Improving Operational Awareness through a Data and Expert Driven Advisory System

This paper presents a case study of a data- and expert driven approach to candidate recognition in production operations and reservoir optimization. The presented Advisory System is capable of capturing expert knowledge from various domains in one consistent and unbiased logic framework and applying it to newly acquired data in order to identify situations where production, injection or reservoir trends deviate from expectations. Furthermore the system uses multivariate correlations extracted from data and expert knowledge to identify reasons for these trend deviations.

The Advisory System is comprised of a process that ranges from raw data acquisition and integration, time stamping and aggregation, to data cleansing, data preprocessing and Complex Event Processing for the identification and tagging of events. The events are processed in a probabilistic expert system to yield likelihoods for the occurrence of certain problems in wells, reservoir or facilities. Based on this result the asset team is assisted in differentiating root causes and assigning respective priorities for activities and interventions. This paper will present how the Advisory System is set up, calibrated and deployed. Sample results will be provided. The underlying technologies and algorithms will be discussed, ranging from Complex Event Processing for data abstraction and event tagging to Bayesian Networks for probabilistic reasoning under uncertainty and for translation of event patterns to proactive problem detection. Also we will thoroughly elaborate on the combination of these algorithms and tools to achieve a continuously executed workflow for reservoir management and production optimization. It will be discussed how the results are used in ongoing operations and how an Advisory System like the one presented facilitates production optimization processes and improves operational decisions.

Andreas Al-Kinani

Andreas Al-Kinani is Managing Partner and Technical Director of myr:conn solutions. Previously he has been working as Petroleum Engineer with Schlumberger Information Solutions (SIS) since 2004. He is an industry wide recognized expert for production optimization and workflow automation, his interest covers reservoir management, production optimization, artificial intelligence applications in petroleum engineering as well as statistical workflows for candidate recognition in brownfields. In his projects, Andreas has integrated data mining approaches with analytical petroleum engineering techniques, and has implemented reservoir and production monitoring and workflow automation systems for multiple oil and gas companies worldwide.

He holds a MSc. Degree in Petroleum Engineering of the Mining University in Leoben, Austria.