The Applied AI and Digitalization in Drilling Group at IEP, Department of Energy and Petroleum Engineering at the University of Stavanger, bridges the gap between industrial domain expertise and latest advances in Artificial Intelligence.
Stemming from the Department of Energy and Petroleum Engineering, Applied Industrial AI group focuses on both problems of the present, and problems of the future. With petroleum continuing to be a backbone of the Norwegian economy we have a duty and responsibility to enable efficient, safe, and responsible operations ranging from oil discovery, drilling, production, and refining. At the same time, know-how developed for the fossil fuel industry is re-used in domains such as drilling of geothermal as well as carbon dioxide storage wells. We believe that best modern solutions involving Artificial Intelligence must leverage understanding of the underlying system, including mechanical engineering, electrical engineering, control theory and domain expertise. As AI matures and we better understand its strengths and limitations, it begins to be seen as a tool with specific advantages and disadvantages, and not the be-all and end-all solution to all the problems. This approach allows us to be closer to the industry and research highly deployable solutions.
Drilling automation
We run multiple projects related to Drilling Automation. This includes automatic event detection (kick), ROP optimization, MSE optimization, and other.
Data quality improvement
There are not data-driven methods without high quality data. We have published multiple papers focusing on data quality within the petroleum industry. This work is related to data filtering, data inputation, and more.
Predicting Senor Data
We have published a number of papers related to predicting sensor data before the actual measurement has taken place. Our case study involved a directional drilling system, which originally consisted of inclination sensor 23 meters behind the bit. With the use of latest techniques in artificial intelligence we were able to successfully predict inclination data all the way to the bit itself using just software.
Drilling simulator (Drillbotics)
Drillbotics is an international competition sponsored by SPE, where students have ability to design their own lab-sized drilling rig and compete against each other. In recent years this competition was expanded to include a development of a drilling simulator, where students can implement different models and learn how they all interact, implement their own modelling and control ideas, and more. University of Stavanger takes part in both competition paths yearly with support from the Applied Industrial AI lab at IEP.
Reinforcement learning in drilling
Reinforcement learning aims at creating an agent, that is capable of operating within a given environment. Our current research is focused on connecting that technology with drilling environment, where AI can be used to perform drilling operations in an optimized manner. Latest machine learning algorithms are being connected with the state-of-the-art drilling simulator, OpenLab, to explore drilling process automation and optimization involving complete set of potential driller actions.
Team
Department of Energy and Petroleum Engineering
Department of Electrical Engineering and Computer Science
Department of Energy and Petroleum Engineering
Department of Energy and Petroleum Engineering