Maintenance of the machinery equipment has been traditionally carried out through the Planned Maintenance System (PMS), which is based mainly on the running hours of engine components. Although the PMS has been mostly effective in preventing major breakdowns, it can be time consuming and expensive when perfectly serviceable parts are replaced with new ones. Smart maintenance systems, on the other hand, aim to increase asset uptime and prevent critical failures by utilizing data from different sources, like sensors, inspection reports and lube oil analysis results, to assess the condition of the equipment. The goal of a smart maintenance solution is long-term cost-savings with longer-lasting, safer, and more reliable performance, more cost-efficient lifecycle, and compliance with environmental legislations.
The Ph.D. project is funded by ShippingLab and is part of Work Package 1. ShippingLab is a nonprofit innovation and project partnership in Blue Denmark with a purpose to establish Denmark as a driver of smart shipping of the future. SDU is the leader of the activity 1.3.2 “Predictive Analytics for maintenance and spare parts planning”.
Using predictive analytics in vessel engine maintenance is not an easy task. The project aims to tackle major challenges of the field such as:
- Developing a database of nominal data which are representative of different operational conditions of the vessel
- Investigating which features are the best estimators of the cylinder condition
- Comparing different Machine Learning techniques and statistical methods for training on the nominal data and predicting on live data
- Assessing the condition of the engine based on residual differences from the nominal behaviour
- Estimating the Remaining Useful Life (RUL) of critical components such as piston rings and cylinder liners
The ultimate goal is to use Smart Maintenance techniques to provide decision support to the marine engineers on-board the vessel and ashore as early as possible.