Question about Buildings.DHC.Loads.BaseClasses.FlowDistribution Control Valve and Pump Modeling

[hoxj180]

Dear MBL Developers,

I hope you are doing well, and thanks for all the work you do on MBL. I have a question about the Buildings.DHC.Loads.BaseClasses.FlowDistribution model in Buildings v12.1.0. I am using it for a district cooling system control analysis and cannot figure out if it implicitly models cooling coil control valves or not when using the ConstantDP pump control. I've listed the reasons for and against implicit modeling that I have found below:

Yes, it implicitly models the cooling coil control valves:

• The schematic in the info tab includes control valves downstream of each cooling coil.

• The description in the info tab says that "The modeling approach aims to minimize the number of algebraic equations by avoiding an explicit modeling of the terminal actuators..." I'm not sure if this means implicit or no modeling.

No, it does not implicitly model the cooling coil control valves:

• The pressure drop through the coil branch is calculated as dpPum = dpMin + dpVal + sum((mDis_flow_i/K_i)^2), where only mDisFlo is the required mass flow through the loop and is the only variable (non-constant) affecting dpPum. To me, this looks like pressure drop will increase as required mass flow increases, which is the opposite of what would happen in a cooling coil control valve.

I also wanted to ask why the pumps (pumFlo) turn on when they do in the ConstantDP control scheme. When I model them, the building pumps turn on only when the chilled water pumps are not providing enough flow. This happens even though the building pumps are receiving a non-zero mass flow signal (the required mass flow). Why do they not turn on whenever there is required mass flow?

Thank you so much. I appreciate any insight you have.

Sincerely,
Julia

[AntoineGautier]

The model Buildings.DHC.Loads.BaseClasses.FlowDistribution represents the building CHW/HW distribution system, optionally with a distribution pump if have_pum=true.
The modeling approach avoids modeling the full pressure-flow coupled network, while still calculating a representative pump operating point – needed for accurate ETS hydronics and incidentally for building pump electricity consumption.
To achieve this, you need to connect this component to instances of e.g. Buildings.DHC.Loads.BaseClasses.Validation.BaseClasses.FanCoil2PipeHeating or Buildings.DHC.Loads.BaseClasses.Validation.BaseClasses.FanCoil2PipeCooling that output the required CHW/HW mass flow rate to meet the zone load.
This setup is illustrated in Buildings.DHC.Loads.BaseClasses.BuildingTimeSeries.
No actual valve model is then instantiated, which is what FlowDistribution precisely avoids.
But the effect of the terminal unit control valves on the pump operating point is taken into account, at least for the configurations described in the "Pump head computation" section of the model documentation.
Taking for example a configuration with a single coil and a two-way valve, and a pump controlled to maintain a constant ∆p at the coil+valve boundaries, the valve ∆p increases as the valve closes while the pressure drop across the pipes and coil tends towards zero with the decreasing flow rate. So the pressure differential available at the circuit boundaries (= pump head) shifts entirely towards the valve. But the pump head itself decreases with the decreasing flow rate.

I struggle to understand the other question about the building pumps and CHW pumps. When using the model FlowDistribution with have_pum=true, the building CHW pumps are included in the model. And yes, according to the above logic, the pumps operate when there is any terminal unit flow request.

[hoxj180]

Hi Antoine,

Thanks so much for the quick response. Let me clarify both questions. I am considering a Buildings.DHC.Loads.BaseClasses.FlowDistribution model with a building pump with a ConstantDP control and no 3-way valve. It is connected to a Buildings.DHC.Loads.BaseClasses.Validation.BaseClasses.FanCoil2PipeCooling model in a district cooling model, and is supplied by chiller plant chilled water pumps. I also want to clarify that I am using Buildings v11.0.0, but it looks like the models are identical to Buildings v12.1.0.

Cooling Coil Modeling
I'm still confused how the cooling coil dp is taken into account in different pump head configurations. Does the parameter kDis[i] change, or is it constant regardless of required mass flow rate? If kDis changes, I can see how pressure drops from cooling coils are considered even if the coils aren't modeled. If kDis is constant, then the pressure drop basically only on frictional losses. Therefore, the dp from the cooling coils are effectively not considered because there is no dp reduction when the valve "opens" at high mass flows.

Building Pump Operation
I am seeing scenarios where the building pumps do not operate when there is a terminal unit flow request. In the below graph, the pressure drop across the building supplied by the district pump (top graph, green line) is greater than the pressure drop within the building (top graph, blue line). When this happens, the building pump turns off (middle graph). However, the required mass flow input to the building pump is still greater than zero. I can't find any equations in the model that make the pump turn off, and I am wondering what/where those equations are.

[AntoineGautier]

Cooling Coil Modeling

Cooling coils are modeled for the sole purpose of the heat transfer and the computation of the required CHW flow.
The hydronic model of the distribution system is entirely decoupled, as explained in the documentation of Buildings.DHC.Loads.BaseClasses.FlowDistribution.
From pump discharge to pump suction, there are several paths along which to sum up the pressure drops to get to the pump head. For a variable-flow distribution where the pump speed is modulated by a PI loop tracking a ∆p setpoint in a remote location, the path used in the model FlowDistribution solely relies on the pipe fixed flow resistances up to the sensor location.
So, yes, kDis is constant as explained in the documentation and as the parameter declaration demonstrates.
Of course you could select alternative paths that include the coils and valves. But we're not interested in those.

A simulation model may be more convincing than this discussion. Please check out the branch issue4534_fixValidation and inspect Buildings.DHC.Loads.BaseClasses.Validation.FlowDistributionPumpControl.
This model includes a comparison of FlowDistribution to an explicit modeling of the distribution network, the distribution pump and the terminal units. You'll see that there is strict equivalence, modulo the control dynamics.

Building Pump Operation

From your plots, I am assuming that your buildings are connected to the DHC system by means of a direct connection, with no bypass (bridge) and no control valve. This is unusual.
This setup effectively puts the district and building pumps in series.
If the district pump operating point is such that it creates a differential pressure at the building boundaries that is higher than necessary to reach the building CHW ∆p setpoint at a remote location, then the building pump speed naturally goes to zero. This is from Buildings.Fluid.Movers.BaseClasses.FlowMachineInterface which solves for the pump speed in case of a prescribed flow rate. And the pump speed solves to zero when the prescribed flow is already met by the district pump ∆p alone.

[hoxj180]

Hi Antoine,

Thanks so much. The building pump operation makes sense to me, and I will check out the simulations for the explicit hydronic modeling. I appreciate your time.

-Julia