Consideration on Data Dispersion for Two-Phase Flow Micromodel Experiments

Transparent man-made porous media, also known as micromodels, are a widely used exploration tool in the field of two-phase flow in porous media (Alireza and Sohrabi in Soc Petrol Eng 166435, 2013; Bondino et al., in International symposium of the society of core analysts held in Napa Valley, California, USA, 2013) to enhance the comprehension of oil recovery mechanisms at pore-scale. Although they have more often been used as qualitative visualization tools to explore the elementary physicochemical features of a given flow mechanism, their utilization as a quantitative tool is interesting especially in industrial context, where they represent an easy and low-cost screening tool for complex recovery mechanisms (low salinity waterflooding, polymer flooding, etc). However, the repeatability of these experiments and thus the possibility to derive quantitative conclusions from them appears not to be investigated in the literature in our field of study. In this work, we explore the dispersion of data such as capillary desaturation curves and secondary waterflood recoveries using micromodels of different sizes and different pore patterns from our laboratory and from an external one. Using datasets with low sampling (low number of repeats of an experiment) and with very large sampling, we document the type of data dispersion, we analyze its reasons and we verify to which extent truly quantitative conclusions can be drawn from these datasets. Our study demonstrates that at low sampling drawing quantitative inferences from our datasets is questionable due to the large uncertainty of the produced data.