Air System Fans[LINK]
Three input objects (Fan:ConstantVolume, Fan:VariableVolume, and Fan:OnOff) provide models for fans – the prime movers in most of the air loop and zonal air conditioning systems in EnergyPlus. For these input objects, EnergyPlus uses a simple polynomialbased curvefit model to describe the relation between the volumetric flow rate and the fan electric power, with no explicit modeling of fan pressure rise. Consequently, ductstaticpressure reset strategies cannot be modeled using this approach. For CAV and VAV central airhandling systems, a fourth input object (Fan:ComponentModel) provides a simple physicsbased model for flowdependent fan pressure rise and detailed models for fan, belt, motor, and variablefrequencydrive efficiencies and energy use. This latter input object provides the capability to analyze the effects of ductstaticpressure reset strategies, as well as changes in fan system element performance. All of these fan models calculate the fan energy usage, which is often a large fraction of HVAC energy use and a significant portion of the building energy consumption. They also calculate the temperature rise in the air stream caused by the static pressure increase as the air goes through the fan. If the motor and belt are in the air stream, waste heat from the motor and belt also acts to raise the air stream temperature.
Each of the four fan models is a forward type: the model inputs describe the fan characteristics and the conditions of the air at the fan inlet; the outputs are the fan electrical power consumption and the conditions of the air at the fan outlet. The model algorithms and data are contained in the Fans module in EnergyPlus.
For the Fan:ConstantVolume, Fan:VariableVolume, and Fan:OnOff models, the user describes the fan by entering values for the design pressure rise across the fan, the design volumetric flow rate, the fan total efficiency, and the fan motor efficiency. The user also needs to specify the fraction of the fan waste heat that will enter the air stream (usually 0 or 1). For the Fan:VariableVolume model, the user must also enter the coefficients of a 4th order polynomial that relates the fan mass flow rate to the fan power consumption.The independent variable is the volumetric flow fraction; the dependent variable is the fan power part load ratio. For multispeed fans, the user must enter a fan power ratio as a function of speed ratio performance curve name. Multispeed fans can only be simulated in a parent object which allows multiple fan speeds (e.g., AirLoopHVAC:Unitary:Furnace:HeatCool, ZoneHVAC:PackagedTerminalAirConditioner, etc.). An optional efficiency curve may also be used when simulating multispeed fans to adjust the fan total efficiency as the fan speed changes.
For the Fan:ComponentModel object, the user describes the fan in more detail, and also describes the duct system characteristics as seen by the fan. In particular, the user specifies a pressure rise curve with four coefficients that relates the fan total pressure rise to the volumetric flow through the fan, the ductstaticpressure setpoint, and the static pressure of the spaces surrounding the ducts. If ductstaticpressure reset is used, the user enters a linear curve with two coefficients that relates the pressure setpoint to the volumetric flow through the fan. For the fan itself, the user specifies the fan geometry (wheel diameter and outlet area), maximum efficiency, the Euler number corresponding to the maximum efficiency, the maximum dimensionless flow, the names of four curves with several coefficients that describe the variation of fan efficiency and dimensionless flow with the Euler number in the normal and stall operation regions of the fan, and a sizing factor (applied to the maximum flow through the fan). For the belt, the user specifies the maximum efficiency (or a curve with five coefficients that defines the maximum efficiency as a function of maximum fan shaft input power), three curves with three coefficients each that relate the belt partload efficiency to belt fractional output torque, the motor/fan pulley diameter ratio, the belt output torque capacity, and a sizing factor (applied to the maximum output torque of the belt). For the motor, the user specifies the maximum efficiency (or a curve with three coefficients that define the maximum efficiency as a function of maximum belt input power), a curve with three coefficients that relate the motor partload efficiency to motor fractional output power, and a sizing factor (applied to the maximum output power of the motor). For the variablefrequencydrive (VFD), the user specifies a curve with three coefficients that relate the VFD partload efficiency to motor fractional input power or to motor fractional speed, and a sizing factor (applied to the maximum output power of the VFD).
The models must decide whether the fan is on or off. The primary on/off trigger is the fan schedule. This is an on/off schedule associated with each fan: a value of 1 indicates the fan is on; a value of 0 indicates the fan is off. The fan schedule can be overruled by flags set by system availability managers. If the flag TurnFansOn is true, a zero fan schedule value will be overridden and the fan will be turned on. If the flag TurnFansOff is true the fan will be forced off. The inlet air mass flow rate must be greater than zero for the fan to be on.
Generally the fan is a passive component: it accepts the mass flow on its inlet node, uses it in its calculations of energy consumption and temperature rise, and passes it to the outlet node. However the fan maximum and minimum airflow rates act as absolute limits on the airflow rate.
For multispeed fans, the parent object determines the fan speed ratio (i.e., the selected speed of the fan motor) and uses this value in conjunction with the fan power ratio performance curve to calculate the full load fan power. This full load fan power is then used to determine the partload performance of the fan and motor assembly.
Simple (Single Speed) Fan Model
The following equations define the model for this fan:
˙Qtot=˙m⋅ΔP/(etot⋅ρair)
˙Qshaft=emotor⋅˙Qtot
˙Qtoair=˙Qshaft+(˙Qtot−˙Qshaft)⋅fmotortoair
hout=hin+˙Qtoair/˙m
wout=win
Tout=PsyTdbFnHW(hout,wout)
On/Off Fan Model
The on/off fan model is similar to the simple fan model with the exception that the on/off fan may cycle on and off during a simulation time step. The cycling rate of the fan is known as the run time fraction. The calculation of run time fraction accounts for the partload losses of other equipment used in the HVAC system. A partload factor (a.k.a. partload ratio) is first calculated for the fan as the ratio of the actual operating mass flow rate to the maximum fan mass flow rate. The run time fraction is then calculated as the partload factor divided by the partload fraction. The partload fraction is determined by other HVAC equipment in the simulation (Ref. DX coil) for use by this specific fan model.
fflow=˙m˙mmax
RTF=fflowPLF
The total fan power is then calculated as the maximum fan power multipled by the run time fraction.
⋅Qtot=RTF[⋅m⋅ΔP/(etot⋅ρair)]
The remaining calculations are the same as those described in the simple singlespeed fan mode.
MultiSpeed Fan Model
The model used to simulate a multispeed fan relies on the general fan laws to correct for speed changes in the fan’s motor and the corresponding change in fan power and fan total efficiency. Two performance curves are used to define the change in power and efficiency as shown below. The power ratio curve must be used to simulate a multispeed fan. The power ratio curve modifies the fan power based on a change in fan speed according to the cubic fan law. Exponents other than 3 are allowed. The efficiency ratio curve is used to correct for changes in nominal efficiency at alternate fan speeds. If either or both of these curves are not provided, the ratio is assumed to be 1.
The power ratio term is evaluated using an exponent performance curve. The form of the exponent curve equation is shown below.
ExponentCurveOutput=C1+C2∗xC3
The exponent performance curve is used to evaluate the power ratio as a function of speed ratio (x in the equation above). For typical fan performance, coefficient C1 = 0 and coefficient C2 = 1. The exponent (C3) is typically equal to 3 (cubed fan law) but other values are allowed.
PowerRatio=f(Nratio)n=ExponentCurveOutput
The efficiency ratio curve is a quadratic or cubic curve used to evaluate the efficiency ratio as a function of speed ratio (x in the following equation). The cubic equation form is shown below.
EfficiencyCurveOutput=C1+C2(x)2+C3(x)3
The curve coefficients are determine by correlating the normalized fan total efficiency (i.e., the actual fan total efficiency divided by the nominal fan total efficiency at a speed ratio of 1) to the speed ratio (i.e., the ratio of actual air flow rate through the fan to the maximum fan air flow rate).
EfficiencyRatio=f(Nratio)=EfficiencyCurveOutput
To account for the change in fan speed and to ensure that the run time fraction is accurately calculated, an adjustment to the previous calculation for run time fraction is made. This adjusts the calculated run time fraction to account for times when the fan speed has been reduced and the new speed is now considered the fan’s maximum speed. The speed adjusted run time fraction is then used in the fan power calculation.
RTF=RTFNratio
∙Qtot=RTF[∙mΔPetotρair](PowerRatioEfficiencyRatio)
Each of the performance curves described above may be used to model the performance of a multispeed fan motor, however, the power ratio curve must be used to envoke the multispeed simulation. These curves are used when the fan is used in an HVAC system having multiple flow rates (i.e., different flow rates in cooling and heating mode). If an HVAC system operates at the same speed in either cooling or heating mode, these curves are not required. When these curves are not used, the associated ratio term in the equation above is assumed to be 1. The remaining calculations are identical to the simple singlespeed fan model described above.
