at center of plate vs. t/l for corresponding results by PLATE are monotonic
simply-supported plate subjected to UDL behavior with increasing mesh size and the errors are
small if a reasonable mesh is used. Once again，the
For the center bending moment in this case，the solutions by soft support cannot be improved simply
bending moment coefficient is independent of the ratio by mesh refinement.
t / l . For the bending moment coefficients which are In conclusion，the bending moment and deflec-
shown in Fig.4，it is noticed that there are noticeable tion from soft support analysis are generally over-
differences between a mesh of 8× 8 and a mesh of estimated and the errors increase rapidly with t / l .
128× 128 for IB element. This phenomenon is not Mesh refinement can only slightly reduce the error but
surprising as the determination of the bending moment the effect is small. It is clear that IB element cannot
• 1004 • 岩石力学与工程学报 2004 年

Fig.5 Deflection coefficient at center of plate vs. number of elements (log scale) for simply-supported plate subjected to UDL

Fig.6 Moment coefficient at center of plate vs. number of elements (log scale) for simply-supported plate subjected to UDL.

converge to the rigorous solution even with mesh there is a strong need to reduce the cost of pile cap
refinement and this is the fundamental limitation of among many engineerings in Hong Kong. In the
this element. For a thick plate with t / l = 0.3 which is present study ， a 1 m thick 4 m × 4 m pile cap
common for many pile cap or some transfer plate in supporting two 2 m long shear walls is analyzed and
Hong Kong，appreciable saving can be achieved if a the layout of the pile cap is shown in Fig.7. For the
more rigorous analysis is adopted. two shear walls，the loadings on each wall are that
distributed load = 4 000 kN/m，and in-plane moment
3 ANALYSIS AND DESIGN OF Mx = 2 000 kN·m
PILE CAP For in-plane moment，the moment is transformed
to equivalent point loads automatically by plate which
Pile cap is an expensive item in construction and is also the current practice for modeling of in-plane
第 23 卷 第6期 Cheng Y M，et al. Thick Plate Analysis with Application to Pile Cap Design • 1005 •

can be achieved for general problems if the base

pressure from soil is considered in pile cap analysis
and design.
y/m

y/m
x/m
Fig.7 Layout of piles in pile cap with 2 shear wall x/m
Fig.8 Base pressure on pile cap due to soil
moment in Hong Kong. The Young’s modulus and
Poisson ratio of concrete are taken as 2.5×107 kN/m2 4 DISCUSSION AND CONCLUSION
and 0.2 respectively.
This pile cap is supported by 9 bore piles with For a proper thick plate analysis of pile cap， the
small diameter 0.6 m and length of 15 m. The soil author suggests that the size of the element used in
parameters are Es = 25 000 kPa and μ = 0.2. The piles analysis should not exceed 5% of the domain size in
are assumed to be affected each other by the ring shear order to determine the deflection and moment with
theory (after Randolph) so that a coupling analysis is good accuracy. The author discovered that there are
performed. A full stiffness matrix has to be solved in many limitations in the soft support which is
this coupling analysis. Soil beneath the pile cap is sometimes adopted for analysis in Hong Kong. In
commonly assumed to take up no loading but a fact，many engineers use thin plate option to model
3
Winkler’s spring of 25 000 kN/m is assumed in the thick plate by commercial program as the treatment of
present study. The maximum and minimum pile load boundary condition for general shape thick plate is
in pile cap using coupling analysis is shown in Table 1. relatively difficult. In the design of PLATE for
research and teaching，this problem has been solved
Table 1 Maximum and minimum pile load in pile cap
effectively and a demonstration copy of PLATE can be
using coupling analysis kN·m/m
obtained at http：//www.cse.polyu.edu.hk/～ceymcheng/.
Condition Max. load Min. load For soft support ， bending moment is usually
over-estimated (there are isolated cases where the
With soil support 2 598 －426
moment is under-estimated) and noticeable saving can
Without soil support 3 309 －318 be achieved if more rigorous analysis based on
modern thick plate elements are used. From limited
In this simple example，the pile cap has taken up case study ， the author has found that the use of
appreciable loadings from superstructure even though classical thin plate element (available in many finite
Ks is not great. That means ， the soil support element programs) for thick plate problem can
underneath the pile cap is effective in reducing the pile actually give better shear force as compared with the
load and hence the cost of construction. The soil soft support thick plate formulation. If soft support
bearing pressure is determined and is shown in Fig.8. condition is used in the analysis，the moment from the
If the bore piles are very large， the contribution of the analysis may still be acceptable while the shear force
pile cap base may be negligible but noticeable saving should be computed using thin plate analysis.
• 1006 • 岩石力学与工程学报 2004 年

