Printer Friendly

Low evaporator airflow detection using fan power for rooftop units.


Proper airflow is an important factor to ensure the continuous and healthy operation of a rooftop unit (RTU). However, low airflow is a common fault in most units (Breuker and Braun 1998). Low airflow can be caused by many reasons, such as a dirty air filter, evaporator coil fouling, ductwork blockage, or loose fan belt. For a rooftop unit equipped with a constant-speed fan, a rule of thumb is that measured airflow lower than 300 cfm/ton (0.04 [m.sup.3]/s/kW) is considered to be a low airflow fault (Cowan 2004). Reduced fan airflow could result in evaporator coil freezing and system trip-off at high burner temperatures. More importantly, it will cause thermal discomfort and increase energy consumption.

Low airflow is a very common issue but is usually not paid attention to until comfort issues occur. Plenty of studies show that the effect of air-side fouling of a heat exchanger on pressure drop is more pronounced than on heat transfer. Detection of dirty filters and evaporator coil fouling can be converted to help with detection of reduced airflow (Lankinen et al. 2003; Bell et al. 2009). There are several ways to detect low airflow, such as pressure or temperature detection. With pressure-based detection methods, an air pressure switch is used to measure the differential pressure across the filter or cooling coil. When the filter or coil is dirty, the blocked airflow results in a higher pressure drop than the normal value. An alarm can be generated to indicate a blocked filter or coil. This method works better under constant-fan-speed operations, especially for large-size air-handling units or rooftop units. However, for small-sized RTUs or units equipped with a variable-speed fan, the pressure switch may not sense the pressure change under low-speed conditions. For rooftop units without a filter pressure switch, the common way to maintain their functions is routine inspection and replacing with a new filter regularly. Besides the pressure-based method, the temperature-based method (Li and Braun 2007; Wichman and Braun 2009) is the most-used method to detect the evaporator airflow restriction. Several temperatures are measured, including evaporator inlet air temperature and condenser inlet air temperature. The evaporator airflow is derived based on these temperature measurements and system models.

Researchers from Massachusetts Institute of Technology (Armstrong et al. 2006) used electrical measurements such as real power and reactive power to detect the reduced airflow fault caused by blockage. The transient power value is collected by a nonintrusive load monitoring (NILM) device. The idea of electrical measurement is pretty valuable, as the electrical signal measurements are more reliable compared with temperature or pressure measurements. However, the NILM device is required, which may not realistic for all RTUs.

As variable-frequency drives (VFDs) have been installed on supply fans more and more widely, they can be treated as an electrical meter to measure the instantaneous fan parameters, such as frequency or power. These readings can be transmitted to a stand-alone controller or building automation system (BAS), which is helpful to further analyze the performance of units and detect the reduction of airflow.


At a fixed speed, the fan power is determined by airflow and fan head. When a VFD is installed for a supply fan, the fan speed should be always considered. Centrifugal fans are commonly used in rooftop units. There are several blade shapes: airfoil, backward inclined or backward curved, radial, and forward curved. For radial and forward-curved fans, the fan power increases when the airflow rate increases. However, for airfoil and backward-inclined or backward-curved fans, there is a point where the fan power is at the maximum value. If the airflow rate is higher than this point, the fan power decreases. Figure 1 shows the fan performance curve for a generic fan. This turning point is on the right of the maximum fan-power point. This means that as long as the airflow varies in the selection range, the fan power still increases as the airflow increases.

Figure 1 shows the fan performance curve from one manufacturer. Assuming the design airflow is 8000 cfm (3.8 [m.sup.3]/s), the fan power will be around 3.2 hp (2.4 kW). If the fan speed is fixed and the airflow is reduced to 6000 cfm (2.8 [m.sup.3]/s) (which is equivalent to a 25% reduction), the fan power will be 2.8 hp (2.1 kW), which is equivalent to a 12.5% reduction. If the airflow is reduced to 4000 cfm (1.9 [m.sup.3]/s), which is equivalent to 50% reduction, the fan power will be around 2.4 hp (1.8 kW), which is equivalent to 25% reduction. This example clearly illustrates that the fan power can reflect the change of airflow.

