Sie sind auf Seite 1von 8

Computers & Structures Vol. 48, No. 5. pp. 80~810, 1993 Printed in Great Britain.

0045-7949/93 $6.00 + 0.00 ,~ 1993 Pergamon Press Ltd

DESIGN OF REINFORCED CONCRETE CYLINDRICAL WATER TANKS FOR MINIMUM MATERIAL COST
G. H. TAN,~" V. THEVENDRAN,~" N. C. DAS GUPTA~" and D. P. THAMBIRATNAM~ tDepartment of Civil Engineering, National University of Singapore, l0 Kent Ridge Crescent, Singapore 0511 :l:School of Civil Engineering, Queensland University of Technology, 2 George Street, Brisbane, Queensland 4001, Australia

(Received 10 August 1992)


Abstract--The minimum material cost design of reinforced concrete cylindrical water tanks according to BS8007 is considered. The material cost takes into account the amount of reinforcement and concrete required. The analysis is simplified by using the beam on elastic foundation (BEF) analogy. The tank wall is modelled as consisting of linear pieeewise slopes. The non-linear constrained minimization problems have been solved numerically by direct search methods using a microcomputer. The results are presented and discussed.

!. INTRODUCTION

Structural optimization has received considerable attention in the literature with tremendous advances in recent years [1, 2]. The idea of optimization to enable more economical structures in design is becoming more important and attractive as construction and material costs increase; shell structures are widely used. However, optimization studies in this area are limited, especially when they involve microcomputer applications. This is probably due to the nature of shell structures whose behaviour is governed by complex differential equations. Problems pertaining to shell structures are usually solved by large finite element packages and therefore iterative computation may be too time consuming. Thevendran and Thambiratnam have made preliminary studies on optimal design of cylindrical tanks using commercial finite element packages and an optimization subroutine [3]. They later modified the analysis by using a numerical approach based on the Runge-Kutta method [4]. However, these approaches have limitations as only bending and hoop stresses are considered as constraints and the problems have been solved using a mainframe computer. In this study, the minimum material cost design of reinforced concrete cylindrical water tanks is considered. The objective of this study is to develop an optimal design procedure that is simple and effective, and can be easily utilized by design engineers having access to microcomputers. The minimum material cost design is sought considering only concrete and steel costs. Such a design may not be the most economical design if other aspects such as construction cost are considered. Nevertheless, it gives a good initial start for a feasible and economical design. 803

The problem basically consists of three parts: namely analysis, design, and optimization. The present shell problem is axisymmetric. Therefore, a simplified finite element method based on beam on elastic foundation (BEF) analogy[6, 7] has been adopted. Earlier studies have shown that the BEF method is efficient, fast, and can be easily implemented on a microcomputer. These features are highly suitable for the present optimization problem. The analysis is divided into two parts: static and free vibration of the tank. The vibration analysis is not the main consideration in this study. However, it is recognized that knowledge of natural frequencies in order to avoid resonance which may cause failure of structures by excessive deflection is important. Therefore, the effect of optimization on natural frequencies of the structure is investigated. The BEF analogy predicts only axisymmetric frequencies. However, earlier studies have shown that when axisymmetric frequencies are elevated, the frequencies in nonaxisymmetric modes also get elevated [8]. In order to achieve feasibility, design is done in accordance with the British code of practice for aqueous retaining structures, BS8007 [5]. The design requirements form the constraints of the optimization problem, resulting in constrained non-linear programming problems. The constrained problems are at first transformed into equivalent unconstrained problems using the exterior point method of the sequential unconstrained minimization techniques (SUMT) developed by Fiacco and McCormick [9] and then solved numerically using a suitable direct search method [10, 11]. The problems in this study are finite optimization problems as they can be defined in terms of a finite number of variables. This differs from classical problems which employ mathematical theory of calculus, variational methods, etc.

