A Practical Method To Determine in Uence Surfaces Using Commercial Software
A Practical Method To Determine in Uence Surfaces Using Commercial Software
A Practical Method To Determine in Uence Surfaces Using Commercial Software
net/publication/241002887
CITATIONS READS
0 19
1 author:
Jackson Kong
City University of Hong Kong
27 PUBLICATIONS 187 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Jackson Kong on 08 March 2017.
Jackson Kong
Division of Building Science and Technology, City University of Hong Kong, Hong Kong, China
Abstract: In this paper, a simple yet practical method is presented for the determination of critical bending
moments of bridge decks subject to traffic loads by using influence surfaces and simple computing tools.
By means of commercial finite element software, the influence surface is first obtained automatically by put-
ting unit loads across the region concerned, and corresponding results are exported to spreadsheets. A
least square fit to the classical thin-plate solutions, based on the computed results, is subsequently deter-
mined. Based on the least-square fit solution, the exact location of the designed vehicle load and the corre-
sponding critical moments can be obtained using simple and readily available optimization tools in Excel.
The principle behind the approach is only based on the least-square approach and the thin-plate theory. The
method does not involve any complicated mathematics or special software and all computations can be eas-
ily done on a spreadsheet tool like Excel, thus facilitating engineers to implement the method in practice.
Key words: influence surfaces, critical moments, bridge decks, Excel, spreadsheets.
Several general approaches have been proposed in the 2 The proposed method
literature to determine influence surfaces. In particular,
a simple method of equivalent loads [3-6], based on the
2.1 Classical thin-plate bending solutions[7]
reciprocal theorem of linear elastic systems, could be
used to directly determine influence surfaces. The me- For a thin plate subject to a unit point load at (xi,yi), the
thod is versatile and computationally more effective solution of the governing equation of plate bending can
than the conventional “unit-load” method, but its ap- be written as the summation of a singular part and a
plications, unfortunately, involves the explicit form of non-singular part, i.e.,
the strain-displacement functions of the finite elements
used, which, for many commercial software, are usu- w(x,y)=βr2 ln r /πD+ non-singular part (3)
ally not known to practicing engineers, thus hindering where the singularity of the solution is attributed to the
its wide applications in practice. A recent literature logarithmic function. (The singular distribution of
survey of methods for obtaining influence lines and/or bending moment Mx in the plate due to a unit load at
influence surfaces are summarized in [6]. In this study, the center is shown in Figure 3). β is a parameter to be
the conventional unit-load method is used. determined. The non-singular part of the solution is de-
As a vehicle to illustrate the concept and procedure of termined by boundary conditions imposed. D denotes
the bending stiffness of the plate and r represents the
distance of a generic point (x,y) from the loaded posi-
tion:
M x ( x, y; xi , yi )
∂ 2 (r 2 ln r ) ∂ 2 (r 2 ln r )
= β( + ν ) + P( xi , yi )
∂x 2 ∂y 2
= βWP + P( xi , yi )
(6)
in which the derivatives of the singular function WP( x,
y ; xi , yi ) can be easily worked out manually or using
Matlab.
In this study, the non-singular part in (6) is approxi-
mated as a polynomial function of order n. A complete
polynomial of order 7 is adopted, that is,
⎧MXS 0 (x1, y1) ⎫ ⎡WP (x0 , y0 ; x1, y1) P1(x1, y1) P2 (x1, y1) .... P36(x1, y1) 0⎤⎧ β ⎫
⎪ ⎪ ⎢ ⎪ in which the closed form expression for the integral
⎪⎪ ... ⎪ ⎢ .... 0⎥⎥⎪⎪ α1 ⎪
⎪ ⎪ ⎪ is found using Matlab. The location of each wheel load
⎨MX0 (x0 , y0 )⎬ = ⎢ 0 1⎥⎨ ⎬
S
0 0 0 ...
⎪ ... ⎪ ⎢ ...
⎥
0⎥⎪ α36 ⎪
is defined by the center of the vehicle and the given
⎪ ⎪ ⎢ ⎪ ⎪
⎢ ... P36(xn , yn ) 0⎥⎦⎪⎩MXS 0 ⎭⎪ seperation of the wheel loads. The total bending mo-
⎩⎪MX0 (xn , yn )⎭⎪ ⎣WP (x0 , y0; xn , yn ) P1(xn , yn ) P2 (xn , yn )
S
∂ 2 (r 2 ln r ) ∂ 2 (r 2 ln r )
β( + ν ) + [P(xi ,yi )]{α}
∂x 2 ∂y 2
(13)
2.4 Optimization using Excel tion, London, 1961.
[3] Ren H.L., Yeo J S, Shu L. Influence function of thin plates
Once the closed form expression of (14) is obtained, the
using finite strips: computational method and applications.
optimization problem as defined by equations (1) and (2)
Journal of Jianghan Petroleum Institute, 2000, 22(1), 56-58.
method is implemented on Excel using its built-in func-
tion “Solver”. (It is part of a suite of commands some- (in Chinese)
times called what-if analysis : A process of changing the [4] Liu S H, Wu H Q. A method to calculate influence function
values in cells to see how those changes affect the out- of member structure inner forces. Journal of Lanzhou Jiao-
come of formulas on the worksheet). With Solver, engi- tong University (Natural Sciences), 2004, 23(1), 1-5. (in
neers can find an optimal value of Mx0 (xc,yc), i.e. equa- Chinese)
tion (1), with constraints applied to the “design” variables [5] Shen W. The generalized Muller-Breslau principle for high-
(xc,yc) i.e. equation (2) to restrict the values. The General- er-order elements. Computers and Structures, 1992, 44,
ized Reduced Gradient (GRG2) nonlinear optimization 207-212.
method is used in “Solver”. A screen dump from the Ex- [6] Orakdogan E., Girgin K.M. A Direct determination of in-
cel file is given in Figure 4. fluence lines and surfaces by FEM. Structural Engineering
and Mechanics, 2005, 20(3), 279-292.
[7] Timsoshenko S. Theory of Plates and Shells. McGraw-Hill,
3 Numerical Examples
New York. 1959.
Numerical examples of square plates with various bound- [8] Kong J. A Practical method to determine critical moments
ary conditions and bridge decks of box girders, slab-on- in bridge decks using influence surfaces and spreadsheets
girders have been studied using the proposed method. It is (submitted to Computers & Structures)
found that, depending on the separation of the wheel
loads and the size of the wheels, the critical locations of
the vehicle that generate the optimal bending moment at
various locations can be easily estimated. Due to the lim- 5 Acknowledgement
ited scope of the paper, details of numerical results will
be presented in the Conference and reported elsewhere The work described in this paper was partially supported by
[8]. a grant from City University of Hong Kong (Project No.
7001774).
References