
Simplification of large-scale solid element model for seismic structural response analysis of buildings
Abstract
The seismic response analysis of large-scale structural systems, including integrated earthquake simulations for urbanized cities, requires developing complex numerical models with over 10 million solid elements. However, such detailed models are often computationally intensive and time-consuming. Therefore, simplified models are necessary to efficiently assess the seismic response of structures in the time domain during the early stages of analysis. The authors are proposing a meta-modelling theory in which a simple mass-spring model (MSM) is constructed from a large-scale solid element model. This paper presents examples of converting a three-dimensional solid element model to a one-dimensional MSM, which is denoted as the consistent mass-spring model (CMSM) in the viewpoint of structural mechanics. The performance of the proposed CMSM is compared with conventional MSM and the frequency-adaptive lumped mass-stick model (LMSM). For comparison, three-storey symmetrical and unsymmetrical (T-shaped and L-shaped) reinforced concrete (RC) buildings are selected. Numerical time history simulations are carried out to check the suitability of the proposed CMSM for three sets of ground motions. The proposed CMSMs solve the same physical problem as the solid element model, using suitable mathematical approximations, and show very good agreement with those of the solid models. The first three natural frequencies of the CMSM were found to match those of the solid element model for both symmetric and asymmetric RC buildings in the study, demonstrating the high accuracy of the developed MSM. It is also shown that such a simplified model is used as an alternative to a large-scale solid element model to estimate the overall responses of the structure at the initial stage of large-scale analysis.
© 2025 S.C. Rathnasiri, J.A.S.C. Jayasinghe, A.J. Dammika, published by National Science Foundation of Sri Lanka
This work is licensed under the Creative Commons Attribution-NoDerivatives 4.0 License.