Accurate simulation of in-situ combustion processes is computationally very challenging because the spatial and temporal scales over which the combustion process takes place are very small. In the last report, we focused heavily on experimental work. The experimental work is being continued and new combustion tube runs will be discussed in the next quarterly report. In this current and eighth report, we report on our numerical as well as experimental work. In the previous reports, we discussed in detail the development of our adaptive parallel pressure solver, which is based on an efficient Cartesian Adaptive Mesh Refinement technique. This methodology allows much higher grid densities to be used near typical fronts than current simulators. Last quarter we improved the computation of upscaled permeabilities and transmissibilities. This quarter we completed testing and implementation and present a couple of new test cases that clearly show the advantage of integration adaptive mesh refinement with upscaling: not only do we get improved global flow results and reduced process dependency, we also obtain improved local flow velocities.