Computational Technology Resources - CCP

Keywords: bridge, pier, vehicle, live load, neural network.

Summary

The structural analysis of bridges and their supporting pier foundations is an intriguing subject that is frequently debated among many structural engineers. Modern highway bridge designs must consider the variability of loads on the bridge and the uncertainty of their application. In particular, the application of vehicular live loads is not straightforward. At any given time, vehicles can traverse the bridge at unknown speeds and paths producing different force effects. Fortunately, the correct application of vehicle loads to the bridge superstructure has been documented by the American Association of State, Highway, and Transportation Officials (AASHTO) as well as by other research institutions. However, the subsequent application of these live loads to the supporting bridge piers is still not well understood and is only briefly addressed by the AASHTO-LRFD Design Specifications.

A common situation arises when determining the maximum force effects on the superstructure and pier foundation. The application of vehicular live loads to the superstructure to achieve the maximum force effects in the superstructure does not necessarily produce the maximum force effects in the pier foundation. In other words, an entirely different live load application may produce the maximum force effects in the pier foundation. An exhaustive study of live load position combinations across the bridge deck can produce hundreds of possible design load cases. The most critical live load positions can then be determined by studying the results of all the live load combinations. In an attempt to circumvent this tedious process, this paper introduces the use of neural networks to predict the live load positions across the bridge width to achieve the maximum force effects in the pier foundation. The neural networks use geometric input parameters that describe the bridge and pier structure and produce truck and lane load positions for up to four design lanes, for output.

A live load generation routine was implemented in this paper to study all possible combinations of load placement along the transverse direction of the bridge superstructure. This live load generation was undertaken to determine the most critical loading for interior piers. A separate finite element analysis of the bridge superstructure was performed for each load case to determine the reactions at the bearing pads at the interior support. The bearing reactions were then transferred to the pier where an analysis was performed for each load case. Neural networks were used to encode the relationship between the girder reactions from the load cases and the maximum force effects in the interior piers.

The training and validation phases were then conducted using data from existing highway bridge designs. The training pairs represented the broad range of input parameters that are possible. Although the size of the training set is limited to a database of existing bridges, the network prediction results are good. The success in network prediction indicates that the networks can predict the critical load positions for new problems with reasonable accuracy. The results are very encouraging although further training and network enhancements are still recommended.

References

1: American Association of State, Highway, and Transportation Officials (AASHTO), 1994. AASHTO-LRFD Bridge Design Specifications, AASHTO, First Edition, Washington, D.C.
2: Consolazio, G. (1995). "Analysis of highway bridges using computer assisted modeling, neural networks, and data compression techniques." PhD Dissertation. University of Florida.
3: Hoit, M., McVay, M., and Hays, C., (2002). FB-Pier Users Manual. Florida Bridge Software Institute. University of Florida.
4: Hoit, M., McVay, M., Hays, C., Andrade, P. (1996). Nonlinear Pile Foundation Analysis Using Florida Pier, Journal of Bridge Engineering, ASCE, Vol. 1, No. 4, pp. 135-142. doi:10.1061/(ASCE)1084-0702(1996)1:4(135)
5: Hecht-Nielsen, R. (1990). Neurocomputing, Addison-Wesley Publishing Company, New York.
6: Wasserman, P. (1989). Neural Computing - Theory and Practice, Van Nostrand Reinhold, New York.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £85 +P&P)

