Three automated pipelines for wear reporting at FLSmidth

Challenge

FLSmidth's wear reports came together through a combination of different programs and manual handling of point clouds, 3D models and analyses across them. Data was not processed in one coherent flow but had to be pieced together across tools – which made the work time-consuming and dependent on the individual employee's experience. Identifying wear patterns, deviations and critical areas required manual inspection, and the quality of the reports therefore varied across projects and employees.

At the same time, the whole process had to run locally for data security, which limited the scope for using cloud-based or licence-paid tools to make the work more efficient.

Solution

Capacit built an automated pipeline that runs data processing and preliminary analysis across three flows, each tailored to the specific part in question (mill or crusher). It sits on the local workflows and uses only free, licence-free software, so the requirements for data security and flexibility are met without new licence costs.

How the solution adds value

The pipelines are built on classic computer vision – no AI – and automate the manual steps from raw point cloud to processed data. That makes the analysis faster and more consistent across projects and employees.