Revolutionising production

Context and objectives

The project set out to build a web app to monitor and optimise a production process based on the growth of living organisms.

It was particularly ambitious, since the process had been optimised intuitively by people for a century.

Outcome

• Much easier monitoring of the production process.
• 10% cost reduction on the faulty production unit, with better output quality.
• Integration of two highly profitable production process adjustments.

Our approach

Step 1: use case framing

• Understanding of the industrial process and definition of the metrics to optimise
• Refinement of end-user needs, definition of the web app's use cases and mock-ups
• Mapping of the key production parameters to monitor closely, and identification of the advanced machine learning methods to implement

Step 2: advanced method development

• Design of a digital twin monitoring more than 100 dynamic and static parameters
• Development of a custom machine learning method to model and optimise the production process, enabling a daily assessment of production performance

Step 3: integrating recommendations

• Replacement of faulty sensors identified through the monitoring features
• Testing of the most promising process improvements, such as adjustments to input timing and quantities
• Quantitative validation of the improvements before their permanent integration