Bayesian approach for the assessment of sufficiency of geotechnical data

Online
Luis-Fernando Contreras
May 13, 2020
Technical Talk

Technical Talk

Luis-Fernando Contreras

May 13
2020
Technical presentation that discusses the common technique used to relate the number of data points with safe values of the design parameters based on the concept of confidence interval from classical statistics.

The characterisation of geotechnical materials for the design of mine and civil slopes requires the collection of data through site and laboratory investigations. The information provided by data contributes to the reduction of the knowledge uncertainty of design parameters. However, the amount of data collected at different stages of project development is normally limited and subjected to budget and time constraints. Proper assessment of data sufficiency at each stage is, therefore, a key aspect of the slope design process. 

The paper discusses the common technique used to relate the number of data points with safe values of the design parameters based on the concept of confidence interval (CI) from classical statistics (i.e. the frequentist approach). This conventional approach is then contrasted with a technique based on the highest density interval (HDI) from Bayesian statistics, which offers a simpler and more intuitive way to judge the sufficiency of data. The discussion is illustrated with examples of analysis of uniaxial compressive strength (UCS) data, and the intact rock strength parameters σci and mi of the Hoek–Brown strength criterion. 

Luis-Fernando Contreras | Geotechnical Engineer | SRK Australia

Luis-Fernando Contreras

Associate Corporate Consultant (Geotechnical Engineering)

Luis-Fernando has over 40 years’ experience in geotechnical analysis and design, tunnel and dam engineering, slope and risk analysis.

CONTACT ME