A S H RA E J O U R N A L

5 0 A S H R A E J o u r n a l Decembe r 1997

The following article was published in ASHRAE Journal, December 1997. © Copyright 1997 American Society of Heating, Refrigerating and Air-ConditioningEngineers, Inc. It is presented for educational purposes only. This article may not be copied and/or distributed electronically or in paper form without permissionof ASHRAE.

Operating andMaintaining Rooftop

Air Conditioners

By David Houghton, P.E.Member ASHRAE

limb up on the roof of a commer-cial building, and you’ll probably

find a packaged air conditioner like theone shown in Figure 1. These units pro-vide cooling for 49% of U.S. commer-cial building space.1 Although they arerelatively durable, they often don’t getthe attention they need to run efficientlyand effectively.

Rooftop air conditioners usually pro-vide between 5–20 tons (17–70 kW) ofcooling. Some commercial buildingsalso use larger prefabricated or custom-built units (20–100 tons [70–351 kW]),smaller units that cross over into the res-idential category (1–5 tons [3–7 kW]),or “split systems” that separate theindoor and outdoor functions. Mostunits include a built-in gas furnace, elec-tric resistance heater, or heat pump capa-bility. This article focuses on the coolingand air handling equipment inside roof-top units.

A typical 10-ton [35 kW] rooftop unitcosts about $2,000 per year to operateand $10,000 to replace. The mainte-

nance steps rec-ommended in thisarticle cost verylittle per year toimplement, and pay dividends in loweroperating costs, better occupant com-fort, and longer equipment life.

Potential for Improvement

Rooftop air conditioners must endureradiative heat, high ambient tempera-tures, leaves, dust, insects, pollen, rain,hail, snow, and even lightning. In manycases no one on-site is responsible foroperating them, so the units are ignoreduntil they break down. Field measure-ments of rooftop unit efficiency showperformance substantially below catalogratings. A project that examined the per-formance of rooftop units in Mississippifound that two 10-ton (35 kW) unitsrated at EER 9.0 operated at EER 6.6 and7.1, respectively,2 while a test in Con-necticut that measured two 10-ton (35kW) units rated at EER 8.7 found aver-age efficiencies of EER 6.6 and 8.6respectively over a three-week period.3

A program of regular rooftop unitmaintenance can improve performance.A project that studied efficiencyimprovements in 25 commercial rooftopunits in New England, for example,resulted in average energy savings of11%, demand reductions of 2%, andpaybacks of just under three years,4 anda similar project in Louisiana that per-formed “complete professional tune-ups” of 23 air conditioners in motels,restaurants, and grocery stores resultedin efficiency improvements rangingfrom 22 to 42%.5 Although these exam-ples are from one-time efforts to

improve efficiency and operation, roof-top unit maintenance is a continual pro-cess, and “tune-ups” need to be doneannually.

The operation and maintenanceopportunities for rooftop units can becategorized according to their two inter-nal fluid loops: air and refrigerant. Air-side maintenance and repairs should bedone before tackling the refrigerationsystem, because most refrigeration prob-lems can’t be effectively fixed until theproper airflow is established. For exam-ple, refrigerant charge measurementswill be inaccurate if airflow is restrictedby dirty filters.

Filters

Filters play two important roles: theyhelp maintain indoor air quality, andthey protect downstream components ofan air handling system (the evaporatorcoil and fan) from accumulating dirt.

Pleated filters made with cotton orsynthetic fabrics cost more but performbetter than flat filters made of fiberglassor spun polyester mats (Figure 2). Thefabric boosts filtration efficiency frombelow 20% to about 30%, and the pleatsincrease the filters’ effective area,reduce pressure drop, and extend usefullife. For example, using 2 in. fiber orpolyester filters in a 10-ton rooftop unitcosts about $100 per year in material andlabor costs, while using 2 in. pleated fab-ric filters costs about $60 per year. Thelonger lifetime of the pleated filters (sixmonths vs. two months) outweighs theirhigher purchase cost.

About the Author

David Houghton, P.E., is a consultingengineer based in Boulder, Coloradospecializing in the design of energy-ef-ficient building systems. Mr. Hough-ton received a B.S. degree in CivilEngineering from the University ofCalifornia at Berkeley and is a Certi-fied Energy Manager.

C

Fig. 1: Although rooftop units are relatively durable, theyoften don’t receive the attention they need to run ef-feciently and effectively.

