Machine Learning Applications to Predict Surface Oil Rates for High Gas Oil Ratio Reservoirs

2021 ◽  
pp. 1-19
Author(s):  
Ahmed Farid Ibrahim ◽  
Redha Al Dhaif ◽  
Salaheldin Elkatatny ◽  
Dhafer Al-Shehri
Keyword(s):  
Machine Learning ◽  
Oil And Gas ◽  
Gas Production ◽  
Coefficient Of Determination ◽  
Accurate Estimation ◽  
Support Vector ◽  
Critical Flow ◽  
Flow Conditions ◽  
Production Rates ◽  
Subcritical Flow

Abstract Well-performance investigation highly depends on the accurate estimation of its oil and gas flow rates. Testing separators and multiphase flow meters are associated with many technical and operational issues. Therefore, this study aims to implement the support vector machine (SVM), and random forests (RF) as machine learning (ML) methods to estimate the well production rate based on chokes parameters for high GOR reservoirs. Dataset of 1,131 data points includes GOR, upstream and downstream pressures (PU, and PD), choke size (D64), and actual data of oil and gas production rates. The data have GOR was up to 9,265 SCF/STB, the oil rate varied from 1,156 and 7,982 BPD. SVM and RF models were built to estimate the production rates. The ML models were trained using seventy percent of the dataset, while the models were tested and validated using thirty percent of the dataset. The dataset was classified to 622 wells that were flowing at critical flow compared to 509 wells that were flowing at subcritical conditions based on a PD/PU ratio of 0.55. Four machine learning models were developed using SVM and RF for subcritical flow and critical flow conditions. Different performance indicators were applied to assess the developed models. SVM and RF models revealed average absolute percent error (AAPE) of 1.3, and 0.7%, respectively in the case of subcritical flow conditions. For critical flow conditions, the AAPE was found to be 1.7% in the SVM model, and 0.8% in the RF model. The developed models showed a coefficient of determination (R2) higher than 0.93. All developed ML models perform better than empirical correlations. These results confirm the capabilities to predict the oil rates from the choke parameters in real-time without the requirement of instrument installation of wellsite intervention.

Prediction of Surface Oil Rates for Volatile Oil and Gas Condensate Reservoirs Using Artificial Intelligence Techniques

2021 ◽  
pp. 1-14
Author(s):  
Redha Al Dhaif ◽  
Ahmed Farid Ibrahim ◽  
Salaheldin Elkatatny
Keyword(s):  
Performance Test ◽  
Oil And Gas ◽  
Volatile Oil ◽  
Critical Flow ◽  
Well Test ◽  
Well Performance ◽  
Flow Conditions ◽  
Production Rates ◽  
Subcritical Flow ◽  
Well Production

Abstract Allocated well production rates are crucial to evaluate the well performance. Test separators and flow meters were replaced with choke formulas due to economic and technical issues special for high gas-oil ratio (GOR) reservoirs. This study implements Adaptive network-based fuzzy logic (ANFIS), and functional networks (FN) techniques to predict the oil rate through wellhead chokes. A set of data containing 1,200 wells was obtained from actual oil fields in the Middle East. The dataset included GOR, upstream and downstream pressure, choke size and actual oil and gas rates based on well test. GOR varied from 1,000 to 9,265 scf/stb, while oil rates ranged between 1,156 and 7,982 stb/d. Around 650 wells were flowing under critical flow conditions, while the rest were subcritical. Seventy percent of the data was used to train the AI models, while thirty percent of the data was used to test and validate these models. The developed AI models were then compared against the previous formulas. For subcritical flow conditions, rate prediction was correlated to both upstream and downstream pressures. While at critical flow conditions, changes in the downstream pressure did not affect the prediction of the production rates. For each AI method, two models were developed for subcritical flow and critical flow conditions. The average absolute percent error (AAPE) in the case of subcritical flow for ANFIS and FN were 0.88, and 1.01%, respectively. While in the case of critical flow, the AAPE values were 1.07, and 1.3% for ANFIS and FN models, respectively. All developed AI models outperform the published formulas, where the AAPE values for published formulas were higher than 34%. The results from this study will greatly assist petroleum engineers to predict the oil and gas rates based on available data from wellhead chokes in real-time with no need for additional operational costs or field intervention.

