A reliable petrophysical rock classification can significantly improve production form organic-rich source rocks, which have been recognized as a major energy resource in the recent years. Complex lithology and rapid vertical variation of petrophysical and compositional properties in these reservoirs result in challenging formation evaluation and production planning. Well logs are good candidates for formation evaluation of shale-gas reservoirs as they provide measurements with a relatively high vertical resolution. However, it has been challenging for the petroleum industry to estimate petrophysical, elastic, and compositional properties and to classify rock types in organic-shale formations using only conventional well-log interpretation techniques. Conventional rock typing techniques are also significantly dependent on core measurements and not reliable in shale-gas reservoirs. In this paper, we introduce a rock typing method based on the estimated petrophysical, compositional, and elastic properties obtained from combined interpretation of well logs and core measurements.
We first jointly interpret photoelectric factor, density, neutron porosity, compressional- and shear-wave slowness, and elemental capture spectroscopy logs to estimate depth-by-depth petrophysical and compositional properties of the organic-rich source formation. We then apply the self-consistent approximation model to estimate depth-by-depth elastic properties of the rock using the estimates of petrophysical and compositional properties. Finally, we use the depth-by-depth estimates of porosity, Total Organic Content (TOC), fluid saturations, volumetric concentrations of mineral constituents, and elastic properties to classify rock types in the reservoir using unsupervised artificial neural network.
We successfully applied the introduced method in the Haynesville shale-gas formation for rock classification. The estimates of porosity, TOC, bulk modulus, shear modulus, and volumetric concentrations of minerals are in agreement with core measurements. We verified the identified rock types using thin-section images. We also showed that well logs can directly be used for rock classification instead of petrophysical/compositional properties estimated from well logs. Direct application of well logs can reduce uncertainty associated with physical models used for well-log interpretation. Implementing an efficient rock classification technique using well logs can potentially improve production from these reservoirs.
Recent developments in technologies such as horizontal drilling and hydraulic fracturing, have transformed organic-shale formations into economic oil and gas resources. The petrophysical and compositional evaluation in these reservoirs has been equally important for the petroleum industry. In the past decade, the recovery from shale-gas resources has increased from about 2% to approximately 50% (King 2010). However, there still remains a great potential for improving the production from these challenging reservoirs.