By utilizing machine learning (ML) data analytics and conducting rigorous testing, VirtusLab achieved a 4.4% reduction in the client's scrap rate, improving manufacturing efficiency. This helped facilitate smarter manufacturing processes overall.
The challenge
Although well-planned, the client’s production process resulted in a high scrap rate. Furthermore, scraps obtained at particular points in production could not be reprocessed or repurposed. After several failed attempts to solve this issue, our clients asked themselves if repetitive rebalancing was worth performing at all! This was the point in time when our client contacted VirtusLab.
The solution
VirtusLab received historical data about the production of two product types and collected more data to fill in some gaps. This helped us understand the manufacturing process, its technology, and environment, bringing us closer to the root cause of the high scrap rate.
We created a machine learning model based on our client’s know-how. Every step within the production process delivered data about step-specific errors. One error stood out during the first balancing attempt in both production processes of the two product types. The data also showed a steady increase in errors before the balancing attempt.
The model delivered predictions showing how changing specific parameters could benefit production. VirtusLab found patterns to optimize production processes and indicate when a part certainly becomes scrap. According to these findings, VirtusLab suggested eliminating more products at earlier stages.
The results
Due to our consultancy, our client saw results in the first four days:
- Decreased overall production time, which also helped increase the number of produced goods.
- Enabled pinpointed analysis of specific parameters that had led to instances of production scrap.
- Scrap rate reduction by 4.4%.
- OEE increase by 4%.
Analysis
Modelling part
