study on biological contaminant control strategies under different ventilation models in hospital...

ARTICLE IN PRESS

0360-1323/$ - se

doi:10.1016/j.bu

�CorrespondE-mail addr

Building and Environment 43 (2008) 793–803

www.elsevier.com/locate/buildenv

Study on biological contaminant control strategies under differentventilation models in hospital operating room

Zhang Rui�, Tu Guangbei, Ling Jihong

Department of Environment, 24th Building, Tianjin University, No.92, Weijin Road, Tianjin 300072, China

Received 28 February 2006; received in revised form 15 November 2006; accepted 15 January 2007

Abstract

Transmission of airborne bacteria is the main factor causing surgical site infection (SSI). Previous researches have provided evidence of

relationships between cleanness of room air and incidence of SSI, but little work has been done to verify the numerical simulation results

of particle dispersion. This paper focuses on the airborne transmission of bacteria in two operating rooms during two surgeries: a surgical

stitching of fractured mandible and a joint replacement surgery. Field measurement was carried out in two newly built ISO class 5

(OR.A) and class 6 (OR.B) operating rooms. Bacteria collecting agar dishes were put in different places of the two operating rooms to get

the deposited bacteria number during the operation. Then numerical simulation was carried out to calculate the particle trajectories using

the Euler–Lagrange approach. Simulation results were compared with field measured data, and acceptable level of consistency was

found. Then we changed the supply air velocity and supply vent area in the OR.B numerical model under same room air change rate, to

compare bacteria colony deposition onto the ‘‘critical area’’, which consisted of three connected surfaces around the surgical site on

patient body. Result showed that improving air flow pattern can reduce particle deposition on critical surface, but its effect is less evident

by increasing the air change rate in a certain amount, and we found that bacteria colony deposition would increase (mainly on upper

surface), if air velocity increases beyond a certain velocity.

r 2007 Elsevier Ltd. All rights reserved.

Keywords: Hospital operating room; Particle; CFD; Ventilation

1. Introduction

Based on Centers for Disease Control (CDC) andPrevention National Nosocomial Infections Surveillance(NNIS) system [1] reports, surgical site infection (SSIs) arethe third most frequently reported nosocomial infection,accounting for 14–16% of all nosocomial infections amonghospitalized patients. In 1980, it was estimated that an SSIincreased a patient’s hospital stay by approximately 10days and cost an additional $2000. Advances in infectioncontrol practices include improved operating room ventila-tion, sterilization methods, barriers, surgical technique,and availability of antimicrobial prophylaxis. To reducethe risk of SSI, a systematic but realistic approach must beapplied with the awareness that this risk is influenced by

e front matter r 2007 Elsevier Ltd. All rights reserved.

ildenv.2007.01.018

ing author. Tel: +1412 651 7384.

ess: [email protected] (Z. Rui).

characteristics of the patient, operation, personnel, andhospital. Microbial contamination of the surgical site is anecessary precursor of SSI. It is agreed in the literature thatthe primary sources of such bacteria are squames, or skinscales or particles, and with a dimension of 5–10 mm [2].These air born coenobiums may deposit on surgical siteand cause potential infections.Since 1960s, when cleanroom technology was applied in

operating rooms for the first time, there has been plausibleevidence showing good cleanness and well organizedairflow pattern in operating room can reduce incidence ofSSI. According to Lidwell’s [3] statistic analysis result, outof 8025 cases of joint replacement surgeries, when usedcleanroom technology only, the SSI incidence droppedfrom 3.4% to 1.6%; when used antibiotic only, theincidence dropped from 3.4% to 0.8%; when used bothcleanroom technology and antibiotic, the incidencedropped to 0.7%. Thus how to improve the performance

www.elsevier.com/locate/buildenv

dx.doi.org/10.1016/j.buildenv.2007.01.018

mailto:[email protected]

ARTICLE IN PRESSZ. Rui et al. / Building and Environment 43 (2008) 793–803794

of cleanrooms to more effectively reduce bacteria concen-tration around surgical site has become an important issuein cleanroom profession.

