Abstract
D. Ing.
This dissertation deals with the important issue of reliability management for 3kV DC Traction
Substations used by the national railway company in South Africa. Maintenance is one of the critical
and most costly phases in the lifecycle of any plant. It looks at the total life cycle of the equipment in a
typical substation, but the focus in the latter chapters is on the maintenance.
Through improved maintenance management, the reliability of the system can be improved. The
approach to maintenance is addressed as a predictive strategy, avoiding even more costly nonproductive
time due to downtime caused by failure or induced by maintenance. Condition monitoring
and assessment is described as one of the effective tools in the maintenance engineer’s armoury to
apply a predictive approach.
A direct link between predictable maintenance and reliability is explored. In the definition of reliability,
concepts such as time and expected performance can be linked to a predictable delivery of the
designed function. In other words, if down time is expected and can be prepared for, it is more
acceptable than the unexpected. In essence, the system is still reliable as it performs according to
expectation.
The concept of predictable maintenance can be applied wider than just the 3kV traction substation.
The process of identifying critical equipment, to measure the condition and to take decisions based on
the rate of change in the condition can be used in any maintenance environment, even outside
electrical. The crucial ingredient to this is to understand that condition monitoring is not based on fixed
values, but the rate at which these values change. This is called Fuzzy logic.
Can we predict the future? If yes, how accurate will the predictions be?