The challenge
The quality of a refractory coating sold to steel plants depends on every step being right, from the amount of each raw material to the time and conditions of mixing. With parts of the work done by hand, each batch came out a little different from the last, and tracing the cause of a quality problem was hard.
What we built
- Control boards we designed and built ourselves to drive the line's equipment and read its sensors
- Full automation of the process, so the line runs each product's recipe without manual steps
- Every board connected to a server, recording each stage and each batch as it happens
- A web dashboard showing line status, production history and alerts
- Machine-learning models on the production data to find optimal settings and make quality more consistent
How we worked
We did the hardware, the software and the on-site commissioning. Because the boards are our own design, the line can be changed or extended with new stages without depending on a foreign supplier.
Results
- A fully automatic line with no manual steps in production
- A full, traceable record of every batch on the server
- Production settings optimised with machine learning
- Home-grown hardware that can be extended