Publications Details

Value of Laboratory Experiments for Code Validations

Wawersik, W.R.

Numerical codes have become indispensable for designing underground structures and interpretating the behavior of geologic systems. Because of the complexities of geologic systems, however, code calculations often are associated with large quantitative uncertainties. This papers presents three examples to demonstrate the value of laboratory(or bench scale) experiments to evaluate the predictive capabilities of such codes with five major conclusions: Laboratory or bench-scale experiments are a very cost-effective, controlled means of evaluating and validating numerical codes, not instead of but before or at least concurrent with the implementation of in situ studies. The design of good laboratory validation tests must identifj what aspects of a code are to be scrutinized in order to optimize the size, geometry, boundary conditions, and duration of the experiments. The design of good and sometimes difficult numerical analyses and sensitivity studies. Laboratory validation tests must involve: Good validation experiments will generate independent data sets to identify the combined effect of constitutive models, model generalizations, material parameters, and numerical algorithms. Successfid validations of numerical codes mandate a close collaboration between experimentalists and analysts drawing from the full gamut of observations, measurements, and mathematical results.