In a single-zone variable-air-volume (VAV) system, the supply fan is usually controlled by the VFD to maintain the space air temperature. The airflow reduction caused by a dirty filter or a dirty evaporator can be identified by comparing the actual fan power and fault-free fan power.

Figure 2 shows the change of working point and power comparison, where c1 and c2 are system resistance curves under normal and fault conditions (dirty filter or evaporator); [P.sub.[omega]1] and [P.sub.[omega]2] are the fan head curves of a forward-curved fan, under speed [omega]1 and [omega]2; [W.sub.[omega]1] and [W.sub.[omega]2] are the fan-power curves under speed [omega]1 and [omega]2. At normal conditions--that is, when both the filter and evaporator are clean--the fan working point is 1 with speed [omega]1. If the filter or evaporator is dirty, then the pressure drop across the filter and evaporator will be higher, and also the system resistance will increase. The system curve will be changed from c1 to c2. The working point becomes 2. It can be seen clearly that the airflow drops from [Q.sub.1] to [Q.sub.2], and the fan power is reduced from [W.sub.1] to [W.sub.2].

The reduced airflow causes the reduction of cooling capacity, which results in the increase of space air temperature. As the fan speed and compressor speed are adjusted to control the space temperature, the thermostat calls for more cooling and, consequently, the VFD speed goes up. Assuming that at speed [omega]2 the airflow increases to [Q.sub.1] and the system can provide the required cooling capacity, the new working point will be 3 and the power curve is changed to [W.sub.[omega]2]. It can be seen that the fan power under the new working condition will be higher than the initial fan power. The increment is as follows:

[DELTA] W = [W.sub.3] - [W.sub.1] (1)

However, the fan power at Point 3 is still lower than at Point 4, which is at the normal system curve. When the fan speed increases to compensate for the lost cooling capacity, the fan power at the new working point is always lower than that at the original curve cl. Therefore, this relationship could be used to detect the low-airflow fault.


It has already been demonstrated that the change of fan airflow is reflected by the variation of fan power. Therefore, low airflow can be indicated by the reduction of fan power. Usually, the fan power is not directly measured. The fan motor power measurement is an approximation of fan power with a difference of belt power loss and motor power loss. The motor input power can be measured by a power meter installed for the fan motor. However, the motor input power often cannot be trended automatically by the third-party controller or BAS.

With the application of VFD, it is feasible to use a VFD to measure the fan power and speed. In recent years, VFDs have been more and more widely applied in the HVAC industry because of the needs of building energy efficiency and the fast development of VFD technology. Most VFDs have the ability to measure either the current or power input of a motor. The measured motor power can be used to estimate the fan power, while the error is the power loss of a motor. The motor operating power could be sent out from VFDs through either analog signal or digital communication (e.g., Modbus, BACnet[R]). In this study, Modbus communication is directly used to receive the fan motor power data.


First, the baseline must be developed. Then, the actual fan power can be measured by VFD to compare with the baseline value. The difference of actual and base values will indicate the magnitude of airflow reduction.

Baseline Development

The rooftop unit can run at cooling mode, heating mode, or fan mode (ventilation mode). The VFD can be installed for fan only or both fan and compressor. In this paper, we focus on the second scenario, that is, a VFD is installed on both fan and compressor. The baseline model is described as follows.

