Uvgi Design Basics

By W.J. KOWALSKI, PE, andWILLIAM P. BAHNFLETH, PhD, PE,*Department of ArchitecturalEngineering,The Pennsylvania State University,University Park, Pa.

Lethal to microorganisms, ultra-violet radiation in the range of2250 to 3020 angstroms is usedin a variety of disinfection ap-

plications, a process referred to as ultra-violet germicidal irradiation (UVGI).

Since the first UVGI system wassuccessfully implemented for disinfect-ing the municipal water system in Mar-seilles, France,1 in 1909, the disinfec-tion of medical equipment using UVGIhas been a common and reliable prac-tice. But unlike water- and equipment-disinfection applications, the disinfec-tion of air streams using UVGI has ahistory of varying success and unpre-dictable performance.

The first laboratory studies onUVGI of air in the 1920s showed such

promise that the elimination of air-borne disease seemed possible. In 1936,Hart used UVGI to sterilize air in a sur-gical operating room.2 In 1937, the firstapplication of UVGI for a school ven-tilation system dramatically reducedthe incidence of measles, with subse-quent applications enjoying similarsuccess.3 Experiments by Riley andO’Grady4 resulted in the elimination oftuberculosis (TB) bacilli from hospital-ward exhaust air.

A plethora of designs that were moreimitative than engineered followedthese early applications. The result wasa mixture of successes and failures. Thisexperience is reflected in variousguidelines that decline to sanction theuse of UVGI as a primary system. A1954 study on the use of UVGI showeda failure to reduce disease in Londonschools. Although limited data areavailable to determine the causes ofearlier design failures, the apparentcloning of UVGI systems without re-gard to operating conditions probablydoomed many installations from thestart.

A review of current industry prac-tices indicates that information on thedesign of UVGI systems lacks the de-tail necessary for engineers to ensureperformance. This article addresses thefactors that determine the design

UVGIDesignBasicsfor Air and Surface Disinfection

1Superscript numerals indicate references listedat end of article.

*William P. Bahnfleth is a member of HPACEngineering’s Editorial Advisory Board.

FIGURE 1. Types of UVGI systemsand approximate share of market.

FIGURE 2. Approximate breakdown ofwhere UVGI air-disinfection systemsare being installed.

Ultraviolet germicidal irradiation lamps can help clean coils and improve indoor air quality

Pho

to c

ourt

esy

of U

ltraV

iole

t Dev

ices

, Inc

.

HPACENGINEERING

100 January 2000 • Heating/Piping/AirConditioning

In-duct systems,27%

Roomrecirculation,

17%Microbial-

growth control,32%

Upper air,25%

Shelters,19%

Hospitals,41%

Clinics,19%

Other, 3%

Prisons,19&%

Heating/Piping/AirConditioning • January 2000 101HPACENGINEERING

parameters of UVGI systems and dis-cuss methods that can be used to sizesystems more effectively.

TYPES OF UVGI SYSTEMSFigure 1 shows the types of UVGI

systems that are sold for building-airand air-handling-unit (AHU) applica-tions and their approximate share ofthe market, based on estimates from anumber of major manufacturers. Theuse of systems for disinfecting air andcontrolling microbial growth is grow-ing in the United States and Europe,according to manufacturers. In theThird World, however, demand for up-per-air-disinfection systems is high be-cause of the TB pandemic, strainedeconomics, and the common use ofnatural ventilation.

As shown in Figure 2, health-care fa-cilities are where the most UVGI sys-tems are installed. Notably absent areschools, office buildings, and publicand residential buildings, even thoughthese are major sources of contagiousrespiratory diseases.

AIR-STREAM-DISINFECTIONAPPLICATIONS

The first step in the design of an air-stream- or surface-disinfection systemis to characterize the application. Thisincludes describing the air stream,identifying the specific surface, and,sometimes, targeting specific microbes,such as TB.

UVGI units commonly are locatedin an AHU downstream from the mix-ing box. Photo A shows a typical air-

stream-disinfection system installeddownstream from the filter bank andupstream from the cooling coils.

