Detecting Faults in Triplex Reciprocating Pumps with Synthetic Data Generated Using Simulink

Triplex reciprocating pumps are often used in the oil and gas industries. Due to their moving parts, pumps are more prone to failure compared to static equipment such as pumps and vessels.

Predictive maintenance enables engineers to service their process equipment at an optimal time whereas preventive maintenance takes a more conservative approach to maintenance scheduling which increases maintenance costs. To develop a predictive maintenance algorithm, raw equipment data is needed to extract condition indicators. To make fault detection models more robust, having enough data representing different fault types is a prerequisite. If such data is not readily available, it can be generated synthetically.

In this work, with a Simulink model, we generated fault and healthy data for a triplex reciprocating pump and trained classifiers to classify fault. Fault data includes cylinder leaks, blocked inlet, and increased bearing friction. For each fault type, fault severity ranged from no fault to a significant fault. For different combinations of fault parameter values, simulations were run in parallel to produce a large dataset of pump output pressure, output flow, motor speed, and current.

From the pump output flow, condition indicators are extracted to train a classifier to detect pump faults. Some fault conditions such as blocking fault and bearing fault were misclassified as no-fault at fault values close to nominal values. Overall, the validation accuracy was 66% and the accuracy to predict that there is a fault was 94%.