Synthesizing Nuclear Magnetic Resonance (NMR) Outputs for Clastic Rocks Using Machine Learning Methods, Examples from North West Shelf and Perth Basin, Western Australia

A nuclear magnetic resonance (NMR) logging tool can provide important rock and fluid properties that are necessary for a reliable reservoir evaluation. Pore size distribution based on T2 relaxation time and resulting permeability are among those parameters that cannot be provided by conventional logging tools. For wells drilled before the 1990s and for many recent wells there is no NMR data available due to the tool availability and the logging cost, respectively. This study used a large database of combinable magnetic resonance (CMR) to assess the performance of several well-known machine learning (ML) methods to generate some of the NMR tool’s outputs for clastic rocks using typical well-logs as inputs. NMR tool’s outputs, such as clay bound water (CBW), irreducible pore fluid (known as bulk volume irreducible, BVI), producible fluid (known as the free fluid index, FFI), logarithmic mean of T2 relaxation time (T2LM), irreducible water saturation (Swirr), and permeability from Coates and SDR models were generated in this study. The well logs were collected from 14 wells of Western Australia (WA) within 3 offshore basins. About 80% of the data points were used for training and validation purposes and 20% of the whole data was kept as a blind set with no involvement in the training process to check the validity of the ML methods. The comparison of results shows that the Adaptive Boosting, known as AdaBoost model, has given the most impressive performance to predict CBW, FFI, permeability, T2LM, and SWirr for the blind set with R2 more than 0.9. The accuracy of the ML model for the blind dataset suggests that the approach can be used to generate NMR tool outputs with high accuracy.

A review of basic well log interpretation techniques in highly deviated wells

Drilling deviated wells has become customary in recent times. This work condenses various highly deviated and horizontal well log interpretation techniques supported by field examples. Compared to that in vertical wells, log interpretation in highly deviated wells is complex because the readings are affected not only by the host bed but also the adjacent beds and additional wellbore-related issues. However, understanding the potential pitfalls and combining information from multiple logs can address some of the challenges. For example, a non-azimuthally focused gamma ray logging while drilling (LWD) tool, used in combination with azimuthally focused density and neutron porosity tools, can accurately tell if an adjacent approaching bed is overlying or underlying. Moreover, resistivity logs in horizontal wells are effective in detecting the presence of adjacent beds. Although the horns associated with resistivity measurements in highly deviated wells are unwanted, their sizes can provide important clues about the angle of the borehole with respect to the intersecting beds. Inversion of horizontal/deviated well logs can also help determine true formation resistivities. Additionally, observed disagreement between resistivity readings with nuclear magnetic resonance (NMR) T2 hydrocarbon peaks can indicate the presence or absence of hydrocarbons. Furthermore, variations in pulsed neutron capture cross sections along horizontal wells, measured while injecting various fluids, can indicate high porosity/permeability unperforated productive zones. Finally, great advances have been made in the direction of the bed geometry determination and geologic modeling using the mentioned deviated well logs. More attention is required toward quantitative log interpretation in horizontal/high angle wells for determining the amount of hydrocarbons in place.

Development of 1D-Synthetic Geomechanical Well Logs for Applications Related to Reservoir Geomechanics in Buzurgan Oil Field

