Fan power underestimation in terminal unit cooling model

[hoxj180]

Describe the bug
When using the BuildingTimeSeries model, I noticed that my total fan power is very low (basically nonexistent) despite dealing with large cooling loads. The fan power and thermo-fluid dynamics are calculated in the Buildings.DHC.Loads.BaseClasses.Validation.BaseClasses.FanCoil2PipeCooling module inside the Buildings.DHC.Loads.BaseClasses.BuildingTimeSeries model.

To recreate this error, I ran the Buildings.DHC.Loads.Cooling.Examples.BuildingTimeSeriesWithETS example model.

To Reproduce
Steps to reproduce the behavior:

1. Open model Buildings.DHC.Loads.Cooling.Examples.BuildingTimeSeriesWithETS
2. Run script 'Simulate and plot.' Plot buiWitETS.bui.terUniCoo.PFan and buiWitETS.bui.terUniCoo.mulPFan.k
3. See that the maximum total fan power does not exceed 10 W for the entire building despite a cooling load of 10kW. This is the maximum total fan power AFTER multiplying by the facMulCoo multiplier (plotted with mulPFan.k).

Expected behavior
I would expect the total fan power in an office building (instantaneous cooling loads around 10kW) to be a lot of energy. A maximum of 6W of instantaneous energy use for all fans in the building doesn't make sense to me.

I can't plot this in the example file, but when I plot the behavior in my own district cooling system model with the same BuildingTimeSeries component, I see that heat is being exchanged across the heat exchanger with very low airflow (in the FanCoil2PipeCooling model). Despite this, the heat transfer across the heat exchanger approximates the required heating load (after dividing by the facMulCoo multiplier), so the PID control sends an output signal near 0 to the fan, which keeps mass flows very low.

I think what could be causing this is the facMulCoo calculation. It currently calculates very high values which split the total cooling load into very small loads that can be handled with very small mass flows. In my model (this wasn't plottable in the example file, sorry), the Qcoo the fan is controlled to meet is only 70W, which seems significantly smaller than a typical zone internal loads. Changing the equation for the facMulCoo calculation might give more reasonable zone Q values and fan powers.

In conclusion, the fan power seems very low, and I think the facMulCoo calculation might be the cause. Also, if it isn't a modeling issue, would it be possible to update the documentation to more fully explain the assumptions around the fan control in the BuildingTimeSeries and FanCoil2PipeCooling models?

Thank you so much for your help, and please let me know what you find. Let me know if you need any other information.

Version

• Modelica Buildings Library: 11.0.0
• Modelica simulation environment: Dymola
• Operating system: Windows

[AntoineGautier]

The issue is that, in the example model Buildings.DHC.Loads.Cooling.Examples.BuildingTimeSeriesWithETS, the time series have a peak cooling load of buiWitETS.QCoo_flow_nominal = 245E3 (W).
As described in the documentation of Buildings.DHC.Loads.BaseClasses.BuildingTimeSeries, this load is split between identical terminal units with heat transfer performance approximated based on design specifications of a reference FCU. The reference FCU has a cooling capacity of buiWitETS.bui.terUniCoo.QCoo_flow_nominal = 5E3 (W) (with a design fan power of about 200 W). The number of FCUs calculated by the model is then buiWitETS.bui.facMulCoo = 45.
Now, during the simulated period in Buildings.DHC.Loads.Cooling.Examples.BuildingTimeSeriesWithETS, the cooling load only peaks at max(buiWitETS.bui.QCoo_flow) = 20E3 (W), requiring only 7 % of the design fan air flow. The FCU model assumes a continuous variation of the fan speed to meet the load and a fixed flow characteristics on the air side. The power thus varies (almost) cubic with the air flow, yielding the low fan power that you noticed.
In reality, the fan would cycle on and off at minimum speed (~10-15 % for EC motors). The simulation only exhibits the average power assuming continuous operation.

That being said, the model Buildings.DHC.Loads.Cooling.BuildingTimeSeriesWithETS is not intended to compute a representative fan power considering that

• there is no parameter to specify the air-side systems,
• have_fan = false and
• the fan power is not exposed as an output connector.

The FCU model is just a means of converting the cooling and heating loads into a CHW/HW flow to connect to the DHC hydronic system.
When using Buildings.DHC.Loads.Cooling.BuildingTimeSeriesWithETS, the assumption is that, since you can provide heating and cooling loads, you may have simulation results from a building model. Best is to use the fan power calculated by this model.
If that's not the case, you can adjust the FCU parameters as described in Buildings.DHC.Loads.BaseClasses.BuildingTimeSeries to better approximate your HVAC system. But you'll still rely on the simplified FCU model and you won't account for, e.g., the ventilation fans.

A more complex alternative would be to develop your own terminal unit model by extending Buildings.DHC.Loads.BaseClasses.PartialTerminalUnit and then redeclare the component terUniCoo within Buildings.DHC.Loads.BaseClasses.BuildingTimeSeries.

[hoxj180]

Ahh, I see. Thank you so much for the explanation. This makes a lot of sense.