OPEN ACCESS
The implementation of models for the prediction of structural behavior due to blast loads has become vital in the evaluation of threats. Most often, these models require complex finite element analysis and a background in structural and blast engineering to use effectively. Although the finite element models (FEMs) are robust and accurate, it is often necessary to evaluate structures and their potential risks in a relatively short time frame, much quicker than the time a complex, non-linear analysis would necessitate. Typically, simplified engineering tools, such as single degree of freedom models, are utilized in these scenarios but, while effective for some blast applications, often lack the fidelity necessary to sufficiently represent the variable (spatial and temporal) loading and non-linear response of the structure from a range of explosive events.
This paper presents a methodology for a blast analysis tool whereby a FEM is used in conjunction with an artificial neural network to develop a physics-based fast running model. The methodology, which can be applied to a variety of engineering problems, is applied to residual capacity predictions for steel columns subjected to a range of vehicle-borne explosive threat scenarios. Through a set of validation scenarios, the model is shown to be an effective blast analysis tool capable of predictions of structural response within seconds and accessible for security and engineering professionals with varying technical backgrounds.
artificial neural network, blast, steel columns, structural models.
[1] Biggs, J., Introduction to Structural Dynamics, McGraw-Hill: New York, 1864.
[2] Morrill, K., Malvar, L., Crawfor, J. & Ferritto, J., Blast resistant design and retrofit of reinforced concrete columns and walls, Structures 2004, Nashville, TN, 22–26 May 2004, pp. 1–8, available at: http://ascelibrary.org/doi/abs/10.1061/40700(2004)154.
[3] Wu, K., Bing, L. & Tsai, K., Residual axial compression capacity of localized blastdamaged RC columns. International Journal of Impact Engineering, 38(1), pp. 29–40, 2011. doi: http://dx.doi.org/10.1016/j.ijimpeng.2010.09.002
[4] Stewart, L.K., Experimental and computational methods for steel columns subjected to blast loads. WIT Transactions on the Built Environment, 126, pp. 157–168, 2012. doi: http://dx.doi.org/10.2495/SU120141
[5] Anderson, J., An Introduction to Neural Networks, The MIT Press: Cambridge, MA, 1995.
[6] Fausett, L., Fundamentals of Neural Networks, Prentice Hall: New York, NY, 1994.
[7] Mays, G.C. & Smith, P.D., Blast Effects on Buildings – Design of Buildings to Optimize Resistance to Blast Loading, Thomas Telford: New York, 1995.
[8] American Institute of Steel Construction, Steel Construction Manual, 13th edn., American Institute of Steel Construction, 2006.
[9] Stewart, L.K., Computational modeling of steel columns subjected to experimentally simulated blasts, International Journal of Computational Methods and Experimental Measurements, 2(3), pp. 225–242, 2014. doi: http://dx.doi.org/10.2495/CMEMV2-N3-235-242
[10] Livermore Software Technology Corporation, LS-DYNA Keyword User’s Manual, LSTC: Livermore, CA, 2007.
[11] US Army Corps of Engineers, BlastX, 2002.
[12] Rigby, S., Tyas, A., Benneyy, T., Warren, J. & Fay, S., Clearing effects on plates subjected to blast loads, ICE – Engineering and Computational Mechanics, 166(3), pp. 140–148, 2013.
[13] Smith, P.D., Rose, T. & Saotongland, E., Clearing of blast waves from building facades, Proceedings of the Institution of Civil Engineers – Structures and Buildings, pp. 193–199, 1999. doi: http://dx.doi.org/10.1680/istbu.1999.31385
[14] Levenberg, K., A method for the solution of certain non-linear problems in least squares, The Quarterly of Applied Mathematics, 2, pp. 164–168, 1944.
[15] Demuth, H., Beale, M. & Hagan, M., Matlab Neural Network Toolbox User’s Guide, The MathWorks: Natick, MA, 1992.
[16] Press, W., Teukolsky, S., Vetterling, W. & Flannery, B., Numerical Recipes in C: The Art of Computing, 2nd edn., Cambridge University Press: Cambridge, MA, 1992.