Although UVGI systems also can beplaced in a return-air duct to deal withrecirculated, contagious pathogens,they are rarely placed in outside-air-supply ducts. Spores, which hail fromthe outdoors, are more efficiently re-moved by filtration alone. An excep-tion exists in cases such as AIDS clin-ics, where environmental bacteriafrom the outdoors could threaten im-munodeficient patients indoors.

SURFACE-DISINFECTIONAPPLICATIONS

UVGI for microbial-growth controlhas been undergoing much study re-cently and has enjoyed success in fieldapplications.5,6 Microbial growth maybe comprised of fungi, bacteria, oreven algae, but never viruses. In Eu-rope, microbial-growth control oncooling coils has been practiced in

breweries since at least 1985. Onemanufacturer recommends placing a15-W lamp 1 m from the surface ofcooling coils or walls where condensa-tion may occur.7

Direct UVGI exposure can sterilizeany surface if given enough time.Theoretically, low-intensity UVGIcould be used for microbial growthbecause the exposure time is ex-tended. In practical applications,however, microbial growth can occurin crevices, shadowed areas such asinsulation, and stagnant water whereUVGI may not completely penetrate.

UVGI can control microbial growthon filters subject to moisture or highhumidity. Photo B shows a test applica-tion of UVGI for controlling microbialgrowth on filters. Photos C and D showan unirradiated and irradiated filterbank, respectively. The unirradiatedfilters show natural contaminationfrom various fungal species, includingAspergillus and Penicillium, while the ir-radiated filters show no evidence of mi-crobial growth. The system in photosB, C, and D used lamps that produce arated intensity of 100 mW/cm2 at 1 mfrom their midpoints.

PHOTO A. UVGI array used for air disinfec-tion. Note the specular reflective surfaces.Photo courtesy of Lumalier Inc., Memphis.

PHOTO B. UVGI lamp array used to disinfecta filter bank. The filters are to the left.

PHOTO C (top): Microbial growth on unirra-diated filters. PHOTO D (bottom): Microbe-free irradiated filters.

Pho

to c

ourt

esy

of A

irgua

rd In

dust

ries

Inc.

, Lou

isvi

lle.

continued on page 103

TYPES OF MICROORGANISMSThe variety of microbes encoun-

tered by a given UVGI system is essen-tially unpredictable. It depends tosome degree on the type of facility andgeographic location.

All viruses and almost all bacteria(excluding spores) are vulnerable tomoderate levels of UVGI exposure. Be-cause viruses are primarily contagiouspathogens that come from humansources, they are found in occupiedbuildings. Bacteria can be contagiousor opportunistic, with many found in-doors; however, some are environmen-tal. Certain facilities, such as agricul-tural buildings, may disseminateunique types of bacteria, such as spore-forming actinomycetes.

Spores, which are larger and moreresistant to UVGI than most bacteria,can be controlled effectively throughthe use of high-efficiency filters. Thecoupling of filters with UVGI is therecommended practice in all health-care settings8 and for UVGI applica-tions in general.

MICROBIAL RESPONSE TOUVGI

A basic review of the mathematicsof UVGI disinfection will assist designengineers. The population S of a spe-cies exposed to any biocidal factor isdescribed by the characteristic loga-rithmic decay equation:

(1)

where:k = standard decay-rate constant,

cm2/mW1sI = intensity of UVGI irradiation,

mW/cm2

t = time of exposure (sec)

The standard decay-rate constantdefines the sensitivity of a microorgan-ism to UVGI and is unique to each mi-crobial species.9 It can be thought of asthe rate constant at an intensity of 1mW/cm2, providing a basis for compar-ing pathogens. The rate constant for E.coli, commonly used for design pur-poses, is 0.000767 cm2 per mW sec.

Equation 1 omits two characteristicsthat may impact the disinfection pro-

Circle 365 on Reader Service Card

UVG I DES IGN B ASICS


S t e kIt() = −

FIGURE 3. Survival curve forStaphylococcus aureus illustrat-ing the shoulder portion and two

distinct stages of decay. (Source: Sharp, G. 1940.

The effects of ultraviolet light on

bacteria suspended in air.

