科研之友
微信
新浪微博
Facebook
分享链接
摘要
Prediction of extremely low permeability in tight reservoirs poses major challenges with traditional methods. Several studies have proposed nuclear magnetic resonance (NMR) permeability predictors, but these often give large errors when applied in tight formations. In this report, we describe a new method with NMR well-log measurements that decomposes the T-2 spectrum into, at most, three Gaussian components. On the basis of parameters from the decomposition, we build a pore-size-based lithofacies model to predict whole-core horizontal permeability. With these parameters, we also modify the empirical Timur-Coates equation (TIM) to predict permeability. The NMR decomposition allows us to predict proportions of shale and silt. Applied to the tight Cardium formation, the parameters correlate strongly with core image and X-Ray-diffraction (XRD) results. In addition to Cardium data, we apply our approach to published data sets with good results, showing that the model gives accurate lithofacies-proportion estimates. To calibrate the model, Cardium probe permeameter data are used to identify facies permeabilities. Arithmetic-averaged permeability with the NMR-based model was calculated to compare with whole-core horizontal permeability. Monte Carlo analysis confirms the agreement between the model and core-permeability values. Our model provides a "bridge" to relate permeability between the probe scale (<1 cm laminations) and core size (>15 cm thin beds). Without the NMR well-log decomposition, Cardium TIM permeability predictions are in error by more than one order of magnitude in most intervals. The major challenge with the TIM model is obtaining an accurate T-2 cutoff value. Compared with coremeasured bound-water saturations, the default 33 ms value is too large for our tight samples. Our NMR decomposition, however, shows good correlation with measured bound-water saturations. With several core samples and NMR parameters, we modified the TIM model and found that it provides very good permeability predictions.
- 出版日期2016-8
