Matching a fan curve in Openmodelica

[Dynteq]

Trying to optimize some ducting configuration I want to model a system of ducts with a fan to simulate the flow and pressure distribution.
In order to choose the right fan I want to add a fan curve from the fan supplier. But how to get the fan to behave like the curve says it should?

Using a SpeedControlled_y pump from the Buildings library I want to match this performance curve:

The test setup was adapted from Buildings.Fluid.Movers.Examples.ClosedLoop_y

I tried testing this by adding a pressure drop (called inlet in the immage) and changing it from very large to very low. (inlet_nominal in the code below) But what ever I do I never get points on the curve. Especially for low pressure drops I get values for flow that exceed 110m3/h (or 0.0235kg/s as calculated). For example when I set inlet_nominal = 20 Pa. The simulation gives a flow of 0.0258kg/s (*4680 = 120m3/h) and a pressure over the fan (fan.dp) of 29Pa. I would expect either a smaller pressure with that flow or a smaller flow with that pressure.

The values for the pressure curve are in the table below. kg/s is calculated with a density of 1.3 kg/m3

m3/h	kg/s	Pa
0	0	175
10	0.00214	170*
70	0.0150	100*
105	0.0224	20*
110	0.02354	0

The values marked * are in the per definition in the code example below where the fan curve is defined. (per(pressure(V_flow = {0.0002, 0.015, 0.0224},dp = {170, 100, 20})))
The other values I tried earlier. The results are similar.

model FanTest01
  package Medium = Buildings.Media.Air;
  parameter Modelica.Units.SI.MassFlowRate m_flow_nominal = 0.0224 "Nominal mass flow rate";
  parameter Modelica.Units.SI.MassFlowRate m_flow_nominal_fan = 0.0128 "110m3/h Nominal mass flow rate for fan";
  parameter Modelica.Units.SI.PressureDifference dp_nominal = 5 "Nominal pressure difference";
  parameter Modelica.Units.SI.PressureDifference dp_nominal_fan = 117 "Nominal pressure difference for fan";
  parameter Modelica.Units.SI.PressureDifference inlet_nominal = 20 "Inlet pressure difference";
  parameter Modelica.Units.SI.Length steek = 0.44 "Horizontale steek tussen modulles";
  parameter Modelica.Units.SI.Length diameter = 0.08 "Buis diameter";
  // FAN
  Buildings.Fluid.Movers.SpeedControlled_y fan(redeclare package Medium = Medium,
  per(pressure(V_flow = {0.0002, 0.015, 0.0224},
    dp = {170, 100, 20})), 
    energyDynamics = Modelica.Fluid.Types.Dynamics.FixedInitial) "Fan" annotation(
    Placement(visible = true, transformation(origin = {70, 30}, extent = {{40, 40}, {60, 60}}, rotation = 0)));
Buildings.Fluid.Sources.Boundary_pT sou(redeclare package Medium = Medium, T = 293.15, nPorts = 2, p = 101300, use_p_in = false) annotation(

Placement(visible = true, transformation(origin = {82, 0}, extent = {{-82, 10}, {-62, 30}}, rotation = 0)));
Buildings.Fluid.FixedResistances.HydraulicDiameter duct01(redeclare package Medium = Medium,

dh = diameter, fac = 1, length = 0.44,

m_flow_nominal = m_flow_nominal) annotation(

Placement(visible = true, transformation(origin = {150, 80}, extent = {{-10, -10}, {10, 10}}, rotation = 0)));
Buildings.Fluid.FixedResistances.HydraulicDiameter duct02(redeclare package Medium = Medium,

dh = diameter, fac = 1, length = 0.44,

m_flow_nominal = m_flow_nominal) annotation(

Placement(visible = true, transformation(origin = {90, 80}, extent = {{-10, -10}, {10, 10}}, rotation = 0)));

Buildings.Fluid.FixedResistances.PressureDrop inlet(redeclare package Medium = Medium,

dp_nominal = inlet_nominal,

m_flow_nominal = m_flow_nominal) annotation(

Placement(visible = true, transformation(origin = {50, 50}, extent = {{-10, -10}, {10, 10}}, rotation = 90)));
Modelica.Blocks.Sources.Ramp ramp(duration = 100, startTime = 10) annotation(

Placement(visible = true, transformation(origin = {90, 110}, extent = {{-10, -10}, {10, 10}}, rotation = 0)));
equation

connect(fan.port_b, duct01.port_a) annotation(

Line(points = {{130, 80}, {140, 80}}, color = {0, 127, 255}));

connect(duct01.port_b, sou.ports[1]) annotation(

Line(points = {{160, 80}, {170, 80}, {170, 20}, {20, 20}}, color = {0, 127, 255}));

connect(ramp.y, fan.y) annotation(

Line(points = {{101, 110}, {120, 110}, {120, 92}}, color = {0, 0, 127}));

connect(fan.port_a, duct02.port_b) annotation(

Line(points = {{110, 80}, {100, 80}}, color = {0, 127, 255}));

connect(inlet.port_a, sou.ports[2]) annotation(

Line(points = {{50, 40}, {50, 20}, {20, 20}}, color = {0, 127, 255}));

connect(inlet.port_b, duct02.port_a) annotation(

Line(points = {{50, 60}, {50, 80}, {80, 80}}, color = {0, 127, 255}));

annotation(

Diagram(coordinateSystem(preserveAspectRatio = true, extent = {{-200, 160}, {200, -240}})),

Documentation(info = "<html>

<p>

Adapted from 'ClosedLoop_y' example.

