Flip-flop problem

[Akram]

Hello everyone,

My model includes the local district heating of a city. Since my model has a limited geographical scope, I have chosen an Integer Programming model instead of a Linear Programming model. Now, I have the flip-flop problem in my model. This means that, when I run the model for a scenario with a constant heat demand over the entire model time horizon, I get relatively different results compared to a scenario with a slow linear heat demand increase (e.g. +2% increase at the end of the model time horizon). The results differ in terms of investment years, new capacity investments and even investment in a different heat production technology.

I would be very grateful if someone worked with this type of models share his/her solutions with me.

Thank you.
//Akram

[Antti-L]

I have not worked with that type of models, but to minimize any flip-flop in MIP solutions, you should always set OPTCR=0, in order to get a truly optimal solution.  For example, if you leave OPTCR to its default value 0.1, in most energy system models that would leave a huge room for different MIP solutions that would satisfy the optimality tolerance, and thus you could really be seeing lots of flip-flopping between model runs (even when using exactly the same scenario data).

[Akram]

I have not worked with that type of models, but to minimize any flip-flop in MIP solutions, you should always set OPTCR=0, in order to get a truly optimal solution.  For example, if you leave OPTCR to its default value 0.1, in most energy system models that would leave a huge room for different MIP solutions that would satisfy the optimality tolerance, and thus you could really be seeing lots of flip-flopping between model runs (even when using exactly the same scenario data).

Thank you Antti_L for your reply. I have actually set OPTCR=0, but I still have the flip-flop problem. 

[Antti-L]

Good.  It was just a reminder, as users do tend to forget about it.

Then the flip-flop is optimal, and I am out of suggestions  .