J. Bact. 38:535-547.)

continued from page 101

cess: the shoulder and the second stage.The shoulder represents the delay inresponse (or threshold dose) of a mi-croorganism subject to UVGI expo-sure. If air velocity is too high and thedose is insufficient, a microbe mayhave a negligible response or even re-cover from the damage. Insufficientdata exists to determine the shoulders,or threshold doses, of most airbornepathogens.

Most microbial populations exhibitcharacteristic two-stage inactivationcurves (Figure 3) in which each stagehas a unique rate constant. The totalsurvival curve is the sum of a fast-decaycurve (the vulnerable majority) and aslow-decay curve (the resistant minor-ity), as follows:

(2)

where:kf = rate constant for fast-decay

populationks = rate constant for slow-decay

populationF = fraction of the total initial pop-

ulation subject to fast-decay response

The resistant fraction of most micro-bial populations is about 0.01 percent,although some studies suggest that itcan be as high as 10 percent for certainspecies.3

A distinction exists between theterms “disinfection” and “sterilization.”Sterilization is defined as the completedestruction of all microbial species.Sterilization sometimes is consideredto be 99.9999-percent eradication, or asix-log (base-10) reduction in micro-bial population. Disinfection, on theother hand, is merely the reduction ofmicrobial population. Because airstreams are generally disinfected, notsterilized, this residual second stageusually can be ignored.

DESIGN PARAMETERSA number of parameters must be

considered when considering UVGIproducts for HVAC designs. The mostimportant factors are the air-flow orHVAC equipment that will be disin-fected, the lamp wattage and distance,and the ventilation system design itself.

Air-stream characteristicsThe characteristics of an air stream

that can impact UVGI design are rela-tive humidity (RH), temperature, andair velocity.

Increased RH is commonly believedto decrease decay rates under ultravio-let (UV) exposure. However, studieson this matter are contradictory andincomplete at present. Fortunately,because most UVGI studies have beenconducted under normal indoor con-ditions, typical room and in-duct ap-

plications are not likely to differgreatly.

Air temperature has a negligibleimpact on microbial susceptibility toUVGI.10 However, it can impact thepower output of UVGI lamps if it ex-ceeds design values.

Operating a UVGI system at air ve-locities above design will degrade thesystem’s effectiveness because of thecooling effect of the air on the lampsurface, which, in turn, will cool theplasma inside of the lamp. UV output


HPACENGINEERING


S t Fe F ek It k Itf s() ( )= + −− −1

FIGURE 5. Calculated additional light intensity from reflections and inter-reflections. Totalintensity is the sum of direct, reflected, and inter-reflected UV light.

FIGURE 4. Survival of E. coli under mixed flow and unmixed flow in square ducts of in-creasing dimension.

is a function of plasma temperaturewhen power input is constant.

Not all UVGI lamps have the sameresponse to cooling effects. Somelamps have different plasma mixtures;overdriven power supplies that respondto plasma temperature; or UV-trans-parent, infrared-blocking shieldingthat limits cooling effects. Data fromthe manufacturer should be consultedto determine the cooling effects or thelimiting design air velocities and tem-peratures within which the lamps canbe operated efficiently.Ventilation system design

A number of ventilation system pa-rameters can impact UVGI design.

Air velocity and air mixing. Doses aredetermined by the time of exposureand UVGI intensity, both of whichare dependent on the velocity profileand the amount of air mixing in theair stream. The velocity profile insideof the duct or chamber depends on lo-cal conditions and may be impossibleto know in advance with any cer-tainty. In any event, the design veloc-ity of a typical UVGI unit is similar tothat for filter banks—about 400 fpm.Sufficient mixing will occur at thesevelocities to temper the effects of anon-uniform velocity profile.

The amount of air mixing that oc-curs will affect system performance toa degree that depends on system con-figuration. This is illustrated in Figure4, which compares survival predic-tions for mixed- and unmixed-flowconditions in square ducts of increas-ing dimension. The error resultingfrom the assumption of complete mix-ing will decrease as system dimensionsincrease.

In systems in which the lamps do notspan the duct’s entire width or length,the assumption of complete mixingalso will result in larger differences,compared to unmixed flow. The im-portant point is that system operationwill lie somewhere between these twoassumptions, which provide limits de-scribing system efficiency.

