Delivery
Predictive Maintenance
Industry
Painting, Production and Maintenance
Area
Distribute maintenance insights worldwide
Impact
Real-time AI-models integrated in an App
Introduction
Optimizing maintenance in a global and complex operation.
Hempel is a global supplier of paints and surface protection solutions across a wide range of industries. The company is particularly known for its solutions in the maritime sector, where its products contribute to increased durability, lower fuel consumption, and reduced operational costs for ships. With operations in more than 80 countries, Hempel continuously works to develop new solutions that both strengthen the business and improve customer operations.
Challanges
Vedligehold baseret på antagelser frem for reel indsigt.
Vedligeholdelse af skibe er en konstant balancegang. Overflader skal serviceres for at undgå slid, ineffektivitet og fejl, men samtidig er det både dyrt og forstyrrende at tage et fartøj ud af drift for tidligt. Udover de direkte omkostninger kan forkert timet vedligeholdelse føre til betydelig tabt omsætning. Udfordringen forstærkes i det maritime miljø, hvor forhold som vejr, belastning og sejlmønstre varierer løbende. Den faktiske tilstand af skibenes overflader er derfor svær at vurdere og ofte delvist skjult. Det gjorde det vanskeligt at planlægge vedligehold baseret på reel tilstand frem for faste intervaller eller antagelser.
The Solution
Data and AI as the foundation for more precise and predictable maintenance.
Hempel’s goal was to use data and AI to shift maintenance from a reactive cost to a more predictable and value-creating discipline. In collaboration with Hempel, Codellent designed, developed, and implemented a data and AI platform in Azure to support this transformation. The platform was built using domain-specific data models and automated data pipelines from relevant sources, ensuring that information could be continuously updated and used in analysis. Based on this, Codellent developed AI models using neural networks to identify patterns and predict maintenance needs. To ensure long-term stability, an MLOps setup was established with continuous retraining, along with integrations into relevant applications and business systems.
Final thoughts
(03)
PROJECTS