A recent LinkedIn post by a PhD Engineer describes a particular numerical modelling package as ‘low level’. Another said that his code is superior because it uses a multivariable plasticity model. Model calibration was ignored in the context of these comments as if some models take care of everything including automatically providing realistic results. I could not help wondering how confident these two commentators are that their higher level models are not just using fragile theories wrapped in pretty math?

Geotechnical engineers are sometimes deferring model calibration because the repetitive adjustments and fine-tuning require patience and attention to detail. However, the geotechnical engineering profession is not alone when it comes to calibration and validation of models. With reference to the stock market, Ray Dalio of the asset management firm, Bridgewater Associates said “things you can do to improve your decision-making is to think through your principles for making decisions, write them out in both words and computer algorithms, back-test them if possible, and use them on a real-time basis to run in parallel with your brain’s decision making.”

We should prioritize back-testing of various damage classes (pillar failure, tunnel deformation, stress induced overbreak) and updating the far-field stress model using techniques like over-coring or acoustic emission (AE) before advancing to multivariable constitutive models. While elastic models require only the three major components of the stress tensor to calibrate the results from an overstressing analysis,  a plastic model has four or more variables to calibrate, which adds to the engineering effort and erodes the limited time available at an operation.

Gerrit Kotze, 28 November 2024