Variable Speed Fan Model
The model for the variable speed fan is similar to the simple singlespeed fan model except for a part load factor that multiplies the fan power consumption.
fflow=˙m/˙mdesign
fpl=c1+c2⋅fflow+c3⋅f2flow+c4⋅f3flow+c5⋅f4flow
˙Qtot=fpl⋅˙mdesign⋅ΔP/(etot⋅ρair)
The rest of the calculation is the same as for the simple fan.
Nomenclature for Simple Models[LINK]
˙Qtot is the fan power in watts;
˙m is the air mass flow in kg/s;
˙mdesign is the design (maximum) air flow in kg/s;
ΔP is the fan design pressure increase in Pascals;
etot is the fan total efficiency;
ρair is the air density at standard conditions in kg/m3;
emotor is the motor efficiency;
˙Qshaft is the fan shaft power in watts;
˙Qtoair is the power entering the air in watts;
Nratio is the ratio of actual fan flow rate (or speed) to maximum fan flow rate (or speed)
hin,hout are the inlet and outlet air stream specific enthalpies in J/kg;
win,wout are the inlet and outlet air stream humidity ratios;
Tout is the outlet air temperature in degrees C;
PsyTdbFnHW is the EnergyPlus psychrometric routine relating enthalpy and humidity ratio to temperature;
fflow is the flow fraction or partload ratio;
fpl is the part load factor.
Component Fan Model
The Fan:ComponentModel object is based upon combining: modified forms of fan, belt, motor, and variablefrequencydrive (VFD) element models (Stein and Hydeman 2004); a simplified fan pressure rise model with distribution system leakage, duct static pressure, and conditioned space pressure control effects included (Sherman and Wray 2010); and a diagnosticbased simplified duct staticpressure reset (SPR) scheme for airhandling systems (Federspiel 2004, 2005).
Fan electric power depends on fan air power (product of the airflow through and pressure rise across the fan), mechanical efficiencies (fan and belt), and electrical efficiencies (motor and drive). For systems with variable flows, none of these parameters is constant and all are interrelated. For example, Figure shows that fan efficiency strongly depends on fan flow as well as pressure rise, although this is not obvious from the manufacturer’s power and speed performance map (top left). Fan efficiency maps can be derived from the manufacturer’s performance map using a commerciallyavailable softwarebased data extraction tool such as DigitizeIt and by applying the following equation to the extracted data:
ηfan=ΔPfan⋅QfanHfan ()
where ΔPfanis the fan pressure rise (Pa); Qfan is the fan flow at standard conditions (m3/s); and Hfan is the fan shaft power (W). Fan speed values also can be derived from the manufacturer’s performance map using a similar softwarebased data extraction tool.
Fan pressure rise must be sufficient to overcome the airhandling system pressure drop, which depends on duct static pressure, on duct and equipment leakage, and on pressure drops across duct and ductlike elements (e.g., dampers, fittings), coils, and filters that are connected to the fan. Duct and ductlike pressure drops increase approximately as the square of the flow through them. However, pressure drops across coils and filters behave differently: they are proportional to the flow raised to a power n, which can approach one for highefficiency filters and wet coils (Liu et al. 2003, Trane 1999).
The relation between system pressure drop and flow defines what is commonly called a “system curve”. When system characteristics change, such as when the duct static pressure set point is varied, a family of system curves results. The intersections of these curves with fan curves (e.g., power as a function of pressure rise and flow) on a pressure versus flow plot define one or more loci of unique fan operating points. Each of these points has an associated fan efficiency, power, and speed.
Fan Pressure Rise Model: To calculate fan pressure rise based on flow through the fan, Sherman and Wray (2010) have developed a simple physicsbased datadriven four parameter duct system model for the purpose of simulating its system curve. The embodiment of the model for a fixed outdoor air fraction that can be applied to constant or variablevolume central airhandling systems is:
ΔPfan,tot=AfprQ2fan+BfprQfan+CfprQfan√Psm−Po+Dfpr(Psm−Po) ()
where ΔPfan,tot is the fan total pressure rise (Pa); Qfan is the fan flow at standard conditions (m3/s); Psm is the duct static pressure set point (Pa); Po is the static pressure of the spaces surrounding the ducts (Pa); and Afpr, Bfpr, Cfpr, and Dfpr are constant coefficients that represent different aspects of the fan pressure rise model as described below.
Fan static pressure rise is determined from the total pressure rise by subtracting the outlet velocity pressure:
ΔPfan=ΔPfan,tot−ρ2(QfanAfan,out)2 ()
where Afan,out is the fan outlet area (m2) and ρ is the air density at the fan inlet (kg/m3).
The first term in Equation looks like the common system curve in which the fan pressure rise is proportional to the square of the fan flow, but here it also depends implicitly on supply and return pressure losses, and in part on the fraction of the fan flow that is outdoor air (essentially “leaks” into and out of the return side of the system). Very often it is the only term considered, but that would only be correct with fixedposition dampers, no distribution system leakage, no linear resistance components, and no duct static pressure control.
The second term accounts for significant flow resistances in the system where the pressure difference is linearly proportional to the flow. Some filters and coils in the return may need this term to be adequately described. This term could be ignored if there are no linear components or if their pressure drops are very small compared to the other terms.
The third term, which depends on the fan flow and square root of the supply duct pressure Psm, accounts in part for air leakage from the supply system when damper positions are fixed or are changed independently of static pressure or fan flow. In this case, reducing or eliminating supply leakage results in a different system curve. This, however, might be only a minor “correction” to the simple system curves generally used. The third term is zero when VAV box dampers are modulated to control flow. Consequently, with variableposition supply dampers, reducing or eliminating supply leakage does not change the system curve.
The last term also accounts in part for leakage from the supply system when damper positions are fixed or are changed independently of static pressure or fan flow. This term indicates that the same fan pressure rise can be achieved by raising the duct pressure and closing dampers. The only change in the system in such a case is that the leakage flow may increase. The coefficient for this term is equal to one when the VAV box dampers are modulated to control flow. In both cases, this term may be the most important “correction” to the simple system curves generally used, especially at low flows.
In principle, especially for research applications of Energy Plus, the four individual fan pressure rise coefficients can be determined using duct design calculation tools for the entire system (e.g., RightCommDuct, UNIDUCT, Ductsize, Varitrane Duct Designer, TDuct). In practice, however, especially for existing buildings, the necessary details may not be known sufficiently. In that case, one can make active measurements at different combinations of fan flows, outside air fractions, damper positions (if they are variable), and duct static pressures and then use nonlinear system identification techniques to fit the data and determine the coefficients. If passive measurements can be made over a sufficiently long time so that a wide range of outside air settings, damper positions, and fan flows is obtained, one could instead regress the data to find all of the parameters in the equation. Unfortunately, there are no standardized test procedures available to draw upon, even though some standards appear to be related (e.g., ASHRAE 1999, 2008). Consequently, field test protocols need to be developed to determine the parameters for the new duct system model. These protocols then need to be integrated into standardized data collection and analysis tools such as Pacific Gas and Electric’s “Universal Translator” tool. The California Energy Commission is funding such a project; data, procedures, and tools from this project will support the EnergyPlus implementation of the fan and duct system models described here.
SPR Model: The model for ductstaticpressure reset (SPR) is based on a simple diagnostic procedure and a linear correlation between duct static pressure and supply fan airflow (Federspiel 2004, 2005). The diagnostic method involves measuring the static pressure at the duct static pressure sensor and the velocity pressure at the fan inlet (represents the fan flow) at multiple points over the fan’s operating range, while the VAV box dampers attempt to control flow in response to a constant thermostat setpoint. The goal of the test is to define the lowest duct static pressure where all VAV boxes are still in control (dampers modulating). In the model, the correlation between Psm and Qfan is as follows for Qfan,min ≤ Qfan ≤ Qfan,max:
Psm=Psm,min+(Psm,max−Psm,min)∗(Qfan−Qfan,min)(Qfan,max−Qfan,min)=C1+C2∗Qfan ()
Where
C1=Psm,min−C2∗Qfan,min and C2=(Psm,max−Psm,min)(Qfan,max−Qfan,min) ()
For Qfan < Qfan,min, Psm = Psm,min;forQfan > Qfan,max, Psm = Psm,max
Fan Efficiency and Shaft Input Power Model: A dimensionless parameter in the form of an Euler number can be used to simplify the description of fan static efficiency variations:
Eu=(ΔPfan∗D4fan)(ρ∗Q2fan) ()
where ΔPfan is the fan static pressure rise (Pa), Dfan is the fan wheel outer diameter (m), ρ is the air density at the fan inlet (kg/m3), and Qfan is the fan flow at standard conditions (m3/s). Eu is nominally the ratio of pressure forces across the fan to inertial forces at the fan wheel exit.
