Printer Friendly

Dedicated Outside Air System (DOAS)--Design vs. Actual Operation (Are Aggressive Energy Targets Achievable?).


The DOAS installed at the UMass/Amherst Police Station was an integral part of the HVAC design to obtain the energy performance projected for its LEED (Leadership in Energy and Environmental Design) Gold rating for this 27,500 gsf (2556 [gm.sup.2]) building. The LEED measurement and verification task is being conducted in house with Physical Plant and Mechanical Engineering Department staff and students. Electrical submetering installed during building construction afforded us the opportunity to study in some detail the system performance.

When compared against the EQuest energy model used for its LEED certification the building in its first year of operation used 1.7 times the predicted electrical energy, after correcting for actual observed plug loads. Natural gas use, for domestic hot water heating only, was minimal and conformed to the model predictions. The electrical submetering data pointed to the HVAC system as being the furthest from it predicted performance, and the DOAS was easily seen to be exceeding its modeled airflow and fan horsepower. For example, it ran at an average of 8000 cfrn (3776 l/s) fairly continuously, generally at 100% fan speed. No more than 15% turndown from minimum to maximum values was observed, even though the great majority of spaces were unoccupied at night. The energy model, assuming effective demand controlled ventilation (DCV), assumed an average airflow of only 4900 cfm (2313 l/s). Fan power consumption was also higher than modeled, running most often at 13.7 kW, 1.7 W/cfm (3.6 W/(l/s)), while the model assumed 4.2 kW at 0.849 W/cfm (1.8 W/(l/s)). The additional dynamic losses in the rooftop unit air handler (RTU) and distribution system at the higher flowrates were a large factor in causing this added power consumption. These were surveyed and reported.

An assessment of system air leakage was conducted using Building Automation System (BAS) reported flow data from variable air volume (VAV) boxes added to the design flows from terminals with open two position zone dampers and compared to RTU flowstation readings. This suggested that 45% of the RTU flow was unaccounted for with the system running at 1.0 in. wc (0.25 kPa) static pressure. The supply ducts had been sealed and pressure tested during construction so evidence pointed to the control dampers as a major source of the lost air.

Not ready to throw out the energy model yet we enlisted the sheet metal contractor to better seal the control dampers, enlisted the air balancer for survey work, and we worked with the designer to look at the occupancy assumptions behind the design ventilation levels. It seemed these were adequately conservative that reductions were possible, so the operating static pressure of the RTU was dropped from 1.0 in. to 0.3 in. wc (0.25 to .075 kPa). Results are reported below.

The operation of the zone demand control ventilation sequences was examined and results are summarized below.

Finally, the heating and cooling operation of the RTU was examined in light of current thinking on DOAS operation (Mumma, 2011; Murphy, 2012). Its initial leaving air temperature control sequence created both reheating loads in summer and recooling loads in winter.

The investigation of this DOAS is reported in four categories which seemed to be responsible for the observed energy performance shortfall,

1. The efficacy of the various demand controlled ventilation (DCV) strategies,

2. The physical control of air in the DOAS,

3. The air horsepower required to move the air and the sources of pressure loss.

4. The temperature and humidity control operating parameters of the rooftop unit.


The system uses an interesting mix of demand control ventilation schemes in the zones, including occupancy sensors, humidity readings, C[O.sub.2] readings and volatile organic compound (VOC) readings, with varying degrees of effectiveness. The air quality measurements are done with a central analyzing unit which samples from the various zones. This use of a single sensor bank makes possible very accurate comparison of indoor readings to an outdoor air reference, or to evaluate latent loads with a rise in dewpoint for example.

The prime mover of the DOAS is a rooftop water source heat pump unit (RTU) running 100% outside air with an enthalpy wheel for heat recovery. With supply and exhaust fan VFD's, and dual compressors with one capacity modulated with pulse width unloading "digital scroll" controls, it is well suited to a variable flow DOAS.

