Discussion. The Analysis of a Four-Span Skew Bridge Using an Electrical Analogue Computer.

Abstract

) wrote that the Authors had produced an interesting application of electrical analogue networks to a complex stress analysis problem.62. The digital computer presented an extremely versatile and economic approach to such problems, and programs had been developed to apply the finite element method with triangular elementsa.? to plate bending problems such as skew bridges. One such program allowed for many complexities, such as:(U) variable thickness plates; (6) orthotropic structure; (c) irregular geometry; (d) elastic column supports; (e) additional girders in any direction.The program had been successfully tested against exact and experimental solutions on a number of occasions and appeared to present a more versatile and economical procedure than that given in the Paper.63. For comparison an analysis had been made of the simplified example described by the Authors, the case with uniform load q over the whole area being chosen. To confirm convergence of the solution three mesh layouts were used, (A), (B), (C) in Fig. 13. The similarity of the automatically plotted contours of M, for (A) and (C) shown in Fig. 14, demonstrated the good convergence of the solution. For a typical case the total time involved in preparation of data, using automatic mesh generation, was less than three hours. The ICT 1905 computer took less than 15 min for case (B) and additional load distributions added approximately one minute per load case. This order of subdivision was probably adequate for most practical cases.64. Fig. 15 showed section plots of displacement along an edge and moment along the centre line. The finite element results showed reasonable agreement with the Authors on the centre line moment (when allowance was made for the averaging procedure of automatic plotting). There was a discrepancy in displacement along the edge. The difference was however not much more than the 5% indicated by the Authors as the margin of error.65. Clearly it was difficult to assess the accuracy of two approximate methods on such a complex structure where no absolute standard existed.66. The distribution of support reactions obtained by the finite element method showed a considerable discrepancy from that given by the Authors (Fig. 16)