Development of CFD-Based Simulation Tools for In Situ Thermal Processing of Oil Shale/Sands
- Assess the capability of Computational Fluid Dynamics (CFD)-based simulation tools to quantitatively predict performance of a modified in situ oil shale treatment process
- Use simulation data and available test data from the ECOSHALE capsule to perform a validation/uncertainty quantification (V/UQ) of the modified in situ process
- Demonstrate how combined experiments and simulations with the V/UQ approach can provide quantified understanding of in situ processes that are convective heat transfer controlled.
Department of Energy, National Energy Technology Laboratory
In situ technologies are currently being explored because of their potential for reducing the environmental footprint of oil shale development. However, the first generation technologies have proven to be energy-intensive, and many unknowns remain relative to optimal heating strategies, potential groundwater contamination, and achievable production rates.
Reservoir simulation tools are typically applied to in situ production processes. However, in the case many oil shale and oil sands applications, the rate limiting step is not porous media flow but the rate of heat transfer in the thermal treatment process. For example, the rate-controlling step in the modified in-situ process is the combined convective-conductive heat transfer throughout the rubblized bed. In this case, there is a distribution of rock size in the production bed and those rocks are packed in such a way that large convective currents heat the bed. Preliminary simulations using a reservoir simulation-type approach (e.g. fluid flow through porous media) showed that such an approach is insufficient to resolve key physics affecting production rates, particularly convective heat flow patterns.
This project takes the novel approach of applying massively parrallel CFD-based simulation tools to a modified in situ process. Rigorous validation/uncertainty quantification (V/UQ) requires both a simulation tool that captures the relevant physical processes and data from a large-scale system. Initially, the focus is on pilot-scale heat transfer data obtained from Red Leaf Resources' ECOSHALE capsule. As data sets from other processes become available, the tools being developed can be applied to those processes as well.
The ECOSHALE capsule, which consists of a clay-lined volume filled with rubblized oil shale and heated by pipes fired with natural gas burners, is simulated using a suite of commercial software tools: Matlab, Gambit, and Star-CCM+. The random shale distribution inside of ECOSHALE capsule is simulated using a discrete element method (DEM) capability in Star-CCM+. Particles are packed randomly based on input particle size distributions and on particle physics. Figure 1 shows the random particle packing obtained from the Star-CCM+ DEM simulation, which is then converted using a Matlab script and Gambit meshing software into a computational domain with convective channels between the particles where fluid flow occurs. With the computational domain as an input, the channel physics are simulated in Star-CCM+ using Direct Numerical Simulation (DNS).
This methodology has been used to produce a CFD simulation of the heat transfer occurring in a simplified computational domain representing the ECOSHALE capsule. Figure 2 shows the thermal distribution of the fluid inside the convective channels as well as pieces of shale. By incorporating an appropriate kerogen pyrolysis model, production rates of gaseous and liquid fuels for a given gas burner firing rate can be computed.
Figure 1 (left):. Representative portion of the ECOSHALE geometry including the randomly packed oil shale particles as well as heating tubes.
Figure 2 (right): Thermal distribution of the convective fluid flow as well as the pieces of shale, in a plane of the representative ECOSHALE geometry.
Once this set of tools has shown its efficacy with a demonstration simulation of the representative ECOSHALE capsule geometry, a V/UQ analysis will be performed involving experimental uncertainty, model uncertainty, operating condition uncertainty and numerical uncertainty with the goal of better understanding the processes that drive production in a modified in situ process.