|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
Abstract Mathematical equations, based on conservation of mass and momentum, are used to determine the ECD at different depths in the wellbore. However, such equations do not consider important factors that have a influence on the ECD such as: (i) bottom hole temperature, (ii) pipe rotation and eccentricity, and (iii) wellbore roughness. Thus, discrepancy between the calculated ECDs and actual ones has been reported in the literature. This paper aims to explore how artificial intelligence (AI) and machine learning (ML) could provide real-time accurate prediction of the ECD, to have more insight and management of wellbore downhole conditions. For this purpose, a supervised ML algorithm, support vector machine (SVM), based on principal components analysis (PCA), was developed. Actual field data of Well-1 including drilling surface parameters and ECDs, measured by downhole sensors, were collected to develop a classical SVM model. The dataset was split with an 80/20 training-testing data ratio. Sensitivity analysis with different SVM parameters such as regularization parameter C, gamma, kernel type (linear, radial basis function "RBF") was performed. The performance of the model was assessed in terms of root mean square error (RMSE) and coefficient of determination (R). Afterward, PCA was applied to the dataset of Well-1 to develop an SVM model using the transformed dataset in PCA space. The performance of the model while using different numbers of principal components was evaluated. The results showed that the classical SVM with the linear kernel predicted the ECD with RMSE of 0.53 and R of 0.97 in the training set, while RMSE and R were 0.56 and 0.97 respectively in the testing set. The PCA-based SVM model, with the linear kernel and four principal components (93.53% variation of the dataset), predicted the ECD with RMSE 0.79 and R of 0.95 in the testing set.
Gamal, Hany (King Fahd University of Petroleum & Minerals, Dhahran, Eastern Province, Saudi Arabia) | Elkatatny, Salaheldin (King Fahd University of Petroleum & Minerals, Dhahran, Eastern Province, Saudi Arabia) | Abdulraheem, Abdulazeez (King Fahd University of Petroleum & Minerals, Dhahran, Eastern Province, Saudi Arabia)
The fourth industrial revolution and its vision for developing and governing the technologies supported artificial intelligence (AI) applications in the different petroleum industry disciplines. Therefore, the objective of this paper is to use the artificial neural network (ANN) to build a model for the rate of penetration (ROP) that considers the effect of drilling parameters,formation lithology, and drill bit design on the ROP performance. The novelty in this study is addressing the influence of poly diamond crystalline (PDC) bit design as the number of blades and cutter size, bit nozzle total flowing area (TFA),and combined different drilled formations on the penetration rate. The well drilling data covered the 8-3/8" phase with more than 1000 readings for each input.The input data are the weight on bit (WOB),revolution per minute (RPM), torque (T), standpipe pressure (SPP),and mudflow rate (Q), mud weight (MWin), gamma-ray (GR), bit design codes as the number of blades and cutter size, bit nozzle, and total flowing area (TFA).The data training to testing ratio was 70: 30%. Another data set from the same filed was used to validate the model and the results showed high accuracy for the ANN-ROP model. The model provides a high performance and accuracy level with correlation coefficient (R) of 0.99, 0.98, and 0.98 and an average absolute percentage error (AAPE) of 4.36 %, 7.06 %, and 8.14 % for training, testing, and validating respectively.
Mahmoud, Ahmed Abdulhamid (King Fahd University of Petroleum & Minerals) | Elkatatny, Salaheldin (King Fahd University of Petroleum & Minerals) | Al-AbdulJabbar, Ahmad (King Fahd University of Petroleum & Minerals) | Moussa, Tamer (King Fahd University of Petroleum & Minerals) | Gamal, Hany (King Fahd University of Petroleum & Minerals) | Shehri, Dhafer Al (King Fahd University of Petroleum & Minerals)
ABSTRACT This study introduces an empirical equation for estimation of the rate of penetration (ROP) while horizontally drilling carbonate formations based on the surface measurable drilling parameters, well log data, and the extracted weights and biases of an optimized artificial neural networks (ANN) model. The ANN model was trained using 3000 datasets of different surface measurable drilling parameters including the torque, rotation speed, and weight-on-bit, with the conventional well log data of the deep resistivity, gamma-ray, and formation bulk density, and their corresponding ROP, the self-adaptive differential evolution algorithm was applied to optimize the ANN model's design parameters. For the training dataset, the ROP was predicted with the optimized ANN model with an average absolute percentage error (AAPE) and a correlation coefficient (R) of 5.12% and 0.960, respectively. The developed empirical equation was tested on another unseen dataset (531 data points) collected from the same training well; where it predicted the ROP with AAPE of 5.80% and R of 0.951. 1. INTRODUCTION The total cost of drilling a hydrocarbon well is time-dependent (Lyons and Plisga, 2004). Rig time, which is affected by many factors, such as rate of penetration (ROP), is considered the most critical parameter for determining the total cost of drilling. Optimizing ROP has a significant impact on reducing the total cost (Barbosa et al., 2019). ROP is affected by several parameters, which can be categorized into controllable and uncontrollable parameters (Hossain and Al-Majed, 2015). The controllable parameters include weight-on-bit (WOB), rotation speed (RPM), pumping rate (GPM), torque (T), and standpipe pressure (SPP) (Eren and Ozbayoglu, 2010; Payette et al., 2017). All abbreviations are listed in Appendix A. The uncontrollable parameters include bit size and drilling fluid type, density, and rheological properties. The uncontrollable parameters affect each other, which complicates the quantification of their effect on ROP (Osgouei, 2007).