scholarly journals Machine Learning for Sensorless Temperature Estimation of a BLDC Motor

Sensors ◽  
2021 ◽  
Vol 21 (14) ◽  
pp. 4655
Author(s):  
Dariusz Czerwinski ◽  
Jakub Gęca ◽  
Krzysztof Kolano
Keyword(s):  
Machine Learning ◽  
Temperature Measurement ◽  
Stochastic Gradient Descent ◽  
Estimation Accuracy ◽  
Coefficient Of Determination ◽  
Percentage Error ◽  
Support Vector ◽  
Bldc Motor ◽  
Temperature Estimation ◽  
Motor Operation

In this article, the authors propose two models for BLDC motor winding temperature estimation using machine learning methods. For the purposes of the research, measurements were made for over 160 h of motor operation, and then, they were preprocessed. The algorithms of linear regression, ElasticNet, stochastic gradient descent regressor, support vector machines, decision trees, and AdaBoost were used for predictive modeling. The ability of the models to generalize was achieved by hyperparameter tuning with the use of cross-validation. The conducted research led to promising results of the winding temperature estimation accuracy. In the case of sensorless temperature prediction (model 1), the mean absolute percentage error MAPE was below 4.5% and the coefficient of determination R2 was above 0.909. In addition, the extension of the model with the temperature measurement on the casing (model 2) allowed reducing the error value to about 1% and increasing R2 to 0.990. The results obtained for the first proposed model show that the overheating protection of the motor can be ensured without direct temperature measurement. In addition, the introduction of a simple casing temperature measurement system allows for an estimation with accuracy suitable for compensating the motor output torque changes related to temperature.

scholarly journals Prediction of Healing Performance of Autogenous Healing Concrete Using Machine Learning

Materials ◽  
2021 ◽  
Vol 14 (15) ◽  
pp. 4068
Author(s):  
Xu Huang ◽  
Mirna Wasouf ◽  
Jessada Sresakoolchai ◽  
Sakdirat Kaewunruen
Keyword(s):  
Machine Learning ◽  
Search Algorithm ◽  
Weather Conditions ◽  
Prediction Performance ◽  
Machine Learning Algorithms ◽  
Coefficient Of Determination ◽  
Gradient Boosting ◽  
Support Vector ◽  
Self Healing ◽  
Artificial Neural Network Ann

Cracks typically develop in concrete due to shrinkage, loading actions, and weather conditions; and may occur anytime in its life span. Autogenous healing concrete is a type of self-healing concrete that can automatically heal cracks based on physical or chemical reactions in concrete matrix. It is imperative to investigate the healing performance that autogenous healing concrete possesses, to assess the extent of the cracking and to predict the extent of healing. In the research of self-healing concrete, testing the healing performance of concrete in a laboratory is costly, and a mass of instances may be needed to explore reliable concrete design. This study is thus the world’s first to establish six types of machine learning algorithms, which are capable of predicting the healing performance (HP) of self-healing concrete. These algorithms involve an artificial neural network (ANN), a k-nearest neighbours (kNN), a gradient boosting regression (GBR), a decision tree regression (DTR), a support vector regression (SVR) and a random forest (RF). Parameters of these algorithms are tuned utilising grid search algorithm (GSA) and genetic algorithm (GA). The prediction performance indicated by coefficient of determination (R2) and root mean square error (RMSE) measures of these algorithms are evaluated on the basis of 1417 data sets from the open literature. The results show that GSA-GBR performs higher prediction performance (R2GSA-GBR = 0.958) and stronger robustness (RMSEGSA-GBR = 0.202) than the other five types of algorithms employed to predict the healing performance of autogenous healing concrete. Therefore, reliable prediction accuracy of the healing performance and efficient assistance on the design of autogenous healing concrete can be achieved.