Heating Mode and Free-Cooling Mode (Outdoor Air Temperature [OAT] <55[degrees]F [12.8[degrees]C]). In both modes, only the fan is running and the compressor is off. The fan-power-speed curve can be built using the measured data (Li 2012):

[W.sub.f,n] = [a.sub.0] [[bar.[omega]].sup.2] + [a.sub.1] [bar.[omega]] + [a.sub.2] (2)

The experimental results reveal that the difference in the fan power between the normal and faulty conditions will be larger at higher speeds than at lower speeds. Therefore, the full-speed fan power [W.sub.f,n,60] is used as the baseline:

[W.sub.f,n,60] = [a.sub.0] [[bar.[omega]].sup.2] + [a.sub.1] [bar.[omega]] + [a.sub.2][|.sub.[bar.[omega]]=1] = [a.sub.0] + [a.sub.1] + [a.sub.2] (3)

Fan Mode. In fan mode, both compressors and heaters are off and only the fan is running at the minimum speed. The fan power at minimum speed is used as the baseline.

Actual Fan-Power Performance

In heating and free-cooling mode the VFD reading will be the fan-only power. The full-speed fan power can be determined by the following:

[W.sub.f,n,60] = [W.sub.f,a,[omega]] x [[eta].sub.m,p[omega]]/[[bar.[omega]].sup.3][[eta].sub.m,60] (4)

where [[eta].sub.m,[omega]] and [[eta].sub.m,60] are motor efficiency under speed ra and 60 Hz. These values can be determined based on the motor speed ratio and load ratio (Li et al. 2015).

In fan mode, the VFD power reading is the fan power at minimum speed:

[W.sub.f,a,min] = [W.sub.VFD] (5)

Low Evaporator Airflow Detection

The low-airflow fault will be detected by comparing the actual fan power and the baseline. The control limit method is used to evaluate the fan power.

The ratio of actual power to its normal value is used:

r = [x.sub.a]/[x.sub.n] (6)

The actual value is [x.sub.a], and its normal value is [x.sub.n]; x is not limited to fan power--it could be speed, total power, or any other parameter. Total power is the sum of fan power and compressor power.

The upper and lower control limits are established using the baseline data:

[r.sub.ul] = max ([x.sub.a,i]/[x.sub.n,i]) (7)

[r.sub.ll] = max ([x.sub.a,i]/[x.sub.n,i]) (8)

At normal conditions, the upper and lower control limits can be obtained from the baseline data:

[,n,ul] = max ([W.sub.f,meas,i]/[W.sub.f,pred,i]) (9)

[,n,ll] = min ([W.sub.f,meas,i]/[W.sub.f,pred,i]) (10)

During actual operations, the fan-power ratio is as follows:

[,a,i] = [W.sub.f,a,i]/[W.sub.f,n,i] (11)

where [W.sub.f,meas,i] and [W.sub.f,pred,i] are the measured and predicted full-speed fan power, respectively. [W.sub.f,a,i] and [W.sub.f,n,i], are the actual and normal full-speed fan-power ratios. Similarly, the upper and lower limits of total power ratio could be calculated as well:

[r.sub.wt,n,ul] = max ([W.sub.t,meas,i]/[W.sub.t,pred,i]) (12)

[r.sub.wt,n,ll] = min ([W.sub.t,meas,i]/[W.sub.t,pred,i]) (13)

[r.sub.wt,a,i] = [W.sub.t,a,i]/[W.sub.t,n,i] (14)

If the ratio is varied between the upper and lower limits, no "change" is detected. Otherwise, the fan-power change will be detected.


A field test was conducted to demonstrate the proposed method. The rooftop unit was designed as a constant-volume system originally and retrofitted to a VAV system before the field testing. A VFD was installed on this unit with communication with a third-party controller to control both fan and compressor speeds. The system performance data were obtained from an energy management and control system (EMCS), which gets the power reading from the VFD through Modbus communication. When the compressor is running, the VFD power is the sum of fan and compressor power. When the compressor is off, the VFD power reading is the fan power. One VFD was used during this testing. The accuracy of frequency and power measurements are 0.1 Hz and 0.1 kW, respectively.

The low-airflow fault is simulated by blocking the evaporator coil with papers.

