SPE Reservoir Evaluation & Engineering
Volume 10, Number 5, October 2007, pp. 527-538

SPE-89992-PA

Prediction of Three-Phase Relative Permeabilities Using a Pore-Scale Network Model Anchored to Two-Phase Data

View full textPDF ( 2,618 KB )

DOI  More information 10.2118/89992-PA http://dx.doi.org/10.2118/89992-PA

Citation

  • Svirsky, D.S., van Dijke, M.I.J. and Sorbie, K.S. 2007. Prediction of Three-Phase Relative Permeabilities Using a Pore-Scale Network Model Anchored to Two-Phase Data. SPE Res Eval & Eng  10 (5): 527-538. SPE-89992-PA.

Discipline Categories

  • 6.3.2 Multi-phase Flow
  • 6.4.2 Gas-Injection Methods

Summary

A first attempt has been made to predict three-phase relative permeability experimental data of a water-wet Berea sandstone obtained by Oak (1990) using the three-phase flow network model for arbitrary wettability developed by van Dijke and Sorbie (2002a). First, the network model is anchored to the corresponding two-phase relative permeability and capillary pressure data using an idealized representation of the pore geometry and a simple parameter-fitting procedure. Then, predictions of three-phase properties are made, which are compared with experimental data as well as previous predictions from different network models. The present study has confirmed that the relatively simple network model, anchored to experimental data, is able to predict three-phase relative permeabilities with reasonable accuracy, comparable to the accuracy of more-complex models. On the basis of these preliminary results, a limited sensitivity study is carried out with respect to different wettability states and two combinations of interfacial tensions (IFTs). This study reveals some new results with respect to the invariance of relative permeability to interfacial-tension combinations and the trend of water relative permeability as a function of the fraction of oil-wet pores in systems of nonuniform wettability.

Introduction

Relative permeabilities in a three-phase system depend not only on phase saturations, but also on the saturation history and may be further complicated by wettability effects. Careful measurement of three-phase relative permeabilities is a very difficult task, which is probably the reason why so few data sets have been reported (Oak 1990, 1991; Egermann et al. 2000; Element et al. 2003; Skauge and Larsen 1994). Empirical correlations, which were designed to predict three-phase relative permeabilities from more readily available two-phase data (as demonstrated in Stone 1970, 1973), are not capable of reproducing this complex behavior. Alternatively, network models predict the three-phase relative permeabilities by simulating the flow processes on the basis of details of pore-space representation, fluids, and pore-scale flow mechanisms. Existing network-model formulations vary from simple bond models with idealized pore geometries to complex models with a detailed geometrical representation of the reconstructed 3D pore space. Theoretically, a more detailed representation of the porous medium and the flow processes should provide a better approximation of the macroscopic flow behavior and better predictive capabilities. However, measuring all the detailed pore-scale parameters, as well as modeling these features, is clearly impossible. Therefore, measurement of three-phase relative permeabilities remains essential, if only for validation of the network models.

Many studies report predictions of two-phase relative permeabilities (for examples, see Øren et al. 1998, and McDougall et al. 2002), but relatively few attempts have so far been made to predict three-phase properties. Two network-model studies have been published (Lerdahl et al. 2000, Piri and Blunt 2002) predicting three-phase experimental data of Oak (1990), (see also Piri and Blunt 2005a, 2005b). Both studies are based on the stochastically reconstructed 3D microstructure of Berea sandstone. To reduce the computational cost of solving the transport equations, the detailed microstructure was converted into a pore network with a simplified pore and throat geometry, which was supposed to preserve all relevant features of the reconstructed pore space. Additionally, both networks incorporate a detailed explicit description of intrapore fluid configurations. Because the above models have been conditioned to a particular rock, no further adjustment of network parameters is needed. Both networks additionally incorporate a detailed, explicit description of the intrapore-fluid configurations and distinguish between receding and advancing contact angles. For a simulation of the water/oil imbibition, the latter were randomly distributed between 10 and 30° (Lerdahl et al. 2000) and between 30and 80o (Piri and Blunt 2005a). Predicted two- and three-phase relative permeabilities of both studies are in reasonable agreement with experimental data.

View full textPDF ( 2,618 KB )

History

  • Original manuscript received: 4 June 2004
  • Meeting paper published: 26 September 2004
  • Revised manuscript received: 30 March 2007
  • Manuscript approved: 10 April 2007
  • Version of record: 20 October 2007