Regarding the simple_modified model.
Despite assets being identical, activity is not shared equally between them. @tsmbland points out that in timeslices where there's excess capacity and not all assets have to be fully active, the solver will max out one asset before moving to the next, leaving some assets with lower/zero activity, and resulting in one of many equally good but arbitrary solutions that the solver could converge on. This is explained in a different context in #1174. This occurs both in a single agent model, and when commodity demands are split between multiple agents.
simple_modified.zip
The differing activity profiles between assets, even when overall capacity remains unchanged, gives rise to costing differences when we calculate for each asset in each milestone year.
Regarding the simple_modified model.
Despite assets being identical, activity is not shared equally between them. @tsmbland points out that in timeslices where there's excess capacity and not all assets have to be fully active, the solver will max out one asset before moving to the next, leaving some assets with lower/zero activity, and resulting in one of many equally good but arbitrary solutions that the solver could converge on. This is explained in a different context in #1174. This occurs both in a single agent model, and when commodity demands are split between multiple agents.
simple_modified.zip
The differing activity profiles between assets, even when overall capacity remains unchanged, gives rise to costing differences when we calculate for each asset in each milestone year.