Impact of adding bounds to variables

When I was trying to solve the model I oftentimes get this kind of warning:

Matrix range     [6e-07, 2e+03]
Objective range  [2e-01, 5e+05]
Bounds range     [5e+10, 5e+10]
RHS range        [1e-03, 1e+100]
Warning: Model contains large rhs
Warning: Model contains large bounds

which comes from the missing upper bounds on power or energy when we adds the generators, storage_units, links, and stores to the network. I suspect that it may lead to numerical issues, so I tested the impact of adding shared bounds to those components, i.e. p_nom_max = 1e6, etc.

It turns out that adding those bounds makes the solution process even more difficult: 1743.04 working units w/o bounds vs 4347.72 working units w/ bounds. I think the current way of defining components is just fine. That been said, the solver behavior might differ with the problem that you are trying to solve, and it might help to add tighter bound specified for each component in a smarter way.

Start by reproducing the large RHS and bounds warnings when adding generators, storage_units, links, and stores, then compare solver working units with and without shared bounds such as p_nom_max = 1e6. No files or tests are named; done would require a clearly scoped change or documented conclusion about component-specific bounds and solver behavior.