Knowledge of the distribution of the rock mechanical properties along the depth of the wells is an important task for many applications related to reservoir geomechanics. Such these applications are wellbore stability analysis, hydraulic fracturing, reservoir compaction and subsidence, sand production, and fault reactivation. A major challenge with determining the rock mechanical properties is that they are not directly measured at the wellbore. They can be only sampled at well location using rock testing. Furthermore, the core analysis provides discrete data measurements for specific depth as well as it is often available only for a few wells in a field of interest. This study presents a methodology to generate synthetic-geomechanical well logs for the production section of the Buzurgan oil field, located in the south of Iraq, using an artificial neural network. An issue with the area of study is that shear wave velocities and pore pressure measurements in some wells are missing or incomplete possibly for cost and time-saving purposes. The unavailability of these data can potentially create inaccuracies in reservoir characterization n and production management. To overcome these challenges, this study presents two developed models for estimating the shear wave velocity and pore pressure using ANN techniques. The input parameters are conventional well logs including compressional wave, bulk density, and gamma-ray. Also, this study presents a construction of 1-D mechanical earth model for the production section of Buzurgan oil field which can be used for optimizing the selected mud weights with less wellbore problems (less nonproductive time. The results showed that artificial neural network is a powerful tool in determining the shear wave velocity and formation pore pressure using conventional well logs. The constructed 1D MEM revealed a high matching between the predicted wellbore instabilities and the actual wellbore failures that were observed by the caliper log. The majority of borehole enlargements can be attributed to the formation shear failures due to an inadequate selection of mud weights while drilling. Hence, this study presents optimum mud weights (1.3 to 1.35 g/cc) that can be used to drill new wells in the Buzurgan oil field with less expected drilling problems.

Evaluation of Mishrif Reservoir in Abu Amood Oil Field, Southern Iraq

The main goal of this study is to evaluate Mishrif Reservoir in Abu Amood oil field, southern Iraq, using the available well logs. The sets of logs were acquired for wells AAm-1, AAm-2, AAm-3, AAm-4, and AAm-5. The evaluation included the identification of the reservoir units and the calculation of their petrophysical properties using the Techlog software. Total porosity was calculated using the neutron-density method and the values were corrected from the volume of shale in order to calculate the effective porosity. Computer processed interpretation (CPI) was accomplished for the five wells. The results show that Mishrif Formation in Abu Amood field consists of three reservoir units with various percentages of hydrocarbons that were concentrated in all of the three units, but in different wells. All of the units have high porosity, especially unit two, although it is saturated with water.

Depositional Environments and Reservoir Evaluation of Otuma Oil Field, Niger - Delta basin, Nigeria

This study was carried out by using well logs to evaluate the depositional environments and hydrocarbon reservoirs in the Otuma oil field, Niger Delta basin. The gamma motif/model within- study interval in the drilled well shows blocky, symmetrical, and serrated shapes which suggest a deltaic front with mouth bar to a regressive - transgressive shoreface delta respectively. A correlation was done on the well logs across the wells and the ten well logs were used to evaluate the petrophysical characteristics of the reservoirs. The reservoirs showed highly porous and permeable channels where the wells were used for the characterization. The ten reservoirs were mapped at a depth range of 2395 m to 2919 m with thicknesses varying from 4m to 135m. The petrophysical results of the field showed that the porosity of the reservoirs ranges between 0.10 to 0.30, and permeability from 48 md to 290 md; the water saturation ranges from 0.39 to 0.52, and hydrocarbon saturation from the field 0.48 to 0.61. The By-passed hydrocarbons identified in low resistivity pay sands D4 and D3 at depth 2649 m to 2919 m, respectively were also evaluated and will be put to production in the field.

Ensemble Learning for Predicting TOC from Well-Logs of the Unconventional Goldwyer Shale

Precise estimation of total organic carbon (TOC) is extremely important for the successful characterization of an unconventional shale reservoir. Indirect traditional continuous TOC prediction methods from well-logs fail to provide accurate TOC in complex and heterogeneous shale reservoirs. A workflow is proposed to predict a continuous TOC profile from well-logs through various ensemble learning regression models in the Goldwyer shale formation of the Canning Basin, WA. A total of 283 TOC data points from ten wells is available from the Rock-Eval analysis of the core specimen where each sample point contains three to five petrophysical logs. The core TOC varies largely, ranging from 0.16 wt % to 4.47 wt % with an average of 1.20 wt %. In addition to the conventional MLR method, four supervised machine learning methods, i.e., ANN, RF, SVM, and GB are trained, validated, and tested for continuous TOC prediction using the ensemble learning approach. To ensure robust TOC prediction, an aggregated model predictor is designed by combining the four ensemble-based models. The model achieved estimation accuracy with R2 value of 87%. Careful data preparation and feature selection, reconstruction of corrupted or missing logs, and the ensemble learning implementation and optimization have improved TOC prediction accuracy significantly compared to a single model approach.