Filtration in Frac Packs and Its Impact on Injector Performance

Abstract

SummaryThe filtration of solids in the injection water in a frac pack is the primary factor controlling the injectivity decline in frac-packed water-injection-well completions. The injectivity may decline rapidly or much more slowly, depending on the degree of frac-pack filtration. The widening and lengthening of frac packs and the associated loss of sand control are affected also. However, there are no experimental data available for the filtration coefficient at high velocities that are typical in frac packs. In this research, the filtration coefficients were measured experimentally in high-velocity flows encountered in frac packs. The solid concentrations and pressure drops across proppant-pack sections were measured and correlated with earlier filtration theory and permeability-decline models. The filtration coefficients were measured at various flow rates and for different proppant sizes. Our experiments show that at high fluid velocities, the filtration coefficient is significantly lower than that estimated from prior correlations that are based on low-velocity filtration.We have developed new empirical correlations for filtration coefficients at high flow rates in frac packs. The improved filtration coefficients were used as the primary input into a well-injectivity model. The proper estimation of the filtration enabled us to more accurately analyze the impact of injection rates and proppant selection on injector performance (i.e., predict long-term injection-well behavior). The effect of particle filtration in the frac pack, and its effect on injector performance, was captured accurately for the first time. The newly presented empirical correlations coupled with a model for frac-pack growth in the injection well allowed us to estimate accurately the fracture dimensions and the long-term water injectivity of frac-packed injectors. These results can be used for frac-pack design, proppant selection, and specification of injection-water quality.