804 2. ANALYSISUSING BEF ANALOGY

G.H. TAN et al. which depends on the material and on the cross-sectional dimensions of the shell. This analogy can also be established by examining the differential equations governing a beam on elastic foundation and an axisymmetrically loaded cylindrical shell. These are respectively given as follows: dz [ d2u"~ (4)

The cylindrical water tank is analysed using a finite clement method based on BEF analogy. This section illustrates the theoretical considerations behind the analysis and the methodology implemented. Consider a thin-walled elastic circular cylinder subjected to axisymmetric radial loading as shown in Fig. l(a). In view of symmetry, every section of the cylinder perpendicular to the axis will remain circular, whilst the radius, R, will undergo a change AR = u. The radial displacement, u, can be regarded as the lateral deflection for a longitudinal strip AB (Fig. la). Also, on account of symmetry, it would suffice to consider the deformation of only one longitudinal strip, AB in this case, whose width is taken equal to unity. The thickness of this strip may vary arbitrarily in the longitudinal direction. The corresponding strain component in the circumferential direction, u/R, due to the radial displacement u, will give rise to circumferential (or hoop) forces, N (Fig. 1b), per unit length along the longitudinal edges of the strip AB, and are given by

~x2~EI-~x2)+kfu=p
d2 f D d2u~

dx (

ghu

(5)

N = Eh(u/R)

(1)

where D = Eh3/12(l- v z) is defined as the flexural rigidity of the shell and p is the intensity of load per unit length. The equivalent stiffness matrices are obtained by the finite element method on the basis of minimum potential energy. Derivation of the complete stiffness matrices are presented in the Appendix. The stiffness method [12] is used to solve for the required displacements. In this method, the structure stiffness matrix is obtained by superposing the individual element stiffness matrices. The fundamental equation of the stiffness method is {Q} = [K]{q}, (6)

in which h is the shell wall thickness, R the radius of the middle surface of the cylinder, and E the modulus of elasticity. The hoop forces are considered positive in tension, whilst the radial deflection and loading are positive in the outward direction. The resultant force P (Fig. I b), of the hoop forces N, on the longitudinal edges of the strip, is in the radial direction, and its value per unit length of the strip is given by

where {Q} is the vector of nodal loads, [K] the structure stiffness matrix, and {q} a vector of nodal displacements. For free vibration analysis of shells, neglecting damping effects, the equation of motion can be given as [12] [M]{4} + [K]{q} = 0. Substituting {q} = [Q] e~'t into eqn (7) yields (7)

N
P R

Ehu
Rz . (2)

This resultant force, P, which opposes the deflection, is proportional to the deflection with (Eh/R 2) as the proportionality factor. Hence, a longitudinal strip of an axisymmetrically loaded cylindrical shell can be regarded as a beam on an elastic foundation having a modulus

(2 [I1 -- [K]-'[Ml){q } = 0,

(8)

ks = ~-7

Eh

(3)

where 2 = 1/(.o2, [M] is the consistent mass matrix for the structure, and [I] is the identity matrix. The frequencies of the structure are obtained by solving the above equations. Derivation of the complete consistent mass matrix are presented in the Appendix.
axisymmetric radial loading

I
I I

I L

x fl

(a)

/U

\,
/

" ~ ,/./~width

Ib)

Fig. 1. Thin-walled cylinder subjected to axisymmetric loading.

Cost-effective concrete water tank design 3. OPTIMIZATION The variables considered in this study are the thicknesses at various cross-sections of the shell. For example, in Fig. 2, there are four optimization variables, h0, h~, h2, and h 3. Generally for water-retaining structures, it is found that serviceability considerations control the design. The maximum crack width allowable is taken as 0.1 mm, for critical aesthetic appearance (BS8007: Clause 2.2.3.3). The structure is also checked for strength at ultimate limit state. To obtain a feasible design, the design requirements are incorporated as constraints. 3.1. Objective function The objective function is the total cost of concrete and steel required for the structure. The cost of concrete is taken as S$110/m 3 and the cost of steel as S$850/tonne. (S$ = Singapore dollars.) 3.2. Constraints The design of the water tank is in accordance with BS8007:1987 'Design of concrete structures for retaining aqueous liquids' [5]. The various constraints considered in this study are as follows: (i) The ultimate strength of concrete is obtained by using BS8110[13] simplified equivalent stress block. A value of 0.45f~, is used for the width of the stress block. The allowable moment, M, is governed by

805

where v~ is the design concrete shear stress (as defined in Table 3,9 in BSSll0:Part 1)[12] and v is the maximum applied shear stress (iii) Allowable hoop tension, T, in the reinforcement is governed by

r ~<A~f,/~,

(lO)

where T is the hoop force, Ah is the steel reinforcement, fy is the steel yield strength taken as 460N/mm 2, and 7, the partial safety factor taken as 1.15. (iv) To be effective in distributing cracking the minimum reinforcement Aa required is governed by A d >1 A<pc~it, ( 11 )

where Petit, the critical steel ratio is taken as 0.0035 for grade 460 steel as defined in Appendix A.2, BS8007:198715] and Ac is the area of the concrete section. (v) Allowable crack width, w (in flexure or tension), is governed by w ~<0.1 mm, (12)

where w is the crack width due to flexural tension or direct tension (hoop forces) as defined in Appendix B, BS8007:1987 [5]. (vi) Minimum thickness of section, h, is governed by h >/200 mm, (13)