Using reflectors. Reflectivity can bean economical way of intensifying theUVGI field in an enclosed duct orchamber. A surface with a reflectivityof 90 percent will reflect 9⁄10 of the lightit receives.

The results of a computer-generatedanalysis of reflectivity are shown inFigure 5. The components of reflectiv-ity—both direct and inter-reflected—will clearly sum to greater than the ini-tial direct intensity. This can occurwhenever the surface is mostly en-closed and highly reflective. Such de-

signs can considerably improve economics.

Two types of reflective surfaces ex-ist: specular and diffuse. Specular sur-faces produce mirror-like reflectionsthat are directionally dependent onthe source, while diffuse surfaces pro-duce non-directional reflections that



spread equally in all directions. Non-glossy white paper is a good exampleof a diffuse surface. Most materialspossess a combination of specular anddiffuse properties and exhibit a de-gree of directional dependence. ForUVGI design purposes, the degree ofdirectional dependence is usually notcritical.

Some materials reflect visible light,but not UV light. Polished aluminumis highly reflective to UV wavelengths,while copper, which reflects most visi-ble light, is transparent in the UVrange.

No simple method of calculatingthe three-dimensional UVGI-inten-sity field for specular reflectors exists.Ray-tracing routines using Monte

Carlo techniques are one approach,but the results do not easily lendthemselves to analysis. However,they can be rather useful for examin-ing complex geometries , such aswhen cooling coils are irradiated.Figure 6 shows ray-tracing diagramsof a UVGI lamp irradiating a bank ofcooling coils from three perspectives.Note how few of the rays penetratethe coils, even after 20 reflections.Also note how the copper tubes ab-sorb many of the rays—although cop-per is transparent to UVGI, the waterinside is not.

Combining with filtration. UVGI sys-tems generally are used in combinationwith HEPA filters, a practice usuallyrecommended for isolation-room ap-plications. For other applications,however, HEPA filters do not offer asignificant enough improvement inmicrobe-removal rates over high-effi-ciency filters to warrant their exclusiveuse with UVGI.


HPACENGINEERING



a

b

c


FIGURE 6. Ray-tracing computer model of a cooling-coil bank irradiated with a UVGI lamp. Rays arecolor-coded from blue to red in order of decreasing intensity. The staggered 5/4 coil tubes are 0.5-in. dia.with six fins per in. Five reflections are shown with 90-percent reflective duct surfaces. Perspectives are (a)isometric, (b) front, and (c) side.All ray tracings were produced using Photopia software from Lighting Technologies, Inc., Boulder CO.

Recirculation systems. UVGI sys-tems that recirculate room air or thatare placed in a return-air duct or mix-ing-air plenum deliver multiple dosesto airborne microorganisms. Al-though the effect is partially depen-dent on the air-change rate, the re-sult i s an ef fective increase inremoval rate in comparison with asingle-pass system.

Calculations of removal rates forUVGI and associated filters in recircu-lation systems can be performed byevaluating the system minute-by-minute, including filtration rates, out-side-air rates, and any microbial contaminants.Lamp considerations

The hardest part of sizing a UVGIsystem is determining the lamp wattagefor the stated disinfection goal. The in-tensity field caused by the lamp and thereflectors must be modeled and aver-aged before Equation 1 is used to pre-dict the disinfection rate.



Calculating the Intensity Field of a UVGI Lamp

The intensity field of a UVGI lamp can be computed using the followingradiation view factor from a differential planar element to a cylinder, per-

pendicular to the cylinder axis (Modest, M.F. 1993. Radiative Heat Transfer.McGraw-Hill, New York.):

The parameters in the equation at left aredefined as follows:

H = x/rL = l/rX = (1 + H)2 + L2

Y = (1 1 H)2 + L2

where:l = length of the lamp segment, cmx = distance from the lamp, cmr = radius of the lamp, cm

The intensity at any point will be the product of the view factor and thesurface intensity of the lamp. The surface intensity is simply the UV poweroutput in watts divided by the surface area in cm2.

