An analytical correlation to describe relative permeability is extremely useful in the absence of laboratory measured data, or when a general representative fluid flow is required. Previously, experimental work has been done to develop new correlations to predict gas-water relative permeability (RP) data for various permeability and porosity media. The newly developed technique consists of quick laboratory measurements to measure end-point saturations and newcorrelations to predict gas-water relative permeability behaviour.

These differ from previous work because they honour the reservoir's rock and fluid properties – such as reservoir permeability, porosity, interfacial tension and gas density. The effectiveness and practicality of applying these new correlations to simulate laboratory measured gas-water relative permeability data (for various permeability and porosity cores) has been demonstrated in SPE 71523.

To conclude, as part of a doctoral dissertation, the correlations were tested on simulation models. The aims of this work segment were twofold:

  1. To investigate the applicability of the newly developed analytical correlations to reservoir simulation and their impact on history matching and forecasting of reservoir production. No attempts were to be made to re-match the production history.

  2. To investigate the effect of applying the correlations pre and post-upscaling of static reservoir model. This part of the investigation demonstrates that in situations where the upscaling of the static model has resulted in significant distortion of the K and ϕ relationship, application of Mulyadi-Amin-Kennaird (will be refered to as MAK through out this paper) correlations, or similar, would be erroneous.

The new correlations were used to generate several sets of gas-water relative permeability data to replace the corresponding history-matched relative permeability data in the pre-upscaled and a post-upscaled producing gas-condensate reservoir models. The implications of the new correlations were assessed from the model's historical and forecasted reservoir production performance.

The results of the specific cases investigated demonstrated that while the new correlations did not significantly affect the history-matched period, it did havesignificant impact on the forecasted production profiles and ultimate recovery. Although these are not unexpected, the differences were significant enough to warrant attention; in particular the manner in which such correlations are used to populate dynamic models.

You can access this article if you purchase or spend a download.