Optimising Laboratory Test Quantities using Bayesian Statistics
Optimising Laboratory Test Quantities using Bayesian Statistics
The use of Bayesian statistics is making its way into routine geotechnics because of increasing availability of regional databases of geotechnical parameter values and increasingly efficient computing. Bayesian statistics can support development of reliable probability density functions (PDF) of geotechnical parameters. These PDFs allow estimation of statistical uncertainties of parameter values, including the mean (BE), median, confidence interval (BEL, BEH) and the prediction interval (LE, HE) which are key inputs for design of offshore wind foundations. This paper describes the use of Bayesian statistics for optimising laboratory test quantities by leveraging existing data. The presented optimisation approach also covers dynamic updating of the (posterior) probability density function for key parameters (in this case undrained shear strength in triaxial testing for clays) and monitoring of laboratory test quantities as site-specific data become available. Achieved optimisation is compared with the conventional approach of Frequentist statistics.
Maddy Murali; Joek Peuchen; Matthieu Constant; Benoit Spinewine; Michel Vrolijk; Peter. Paul Lebbink
5th International Symposium on Frontiers in Offshore Geotechnics (ISFOG2025)
5 - Data Analytics and Machine Learning
https://doi.org/10.53243/ISFOG2025-434
Contact Information
- Secretary General - Dr A M McNamara
- Email: [email protected]
- Phone: +44 20 7040 8154
- City, University of London
- Northampton Square
- London EC1V 0HB. United Kingdom