Hierarchical Bayesian Model for Load Test Database with Extremely Sparse Data

For full-scale load tests (e.g., footing, driven pile, drilled shaft, etc.), it is common that a very limited number (e.g., 1 or 2) of load tests is conducted for a given site. The consequence is that a load test database usually contains sites with extremely sparse data. The hierarchical Bayesian model (HBM) originally proposed by Ching et al. (2021) may not be suitable to analyzea load-test database. This paper proposes a novel HBM, called the C_g-HBM, to deal with a load-test database with extremely sparse data. The idea of this novel HBM and its model structure will be discussed and elaborated. The effectiveness of this novel HBM will be demonstrated using a spread foundation database.

Keywords: Load test database, Hierarchical Bayesian model, Data-driven geotechnics.