scholarly journals Modeling of Aboveground Biomass with Landsat 8 OLI and Machine Learning in Temperate Forests

Forests ◽  
2019 ◽  
Vol 11 (1) ◽  
pp. 11
Author(s):  
Pablito M. López-Serrano ◽  
José Luis Cárdenas Domínguez ◽  
José Javier Corral-Rivas ◽  
Enrique Jiménez ◽  
Carlos A. López-Sánchez ◽  
...  
Keyword(s):  
Machine Learning ◽  
Aboveground Biomass ◽  
Goodness Of Fit ◽  
Accurate Estimation ◽  
Support Vector ◽  
Landsat 8 ◽  
Sensing Applications ◽  
Learning Techniques ◽  
Physical Variables ◽  
Selection Of

An accurate estimation of forests’ aboveground biomass (AGB) is required because of its relevance to the carbon cycle, and because of its economic and ecological importance. The selection of appropriate variables from satellite information and physical variables is important for precise AGB prediction mapping. Because of the complex relationships for AGB prediction, non-parametric machine-learning techniques represent potentially useful techniques for AGB estimation, but their use and comparison in forest remote-sensing applications is still relatively limited. The objective of the present study was to evaluate the performance of automatic learning techniques, support vector regression (SVR) and random forest (RF), to predict the observed AGB (from 318 permanent sampling plots) from the Landsat 8 Landsat 8 Operational Land Imager (OLI) sensor, spectral indexes, texture indexes and physical variables the Sierra Madre Occidental in Mexico. The result showed that the best SVR model explained 80% of the total variance (root mean square error (RMSE) = 8.20 Mg ha−1). The variables that best predicted AGB, in order of importance, were the bands that belong to the region of red and near and middle infrared, and the average temperature. The results show that the SVR technique has a good potential for the estimation of the AGB and that the selection of the model hyperparameters has important implications for optimizing the goodness of fit.

scholarly journals Investigating the Applications of Machine Learning Techniques to Predict the Rock Brittleness Index

2020 ◽  
Vol 10 (5) ◽  
pp. 1691 ◽  
Author(s):  
Deliang Sun ◽  
Mahshid Lonbani ◽  
Behnam Askarian ◽  
Danial Jahed Armaghani ◽  
Reza Tarinejad ◽  
...  
Keyword(s):  
Machine Learning ◽  
Point Load ◽  
P Wave ◽  
Machine Learning Techniques ◽  
Brittleness Index ◽  
Coefficient Of Determination ◽  
Support Vector ◽  
Ann Model ◽  
Learning Techniques ◽  
Rock Brittleness

Despite the vast usage of machine learning techniques to solve engineering problems, a very limited number of studies on the rock brittleness index (BI) have used these techniques to analyze issues in this field. The present study developed five well-known machine learning techniques and compared their performance to predict the brittleness index of the rock samples. The comparison of the models’ performance was conducted through a ranking system. These techniques included Chi-square automatic interaction detector (CHAID), random forest (RF), support vector machine (SVM), K-nearest neighbors (KNN), and artificial neural network (ANN). This study used a dataset from a water transfer tunneling project in Malaysia. Results of simple rock index tests i.e., Schmidt hammer, p-wave velocity, point load, and density were considered as model inputs. The results of this study indicated that while the RF model had the best performance for training (ranking = 25), the ANN outperformed other models for testing (ranking = 22). However, the KNN model achieved the highest cumulative ranking, which was 37. The KNN model showed desirable stability for both training and testing. However, the results of validation stage indicated that RF model with coefficient of determination (R2) of 0.971 provides higher performance capacity for prediction of the rock BI compared to KNN model with R2 of 0.807 and ANN model with R2 of 0.860. The results of this study suggest a practical use of the machine learning models in solving problems related to rock mechanics specially rock brittleness index.