By plotting the normalized fan static efficiency (static efficiency / maximum static efficiency) versus the logarithm base 10 of the normalized Euler number (Eu / Eu at maximum static efficiency), the dimensionless performance of various fan sizes is very similar (as one might expect from the “fan laws”), but so also is the dimensionless performance of different types of fans (e.g., singleinlet plenum fans, doubleinlet housed centrifugal fans, mixed flow fans, vane axial fans, fans with backward or forward curved blades). An example of this correlation for the “normal operation” (nonstall) and stall regions of eight fans is shown in Figure.
This model uses a continuous function to represent the normalized fan efficiency (ηfan) variation. The normalized exponentialconditioned skewnormal functional relationship is:
ηfan(xfan)ηfan,max=e(−0.5∗Z21)[1+Z2Z2∙erf(Z2√2)e(−0.5∗Z23)[1+Z3Z3∙erf(Z3√2)
where
xfan = log10(Eu / Eumax)
Z1 = (xfan  afan) / bfan; Z2 = ((e(cfan∗xfan)*dfan*xfan)  afan) / bfan; Z3 = afan / bfan
For conceptual design when only rough estimates are needed, a generic curve (also shown in Figure) can be used and then one only needs to know the maximum efficiency (ηfan,max) and Eu at that maximum (Eumax) to entirely model the fan efficiency and hence fan power. The dimensionless coefficients for the generic normalized fan efficiency curve are as follows:
afan = 2.732094, bfan = 2.273014,cfan = 0.196344, dfan = 5.267518
In this case, the coefficient of determination R2 is 0.994. Also, the average and RMS differences between the efficiency values based on extracted data and the fitted generic curve are, respectively, about 0.5% and 1.4%; maximum differences are about ±9%.
If more accuracy is needed for a specific fan, a similarly shaped curve can be developed for that fan (using extracted data, Equation , and leastsquares regression techniques), segmented into normal operation and stall regions with specific coefficients fitted for each region. Figure shows an example, using the data from Figure.
For this specific fan, the coefficients are:
Normalized Fan Efficiency Coefficients for Specific BackwardCurved Plenum Fan
Normal (NonStall) 
0.072613 
0.833213 
0 
0.013911 
Stall 
2.354091 
2.117493 
0 
2.753264 
In this case, the average and RMS differences between the efficiency values based on extracted data and fitted specific curve (R2 is 0.999) are, respectively, about 0.1% and 0.3%; maximum differences are about ±0.7%. Field tests are especially needed to determine the coefficients for installed fans because fans are susceptible to “system effects” (e.g., inlet obstructions) that are not included in the manufacturer test data (AMCA 1990a).
To determine fan efficiency and fan shaft power at a particular time step, first calculate the fan pressure rise (ΔPfan) using the time step fan flow (Qfan) and Equation (and also using Equations and if there is SPR). Next, calculate Eu using Equation and then xfan (log10 normalized Eu), which is based on Eu and the specified Eumax (Eu at maximum efficiency) for the fan. The corresponding normalized efficiency (ηfan(xfan) / ηfan,max) is obtained using Equation . The fan efficiency (ηfan(xfan)) therefore is:
ηfan(xfan)=ηfan,max[ηfan(xfan)ηfan,max] ()
The fan shaft input power (mechanical, W) is:
Hfan=ΔPfan⋅Qfanηfan(xfan) ()
Fan Shaft Speed and Torque Modeling: For rotating elements, power (H) is the product of torque (τ) and rotational speed (ω), or conversely, torque is power divided by rotational speed (τ = H / ω).
The Stein and Hydeman variablefrequencydrive (VFD) component model correlates VFD efficiency as a linear function of VFD fractional output power (i.e., motor input power). Available data for about 50 drives from Saftronics were apparently used to develop their model, but those data represent VFD efficiency as a function of motor fractional speed (ωmotor / ωmotor,max, or nominally, drive output frequency divided by maximum output frequency, if motor slip is ignored).
To make use of the available data, the Stein and Hydeman linear correlation must intrinsically make an assumption that motor speed and torque have some fixed relationship. Although not documented, their assumption might be the common belief that fractional torque (τ/ τmax) for a motor is simply the square of its fractional speed. For fans serving duct systems with components such as filters and coils, with relatively low pressure drops elsewhere in the system, and for systems that have a nonzero controlled duct static pressure, this assumption may be inappropriate. Consequently, to make use of the available data and to avoid such assumptions, one needs to know the fraction of full speed at which fan components operate.
For the fan, dimensionless flow (φ) can be defined as (ASHRAE 1993):
φ=Qfanωfan⋅D3fan ()
where ωfan is the fan speed (rad/s). This parameter can be calculated for each operating point on the fan manufacturer’s performance map that represents fan speed as a function of flow and fan pressure rise. To simplify the data representation to a single curve (as we have done for fan efficiency), one can plot normalized dimensionless flow (dimensionless flow divided by maximum dimensionless flow, (φ(xfan) / φmax) as a function of the log base 10 of the normalized Euler (Eu) parameter (xfan). Figure shows an example plot derived from the manufacturer’s data in Figure.
Equation describes the sigmoidal functional form for normalized φ:
φ(xfan)φmax=(Aspd+Bspd/((1+e[(Cspd−xspd)/(Cspd−xspd)DspdDspd])Espd) ()
where Aspd, Bspd, Cspd, Dspd, and Espd are coefficients for the fan. Separate sets of coefficients for the normal operation and stall regions can be specified. For the curve shown in Figure, these coefficients are:
Dimensionless Flow Coefficients – Specific BackwardCurved Fan
Normal (NonStall) 
0 
1.001423 
0.123935 
0.476026 
1 
Stall 
0 
5.924993 
1.916316 
0.851779 
1 
For a generic centrifugal fan with backwardcurved blades, the coefficients are:
Dimensionless Flow Coefficients – Generic BackwardCurved Fan
Normal (NonStall) 
0.551396 
1.551467 
0.442200 
0.414006 
0.234867 
Stall 
0.000608 
0.586366 
0.021775 
0.063218 
0.072827 
For any operating point of flow (Qfan) and fan pressure rise (ΔPfan), one can calculate the corresponding Euler number (Eu) for the given fan, and then determine the corresponding normalized dimensionless flow (φ(xfan) / φmax) from the sigmoidal dimensionless flow function (Equation ). With φ(xfan) / φmax determined and knowing φmax for the fan, the dimensionless flow is:
φ(xfan)=φmax[φ(xfan)φmax] ()
With φ(xfan) determined, the fan rotational speed (rad/s) is:
ωfan=Qfan(φ(xfan)∗D3fan) ()
Fan shaft torque (N∙m), which the belt drive must supply to the fan shaft, is then:
τfan=Hfanωfan ()
The fraction of fullload driven torque for the belt (τfan / τbelt,max), which is typically called “belt load”, is thus τfan divided by the belt torque capacity (τbelt,max). For a particular belt type and crosssection, belt torque capacity can be determined from manufacturer’s information such as a shaft speed versus power chart.
Ignoring belt slip, motor shaft speed (ωmotor, rad/s) can then be determined using the fan speed (ωfan) and the motor/fan pulley diameter ratio (Dpulley,motor / Dpulley,fan), which is typically called the “drive” ratio:
ωmotor=ωfan(Dpulley,motorDpulley,fan) ()
Belt, Motor, and VariableFrequencyDrive Efficiency and Input Power Models – Overview: The models for belt drives, motors, and VFDs that Stein and Hydeman included in their airhandling system model represent maximum efficiency as a function of power input to the adjacent downstream component (e.g., the power input to a fan shaft by a belt drive), but they do not include partload models. The partload models are needed, because partload efficiency for these components can fall off rapidly at low load (e.g., to zero at zero load) and many systems operate occasionally (and some much of the time) at low loads, in part because of current practices that result in substantial oversizing of components.
Belt Efficiency and Input Power Model: Figure shows three maximum efficiency (ηbelt,max) curves for belts (low, medium, and high) as a function of maximum fan shaft torque. This set of efficiency curves is based on belt drive loss data from AMCA Publication 203 (1990b), which reportedly is an aggregation of data from over 400 tests.
To determine ηbelt,max if data for a specific belt are not available, first use the maximum fan shaft input power (Hfan,max) for the load spectrum to calculate the natural logarithm of belt power capacity:
xbelt,max =ln(Hfan,max) with Hfan,max expressed in terms of hp ()
Then, use Equations and , along with a choice of low, medium, or high efficiency coefficients from Table, to calculate ηbelt,max.
