[hase Mabe Oy ‘Anew method to assign initial cable forces for prestressed concrete cable-stayed bridges Dr, Dewei Chen, born 1936, received bis degrees from Tongii University of ‘Shanghai. He is now an Associate Professor of the department Summary ‘The determination of initial cable forces in a prestressed concrete cable-stayed bridge for a given vertical profile of deck under its dead load is an important but difficult task that affects the overall design of the bridge. A new method utilizing the idea of force equilibrium is presented in this paper for their determination. The method can easily account for the effect of prestressing and the additional bending moments due to the vertical profile of the bridge deck. Itis much more rational and simpler than the traditional “zero displacement” method, and it is able to achieve bending moments in the bridge deck approaching those in a continuous beam over rigid simple supports. Introduction ‘The cable-stayed bridge is a modern form of bridge which is both economical and aesthetic. It has been extensively employed in the construction of long-span bridges in the past few decades. However this kind of structures are highly statically indeterminate, and therefore ‘many schemes of initial cable forces are possible. In the particular case of prestressed ‘concrete cable-stayed bridges, it is especially important to choose an appropriate scheme of initial cable forces while the bridge is under dead load only. Owing to shrinkage and creep, the deflections will change with the passage of time and the intemal forces may also redistribute. Should an inappropriate scheme of initial cable forces be chosen, an unfavourable pattern of intemal forces may be locked in subsequently, for which there may be no simple solution. Theoretically it is possible to search for a “stable” scheme of initial cable forces under which there is the minimum redistribution of internal forces and time-dependent displacements. However it is usually very difficult in view of the many factors affecting the subsequent time- dependent deformations. For example, many cable-stayed bridges are constructed using cast insitu segmental cantilever construction, which gives rise to complex effects of shrinkage and creep because of the different ages of concrete. The presence of longitudinal prestressing also ‘complicates the problem further. Inevitably some simplifying assumptions have to be made.Review of Existing Methods ‘The scheme of initial cable forces giving rise to bending moments in the bridge deck approaching those of an equivalent continuous beam with all the supports from cables and towers considered as rigid simple supports is generally acknowledged to be both rational and practical, as the long term behaviour of the bridge is reasonably “stable”. The problem hinges ‘upon how to achieve this scheme of initial cable forees. There are two main categories of methods in achieving an appropriate scheme of initial cable forces in prestressed concrete cable-stayed bridges [1-6], namely the optimization method [2-5] and the “zero displacement” method [6]. In the optimization method [2-5], the initial cable forces are chosen based on the optimization of certain objective functions which may either be related to structural efficiency or economy. In this method, the total strain energy is often one of the objective functions to be minimized. It is necessary to impose the constraints for optimization very carefully, or else the resulting schemes may sometimes become impractical. On the other hand, the traditional “zero displacement” method [6] is more straight forward in theory, and it enables the designer to fine-tune the initial cable forces as well as the structural configuration. If a straight and horizontal bridge deck is supported on a number of stay cables, the horizontal components of the cable forces have little effect on the bending ‘moments of the deck, and hence the bending moments are primarily governed by the vertical components of the cable forces and the dead load. In the “zero displacement” method, an appropriate scheme of initial cable forces is obtained by making the deflections at the cable anchorages vanish. When the deck gradient is negligible, the resulting bending moments in the deck are essentially those of an equivalent continuous beam with all supports from cables and towers considered as rigid simple supports. However, when the vertical