Search

Determination of pore pressure (PP), a key reservoir parameter that is beneficial for evaluating geomechanical parameters of the reservoir, is so important in oil and gas fields development. Accurate estimation of PP is also essential for safe drilling of oil and gas wells since PP data are used as the input for safe mud window determination. In the present study, empirical equations along with machine learning methods, namely random forest algorithm, support vector regression (SVR) algorithm, artificial neural network (ANN) algorithm, and decision tree (DT) algorithm, are employed for PP prediction applying well log data. To this end, 2827 data records collected from three wells (Well A, Well B, and Well C) drilled in one of the Middle East oil fields are used. The dataset of Wells A and B is used for models’ training, validating, and testing, while Well C dataset is applied for evaluating the models’ generalizability in PP prediction in the field under study. To construct the predictive algorithms, 12 input variables are initially considered in the study. A feature selection analysis is conducted to find the most influential input variables set for developing PP predictive models. The results obtained suggest that the 9-input-variable set is the most efficient combination of inputs used in the ML models construction. Among all the four ML algorithms proposed, the DT algorithm presents the most accurate predictions for PP, delivering R2and RMSE values of 0.9985 and 14.460 psi, respectively. Furthermore, the model generalization analysis results reveal that the 9-input-variable DT model developed can be used for PP prediction throughout the field of study since it presented an excellent accuracy performance in predicting PP when applied to Well C dataset.

The mud weight window (MW) determination is one of the most important parameters in drilling oil and gas wells, where accurate design can secure the drilled well and deliver a stable borehole. In this paper, novel algorithms based on the most influential set of input features are developed to predict pore pressure, including rate of penetration (ROP), deep resistivity (ILD), density (RHOB), photoelectric index (PEF), corrected gamma ray (CGR), compression-wave velocity (Vp), weight on bit (WOB), shear-wave velocity (Vs) and pore compressibility (Cp). The algorithms used in this study are as follows: 1) machine learning algorithms (ML), these are the K-nearest neighbor (KNN) algorithm, weighted K-Nearest Neighbor (WKKNN), and distance weighted KNN (DWKNN); 2) hybrid machine learning algorithms (HML), which include the combination of three ML with particle swarm optimization (PSO) (KNN-PSO, WKNN-PSO and DWKNN-PSO). The 2875-record dataset used in this study was collected from three wells (S1, S2 and S3) in one of the gas reservoirs (Tabnak field) in Iran. After comparing the performance accuracy of all algorithms, DWKNN-PSO has the best performance accuracy compared to other algorithms presented in this paper (for the total dataset of wells S1 and S2: R2=0.9656and RMSE = 12.6773 psi). Finally, the generalizability of the best predictive algorithm for PP, DWKNN-PSO, is evaluated by testing the proposed algorithm on an unseen dataset from another well (S3) in the field under study, where the DWKNN-PSO algorithm provides PP predictions in well S3 with high accuracy, R2 = 0.9765 and RMSE = 9.7545 psi, confirming its ability to be used in PP prediction in the studied field.

Abstract: Shear wave velocity (VS) data from sedimentary rock sequences is a prerequisite for implementing most mathematical models of petroleum engineering geomechanics. Extracting such data by analyzing finite reservoir rock cores is very costly and limited. The high cost of sonic dipole advanced wellbore logging service and its implementation in a few wells of a field has placed many limitations on geomechanical modeling. On the other hand, shear wave velocity VStends to be nonlinearly related to many of its influencing variables, making empirical correlations unreliable for its prediction. Hybrid machine learning (HML) algorithms are well suited to improving predictions of such variables. Recent advances in deep learning (DL) algorithms suggest that they too should be useful for predicting VSfor large gas and oil field datasets but this has yet to be verified. In this study, 6622 data records from two wells in the giant Iranian Marun oil field (MN#163 and MN#225) are used to train HML and DL algorithms. 2072 independent data records from another well (MN#179) are used to verify the VSprediction performance based on eight well-log-derived influencing variables. Input variables are standard full-set recorded parameters in conventional oil and gas well logging data available in most older wells. DL predicts VSfor the supervised validation subset with a root mean squared error (RMSE) of 0.055 km/s and coefficient of determination (R2) of 0.9729. It achieves similar prediction accuracy when applied to an unseen dataset. By comparing the VSprediction performance results, it is apparent that the DL convolutional neural network model slightly outperforms the HML algorithms tested. Both DL and HLM models substantially outperform five commonly used empirical relationships for calculating VSfrom Vprelationships when applied to the Marun Field dataset. Concerns regarding the model's integrity and reproducibility were also addressed by evaluating it on data from another well in the field. The findings of this study can lead to the development of knowledge of production patterns and sustainability of oil reservoirs and the prevention of enormous damage related to geomechanics through a better understanding of wellbore instability and casing collapse problems. Graphical abstract:

One of the challenges in reservoir management is determining the fracture density (FVDC) in reservoir rock. Given the high cost of coring operations and image logs, the ability to predict FVDC from various petrophysical input variables using a supervised learning basis calibrated to the standard well is extremely useful. In this study, a novel machine learning approach is developed to predict FVDC from 12-input variable well-log based on feature selection. To predict the FVDC, combination of two networks of multiple extreme learning machines (MELM) and multi-layer perceptron (MLP) hybrid algorithm with a combination of genetic algorithm (GA) and particle swarm optimizer (PSO) has been used. We use a novel MELM-PSO/GA combination that has never been used before, and the best comparison result between MELM-PSO-related models with performance test data is RMSE = 0.0047 1/m; R2 = 0.9931. According to the performance accuracy analysis, the models are MLP-PSO < MLP-GA < MELM-GA < MELM-PSO. This method can be used in other fields, but it must be recalibrated with at least one well. Furthermore, the developed method provides insights for the use of machine learning to reduce errors and avoid data overfitting in order to create the best possible prediction performance for FVDC prediction.

