Challenge details:

Slag pots are critical assets used for handling and transporting molten slag from blast furnace operations. Repeated exposure to extreme temperatures, thermal cycling, mechanical stresses, and impact loading leads to crack formation over time. Currently, inspection is carried out through manual visual observation by maintenance and operational personnel, making the process subjective, time-consuming, and prone to missed defects. Failure to identify and assess cracks at an early stage can result in slag leakage, pot failure, equipment damage, production disruption, and serious safety hazards. The plant operates approximately 20 slag pots, and crack occurrence is generally concentrated at specific recurring locations on the pot body.

We are seeking an automated, non-contact inspection solution capable of detecting, classifying, and reporting cracks on slag pots with high reliability and minimal human intervention. The solution may utilize computer vision, thermal imaging, AI/ML analytics, 3D imaging, or other suitable technologies but should not require sensors to be mounted directly on the slag pot. Automated inspection reporting, defect ranking, and digital record generation are required. Predictive failure detection is desirable; however, the primary objective is accurate crack detection and classification in a high-temperature steel plant environment, where slag pot surface temperatures may be approximately 500-700°C or less during inspection.