profile of the bridge deck is significant by reason of traffic requirements or otherwise, the basis of this ‘method is itself questionable. As the horizontal components of the cable forces will induce additional bending moments in the deck, the resulting bending moments are likely to cause substantial redistribution in the long run. In this case, what really matter are the bending ‘moments because they will affect the ong term behaviour of the bridge. Whether the corresponding displacements are zero or not is immaterial, as they can be adequately controlled by appropriate precamber or preset of the deck during construction. Jn this paper, a new method utilizing the idea of force equilibrium is presented for the determination of a “stable” scheme of initial cable forces. The method can easily account for the effect of prestressing and the vertical profile of the bridge deck, and therefore it is much ‘more rational as well as simpler than the traditional “zero displacement” method. Two mumerical examples using real cases of prestressed concrete cable-stayed bridges are presented to demonstrate the versatility of the proposed method. The Force Equilibrium Method In the force equilibrium method, the cable-stayed bridge is modelled as a planar structure. ‘The method works on an evolving substructure eventually comprising the bridge deck and towers, and searches for a set of cable forces which will give rise to desirable bending ‘moments at selected locations of the substructure. As the method works only on the equilibrium of forces rather than deformation, there is no need to deal with non-linearity caused by cable sag and other effects. The method is therefore computationally efficient.First of all, certain sections of the bridge deck and tower are chosen as control sections where the bending moments are adjusted by varying the cable forces. Consider a typical single tower cable-stayed bridge, as shown in Figure 1, in which the connection between the bridge Figure 1. A typical single tower cable-stayed bridge. CETTE TEER ETE Figure 2, Stage 1 model for cable-stayed bridge shown in Fig: 1. deck and tower is monolithic. To established the target bending moments, only the bridge deck is considered. All supports from the cables and tower are replaced by rigid simple supports, as shown in Figure 2. This is regarded as the Stage 1 model for the sake of subsequent discussions. The prestressing to be applied during construction is also taken into account. The bending moments caused by dead load in the bridge deck under such modified support conditions are then taken to be the target bending moments. It is noted that the prestressing to be introduced after the completion of the bridge deck is not taken into account hete. These target bending moments are adopted because the effects of creep and shrinkage of concrete tend to change the bending moments towards these target values in the long term anyway [1]. If the initial bending moments in the towers can be controlled at the same time, the sclieme of initial cable forces is reasonably stable. It is further assumed here that factors such as the differences in age among deck segments are insignificant in the long term and therefore they are neglected. Fig. 3 shows the same bridge as in Fig. 1, except that all. cables are taken away and replaced Figure 3. Model for stage 2 and stage 3 for cable-stayed bridge. by the internal forces. This simplified model applies to both Stages 2 and 3. The only difference between these two stages lies in the degree of sophistication. The cable forces are taken as independent variables for adjustment of bending moments at the control sections. ‘Normally the bending moment at each deck section where a cable is anchored is treated as a ‘control parameter. It should be pointed out that wherever a model consists of a back-stay anchored at the deck above an end pier, where the deck carries no bending moment, the corresponding cable force can be treated as an additional variable to improve the structural efficiency further. For example, the bending moment at the deck-tower junction or that at the tower base may be taken to be an additional control parameter as they are critical sections affecting the long term behaviour. The target bending moments at the deck sections are those ‘obtained from the Stage 1 model whereas the target bending moment at the chosen tower section is normally set as zero.