By Patrick Andersen
Everything physical and manmade will eventually break or malfunction given enough time without attention. Often this happens unexpectedly and must be repaired after the fact, but usually such a fault can be avoided by preventative maintenance. In industries where malfunctions would be far more expensive to fix than avoid, they normally perform this maintenance regularly, having pre-planned routines in place to ensure everything gets its proper attention. However, as many of us feel when we are unfortunate enough get stuck behind road works during peak hour traffic, some maintenance schedules are better than others and finding a good schedule could certainly save a lot of time, money or trouble.
The subject of my project was to find the best maintenance schedule that meets all the requirements so that all the required maintenance jobs get done, but which minimises the effect of performing all the jobs on the system on which they’re being performed. The specific application I was concerned with was the Hunter Valley Coal Chain (HVCC) which is made up of a series of rail track and processing facilities through which coal is transported from mines in the Hunter Valley all the way to ship loading terminals at the port in Newcastle. When a part of the chain undergoes maintenance, it can no longer send any coal through which limits the amount of coal that gets sent to the terminals. It’s possible to have different maintenance schedules with the jobs scheduled at different times that result in different amounts of coal that can be sent through the HVCC over the course of a year, so my project was concerned with finding a way to determine the best schedule that will result in the most coal being sent through the network.
In particular, I looked at what would happen if we considered that the HVCC had storage facilities such as coal stockpiles which could store the coal while part of the chain was down and then send it through the chain once there was some free capacity. By incorporating this extra feature, the problem of finding a good schedule becomes a lot more difficult since a lot of the mathematical properties of the problem that we were using to find good solutions previously are no longer true for this storage problem. I was tasked with looking at the features of this problem and performing some computational experiments in order to try and help our understanding of the storage problem so that we may be able to come up with ways of finding good solutions in the future.
I really enjoyed working with this topic because it’s a very recent problem which meant that there were a lot of things for me to do and a lot of different avenues I could explore. The fact that the problem has a clear application made it easier for me to visualise as well as making my work far more exciting as anything I contribute has the potential to be used in the real world.
Patrick Andersen was one of the recipients of a 2013/14 AMSI Vacation Research Scholarship.