User's Guide


This package contains models for fans and pumps (movers). The same models can be used for fans or pumps.

Model description

The models consider the pressure rise, flow rate, speed, power consumption, and heat dissipation based on the user's specification. They can take pressure rise (head), mass flow rate, or speed (absolute or relative) as control signal, and compute resulting quantities based on user-provided performance curves.

While the models in the package Buildings.Fluid.Movers allow full customization, preconfigured models that use the same underlying physical equations are available in the package Buildings.Fluid.Movers.Preconfigured. The models in Buildings.Fluid.Movers can also be parameterized with the data records from Buildings.Fluid.Movers.Data.

A detailed description of the fan and pump models can be found in Wetter (2013). The models are implemented as described in this paper, except that equation (20) is no longer used. The reason is that the transition (24) caused the derivative

d Δp(r(t), V(t)) ⁄ d r(t)

to have an inflection point in the regularization region r(t) ∈ (δ/2, δ). This caused some models to not converge. To correct this, for r(t) < δ, the term V(t) ⁄ r(t) in (16) has been modified so that (16) can be used for any value of r(t).

Below, the models are briefly described.

Performance data

The models use performance curves that compute pressure rise, electrical power draw and efficiency as a function of the volume flow rate and the speed. The following performance curves are implemented:

Independent variable Dependent variable Record for performance data Function
Volume flow rate Pressure flowParameters pressure
Volume flow rate Efficiency
(hydraulic or motor)
efficiencyParameters efficiency
Motor part load ratio Motor efficiency* efficiencyParameters_yMot efficiency_yMot
Volume flow rate Power** powerParameters power

Notes (applicable to Buildings.Fluid.Movers.FlowControlled_dp and Buildings.Fluid.Movers.FlowControlled_m_flow):

These performance curves are implemented in Buildings.Fluid.Movers.BaseClasses.Characteristics, and are used in the performance records in the package Buildings.Fluid.Movers.Data. The package Buildings.Fluid.Movers.Data contains different data records.

Models that use performance curves for pressure rise

The model Buildings.Fluid.Movers.SpeedControlled_y takes as an input a control signal between 0 and 1. From this input and the current flow rate, they compute the pressure rise. This pressure rise is computed using a user-provided list of operating points that defines the fan or pump curve at full speed. For other speeds, similarity laws are used to scale the performance curves, as described in Buildings.Fluid.Movers.BaseClasses.Characteristics.pressure.

For example, suppose a pump needs to be modeled whose pressure versus flow relation crosses, at full speed, the points shown in the table below.

Volume flow rate [m3⁄s] Head [Pa]
0.0003 45000
0.0006 35000
0.0008 15000

Then, a declaration would be

  Buildings.Fluid.Movers.SpeedControlled_y pum(
    redeclare package Medium = Medium,
                 dp    ={45,35,15}*1000))
    "Circulation pump";

This will model the following pump curve for the pump input signal y=1.


See Buildings.Fluid.Movers.Validation.PressureCurve for a small example that validates the pressure curve specification.

Models that directly control the head or the mass flow rate

The models Buildings.Fluid.Movers.FlowControlled_dp and Buildings.Fluid.Movers.FlowControlled_m_flow take as an input the pressure difference or the mass flow rate. This pressure difference or mass flow rate will be provided by the fan or pump, i.e., the fan or pump has idealized perfect control and infinite capacity. Using these models that take as an input the head or the mass flow rate often leads to smaller system of equations compared to using the models that take as an input the speed.

These models can be configured for three different control inputs. For Buildings.Fluid.Movers.FlowControlled_dp, the head is as follows:

Similarly, for Buildings.Fluid.Movers.FlowControlled_m_flow, the mass flow rate is as follows:

These two models do not need to use a performance curve for the flow characteristics. The reason is that

However, the computation of the electrical power consumption requires the mover speed to be known and the computation of the mover speed requires the performance curves for the flow and efficiency/power characteristics. Therefore these performance curves do need to be provided if the user desires a correct electrical power computation. If the curves are not provided, a simplified computation is used, where the efficiency curve is used and assumed to be correct for all speeds. This loss of accuracy has the advantage that it allows to use the mover models without requiring flow and efficiency/power characteristics.

The model Buildings.Fluid.Movers.FlowControlled_dp has an option to control the mover such that the pressure difference set point is obtained across two remote points in the system. To use this functionality parameter prescribeSystemPressure has to be enabled and a differential pressure measurement must be connected to the pump input dpMea. This functionality is demonstrated in Buildings.Fluid.Movers.Validation.FlowControlled_dpSystem.

The models Buildings.Fluid.Movers.FlowControlled_dp and Buildings.Fluid.Movers.FlowControlled_m_flow both have a parameter m_flow_nominal. For Buildings.Fluid.Movers.FlowControlled_m_flow, this parameter is used for convenience to set a default value for the parameters constantMassFlowRate and massFlowRates. For both models, the value is also used for the following:

However, otherwise m_flow_nominal does not affect the mass flow rate of the mover as the mass flow rate is determined by the input signal or the above explained parameters.

Electrical power consumption

All models compute the motor power draw Pele, the hydraulic power input hyd, the flow work flo and the heat dissipated into the medium . Based on the first law, the flow work is

flo = | V̇ Δp |,

where is the volume flow rate and Δp is the pressure rise. In order to prevent the model from producing negative mover power when either the flow rate or pressure rise is forced to be negative, the flow work flo is constrained to be non-negative. The regularisation starts around 0.01% of the characteristic maximum power max = V̇max Δpmax. See discussions and an example of this situation in IBPSA, #1621.

The heat dissipated into the medium is as follows: If the motor is cooled by the fluid, as indicated by per.motorCooledByFluid=true, then the heat dissipated into the medium is

Q̇ = Pele - Ẇflo.