The design airflow for the DOAS is 7750 cfm (3658 l/s), and the sum of the terminal design flows was 9870 cfm (4659 l/s). It supplies air typically to the return duct of the individual room water source heat pumps which are above the dropped ceiling. In a few of the high ventilation spaces (locker rooms and cell block) the DOAS air is ducted straight to the local heat pump without any room return, thereby creating a series fan situation. These high ventilation spaces utilize VAV boxes for supply and exhaust air control. The majority of spaces use a two position rectangular control damper in the DOAS supply duct only. (It appears VAV box control for all terminals was value engineered out in favor of these two position dampers.)

The rooftop unit and all zones are controlled and monitored by the building automation system (BAS). Figure 1 shows a BAS graphic of the RTU, and Table 1 shows typical operating air conditions in cooling mode. Figures 2 and 3 show the mixed air terminal heat pump units and the 100% outside air (series fan) terminal units respectively. Although the series fan arrangement may be somewhat unusual it seems to have merit in reducing the overall system static pressure required to meet the need of the most demanding far terminals.

The distribution systems are conventionally designed using friction rates less than 0.1"/100 ft. (763 Pa/100 m). in all cases, however no continuous vertical chases were available so significant horizontal offsets were needed. As with many VAV systems relying on diversity to meet instantaneous requirements, air balancing was challenging. It was carried out by shutting off some of the larger loads anticipated to have intermittent demand and setting manual dampers ahead of the remaining two position control dampers for their design airflow. Control of DOAS air to these fixed damper served spaces is necessarily a bit inaccurate, depending on varying duct pressure.

Electrical submetering is provided at each local panel, where panels are dedicated to lighting, plug load, or mechanical load. The RTU, elevator, pumps, and electric heat were not submetered since they did not fall within the specification requirement of metering all panels. Designed connection of electrical submetering to BAS may be completed someday.


Occupancy Sensor Control

Certainly the simplest and arguably the most effective in this first year's operation, this relies on a room occupancy sensor common to the lighting controls to open the DOAS supply damper.

C[O.sub.2] Sensor control

Using the central air quality monitoring system this will open the two position DOAS terminal control damper at a C[O.sub.2] setpoint of 800 ppm. These DOAS dampers rarely opened. Some combination of damper leakage, or open doorways to adjacent spaces likely contributed to this result.

Humidity control

A humidity based control sequence was attempted for the locker rooms, with the intent of ventilating to the fixed ASHRAE 62.1 requirements but adding more supply and exhaust upon a rise in humidity, with a goal of drying the area after shower use. The initial setpoint however was 50% rh, and in summer the ventilation air from the DOAS was not dry enough to achieve that level. Consequently the extra ventilation ran most of the summer to no particular avail. Also, the wall mounted sensing point for humidity is right inside the entrance door, little if at all affected by shower generated humidity.

VOC control

The installed air quality sensing system has the ability to measure volatile organic compounds at any of the sensing locations with resolution down to fractional ppm levels. This was used in the evidence room, with the design intent of clearing the air following dumping of alcoholic evidence down the drain. It activated a half dozen times during the year, with a setpoint of 0.5 ppm which corresponds to LEED air quality requirements. It also activated for a longer stretch when the VOC sensor zero calibration point drifted up past 0.5 ppm . Basing this sequence on a rise above ambient might be more helpful, or alarming if outdoor ambient begins reading high. This sequence was also used in the entrance lobby to add a high percentage of outside air there upon a particularly odiferous situation. It never triggered.

Wall switch control

For the cell block area the design team settled on a wall switch to be part of the unoccupied control which would shut down the supply air and the local heat pump. The "heating/cooling" designation on the switch seems to be confusing, (certainly it is to me), and it toggles back and forth somewhat randomly but predominantly occupied.


The control sequences described above operate either two position control dampers or VAV boxes. Consequently the system is a mix of pressure independent VAV boxes and pressure dependant fixed opening dampers. While the ductwork passed its pressure test and the VAV boxes are expected to achieve good air control even down to shutoff, it became apparent that the two position dampers were not performing nearly as well.

