Multisensor data fusion methods for petroleum engineering applications

Abstract

Small amount of sand in oil pipelines can result in significant erosion in a very short time period. This Produced sand is a serious problem in many production situations. Installation of a system to monitor and quantify sand production from a well would be valuable to assist in optimizing well productivity and to detect sand as early as possible. We present a multi-sensor framework for sand detection. Wireless acoustic sensors are applied in networked data fusion systems for sand detection. The framework is designed to collect real time data from oil pipeline using acoustic sensors and flow analyzer. Fusion was implemented using two methods; Fuzzy Art (FA) and Moving Average Filter (MAF). A test bed was established from ten acoustic sensors. The flow rate was monitored as well in order to collect the data with the same flow rate. For each acoustic sensor the average percentage error between the observed sand rate and the actual sand rate is very high and inconsistence. However, using the fusion methods, the result shows that the average percentage error of the fusion methods is decreased.