‘The above arguments can also be extended to other configurations of cable-stayed bridges. In a symmetric single tower cable-stayed bridge without back-stays anchored above end piers, the bending moments in the tower should normally be zero under dead load, and therefore there is no need to treat any of these as a control parameter. In a two-tower cable-stayed bridge of symmetric arrangement, itis only necessary to consider one half of the bridge with appropriate boundary conditions at the middle section to account for symmetry, and the above reasoning can similarly be applied. ‘The main purpose for setting up the Stage 2 model is to evaluate the approximate influence coefficients, which are the bending moments at the control sections caused by a unit load in a certain cable. In order not to introduce the non-linearity of cable stiffnesses, some simplifying assumptions are made, The self-weight of each cable is neglected, and hence the forces at the ends are roughly equal. The bending moments in the deck are primarily determined by the cable forees acting on the deck, and to a lesser degree by the cable forces acting on the tower. Therefore the cable forces acting on the tower are neglected in the calculation of bending moments in the deck. Similarly in the calculation of bending moment at the control section at the tower, only the cable forces acting on the tower are taken into account. ‘The errors introduced by these simplifying assumptions will be almost eliminated by iterations in the next stage. Considering the equilibrium of the Stage 2 model, the following equation can be written {at} = [on] {7} + {t*} wo where {M°} is an N-dimensional vector containing the target bending moments M; derived from the Stage | model, [m] is an NxN matrix containing approximate influence coefficients 1m, for the Stage 2 model in which m, is the bending moment at the i* control section caused by a unit force in the * cable, {7} is an N-dimensional vector containing the cable forces 1,, {M*} is an N-dimensional vector containing the bending moments M' caused by only dead load and prestress in the Stage 2 model, V is the number of cables considered in the ‘model, 7 is the subseript comesponding to the i control section and j is the subscript corresponding to the; cable. 1f {14°} contains the bending moments of the equivalent continuous beam on rigid simple supports as obtained from the Stage 1 model, and the control sections are well chosen so that the matrix [m] is non-singular, an initial estimate of the cable forces {7°} can be calculated from the Stage 2 model as {1°} =bod"({t"} -{a04]) @ However the cable forces {7°} obtained above are only rough estimates as the Stage 2 ‘model does not take into account the interaction among tower, cables and deck, It is therefore necessary to build the Stage 3 model. In the Stage 3 model, the interaction among tower, cables and deck is taken into account by iterations. ‘The cable forces at the deck anchorages are taken as independent variables in the optimization process, and the self-weight of each cable can also be introduced. Using the initial estimate of the cable forces {7°}, as well as the bending moments {4*} caused by dead load and prestress in the Stage 2 model, the updated deck bending moments {44"} can be calculated from the Stage 3 model. Such bending moments are normally different from thetarget bending moments {44°}, and hence it is necessary to introduce some adjustments {AT"} of the cable forces given by {ar} = tnt" ({ae)}-(0e}) ® Using the updated cable forces {7"} given by {r'}={7°}+ {ar} @ the updated deck bending moments {24°} can then be calculated again from the Stage 3 model. Notice that the approximate influence matrix [m] for the Stage 2 model has been used in the Stage 3 model, and hence’ further iterations are necessary. Further adjustments {A7?