To compute the intensity at any distance from the midpoint of a lamp,multiply the above equation by 2. From any location other than the midpoint,divide the lamp into two unequal segments and add the two view factors.View-factor algebra (see reference) can be used for other locations. If weassume that complete mixing occurs, then the intensity field for any duct canbe computed by averaging the field in all three dimensions.

FH

ATANL

H

X H

XYATAN

X HY H

LH

ATANHH

LH

d1 22

1

1

2 11

11

− =−

+ − −+

− −+

π

π

π

( )( )

Lamp-intensity field. An exact de-scription of the lamp-intensity field isnecessary to accurately determine thedose that is to be delivered to an air-borne microorganism. Lamp ratings of-ten are the sole parameter used for siz-ing a UVGI installation. Although thismay be a conservative approach whendistances to the lamp exceed 1 meter,oversizing and prohibitive economicscan result.

If complete mixing is assumed, thenany intensity field can be described bythe single value of average intensity.This requires computing the intensityat every point in a three-dimensionalmatrix defining the duct. We need toknow the field caused by the lamp

and, if necessary, the field caused bythe reflections. Although the inverse-square law has been used for this pur-pose, it has proven to be inaccurateclose to the lamp. An improved ap-proach is to use the radiation view fac-tor from a differential planar elementto a cylinder as detailed in the sidebarCalculat ing the Intensity Field of aUVGI Lamp. Ignoring reflectivity, theaverage intensity field can be conser-vatively computed by applying Equa-tion 3 to a three-dimensional matrix.

There are view factors that can beused for computing the reflected inten-sity from flat parallel or perpendicularsurfaces. Consult any thermal-radia-tion textbook for such view factors.


UVGI DES IGN BASICS

Width Height Airflow Lamp ReflectivityKill rate, percent

cm cm m3/min UV watts percent Minimum Maximum

100 50 60 12 50 45 48

75 63 74

90 74 96

100 50 60 24 50 69 72

75 85 93

90 92 99

100 50 60 36 50 81 86

75 93 98

90 97 99

100 100 120 36 50 61 64

75 72 76

90 79 83

100 100 120 48 50 70 75

75 81 85

90 87 91

100 100 120 56 50 75 80

75 85 89

90 90 94

200 200 480 96 50 47 59

75 56 68

90 62 73

200 200 480 144 50 58 74

75 68 82

90 74 86

Travel time = 0.5 sec; lamp length = 72 cm; radius = 1.9 cm

TABLE 1. Predicted disinfection rates for typical systems.

HPACENGINEERING


UVGI Economics

Table 2 summarizes the costs associ-ated with purchasing, installing, and

operating two types of UVGI systems:an air-stream-disinfection (AD) systemand a microbial-growth-control (MGC)system. The ventilation systems forboth are identical. These systems weresized using the techniques described inthe accompanying article, with

predicted disinfection rates as shown.The location used in the energy

analysis is Philadelphia, with the heatadded by the lamps resulting in acooling energy penalty for 30 percent ofthe year. No credit is taken for energyinput during the heating season.Clearly, the first cost of each of thesesystems is minor, with the maintenancecost eclipsing the energy cost.

Although the MGC system uses lesswattage, it operates continuously, whilethe AD system operates only when thebuilding is occupied. The power require-ments of the former system are appro-priate for disinfection of duct surfaces orfilter faces, but not necessarily forcooling costs.

A critical energy difference betweenthese systems occurs because the ADsystem has an ASHRAE 25-percentfilter, while the MGC system has a dustfilter only. Because the short exposuretime in an AD system may not effec-tively reduce spore levels, it becomescost-effective to use a higher-efficiencyfilter to control spores. The MGCsystem renders spores inactive withcontinuous (24-hr) exposure and, as aresult, needs only a dust filter forpurposes of cleanliness.

TABLE 2. Economic evaluation of typical UVGI systems(Corrected from original publication, January 2000)

FIGURE 7. A comparison of UVGI air-stream-disinfection (AD) and microbial-growth-control(MGC) systems for a 20-year life cycle.