The quartic polynomial curves in Figure and their coefficients are as follows:
ηbelt,max,ln=c1+c2⋅xbelt,max+c3⋅x2belt,max+c4⋅x3belt,max+c5⋅x4belt,max ()
Belt Maximum Efficiency Curve Coefficients
High (Low Loss) 
6.502E2 
2.475E2 
6.841E3 
9.383E4 
5.168E5 
Medium (Medium Loss) 
9.504E2 
3.415E2 
8.897E3 
1.159E3 
6.132E5 
Low (High Loss) 
1.422E1 
5.112E2 
1.353E2 
1.814E3 
9.909E5 
The belt maximum efficiency is therefore:
ηbelt,max=eηbelt,max,ln ()
The normalized partload belt efficiency model is provided to modify the maximum efficiency of the belt. This model is based on partload data for belts in ACEEE’s handbook on motor systems (Nadel et al. 2002). To determine the normalized belt efficiency at partload (ηbelt(xbelt)/* ηbelt,max), use the fan shaft (belt output) fractional torque (xbelt* = τbelt / τbelt,max; τbelt = τfan) as the belt fractional load in Equation or (select the equation for the region that contains xbelt) and use coefficients from Table that correspond to that region and the belt type: VBelt or synchronous (toothed). Figure shows a graphical representation of the curves defined by these equations and coefficients.
The belt normalized (partload) efficiency curves in Figure for Regions 1 and 3 (single rectangular hyperbola type 2) and Region 2 (exponential decay), respectively, and their coefficients are as follows:
Region 1 (0 < = xbelt < xbelt,trans) and Region 3 (xbelt > 1):
ηbelt(xbelt)ηbelt,max=[abelt⋅xbeltbbelt+xbelt]+cbelt⋅xbelt ()
Region 2 (xbelt,trans < = xbelt < = 1):
ηbelt(xbelt)ηbelt,max=abelt+bbelt⋅e(cbelt⋅xbelt) ()
where xbelt = τ belt / τ belt,max
Belt efficiency therefore is:
ηbelt(xbelt)=ηbelt,max[ηbelt(xbelt)ηbelt,max] ()
Belt input power (mechanical, W) at the motor shaft is:
Hbelt=Hfanηbelt(xbelt) ()
Belt Normalized Efficiency Curve Coefficients
Belt 
0.167 
1 
0.920797 
0.026269 
0.151594 


2 
1.011965 
0.339038 
3.43626 


3 
1.037778 
0.010307 
0.026815 
Synchronous 
0.137 
1 
0.982167 
0.049135 
0.158164 


2 
1.002134 
0.531885 
5.295707 


3 
1 
0 
0 
Motor Efficiency and Input Power Model: Figure shows three maximum efficiency (ηmotor,max) curves for motors (low, medium, and high) as a function of rated motor output power (belt input power). This set of efficiency curves is based on maximum efficiency data in DOE’s MotorMaster+ database (2003) from about 800 tests. It is provided for use in the absence of manufacturer’s data.
To determine ηmotor,max if data for a specific motor are not available (e.g., as listed in Table), first use the maximum belt input power (Hbelt,max) for the load spectrum (multiplied by whatever oversizing factor may be desired) to calculate the natural logarithm of belt power capacity:
xmotor,max =ln(Hbelt,max) with Hbelt,max expressed in terms of hp ()
Then, use Equation , along with a choice of low, medium, or high efficiency coefficients from Table, to calculate ηmotor,max.
The maximum motor efficiency curves in Figure (single rectangular hyperbola type 1) and their coefficients are as follows:
ηmotor,max=[amotor,max⋅xmotor,maxbmotor,max+xmotor,max]+cmotor,max ()
Motor Maximum Efficiency Curve Coefficients
HighEfficiency 
0.196205 
3.653654 
0.839926 
MidEfficiency 
0.292280 
3.368739 
0.762471 
LowEfficiency 
0.395895 
3.065240 
0.674321 
The normalized partload motor efficiency model is provided to modify the maximum efficiency of the motor. This model is based on partload data for motors from DOE’s MotorMaster+ database (2003). To determine the normalized motor efficiency at partload (ηmotor(xmotor)/* ηmotor,max), use the motor fractional power output (xmotor = Hbelt* / Hbelt,max) as the fraction of motor output power in Equation and use coefficients from Table, or for a specific motor, determined from DOE MotorMaster+ data or from manufacturer’s data. Figure shows a graphical representation of eight example curves defined by these equations and coefficients.
The example motor normalized efficiency curves (single rectangular hyperbola type 2) and their coefficients as a function of motor load fraction in Figure are determined from DOE MotorMaster+ data and are as follows:
ηmotor(xmotor)ηmotor,max=[aPLmotor⋅xmotorbPLmotor+xmotor]+cPLmotor⋅xmotor ()
Example Motor Normalized Efficiency Curve Coefficients
8 
1 
0.6675 
1.096694 
0.097126 
0.002011 
4 
1 
0.7787 
1.092165 
0.08206 
0.0072 

5 
0.84 
1.223684 
0.08467 
0.135186 

10 
0.8745 
1.146258 
0.045766 
0.110367 

25 
0.8991 
1.137209 
0.050236 
0.08915 

50 
0.9129 
1.088803 
0.029753 
0.064058 

75 
0.9259 
1.07714 
0.029005 
0.04935 

100 
0.9499 
1.035294 
0.012948 
0.024708 

125 
0.9527 
1.030968 
0.010696 
0.023514 
Motor efficiency therefore is:
ηmotor(xmotor)=ηmotor,max[ηmotor(xmotor)ηmotor,max] ()
Motor input power (electrical, W) is:
Hmotor=Hbeltηmotor(xmotor) ()
VFD Efficiency and Input Power Model: For VFDs, published performance data are limited. Data from DOE (2008) suggest using a functional relation similar to that used for motors to represent VFD efficiency (ηVFD) as a function of the fraction of fullload motor input power (xVFD = Hmotor / Hmotor,max), as shown in Figure, or as a function of the fraction of fullload speed (xVFD = ωmotor / ωmotor,max), as shown in Figure.
The performance curve (single rectangular hyperbola type 2) used here for VFD efficiency is:
ηVFD=[aVFD⋅xVFDbVFD+xVFD]+cVFD⋅xVFD ()
Example coefficients derived from the DOE data as a function of the fraction of fullload motor input power are listed in Table.
Example VFD Efficiency Curve Coefficients
3 
0.978856 
0.034247 
0.007862 
5 
0.977485 
0.028413 
0.002733 
10 
0.978715 
0.022227 
0.001941 
20 
0.984973 
0.017545 
0.000475 
30 
0.987405 
0.015536 
0.005937 
50 
0.987910 
0.018376 
0.001692 
60 
0.971904 
0.014537 
0.011849 
75 
0.991874 
0.017897 
0.001301 
100 
0.982384 
0.012598 
0.001405 
>= 200 
0.984476 
0.009828 
0.004560 
VFD input power (electrical, W) is:
HVFD=HmotorηVFD ()
System Total Efficiency: The combined efficiency of the fan system components (i.e., fan, belt, motor, and VFD) is:
ηTOTAL=ηfan⋅ηbelt⋅ηmotor⋅ηVFD (=Qfan⋅ΔPfanHVFD) ()
Heat Loss to Air: To calculate the temperature rise from waste heat entering the air stream from the fan, belt, and motor, it is assumed that the userspecified “motor in air fraction” applies to the belt and motor but not to the VFD. The power “lost” to the air (W) is:
˙Qtoair=Hfan+(Hmotor−Hfan)⋅fmotortoair ()
AMCA. 1990a. “Fans and Systems”. Publication 20190. Arlington Heights, IL: Air Movement and Control Association International.
AMCA. 1990b. “Field Performance Measurement of Fan Systems”. Publication 20390. Arlington Heights, IL: Air Movement and Control Association International.
ASHRAE. 1993. HVAC 2 Toolkit: Algorithms and Subroutines for Secondary HVAC System Energy Calculations. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
ASHRAE. 1999. “ANSI/ASHRAE Standard 120 Method of Testing to Determine Flow Resistance of HVAC Ducts and Fittings”. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
ASHRAE. 2008. “ANSI/ASHRAE Standard 126 Method of Testing HVAC Air Ducts and Fittings”. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
DOE. 2003. “MotorMaster 4.0 User Guide”. U.S. Department of Energy, Industrial Technologies Program.
DOE. 2008. “Energy Tips – Motor: Motor Tip Sheet #11”. U.S. Department of Energy, Industrial Technologies Program. June.
Federspiel, C. 2004. “Detecting Optimal Fan Pressure”. Final Report of Federspiel Controls to the CEC Energy Innovations Small Grant Program. Grant #: 0203.
Federspiel, C. 2005. “Detecting Critical Supply Duct Pressure”. ASHRAE Transactions, Vol. 111, Part 1. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
Heredos, Francis P. 1987. Selection and Application of Multispeed Motors, IEEE Transactions on Industry Applications, Vol. 1A23, No. 2, March/April.
Liu, M., D.E. Claridge, and S. Deng. 2003. “An Air Filter Pressure Loss Model for Fan Energy Calculation in AirHandling Units”. International Journal of Energy Research. Vol. 27, pp. 589600.