D e c emb e r 1 9 9 7 A S H R A E J o u r n a l 51

Rooftop unit filter racks are either 1or 2 in. (25 or 50 mm) thick. Two-inch(50 mm) pleated filters are better, sincethe greater media volume provides moresurface to trap pollutants, and theirlower face velocity causes less pressuredrop. Some 1 in. (25 mm) racks can beretrofitted to 2 in. (50 mm) simply byrotating the rail that holds the filters inplace—an easy way to upgrade a unit’sfiltration system.

Filter-changing intervals can be basedon the pressure drop across the filter, cal-endar scheduling, or visual inspection.Although the latter two methods are themost common for rooftop units, measur-ing air-side pressure drop is the mostreliable way to rate filter loading. Atechnician can install pressure-tap tub-ing and then use a hand-held pressuremeter or manometer to check filter sta-tus; when the pressure drop is higherthan a specified level—typically, about0.5 to 0.75 in. of water gage (125 to 188Pa) above the brand-new pressuredrop—it’s time for new filters. Facilitieswith regular filter loading can use pres-sure measurements to establish theproper filter change interval and use cal-endar scheduling thereafter. Scheduledintervals should be between one and sixmonths, depending on the pollutantloading from indoor and outdoor air andthe filter type.

Conventional wisdom holds that dirtyfilters reduce the efficiency of rooftopunits, but in reality, system effects canresult in very small energy savings.Analysis of a 10-ton (35 kW) unit showsthat a static pressure increase of 1 in. ofwater gage (250 Pa) from dirty filtersreduces compressor efficiency but actu-ally boosts fan efficiency, with a net pen-alty of only about $21 per year (1%) inenergy costs. However, dirty filters alsoreduce total airflow by 23% and cutcooling capacity by 7%. Although it’sworthwhile to change filters regularly,don’t expect big energy savings.

Evaporator Coil

Dirt on the evaporator coil causes twoproblems: it reduces system airflow, andit directly degrades the coil’s heat trans-fer efficiency, which significantly cutscooling capacity. Field work in Louisi-ana showed that the evaporator coils in87% of 23 units investigated for tune-ups needed cleaning.6

It is worthwhile to inspect the coil atleast annually to make sure that the fil-ters have been doing their job. Check

coil cleanliness by measuring supply fanamperage and filter/coil pressure drop(with fresh filters). If the amps are lowerand pressure drop is higher than lastyear’s measurement (also with fresh fil-ters), then the flow through the coil islower—which means the coil is dirtyand needs cleaning. Evaporator coilcleaning should be done with a powerwasher.

Supply Fan

Fans in older rooftop units havesleeve bearings, which are oiled metal-to-metal running surfaces. These shouldbe lightly oiled two or three times peryear with the recommended lubricant. Ithelps to place a label near the bearingswith the lubrication interval, lubricanttype, and a service log.

Newer fans are equipped with self-lubricating bearings—sealed-cassetteball bearing cartridges pre-loaded withgrease. There is no way to re-greasethese bearings, so when they finallyfail—typically after several years of ser-vice—the bearing cassette must bereplaced. Warning signs of impendingbearing failure are excessive noise,vibration, or heat emanating from thebearing.

Conventional greased ball bearingsare occasionally found in rooftop units.The most common problem with thesebearings is over-greasing, which can beas damaging as under-greasing.7 The

proper procedure is to open the drainplug and inject grease through the fill fit-ting until clean grease comes out of thedrain. Take care not to get grease or oilon the pulley wheels or belt, as thiscauses a slip-stick action.

Most HVAC technicians have at leastone story of finding a fan motor runningin the wrong direction. Centrifugal fanswill supply some air even if runningbackwards (about 50% of rated flow), sothis situation may not be apparent. Themost common cause of reverse fan oper-ation is switched wire leads on themotor. Clear labels on the fan housing,pulleys, motor, and wires can help pre-vent this problem.

Fan Belt

Loose belts slip on the pulley wheels(Figure 3), causing torque loss and rapidbelt wear, while tight belts put an exces-sive lateral load on the motor, causingrapid bearing wear. Proper belt tensioncan be achieved with a deflection straingauge, but most technicians adjust ten-sion simply by pressing on the belt witha finger. Either method works well ifperformed consistently. Belts shouldalso be aligned with a straightedge.

Some technicians advocate annual orbi-annual belt changes, while others letthem run until they break. Since a typicalbelt set for a 10-ton (35 kW) unit’s sup-ply fan costs about $5 to $10, while aservice call to replace a broken belt costs

A I R C O N D I T I O N E R

Fig. 2: Pleated filters offer better filtration efficiency and lower total operatingcosts compared to the mat filters shown here.

