Degradation Modelling of and Optimising the Timing of Replacements for Batch Vacuum Pans

Replacing a pan involves the expenditure of significant capital for a sugar mill. Replacing a pan too late may result in excessive downtime, maintenance costs, and risk of catastrophic failure. On the other hand, replacing a pan too early will lead to wasting residual life and an unnecessary allocation of capital funds that may have been spent better elsewhere in the mill. This paper reports on the development of a replacement policy for batch vacuum pan components based on a stochastic model of degradation. Degradation data, principally wall-thickness measurements, were collected from the vacuum pans of an Australian sugar factory and used to develop component degradation models. Unlike the conventional approach of using a line of best fit to identify the end of life of the pan, the methods adopted account for the uncertainties due to seasonal operating conditions and inherent uncertainty in the degradation model parameters. The quantification of the uncertainty in identifying the end of life of a vacuum pan has shown that there is significant risk of a pan failing earlier than the straight-line prediction. Employing this quantification of the risk, a component replacement plan was developed by optimising the replacement of each component individually and subsequently optimising the replacement plan for the entire pan. This strategy is demonstrated using a case study with and without parametric uncertainty to evaluate its impact on maintenance optimisation. Including parametric uncertainty leads to the determination of greater risk earlier, proposing the replacement of components earlier than when parameters are considered as ‘known’. It is, therefore, important to consider parametric uncertainty in the planning of pan component replacements to better manage risk.