Data-Driven Analytics for Early Warnings in Reservoir and Well Performance

Abstract

This study is aimed to detect deviations by using an algorithm fast enough to process data and to elicit information as raw data is being gathered in order to calculate past and current performance parameters and to calculate data-driven future performance. Data manipulation is normally limited to gathering for history matching purposes, but it is not usually processed with data mining techniques to detect performance pattern changes, to discriminate between stable and non-stable periods, to detect changes of fluid density and to perform production forecasting. The algorithm is developed in R language due to the easiness to execute statistical operations with a great deal of data.