Ahmed, Abdulmalek (King Fahd University of Petroleum & Minerals) | Elkatatny, Salaheldin (King Fahd University of Petroleum & Minerals) | Ali, Abdelwahab (King Fahd University of Petroleum & Minerals) | Abdulraheem, Abdulazeez (King Fahd University of Petroleum & Minerals)
Drilling high-pressure high-temperature (HPHT) wells lead to many difficulties and issues. One of the most difficulties during the drilling is the loss of circulation. 40% of the drilling's cost is belong to the drilling fluid, so the loss of these fluid causes an increasing in the total drilling operation's cost. Circulation loss has many consequences related to well control which can lead to the worst case of blowout. There are several approaches to avoid loss of return, one of these approaches is preventing the occurrence of the losses by identifying the lost circulations zones. However, most of these approaches are difficult to be applied due to some constraints in the field.
The purpose of this work is to apply two artificial intelligence (AI) techniques to identify the zones of lost circulation. A data of real-time surface drilling parameters from three wells were obtained using real-time drilling sensors. The two methods of AI are functional networks (FN) and artificial neural networks (ANN). Well (A) was utilized to build the two AI models by dividing their data into training and testing. Then, well (B) was utilized to validate the developed AI models.
A high accuracy was achieved by the two AI models based on the root mean squared error (RMSE), confusion matrix and correlation coefficient (R). Both AI models identified the zones of lost circulation in Well A with high accuracy where the R is more than 0.98 and RMSE is less than 0.09. ANN is the most accurate model with R = 0.99 and RMSE = 0.05. Moreover, ANN was able to predict the lost circulation zones in the unseen well B (R = 0.952 and RMSE = 0.155).
Alsabaa, A. (King Fahd University of Petroleum & Minerals) | Gamal, H. A. (King Fahd University of Petroleum & Minerals) | Elkatatny, S. M. (King Fahd University of Petroleum & Minerals) | Abdulraheem, A. (King Fahd University of Petroleum & Minerals)
ABSTRACT Tracking the rheological properties of the drilling fluid is a key factor for the success of the drilling operation. The conventional tests that are usually be conducted by the mud engineers have limited resolution of the rheological data. The main objective of the paper is to relate the most frequent mud measurements as mud weight (MWT) and Marsh funnel viscosity (MFV) to the less frequent mud rheological measurements as plastic viscosity (PV), yield point (YP), behavior index (n), and apparent viscosity (AV). Artificial intelligence (AI) is the best tool for modeling such a large number of recorded heuristic data from which the artificial neural networks (ANN) was chosen to be the optimization method. In addition, the study developed empirical correlations for determining the mud rheological properties. 369 real field measurements were used to build the ANN model which were collected from 56 different wells during drilling operations of different sections with different sizes. The results showed a correlation coefficient (R) that exceeded 0.9 between measured and predicted values and with an average absolute percentage error (AAPE) below 8%. The correlations may track on real-time the rheological properties for all-oil mud that allows better control for the drilling operation problems. 1. INTRODUCTION During the drilling operations, drilling fluids are mainly used to provide many functions. The primary function of the drilling fluid is to control the formation pressure of the drilled zone (Knox and Jiang, 2005). In addition, the drilling fluid is used to lubricate the drill bit, carry the drilled cuttings, and format a filter cake to support the hole and other functions. From an economic point of view, the drilling fluids cost share 25–40 % of the total well drilling cost (Chilingarian et al., 1983). Therefore, designing and monitoring the drilling fluids parameters are very critical for drilling operations. Bad mud design or non-precise monitoring for the drilling fluid will cause drilling problems, and therefore, will add an extra cost to the drilling expenditures (Jenkins and Crockford, 1975; Okoro et al., 2018). Consequently, monitoring the drilling fluid on real-time will help to complete the drilling operation with a successful technical and economic program. The all-oil mud is a type of oil-based drilling fluids (OBM) and it is mainly composed of oil as a continuous phase with no water (Amani et al., 2012). The other type of OBM is the invert emulsion mud which is containing an amount of water and this is the major difference between the two types. The all-oil mud is a mud system composed of oil with a low level of dispersed additives. The field application of the all-oil system showed successful case studies with better mud rheology and hole cleaning than the invert emulsion mud system. All-oil muds are used to drill difficult wells, such as long sections of high angle and high-pressure high-temperature (HPHT) wells. The application of the all-oil system provides a higher penetration rate, excellent lubrication properties, thermal stability, improves hole cleaning, and reduces the bit wear significantly (Fraser, 1992).