Reduction of the cost of construction is 5 Qin Q H. The Trefftz Finite and Boundary Element Method[M].

particularly important in time of recession. Since New York：WIT Press，2000

transfer plate and pile cap are expensive construction 6 Zienkiewicz O C，Taylor R L. The Finite Element Method 4th
Edition[M]. New York：McGraw-Hill，1991
items， engineers should exercise their knowledge and
7 Bathe K J. A four node plate bending element based on
judgment in the analysis. By performing a more
Mindlin/Ressiner plate theory and mixed interpolation[J]. International
realistic modeling of the problem，appreciable saving
Journal for Numerical Methods in Engineering，1985，21：367～383
in the cost can be achieved. The cumulative effects of
8 Hughes T J R，Hinton E. Finite Element Methods for Plates and Shell
more rigorous and more realistic analysis can be Structures[M]. London：Pineridge Press，1986
beneficial in providing a better and more economic 9 Hughes T J R. The Finite Element Method[M]. London：Prentice
solution. It should be emphasized that reduction of Hall，1987

internal forces means reducing unnecessary waste 10 Ibrahimbegovic A. Quadrilateral finite elements for analysis of thick

instead of reducing the safety of the structure. Factor and thin plates[A]. In：Computer Methods in Applied Mechanics and

of safety should not rely on over-design of the Engineering[C]. [s. l.]：[s. n.]，1993，195～209

11 Jirousek J，Wroblewski A，Szybinski B. A new 12 d.o.f. quadrilateral

structure.
element for analysis of thick and thin plates[J]. International Journal

for Numerical Methods in Engineering，1995，38：2 619～2 638

REFERENCES
12 Cheng Y M. Flexible pile cap analysis[R]. Hong Kong：Hong Kong
Polytechnic University，2003
1 Reddy J N. Theory and Analysis of Elastic Plates[M]. New York：[s.
13 Cheng Y M. A flexible and accurate thick and thin plate element[J].
n.]，2000
Computers and Structures，2003(submitted for publication)
2 Szilard R. Theory and Analysis of Plates[M]. New York：Prentice
14 Computers and Structures Inc. Safe 7.0 user’s and verification
Hall，1974
manual[R]. [s. l.]：Computers and Structures Inc.，2002
3 Timoshenko S，Krieger W. Theory of Plates and Shells[M]. New 15 Computers and Structures Inc. Sap2000 8.0 user’s manual[R]. [s. l.]：
York：McGraw-Hill，1959 Computers and Structures Inc.，2002
4 Urugal A C. Stress in Plates and Shells 2nd Edition[M]. New York： 16 Rieg F. Z88 The Compact Finite Element System ver.10[M]. [s. l.]：
McGraw-Hill，1999 [s. n.]，2002

新书简介

《海河平原地下水演变与对策》一书由张光辉，费宇红等编著，科学出版社 2004 年出版，365 页，估价：35.00 元。

该书论述了海河平原地下水在该区社会经济发展中的作用和地位，界定了地下水系统划分与评价分区，从不同的时间尺
度论述海河地下水系统与相邻圈间水通量演化规律，探讨了地下水循环变化与地质环境、生态环境之间互动机理和未来发展
趋势，阐述了海河平原地下水调蓄功能特征、潜力和优化模式，以及调蓄战略与可持续利用条件。
读者对象：干旱、半干旱地区水资源及环境研究的科研、规划、管理等人员，大学院校相关专业师生。
《近海环境流体动力学数值模型》一书由孙文心著，科学出版社 2004 年出版，394 页，估价：40.00 元。
该书在介绍近海环境流体动力学、数值计算及建模方法基本知识的基础上，以主要篇幅介绍了近岸海洋动力学的一些数
值模型、经典二维的数值模型、各种三维数值模型及国际上几个著名的近岸海洋动力学数值模型。另外，书中还介绍了建立
岸界综合及动岸边界数值模型的几种方法，以及作为环境动力学基础的特质输运核扩散模型。
读者对象：海洋环境动力学的科技工作者，海洋科学和环境科学相关专业研究生、本科生。
《空气污染气象学》一书由蒋维楣编著，南京大学出版社 2003 年出版，432 页，定价：50.00 元。
该书在原教材《空气污染气象学教程》的基础上，保持了原教材讲授空气污染气象学基本原理、处理方法和应用的内容，
教学需要，充分引入了学科发展形成的新领域、新方法等方面的内容。
读者对象：相关专业本科生、研究生，相关部门工作人员。