Page 17 - UNAM Virtual Graduation e-Book (April2021)
P. 17
FACULTY OF ENGINEERING &
INFORMATION TECHNOLOGY
DOCTOR OF PHILOSOPHY IN ENGINEERING
(MECHANICAL ENGINEERING)
CANDIDATE: KIMERA David
CURRICULUM VITAE
David Kimera is able to work independently or as part of a team. He is
time-conscious, result oriented, self-motivated, has a quick grasp of issues,
responsive to change and thrives on challenges. He is a qualified engineering
professional and can multitask. David lectured and researched in various
areas of engineering both in academia and engineering practice.
CANDIDATE’S DISSERTATION
MAINTENANCE OPTIMIZATION DYNAMICS FOR MARINE MECHANICAL SYSTEMS
The doctoral study was undertaken and completed under the Supervision of Dr. F.N. Nangolo, Faculty of
Engineering & Information Technology, of the University of Namibia.
The research presented, addresses the various aspects of maintenance practices, strategies, platforms and
optimization techniques that are currently used and proposed for the marine industry. Four themes that
underlie maintenance of marine mechanical consists, were explored. These included a critical review of
the various maintenance practices, tools and parameters for marine vehicle mechanical Plants, Machinery
and Equipment (PME) systems. Thirteen maintenance parameters were identified and, the most important
maintenance parameters are maintenance costs, reliability and safety. Maintenance models that have been
developed have been validated using one system without considering the sub-components. A reliability and
degradation analysis was carried out on aging/aged fishing vessels at ship repair yards at Walvis Bay. Reliability
and degradation results indicated that capstans have higher reliability levels compared to cranes and winches.
This is attributed to the age of the fishing vessels and maintenance laxity towards deck machinery. Maintenance
of aging fishing vessel’ deck machinery should be given priority in order to extend their remaining useful time.
The third focus, was to develop a reliable predictive maintenance tool that does not use sensor technology,
but rather machine learning techniques. Concentration of the predictive maintenance tool, focused on dock
ballast pump on floating docks. Using unsupervised machine learning approach, a predictive maintenance
tool was developed to predict when maintenance is due. Pump operating parameter variations, control charts
and scatter plots were designed to form the basis of an early warning maintenance tool. Finally, a maintenance
optimization model was developed. The rationale of the maintenance model was to optimize the maintenance
costs without compromising on the reliability of the mechanical systems. Two performance control thresholds
(system reliability and maintenance interval) were set up as decision variables, which eventually dictated which
maintenance policy action to adopt. Based on a 20 year lifespan, the optimal maintenance policy implied a
possibility of an annual maintenance cost saving of 13.2% as compared to the current maintenance policies
used.
17