Improved Performance Models Application to Field Scale GCS Simulation

The numerical simulation plays an increasingly important role in different stages of geological carbon storage (GCS) applications both in the research and industrial projects. In particular, storage verification, risk mitigation, monitoring and matching activities often imply the necessity to perform massive computations at different description scale. It turned out that for the practically valuable GCS cases the typical time and space scales are significantly larger compared to typical reservoir simulation. So, from one side, a number of obstacles related to available computational resources may become prohibitive when complex reservoir simulation approaches are in use for a GCS application. From the other side, this makes inevitable the development of simplified methods with improved computational performance like, for instance, the so-called vertically integrated (VI) models based on partial integration of the system of multiphase flow and energy equations. The main idea of the method is the reduced model dimension so that an initial full 3D model can be substitute by equivalent 2D one; for the latter, despite several known limitations, the dedicated approaches and computer models were developed, tested and applied for the GCS simulation. The main objectives of our work were to develop and test the reduced 2D models with improved computational performance and sufficient accuracy and to study their domain of applicability for field scale simulation in rather different context.

Strictly speaking, to avoid some discrepancies in model consideration compared to “standard” 2D multiphase flow formulation the application of the partial integration technique requires additional assumptions. Being one of them the vertical flow equilibrium assumption removes the ambiguity in phase vertical distribution which is necessary to close the 2D flow problem formulation. At the initial step the general workflow was developed for 3D model transformation (via integration over dip-normal direction and subsequent application of the vertical equilibrium assumption), for the case of sharp interface between the carbon dioxide phase and the brine-in-place phase. This workflow includes the adaptation of single and multiphase transport properties, the material (saturation dependent) functions and the procedure of vertical profiles reconstruction for saturation(s) and pressure. The summary of principal aspects of the workflow including particular approaches and procedures is reported. It should be specified that at this stage the resulting governing equations relating (vertically) averaged saturations dynamics to the pressure at the aquifer top, may be applied using a dedicated reservoir simulator for a field-scale study of the GCS in a deep saline aquifer. It is shown that taking advantage of these processed data the robust estimation of storage capacity, reservoir pressure variation, CO₂ plume dynamics could be carried out.

Two large scale applications are presented and discussed. Firstly, the VI model was tested for 3D synthetic case within relatively simple geological context. Horizontal and tilted deep saline aquifer geometry is used in highly variable operational and numerical environment to analyze the dynamic CO₂ storage cases and directly compare the full 3D to 2D reduced model results. Then, the similar storage scenarios are studied for the available geological model and realistic injection conditions close to the conditions of Utsira storage site (Sleipner area, North Sea). Secondly, the reduced 2D model was applied in more complex geological context for hydro-mechanical coupling problem with significant pressure build-up. This particular study was chosen to demonstrate other possible application for the reduced GCS models. A coupled fluid and mechanical model of a deep saline aquifer, containing cap and base rock, upper overlying and lower basal-aquifer units, and two-phase flow within the storage aquifer, was used for the simulation.

To conclude, the results of this study provide evidence of computational advantages for the VI flow models while remaining in good agreement to non-reduced dimension models. The corresponding initial data generation workflow is presented and discussed in detail. The models applicability for direct and coupled hydro-mechanical (and possibly more complex) simulations is demonstrated together with some preliminary estimations of their application limit.