Use of Maintenance Data from Periodic Intervals for Maintenance Optimisation
Abstract
Application and collection of data from maintenance work as a systematic process is paramountfor the continuous improvement of any maintenance program. Teekay Petrojarl, the world slargest FPSO operator, is working to improve their maintenance reporting system. The aimof this thesis is to direct Teekay Petrojarl towards the best practice for collection and use of maintenance data.
In relation to estimation of parameters for maintenance optimisation, there are not manythat discuss the applicability of data in the literature Nevertheless, a suggestion for structuringof the data is presented. The study also presents suggestions for which data that need tobe collected from preventive and corrective maintenance.
A lifetime model, Weibull, and two degradation models, Markov and the P-F model,are presented and reviewed. These can be used to estimate parameters for maintenanceoptimisation, given that the company can support with the right data in a structured way.Challenges with using maintenance data for these models are also introduced.
The status of the reporting system in Teekay Petrojarl is also evaluated. Teekay Petrojarlhas worked to implement new categories to be used for failure reporting, but have not startedto use them yet. The evaluation reveals several weaknesses in the current system. Most ofthe existing data from the past 20 years is deemed unusable for estimation of the parametersneeded to perform maintenance optimisation. The computerised maintenance management isin addition not very flexible, which makes it a challenge to change the maintenance reportingforms.
Lastly, measures to achieve best practice are presented and discussed. Reliability centredmaintenance analyses are suggested to be done for the equipment on the vessels. Theseanalyses will aid to map the important maintainable items and their failure modes. Thesemeasures are extensive and will likely be costly, if they are prioritised. Further work forTeekay Petrojarl is suggested in order to aid the development of the maintenance reportingsystem.