Customer Case Study: Scrap Parts Reduction
A global manufacturer sees 10x ROI from using decisionOS to explain and avoid failures in their production lines
A global manufacturer of consumer appliances and industrial components
Root cause analysis of faults in manufacturing processes
Planning, optimization, and root-cause-analysis (RCA) are critical processes in ensuring efficient production and preventing recurring issues in manufacturing. However, these processes are often time- consuming and manual, making them prone to error and requiring domain experts to spend countless hours tracing back root causes of issues.
Despite the company having an abundance of data available from process development and manufacturing, the data was high dimensional in nature and made it challenging for humans to spot relationships. As such the data was hardly used to support these processes leading to inefficiencies.
Without an efficient and automated approach, the company’s engineering team was wasting valuable time tracing back root causes of issues and rarely finding the problem. This led to increased downtimes, scrap rates, and customer returns, affecting the company’s reputation and bottom line.
1Focus on predictions
Machine learning can help forecast when a part may be scrap but cannot explain why and how to avoid it
2Open source isn't ready
Open-sourced causal discovery struggles with high-dimensional data and unobserved variables
3Extracting expert knowledge
The client wanted to combine causal discovery with the valuable knowledge from process engineers
The manufacturer partnered with causaLens to help build an automated root cause analysis system that leveraged the manufacturing data.
- Using our human guided causal discovery methodology, engineers and process experts were empowered to embed their unique knowledge of the manufacturing line into the discovery process achieving results that merge the best of data-driven approaches and expert knowledge
- The tool leverages this graph to automatically trace back the root causes of different scrap parts, batches of scrap parts or parts that were wrongly identified as not scrap
- Results are shown through a user friendly decisionApp that can be easily customized to each process line workflow
Results and Benefits
Industry experts estimate the cost of poor quality (COPQ) to be as high as 20% of revenue for manufacturers – even a 1% improvement will create enormous value.
Automated fault analysis and root cause analysis can deliver a significant return on investment for our manufacturing clients – projected to be well over 10x their investment in causaLens when applied at scale. They empower the client to address the most significant cost of quality drivers like waste & scrap, complaints and returns.
The flexibility of the platform allows the client to quickly and easily adapt the solution for new process lines and manufacturing plants – further enhancing the financial return on their causaLens investment.