B.J. Services - Lab Prediction
Below is a case study that shows EXC’s ability to develop an effective predictive analytics and data analysis solution for lab prediction needs.
Problem and Project Goal
When a company uses acid fracturing to break up a well, they typically need to use a chemical package to prevent corrosion on the metal pipe when they pump acid through it. B.J. Services develops the necessary chemical packages for that process.
Unfortunately, only specific packages will work under certain well conditions, so B.J. Services must perform several tests for each well to make sure they provide the right package. This process is time and money-intensive, so B.J. Services wanted a way to streamline the process and remove the testing bottleneck. They turned to EXC for a solution.
EXC Solution
EXC met with B.J. Services' research team and discussed the problem in great detail. Based on information we gathered, we realized that statistical analysis would not be enough. Acid corrosion data typically involves hundreds of thousands of data combinations, each of which have non-linear relationships.
In order to overcome these obstacles and provide a good simulation of corrosion chemistry, we used machine learning techniques to automatically determine and mimic corrosion behavior. Our end solution provided 80 - 99% accuracy when predicting if a specific corrosion test would fail or pass.
ROI Benefit
EXC's solution allows B.J. Services' personnel to make quick, accurate decisions on corrosion packages without the need for costly, time-consuming lab tests. The solution should save B.J. Services hundreds of thousands of dollars in the next few years.