	Computational & Technology Resources an online resource for computational, engineering & technology publications
	not logged in - login
Front Page Browse CCP CSETS CTR IJRT Other Authors Search Purchase Guide FAQ Contact us	Civil-Comp Proceedings ISSN 1759-3433 CCP: 76 PROCEEDINGS OF THE THIRD INTERNATIONAL CONFERENCE ON ENGINEERING COMPUTATIONAL TECHNOLOGY Edited by: B.H.V. Topping and Z. Bittnar Paper 79 Bridge Pier Live Load Analysis using Neural Networks M.E. Williams+ and M.I. Hoit* +Florida Bridge Software Institute, University of Florida, Gainesville, Florida, USA College of Engineering, University of Florida, Gainesville, Florida, USA doi:10.4203/ccp.76.79 purchase the full-text of this paper Full Bibliographic Reference for this paper M.E. Williams, M.I. Hoit, "Bridge Pier Live Load Analysis using Neural Networks", in B.H.V. Topping, Z. Bittnar, (Editors), "Proceedings of the Third International Conference on Engineering Computational Technology", Civil-Comp Press, Stirlingshire, UK, Paper 79, 2002. doi:10.4203/ccp.76.79 Keywords:* bridge, pier, vehicle, live load, neural network. Summary The structural analysis of bridges and their supporting pier foundations is an intriguing subject that is frequently debated among many structural engineers. Modern highway bridge designs must consider the variability of loads on the bridge and the uncertainty of their application. In particular, the application of vehicular live loads is not straightforward. At any given time, vehicles can traverse the bridge at unknown speeds and paths producing different force effects. Fortunately, the correct application of vehicle loads to the bridge superstructure has been documented by the American Association of State, Highway, and Transportation Officials (AASHTO) as well as by other research institutions. However, the subsequent application of these live loads to the supporting bridge piers is still not well understood and is only briefly addressed by the AASHTO-LRFD Design Specifications. A common situation arises when determining the maximum force effects on the superstructure and pier foundation. The application of vehicular live loads to the superstructure to achieve the maximum force effects in the superstructure does not necessarily produce the maximum force effects in the pier foundation. In other words, an entirely different live load application may produce the maximum force effects in the pier foundation. An exhaustive study of live load position combinations across the bridge deck can produce hundreds of possible design load cases. The most critical live load positions can then be determined by studying the results of all the live load combinations. In an attempt to circumvent this tedious process, this paper introduces the use of neural networks to predict the live load positions across the bridge width to achieve the maximum force effects in the pier foundation. The neural networks use geometric input parameters that describe the bridge and pier structure and produce truck and lane load positions for up to four design lanes, for output. A live load generation routine was implemented in this paper to study all possible combinations of load placement along the transverse direction of the bridge superstructure. This live load generation was undertaken to determine the most critical loading for interior piers. A separate finite element analysis of the bridge superstructure was performed for each load case to determine the reactions at the bearing pads at the interior support. The bearing reactions were then transferred to the pier where an analysis was performed for each load case. Neural networks were used to encode the relationship between the girder reactions from the load cases and the maximum force effects in the interior piers. The training and validation phases were then conducted using data from existing highway bridge designs. The training pairs represented the broad range of input parameters that are possible. Although the size of the training set is limited to a database of existing bridges, the network prediction results are good. The success in network prediction indicates that the networks can predict the critical load positions for new problems with reasonable accuracy. The results are very encouraging although further training and network enhancements are still recommended. References 1 American Association of State, Highway, and Transportation Officials (AASHTO), 1994. AASHTO-LRFD Bridge Design Specifications, AASHTO, First Edition, Washington, D.C. 2 Consolazio, G. (1995). "Analysis of highway bridges using computer assisted modeling, neural networks, and data compression techniques." PhD Dissertation. University of Florida. 3 Hoit, M., McVay, M., and Hays, C., (2002). FB-Pier Users Manual. Florida Bridge Software Institute. University of Florida. 4 Hoit, M., McVay, M., Hays, C., Andrade, P. (1996). Nonlinear Pile Foundation Analysis Using Florida Pier, Journal of Bridge Engineering, ASCE, Vol. 1, No. 4, pp. 135-142. doi:10.1061/(ASCE)1084-0702(1996)1:4(135) 5 Hecht-Nielsen, R. (1990). Neurocomputing, Addison-Wesley Publishing Company, New York. 6 Wasserman, P. (1989). Neural Computing - Theory and Practice, Van Nostrand Reinhold, New York. purchase the full-text of this paper (price £20) go to the previous paper go to the next paper return to the table of contents return to the book description purchase this book (price £85 +P&P)
Back to top	©Civil-Comp Limited 2023 - terms & conditions