刊·见 | 石油科学SCI期刊，探索能源需求下的技术突破

Drilling rate index (DRI) is a fundamental parameter in the investigation of rock drillability, as drillability is considered one of the main problems in rock engineering. Several researchers have continuously tried to analyze and correlate rock DRI, but the problem remains unchanged. This study elucidates the machine learning approaches, namely long short term memory (LSTM), simple recurrent neural network (RNN) and random forest algorithm (RFA) to predict DRI of rocks using multivariate inputs, that is, uniaxial compressive strength in MPa; Brazilian tensile strength (BTS) in MPa; brittleness value (S20); Sievers’ J value (Sj); modulus ratio (MR); shore hardness (SH), porosity (n) in %; shimazeks F abrasitivity in N/mm; and equivalent quartz content in %. For all proposed methods, the original dataset was divided into 70% for training and the remaining 30% for testing. Next, the performance indices, such as correlation coefficient (R2), root mean square error (RMSE), variance accounts for (VAF) and a-20 index of each proposed method were determined to examine the accuracy of the predicted data. In this study, according to the results of LSTM, simple RNN and RFA methods, the LSTM revealed the best prediction output for DRI with the strongest R2, the lowest RMSE, the largest VAF and an appropriate a-20 index values as 0.999, 0.13416, 0.997, and 0.999 in the training stage and 0.998, 0.19479, 0.996, and 0.997 in the testing stage, respectively. Therefore, LSTM is an applicable machine learning approach that can be applied to accurately predict the DRI.

The radius of investigation (ROI) is an essential parameter in many reservoir engineering applications. Inconsistencies and ambiguities still exist between different ROI evaluation methods, especially for the dual-porosity (DP) system representing naturally fractured reservoirs. This paper revisited the impact of inter-porosity flow in the DP system and presented ROI equations in the early and late flow periods based on the Laplace space solution of the Warren-Root dual-porosity model. A new comprehensive ROI expression for the DP system is derived which accounts for both storativity ratio (ω) and inter-porosity flow coefficient (λ). The new method in this paper reveals that the log-log plot of ROI vs. time for DP models exhibits two parallel straight lines linked by a smooth transition curve. The distance of these two straight lines is dependent on ω, and ROI in the intermediate flow period is highly dependent on the value of λ. The new ROI equation for the DP system can degenerate to similar ROI formula for the single-porosity system in early and late flow periods. Comparison analysis of the new method with other approaches is conducted and the validation of the new method is realized by analytical and numerical matching.

Predicting shale gas production is challenging due to varying and unclear influencing factors. In this work, we explore the average daily production rate (ADPR) and its determinants by analyzing data from 119 horizontal wells in the Weiyuan block. Initial data analysis revealed weak Spearman correlations between ADPR and geological and engineering parameters. Then, we develop four feed-forward deep learning models to predict ADPR and compare them. One of the proposed models utilizes geological and engineering data while the other three models utilize additional early-stage production data. The model with test-stage gas production can reach R2>0.95 while the model without production data can also reach R2>0.88, indicating ADPR can be efficiently predicted using only geological and engineering parameters. Moreover, the multiplication of the thickness of the target formation and the drilled length in the high-quality reservoir, as well as total organic carbon content (TOC), are the two most important influencing factors of ADPR apart from test-stage gas production. In contrast, the first-month flowback ratio is not important and cannot improve model performance. These findings can provide experts with theoretical suggestions for shale gas development. The workflow is also beneficial to future research by repeating it on larger datasets with more parameters.