Analogue and numerical modelling of rock deformation due to subsurface compressed gas storage 

Abstract

Storage of compressed gases in lined rock caverns (LRC) has been proposed in order to buffer the peaks of energy intake/output for industrial/renewable systems. LRCs consist of a thin (1-2 cm) steel liner to ensure gas tightness, surrounded by a concrete layer (0.5-2 m) to transmit stresses to the host rock which acts as a pressure vessel. Advantages of storage in such underground LRCs when compared to storage in tanks on the surface or on the seabed include (1) increased safety, and potentially (2) increased storage capacity. Pilot studies exploring the feasibility for LRC storage have been carried out showing the potential for the technology (e.g. Grängesberg, Sweden​; Skallen, Sweden; ANGAS, Japan). All three test facilities demonstrated that it is possible to reach a gas pressure that is at least 20 times larger than the lithostatic pressure without compromising the integrity of the LRC. While these pilot tests are valuable for demonstrating the potential of the LRC technology, it is not possible to extrapolate safe operating conditions for future scenarios (e.g. shallower depths, different rock types). As a result, widespread adoption of LRC storage is hindered by uncertainty. Our focus is operating pressure, which scales roughly with storage capacity, and thus has a direct impact on LRC profitability. In this study we will present results of analogue and numerical models. We focus on brittle deformation in the host rock due to the load exerted by a pressurized cavity. Our specific goal is to determine the extent of safely acceptable brittle deformation, and to identify useful indicators of storage integrity to be monitored during initial pressurization tests and continuous operation. Our analogue modelling of this system identifies key parameters that influence the potential success of these projects including (1) proximity to surface, (2) strength of the host rock, (3) mechanical anisotropy, (4) injection rate and amount, and (5) type of liner(s) applied. For the analogue modelling we use gelatin to represent a shallow competent host rock (e.g. granite, gneiss). A hole placed centrally on one side of the cell allows for the injection of compressed air. Utilizing a balloon placed within the hole, the compressed air acts on the liner until the stress applied by the forced air results in visible strain (i.e. fracturing). For the numerical modelling we use a 2D, visco-elasto-plastic, finite difference, hydro-mechanical code utilizing a pseudo-transient solver running on GPUs. The main goal is to cross-validate the results of the two independent methods, and to provide a straightforward way to extrapolate from laboratory simulation to real world conditions.