There is not an international standard for operatingcleanroom. The first standard regarding biological clean-room is carried out by NASA—NASA standard for CleanRooms and Work Station for Medical Controlled Envir-onment (NHB 5340.2). It sets criterion on concentration ofparticles and bacteria colonies in different classes ofcleanroom. Based on that standard, USA, England,Switzerland and Japan carried out their standard on cleanoperating room in 1960s. In US, American Society ofHeating, Refrigerating and Air-Conditioning Engineers(ASHRAE) and The US Department of Health andHuman Services carried out similar guidelines, which setcriterion on the efficiency of filter in different classes ofoperating rooms. Japan standard sets criterion on terminalfilter efficiency, minimum air change rate and maximumcount of bacteria colonies in the operating room. Hollandstandard VCCN 2001, England standard DHSS (1986),Switzerland standard SKI (1987) and China standardGB50333-2002 specify maximum count of bacteria coloniesand set specification on filter and HVAC system. Oneparticular standard is Health Technical Memorandum(HTM) developed by UK National Health Service Estates.HTM 2025 specifies the maximum operational bacteriacount, around different areas of the operating room. Therequired index in this standard is difficult to test and it maypotentially induce risk to the patient under operation. Butthe ultimate goal of all these standards is to control theairborne bacteria colonies around surgical site to reducechance of infection.

Some works have been done on contaminant dispersionin operating room and indoor environment both experi-mentally and numerically. Chih-Shan Li et al. [4] carriedout a series of field tests with Andersen 1-STG sampler,and got the bacterial and fungal concentrations in differentparts of a hospital, including class 100 and class 10,000clean rooms. Lidwell [5] and Schmidt [6] did lot experi-ments in operating rooms, comparing bacteria concentra-tion in different air flow patterns. But in their literature,HVAC system’s parameter was not included, so their resultcould not establish definitive recommendation for actualdesign of HVAC system in operating room.

During-surgery test was sparse, because of considerationon potential risks that would bring to the patient.Currently the during-operation test could only be carriedout in surrounding areas of the surgical bed. We couldbarely know the actual bacteria concentration around thesurgical site. A new powerful tool CFD (computation fluiddynamic) has been proved an effective way to investigateair flow and contaminant dispersion in indoor environ-ment. Lot of work has been done on CFD simulation ofcontaminant dispersion. Farhad et al. [2] simulated air flowpattern and particle trajectories in an operating room withdifferent diffuser types and different range of air changerates. The particle in consideration is 10 mm in size. Three