5 2 A S H R A E J o u r n a l Decembe r 1997

$65 or more, it makes sense to preemptbreakage with scheduled replacement.

An easy upgrade that can improvedrive train efficiency by 2 to 10% is toswitch from standard to cogged V-belts.8

The cost premium for cogged V-belts isabout 20%.

Fan Motor

Supply fan motors installed by theoriginal equipment manufacturer aregenerally standard-efficiency inductionmotors. It pays to specify premium-effi-ciency motors for new rooftop units or asreplacements for burnouts in existingequipment. For example, using the bestavailable 2 hp (1.5 kW) induction motor(89.5% efficiency instead of 81.5%) in a10-ton (35 kW) rooftop unit will saveabout $60 per year in operating costs foran extra cost of about $85—a payback of17 months.

Outside Air Dampers

Improper damper operation is a verycommon rooftop unit problem. A studyof 13 rooftop units on small commercialbuildings found that none of them hadproperly operating outside-air dampers.9

This problem can have major energyconsequences in regions that could takeadvantage of economizer operation, andpotentially serious indoor air qualityimpacts in all climates.

Damper servicing consists of clean-ing, lubricating, and testing dampermovement, which costs about $10 to $20of a technician’s time. If this preventsone of the five-ton (17 kW) compressorsin a 10-ton (35 kW) unit from runningfor 500 hours per year, it will save about$185 per year.

After cleaning and lubrication, adamper should be run through its fullrange of motion. Then the economizer setpoint should be checked. Although manyeconomizers are set at about 60°F (16°C),the set point can be as high as the returnair temperature (about 74° F [23°C]) toprovide beneficial ventilation. In high-humidity climates or where outside air isvery polluted, however, it may not makesense to maximize outside airflow at lowdry-bulb temperatures.

Cabinet Integrity

Many rooftop units spill expensivechilled air onto the roof through cabinetleaks. Most rooftop units are coveredwith access panels held in place by sheetmetal screws, but often the panels have

only one or two screws left after a fewcareless service calls. A cordless drillwith the right nut driver makes panelaccess quick and easy, and is conduciveto screw replacement. Techniciansshould also keep a bag of screws on handto replace those that are missing. Losing200 cfm (94 L/s) from a 10-ton (35 kW)rooftop unit reduces cooling and airflowcapacity by about 5% and wastes morethan $100 per year in energy costs.10

Refrigerant Charge

Methods for verifying and correctingthe refrigerant charge in a direct expan-sion (DX) cooling system range frommeasuring the length of sweating pipesto peering into sight glasses. A survey of25 refrigerant circuits in 18 rooftop unitsfound that 10 (40%) were overchargedand eight (32%) were undercharged.11

Undercharged systems usually resultfrom a leak, while overcharged systemscan occur when a technician charges theunit on a cool day or adds refrigerant to“correct” evaporator coil icing that ismore likely the result of low airflow.Figure 4 shows the impact of incorrectrefrigerant charge on unit efficiency.

The most accurate way to check andcorrect refrigerant charge is by measur-ing superheat and/or subcooling (or, ifthe unit has been evacuated, by weigh-ing in the correct amount of refrigerant).Superheat and subcooling measure-ments, however, are only meaningfulwhen correlated with the loads on thecondenser and evaporator. Techniciansshould have a blanket or other means ofblocking condenser airflow to simulatedesign ambient conditions, or a lookup

table from the manufacturer that speci-fies the proper superheat for differentambient temperatures. The superheat formost DX systems should be between10°F and 20°F (6°C and11°C).

Because the idea of superheat is toprotect the compressor, some manufac-turers now specify superheat valuesmeasured at the suction line entrance tothe compressor rather than the tradi-tional location at the expansion valvetemperature bulb. The superheat differ-ence between these two locations can beseveral degrees—enough to make a sig-nificant error in refrigerant charge.

Expansion Valve

Some service technicians blame thethermostatic expansion valve (TXV) forsystem malfunctions, but adjustments tothe internal spring’s screw setting should

MAINTENANCE TIPS

Finding a maintenance service con-tractor who exhibits quality andintegrity is essential for high qualitymaintenance. Be sure to check refer-ences, ask to see sample reports, andfind out if the contractor is commit-ted to training their technicians andequipping them with the right tools.A thorough maintenance programincludes checkups before each heat-ing and cooling season and completereports showing the findings of theservice technicians.

• Don’t let service contractorskeep the only copy of performancedocuments.

• Keep performance information(wiring diagrams, fan curves, etc.)close to the unit.

• Include a service log sheet and arecord of “alarm” conditions foreach unit.