scholarly journals Heuristic methods applied in reference evapotranspiration modeling

2018 ◽  
Vol 42 (3) ◽  
pp. 314-324 ◽  
Author(s):  
Daniel Althoff ◽  
Helizani Couto Bazame ◽  
Roberto Filgueiras ◽  
Santos Henrique Brant Dias
Keyword(s):  
Reference Evapotranspiration ◽  
Performance Criteria ◽  
Superior Performance ◽  
Weather Data ◽  
Coefficient Of Determination ◽  
Accurate Estimation ◽  
Support Vector ◽  
Computational Techniques ◽  
Heuristic Methods ◽  
Scarce Data

ABSTRACT The importance of the precise estimation of evapotranspiration is directly related to sustainable water usage. Since agriculture represents 70% of Brazil’s water consumption, adequate and efficient application of water may reduce the conflicts over the use of water among the multiple users. Considering the importance of accurate estimation of evapotranspiration, the objective of the present study was to model and compare the reference evapotranspiration from different heuristic methodologies. The standard Penman-Monteith method was used as reference for evapotranspiration, however, to evaluate the heuristic methodologies with scarce data, two widely known methods had their performances assessed in relation to Penman-Monteith. The methods used to estimate evapotranspiration from scarce data were Priestley-Taylor and Thornthwaite. The computational techniques Stepwise Regression (SWR), Random Forest (RF), Cubist (CB), Bayesian Regularized Neural Network (BRNN) and Support Vector Machines (SVM) were used to estimate evapotranspiration with scarce and full meteorological data. The results show the robustness of the heuristic methods in the prediction of the evapotranspiration. The performance criteria of machine learning methods for full weather data varied from 0.14 to 0.22 mm d-1 for mean absolute error (MAE), from 0.21 to 0.29 mm d-1 for root mean squared error (RMSE) and from 0.95 to 0.99 coefficient of determination (r²). The computational techniques proved superior performance to established methods in literature, even in scenarios of scarce variables. The BRNN presented the best performance overall.

Effects of Pipe Diameter and Stokes Number on Erosion in Elbows

2020 ◽  
Author(s):  
Soroor Karimi ◽  
Alireza Asgharpour ◽  
Elham Fallah ◽  
Siamack A. Shirazi
Keyword(s):  
Experimental Data ◽  
Oil And Gas ◽  
Learning Algorithm ◽  
Large Diameter ◽  
Stokes Number ◽  
Gas Production ◽  
Flow Conditions ◽  
Cfd Simulations ◽  
Oil And Gas Production ◽  
Large Diameter Pipes

Abstract Large diameter pipes and elbows are vastly used in industry especially in mining and oil and gas production. Solid particle erosion is a common issue in these pipelines, and it is important to predict it to avoid failures. Currently, laboratory experiments reported in the literature are limited to diameters less than 4 inches. Therefore, there is not much experimental data available for large diameter elbows. However, the erosion can be predicted by CFD simulations and applying erosion equations in such elbows. The goal of this project is to examine the effects of elbow diameter and Stokes number on erosion patterns and magnitude for various flow conditions for elbow diameters of 2, 4, 8, and 12 inches. The approach of this work is to first perform CFD simulations of liquid-solid and gas-solid flows in 2-inch and 4-inch elbows, respectively, and evaluate the results by available experimental data. Then CFD simulations are carried for 2, 4, 8, and 12-inch standard elbows for various Stokes numbers corresponding to gas dominant flows with the velocity of 30 m/s, and liquid dominant flows with the velocities of 6 m/s. For gas dominant flows erosion in air and for liquid dominant flows erosion in water is investigated. All these simulations are carried for four particle sizes of 25, 75, 150, and 300 microns. The results indicate that Stokes number and diameter of elbows have significant effects on erosion patterns as well as magnitudes in this geometry. This work will have various applications, including validating mechanistic models of erosion predictions in elbows and developing an Artificial Intelligence (machine learning) algorithm to predict erosion for various flow conditions. Such algorithms are limited to the range of conditions they are trained for. Therefore, it is important to expand the database these codes are accessing. Overall, the CFD database of large diameter elbows will reduce the computational costs in the future.

