Hi. I have the following scenario.
I have 10 machines which we maintain for a customer. Each machine lasts 10 years, and each one requires 2 overhauls, roughly every 4 years, the timing of which depends on level of use, operation, condition etc. Each overhaul takes around 100 days. Assume an infinite capacity i.e. there's no limit to the amount of machines that can be overhauled at once.
When a machine is sent for overhaul, the customer needs a spare machine to cover for when the original machine requires an overhaul.
I'm trying to calculate the minimum level of dedicated spare machines needed.
The problem I have is the staggered nature of the overhauls. Lets say we get to year 4, I would not expect all 10 machines to be brought in at the same time, therefore needing 10 spare machines. The 10 repairs would be staggered over the year, to minimise the requirement of a dedicated spare machine.
How do I calculate what that dedicated level needs to be? So far i've looked at:
I have 10 machines which we maintain for a customer. Each machine lasts 10 years, and each one requires 2 overhauls, roughly every 4 years, the timing of which depends on level of use, operation, condition etc. Each overhaul takes around 100 days. Assume an infinite capacity i.e. there's no limit to the amount of machines that can be overhauled at once.
When a machine is sent for overhaul, the customer needs a spare machine to cover for when the original machine requires an overhaul.
I'm trying to calculate the minimum level of dedicated spare machines needed.
The problem I have is the staggered nature of the overhauls. Lets say we get to year 4, I would not expect all 10 machines to be brought in at the same time, therefore needing 10 spare machines. The 10 repairs would be staggered over the year, to minimise the requirement of a dedicated spare machine.
How do I calculate what that dedicated level needs to be? So far i've looked at:
- intermittent demand forecasting, using a probability distrubution like poisson. The problem with that is it assumes a constant repair rate and doesn't factor in the staggered queuing of repairs.
- Empirical approach using real data from the overhaul shop to see for a particular customer and their given level of operation, how many machines are in the overhaul shop at any point in time - so i can use the actual data, create a line of best fit which will be a wave pattern with 2 peaks, and simply look at the peak. Then i run monte carlo to see how much this could vary by.
- Queuing theory?