These dampers were described as class 2 in submittals, which would correspond to a leakage rating of 10 cfm/sf (51 l/s/[m.sup.2]) at 1.0 in wc (.25 kPa) static pressure, perhaps a bit high, but damper frames were typically screwed into the duct walls on two sides and the frame was unsealed. The dampers were sized for flow velocity, and the resulting size of the dampers was relatively large and this may have added to leakage volumes. When the majority of these dampers were sealed, during operation at 0.3 in. wc (.075 kPa) static pressure, airflow was seen to reduce from 5400 cfm (2549 l/s) average to 4200 cfm (1982 l/s) average. Still, a comparison of RTU flows vs. measured and design flows for the terminals suggested a 30% unaccounted for airflow, but at least the absolute value of airflow is lower. After giving some thought to this issue and comparing damper ratings and common applications, it seems perhaps using smaller round dampers, as common in VAV boxes would have alleviated the installation issue and delivered better leakage rates,


Our air balancer was brought back in to help evaluate the effects of different operating strategies and the following table reports the static pressure surveys taken at the RTU when operating at the initial 1.0 in. wc (0.25 kPa) wc and then at 0.3 in. wc (.075 kPa).. Controlling static pressure is measured part way out the middle (first) floor supply duct, in keeping with the 2/3 tradition. It was interesting to note the duct static pressure going negative at lower airflow delivery rates since the heat pump fans in series continue to pull their constant volume out of the supply duct. The 0.3 in. wc (.075 kPa) setting seemed to assure positive supply duct pressure at all locations.

It became clear that the largest single source of pressure drop in the RTU was the heat recovery wheel. It was seen to be helpful to utilize the installed bypass dampers (used for frost control) to reduce wheel pressure drop when it is not in service. Since this amounts to multiple thousands of hours per year, when economizer cooling is desired, it should be worth the effort to modify the control sequence here. The bypass damper is quite small, but there is room for additional damper capacity in the mounting plate. The RTU was profiled with the dampers closed and opened, and results are shown in Table 2.


Some basic needs can be identified: humidity control is desired, (dehumidification only in this case) and the need to provide air at a temperature that it can be introduced to the rooms without creating discomfort or overloading local heat pumps. Regarding dehumidification, initially the RTU internal dehumidification sequence was employed, which was set up to run one compressor continuously for the cooling coil and then controlling DAT with a hot gas reheat coil. In an attempt to reduce RTU compressor power we are now simply holding a 60[degrees]F (15.6 [degrees]C) DAT during high outdoor dewpoint times, which delivers indicated 55[degrees]F (12.8[degrees]C) dewpoint supply air, and little evidence is seen of significant latent loads in the building (no spaces show particularly elevated dewpoints). The disadvantage with this may be the additional reheating requirments imposed on the high ventilation areas such as locker rooms.

Regarding the temperature of the DOAS supply air we have an advantage here that in most cases the air is introduced to a return duct and then is distributed by the local heat pump air distribution (constituting then 10-15% of the local air stream). In some cases however, where high volumes of outside air are used in locker rooms and cell blocks, the DOAS air is the sole air source and determines the temperature of the room supply air when the local heat pump is cycled off. This defines a lower limit of RTU supply air temperature, we'll assume 45[degrees]F (7.2[degrees]C). Fortunately this can generally be supplied in cold weather with heat wheel operation alone. There does not appear to be an upper limit of supply air temperature, so when dehumidification is not required we now let the supply temperature float upward using only heat recovery.

Operation of the heat wheel, and potential use of hot gas heat recovery in the RTU justify some consideration of whether the DAT will on the average help or hinder building space temperature control. Certainly there are local temperature control deadbands that can be used to advantage, and in high overall heating or cooling load situations the heat recovery DAT setpoints should be as low or high respectively as possible. In this particular building the spaces using large volumes of DOAS air are much more sensitive to the DAT and will reheat or recool significantly if DAT is not close to optimum. Fortunately these are interior spaces with somewhat constant thermal loads, so a target DAT of say 65F seems ideal. Attempting to hit this temperature in winter will maximize heat wheel benefit, and the same will hold true in summer.


