Civil Engineering Reference
In-Depth Information
Figure 9.14
Analytical results of a shear wall with various sizes of finite element meshes
concrete structures by first validating the program by comparing the predictions to the tests of
panels, beams, framed shear walls, bridge columns, and wall buildings.
Four example input files of SCS are shown in Appendix A. The first example is the static
analysis of the prestressed concrete beam B1, as described in Section 10.2. The second is
the reversed cyclic analysis of post-tensioned column C5C, as described in Section 10.4. The
third one is the reversed cyclic analysis of the framed shear wall FSW13, as described in
Section 10.3.1. The last is the dynamic analysis of the shear wall STN under seismic loading,
as presented in Section 10.5.
For nonlinear finite element analysis of RC/PC plane stress structures, the following dis-
cussion is worthy of study.
1. Mesh size. Because cracks are fully smeared in the CSMM, the finite element that addresses
cracked concrete with embedded steel bars and tendons can be considered as a continuum.
To examine the proper size of the elements used in finite element analysis, a reinforced
concrete shear wall FSW6 described in Section 10.3.1was analyzed using different mesh
sizes. The analytical results are shown in Figure 9.14.
Figure 9.14 shows that the analytical result obtained using only one wall element gives
much higher prediction than the remaining three curves with 4, 9 and 25 elements. The
prediction using 4 wall elements is slightly greater than those of 9 elements and 25 elements.
The two analytical curves using 9 elements and 25 elements are barely distinguishable.
The rapid convergence of the analytical curves with increasing number of elements
was also observed by Okamura and Maekawa (1991) when the smeared crack model was
applied to the finite element analysis of shear walls. Thus, a relatively small number of RC
plane stress elements is adequate to obtain accurate outcomes using the developed SCS
program and also saves precious computation time.
2. Nonlinear solutions. The modified Newton-Raphson method with Krylov subspace ac-
celeration (Carlson and Miller, 1998) was found to converge faster and was more stable