M~<M~,
where f,,, is concrete cube strength, M is applied moment, and M~ is the ultimate moment of resistance as defined in BS8110:Clause 3.4.4 (Part 1)[12]. (ii) The concrete section is designed such that the wall thickness is able to resist ultimate shear forces. The allowable shear, v, is governed by v <~v<, (9)

where h is defined as the thickness of the shell section. (vii) s, the bar spacings of the section, is governed by (BS8007:Clause 2.6.2.3)[5] s ~<300 mm. (14)

(viii) c, the allowable concrete cover, is governed by (BS8007:Clause 2.7.6)[5] c/> 40 mm. 3.3. Methodology The resulting optimization problems become constrained non-linear minimization problems. A constrained minimization problem may be converted into an equivalent unconstrained one using the 'exterior point' method of the SUMT developed by Fiacco and McCormick [9]. Accordingly, a problem of minimizing a function f ( x ) subject to m constraints gj(x)/> 0 ( j = 1, 2 . . . . . m) is solved by considering the problem of minimizing f ( x , rj) = f ( x ) + ~ rj[gj(x) - [gj(x)I]2
j=l

(15)

h3

-h2
._ L _

h~

Fig. 2. Tank wall modelled as consisting of linear piecewise slopes.

(16)

806

G.H. TAN et al. and the dead load factors are taken as 1.0. In the analysis, a longitudinal strip of the water tank is discretized into five elements of equal lengths. The tank is assumed to be fixed at the bottom and free at the top. Two examples have been considered. The second example includes comparison of direct search methods mentioned in Sec. 3.3.

over monotonically increasing sequences of rj. The resulting unconstrained minimization problems can be solved using direct search methods. The direct search methods have been chosen against gradient techniques as these methods can be easily and quickly programmed, are computationally compact and require minimal storage which is important in microcomputer applications. Furthermore it does not require explicit evaluation of any partial derivatives but relies solely on evaluation of the objective function. All these characteristics are highly favourable for utilization by practising engineers. Direct search methods [10, 11] considered in this study include alternate variable search (AVR), Hooke-Jeeves and Powelrs conjugate directions. Included as comparison is the complex method [11] which can solve the problem automatically without SUMT. The optimization methods adopted may not always lead to global optimum values. This shortcoming may be overcome by using different initial starting points and choosing the best optimum values. 3.4. Solution procedure The solution procedure consists of the following steps: (i) An initial starting value for shell thicknesses as shown in Fig. 2. (ii) Analyse and design in accordance to BS8007. Search for the optimum steel section and the spacing required from the database created. (iii) Check the constraint for violations and calculate the total material costs. (iv) Check if the optimization convergence criteria are satisfied. (v) If step (iv) is not satisfied, the optimization variables are adjusted accordingly and steps (ii)-(iv) are repeated.
4. NUMERICAL EXAMPLES

Example 1
A water tank of height 10 m and radius 20 m, is considered in this example. The Young's modulus and the Poisson's ratio of concrete are taken as 28kN/mm 2 and 0.167, respectively. The specific weight of water is 9.81 kN/m 3 and that of concrete is 24 kN/m 3, while the grade of concrete is 35. The steel reinforcement design is based on maximum bending and hoop stresses. This reinforcement is then applied throughout the tank. Four cases with different types of variation in wall thickness have been considered:

Case 1. A uniform thickness tank: There is only one design variable. Case 2. Linear variation from top to bottom (one slope): There are two design variables. Case 3: Two linear variations in wall thickness (two slopes): There are three design variables. Case 4: Three linear variations in wall thickness (three slopes): There are four optimization variables.
The detailed results of optimization using AVR are summarized irr Table 1 and variation of lowest axisymmetric frequency is shown in Fig. 3.

Example 2
The four cases considered are the same as those in Example 1 except that the steel reinforcement design is based on maximum stress in each element. The reinforcement required is then applied accordingly to each element. A comparison of a few direct search methods are made and the results are summarized in Table 2. The detailed results of optimization using AVR are summarized in Tables 3 and 4. Variation of lowest axisymmetric frequency is also shown in Fig. 3.

