Recovery Estimates from Micromodel Experiments: Processing, Uncertainty, and Rate-dependence

Abstract

Micromodels are commonly utilized to investigate the fundamentals of multiphase displacements and oil mobilization. Definitely, the utility of micromodels has been well demonstrated in the literature. Yet, while the generic workflows are mutual, there is no standard protocol. Therefore, the primary objective of this work was to develop reliable protocols for micromodel experimentations. These protocols are developed within the context of investigating flow-rate effects on oil trapping and recovery, which represents a supplementary objective. The presented experimental work utilized a high pressure and high temperature setup. A metalloid pattern with a pore-volume of 0.08 mL constitutes the porous-media micromodel. The model is positioned vertically, which permits investigation of gravity effects. Displacement experiments were performed to establish the image processing workflow. Those experiments were performed at different injection rates for fixed volumes starting from 10 mL up to 50 mL. All experiments were replicated to assess the associated uncertainties. Initial conditions were established via drainage of connate brine by dead crude oil followed by imbibition of injection brine. The performed experiments established a preferred workflow for image processing that includes in order: thresholding, despeckling, and binary conversion. Thresholding limits were found to be dependent on the camera including its position and focal length. The final binary images can be used for oil recovery estimation based on areal analyses. High rate experiments demonstrated better repeatability. Prolonged injection helped reduce variations in recovery estimates between replicates. At the investigated macroscopic scale and in light of associated uncertainties, recovery was found to be negligibly dependent on injection rate up to a critical flow-rate of around 1 mL/min above which recovery increases with higher injection rates. A trend that is consistent with capillary desaturation. This paper demonstrates the procedure to establish a micromodel image processing protocol. It also illustrates the possible uncertainties associated with recovery estimates obtained from such images. Finally, key observations and recommendations with respect to the significance of high throughput and replications were uncovered.