• Records and record containersmust be rugged enough to withstandabuse. Consider using reinforced,heavy-duty paper sheets and bind-ers, and waterproof metal boxes.

• Highly visible signage encour-ages the use of service logs and othertracking documents.

• Contractual agreements withservice providers can enforce use ofservice logs and other tracking doc-uments.

Fig. 3: Loose fan belts cause rapid beltwear and loss of torque, while tightfan belts cause rapid bearing wear.

D e c e m b e r 1 9 9 7 A S H R A E J o u r n a l 53

be attempted only after other possible problems have beenruled out. The valve’s temperature sensing bulb should alwaysbe attached to the suction line with copper straps (rather thansteel hose clamps or plastic “zip-ties”) to ensure complete heattransfer between the bulb and the pipe.12

CompressorNinety percent of rooftop unit replacements are because of

failures in the hermetic motor-compressor.13 Replacing a com-pressor is major surgery, costing $1,500 to $5,000 or moredepending on tonnage. Electrical and oil tests should be part ofannual checkups for rooftopunits.

Electrical testing checkscompressor motor health bymeasuring the ground resis-tance of motor windingswith a megohmmeter (“meg-ger”), a process that requiresa $300 instrument and takesabout 10 minutes. Low meg-ger readings (typically,below 100 MΩ) call for“drying out” the system byinstalling a fresh filter-dryerand/or dehydrating the sys-tem with deep evacuation.14

Oil testing is performedon site by plugging a smallindicator vial onto aschrader-valve service port. If the oil contains acid, the vialchanges color. Test vials cost about $10 each and can be reuseduntil a compressor fails the test. The most important time totest oil for acidity is after a compressor failure, but it can alsobe used as part of the annual checkup. Systems that fail this testshould be fitted with one or more filter-dryers designed toclean acid and moisture out of the system.

If the wiring to the unit is undersized or otherwise compro-mised, the compressor motor can be damaged by low voltageconditions (the motor responds to low voltage by drawingmore current, which dramatically increases resistive heating inthe motor’s windings). If this problem is suspected and thecause of the voltage problem is too difficult or expensive to fix,a protective phase monitor can be installed in the rooftop unitjust before the unit’s main breaker, at a cost of $200 to $300.15

These devices cut power to the unit when voltage is out of tol-erance (typically, plus or minus 10%) and restart when condi-tions are safe again.

Condenser CoilA dirty condenser coil that raises condensing temperature

from 95°F (35°C) to 105°F (41°C) will cut cooling capacity by7% and increase power consumption by 10%, with a net (com-pressor) efficiency reduction of 16%.16 Such performancedegradation on a 10-ton (35 kW) unit operating at 9.0 EER for2,000 hours per year wastes about $250 per year in operatingcosts.17

It costs about $50 in labor and materials to clean the con-denser with a power washer that feeds cleaning solution into ahigh-pressure water flow. (Spray-on cleaning solutions that

are intended to be used with a brush and a hose will not do agood enough job of cleaning the coils, even though they maybrighten the outer surface.)18 This is money well spent—cleaning the unit in this example has a payback of just over twomonths, with a net annual savings of $200. A serious con-denser cleaning will include before-and-after measurements ofthe temperature difference across the coil to verify the effec-tiveness of the job.

Condenser Fan and MotorMost condenser fan motors are equipped with sealed-cas-

sette ball bearings that needno lubrication. Excessivenoise, vibration, or heat atthe bearing indicate that thecassette needs to bereplaced.

Rapid on-off cycling of acondenser fan (three minutesor less) leads to poor controlof the refrigeration systemand can wear out the fanmotor prematurely. Thisproblem is often caused bya narrow deadband on thehead pressure controller forthe condenser fan. Ahealthy deadband (20 to 50psi [138 to 345 kPa])between cut-in and cut-out

pressure set points prevents rapid condenser fan cycling.19

Like all air conditioners, rooftop units are complexmachines that quickly consume their purchase price in opera-tion costs. Maintaining and servicing these units to reduceenergy costs and avoid expensive repairs is a significant tech-nical and organizational challenge that deserves the full atten-tion of facility managers and service providers.

References1. U.S. Department of Energy, Energy Information Administra-

tion, “Commercial Buildings Characteristics 1992,” DOE/EIA-0246(92), page 201 (April 1994).

2. Mukesh Khattar and Michael Brandemuehl, “DehumidificationPerformance of Unitary Rooftop Air Conditioning Systems: KmartDemonstration,” EPRI TR-106066, 3565-06, Final Report (May1996), Electric Power Research Institute, Palo Alto CA, 415-855-2514, p. 4-7.

