Reactive Transport Modeling of Swelling Processes in Clay‐sulfate Rocks

Abstract

Swelling of clay‐sulfate rocks often causes large problems in geotechnical applications such as tunneling. The primary mechanism inducing the increase in rock volume is the chemical transformation of anhydrite to gypsum, which is triggered by the ingress of groundwater. In the present study, a novel conceptual and numerical modeling approach is developed that emphasizes the effect of groundwater flow in conjunction with the associated availability of water and changing geochemical conditions on the chemical transformation of anhydrite to gypsum. A reactive transport model was developed and hydraulic, reactive, and solute transport as well as mechanical model parameters were estimated through an inversion process, constrained by geodetic ground heave measurements from a study site in Staufen, Germany. The conceptual model of the swelling process was implemented numerically through a dual‐domain modeling approach, whereby the mobile domain accounts for solute transport along discontinuities, and the immobile reactive domain represents the matrix. A mass transfer process accounts for diffusive and/or capillary water transport into the matrix, where the rate‐limited transformation of anhydrite to gypsum takes place. The model calculates heave at the land surface depending on water inflow, the transformation of anhydrite into gypsum and the local stress conditions exerted by overburden pressure. The results show that the proposed reactive transport modeling approach is suitable to quantify the observed swelling‐induced heave at the study site with a plausible parameterization. The study also highlights that diffusion is a decisive factor for the effective rate of anhydrite dissolution and, therefore, the overall chemical transformation process.

Coupled Hydro-Mechanical Modeling of Swelling Processes in Clay–Sulfate Rocks

“…The frequency distribution of the measured heave values in both time and space obtained from the geodetic monitoring network, which corresponds to a total of 3431 observation points (Schweizer et al 2018), is depicted in Fig. 1.…”

“…The heave measurements (3431 observation points) were used to calibrate the numerical model and evaluate the accuracy of the presented modeling approach. However, in the compiled dataset (Schweizer et al 2018), the number of measured heave data was limited at the start of the campaign, and more data were acquired after installing all the observation points. The analysis showed that the dataset has high variability and ambiguity for various distances from the center, indicated by the large spread of heave values (see Sect.…”

“…During the leakage period, water flows into the swelling layer (−104.5 m ≤ y ≤ −62 m) from the left side of the domain with an inflow rate of 1.3.10 −1 kg s −1 (Schweizer et al 2018). During the mitigation phase, the leakage (water inflow) is stopped, while the mechanical behavior (heave) is observed further.…”

“…However, it should be noted that we did not aim to model the classical case of two fluids with different densities (gas and water), but rather using the initial volumetric water content as a proxy for the inflowing water that triggers the swelling. With this approach, we are able to account for the circumstance that swelling does not occur prior to the ingress of water into the swelling layer (Schweizer et al 2018).…”

“…The left side (−104.5 m ≤ y ≤ −62 m) was set to a fixed inflow rate ( 1.3.10 −1 kg s −1 ) during the first 790 days, and then it was set to a no-flow boundary to account for the mitigation period. The inflowing water flow rate was defined on the basis of Schweizer et al (2018) findings. The ambient and inflowing water was assumed as pure water, having a density of 1000 kg m −3 .…”