} may be obtained by far} =[nt"({a7} -{40}) © resulting in more accurate cable forces {7°} given by {r}={r"}+ {ar} far} ) ‘This process can be repeated until the updated deck bending moments {4"} converge to {4°}. This is summarized in the flow chart shown in Figure 4. Numerical Examples ‘wo numerical examples are presented to demonstrate that the present method is both rational and reliable. Both are taken from existing prestressed concrete cable-stayed bridges in China, but some minor simplifying modifications are made. Example 1. A single tower prestressed concrete cable-stayed bridge with harp arrangement The first example is a single tower prestressed concrete cable-stayed bridge, situated in ‘Ningbo City, China, with spans of 90m and 105m. The moment of inertia (Zp), the cross sectional area (4p) and the Young’s modulus (Ep) of the deck are 4.706m', 12.145m* and 3.5x10°kN/m” respectively. The stay cables are of the harp arrangement. Three types of stay cables are used, and their respective cross sectional areas (4s) are 0.013m°, 0.0166m* and 0.0208m?. The Young’s modulus of the stay cables (Es) is 2.1x10°KN/m”. The tower is stepped with the biggest section below the bridge deck and the smallest section over the length where the cables are anchored. The moments of inertia (/r) of the tower are 11.212m', 19,939m! and 79.688m‘. The corresponding cross sectional areas (47) are 14.46m”, 19.0m” and 45m? respectively, while the Young’s modulus (E7) is 3x10°KN/m’. The information on the prestressing is omitted for brevity. Three different vertical profiles of the bridge deck have been considered. The bridge deck is straight and horizontal in Case 1. In Cases 2 and 3, both the vertical profiles consist of a symmetric parabolic summit curve of 180m horizontal length and a straight tangent of 15m.‘The bighest point is precisely at the tower location. The gradients of the straight tangents for Cases 2 and 3 are 3% and 9% respectively. Figure 5 shows an elevation of the bridge for Case 3. Figure 5. A single tower P.C. cable-stayed bridge with harp arrangement. ‘The present method was applied to optimize the bending moments in the bridge deck for the three cases, and the tolerance value used to terminate iterations was 5kNm. The results for Case 3 are shown graphically in Figures 6-8. Notice that the deck bending moments after optimization agree well with the target values obtained from an equivalent beam on rigid simple supports, except at the tower section which was not chosen as a control section. The abrupt jumps in bending moment are caused by prestressing. ‘The bending moment at the tower base is also very close to zero after optimization. The three cases were also analyzed by the “zero displacement” method using 0.001m as the Figure 6. Internal forces in bridge deck of example 1 (Case 3). 4) Target bending moment in deck in kNm; (b) Bending moment in kNm; (©) Shear force in KN: (@) Axial force in KN. 3 luual desu l] Figure 7. Cable forces of Example 1 (Case 3) in kN. I r me fF Pre » 9 6 Figure 8, Internal forces in tower of Example 1 (Case 3). @) Bending moment in kNm; (6) Shear force in KN; (¢) Axial force in KN.tolerance value to terminate iterations. The initial cable forces for the three cases obtained by the present method are tabulated in Table 1 and compared to those obtained by the “zero displacement” metliod. It is observed that when the bridge deck has no slope, ie. Case 1, results from the above two methods are effectively the same. There are, however, marked differences in the other two cases, especially in the cables close to the tower. Case Cased Cased Cable [Present] ZaroDisp. | Present | ZeroDixp. | Present | Zero Disp. No. ‘Method | Method | Method | Method | Method | Method 7 Tas 1045 Tas 1398 15458 T5351 z 3931 31 390 3975 cad F065 3 ‘ea cay BD wis 9130 oie 4 oT ora aie 6300 66 wea 3 TaD Tat 7300 Fa00 726 7s é 00 00 oo eis aT 6630 7 FT Tat 7561 7590 788 Fit 3 Cag ere oes waar | eae ei7 9 e707 TOT ois aie sas ae 0 2889 289 ar isa 3 Dae i 14818 14618 Tae 18s Tis B78 @ aed 14a Tse 14339 Tne 7 ® 2887 2888 Bas BaF 3961 aie @ Gig 705 ws aia saa 30 iB 5680 ‘6680 607 a9 Ca otis 6 it 7a BR 752 70 7299 7 oe oy BaF e866 oor Cari ie 704 7064 7085 7025 a oat @ Fat 7304 740 7339 7a Tat 20 CE oe 7027 "7082 7028 70a 2 1076 10768 1922 T0912 T1560 Tis24 a and ‘eed 3306 Bit 9508 et Table 1 Initial cable forces for Example I (KN) Example 2, A single tower prestressed concrete cable-stayed bridge with semi-fan arrangement shown in Fig. 9 ‘The second example is also a single tower prestressed concrete cable-stayed bridge, situated in Jilin Province, China, with spans of 95m and 132m, The moment of inertia (Ip), the cross sectional area (An) and the Young’s modulus (Ep) of the deck are 5.1m", 10.579m” and 3.5x10’kN/m? respectively. The stay cables are of the semi-fan arrangement. Three types of stay cables are used, and their respective cross sectional areas (4s) are 0.020m’, 0.019m” and 0.013m?, The Young’s modulus of the stay cables (Es) is 