The minimum cost design of cylindrical water tanks has been considered. To calculate the ultimate Ioadings, the dead load factor is taken as 1.4 and the live load factor as 1.6. For service loads, both the live

Table I. Summary of results for Example l Distance from base (m) 0 10 0 4 10 0 2 6 10 Wall Volume of Hoop thickness c o n c r e t e reinforcement (m) (ma) (mm2) 0.58 0.78 0.26 0.78 0.44 0.20 0.78 0.25 0.20 0.20 742.9 664.5 553.6 347.2 4188 3141 3141 2010 Flexural reinforcement (mm2) 3590 3216 4020 5360 Cost (S$) 155,000 134,000 128,000 105,000

Case l 2 3 4

Cost-effective concrete water tank design


800 7S0
"~

807

700

No. of slopes

Fig. 3. Variation of frequency with number of slopes.

5. D I S C U S S I O N

Optimization of cylindrical water tanks has been studied using a simple finite element method based on BEF analogy and utilizing direct search methods of optimization. The entire procedure has been programmed using a 80486 microcomputer. The average computation time for case 1 is 4 sec, case 2 is 7.5 sec, case 3 is 12 sec, and case 4 is 19 sec. In the analysis of all the cases in the two examples, five elements were used to discretize a strip of the tank as earlier studies[7] have shown that the results obtained provided sufficient accuracy and convergence. However, for free vibration analysis, 20-30 elements are required. The optimization is illustrated in the two examples presented with different cases to show the flexibility of the method. There are two approaches possible to achieve an optimal design. One approach is to consider the wall thickness of the tank and the steel reinforcements as optimization variables; the other is to consider only the wall thickness of the tank. The results from both approaches are relatively the same as the latter approach still considers cost of steel as a component of the objective function, thereby indirectly influencing the results. However, the first approach has been found to be more time consuming and more prone to errors as the number of the optimization variables are higher. Therefore the second approach is recommended. The results in Tables 1, 3, and 4 show the effects of varying the number of design variables. Material cost decreases as degree of variation increases or in other words as the number of slopes in the wall increases. However, volume of concrete does not

follow this pattern as shown in Table 3. This is as expected as the optimization is based on minimum material cost and not minimum weight. The best section of tank may be curved in shape, with the thickness of the wall maximum at the bottom of the shell. This is due to the maximum bending stresses at the base of the tank. However, in terms of practicality, it would be difficult to construct a tank with curved wall. Therefore in this study, the wall was modelled as consisting of single or multiple linear piecewise slopes. The final geometrical configurations of tank are given in Tables 1 and 3. For comparison, case 1 of Example 1 is taken as the benchmark and the amount of savings that can be achieved in the different cases is illustrated in Table 4. Case 1 of Example 1 represents the simplest case of optimization with the least number of design variables. The shell is of uniform thickness and steel reinforcements based on maximum stresses are used for the entire tank. At the other extreme is case 4 of Example 2 which corresponds to the highest degree of optimization. The wall of the tank consists of three linearly varying slopes. The reinforcement designs are based on maximum stresses at each element. When the two cases are compared there is a distinct savings of $66,000 or 42.6% reduction in cost (Table 5) if the latter is favoured. However, the degree of difficulty in construction increases as the number of slopes increases. A possible compromise would be to design the tank with a single linear varying slope. As shown in Table 5, possible savings of 13.5% and 31.0% can be achieved respectively in case 2 of Examples 1 and 2. The examples above illustrated different feasible designs which satisfy constraints listed in Sec. 3.2. Other constraints may be added according to the requirement of the design engineer. The reinforcements required are obtained from a created database which stores commonly used bar sizes and spacings to ensure practicality. Table 2 shows the results of the comparison made between the different search methods. The methods include simple method like the alternate variable search to more sophisticated methods such as, Powell's method which is based on setting up mutually conjugate directions. The optimized costs are obtained by taking the best value of three different starting values. It can be observed that the final optimized cost obtained from various techniques do

Table 2. Comparison of various direct search methods for Example 2 Case


1 2 3

Method AVR Hooke and Jeeves Powell Complex

Cost (S$) 105,076 102,942 105,960 102,712

Iters. 59 63 94 43

Cost (S$) 91,968 93,509 93,528 93,541

Iters. 126 101 105 87

Cost (S$) 89,051 86,420 88,984 --t

Iters. 170 172 156 --f

Iters. denotes the number of objective function evaluations. tFail to converge to a reasonable value.

