ABSTRACT
Fluid losses into formations are a common operational issue that is frequently encountered when drilling across naturally or induced fractured formations. This could pose significant operational risks, such as well control, stuck pipe, and wellbore instability, which, in turn, lead to an increase in well time and cost. This research aims to use and evaluate different machine learning techniques, namely, support vector machine (SVM), random forest (RF), and K nearest neighbor (K-NN) in predicting the loss of circulation rate (LCR) while drilling using solely mechanical surface parameters and interpretation of the active pit volume readings. Actual field data of seven wells, which had suffered partial or severe loss of circulation, were used to build predictive models with an 80:20 training-to-test data ratio, while Well No. 8 was used to compare the performance of the developed models. Different performance metrics were used to evaluate the performance of the developed models. The root-mean-square error (RMSE) and correlation coefficient (R) were used to evaluate the performance of the models in predicting the LCR while drilling. The results showed that K-NN outperformed other models in predicting the LCR in Well No. 8 with an R of 0.90 and an RMSE of 0.17.
ABSTRACT
Measuring oil production rates of individual wells is important to evaluate a well's performance. Multiphase flow meters (MPFMs) and test separators have been used to estimate well production rates. Due to economic and technical issues with MPFMs, especially for high gas-oil ratio (GOR) reservoirs, the use of a choke formula for estimating well production rate is still popular. The objective of this study is to implement different artificial intelligence (AI) techniques to predict the oil rate through wellhead chokes. Support-vector machine (SVM) and random forests (RF) were used to generate different models to predict the production rates for high GOR and WC wells. A set of data (548 wells) was obtained from oil fields in the Middle East. GOR varied from 1000 to 9351 scf/stb, and WC ranged from 1 to 60%. Around 300 wells were flowing under critical flow conditions, while the rest were subcritical. Hence, two cases were studied using each AI model. Seventy percent of the data was used to train both RF and SVM models, while 30% of the data was used to test and validate these models. The developed RF and SVM models were then compared against the previous empirical formulas. The RF model in both critical and subcritical flow conditions was able to perfectly match the actual oil rates. SVM was able to predict the general trend for the oil rates but missed some of the sharp changes in the oil rate trend. The average absolute percent error (AAPE) values in the subcritical flow for SVM and RF were 1.7 and 0.7%, respectively, while in the critical flow, the AAPE values were 1.4 and 0.75% for SVM and RF models, respectively. SVM and RF models outperform the published formulas by 34%. The results from this study will help to estimate the real-time oil and gas rates based on the available data from wellhead chokes without the need for field intervention.