Two-phase flow rate estimation of liquid and gas flow through wellhead chokes is essential for determining and monitoring production performance from oil and gas reservoirs at specific well locations. Liquid flow rate (QL) tends to be nonlinearly related to these influencing variables, making empirical correlations unreliable for predictions applied to different reservoir conditions and favoring machine learning (ML) algorithms for that purpose. Recent advances in deep learning (DL) algorithms make them useful for predicting wellhead choke flow rates for large field datasets and suitable for wider application once trained. DL has not previously been applied to predict QLfrom a large oil field. In this study, 7245 multi-well data records from Sorush oil field are used to compare the QLprediction performance of traditional empirical, ML and DL algorithms based on four influencing variables: choke size (D64), wellhead pressure (Pwh), oil specific gravity (γo) and gas–liquid ratio (GLR). The prevailing flow regime for the wells evaluated is critical flow. The DL algorithm substantially outperforms the other algorithms considered in terms of QLprediction accuracy. The DL algorithm predicts QLfor the testing subset with a root-mean-squared error (RMSE) of 196 STB/day and coefficient of determination (R2) of 0.9969 for Sorush dataset. The QLprediction accuracy of the models evaluated for this dataset can be arranged in the descending order: DL > DT > RF > ANN > SVR > Pilehvari > Baxendell > Ros > Glbert > Achong. Analysis reveals that input variable GLR has the greatest, whereas input variable D64has the least relative influence on dependent variable QL.

Pressure–Volume–Temperature (PVT) characterization of a crude oil involves establishing its bubble point pressure, which is the pressure at which the first gas bubble forms on a fluid sample while reducing pressure at a stabilized temperature. Although accurate measurement can be made experimentally, such experiments are expensive and time-consuming. Consequently, applying reliable artificial intelligence (AI)/machine learning methods to provide an accurate mathematical prediction of an oil’s bubble point pressure from more easily measured characteristics can provide valuable cost and time savings.

Fluid-flow measurements of petroleum can be performed using a variety of equipment such as orifice meters and wellhead chokes. It is useful to understand the relationship between flow rate through orifice meters (Qv) and the five fluid-flow influencing input variables: pressure (P), temperature (T), viscosity (μ), square root of differential pressure (ΔP^0.5), and oil specific gravity (SG). Here we evaluate these relationships using a range of machine-learning algorithms applied to orifice meter data from a pipeline flowing from the Cheshmeh Khosh Iranian oil field. Correlation coefficients indicate that (Qv) has weak to moderate positive correlations with T, P, and μ, a strong positive correlation with the ΔP^0.5, and a weak negative correlation with oil specific gravity. In order to predict the flow rate with reliable accuracy, five machine-learning algorithms are applied to a dataset of 1037 data records (830 used for algorithm training; 207 used for testing) with the full input variable values for the data set provided. The algorithms evaluated are: Adaptive Neuro Fuzzy Inference System (ANFIS), Least Squares Support Vector Machine (LSSVM), Radial Basis Function (RBF), Multilayer Perceptron (MLP), and Gene expression programming (GEP). The prediction performance analysis reveals that all of the applied methods provide predictions at acceptable levels of accuracy. The MLP algorithm achieves the most accurate predictions of orifice meter flow rates for the dataset studied. GEP and RBF also achieve high levels of accuracy. ANFIS and LSSVM perform less well, particularly in the lower flow rate range (i.e., <40,000 stb/day). Some machine learning algorithms have the potential to overcome the limitations of idealized streamline analysis applying the Bernoulli equation when predicting flow rate across an orifice meter, particularly at low flow rates and in turbulent flow conditions. Further studies on additional datasets are required to confirm this.

None of the various published models used to predict oil production rates through wellhead chokes from fluid composition and pressures can be considered as a universal model for all regions. Here, a model is provided to predict liquid production-flow rates for the Reshadat oil field offshore southwest Iran, applying a customized genetic optimization algorithm (GA) and standard Excel Solver non-linear and evolutionary optimization algorithms. The dataset of 182 records of wellhead choke measurements spans liquid flow rates from < 100 to 30,000 stock tank barrels/day. Each data record includes measurements of five variables: liquid production rate (QL), wellhead pressure, choke size, basic sediment and water, and gas–liquid ratio. 70% of the dataset (127 data records) was used for training purposes to establish the prediction relationships, and 30% of the dataset (55 data records) was utilized for independently testing the accuracy of the derived relationships as predictive tools. The methodology applying either the customized GA or standard Solver optimization algorithms, demonstrates significant improvements in QL-prediction accuracy with the lowest APD (− 7.72 to − 2.89), AAPD (7.33–8.51), SD (288.77–563.85), MSE (91,871–316,429), and RMSE (303.1–562.52); and the highest R2(greater than 0.997) compared to six previously published liquid flow-rate prediction models. As a general result, the novel methodology is easily applied to other field/reservoir datasets, to achieve rapid practical flow prediction applications, and is consequently of worldwide significance.