Design and Implementation of a Digital Twin for Live Petroleum Production Optimization
INTRODUCTION
High-frequency live connection to data feed from sensors on equipment offers several benefits such as facilitating human surveillance for better asset management , identification and diagnosis of abnormalities and suboptimal operation . However , normal operation does not necessarily imply optimal operation . To obtain optimal production , the asset must be operated at appropriate set-points . For dynamic assets or systems , the optimum set-point changes with time . In such cases , the live data feed associated with the equipment ’ s IIOT network can be harnessed to develop a dynamic set-point optimization mechanism .
Existing literature on Petroleum Production optimization on Artificial lift wells set-points is heavily focused on manual simulation , design and recommendation by experts , or through semiautomated batch implementation of Evolutionary 1 , statistical or machine learning models . Current literature on Digital Twin implementations 2 in Oil and Gas present a broader picture on overall production process optimization 3 , but not on dynamic individual asset level set-point optimization . Fully automated set-point recommendation requires a data processing engine integrated with a simulation engine that can manage , process and generate large volumes of data . Current literature does not provide design details of critical individual components to implement a fully-automated data processing and simulation engine . This paper attempts to address this deficiency .
Digital Twins representing systems of assets can be valuable in determining optimal operating set-points . The integration of live IIOT data-feed with a Digital twin system offers several challenges and requires a detailed design for effective implementation . The following aspects of the implementation are detailed in this paper : Data processing : profiling , clean-up , transformation and cloud-database maintenance for multiple assets , high frequency data Simulation : Automated cloud-database triggered field data relevant massive scale simulation ( 60,000 + per day ).
1
Garcia , Artur Posenato , and Vinícius Ramos Rosa . " A Genetic Algorithm for Gas Lift Optimization with Compression Capacity Limitation ." Paper presented at the SPE Latin America and Caribbean Petroleum Engineering Conference , Mexico City , Mexico , April 2012 . doi :
https :// doi . org / 10.2118 / 153175-MS
2
LaGrange , Elgonda " Developing a Digital Twin : The Roadmap for Oil and Gas Optimization ." Paper presented at the SPE Offshore Europe Conference and Exhibition , Aberdeen , UK , September 2019 . doi :
https :// doi . org / 10.2118 / 195790-MS
3
Okhuijsen , Bob , and Kevin Wade . " Real-Time Production Optimization - Applying a Digital Twin Model to Optimize the Entire Upstream Value Chain ." Paper presented at the Abu Dhabi International Petroleum Exhibition & Conference , Abu Dhabi , UAE , November 2019 . doi :
https :// doi . org / 10.2118 / 197693-MS
- 36 - March 2021