Air-stream Microbial-growth Type of Application disinfection control

Design airflow 10,000 cfm 10,000 cfm

Velocity 413 fpm 413 fpm

Predicted disinfection 90 percent Serratia 99.99 percent AspergillusUVGI lamp model GPH436T5 TUV18W

Number of lamps 2 1

Height 150 cm 150 cm

Width 150 cm 150 cm

Length 150 cm 150 cm

Lamp total power (each) 36 W 18 W

Hours of operation 3744 hr 8760 hr

Energy costs

Heat generated 0.072 KW 0.018 KW

Cooling load 189 KWH 110 KWH

Total dP (lamps, fixtures, filters) 0.560 in. WG 0.290 in. WG

Total fan energy (80 percent eff.) 8016 KWH 4151 KWH

Electrical energy 270 KWH 158 KWH

Cooling load energy 189 KWH 110 KWH

Total energy 8475 KWH 4419 KWH

Rate 8 cents per KWH 8 cents per KWH

Annual energy cost $678 $354

Maintenance costs

Average tube life 9000 hr 9000 hr

Tube hours per year 7488 hr 8760 hr

Replacements per year 0.83 0.97

Cost per tube $85 $85

Annual cost $71 $89

Annual filter-replacement cost $33 $6

Maintenance (assumed) $200 $200

Annual maintenance cost $949 $642

First costs

UVGI system (AU prices) $765 $550

Labor (estimated) $1000 $1000

Total installation cost $1765 $1550

Life cycle 20 years 20 years

Interest rate 8 percent 8 percent

Life cycle cost $180 $158

Total annual cost $1806 $1154

Economic evaluation of typical UVGI systems

First, use Equation 3 to determine theintensity at the flat surface. Then, usethe appropriate view factor to deter-mine the reflected intensity after mul-tiplying by the reflectivity.

Table 1 presents a comparison ofUVGI systems that were sized usingthe view-factor method and may be

used to approximate the performanceof similar systems.

CONCLUSIONSAlthough simplistic, the methodol-

ogy presented here is more accuratethan any previously published methodfor sizing UVGI systems. The authors

Circle 369 on Card; see HPAC Info-dex, p. 129


HPACENGINEERING


Circle 506 on reader service card if thisarticle was useful; circle 507 if it was not.

hope that these principles will lead tosuccessful applications and avoidanceof the design problems that have ham-pered the industry and perplexed engi-neers. Although the goal of eliminat-ing airborne disease might remainunachievable, the information pre-sented here may help lead the industryback to the path of continuous improvement.

ACKNOWLEDGEMENTSThe authors wish to thank the fol-

lowing for providing information andsupport for this article: Ultraviolet De-vices Inc., Lumalier Inc., AmericanUltraviolet Inc., Steril-Aire Inc., In-sect-O-Cutor Inc., and Airguard Industries Inc. HPAC

REFERENCES1) AWWA. 1971. Water Quality and

Treatment. McGraw-Hill, New York.2) Sharp, G. 1939. The lethal action of

short ultraviolet rays on several commonpathogenic bacteria. J. Bact. 37:447-459.

3) Riley, R.L. 1972. The ecology of in-door atmospheres: Airborne infection inhospitals. J. Chron. Dis. 25:421-423.

4) Riley, R.L., and F. O’Grady. 1961.Airborne Infection. The Macmillan Co.,New York.

5) Shaughnessy, R., E. Levetin, and C.Rogers. 1999. The effects of UV-C on bio-logical contamination of AHUs in a com-mercial office building: Preliminary results.Indoor Environment ‘99:195-202.

6) Scheir, R., and F.B. Fencl. 1996. Us-ing UVC Technology to Enhance IAQ.Heating/Piping/Air Conditioning. February1990.

7) Philips. 1985. UVGI Catalog and De-sign Guide, Catalog No. U.D.C. 628.9,Netherlands.

8) ASHRAE. 1991. Health Facilities.ASHRAE (ed.), ASHRAE Handbook ofAppl., Atlanta.

9) Jensen, M.M. 1964. Inactivation ofairborne viruses by ultraviolet irradiation.Appl. Microb. 12(5):418-420.

10) Rentschler, H.C., R. Nagy, and G.Mouromseff. 1941. Bactericidal effect of ul-traviolet radiation. J. Bact. 42:745-774.

Documents

Uvgi Design Basics