Nadel, S., R.N. Elliot, M. Shepard, S. Greenberg, G. Katz, and A.T. de Almeida. 2002. “EnergyEfficient Motor Systems: A Handbook on Technology, Program, and Policy Opportunities” 2nd Edition. Washington, DC: American Council for an Energy Efficient Economy. p.188.
Sherman, M.H. and C.P. Wray. 2010. “Parametric System Curves: Correlations Between Fan Pressure Rise and Flow for Large Commercial Buildings”. Lawrence Berkeley National Laboratory Report, LBNL3542E.
Stein, J. and M.M. Hydeman. 2004. “Development and Testing of the Characteristic Curve Fan Model”. ASHRAE Transactions, Vol. 110, Part 1. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
Trane. 1999. “DeltaFlo Coils: Data Catalog PLAHCOIL000D2799”. LaCrosse, WI: The Trane Company. July.
Air System Fans[LINK]
Overview[LINK]
Three input objects (Fan:ConstantVolume, Fan:VariableVolume, and Fan:OnOff) provide models for fans – the prime movers in most of the air loop and zonal air conditioning systems in EnergyPlus. For these input objects, EnergyPlus uses a simple polynomialbased curvefit model to describe the relation between the volumetric flow rate and the fan electric power, with no explicit modeling of fan pressure rise. Consequently, ductstaticpressure reset strategies cannot be modeled using this approach. For CAV and VAV central airhandling systems, a fourth input object (Fan:ComponentModel) provides a simple physicsbased model for flowdependent fan pressure rise and detailed models for fan, belt, motor, and variablefrequencydrive efficiencies and energy use. This latter input object provides the capability to analyze the effects of ductstaticpressure reset strategies, as well as changes in fan system element performance. All of these fan models calculate the fan energy usage, which is often a large fraction of HVAC energy use and a significant portion of the building energy consumption. They also calculate the temperature rise in the air stream caused by the static pressure increase as the air goes through the fan. If the motor and belt are in the air stream, waste heat from the motor and belt also acts to raise the air stream temperature.
Model[LINK]
Each of the four fan models is a forward type: the model inputs describe the fan characteristics and the conditions of the air at the fan inlet; the outputs are the fan electrical power consumption and the conditions of the air at the fan outlet. The model algorithms and data are contained in the Fans module in EnergyPlus.
Inputs and Data[LINK]
For the Fan:ConstantVolume, Fan:VariableVolume, and Fan:OnOff models, the user describes the fan by entering values for the design pressure rise across the fan, the design volumetric flow rate, the fan total efficiency, and the fan motor efficiency. The user also needs to specify the fraction of the fan waste heat that will enter the air stream (usually 0 or 1). For the Fan:VariableVolume model, the user must also enter the coefficients of a 4th order polynomial that relates the fan mass flow rate to the fan power consumption.The independent variable is the volumetric flow fraction; the dependent variable is the fan power part load ratio. For multispeed fans, the user must enter a fan power ratio as a function of speed ratio performance curve name. Multispeed fans can only be simulated in a parent object which allows multiple fan speeds (e.g., AirLoopHVAC:Unitary:Furnace:HeatCool, ZoneHVAC:PackagedTerminalAirConditioner, etc.). An optional efficiency curve may also be used when simulating multispeed fans to adjust the fan total efficiency as the fan speed changes.
For the Fan:ComponentModel object, the user describes the fan in more detail, and also describes the duct system characteristics as seen by the fan. In particular, the user specifies a pressure rise curve with four coefficients that relates the fan total pressure rise to the volumetric flow through the fan, the ductstaticpressure setpoint, and the static pressure of the spaces surrounding the ducts. If ductstaticpressure reset is used, the user enters a linear curve with two coefficients that relates the pressure setpoint to the volumetric flow through the fan. For the fan itself, the user specifies the fan geometry (wheel diameter and outlet area), maximum efficiency, the Euler number corresponding to the maximum efficiency, the maximum dimensionless flow, the names of four curves with several coefficients that describe the variation of fan efficiency and dimensionless flow with the Euler number in the normal and stall operation regions of the fan, and a sizing factor (applied to the maximum flow through the fan). For the belt, the user specifies the maximum efficiency (or a curve with five coefficients that defines the maximum efficiency as a function of maximum fan shaft input power), three curves with three coefficients each that relate the belt partload efficiency to belt fractional output torque, the motor/fan pulley diameter ratio, the belt output torque capacity, and a sizing factor (applied to the maximum output torque of the belt). For the motor, the user specifies the maximum efficiency (or a curve with three coefficients that define the maximum efficiency as a function of maximum belt input power), a curve with three coefficients that relate the motor partload efficiency to motor fractional output power, and a sizing factor (applied to the maximum output power of the motor). For the variablefrequencydrive (VFD), the user specifies a curve with three coefficients that relate the VFD partload efficiency to motor fractional input power or to motor fractional speed, and a sizing factor (applied to the maximum output power of the VFD).
Control[LINK]
The models must decide whether the fan is on or off. The primary on/off trigger is the fan schedule. This is an on/off schedule associated with each fan: a value of 1 indicates the fan is on; a value of 0 indicates the fan is off. The fan schedule can be overruled by flags set by system availability managers. If the flag TurnFansOn is true, a zero fan schedule value will be overridden and the fan will be turned on. If the flag TurnFansOff is true the fan will be forced off. The inlet air mass flow rate must be greater than zero for the fan to be on.
Generally the fan is a passive component: it accepts the mass flow on its inlet node, uses it in its calculations of energy consumption and temperature rise, and passes it to the outlet node. However the fan maximum and minimum airflow rates act as absolute limits on the airflow rate.
For multispeed fans, the parent object determines the fan speed ratio (i.e., the selected speed of the fan motor) and uses this value in conjunction with the fan power ratio performance curve to calculate the full load fan power. This full load fan power is then used to determine the partload performance of the fan and motor assembly.
Simulation[LINK]
Simple (Single Speed) Fan Model
The following equations define the model for this fan:
˙Qtot=˙m⋅ΔP/(etot⋅ρair)
˙Qshaft=emotor⋅˙Qtot
˙Qtoair=˙Qshaft+(˙Qtot−˙Qshaft)⋅fmotortoair
hout=hin+˙Qtoair/˙m
wout=win
Tout=PsyTdbFnHW(hout,wout)
On/Off Fan Model
The on/off fan model is similar to the simple fan model with the exception that the on/off fan may cycle on and off during a simulation time step. The cycling rate of the fan is known as the run time fraction. The calculation of run time fraction accounts for the partload losses of other equipment used in the HVAC system. A partload factor (a.k.a. partload ratio) is first calculated for the fan as the ratio of the actual operating mass flow rate to the maximum fan mass flow rate. The run time fraction is then calculated as the partload factor divided by the partload fraction. The partload fraction is determined by other HVAC equipment in the simulation (Ref. DX coil) for use by this specific fan model.
fflow=˙m˙mmax
RTF=fflowPLF
The total fan power is then calculated as the maximum fan power multipled by the run time fraction.
⋅Qtot=RTF[⋅m⋅ΔP/(etot⋅ρair)]
The remaining calculations are the same as those described in the simple singlespeed fan mode.
MultiSpeed Fan Model
The model used to simulate a multispeed fan relies on the general fan laws to correct for speed changes in the fan’s motor and the corresponding change in fan power and fan total efficiency. Two performance curves are used to define the change in power and efficiency as shown below. The power ratio curve must be used to simulate a multispeed fan. The power ratio curve modifies the fan power based on a change in fan speed according to the cubic fan law. Exponents other than 3 are allowed. The efficiency ratio curve is used to correct for changes in nominal efficiency at alternate fan speeds. If either or both of these curves are not provided, the ratio is assumed to be 1.
The power ratio term is evaluated using an exponent performance curve. The form of the exponent curve equation is shown below.
ExponentCurveOutput=C1+C2∗xC3
The exponent performance curve is used to evaluate the power ratio as a function of speed ratio (x in the equation above). For typical fan performance, coefficient C1 = 0 and coefficient C2 = 1. The exponent (C3) is typically equal to 3 (cubed fan law) but other values are allowed.
PowerRatio=f(Nratio)n=ExponentCurveOutput
The efficiency ratio curve is a quadratic or cubic curve used to evaluate the efficiency ratio as a function of speed ratio (x in the following equation). The cubic equation form is shown below.
EfficiencyCurveOutput=C1+C2(x)2+C3(x)3
The curve coefficients are determine by correlating the normalized fan total efficiency (i.e., the actual fan total efficiency divided by the nominal fan total efficiency at a speed ratio of 1) to the speed ratio (i.e., the ratio of actual air flow rate through the fan to the maximum fan air flow rate).
EfficiencyRatio=f(Nratio)=EfficiencyCurveOutput
To account for the change in fan speed and to ensure that the run time fraction is accurately calculated, an adjustment to the previous calculation for run time fraction is made. This adjusts the calculated run time fraction to account for times when the fan speed has been reduced and the new speed is now considered the fan’s maximum speed. The speed adjusted run time fraction is then used in the fan power calculation.
