Automatic Filtering and Monitoring of Real-Time Reservoir and Production Data

Abstract

Abstract This paper addresses data filtering and compression of reservoir and production data. Up to now such filtering has been done either manually by the engineers or through filtering techniques - such as wavelets - by expert users. In both cases the data filtering and compression is quite a time consuming exercise for the engineers. It has been demonstrated that filtering and compression of data quite successfully can be done using wavelet techniques. However, automatic filtering and compression is needed in order for the wavelet techniques to be applicable for real-time reservoir and production management. The "automated wavelet" has not been available due to manual determination of filter settings and the large computational effort associated by running wavelet transform rapidly in a real-time environment. This paper presents and demonstrates a methodology for solving all these problems, allowing wavelet-based de-noising and compression of large amount of data with a minimum of non-expert engineering effort and a minimum of computational effort. The methodology has been successfully tested on several real-time reservoir and production data sources. The filtered data can be used for real-time monitoring, and we demonstrate how transients, underlying trends, noise level and slugging (oscillations) can be monitored using the real-time wavelet transform. A slug number is introduced for slug detection, while a pattern recognition approach is applied for transient detections.