Yash Mandwariya

BTech: NIT Surat, India, 2021
Joined: MTech, IIT Gandhinagar, India, 2023

Area: Generation of Statistically Consistent Demand Vectors by Inducing Artificial Acceleration and Its Impact on the Underlying Distribution

Performance and risk assessment requires thousands to millions of simulations to generate stable loss distributions. Construction of such a large number of correlated engineering demand parameters (EDPs) using a non-linear response history analysis is impractical due to the rarity of appropriate ground motion records and the computational complexity involved. Therefore, blind analysis is used, which can generate a large sample size of simulations by taking a limited set of input demand vectors. However, to achieve the parent probability structure, an unusually large sample size of simulations is required. An option to accelerate the recovery of the underlying probability structure with fewer simulations is introduced in the blind analysis framework. Inducing such an artificial acceleration into the framework may cause deviation from the actual marginal distribution. In this thesis, a five-story reinforced concrete two-dimensional frame is considered, also taking the material non-linearity into account. This frame is analysed against the two suites of ground motion records consistent with the target spectrum, Suite-A:25 records and Suite-B:100 records. EDP responses obtained from Suite-A are used to construct the parent probability structure, and blind analysis is performed to generate the simulations of demand vectors. Demand vectors are also constructed directly based on the nonlinear response history analysis using Suite-B ground motion records. Assessment of blind analysis is done by comparing the marginal and multivariate distribution parameters by comparing it with those directly computed from the structural analysis of Suite-B records. It has been found that blind analysis with artificially induced acceleration generates statistically consistent demand vectors.

Publications: