Wrong calculation of nominal mass flow rate in PartialSolarCollector

[mwetter]

This closes issue #1956

[mwetter]

@jelgerjansen : There is still something wrong with the parameter calculation. Can you please run

simulateModel(
  IBPSA.Fluid.SolarCollectors.Examples.FlatPlate(solCol(nColType=IBPSA.Fluid.SolarCollectors.Types.NumberSelection.Area, totalArea=10, nPanelsPar=4))",
  stopTime=86400,
  tolerance=1e-06,
  resultFile="FlatPlate");

which triggers the assertion for which I added more output. This was already wrong before your change, but should be fixed too.

[jelgerjansen]

@mwetter the current implementation requires the user to input both the total number of collector panels (either via nPanels or totalArea) and the number of collector panels in series and parallel (nPanelsSer and nPanelsPar). This is an "overspecified problem".

As a solution, I kept nPanelsSer as a user-defined parameter in case of an array and calculated the other parameter nPanelsPar_internal = nPanels_internal/nPanelsSer_internal. This also avoids the need for an assert to check whether nPanels=nPanelsSer*nPanelsPar.

However, this implementation is not yet ideal, as totalArea is not necessarily an exact multiple of the area of a single collector. The user might have an array of collectors which is 5 times the area of a single collector, and model this as 2 parallel branches of 2.5 collectors in series. This is not possible in the current implementation as nPanelsSer is defined as an integer. Therefore, I suggest defining nPanelsSer as a Real parameter.

[mwetter]

@jelgerjansen : I agree that the problem is over-specified, that we should remove one parameter.

I however suggest to remove nPanelsSer, and keep instead nPanelsPar, and make nPanelsPar an Integer. This way, we have always an integer value for the number of parallel circuits, which seems more logical from a construction point of view. We can then keep nPanelsSer_internal as a Real, and if it ends up to not be equal to an integer within say 1%, issue a assertion with a level warning.

With this approach, if a user specifies to use totalArea, we have an integer number for nPanelsPar, and perhaps a fraction for nPanelSer, in which case we warn the user and proceed with the simulation.

Does this sound good?

[jelgerjansen]

@mwetter I agree and implemented the proposed changes. Feel free to update the models if something is still missing or if you want to rephrase the warning message.

[jelgerjansen]

@mwetter I assigned @FWuellhorst to do a review (as he also faced this issue recently). Can we merge this PR after his review? Or do you still want to do a final review yourself?

[FWuellhorst]

Looks good!

[mwetter]

The changes now look good. I updated the conversion script, tested it, and also removed nPanelsSer from the user guide.

[mwetter]

This is ready to merge if @FWuellhorst and @jelgerjansen agree.