If per.motorCooledByFluid=false, then the motor is outside the fluid stream, and only the shaft, or hydraulic, work hyd enters the thermodynamic control volume. Hence,

Q̇ = Ẇhyd - Ẇflo.

The efficiencies are defined as

η = Ẇflo ⁄ Pele = ηhyd   ηmot
ηhyd = Ẇflo ⁄ Ẇhyd
ηmot = Ẇhyd ⁄ Pele

where η is the total efficiency, ηhyd is the hydraulic efficiency, and ηmot is the motor efficiency. From the definition one has

η = ηhyd   ηmot.

Hydraulic efficiency

The following options are used to specify how ηhyd is computed.

These options are tested in Buildings.Fluid.Movers.BaseClasses.Validation.HydraulicEfficiencyMethods.

The model uses EulerNumber as the default option unless a pressure curve is not provided. In this case, the model overrides it and uses NotProvided instead.

The user can use the same options to specify the total efficiency η instead by setting per.powerOrEfficiencyIsHydraulic=false. This changes the default constant value to η=0.49 and also imposes an additional constraint of ηhyd ≤ 1 to prevent the division ηhyd = η ⁄ ηmot from producing efficiency values larger than one. This configuration is tested in Buildings.Fluid.Movers.BaseClasses.Validation.TotalEfficiencyMethods.

Although the Euler number method is defined for ηhyd, this implementation applies it also to η and Pele as an approximation. The basis is that ηmot is mostly constant for motors larger than about 3.5 kW or 5 HP except when the motor part load drops below around 40%, (see the documentation of Buildings.Fluid.Movers.BaseClasses.Characteristics.motorEfficiencyCurve) which shows that η and ηhyd are roughly linear to each other for motors of this size.

Motor efficiency

The following options are used to specify how ηmot is computed.

These options are tested in Buildings.Fluid.Movers.BaseClasses.Validation.MotorEfficiencyMethods.

By default, the model uses the GenericCurve to obtain more accurate results with variable ηmot. There are two exceptions:

Start-up and shut-down transients

All models have a parameter use_inputFilter. This parameter affects the fan output as follows:

  1. If use_inputFilter=false, then the input signal y (or m_flow_in, or dp_in) is equal to the fan speed (or the mass flow rate or pressure rise). Thus, a step change in the input signal causes a step change in the fan speed (or mass flow rate or pressure rise).
  2. If use_inputFilter=true, which is the default, then the fan speed (or the mass flow rate or the pressure rise) is equal to the output of a filter. This filter is implemented as a 2nd order differential equation and can be thought of as approximating the inertia of the rotor and the fluid. Thus, a step change in the fan input signal will cause a gradual change in the fan speed. The filter has a parameter riseTime, which by default is set to 30 seconds. The rise time is the time required to reach 99.6% of the full speed, or, if the fan is switched off, to reach a fan speed of 0.4%.

The figure below shows for a fan with use_inputFilter=true and riseTime=30 seconds the speed input signal and the actual speed.


Although many simulations do not require such a detailed model that approximates the transients of fans or pumps, it turns out that using this filter can reduce computing time and can lead to fewer convergence problems in large system models. With a filter, any sudden change in control signal, such as when a fan switches on, is damped before it affects the air flow rate. This continuous change in flow rate turns out to be easier, and in some cases faster, to simulate compared to a step change. For most simulations, we therefore recommend to use the default settings of use_inputFilter=true and riseTime=30 seconds. An exception are situations in which the fan or pump is operated at a fixed speed during the whole simulation. In this case, set use_inputFilter=false.

Note that if the fan is part of a closed loop control, then the filter affects the transient response of the control. When changing the value of use_inputFilter, the control gains may need to be retuned. We now present values control parameters that seem to work in most cases. Suppose there is a closed loop control with a PI-controller Buildings.Controls.Continuous.LimPID and a fan or pump, configured with use_inputFilter=true and riseTime=30 seconds. Assume that the transient response of the other dynamic elements in the control loop is fast compared to the rise time of the filter. Then, a proportional gain of k=0.5 and an integrator time constant of Ti=15 seconds often yields satisfactory closed loop control performance. These values may need to be changed for different applications as they are also a function of the loop gain. If the control loop shows oscillatory behavior, then reduce k and/or increase Ti. If the control loop reacts too slow, do the opposite.

Fluid volume of the component

All models can be configured to have a fluid volume at the low-pressure side. Adding such a volume sometimes helps the solver to find a solution during initialization and time integration of large models.

Enthalpy change of the component

If per.motorCooledByFluid=true, then the enthalpy change between the inlet and outlet fluid port is equal to the electrical power Pele that is consumed by the component. Otherwise, it is equal to the hydraulic work Whyd. The parameter addPowerToMedium, which is by default set to true, can be used to simplify the equations. If addPowerToMedium = false, then no enthalpy change occurs between inlet and outlet. This can lead to simpler equations, but the temperature rise across the component will be zero. In particular for fans, this simplification may not be permissible.

Differences to models in Modelica.Fluid.Machines

The models in this package differ from Modelica.Fluid.Machines primarily in the following points:


Michael Wetter. Fan and pump model that has a unique solution for any pressure boundary condition and control signal. Proc. of the 13th Conference of the International Building Performance Simulation Association, p. 3505-3512. Chambery, France. August 2013.

Hongxiang Fu, David Blum, Michael Wetter. Fan and Pump Efficiency in Modelica based on the Euler Number. Proc. of the American Modelica Conference 2022, p. 19-25. Dallas, TX, USA. October 2022.

EnergyPlus 9.6.0 Engineering Reference

Extends from Modelica.Icons.Information (Icon for general information packages).