3. Scott Silver, Philip Fine, and Fred Rose, “Performance Moni-toring of DX Rooftop Cooling Equipment,” Energy Engineering, vol.87, no. 5 (1990), pp. 32-41.

4. Martha Hewett, David Bohac, Russell Landry, Timothy Dun-sworth, Scott Englander, and George Peterson, “Measured Energyand Demand Impacts of Efficiency Tune-Ups for Small CommercialCooling Systems,” Proceedings, ACEEE 1992 Summer Study onEnergy Efficiency in Buildings (1992), pp. 3.139-3.140.

5. Michael Carl and Joseph Smilie, “How Maintenance ImpactsAir Conditioning Performance and Demand,” proceedings of the1992 International Winter Meeting of The American Society of Agri-cultural Engineers, Nashville, TN, (December 1992), p. 7.

6. Carl and Smilie [5].

A I R C O N D I T I O N E R

Fig. 4: Because most rooftop units are “critical-charge” sys-tems (there is no liquid receiver to buffer the flow of refrig-erant) they are very sensitive to the amount of refrigerant inthe system.

5 4 A S H R A E J o u r n a l Decembe r 1997

7. Bill Howe et. al., E SOURCE DrivepowerTechnology Atlas, (1996), p. 92.

8. Howe [7].

9. Alan Vick, John Proctor, and FrankJablonski, “Evaluation of a ‘Super Tune-Up’Pilot Program for Forced-Air Furnaces inSmall Commercial Buildings,” Proceedings,International Energy Program EvaluationConference, Chicago Illinois, p. 503 (1991).

10. This flow rate would result from atotal leakage area of 10 in.2 (6450 mm2) andexit velocity of 33 mph (15 m/s). Assuming4000 cfm (1888 L/s) unit capacity, 5% cabi-net leakage, EER of 9.0, runtime of 2000hours per year, and electricity cost of 8¢/kWh, the annual cost of air leakage is$106.40.

11. Hewett et. al. [4], pg. 3.131.

12. Dale Rossi, personal communication(September 20, 1996), Chief TechnicalOfficer, Four Seasons Mechanical Inc., 1979Stout Drive, Ivyland PA 18974, tel 215-672-9600, fax 215-671-9658, e-mail dtrossi@acrx.com, web www.acrx.com.

13. David E. Stouppe and Tom Y. S. Lau,“Refrigeration and Air Conditioning Equip-ment Failures,” The Locomotive (the quar-terly magazine of the Hartford Steam BoilerInspection and Insurance Co., Hartford CT),Spring 1988 vol 6 no. 1, pp. 3-9.

14. Leon Neal, personal communication(September 17, 1996), Senior Product Engi-neer, North Carolina Alternative Energy Cor-poration, 909 Capability Drive, Suite 2100,Raleigh NC 27606-3870, tel 919-857-9018,fax 919-832-2696.

15. Keith Clark, personal communication(August 28, 1996), Service Manager, DesignMechanical, 5637 Arapahoe Road, BoulderCO 80303, tel 303-449-2092, fax 303-449-8739.

16. Robert W. Roose, Handbook ofEnergy Conservation for Mechanical Sys-tems in Buildings (New York: Van NostrandReinhold Company, 1978) p. 281. Data is fora 15-ton (53 kW) reciprocating R-22 com-pressor. At 95°F (35°C) condenser tempera-ture, head pressure is 181.8 psig ((1253 kPa),capacity is 18.3 tons (64 kW), and compres-sor input power is 14.3 boiler hp (140 kW).At 105°F (40°C) condenser temperature,head pressure is 210 psig (1448 kW), capac-ity is 17.0 tons (60 kW), and compressorinput power is 15.9 boiler hp (156 kW). (Allvalues at 45°F [7°C] suction temperature and76.6 psig (528 kPa) suction pressure.)

17. Assuming 75% of energy goes to thecompressor, and total energy costs of 8¢/kWh.

18. Keith Clark [15].

19. Ira Richter, “Condenser Short-Cycling,” Refrigeration Service & Contract-ing, vol.64, no. 8, p. 34 (August 1996).

Acknowledgements

This article draws on research performedfor E Source, an independent source of infor-mation on energy technologies based inBoulder, Colorado.

Please circle the appropriate numberon the Reader Service Card at the backof the publication.

Extremely Helpful ........................462

Helpful .......................................463

Somewhat Helpful .......................464

Not Helpful.................................465

Advertisement in the print edition formerly in this space.