A new way of compositional simulation without phase labeling

Saeid Khorsandi, Liwei Li, Russell T. Johns

Research output: Contribution to journalArticlepeer-review

4 Scopus citations


Current relative permeability models rely on labeling a phase as “oil” and “gas” and cannot therefore capture accurately the effect of compositional variations on relative permeabilities and capillary pressures in enhanced oil recovery processes. Discontinuities in flux calculations caused by phase labeling problems not only cause serious convergence and stability problems but also affect the estimated recovery factor owing to incorrect phase mobilities. We developed a fully compositional simulation model using an equation of state (EoS) for relative permeabilities (kr) to eliminate the unphysical discontinuities in flux functions caused by phase labeling issues. The model can capture complex compositional and hysteresis effects for three-phase relative permeability. Each phase is modeled separately based on physical inputs that, in part, are proxies to composition. Phase flux calculations from one gridblock to another are also updated without phase labels. The tuned kr-EoS model and updated compositional simulator are demonstrated for simple ternary cases, multicycle three-phase water-alternating-gas (WAG) injection, and three-hydrocarbon-phase displacement with complex heterogeneity. The approach improves the initial estimates and convergence of flash calculations and stability analyses, as well as the convergence in the pressure solvers. The new compositional simulator allows for high-resolution simulation that gives improved accuracy in recovery estimates at significantly reduced computational time.

Original languageEnglish (US)
Pages (from-to)940-958
Number of pages19
JournalSPE Journal
Issue number2
StatePublished - Apr 2021

All Science Journal Classification (ASJC) codes

  • Energy Engineering and Power Technology
  • Geotechnical Engineering and Engineering Geology


Dive into the research topics of 'A new way of compositional simulation without phase labeling'. Together they form a unique fingerprint.

Cite this