Please write proper unit in PI-timeseries export (applies to version 2.3 and higher)

[SGeeversAtVortech]

In GitLab by @gijsber on Jan 22, 2020, 09:40

Currently pi_mixin writes unit_unknown in the timeseries header during export.

please write a proper unit if known based on the SI specification as used in the Modlica model.
Currently both flow and volume, typically expressed in m3/s and m3, have the same unit_unknown, which gives trouble on the Fews-side when I have to convert this to ML/day (multiply by 86.4) vs. ML (multiply by 0.01)

Please fix in version 2.3 and higher

[SGeeversAtVortech]

In GitLab by @gijsber on Jan 22, 2020, 14:49

This is rather problematic as pymoca does not parse all this Modelica SI-units stuff.
Three solutions could be envisioned:

1. extend pymoca to deal with it (= lot's of work)
2. extend rtcDataConfig with a unit field
3. for optimization:

• the pi.post() function retrieves a unit from the timeseries_import dict unless the unit is specified at during a set_timeseries operation.
  including your result timeseries as part of the timeseries_import ensures that the proper header info including unit is utilized

4. for simulation, inclusion of the result timeseries as part of timeseries_import creates a warning and computation failure as the code currently handles all those series as input and it cannot deal with the nan's specified. the solution is:

• in the post() function of the script, add the following before calling super().post():

    extra_units = {}
  
    for var in self.get_output_variables():
        if '.Q' in var or '_Q' in var:
            extra_units[var] = 'm3/s'
        elif '.V' in var or '_V' in var:
            extra_units[var] = 'm3'
  
    for var, unit in extra_units.items():
        for alias in self.alias_relation.aliases(var):
            self.timeseries_import.set_unit(alias, unit, 0)
  

[SGeeversAtVortech]

In GitLab by @rsinnige on Jan 27, 2020, 16:12

See issue #1112