The standard torque and drag (T&D) modeling programs have been extensively used in the oil and gas industry to predict and monitor the T&D forces. In the majority of cases there has been variability in the accuracy between the pre-calculated (based on a T&D model) and actual T&D values, because of the dependence of the model's predictability on guessed inputs (matching parameters) which may not be correctly predicted. Therefore, to have a reliable model, program users must alter the model inputs, and mainly the friction coefficient to match the actual T&D. This, however, can conceal downhole conditions such as cutting beds, tight holes and sticking tendencies. The objective of this study is to develop an intelligent machine to predict the continuous profile of the surface drilling torque to enable detection of operational problems ahead of time. This paper details the development and evaluation of an intelligent system which could promote safer operation and extend the response time limit to prevent undesired events. Actual field data of Well-1, starting from the time of drilling a 5-7/8-inch horizontal section until one day prior to the stuck pipe incident, was used to train and test three models: random forest, artificial neural network, and functional network, with an 80/20 training-to-testing data ratio, to predict the surface drilling torque. The independent variables for the model are the drilling surface parameters, namely: flow rate (Q), hook load (HL), rate of penetration (ROP), rotary speed (RS), standpipe pressure (SPP), and weight-on-bit (WOB). The prediction capability of the models was evaluated in terms of correlation of coefficient (R) and average absolute error percentage (AAPE). The model with the highest R and lowest AAPE was selected to continue with the analysis to detect downhole abnormalities. The best developed model was used to predict the surface drilling torque in the last day leading up to the incident in Well-1, which represents the normal and healthy trend. Then the model was coupled with a multivariate metric distance called “Mahalanobis” to be used as a classification tool to measure how close an actual observation is to the predictive normal and healthy trend. Based on a pre-determined threshold, each actual observation was labeled “NORMAL” or “ANOMAL”. Well-2 with a stuck pipe incident was used to assess the capability of the developed system in detecting downhole abnormalities. The results showed that in Well-1, where a stuck pipe incident was reported, a continuous alarm was detected by the developed system nine hours before the drilling crew observed any abnormality, while the alarm was detected seven hours prior to any observation by the crew in Well-2. The developed intelligent system could help the drilling crew to detect downhole abnormalities in real time, react and take corrective action to mitigate the problem promptly.
A continuous growth in the global economy and population requires a sustainable energy supply. Maximizing recovery factor out of the naturally occurring hydrocarbons resources has been an active area of continuous development to meet the globally increasing demand for energy. Coalbed methane (CBM), which is one of the primary resources of natural gas, associates complex storage mechanisms and requires some advanced recovery techniques, rendering conventional reserve assessment methods insufficient. This work presents a literature review on CBM in different aspects. This includes rock characteristics such as porosity, permeability, adsorption capacity, adsorption isotherm, and coal classification. In addition, CBM reservoirs are compared to conventional reservoirs in terms of reservoir quality, reservoir properties, accumulation, and water/gas saturation and production. Different topics that contribute to the production of CBM reservoirs are also discussed. This includes production mechanisms, well spacing, well completion, and petrophysical interpretations. The main part of this work sheds a light on the available techniques to determine initial-gas-in-place in CBM reservoirs such as volumetric, decline curve, and material balance. It also presents the pros and cons of each technique. Lastly, common development and economic challenges in CBM fields are listed in addition to environmental concerns.
The sonic data provides significant rock properties that are commonly used for designing the operational programs for drilling, rock fracturing, and development operations. The conventional methods for acquiring the rock sonic data in terms of compressional and shear slowness (ΔTc and ΔTs) are considered costly and time-consuming operations. The target of this paper is to proposed machine learning models for predicting the sonic logs from the drilling data in real-time. Decision tree (DT) and random forest (RF) were employed as train-based algorithms for building the sonic prediction models for drilling complex lithology rocks that have limestone, sandstone, shale, and carbonate formations. The input data for the models include the surface drilling parameters to predict the shear and compressional slowness. The study employed data set of 2888 data points for building and testing the model, while another collected 2863 data set was utilized for further validation for the sonic models. Sensitivity investigations were performed for DT and RF models to confirm optimal accuracy. The correlation of coefficient (R), and average absolute percentage error (AAPE) were used to check the models' accuracy between the actual values and models` outputs, in addition to, the sonic log profiles. The results indicated that the developed sonic models have a high capability for the sonic prediction from the drilling data as DT model recorded R higher than 0.967 and AAPE less than 2.76% for ΔTc and ΔTs models, while RF showed R higher than 0.991 with AAPE less than 1.07%. The further validation process for the developed models indicated the great results for the sonic prediction and RF model outperformed DT models as RF showed R higher than 0.986 with AAPE less than 1.12% while DT prediction recorded R greater than 0.93 with AAPE less than 1.95%. The sonic prediction through the developed models will save the cost and time for acquiring the sonic data through the conventional methods and will provide real-time estimation from the drilling parameters.
The rock unconfined compressive strength (UCS) is one of the key parameters for geomechanical and reservoir modeling in the petroleum industry. Obtaining the UCS by conventional methods such as experimental work or empirical correlation from logging data are time consuming and highly cost. To overcome these drawbacks, this paper utilized the help of artificial intelligence (AI) to predict (in a real-time) the rock strength from the drilling parameters using two AI tools. Random forest (RF) based on principal component analysis (PCA), and functional network (FN) techniques were employed to build two UCS prediction models based on the drilling data such as weight on bit (WOB), drill string rotating-speed (RS), drilling torque (T), stand-pipe pressure (SPP), mud pumping rate (Q), and the rate of penetration (ROP). The models were built using 2,333 data points from well (A) with 70:30 training to testing ratio. The models were validated using unseen data set (1,300 data points) of Well (B) which is located in the same field and drilled across the same complex lithology. The results of the PCA-based RF model outperformed the FN in terms of correlation coefficient (R) and average absolute percentage error (AAPE). The overall accuracy for PCA-based RF was R of 0.99 and AAPE of 4.3 %, and for FN yielded R of 0.97 and AAPE of 8.5%. The validation results showed that R was 0.99 for RF and 0.96 for FN, while the AAPE was 4 and 7.9 % for RF and FN models, respectively. The developed PCA-based RF and FN models provide an accurate UCS estimation in real-time from the drilling data, saving time and cost and enhancing the well stability by generating UCS log from the rig drilling data.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.