2.1x10°KN/m’, The tower is stepped in a manner similar to Example 1, and the moments of inertia (;) are 17.92m', 24.01m' and 47.73m*. The corresponding cross sectional areas (47) are 17.92m’, 13.44m?and 37.20m? respectively, while the Young’s modulus (Ey) is 3x10"KN/m’. The effects of prestressing is not considered in this example for simplicity. The vertical profile consists of a symmetric parabolic summit curve of 190m horizontal length and a straight tangent of 37m. The highest point is again precisely at the tower location. The gradient of the straight tangent is 6%. The results obtained using the present method ate shown in Figure 6, indicating very good agreement between the deck bending moments and the target values, shown in Figure 9- 12. Fa @ Figure 10. Internal forces in bridge deck of Example 2. 4a) Target bending moment in kNm; (b) Bending moment in kNm; (©) Shear force in kN; (d) Axial force in KN. J fiut, somal Figure 11. Cable forces of Example 2 in kN. mY eo Figure 12, Internal forces in tower of Eeample 2 (a) Bending moment in kN; (6) Shear force in KN; (©) Axial force in kN.Conclusions ‘Anew method utilizing the idea of force equilibrium is presented for the determination of an ‘optimum scheme of initial cable forces in a prestressed concrete cable-stayed bridge for a given vertical profile of deck under its dead load as well as prestress. In the proposed method, the stiffnesses of the cables do not enter into the calculations, and it therefore obviates the need for introducing non-linearity into the algorithm. ‘The bending moments, rather than the displacements, of the deck are taken as parameters to be controlled. The additional bending moments caused by the vertical profile of the deck. can also be taken into account. Two real prestressed concrete cable-stayed bridges have been investigated using the proposed method, which demonstrate that it is both rational and practical. tis also observed that, as far as the initial bending moments of the tower are concerned, the harp arrangement is less favourable than the fan or semi-fan arrangement, as the cables are anchored over a larger length in the former case. The proposed method is also a handy too! for optimizing the bending moments in the tower. Acknowledgements ‘The financial support of the block grant from “the Scale B”, The Scientific Committee of PR. of China is acknowledged. References [1]. Analysis of Secondary Stresses in Prestressed Concrete Cable-stayed Bridges due to Creep (in Chinese), Shanghai Institute of Design and Research in Municipal Engineering, P.R. China, 1983. [2]. Furukawa, K., Sugimoto, H., Egusa, T., Inoue, K. and Yamada, Y., Studies on optimization of cable prestressing for cable-stayed bridges. Proceedings of International Conference on Cable-stayed Bridges, Bangkok, 1987, 723-734. [3]. LuQ and Xu Y.G., Optimum tensioning of cable-stays (in Chinese). Chinese Journal of Highway and Transport, P.R. China, 1990, 3(1), 38-48. [4]. Simoes, L.M.C. and Negrao, JH.O., Optimization of cable-stayed bridges with box- girder decks. Proceedings of the 1997 Sth International Conference on Computer ‘Aided Optimum Design of Structures, Rome, taly, 1997, 21-32. [5]. Negrao, J.H.O. and Simoes, L.M.C., Optimization of cable-stayed bridges with three- dimensional modelling. Computers and Structures, 1997, 64 (1-4), 741-758, [6]. Wang, P.H., Tseng, T.C. and Yang, C.G., Initial shape of cable-stayed bridges. Computers and Structures, 1993, 46(6), 1095-1106.Begin 1 Input properties of eable-stayed bridge (Figure 1) ‘Set up the Stage 1 model (Figure 2) y (Calculate the target bending moments {M*} ‘rom the Stage 1 model (Figure 2) 1 Set up the Stage 2 model (Figure 3) i Calculate the approximate influence matrix [m] from the Stage 2 model y Calculate the bending moments {247} caused by dead load and prestress from the Stage 2 model Calculate an initial estimate of the cable forces {7°} from {2°} = by! EM} -) ¥ Set up the Stage 3 model (Figure 3) ¥ Calculate the updated bending moments {M?} caused by {M*} and {7°} from the Stage 3 model y Set i=1 ¥ Calculate the adjustment {A7"} to the cable forces from {47° = (my? (LM '}-(00°)) Calculate the updated cable forces {7} from {2} = {T°} + EE ATS t Set i=i+1 ¢ Calculate the updated bending moments {41°} caused by {M%, and {T*1} from the Stage 3 model No ‘Check convergence WM} < 5 2 } Yes End Figure 4. Flow chart describing the present method. 10