The impact of uncertain centrifuge capillary pressure on reservoir simulation

Sam Subbey, Mike Christie, Malcolm Sambridge

    Research output: Contribution to journalArticlepeer-review

    7 Citations (Scopus)


    The capillary pressure in a reservoir determines the saturation distribution, and hence the total in situ volumes of fluids (oil/water/gas). The accurate knowledge of the capillary pressure distribution is one of the primary factors that may be decisive in the reliable estimation of hydrocarbon reserves. The centrifuge procedure provides laboratory data, which can be inverted to derive capillary pressure curves in laboratory conditions. The derived laboratory capillary pressure curves are then scaled up for full-field simulation of petroleum reservoirs. The inversion procedure is uncertain and gives nonunique capillary pressure curves. The standard industrial practice, however, has been to derive a single capillary pressure curve, and ignoring the uncertainty. Since the capillary pressure is uncertain, estimates of important reservoir parameters dependent on capillary pressure distribution are uncertain. This paper shows how the uncertainty in centrifuge capillary pressure can be quantified. It also shows how this uncertainty propagates through the scale-up process and impacts on the estimate of the oil recovery potential. We illustrate using a simple, two-phase (oil/water) synthetic reservoir model. © 2004 Society for Industrial and Applied Mathematics.

    Original languageEnglish
    Pages (from-to)537-557
    Number of pages21
    JournalSIAM Journal on Scientific Computing
    Issue number2
    Publication statusPublished - 2005


    • Capillary pressure
    • Centrifuge
    • Ill-posed
    • Inverse
    • Linear integral equation
    • Measured data
    • Stochastic algorithm
    • Synthetic data
    • Uncertainty
    • Volterra
    • Voronoi cells


    Dive into the research topics of 'The impact of uncertain centrifuge capillary pressure on reservoir simulation'. Together they form a unique fingerprint.

    Cite this