Observation, Prediction, and Analysis of a Laboratory Two-Well Chemical Flood

Abstract

The concept of volumetric linear scaling was developed to provide a simple oil recovery prediction procedure that adequately described the important aspects of chemical flooding for single and multiple slug systems. This approach has been used successfully to model several chemical flood systems in both field cores and synthetic rock.1,2 When coupled with a flow description, linear flood oil saturation data have also been used to predict oil recovery performance of laboratory pattern floods.3,4 This simple scaling approach will be used here to describe chemical flood performance in a novel unconfined laboratory flood monitored by a microwave saturation scanner. An unconfined two-well geometry was chosen for study because it provided a severe test of the simple streamtube flow description employed. A two-well pattern also represents the simplest of the class of small isolated patterns sometimes employed in field pilot tests. Interpretation and modeling of these isolated patterns has historically been difficult. Extension of the scaling theory to a two-well pattern would provide the engineer with another tool for analysis of such floods. Ten curvilinear unit mobility streamtubes were employed to describe flow in the laboratory experiment. Predictions of 2-propanol saturations measured during miscible injections in the two-well model were compared to observations to check the validity of the streamtube net. Expected effects of the two-well geometry on oil recovery from the laboratory model are discussed in light of the large variation in size of the individual streamtubes in the flow description. Oil saturation profiles were measured during a linear chemical flood involving injection of a small surfactant slug followed by a small polymer slug and continuous drive brine. These So profiles were then scaled along the unit mobility streamtube net. Oil saturation contours, endpoint conditions, oil breakthrough time, and oil production history predictions were made. Comparison of these predictions to the observed performance of the eight day chemical flood carried out in the large two-well model indicate that the scaling concept can be used to model this type of flood. The fixed streamtube description served as a useful first approximation for the unconfined pattern geometry even in the presence of observable crossflow between stream-tubes.