For the purposes of achieving energy performance goals it seems difficult to overstate the value of an energy model, predicting usage in categories and system components, and then the submetering and the LEED M&V task which drives the use of the energy model in auditing actual performance.

No differently than any of our other recent constructions, the systems of this building did not perform up to energy expectations out of the box, and more work remains. Good energy performance requires attention to a set of details which can be unfamiliar to our design and construction and operations teams, particularly when newer systems like DOAS with heat recovery are implemented in different situations. Constructability, operability, and first cost are all necessary considerations.

Valuable tools exist in the industry however, and coordinating the functions of design, energy modeling, construction, commissioning, and operations is clearly essential, regardless of which entity drives the coordination. A comprehensive design intent document with basis of design information specifying details of anticipated energy performance would go a long ways to accomplishing this coordination.

As we learn here effective DCV strategies can be challenging to implement, though tools such as the central air quality monitoring system are valuable resources, and occupancy sensors are getting better all the time. The human component consisting of building occupants is also invaluable, but difficult to design for. Commissioning of these systems would be nearly impossible without BAS monitoring.

Excellent off the shelf equipment is available to implement high performing DOAS, and air horsepower can be greatly reduced if the airside equipment can be oversized for the actual operating flows encountered. In this case as long as operating airflows are approximately 50% of unit design the energy model target for supply fan power, 0.469 w/cfm (0.99 w/l/s) can be achieved. By comparison the ASHRAE 90.1-2010 requirements for design brake horsepower would be closer to 1.2 w/cfm (2.5 w/l/s) in such a VAV system even before credits for ducting and heat recovery were taken. It should be noted that the 0.3" in. wc (0.075 kPa) static pressure setting for this operation would have led to serious air shortfalls in the larger demanding zones if it were not for the local fans in series.

Energy used for temperature control in a DOAS is very climate sensitive, and benchmarks were not apparent, but best practices would dictate floating DAT as much as possible to avoid mechanical heating or cooling.


We acknowledge the assistance and cooperation of the UMass/Amherst Facilities Planning Department, the UMass Building Authority, the mechanical designer Consulting Engineering Services Inc., the LEED consultant Enermodal Engineering, mechanical contractor Adams Plumbing and Heating, balancing contractor Performance Testing and Balancing, and temperature controls contractor Johnson Controls. These parties continue to enable the ever improving energy performance of this building.

BAS     =  Building Automation System
cfm     =  Cubic Feet per Minute
DAT     =  Discharge Air Temperature
DCV     =  Demand Controlled Ventilation
DOAS    =  Dedicated Outdoor Air System
In. wc  =  Inches water column (air pressure)
kW      =  kilowatts
LEED    =  Leadership in Energy and Environmental Design
rh      =  Relative Humidity
RTU     =  Rooftop air handling unit.
OAT     =  Outside Air Temperature
W       =  watts


Acker, B. and Wymelenberg, K.VD, 2011, Demand Control Ventilation: Lessons from the Field- How to Avoid Common Problems, ASHRAE Transactions, 117, (1) pp. 502-508.

Mumma, S.A., 2011. DOAS Misconceptions, ASHRAE Journal, August 2011, pp. 76-79.

Murphy, J., 2012. Total Energy Wheel Control in a Dedicated Outside Air System, ASHRAE Journal, March 2012, pp. 46-58.

Jason J. Burbank, PE

Associate Member ASHRAE

Justin M. Marmaras

Student Member ASHRAE

Dragoljub B. Kosanovic, PhD

Jason Burbank serves as Campus Energy Engineer for the UMass/Amherst Physical Plant, Justin Marmaras is a Staff Engineer and Graduate Student at the University of Massachusetts/Amherst, and Dragoljub (Beka) Kosanovic is Director of the Industrial Assessment Center in the Mechanical Engineering Department at UMass/Amherst, Amherst, Massachusetts..
Table 1 Operating parameters in dehumidification mode

