weird experience with the COMMODITY_BALANCE_LT for "side-products"

[SongminYu]

Dear MESSAGE team, I have a question about the activation of COMMODITY_BALANCE_LT for commodities in the model. It is from a weird experience of the "side-products" in my model.

So, in an earlier version of my model, I gave some technologies some output commodities. I call them "side-products" because they are neither "demanded" nor used as inputs for other technologies. I included them just for accounting use. The model was running smoothly until I was trying to remove one of them. In this specific case, the output of the commodity was 0.3 in the first place. The model became infeasible when I removed it, or changed the value to 0.00000001. However, when I changed it to 0.1, the model was solved successfully again.

This was very weird to me because I thought the "side-products" are not constrained. They are not demanded or required. But, after checking the documentation, I realized that for the commodities, there is also the COMMODITY_BALANCE_LT constraint. I removed all such "side-products" and my model can be solved successfully again.

But, another question came: why could my model be solved previously? The "side-products" all had positive values in the output when the model was solved. Are there some conditions for the activation of the COMMODITY_BALANCE_LT for commodities?

[glatterf42]

Hi @SongminYu, thanks for this question. I've passed it on to the MESSAGE team this morning and their reply was that you should study exactly what was infeasible in your model to understand what happened when you removed one of your "side-products". If GAMS doesn't automatically report this, you can provide the iis option to it as either 1 (for a report when the model becomes infeasible) or 2 for looking for infeasibilities without trying to solve the model. I think you can e.g. pass that to Scenario.solve(iis=1).