On the Representativity of Rock Salt Specimens During Laboratory Tests

Experimental measurements during laboratory tests on salt specimens are important for the design of underground facilities in salt formations. These measurements, especially of the volumetric strain, can be influenced not only by the errors inherent to the experimental work, but also by the material heterogeneity. The natural variability due to the presence of impurities can lead to a non-negligible data scattering in the volumetric strain measurements and therefore raises questions concerning the specimens representativity. In this work, we study the effect of the presence of insoluble materials on the representativity of salt specimens during laboratory tests. An experimental investigation of this matter is delicate due to the shortcomings of laboratory work, thus we suggest a numerical method where salt specimens are modeled as pure halite matrices with insoluble inclusions. The insolubles distribution is described via a leveled Gaussian random field. Instead of using three different constitutive models, we use one law that is flexible enough to be adapted for each material involved (pure halite, inclusions and rock salt) with the adequate choice of parameter set. The modeling approach is presented, compared to the morphological modeling approach, and validated. It is employed to study the natural variability effect on strain measurements and to specify the size requirements to achieve representativity. The study proves that using non-representative specimens can misguide the rock characterization process and might even lead to optimistic design criteria. We propose a methodology that combines numerical and experimental work to overcome representativity issues.