RTF=RTFNratio
∙Qtot=RTF[∙mΔPetotρair](PowerRatioEfficiencyRatio)
Each of the performance curves described above may be used to model the performance of a multispeed fan motor, however, the power ratio curve must be used to envoke the multispeed simulation. These curves are used when the fan is used in an HVAC system having multiple flow rates (i.e., different flow rates in cooling and heating mode). If an HVAC system operates at the same speed in either cooling or heating mode, these curves are not required. When these curves are not used, the associated ratio term in the equation above is assumed to be 1. The remaining calculations are identical to the simple singlespeed fan model described above.
Variable Speed Fan Model
The model for the variable speed fan is similar to the simple singlespeed fan model except for a part load factor that multiplies the fan power consumption.
fflow=˙m/˙mdesign
fpl=c1+c2⋅fflow+c3⋅f2flow+c4⋅f3flow+c5⋅f4flow
˙Qtot=fpl⋅˙mdesign⋅ΔP/(etot⋅ρair)
The rest of the calculation is the same as for the simple fan.
Nomenclature for Simple Models[LINK]
˙Qtot is the fan power in watts;
˙m is the air mass flow in kg/s;
˙mdesign is the design (maximum) air flow in kg/s;
ΔP is the fan design pressure increase in Pascals;
etot is the fan total efficiency;
ρair is the air density at standard conditions in kg/m3;
emotor is the motor efficiency;
˙Qshaft is the fan shaft power in watts;
˙Qtoair is the power entering the air in watts;
Nratio is the ratio of actual fan flow rate (or speed) to maximum fan flow rate (or speed)
hin,hout are the inlet and outlet air stream specific enthalpies in J/kg;
win,wout are the inlet and outlet air stream humidity ratios;
Tout is the outlet air temperature in degrees C;
PsyTdbFnHW is the EnergyPlus psychrometric routine relating enthalpy and humidity ratio to temperature;
fflow is the flow fraction or partload ratio;
fpl is the part load factor.
Component Fan Model
The Fan:ComponentModel object is based upon combining: modified forms of fan, belt, motor, and variablefrequencydrive (VFD) element models (Stein and Hydeman 2004); a simplified fan pressure rise model with distribution system leakage, duct static pressure, and conditioned space pressure control effects included (Sherman and Wray 2010); and a diagnosticbased simplified duct staticpressure reset (SPR) scheme for airhandling systems (Federspiel 2004, 2005).
Fan electric power depends on fan air power (product of the airflow through and pressure rise across the fan), mechanical efficiencies (fan and belt), and electrical efficiencies (motor and drive). For systems with variable flows, none of these parameters is constant and all are interrelated. For example, Figure shows that fan efficiency strongly depends on fan flow as well as pressure rise, although this is not obvious from the manufacturer’s power and speed performance map (top left). Fan efficiency maps can be derived from the manufacturer’s performance map using a commerciallyavailable softwarebased data extraction tool such as DigitizeIt and by applying the following equation to the extracted data:
ηfan=ΔPfan⋅QfanHfan ()
where ΔPfanis the fan pressure rise (Pa); Qfan is the fan flow at standard conditions (m3/s); and Hfan is the fan shaft power (W). Fan speed values also can be derived from the manufacturer’s performance map using a similar softwarebased data extraction tool.
Fan pressure rise must be sufficient to overcome the airhandling system pressure drop, which depends on duct static pressure, on duct and equipment leakage, and on pressure drops across duct and ductlike elements (e.g., dampers, fittings), coils, and filters that are connected to the fan. Duct and ductlike pressure drops increase approximately as the square of the flow through them. However, pressure drops across coils and filters behave differently: they are proportional to the flow raised to a power n, which can approach one for highefficiency filters and wet coils (Liu et al. 2003, Trane 1999).
The relation between system pressure drop and flow defines what is commonly called a “system curve”. When system characteristics change, such as when the duct static pressure set point is varied, a family of system curves results. The intersections of these curves with fan curves (e.g., power as a function of pressure rise and flow) on a pressure versus flow plot define one or more loci of unique fan operating points. Each of these points has an associated fan efficiency, power, and speed.
Example Fan Performance Maps  Manufacturer’s Data from Loren Cook Company, plus Derived Static Efficiency (ThreeDimensional and Contours) (Dashed Parabolic Curve is “Do Not Select Line”)
Fan Pressure Rise Model: To calculate fan pressure rise based on flow through the fan, Sherman and Wray (2010) have developed a simple physicsbased datadriven four parameter duct system model for the purpose of simulating its system curve. The embodiment of the model for a fixed outdoor air fraction that can be applied to constant or variablevolume central airhandling systems is:
ΔPfan,tot=AfprQ2fan+BfprQfan+CfprQfan√Psm−Po+Dfpr(Psm−Po) ()
where ΔPfan,tot is the fan total pressure rise (Pa); Qfan is the fan flow at standard conditions (m3/s); Psm is the duct static pressure set point (Pa); Po is the static pressure of the spaces surrounding the ducts (Pa); and Afpr, Bfpr, Cfpr, and Dfpr are constant coefficients that represent different aspects of the fan pressure rise model as described below.
Fan static pressure rise is determined from the total pressure rise by subtracting the outlet velocity pressure:
ΔPfan=ΔPfan,tot−ρ2(QfanAfan,out)2 ()
where Afan,out is the fan outlet area (m2) and ρ is the air density at the fan inlet (kg/m3).
The first term in Equation looks like the common system curve in which the fan pressure rise is proportional to the square of the fan flow, but here it also depends implicitly on supply and return pressure losses, and in part on the fraction of the fan flow that is outdoor air (essentially “leaks” into and out of the return side of the system). Very often it is the only term considered, but that would only be correct with fixedposition dampers, no distribution system leakage, no linear resistance components, and no duct static pressure control.
The second term accounts for significant flow resistances in the system where the pressure difference is linearly proportional to the flow. Some filters and coils in the return may need this term to be adequately described. This term could be ignored if there are no linear components or if their pressure drops are very small compared to the other terms.
The third term, which depends on the fan flow and square root of the supply duct pressure Psm, accounts in part for air leakage from the supply system when damper positions are fixed or are changed independently of static pressure or fan flow. In this case, reducing or eliminating supply leakage results in a different system curve. This, however, might be only a minor “correction” to the simple system curves generally used. The third term is zero when VAV box dampers are modulated to control flow. Consequently, with variableposition supply dampers, reducing or eliminating supply leakage does not change the system curve.
The last term also accounts in part for leakage from the supply system when damper positions are fixed or are changed independently of static pressure or fan flow. This term indicates that the same fan pressure rise can be achieved by raising the duct pressure and closing dampers. The only change in the system in such a case is that the leakage flow may increase. The coefficient for this term is equal to one when the VAV box dampers are modulated to control flow. In both cases, this term may be the most important “correction” to the simple system curves generally used, especially at low flows.
In principle, especially for research applications of Energy Plus, the four individual fan pressure rise coefficients can be determined using duct design calculation tools for the entire system (e.g., RightCommDuct, UNIDUCT, Ductsize, Varitrane Duct Designer, TDuct). In practice, however, especially for existing buildings, the necessary details may not be known sufficiently. In that case, one can make active measurements at different combinations of fan flows, outside air fractions, damper positions (if they are variable), and duct static pressures and then use nonlinear system identification techniques to fit the data and determine the coefficients. If passive measurements can be made over a sufficiently long time so that a wide range of outside air settings, damper positions, and fan flows is obtained, one could instead regress the data to find all of the parameters in the equation. Unfortunately, there are no standardized test procedures available to draw upon, even though some standards appear to be related (e.g., ASHRAE 1999, 2008). Consequently, field test protocols need to be developed to determine the parameters for the new duct system model. These protocols then need to be integrated into standardized data collection and analysis tools such as Pacific Gas and Electric’s “Universal Translator” tool. The California Energy Commission is funding such a project; data, procedures, and tools from this project will support the EnergyPlus implementation of the fan and duct system models described here.