particle source points were analyzed, and tracked. Theresults showed that ventilation systems that providelaminar flow conditions were the best choice. A facevelocity of around 30–35 fpm (0.15–0.18m/s) was sufficientfrom the laminar diffuser array, provided that the size ofthe diffuser array was appropriate. Our investigation herewould research into higher supply air velocity, which wasfiled test velocity in the two rooms. Chen et al. [10] assumedpoint sources in a room and reported that a higher airinflow rate and a large air inlet area were desirable forcontaminant control but detrimental to the thermalcomfort of the staff; particle concentrations in variousparts of the room were very sensitive to the location of theparticle sources. In these work, point source was adopted,this assumption was an inadequate approximation forsituations like the one investigated here, in which thebacteria emission model was based on experimental testeddata of operating staff, which will be described in detaillater. Chow et al. [11] assumed a bacteria emission rate of100CFU/min for each operating staff in a non-standardoperating room, with a supply diffuser screened with aperforated steel plate. Field test data was set as theboundary condition of supply velocity profile under thediffuser. Simulation was done on temperature distribution,air flow pattern and the contaminant dispersion. Resultsshowed that bacteria concentration was very sensitive tothe position of lamps. In their simulation field tested supplyvelocity profile enhanced their result’s reliability, but theydid not testify their particle transport simulation result,and the operation room in their work was a non-standardoperation room, which limited its utilization in modernoperation rooms. Woloszyn et al. [12] compared experi-mentally measured tracer-gas concentration and simulationresult in an operating room with a diagonal air-distributionsystem. Tracer-gas method has been a very handy methodto observe contaminant dispersion in a space, but as we allknow the particle’s dispersion characteristic was not thesame with gases, so the tracer gas method cannot representso well the airborne bacteria diffusion in an operatingroom. Our investigation here uses the Lagrangian for-mulation to calculate the trajectories of particle, and finds away to testify the deposited bacteria number. Woods et al.[13] developed a two-compartment model in cleanroom air.They divided the room into two parts, one was ‘‘micro-environment’’ and was defined as the space bounded by thepatient, the surgical team around the operating table, andthe surgical lamps above the patient; the other was ‘‘mini-environment’’, which was the space within the operatingroom and enveloped the micro-environment. The idea ofdividing operating room into two parts was constructive,but the ‘‘mini-environment’’ may be too large, we defined a‘‘critical area’’, which consisted of three connected areaaround the surgical site on patient body. We calculated thenumber of bacteria deposited on these three surfaces. Afterall, the surgical site bacteria concentration was crucial tothe rate of SSI. This body of work is important and hasimproved some contamination control technologies in

ARTICLE IN PRESSZ. Rui et al. / Building and Environment 43 (2008) 793–803 795

operating room, but their limitations, which stated above,prevented their application to modern operation room.

One objective of this paper is to present the operationaldeposit bacteria data in two newly built local ISO class 5(OR.A) and local ISO class 6 (OR.B) operating room. Theother is to evaluate the adequacy of Euler–Lagrangeapproach for making pollutant dispersion predictions inoperating room, based on the standard k-epsilon (k–e) twofunction turbulence model calculated flow field. Thebenefits of experimentally verified computational modelsof air flow and pollutant transport are well understood.Within their domain of applicability, such models offeredpractical means to explore pollutant dispersion in occupiedindoor spaces across a wide range of parameters, whichotherwise would be prohibitively expensive, time consum-ing, or simply impossible if based solely on experimentalinvestigation.

2. Computational model

Currently there are two approaches for the numericalcalculation of multiphase flows: the Euler–Lagrangeapproach and the Euler–Euler approach. In the Euler–Euler approach, the different phases are treated mathema-tically as interpenetrating continua. Since the volume of aphase cannot be occupied by the other phases, the conceptof phasic volume fraction is introduced. These volumefractions are assumed to be continuous functions of spaceand time and their sum is equal to one. Conservationequations for each phase are derived to obtain a set ofequations, which have similar structure for all phases.These equations are closed by providing constitutiverelations that are obtained from empirical information,or, in the case of granular flows, by application of kinetictheory. In Euler–Lagrange approach, the fluid phase istreated as a continuum by solving the time-averagedNavier–Stokes equations, while the dispersed phase issolved by tracking a large number of particles, through thecalculated flow field. The dispersed phase can exchangemomentum, mass, and energy with the fluid phase. Particletrajectory is predicted by integrating the force balance onthe particle, which is written in a Lagrangian referenceframe. This force balance equates the particle inertia withthe forces acting on the particle, and can be written as

dUi

dt¼

18n

rpd2p

CDprdpjUip �Uij

24nðUi �UipÞ þ

giðrp � rÞ

rp

þ F i.

(1)

The calculating procedure is:

1.
Solve the continuous-phase flow. 2. Create the discrete-phase injections. 3. Solve the coupled flow. 4. Track the discrete-phase injections, using plots or
reports.

