A methodology for predicting the total average hourly maintenance cost of tracked hydraulic excavators operating in the UK opencast mining industry.
Research into the financial management of construction plant and equipment
maintenance is scant, despite the increased utilisation of mechanisation to augment
productivity in recent years. This thesis addresses the shortage of meaningful
research by developing a methodology for predicting the total average hourly
maintenance costs of tracked hydraulic excavators operating in opencast mining.
Initial pilot and field studies conducted revealed that maintenance management (in
the form of record keeping and attitude to used oil analysis) within the plant hire
and general construction industry was generally poor. Hence, the decision was
made to focus the research upon plant operated by opencast mining contractors.
Here, plant managers were found to utilise an optimum blend of predictive and
fixed-time-to maintenance and also maintain a depth of machine history file data.
Modelling total maintenance costs using multiple regression (MR) analysis at the
five percent level of significance identified four key predictor variables. These
were: machine weight; attitude to used oil analysis (regular use or not); type of
industry (opencast coal or slate); and type of machine (backacter or front shovel).
However, in order to determine the model's robustness an alternative modelling
technique, namely artificial neural networks (ANN) was applied using the same
variables identified as significant predictor variables in the regression analysis.
Performance analysis conducted on the predictive power of both MR and ANN
models revealed that overall the ANN model exhibited greater predictive
The thesis concludes with direction for future research and moreover, identifies the
need for a more fastidious approach to maintenance management.