Because of a question raised by Ed Julian and some testing I was doing for OG&E, I checked into the implications of having a maximum inventory level that is close to the optimal inventory target. What I found is something that I knew many years ago but had forgotten.
The bottom line is you need to be careful and not set your maximum inventory level at a value that is close to the optimal target. If you find that the optimal target is close to the maximum inventory level you should probably raise the maximum inventory level. The reason that this is so is the following.
If you go to the input screen where maximum inventory level is set and press F1 you get context sensitive help. This is what the help screen says:
“There is a limit to how much fuel a utility can accommodate at an inventory site. You specify this limitation under Maximum Inventory Level; this is the largest amount of fuel that would ever be held at your inventory site.”
“UFIM treats this maximum level very seriously. The model assumes that if inventory levels ever exceed the maximum, the extra fuel above the maximum is lost. In other words, the extra fuel is purchased, but it can never be used. It is therefore very important to set a realistic Maximum Inventory Level.”
“Note that the Maximum Inventory Level is expressed as an integer multiple of STEP.”
The facts are the following. There is a penalty associated with ordering fuel that can result in exceeding “Maximum Inventory” (exceeding maximum inventory level can occur when burn is lower than expected or deliveries are higher than expected). This can result in the “optimal” inventory being lower than it otherwise would be if “Maximum Inventory” were set to a higher value.
Remember the fundamental decision rule in UFIM is:
Order at Start of Month = Target Inventory for End of Month + E(Burn) + E(Supply Shortfall) - Total Inventory on Hand