Prediction of Normalized Modulus Reduction Curve Based on Limited Measurement Data

The precision and reliability of the normalized modulus reduction curve in laboratory tests can be significantly compromised by the scarcity of specimens as well as variations among them, which can further impact the accuracy of seismic response analysis. In this paper, a Bayesian updating framework is proposed to predict the normalized modulus reduction curve with limited measurement data. Different prior distributions and measurement errors are investigated. Results show that the proposed technique is effective and efficient in handling limited data, reducing the uncertainty of variables through measurements, and improving the accuracy of the normalized modulus reduction curve with the acquisition of more measurements.