                                      Outside Air   Discharge Air

Temperature, [degrees]F ([degrees]C)   82.5 (28.1)   59.7 (15.4)
Relative Humidity, %                   49.5          84.7
      CO, ppm                           0.0           0.0
      CO2,ppm                         390.1         393.1
Dewpoint, [degrees]F ([degrees]C)      61.7 (16.5)   55.2 (12.9)
Enthalpy, BTU/lb                       33.5(77.9)    24.9 (57.9)
Total VOC, ppm                          0.17          0.18

                                      Return/Exhaust Air

Temperature, [degrees]F ([degrees]C)   72.7 (22.6)
Relative Humidity, %                   52.8
      CO, ppm                           0.0
      CO2,ppm                         416.5
Dewpoint, [degrees]F ([degrees]C)      54.1 (12.3)
Enthalpy, BTU/lb                       27.6 (64.2)
Total VOC, ppm                          0.18

Table 2 Static Pressure Profile at RTU--Following terminal unit damper

Condition/Location           0.3 in. wc         03 in. wc
                             (.075 kPa)         (.075 kPa)
                             Bypass Open        Bypass Closed

OA Flow, cfm (l/s)           4973 (2347)        5014 (2367)
OA Intake in.wc (kPa)          -0.066(-0.016)     -0.055 (-0014)
Lvg. Prefilter in.wc (kPa)     -0.14 (-0.0035)    -0.23 (-0.057)
Lvg. Heat Wheel in.wc          -0.68 (-0.17)      -0.97 (-0.24)
Lvg. Final Filters in.wc       -0.89 (-0.22)      -1.13(-0.28)
Lvg. Coil in.wc (kPa)          -1.09 (-0.27)      -1.36 (-0.34)
Lvg. Supply Fan in.wc           0.55 (0.14)        0.55 (0.14)
Supply Duct Static in.wc        0.3 (0.075)        0.3 (0.075)
Exhaust Air Flow, cfm (l/s)  4900 (2313)        4884 (2305)
Exh Entrg. Wheel in.wc         -0.5 (-0.12)       -0.57 (-0.14)
Exh Lvg. Wheel in.wc (kPa)     -1.35 (-0.34)      -1.37 (-0.34)
Exh Fan Lvg. in.wc (kPa)        0.03 (0.0075)      0.02 (0.0050)
Supply Fan kW                   2.04               2.39
Supply Fan W/cfm                0.410 (0.87)       0.476 (1.01)

Condition/Location           1.0 in. wc         1.0 in wc
                             (.25 kPa)          (.25 kPa)
                             Bypass Open        Bypass closed

OA Flow, cfm (l/s)           7667(3619)         7416 (3500)
OA Intake in.wc (kPa)          -0.096 (-0.024)    -0.075 (-0.019)
Lvg. Prefilter in.wc (kPa)     -0.35 (-0.087)     -0.42 (-0.10)
Lvg. Heat Wheel in.wc          -1.12 (-0.28)      -1.54 (-0.38)
Lvg. Final Filters in.wc       -1.54 (-0.38)      -1.88 (-0.47)
Lvg. Coil in.wc (kPa)          -1.87 (-0.47)      -2.22 (0.55)
Lvg. Supply Fan in.wc           1.39 (0.35)        1.4 (0.35)
Supply Duct Static in.wc        1.0 (0.25)         1.0 (0.25)
Exhaust Air Flow, cfm (l/s)  7576 (3576)        7480 (3531)
Exh Entrg. Wheel in.wc         -1.22 (-0.30)      -1.27 (-0.32)
Exh Lvg. Wheel in.wc (kPa)     -2.58 (-0.64)      -2.64 (-0.66)
Exh Fan Lvg. in.wc (kPa)        0.05 (0.012)       0.03 (0.0075
Supply Fan kW                   5.33               5.91
Supply Fan W/cfm                0.695 (1.47)       0.800 (1.69)
COPYRIGHT 2013 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 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Burbank, Jason J.; Marmaras, Justin M.; Kosanovic, Dragoljub B.
Publication:ASHRAE Conference Papers
Date:Dec 22, 2013
Previous Article:Life Cycle Performance Costing Based Building Design Decision Support.
Next Article:Comparison of Data Center Classes On Energy Use.

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