Creep of Rock Salt Under a Large Range of Deviatoric Stresses

The design of underground facilities in rock salt relies on constitutive models whose parameters are derived from laboratory tests and (when possible) recalibrated on the basis of field data. In this sense, laboratory tests should cover the load paths expected in real-scale applications. However, the literature review proves that most tests have been conducted at stress deviators greater than about 5 MPa, which is not representative of the whole spatial and temporal spans of the facilities. In this paper, results of confined multi-stage creep tests performed on salt samples from two different origins are shown. Stress deviators between 0.5 and 20 MPa have been applied to different samples under a constant temperature of 50 ∘ C. The experimental procedure allows to reduce uncertainty associated with sample-to-sample variability, sample preparation and preconditioning. The analysis of the experimental data shows that the stress dependency of the creep rate is different at low and high differential stresses. In particular, the creep rate is underestimated by several orders of magnitude when the creep tests are performed at high differential stress levels and the trends are extrapolated towards low and moderate levels. On the basis of these results, the Lemaitre and RTL constitutive models for rock salt have been modified, allowing for more accurate predictions of underground facilities. Properly designed laboratory tests are necessary to limit extrapolation; however, the duration of such experiments is negligible compared to the lifespan of the underground facilities, and long-term aspects should be further investigated. Highlights Confined multi-stage creep tests were performed covering stress deviators from 0.5 to 20 MPa. Rock salt specimens from two different origins were tested. The deviatoric stress dependency of the creep rate is different at low and high deviatoric stresses. Two phenomenological viscoplastic models for rock salt are extended to better reproduce creep behavior. Although stress extrapolation is addressed, time extrapolation remains a major challenge.