Case Study: Baker Hughes achieves $10M+ savings and 10× faster development with MathWorks MATLAB predictive maintenance solution

Baker Hughes Develops Predictive Maintenance Software for Gas and Oil Extraction Equipment Using Data Analytics and Machine Learning

Baker Hughes needed a way to predict failures of positive displacement pumps used at well sites to avoid costly downtime, spare-truck deployment, and premature part replacement. To handle up to a terabyte of high-rate sensor data and identify the signals that indicate wear, Baker Hughes worked with MathWorks and used MATLAB (along with relevant toolboxes) to develop a pump health monitoring system.

Working with a MathWorks support engineer, the team used MATLAB, Statistics and Machine Learning Toolbox, and Deep Learning Toolbox to parse proprietary binary sensor files, automate overnight processing of ~1 TB of data, perform spectral and filtering analyses, and train a neural network to predict pump failures. MathWorks’ tools enabled validated field predictions and delivered measurable impact: projected savings of more than $10 million (about a 30–40% reduction in related costs), and development time reduced roughly tenfold.

Baker Hughes

Gulshan Singh

Reliability Principal and Team Lead for Drilling Services