</p>

</html>"),

__Dymola_Commands(file = "modelica://Buildings/Resources/Scripts/Dymola/Fluid/Movers/Examples/ClosedLoop_y.mos" "Simulate and plot"),

experiment(StopTime = 200, Tolerance = 1e-06, StartTime = 0, Interval = 1),

uses(Buildings(version = "9.1.0"), Modelica(version = "4.0.0")),

Icon(coordinateSystem(extent = {{-200, -260}, {200, 160}}, preserveAspectRatio = true)),

version = "");

end FanTest01;

What should I change to get simulation results closer to the desired fan curve from the supplier? It could well be I'm doing something silly wrong.

[hcasperfu]

Greetings!
The fan curve can only be observed at constant speed. In your model, the speed signal to the fan is a ramp signal with varied speed.
In v9.1.0 there is one lower-level validation model Buildings.Fluid.Movers.BaseClasses.Validation.FlowMachineInterface from which you can look at the fan curve by plotting eff.dp vs eff.V_flow. However, this model validates the low-level class that handles the pressure curve. At the level of a full mover, things get more complicated.
You may refer to Buildings.Fluid.Movers.Validation.NegativePressureOrFlow to see how to force a pressure at constant speed to see the mover react to it with a certain flow rate, hence validating the pressure curve. But this model will be removed in the next release.
I will open an issue to add a validation model that can directly display the fan/pump curve, since this seems to be asked quite a few times recently.

[Dynteq]

Thank you for your answer Hcasperfu,
I'm not sure I understand how this works completely.
In my understanding a fan curve shows how a fan will perform at a certain power (usually 100%). To find the flow and pressure you would plot the resistance curve (also flow vs pressure) of the connected system in the same graph. The two lines intersect at the equilibrium of pressure and flow.

The Ramp signal in my model is not important right now. For the first check I only look at the constant pressure after t=100 where the input for the pump is 1. I run the simulation with different parameters. If I set the resistance very high then I get a high pressure and a small flow more or less as expected. But if I set the resistance low (small dp@inlet and short ducts) the the flow and pressure are both above the curve. That is not the expected behavior.
It could be I'm doing something wrong in the simulation setup, or my understanding of how it works physically is incorrect, or it is a numeric quirk of how the thing is modeled.
Any hints are welcome.

[hcasperfu]

Greetings again!
The fan curve shows the pressure-flow relationship at a constant speed, instead of power. Other than that, I think your description of the physics is correct.
However, I think combining the fan curve and the system curve to validate the fan curve seems to bring in unnecessary complications for this purpose. A simple validation model that displays the pressure curve of the fan is currently being reviewed and should be available on the master branch soon (see #3376) and will go into the next release. The model will be Buildings.Fluid.Movers.Validation.PressureCurve.
Concerning the fan's operating point going above the curve, this is partly because the model does not restrict the speed of the pump to an upper limit, i.e. the Nrpm input is allowed to go above speed_rpm_nominal and therefore "above the curve".
Casper

[hcasperfu]

Greetings!
As a followup, there will be a new validation model Buildings.Fluid.Movers.Validation.PressureCurve that outputs the pressure curve from the mover model. Simply plot mov.dpMachine vs mov.VMachine_flow.
This is already available on the master branch (368f3a3 or later).

Cheers,
Casper

[Dynteq]

Hi Casper,
Nice to hear the curve can be plotted in the near future, that will make validation a lot more intuitive.
A quick follow up question about the Nrpm or speed_rpm_nominal. I don't see these anywhere in the SpeedControlled_y pump. I assumed that the y input controlled either power or speed, and setting it to 1 would be nominal.
What does y control if not rpm?
Guess I missed the rpm part in SpeedControlled_Nrpm. Would SpeedControlled_Nrpm be better suited? And does this take the rpm as input? (and thereby prevent operation above the curve if set to nominal)
Cheers,
Martijn

[hcasperfu]

The y input is the normalised speed, i.e. y = Nrpm / Nrpm_nominal. Because internally the mover model converts the Nrpm signal (with an RPM unit) to a y signal (with a unit of "1") anyway, for the model with a y-input, any RPM information is redundant. The _y model always has y_nominal = 1.
Please note that the _Nrpm model will be moved to the Obsolete package in the upcoming release. This is to simplify the implementation and for its lack of use (see ibpsa/modelica-ibpsa#1704). In the linked post we also briefly discussed whether we want to put limits to the input signal, so that y is limited between 0 and 1, as you suggested. This was rejected because (a) this, again, would complicate the implementation and (b) such a limiter is better positioned in the control sequence outside of the component model.
If it is more convenient for you to use a RPM input, it can be scaled to a y signal by a gain block, i.e. Modelica.Blocks.Math.Gain or Buildings.Controls.OBC.CDL.Continuous.MultiplyByParameter. To limit the control signal within 0 to 1, use Buildings.Controls.OBC.CDL.Continuous.Limiter.

[Dynteq]

Hi Casper,
Thanks for your time explaining. Sorry but I'm still a bit lost. In my model I have a ramp on the input of y. that rises to 1 and then stays there. So I don't understand how the fan can perform above the curve.

Seeing the discussion in the 1704 thread you linked, I don't think switching to Nrpm would help me.