Miscible-gas foam field trials have employed a variety of injection strategies, with mixed results. Using simulation, we compare foam-injection strategies in homogeneous reservoirs with a variety of foam models. The optimal injection strategy for overcoming gravity override with foam in a homogeneous reservoir is alternating injection of separate, large slugs of gas and liquid at fixed, maximum-allowable injection pressure. This strategy minimizes both gravity override and time of injection, with minimal rise in injection-well pressure. Injection of gas at maximum pressure can partially reverse the effects of gravity slumping of surfactant during injection of liquid. The process is remarkably insensitive to the detailed properties of the foam, as long as foam does not collapse abruptly and completely at a "limiting water saturation" $Sw*$. However, care is needed to exclude the effects of numerical artifacts in simulating such a process, especially if foam collapses abruptly and completely at $Sw*$.

An idealized model for the process reveals the mechanisms responsible for process success, why injection at fixed injection pressure is better than injection at fixed rate, why details of foam behavior play a secondary role in sweep efficiency, and why numerical artifacts can be difficult to identify.