The risk of confusing model calibration and model validation with model acceptance

Authors: Elmo, D Download Paper Open access courtesy of:

DOI https://doi.org/10.36487/ACG_repo/2335_19

Cite As:
Elmo, D 2023, 'The risk of confusing model calibration and model validation with model acceptance', in PM Dight (ed.), SSIM 2023: Third International Slope Stability in Mining Conference, Australian Centre for Geomechanics, Perth, pp. 333-342, https://doi.org/10.36487/ACG_repo/2335_19



Abstract:
This paper examines the meaning of calibration and validation in rock engineering design, highlighting several challenges and limitations associated with these processes. There exist two fundamental limitations: i) the inability to rely on engineering judgement as a substitute for proper calibration and validation, and ii) the use of qualitative characterisation methods introduces subjectivity in the data subsequently used for calibration and validation. Furthermore, varying modelling conceptualisations result in a paradoxical situation whereby the same problem analysed using different numerical models requires a different set of parameters, which can all be claimed to be calibrated. The author acknowledges that some of the points raised in this paper may encounter objections. However, by ignoring the epistemic limits of calibration and validation, there is the risk of letting engineering faith become the excuse behind the tendency to replace model validation with model acceptance.

Keywords: model calibration, model validation, model acceptance

References: