Waste Reduction In Hard Drive Manufacturing

One of the key components of these devices is a silicone wafer, which acts as a semiconductor within the drives. These wafers are highly sensitive to their environment: factors such as temperature, air quality and air moisture can all lead to the material being compromised and unfit for use.

The manufacturing process is extremely complex and involves hundreds of separate processes, it takes approximately 60 days to turn the raw materials into the finished product. Each silicone wafer costs between £20–£50, and in some cases failure rates can reach 10%.

For these reasons, our client monitors the manufacturing process using environmental sensors, to ensure it stays within the required parameters for production. The challenge it faces, is processing the volume of data these sensors collect. Their current systems often struggle to analyse the information efficiently, which can mean by the time an environmental abnormality is detected, it is too late and the silicon wafer is damaged.

Brainpool proposed the implementation of a predictive ML system which is capable of analysing the incoming environmental data in real time, whilst forecasting any potential changes in the manufacturing environment.

In order to achieve this, Brainpool utilised a significant amount of training (historical) data from our client’s sensors which would allow the system to form an understanding of what precursors may indicate an unexpected change on the production line.

By utilising the data provided, Brainpool designed a predictive ML system which could forecast potential changes in the manufacturing environment with a goal to reduce the Client’s waste materials and improve overall efficiency.

The ML system was subsequently taken over by the Client’s team for implementation.