808 Table 3. Summary of results for Example 2 Distance from base Case (m) 1 2 3 4 0 10 0 4 10 0 2 6 10 Wall thickness (m) 0.39 0.59 0.21 0.74 0.21 0.20 0.65 0.29 0.20 0.20 Volume of concrete (m 3) 494.5 504.5 390.0 350.0

G . H . TAW et al. Table 5. Cost comparison Example/case Cost (S$) 115,000 105,000 92,000 89,000 Acknowledgement--The authors wish to acknowledge that this study has been facilitated by a research grant (No. RP 890649) from the National University of Singapore. Percentage savings 1/2 1/3 1/4 2/1 2/2 2/3 2/4 13.5 17.4 32.2 25.8 32.3 40.6 42.6

REFERENCES

not differ more than 3% in all the cases considered with the exception of the complex method which failed to converge to a optimum value in case 3. Therefore, for the present problem under study a simple optimization routine like the A V R can be adopted to get reasonable results. The variation of the lowest axisymmetric frequencies are shown in Fig. 3. Generally tanks comprising of linear slopes have higher frequencies as compared to uniform tanks. For example, a comparison between a tank of a single slope and a tank with uniform thickness shows an average elevation of 21%. Therefore, if a single slope tank is chosen for design, material cost savings (31%) and elevation of frequency (21%) can be achieved simultaneously. Frequency analysis is important especially if, in the design it is required to isolate the structure from possible dynamic excitation. The results obtained are relevant as it was noted earlier during optimization [8] that both the lowest axisymmetric frequency and natural frequency increase simultaneously by the same order. 6. CONCLUSION The B E F method coupled with direct search optimization routines has been found to be effective and fast in this study. Simple methods like alternate variable search can be incorporated to give reasonable results. Therefore, practising engineers who wish to obtain an initial feasible and economic design can easily implement the procedure in a microcomputer without resorting to large finite element packages and complicated optimization routines. Table 4. Detail of reinforcements in each element in Example 2 Case 1 2 3 4 Element Hoop(mm 2) Flex. (mm 2) Hoop (mm 2) Flex. (ram 2) Hoop (mm 2) Flex. (mm 2) Hoop(ram 2) Flex. (mm 2) 1 3216 8042 2010 4908 2010 4188 1786 893 2 5360 731 4020 893 3216 893 2680 2680 3 5360 731 4020 804 5360 670 4020 670 4 5360 731 5360 731 5360 670 4020 670 5 2010 731 2296 670 3216 670 3216 670

1. G. I. N. Rozvany and B. L. Karihaloo, Structural Optimization. Kluwer, Norwell (1988). 2. C. A. Brebbia and S. Hernandez, Computer Aided Optimum Design of Structures: Recent Advances. Computational Mechanics Publications, Southampton (1989). 3. V. Thevendran and D. P. Thambiratnam, Minimum weight design of cylindrical water tanks. Int. J. Numer. Meth. Engng 23, 1679-1691 (1986). 4. V. Thevendran and D. P. Thambiratnam, Optimal shape of cylindrical concrete water tanks. Comput. Struct. 26, 805-810 (1987). 5. British Standards Institution, Code of Practice for Design of Concrete Structures for Retaining Aqueous Liquids. BS8007 (1987). 6. M. Hetenyi, Beam on Elastic Foundation. University of Michigan Press, Ann Arbor, MI (1961). 7. D. P. Thambiratnam, Application of a simple finite element method using the 'BEF' analogy for the analysis of axisymmetric shell structures. National University of Singapore Symposium on the Finite Element Method,. 36-40 (1989). 8. D. P. Thambiratnam, V. Thevendran, S. L. Chang and S. L. Lee, Maximization of natural frequencies of cylindrical shells. Engng Optimiz. 13, 85-97 (1988). 9. A. V. Fiacco and G. P. McCormick, Nonlinear Programming: Sequential Unconstrained Minimization Techniques. John Wiley, New York (1968). 10. S. L. S. Jacoby, J. S. Kowalik and J. T. Pizzo, Iterative Methods for Non-linear Optimization Problems. Prentice-Hall, Englewood Cliffs, NJ (1972). 11. P. E. Gill and W. Murray, Numerical Methods for Constrained Optimization. Academic Press, London (1978). 12. A. Ghali and A. M. Neville, Structural Analysis', 2nd Edn. Chapman & Hall, London (1978). 13. British Standards Institution, Code of Practice for Design and Construction. BS8110:Part I (1985). APPENDIX In order to derive the stiffness matrix, a typical element with two degrees of freedom per node is considered as shown in Fig. A.l and a complete cubic displacement function, v, is assumed, and is given by {v} = [Ixx2x3l{a } = [A (x )l{a } (17)

in which {a} is a vector of undetermined coefficients. In terms of the nodal displacement {q}, the displacement function Iv], is given by {v} = [A][CI-'{q} in which {q}=[v I 01 v2 02]. (19) (18)

