New Insights From an Old Method After History Matching a Newly Designed 1-D Cyclic Steam Stimulation Experiment

Abstract

Abstract A cyclic steam stimulation (CSS) laboratory experiment was conducted with dead heavy oil. Four cycles of steam injection and fluid production were performed, at reservoir pressure, in order to assist in the numerical modelling and understanding of the main mechanisms involved in the process. This was an important part to developing a base model for a broader project evaluating CSS steam-hybrid experiments with live oil. Experimental data, history matching approach and results, as well as key insights are presented. An experimental setup, originally designed to evaluate CSS hybrid processes, was improved by fitting a sight glass to identify the fluids flowing out of the opposite core end (into a ballast system), during injection cycles. Dead oil was used to facilitate the analysis of this experiment. Relative permeability curves were tuned to history match each cycle sequentially. Injection periods were matched before production ones in order to estimate the amounts of oil and water displaced to the ballast during injection (unknown although total liquid volumes in the ballast were continuously recorded), which were later injected back into the core during production periods. A one-dimensional grid successfully represented the core section while the ballast system was modelled with a production and an injection well. Experimental data such as temperature profiles, pressures and rates were honored. A volumetric ratio of 40% water and 60% oil appeared to be the typical composition of the fluid received by the ballast during injection periods based on simulation results. Fluids reinjected from the ballast back into the core were modelled as an emulsion (i.e., a water-oil mixture). Relative permeability curves were the same for injection and production periods within the same cycle, except for an increased critical water saturation during the last two production periods. One set of relative permeability curves was obtained for each of the four cycles, and are presented in this work. The need to have different curves per each cycle suggests a different flow mechanism was taking place during the CSS test. It appears that the injected steam, after condensing to water, partially emulsified with the heavy oil in the core. Although all the cycles of the CSS experiment were successfully matched using water-oil relative permeability curves, questions about their sufficiency to model heavy oil recovery with steam processes arise. New insights are discussed based on this work and available literature. A CSS experiment conducted on a recently commissioned CSS laboratory setup, that mimics the cyclic movement of reservoir fluids with a ballast system, was successfully history matched using a non-traditional approach. The fluids displaced out of the core-into the ballast-during steam injection were re-injected as a water-oil emulsion. New insights from this work underline the need to rethink the traditional way of modelling heavy oil recovery with steam, where emulsion formation typically occurs.