Affiliation:
1. Federal Rural University of Rio de Janeiro
2. Petrobras
Abstract
AbstractOnline drilling fluid measurement technologies are popping up in the industry as an essential tool for drilling automation, while online density measurements are widespread, the availability of rheology measurements is increasing fast and additional properties (o/w ratio, solids content, electrical stability, filtration, etc) appear as field trials. This article presents the concept of a supervisory/ advisory systems dedicated to support the detection of abnormal events and to provide guidelines for fluid treatment actions.The proposed methodology consisted of two stages: experimental data acquisition in a flow loop and data processing for the validation of the algorithm. In the data acquisition stage, multiple properties of the drilling fluids were continuously measured by using automatic sensors. In the second stage, the drilling fluid's properties were processed in a fault detection algorithm. The algorithm used Principal Component Analysis (PCA) to train the process model through the calculation of the principal components of the steady state of the fluid, which represents the healthy state of the drilling fluid.Once the process was trained, the algorithm monitored new data samples obtained in the data acquisition stage and compared them to the trained model by calculation of the mean square prediction error (MSPE) of the model and the T² of Hoteling. Persistent changes in MSPE and T² values indicated that an anomaly was occurring in the drilling fluid. The new methodology was validated based on the data obtained in a flow loop where fluid properties were monitored using online sensor under different operational conditions. The algorithm was able to detect faults and anomalies in the drilling fluid even identifying the source of the anomalies through the decomposition of the MSPE and T² statistics. The proposed algorithm performed well in real-time conditions, pointing out that it can be used as a diagnostic tool in-field oil well drilling operations.
Reference15 articles.
1. Arghad, A., Esmael, Bilal; Fruhwirth, Rudolf. (2010). Abnormal Oil Well Drilling Operations Detection Using Smallest Principal Components. The 3rd International Conference on Computational Intelligence and Industrial Application
2. Bergh, L. G., Acosta, S. (2009). On-Line Fault Detection on a Pilot Flotation Column Using Linear PCA Models. 10th International Symposium on Process Systems Engineering: Part A, 1437–1442. doi: 10.1016/s1570-7946(09)70630-3
3. Fjetland, A. K., Zhou, J., Abeyrathna, D., Gravdal, J. E. (2019). Kick Detection and Influx Size Estimation during Offshore Drilling Operations using Deep Learning. 2019 14th IEEE Conference on Industrial Electronics and Applications (ICIEA). doi: 10.1109/iciea.2019.8833850
4. Model-based fault-detection and diagnosis – status and applications;Isermann;Annual Reviews in Control,2005
5. A user's guide to principal components. Wiley series in probability and mathematical statistics;Jackson,1991