Chemical osmosis in two‐phase flow and salinity‐dependent capillary pressures in rocks with microporosity. Issue 2 (15th February 2014)

• Record Type: Journal Article
• Authors: Schmid, K. S.; Gross, J.; Helmig, R.
• Is Part Of: Water resources research. Volume 50:Issue 2(2014:Feb.)
• Page Start: 763
• Page End: 789
• Publication Date: 2014-02-15
• Subjects: Hydrology -- Periodicals; 333.91
• DOI: 10.1002/2013WR013848
• Languages: English
• ISSNs: 0043-1397

Abstract

<abstract abstract-type="main"> <title>Abstract</title> <p>The situation of multiphase flow with varying salinity through rock formations with microporosity plays a key role in many applications. Experimental data for single‐phase flow through rocks with microporosity show that chemical osmosis can lead to osmotic pressures in the range of several megapascal, but the effect of osmosis on multiphase flow so far has received little attention. Pore networks can be used to investigate these effects, but crucially depend on expressions for capillary entry pressures. Here we extend the classical theory for capillary entry pressures to the case where chemical potentials play a role. The inclusion of osmosis results into a "capillary‐osmotic" pressure that also depends on salinity differences and temperature. Consequently, also the pore‐scale events of piston‐like displacement and snap‐off depend on salinity contrasts and temperature and not only on pore geometry as has been assumed so far. Examples show that even small salinity differences can lead to significantly different entry pressures and changed pore invasion sequences compared to if osmosis is absent. We show that the ensemble behavior with osmosis often is identical to the case where the medium partly has become more water wet, which implies that osmosis might have a strong impact on laboratory‐scale quantities, but also that their detection in experiments will be challenging. Hence, chemical gradients could be important drivers of multiphase flow in rocks with microporosity and should be included into flow models, which currently is not the case.</p> </abstract>