Baseline Development. When the room temperature reached its setpoint and the thermostat was set at the "fan on" mode, the compressor was off and only the fan was running, at minimum speed. The minimum speed of the tested unit was 45 Hz. The average fan power at minimum speed was 0.62 kW, as shown in Figure 3. Figure 4 shows the distribution of total power ratio during the baseline period. Of total data, 98.7% are located within [+ or -] 6% of the predicted value. The minimum value of total power ratio in this case is 0.94. Therefore, [r.sub.wt,n,ll] = 0.94.

During this test, the dirty evaporator coil was simulated by blocking the coil with papers which were added every 10 minutes. The minimum fan speed was 45 Hz. Because the test was carried out in late July and early August, the supply fan ran with the compressor together in the cooling mode or ran itself in ventilation mode. There is no free-cooling mode available. Therefore, the minimum-speed fan power was used to develop the fan-power baseline.

Figure 5 shows the fan-power variation during the test. The x-axis represents the time, and the y-axis represents the actual fan power. The test started with one paper blocking the coil, which was approximately equivalent to 6% coil blockage, at 8:05 a.m. More papers were used to block the coil as the time increased. The blockage percentages are marked in Figure 5. When the evaporator is clean or no coil was blocked, the fan power was about 0.62 kW. It is clearly shown that the fan power decreased with more coil surface blocked. At around 9:50 p.m., around 84% of the evaporator coil was blocked. The fan power was 0.56 kW at this time. The corresponding airflow was 633 cfm (0.3 [m.sup.3]/s), which was 41% of the normal value.

Figure 6 shows the total power ratio variation during the test. The unit was running at cooling mode during this period of time. Therefore, both fan and compressor were running. This test started with no paper blocking the coil at 9:00 a.m., and more papers were added later. The blockage percentage was also marked in Figure 6. At around 2:00 p.m., the total power ratio was 0.9, which was less than the low limit of the baseline, which was 0.94. The blockage percentages are about 36% where the supply fan power was 0.58, which was lower than the baseline limit of 0.94. Because both fan power and total power ratio requirements are satisfied, the low-airflow problem is detected.

Figures 5 and 6 present the performance of fan power and total power reduction during the test. Both of them have a reduction tendency. Table 1 lists the baseline and actual measurement for fan power and total power ratio. According to Equations 12, 13, and 14, the low-airflow fault was detected by the change of fan power.


The principle was introduced on how to use fan-power reduction to detect airflow reduction for a rooftop unit. The detection algorithm was described in detail. Then, a field test was performed to demonstrate the proposed algorithm. The results show that the airflow reduction was able to be detected only by comparing the actual fan power with its baseline. This method provides a new direction compared with the existing method for low-airflow detection using fan power from the VFD. The results of testing indicate that the fan power might be quite small for small-sized RTUs. Therefore, the VFD reading needs to be calibrated to ensure the accuracy is consistent for both baseline and actual measurements. Further field testing is also necessary to determine whether this method is feasible for a backward-curved or airfoil fan.


[eta] = efficiency

[omega] = speed

[bar.[omega]] = speed ratio

[a.sub.0], [a.sub.1], [a.sub.2] = coefficient

c = curve

P = pressure

Q = airflow

r = ratio

W = fan power

x = performance parameters


1,2 ... = different work conditions

a = actual

f = fan

f, min = fan minimum

i = individual measurement point

ll = lower limit

m = motor

min = minimum

meas = measured value

n = normal

pred = predicted value

t = total

tf = fan total

ul = upper limit

wf = fan power

wt = total power


ASHRAE. 2008. ASHRAE Handbook--HVAC systems and equipment. Atlanta: ASHRAE.

Armstrong, P.R., C.R. Laughman, S.B. Leeb, and L.K. Norford. 2006. Detection of rooftop cooling units faults based on electrical measurements. Science and Technology for the Built Environment 12(1):151-75.

