{"title":"利用基于实时数据的机器学习算法预测等效循环密度","authors":"Abdelrahman Kandil, S. Khaled, T. Elfakharany","doi":"10.3934/energy.2023023","DOIUrl":null,"url":null,"abstract":"Equivalent circulation density (ECD) is one of the most important parameters that should be considered while designing drilling programs. With increasing the wells' deep, offshore hydrocarbon extraction, the costly daily rate of downhole measurements, operating restrictions, and the fluctuations in the global market prices, it is necessary to reduce the non-productive time and costs associated with hole problems resulting from ignoring and incorrect evaluation of ECD. Therefore, optimizing ECD and selecting the best drilling parameters are curial tasks in such operations. The main objective of this work is to predict ECD using three machine learning algorithms: an artificial neural network (ANN) with a Levenberg-Marquardt backpropagation algorithm, a K neighbors regressor (knn), and a passive aggressive regressor (par). These models are based on 14 critical operation parameters that have been provided by downhole sensors during drilling operations such as annular pressure, annular temperature, and rate of penetration, etc. In the study, 4663 data points were selected and included, where 80% to 85% of the data set has been used for training and validation according to the algorithm, and the remaining data points were reserved for testing. In addition, several statistical tests were used to evaluate the accuracy of the models, including root mean square error (RMSE), correlation coefficient (R2), and mean squared error (MSE). The results of the developed models show various consistencies and accuracy, while the ANN shows a high accuracy with an R2 of nearly 0.999 for the training, validation, and testing, as well as the overall of them. The RMSE is 0.000211, 0.000253, 0.00293, and 0.00315 for overall, training, validation, and testing, respectively. This work expands the use of artificial intelligence in the gas and oil industry. The developed ANN model is more flexible in response to challenges, reduces dependence on humans, and thus, reduces the chance of human omission, as well as increasing the efficiency of operations.","PeriodicalId":45696,"journal":{"name":"AIMS Energy","volume":null,"pages":null},"PeriodicalIF":1.8000,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Prediction of the equivalent circulation density using machine learning algorithms based on real-time data\",\"authors\":\"Abdelrahman Kandil, S. Khaled, T. Elfakharany\",\"doi\":\"10.3934/energy.2023023\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Equivalent circulation density (ECD) is one of the most important parameters that should be considered while designing drilling programs. With increasing the wells' deep, offshore hydrocarbon extraction, the costly daily rate of downhole measurements, operating restrictions, and the fluctuations in the global market prices, it is necessary to reduce the non-productive time and costs associated with hole problems resulting from ignoring and incorrect evaluation of ECD. Therefore, optimizing ECD and selecting the best drilling parameters are curial tasks in such operations. The main objective of this work is to predict ECD using three machine learning algorithms: an artificial neural network (ANN) with a Levenberg-Marquardt backpropagation algorithm, a K neighbors regressor (knn), and a passive aggressive regressor (par). These models are based on 14 critical operation parameters that have been provided by downhole sensors during drilling operations such as annular pressure, annular temperature, and rate of penetration, etc. In the study, 4663 data points were selected and included, where 80% to 85% of the data set has been used for training and validation according to the algorithm, and the remaining data points were reserved for testing. In addition, several statistical tests were used to evaluate the accuracy of the models, including root mean square error (RMSE), correlation coefficient (R2), and mean squared error (MSE). The results of the developed models show various consistencies and accuracy, while the ANN shows a high accuracy with an R2 of nearly 0.999 for the training, validation, and testing, as well as the overall of them. The RMSE is 0.000211, 0.000253, 0.00293, and 0.00315 for overall, training, validation, and testing, respectively. This work expands the use of artificial intelligence in the gas and oil industry. The developed ANN model is more flexible in response to challenges, reduces dependence on humans, and thus, reduces the chance of human omission, as well as increasing the efficiency of operations.\",\"PeriodicalId\":45696,\"journal\":{\"name\":\"AIMS Energy\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":1.8000,\"publicationDate\":\"2023-01-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"AIMS Energy\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.3934/energy.2023023\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"ENERGY & FUELS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"AIMS Energy","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3934/energy.2023023","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
Prediction of the equivalent circulation density using machine learning algorithms based on real-time data
Equivalent circulation density (ECD) is one of the most important parameters that should be considered while designing drilling programs. With increasing the wells' deep, offshore hydrocarbon extraction, the costly daily rate of downhole measurements, operating restrictions, and the fluctuations in the global market prices, it is necessary to reduce the non-productive time and costs associated with hole problems resulting from ignoring and incorrect evaluation of ECD. Therefore, optimizing ECD and selecting the best drilling parameters are curial tasks in such operations. The main objective of this work is to predict ECD using three machine learning algorithms: an artificial neural network (ANN) with a Levenberg-Marquardt backpropagation algorithm, a K neighbors regressor (knn), and a passive aggressive regressor (par). These models are based on 14 critical operation parameters that have been provided by downhole sensors during drilling operations such as annular pressure, annular temperature, and rate of penetration, etc. In the study, 4663 data points were selected and included, where 80% to 85% of the data set has been used for training and validation according to the algorithm, and the remaining data points were reserved for testing. In addition, several statistical tests were used to evaluate the accuracy of the models, including root mean square error (RMSE), correlation coefficient (R2), and mean squared error (MSE). The results of the developed models show various consistencies and accuracy, while the ANN shows a high accuracy with an R2 of nearly 0.999 for the training, validation, and testing, as well as the overall of them. The RMSE is 0.000211, 0.000253, 0.00293, and 0.00315 for overall, training, validation, and testing, respectively. This work expands the use of artificial intelligence in the gas and oil industry. The developed ANN model is more flexible in response to challenges, reduces dependence on humans, and thus, reduces the chance of human omission, as well as increasing the efficiency of operations.
期刊介绍:
AIMS Energy is an international Open Access journal devoted to publishing peer-reviewed, high quality, original papers in the field of Energy technology and science. We publish the following article types: original research articles, reviews, editorials, letters, and conference reports. AIMS Energy welcomes, but not limited to, the papers from the following topics: · Alternative energy · Bioenergy · Biofuel · Energy conversion · Energy conservation · Energy transformation · Future energy development · Green energy · Power harvesting · Renewable energy