Cost-effective concrete water tank design

809

Y Pi ~

t
Pj

The same concept is used here for free vibration analysis. The stiffness matrix derived above will be used in the computation of fundamental frequency with a slight modification made to include axial effects. This is required to obtain the lower modes of vibration which are predominantly axial in nature. Define the axial displacement, w, as {w} = [lxl{f} = [F(x)]{f} (31)

(T"

,2

in which {f} is a vector of undetermined coefficients. In terms of nodal displacement {u} in the axial direction, the displacement function is given by, Fig. A. 1. Sign convention for a typical element. The connectivity matrix [C], which involves the nodal coordinates. The bending moment, M~, is given by

{w}=[FlIC] '{u}.

(32)

M, = ~ = OISlIC]with matrix B given by [B]-Potential energy, d2[A (X)] dx 2 =[0 0

d2v

I {q}

(20)

The connectivity matrix [(~] involves the axial nodal coordinates. The strain energy due to axial deformation for a typical element is given by

U =2
2 6x]. (21)

Efo E2dV"
[Hlrh[Hldx

(33)

where strain vector, {e } = d{w}/dx = d[FJ/dx [6"] I{u}. [F] is as defined in eqn (35). Simplification of the above equation and substitution of the necessary parameters yield

UB, due

to bending is given by

Us= ~ J~ ~x2~M~dx
which upon using eqns (18) and (20) becomes

1 ('L (d2v)

U=~{u}r([c] (22)

[('] '{u},

(34)

where [H] =d[F(x)l/dx and E is the Young's modulus. Hence the stiffness matrix [k,,] (due to axial effect) is defined as (23) where [I?,,1 = E

Ue=~{q}r([C]

')r{foL[B]rO[B]dx}[C]-'{q }.

[km] = ([d]-')'It?m] [d]-',

(35)

The stiffness matrix [ks] for the beam element is thus given by [ks] = ([C]-l)r[kn] [C]-~, where (24)

[Hlrh[H]dx.

(36)

[kBl =
Potential energy, given by

[BlrD[BI dx.

(25)

UF, due

to the 'foundation modulus' is

Ur = ~ where yield

if0:

{v}rki{v} dx,

(26)

k: is

given by eqn (3). Using eqn (18) in (26) will

This (2 x 2) matrix is added to the earlier derived (4 4) matrix to give a (6 x 6) matrix. In addition to the stiffness matrix, it is also necessary to derive the element mass matrix. The element mass matrix is a matrix of equivalent nodal masses that dynamically represent the actual distributed mass of the element. The derivation of mass matrix can be separated into two parts. The first part of the mass matrix considers the contribution of kinetic energy due to lateral velocity [ml] and the second part considers the contribution from axial velocity [m,,]. Both are then combined to obtain the element mass matrix. For the first part there are four degrees of freedom to an element. The kinetic energy of an element is given by

Us=~{q}r([c]-')r{ff [A]rk:[A]dx}[C]-'{q}.

(27)

T= ~

{~)}rp,dV{li},

(37)

The stiffness matrix pertaining to the equivalent foundation modulus is thus given by

where {f~}=d{v}/dt and Pc is the density of concrete. Substitution of eqn (18) into (31) and further simplification yields

[kr] =
where

([C]-')r[kr] [C]-',

(28)

r =~{OIT([c] -' )

[Cl-'{q}.

(38)

The mass matrix, [rail, is therefore given by

[kvl = f ] In ]rk:[A l dx.

(29) where

[ml] = ([C] ')r[th,][C] ',

(39)

Finally the complete stiffness matrix for the element is given by [k] = [kB] + [kr]. (30)

[n~l = p, IL[.41 rh[4] dx.


J0

(4O)

810

G. H. TAN et al.
Hence the mass matrix is given by

Kinetic energy of an element due to axial velocity is defined as

T=~

{~i,}rpcdV{~b}

(41) where (42)

[too] = ([]-~)~[,~o]()-',

(43)

which upon using eqn (32) becomes

T=2{ft}r([~]-t)r+{~[A]rh[A]dx}[~]-'{f~ }.

[,~o] = pc

jo

I L [Fr]h[r] dx.

(44)

Das könnte Ihnen auch gefallen