Bell, I.H., E.A. Groll, H. Konig, and T. Odrich. 2009. Experimental analysis of the effects of particulate fouling on heat exchanger heat transfer and air side pressure drop for a hybrid dry cooler. Proceeding of International Conference on Heat Exchanger Fouling and Cleaning VIII, 175-81.

Breuker, M.S., and J.E. Braun. 1998. Common faults and their impacts for rooftop air conditioners. Science and Technology for the Built Environment 4(3):303-18.

Cowan, A. 2004. Review of recent commercial roof top unit field studies in the Pacific Northwest and California. Portland, Oregon: New Buildings Institute. https://newbuild _Report_R3_.pdf.

Lankinen, R., J. Suihkonen, and P. Sarkomaa. 2003. The effect of air side fouling on thermal hydraulic characteristics of a compact heat exchanger. International Journal of Energy Research 27(4):349-61.

Li, H., and J.E. Braun. 2007. A methodology for diagnosing multiple simultaneous faults in vapor-compression air conditioner. Science and Technology for the Built Environment 13(2):369-95.

Li, Y. 2012. Electrical signal based fault detection and diagnosis for rooftop units. ETD collection for University of Nebraska-Lincoln. AAI3546875.

Li, Y., M. Liu, J. Lau, and B. Zhang. 2015. A novel method to determine the motor efficiency under variable speed operations and partial load conditions. Applied Energy 144:234-40.

Wichman, A., and J.E. Braun. 2009. Fault detection and diagnosis for commercial coolers and freezers. Science and Technology for the Built Environment 15(1):77-99.


Steve Liescheidt, Engineer, SPPECSS Consulting LLC, St. Louis, MO: How was the building "power factor" taken into account in the building of the text?

Yunhua Li: We measured actual power, not amps; therefore, this was not a factor. The power is directly measured by VFD. Gang Wang, Assistant Professor, University of Miami, Coral Gables, FL: 1) How did you convert the VFD power ratio to the fan power ratio? 2) How did you consider the motor efficiency?

Li: 1) When both the fan and the compressor are running, the VFD power is the sum of fan power and compressor power. When the compressor is off, the VFD power is the fan power only. There is no need to convert the VFD total power ratio to the fan power ratio because the baseline at each mode are different. The fan power ratio is used in heating mode and free cooling mode. The total power ratio is used in mechanical cooling mode.

2) In fan mode, the motor efficiency is assumed to be the same at the minimum fan speed operation. So, we use VFD output power to estimate the fan power. At other modes, the motor efficiency at different speeds is calculated using the equation provided (by reference shown in context).

Yunhua Li, PhD

Associate Member ASHRAE

Josephine Lau, PhD


Bei Zhang, PhD

Associate Member ASHRAE

Mingsheng Liu, PhD, PE


Yunhua Li is a mechanical product engineer, Bei Zhang is a project engineer, and Mingsheng Liu is the president and CTO at Bes-Tech Inc., Omaha, NE. Josephine Lau is an associate professor in the Department of Architectural Engineering at the University of Nebraska-Lincoln, Omaha, NE.

Table 1. Fan-Power Comparison, Low-Airflow Reduction

Parameter       [W.sub.f,min]             [r.sub.wt]

Baseline    0.62 ([W.sub.f,min,n])   0.94 ([r.sub.wt,n,ll])
Actual      0.58 ([W.sub.f,min,a])    0.9 ([r.sub.wt, a])
COPYRIGHT 2016 American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc. (ASHRAE)
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Li, Yunhua; Lau, Josephine; Zhang, Bei; Liu, Mingsheng
Publication:ASHRAE Transactions
Article Type:Report
Date:Jul 1, 2016
Previous Article:Modeling airflow through a perforated duct.
Next Article:Characterizing the performance of fixed-airflow series fan-powered terminal units using a mass and energy balance approach.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters