Stuck Pipe Detection For North Sumatera Geothermal Drilling Operation Using Artificial Neural Network

Abstract

One of the most common problems encountered during geothermal drilling operations is stuck pipe. The risk of stuck pipe is higher in geothermal drilling operations since geothermal drilling targets the lost circulation zone at reservoir depth. The stuck pipe problem can cause a significant increase in drilling time and costs. The cost of a stuck pipe includes the time and money spent on extracting the pipe, fishing the parted BHA, and the effort required to plug and abandon the hole. Therefore preventing stuck pipes is far more cost effective than the most effective freeing procedures.  Many researchers are working to identify the symptoms to reduce the risk of a stuck pipe. Due to the complexion of stuck pipe’s symptoms and indicators, some researcher proposed artificial intelligence (AI) as the tool to predict stuck pipes. Although researches have been made to build systems employing artificial intelligence (AI) to avoid stuck pipe occurrences in oil and gas drilling operations, few works have been done for geothermal drilling operations. This paper describes a study that employed Artificial Neural Networks (ANN) approaches to predict stuck pipe incidents. Field data were collected from 6 geothermal wells drilled in North Sumatera fields. ANN was used to construct models to forecast stuck pipe incidents. The investigation found that ANN showed good performance with 84% accuracy and 74% recall for the limited training dataset. These ANN approaches provide good predictions that can help increase response time and accuracy in preventing stuck pipes.