CO2 Storage is one of the key technologies to mitigate climate change at a large scale and
ensuring that the injected CO2 stays trapped underground is one of the main challenges.
It is critical to develop fast and more physically accurate methods for CO2 storage
simulations, otherwise computation times become prohibitive, especially when geological
uncertainty is large, as in deep saline aquifers.
Injection strategies and geological uncertainty have an impact on how much CO2 can
be trapped as residual saturation. Fast and accurate simulators such as the one in
this work are necessary to run the large number of simulations used in optimising CO2
sequestration. Our existing research streamline simulator has been extended with two
improved thermodynamic models to maintain thermodynamic equilibrium along the
streamlines. This minimises time-step size dependence and convergence errors. 1D
simulation along the streamline was compared against analytical solution. Models were
validated on 2D and 3D sections of the SPE 10th Comparative Model using water
alternating gas (WAG) injection followed by chase brine. Results show that both new
thermodynamic algorithms are faster (lower CPU cost) and have a faster convergence
of results than the previous algorithm.
Based on the validated model, we run 3D simulations for a single well strategy for
the stage 2 CO2CRC Otway Project to test residual trapping. Simulation results were
compared to TOUGH2 (finite-difference simulator) simulation results to study numerical
dispersion, convergence of results and CPU times. Streamline simulations decreased
computational time by a factor of five but results were not in agreement. Streamline simulations
simulate advection processes accurately. However, there are other non-advective
processes, such as diffusion, dispersion and buoyancy effects, which streamlines cannot
simulate properly. This could cause the differences between streamline and TOUGH2
simulation results. Incompressibility was one of the main assumptions of the streamline-based simulator
and this could pose some challenges when trying to simulate CO2 sequestration projects
where injection strategy is complex. The CO2 streamline code was extended to add
compressibility. Supercritical CO2 is slightly compressible so including compressibility
in the streamline code is important to be able to model the physics more accurately.
Streamlines can now end anywhere in the reservoir. Expansion or contraction of fluids
can create source or sink cells which act as injection/production cells. Initially the
pressure profile obtained numerically was compared to the analytical solution for radial
single-phase flow and 1D simulations were run to study the effect of compressibility
on the saturation profile. 2D simulations of a slightly compressible case on the SPE10
geological model were compared to ECLIPSE simulations, resulting in good matching.
Then, the 3D Otway field case was re-simulated using the compressible code and results
were compared to those obtained by TOUGH2 without obtaining a good agreement due
to the complexity of the case.
With most of the storage potential being in geological formations which are poorly characterised,
monitoring will be a central part of any CO2 storage project. We have adapted
a new approach for streamline-based history matching which exploits the analogy between
the propagation of a wave front and the pressure front in the reservoir. This
approach uses diffusive time-of-flight which determines the velocity at which pressure
propagates as a function of static and fluid properties. This tool enables us to reconcile
response data with static geological data at an earlier time, improving the management
of the project. This approach has been applied to drawndown-buildup well test for a 2D
synthetic case and a 3D real field case. Results for both cases were satisfactory, showing
a clear improvement in the pressure matching after the 10th iteration in most cases.