| Popis: |
When using conventional petrophysical interpretation techniques on a clean sandstone reservoir in South-East Asia it was found that structure and heterogeneity in the formation caused the resistivity based Sw calculations to grossly underestimate hydrocarbons in place. The sands are layered and, although the porosity remains fairly constant, the permeability is highly variable with laminations of low-permeability water-bearing sands and high-permeability oil-bearing sands. The thickness of the layers is below the resolution of most log measurements. This paper describes a technique for petrophysical interpretation of these reservoirs based on three-dimensional modelling of the formation properties. Extensive log and core data sets were available. Information from core was used to distinguish the different types of sand, NMR-T2 distributions were used to quantify the ratio of sand types at each depth and a saturation-height function was used to determine saturations for the different layers, based on height above the free water level. Synthetic log responses were created from the resulting models and inversion processing based on actual log responses was then applied. Capillary pressure measurements showed that the low permeability sand layers had very high entry pressures, causing them to be water bearing through many intervals. The conventional resistivity logs were dominated by these thin water bearing layers, showing little effect of the presence of the oil in the high permeability layers. This is a similar effect to that commonly seen in laminated shaly sand reservoirs. UV fluorescence core photographs were found to be essential to the study as they clearly indicate the location of the oil and the oil water contacts in the cored intervals, even when the resistivity logs remain too low to differentiate the changes in fluids. This meant that the modelling could be verified in the cored intervals and then applied, based on the bi- and tri-modal T2 distributions, in the non-cored intervals. By modelling the formation properties at multiple scales, and taking formation anisotropy into account, the technique allowed good matches to actual log responses to be achieved and also provided good information for upscaling of the petrophysical properties to the reservoir model scale. This case study illustrates a useful technique for situations where conventional deterministic and optimising routines don't give reliable results. It also provides a methodology, which can be built on in the future, to provide a more robust interpretation model for the many situations where formation structure and anisotropy affect log responses. |
