Since decades, numerical models have been used to predict the behavior of the subsurface in civil and geological engineering projects. The validity of these codes must be tested against field observations or, more often, apparently simple laboratory experiments, whose outcome is typically known to the modeler. Therefore, there is a distinct need for an objective procedure to verify if codes are capable to mirror and predict supposedly simple, physical processes.
In this Study we have determined microstructural properties, geometric and grain densities, ultrasonic wave velocities, static moduli, uniaxial compressive strengths, tensile strengths, the time dependence of geomechanical properties, thermal and hydraulic properties.
Accurate observation is the basis of natural science, leading to sound understanding of processes. This also applies to rock mechanics. Numerical simulation of rock mechanics systems should therefore ultimately comply with this basis of natural science, in particular if numerical simulations are applied in sensitive contexts such as radioactive waste disposal. Therefore, any numerical model shall be validated against observational data representing relevant processes to prove the robustness of the numerical implementation and create trust in the results of forward simulations. The international project DECOVALEX (DEvelopment of COupled models and their VALidation against EXperiments) was initiated some thirty years ago and had exactly this basis of natural science in mind. Today, simulation tasks have become even more complex and it is essential to validate their underlying assumptions and outcomes against data. In DECOVALEX this is realized with large scale experimental data. However, simulation software uses different mathematical frameworks that are not necessarily based on physics. Here, the initiative as described in this paper jumps in.
We are in the process of generating a comprehensive, high-accuracy benchmark set of mechanical and hydraulic laboratory experiment on granitic rock. As rock behavior is very sensitive to specimen preparation and loading configurations (Koelen et al. 2021) the data set is carefully documented in terms of sample collection, specimen preparation, experimental procedure, results, and data processing. The dataset aims at providing a validation to numerical codes for backward tuning and forward simulation. In the final stage the dataset will, hence, include open-label experimental results for model tuning and single-blinded results for validation of forward simulations. This will generate trust in simulations on predicting rock mass behavior in sensitive applications.
