Implementation of AI-models for Predictive Maintenance
Codellent
Category
Overview
AI-models developed
Re-training with MLOps implemented
Second response time
Challenge
Ship maintenance is a constant balancing act. Surfaces must be serviced to avoid wear, inefficiency, and failures, but at the same time it is both costly and disruptive to take a vessel out of operation too early. Beyond the direct costs, poorly timed maintenance can lead to significant lost revenue.
The challenge is amplified in the maritime environment, where conditions such as weather, load, and sailing patterns vary continuously. The actual condition of ships’ surfaces is therefore difficult to assess and often partially hidden. This made it challenging to plan maintenance based on actual condition rather than fixed intervals or assumptions.
Solution
For Hempel, the objective 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 transition.
The platform was built with domain-specific data models and automated data flows from relevant data sources, enabling information to be continuously updated and used in analyses. On this basis, Codellent developed AI models based on neural networks to identify patterns and predict maintenance needs. To ensure the models could operate reliably over time, an MLOps setup was established with continuous retraining as well as integrations with relevant applications and business systems.
The Result
The project has established a new foundation for working in a more systematic and data-driven way with maintenance in the maritime sector. Hempel is now able to plan service based on actual conditions rather than fixed intervals, improving both decision quality and resource utilization.
The delivery has also served as a pioneering project, demonstrating how data and AI can be applied to develop more efficient processes, strengthen insights, and support new, more flexible service solutions in practice.