scholarly journals Air Quality Index and Air Pollutant Concentration Prediction Based on Machine Learning Algorithms

2019 ◽  
Vol 9 (19) ◽  
pp. 4069 ◽  
Author(s):  
Huixiang Liu ◽  
Qing Li ◽  
Dongbing Yu ◽  
Yu Gu
Keyword(s):  
Machine Learning ◽  
Air Pollution ◽  
Air Quality ◽  
Regression Models ◽  
Quality Index ◽  
Air Pollutant ◽  
Air Quality Index ◽  
Machine Learning Algorithms ◽  
Coefficient Of Determination ◽  
Support Vector

Air pollution has become an important environmental issue in recent decades. Forecasts of air quality play an important role in warning people about and controlling air pollution. We used support vector regression (SVR) and random forest regression (RFR) to build regression models for predicting the Air Quality Index (AQI) in Beijing and the nitrogen oxides (NOX) concentration in an Italian city, based on two publicly available datasets. The root-mean-square error (RMSE), correlation coefficient (r), and coefficient of determination (R2) were used to evaluate the performance of the regression models. Experimental results showed that the SVR-based model performed better in the prediction of the AQI (RMSE = 7.666, R2 = 0.9776, and r = 0.9887), and the RFR-based model performed better in the prediction of the NOX concentration (RMSE = 83.6716, R2 = 0.8401, and r = 0.9180). This work also illustrates that combining machine learning with air quality prediction is an efficient and convenient way to solve some related environment problems.

scholarly journals A COMPARISON OF MACHINE-LEARNING REGRESSION ALGORITHMS FOR THE ESTIMATION OF LAI USING LANDSAT - 8 SATELLITE DATA

2019 ◽  
Vol XLII-4/W16 ◽  
pp. 679-683
Author(s):  
V. P. Yadav ◽  
R. Prasad ◽  
R. Bala ◽  
A. K. Vishwakarma ◽  
S. A. Yadav ◽  
...  
Keyword(s):  
Machine Learning ◽  
Satellite Data ◽  
Vegetation Index ◽  
Learning Algorithms ◽  
Machine Learning Algorithms ◽  
Accurate Estimation ◽  
Support Vector ◽  
Landsat 8 ◽  
Area Index ◽  
Global Circulation Models

Abstract. The leaf area index (LAI) is one of key variable of crops which plays important role in agriculture, ecology and climate change for global circulation models to compute energy and water fluxes. In the recent research era, the machine-learning algorithms have provided accurate computational approaches for the estimation of crops biophysical parameters using remotely sensed data. The three machine-learning algorithms, random forest regression (RFR), support vector regression (SVR) and artificial neural network regression (ANNR) were used to estimate the LAI for crops in the present study. The three different dates of Landsat-8 satellite images were used during January 2017 – March 2017 at different crops growth conditions in Varanasi district, India. The sampling regions were fully covered by major Rabi season crops like wheat, barley and mustard etc. In total pooled data, 60% samples were taken for the training of the algorithms and rest 40% samples were taken as testing and validation of the machinelearning regressions algorithms. The highest sensitivity of normalized difference vegetation index (NDVI) with LAI was found using RFR algorithms (R2 = 0.884, RMSE = 0.404) as compared to SVR (R2 = 0.847, RMSE = 0.478) and ANNR (R2 = 0.829, RMSE = 0.404). Therefore, RFR algorithms can be used for accurate estimation of LAI for crops using satellite data.

Sign in / Sign up

Export Citation Format

Share Document