Data driven analysis in oil and gas operations - Datadrevne analysemetoder i olje- og gassproduksjon
Abstract
Knowledge about the production system and models of relevant parts of the productionnetwork can improve the decision-making process in offshore oil and gas production. Thisthesis investigates how multivariate projection methods may be used to analyze historicalproduction data for monitoring and production optimization purposes. Two multivariateprojection methods, principal component analysis and partial least squares regression, areused to analyze a data set from the process historian of an offshore production system onthe Norwegian continental shelf.The results presented in this thesis show that the methods may assist production optimizationin several ways. Both methods are robust with respect to correlated variables,noise and missing values, and the methods are well suited for exploratory analysis ofhistorical production data. Furthermore, principal component analysis may be used as amonitoring tool for detecting abnormal operating conditions, and partial least squares regressionmay be used to predict individual flow rates from the wells and the total flowrates from the platform with choke measurements and pressure measurement as explanatoryvariables. Examples of relevant applications are illustrated and discussed.