SPR Model: The model for ductstaticpressure reset (SPR) is based on a simple diagnostic procedure and a linear correlation between duct static pressure and supply fan airflow (Federspiel 2004, 2005). The diagnostic method involves measuring the static pressure at the duct static pressure sensor and the velocity pressure at the fan inlet (represents the fan flow) at multiple points over the fan’s operating range, while the VAV box dampers attempt to control flow in response to a constant thermostat setpoint. The goal of the test is to define the lowest duct static pressure where all VAV boxes are still in control (dampers modulating). In the model, the correlation between Psm and Qfan is as follows for Qfan,min ≤ Qfan ≤ Qfan,max:
Psm=Psm,min+(Psm,max−Psm,min)∗(Qfan−Qfan,min)(Qfan,max−Qfan,min)=C1+C2∗Qfan ()
Where
C1=Psm,min−C2∗Qfan,min and C2=(Psm,max−Psm,min)(Qfan,max−Qfan,min) ()
For Qfan < Qfan,min, Psm = Psm,min;forQfan > Qfan,max, Psm = Psm,max
Fan Efficiency and Shaft Input Power Model: A dimensionless parameter in the form of an Euler number can be used to simplify the description of fan static efficiency variations:
Eu=(ΔPfan∗D4fan)(ρ∗Q2fan) ()
where ΔPfan is the fan static pressure rise (Pa), Dfan is the fan wheel outer diameter (m), ρ is the air density at the fan inlet (kg/m3), and Qfan is the fan flow at standard conditions (m3/s). Eu is nominally the ratio of pressure forces across the fan to inertial forces at the fan wheel exit.
By plotting the normalized fan static efficiency (static efficiency / maximum static efficiency) versus the logarithm base 10 of the normalized Euler number (Eu / Eu at maximum static efficiency), the dimensionless performance of various fan sizes is very similar (as one might expect from the “fan laws”), but so also is the dimensionless performance of different types of fans (e.g., singleinlet plenum fans, doubleinlet housed centrifugal fans, mixed flow fans, vane axial fans, fans with backward or forward curved blades). An example of this correlation for the “normal operation” (nonstall) and stall regions of eight fans is shown in Figure.
Normalized Efficiency Curves for Eight Fans in Dimensionless Space *(BC = backward curved, FC = forward curved; SI = single inlet, DI = double inlet)*
This model uses a continuous function to represent the normalized fan efficiency (ηfan) variation. The normalized exponentialconditioned skewnormal functional relationship is:
ηfan(xfan)ηfan,max=e(−0.5∗Z21)[1+Z2Z2∙erf(Z2√2)e(−0.5∗Z23)[1+Z3Z3∙erf(Z3√2)
where
xfan = log10(Eu / Eumax)
Z1 = (xfan  afan) / bfan; Z2 = ((e(cfan∗xfan)*dfan*xfan)  afan) / bfan; Z3 = afan / bfan
For conceptual design when only rough estimates are needed, a generic curve (also shown in Figure) can be used and then one only needs to know the maximum efficiency (ηfan,max) and Eu at that maximum (Eumax) to entirely model the fan efficiency and hence fan power. The dimensionless coefficients for the generic normalized fan efficiency curve are as follows:
afan = 2.732094, bfan = 2.273014,cfan = 0.196344, dfan = 5.267518
In this case, the coefficient of determination R2 is 0.994. Also, the average and RMS differences between the efficiency values based on extracted data and the fitted generic curve are, respectively, about 0.5% and 1.4%; maximum differences are about ±9%.
If more accuracy is needed for a specific fan, a similarly shaped curve can be developed for that fan (using extracted data, Equation , and leastsquares regression techniques), segmented into normal operation and stall regions with specific coefficients fitted for each region. Figure shows an example, using the data from Figure.
Example Normalized Efficiency Data for One Fan
For this specific fan, the coefficients are:
In this case, the average and RMS differences between the efficiency values based on extracted data and fitted specific curve (R2 is 0.999) are, respectively, about 0.1% and 0.3%; maximum differences are about ±0.7%. Field tests are especially needed to determine the coefficients for installed fans because fans are susceptible to “system effects” (e.g., inlet obstructions) that are not included in the manufacturer test data (AMCA 1990a).
To determine fan efficiency and fan shaft power at a particular time step, first calculate the fan pressure rise (ΔPfan) using the time step fan flow (Qfan) and Equation (and also using Equations and if there is SPR). Next, calculate Eu using Equation and then xfan (log10 normalized Eu), which is based on Eu and the specified Eumax (Eu at maximum efficiency) for the fan. The corresponding normalized efficiency (ηfan(xfan) / ηfan,max) is obtained using Equation . The fan efficiency (ηfan(xfan)) therefore is:
ηfan(xfan)=ηfan,max[ηfan(xfan)ηfan,max] ()
The fan shaft input power (mechanical, W) is:
Hfan=ΔPfan⋅Qfanηfan(xfan) ()
Fan Shaft Speed and Torque Modeling: For rotating elements, power (H) is the product of torque (τ) and rotational speed (ω), or conversely, torque is power divided by rotational speed (τ = H / ω).
The Stein and Hydeman variablefrequencydrive (VFD) component model correlates VFD efficiency as a linear function of VFD fractional output power (i.e., motor input power). Available data for about 50 drives from Saftronics were apparently used to develop their model, but those data represent VFD efficiency as a function of motor fractional speed (ωmotor / ωmotor,max, or nominally, drive output frequency divided by maximum output frequency, if motor slip is ignored).
To make use of the available data, the Stein and Hydeman linear correlation must intrinsically make an assumption that motor speed and torque have some fixed relationship. Although not documented, their assumption might be the common belief that fractional torque (τ/ τmax) for a motor is simply the square of its fractional speed. For fans serving duct systems with components such as filters and coils, with relatively low pressure drops elsewhere in the system, and for systems that have a nonzero controlled duct static pressure, this assumption may be inappropriate. Consequently, to make use of the available data and to avoid such assumptions, one needs to know the fraction of full speed at which fan components operate.
For the fan, dimensionless flow (φ) can be defined as (ASHRAE 1993):
φ=Qfanωfan⋅D3fan ()
where ωfan is the fan speed (rad/s). This parameter can be calculated for each operating point on the fan manufacturer’s performance map that represents fan speed as a function of flow and fan pressure rise. To simplify the data representation to a single curve (as we have done for fan efficiency), one can plot normalized dimensionless flow (dimensionless flow divided by maximum dimensionless flow, (φ(xfan) / φmax) as a function of the log base 10 of the normalized Euler (Eu) parameter (xfan). Figure shows an example plot derived from the manufacturer’s data in Figure.
Example Normalized Dimensionless Flow Data for One Fan
Equation describes the sigmoidal functional form for normalized φ:
φ(xfan)φmax=(Aspd+Bspd/((1+e[(Cspd−xspd)/(Cspd−xspd)DspdDspd])Espd) ()
where Aspd, Bspd, Cspd, Dspd, and Espd are coefficients for the fan. Separate sets of coefficients for the normal operation and stall regions can be specified. For the curve shown in Figure, these coefficients are:
For a generic centrifugal fan with backwardcurved blades, the coefficients are:
For any operating point of flow (Qfan) and fan pressure rise (ΔPfan), one can calculate the corresponding Euler number (Eu) for the given fan, and then determine the corresponding normalized dimensionless flow (φ(xfan) / φmax) from the sigmoidal dimensionless flow function (Equation ). With φ(xfan) / φmax determined and knowing φmax for the fan, the dimensionless flow is:
φ(xfan)=φmax[φ(xfan)φmax] ()
With φ(xfan) determined, the fan rotational speed (rad/s) is:
ωfan=Qfan(φ(xfan)∗D3fan) ()
Fan shaft torque (N∙m), which the belt drive must supply to the fan shaft, is then:
τfan=Hfanωfan ()
The fraction of fullload driven torque for the belt (τfan / τbelt,max), which is typically called “belt load”, is thus τfan divided by the belt torque capacity (τbelt,max). For a particular belt type and crosssection, belt torque capacity can be determined from manufacturer’s information such as a shaft speed versus power chart.
Ignoring belt slip, motor shaft speed (ωmotor, rad/s) can then be determined using the fan speed (ωfan) and the motor/fan pulley diameter ratio (Dpulley,motor / Dpulley,fan), which is typically called the “drive” ratio:
ωmotor=ωfan(Dpulley,motorDpulley,fan) ()
Belt, Motor, and VariableFrequencyDrive Efficiency and Input Power Models – Overview: The models for belt drives, motors, and VFDs that Stein and Hydeman included in their airhandling system model represent maximum efficiency as a function of power input to the adjacent downstream component (e.g., the power input to a fan shaft by a belt drive), but they do not include partload models. The partload models are needed, because partload efficiency for these components can fall off rapidly at low load (e.g., to zero at zero load) and many systems operate occasionally (and some much of the time) at low loads, in part because of current practices that result in substantial oversizing of components.
Belt Efficiency and Input Power Model: Figure shows three maximum efficiency (ηbelt,max) curves for belts (low, medium, and high) as a function of maximum fan shaft torque. This set of efficiency curves is based on belt drive loss data from AMCA Publication 203 (1990b), which reportedly is an aggregation of data from over 400 tests.
To determine ηbelt,max if data for a specific belt are not available, first use the maximum fan shaft input power (Hfan,max) for the load spectrum to calculate the natural logarithm of belt power capacity:
xbelt,max =ln(Hfan,max) with Hfan,max expressed in terms of hp ()
Then, use Equations and , along with a choice of low, medium, or high efficiency coefficients from Table, to calculate ηbelt,max.
Belt Maximum Efficiency vs. Fan Shaft Power Input
The quartic polynomial curves in Figure and their coefficients are as follows:
ηbelt,max,ln=c1+c2⋅xbelt,max+c3⋅x2belt,max+c4⋅x3belt,max+c5⋅x4belt,max ()
The belt maximum efficiency is therefore:
ηbelt,max=eηbelt,max,ln ()
The normalized partload belt efficiency model is provided to modify the maximum efficiency of the belt. This model is based on partload data for belts in ACEEE’s handbook on motor systems (Nadel et al. 2002). To determine the normalized belt efficiency at partload (ηbelt(xbelt)/* ηbelt,max), use the fan shaft (belt output) fractional torque (xbelt* = τbelt / τbelt,max; τbelt = τfan) as the belt fractional load in Equation or (select the equation for the region that contains xbelt) and use coefficients from Table that correspond to that region and the belt type: VBelt or synchronous (toothed). Figure shows a graphical representation of the curves defined by these equations and coefficients.
Belt Normalized Efficiency vs. Belt Fractional Load)
The belt normalized (partload) efficiency curves in Figure for Regions 1 and 3 (single rectangular hyperbola type 2) and Region 2 (exponential decay), respectively, and their coefficients are as follows:
Region 1 (0 < = xbelt < xbelt,trans) and Region 3 (xbelt > 1):
ηbelt(xbelt)ηbelt,max=[abelt⋅xbeltbbelt+xbelt]+cbelt⋅xbelt ()
Region 2 (xbelt,trans < = xbelt < = 1):
ηbelt(xbelt)ηbelt,max=abelt+bbelt⋅e(cbelt⋅xbelt) ()
where xbelt = τ belt / τ belt,max
Belt efficiency therefore is:
ηbelt(xbelt)=ηbelt,max[ηbelt(xbelt)ηbelt,max] ()
Belt input power (mechanical, W) at the motor shaft is:
Hbelt=Hfanηbelt(xbelt) ()
Motor Efficiency and Input Power Model: Figure shows three maximum efficiency (ηmotor,max) curves for motors (low, medium, and high) as a function of rated motor output power (belt input power). This set of efficiency curves is based on maximum efficiency data in DOE’s MotorMaster+ database (2003) from about 800 tests. It is provided for use in the absence of manufacturer’s data.
Maximum Motor Efficiency vs. Belt Power Input (Motor Output)
To determine ηmotor,max if data for a specific motor are not available (e.g., as listed in Table), first use the maximum belt input power (Hbelt,max) for the load spectrum (multiplied by whatever oversizing factor may be desired) to calculate the natural logarithm of belt power capacity:
xmotor,max =ln(Hbelt,max) with Hbelt,max expressed in terms of hp ()
Then, use Equation , along with a choice of low, medium, or high efficiency coefficients from Table, to calculate ηmotor,max.
The maximum motor efficiency curves in Figure (single rectangular hyperbola type 1) and their coefficients are as follows:
ηmotor,max=[amotor,max⋅xmotor,maxbmotor,max+xmotor,max]+cmotor,max ()
The normalized partload motor efficiency model is provided to modify the maximum efficiency of the motor. This model is based on partload data for motors from DOE’s MotorMaster+ database (2003). To determine the normalized motor efficiency at partload (ηmotor(xmotor)/* ηmotor,max), use the motor fractional power output (xmotor = Hbelt* / Hbelt,max) as the fraction of motor output power in Equation and use coefficients from Table, or for a specific motor, determined from DOE MotorMaster+ data or from manufacturer’s data. Figure shows a graphical representation of eight example curves defined by these equations and coefficients.
Motor Normalized Efficiency vs. Motor Load Fraction
The example motor normalized efficiency curves (single rectangular hyperbola type 2) and their coefficients as a function of motor load fraction in Figure are determined from DOE MotorMaster+ data and are as follows:
ηmotor(xmotor)ηmotor,max=[aPLmotor⋅xmotorbPLmotor+xmotor]+cPLmotor⋅xmotor ()
Motor efficiency therefore is:
ηmotor(xmotor)=ηmotor,max[ηmotor(xmotor)ηmotor,max] ()
Motor input power (electrical, W) is:
Hmotor=Hbeltηmotor(xmotor) ()
VFD Efficiency and Input Power Model: For VFDs, published performance data are limited. Data from DOE (2008) suggest using a functional relation similar to that used for motors to represent VFD efficiency (ηVFD) as a function of the fraction of fullload motor input power (xVFD = Hmotor / Hmotor,max), as shown in Figure, or as a function of the fraction of fullload speed (xVFD = ωmotor / ωmotor,max), as shown in Figure.
VFD Efficiency vs. Fraction of Motor FullLoad Input Power (*Source: DOE 2008*)
VFD Efficiency vs. Fraction of FullLoad Motor Speed (*Courtesy of Saftronics*)
The performance curve (single rectangular hyperbola type 2) used here for VFD efficiency is:
ηVFD=[aVFD⋅xVFDbVFD+xVFD]+cVFD⋅xVFD ()
Example coefficients derived from the DOE data as a function of the fraction of fullload motor input power are listed in Table.
VFD input power (electrical, W) is:
HVFD=HmotorηVFD ()
System Total Efficiency: The combined efficiency of the fan system components (i.e., fan, belt, motor, and VFD) is:
ηTOTAL=ηfan⋅ηbelt⋅ηmotor⋅ηVFD (=Qfan⋅ΔPfanHVFD) ()
Heat Loss to Air: To calculate the temperature rise from waste heat entering the air stream from the fan, belt, and motor, it is assumed that the userspecified “motor in air fraction” applies to the belt and motor but not to the VFD. The power “lost” to the air (W) is:
˙Qtoair=Hfan+(Hmotor−Hfan)⋅fmotortoair ()
References[LINK]
AMCA. 1990a. “Fans and Systems”. Publication 20190. Arlington Heights, IL: Air Movement and Control Association International.
AMCA. 1990b. “Field Performance Measurement of Fan Systems”. Publication 20390. Arlington Heights, IL: Air Movement and Control Association International.
ASHRAE. 1993. HVAC 2 Toolkit: Algorithms and Subroutines for Secondary HVAC System Energy Calculations. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
ASHRAE. 1999. “ANSI/ASHRAE Standard 120 Method of Testing to Determine Flow Resistance of HVAC Ducts and Fittings”. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
ASHRAE. 2008. “ANSI/ASHRAE Standard 126 Method of Testing HVAC Air Ducts and Fittings”. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
DOE. 2003. “MotorMaster 4.0 User Guide”. U.S. Department of Energy, Industrial Technologies Program.
DOE. 2008. “Energy Tips – Motor: Motor Tip Sheet #11”. U.S. Department of Energy, Industrial Technologies Program. June.
Federspiel, C. 2004. “Detecting Optimal Fan Pressure”. Final Report of Federspiel Controls to the CEC Energy Innovations Small Grant Program. Grant #: 0203.
Federspiel, C. 2005. “Detecting Critical Supply Duct Pressure”. ASHRAE Transactions, Vol. 111, Part 1. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
Heredos, Francis P. 1987. Selection and Application of Multispeed Motors, IEEE Transactions on Industry Applications, Vol. 1A23, No. 2, March/April.
Liu, M., D.E. Claridge, and S. Deng. 2003. “An Air Filter Pressure Loss Model for Fan Energy Calculation in AirHandling Units”. International Journal of Energy Research. Vol. 27, pp. 589600.
Nadel, S., R.N. Elliot, M. Shepard, S. Greenberg, G. Katz, and A.T. de Almeida. 2002. “EnergyEfficient Motor Systems: A Handbook on Technology, Program, and Policy Opportunities” 2nd Edition. Washington, DC: American Council for an Energy Efficient Economy. p.188.
Sherman, M.H. and C.P. Wray. 2010. “Parametric System Curves: Correlations Between Fan Pressure Rise and Flow for Large Commercial Buildings”. Lawrence Berkeley National Laboratory Report, LBNL3542E.
Stein, J. and M.M. Hydeman. 2004. “Development and Testing of the Characteristic Curve Fan Model”. ASHRAE Transactions, Vol. 110, Part 1. Atlanta: American Society of Heating, Refrigerating and AirConditioning Engineers, Inc.
Trane. 1999. “DeltaFlo Coils: Data Catalog PLAHCOIL000D2799”. LaCrosse, WI: The Trane Company. July.
Documentation content copyright © 19962016 The Board of Trustees of the University of Illinois and the Regents of the University of California through the Ernest Orlando Lawrence Berkeley National Laboratory. All rights reserved. EnergyPlus is a trademark of the US Department of Energy.
This documentation is made available under the EnergyPlus Open Source License v1.0.