The Euler–Lagrange approach is appropriate for model-ing particles, bubbles or droplets that occupied a volumefraction that is less than 10%, even their mass fraction mayfar exceed that number. Thus we choose the Euler–La-grange model for our bio-particle simulation in our model.The dispersion of particles due to turbulence in the fluid

phase is predicted using the stochastic tracking model, whichpredicts the turbulent dispersion of particles by integratingthe trajectory equations for individual particles, using thefluid velocity uþ u0ðtÞ, along the particle path during theintegration. In our simulation we use the Discrete RandomWalk Model. In this model, the fluctuating velocitycomponents are discrete piecewise constant functions of time.Particle simulations have been applied in many fields of

research and were testified in their specific field, like inchemical engineering [15], and in outdoor environment [14].Good consistency was found between simulation results andexperimental data. In their work, high concentration ofparticle in large scale outdoor environment simulation or withchemical reaction were investigated. There were some worksdone on pollutant dispersion simulation in indoor environ-ment. Lu et al. [7] calculated particle trajectories usingLagrangian approach, based on Newton’s 2nd Law ofmotion. Different sized particles’ trajectories in two roomswere calculated. The computational result agreed withexperimental data well, and it verified the common thoughtthat gravity has more influence on larger sized particles thanon smaller ones. Zhao et al. [8] in Tsinghua University usedsimilar method to simulate the flying foam’s dispersion inindoor environment during SARS explosion. Siegel et al. [9]developed and tested a model of particle deposition in typicalHVAC heat exchangers for an isothermal condition. Theyfound the amount that the model under-predicted themeasured data increases at higher velocities and especiallyfor larger particles. Beghein et al. [16] used LES method todevelop the flow field and Lagrangian approach to calculateparticle dispersion. They concluded that the computationalresults are in good agreement with the experimental data.This Lagrangian model can thus be used with confidence forinvestigating particles dispersion in a ventilated cavity. Allthese work has encouraged the use of Lagrangian approachin particle calculation in operating room.For turbulent flow modeling, we use the standard K–e

model, which is widely used in air flow simulation. Thiswas deemed more efficient than the low-Reynolds-numbermodels, which demand a finer grid at locations close tosolid surfaces for enhancing simulation stability. The basicequation is listed below.Mass:

qUj

qxj

¼ 0. (2)

Momentum:

Uj

qUi

qxj

¼ �1

rqP

qxj

þqqxj

ðnþ ntÞqUj

qxi

þqUi

qxj

� �� bgtT .

(3)

ARTICLE IN PRESSZ. Rui et al. / Building and Environment 43 (2008) 793–803796

Turbulence kinetic energy:

Uj

qk

qxj

¼qqxj

nþnt

sk

� �qk

qxj

� �þ nt

qUj

qxi

þqUi

qxj

� �qUi

qxj

þ bgi

nt

Prt

qT

qxi

� �. ð4Þ

Dissipation rate of turbulence kinetic energy:

Uj

q�qxj

¼qqxj

nþnt

s�

� �q�qxj

� �þ�

kC1nt

qUj

qxi

þqUi

qxj

� �qUi

qxj

� �

� C2�2

kþ C3

�

kbgi

nt

Prt

qT

qxi

. ð5Þ

Turbulent viscosity:

nt ¼ CD

k2

�. (6)

Energy:

Ui

qT

qxi

¼qqxi

kt þnt

Prt

� �qT

qxi

þ Sh, (7)

where C1 ¼ 1.44, C2 ¼ 1.92, C3 ¼ 0, CD ¼ 0.09, Prt ¼ 0.9.

Ui, Uj
velocity components
xi, xj
rectilinear orthogonal coordinates
P
pressure r fluid density n molecular viscosity nt turbulent viscosity b Volumetric expansion coefficient gi gravitational constant in i direction T Temperature K Turbulent kinetic energy � Dissipation rate of k
C1, C2
empirical constant in generation/destructionterm of e equation
C3
empirical constant in buoyant term of eequation
CD
empirical constant for eddy viscosity kt Thermal diffusivity dp particle diameter
rp
particle density
CDp
drag coefficient
Fi
additional acceleration (force/unit particlemass) term
3. Boundary conditions

The actual velocity profile across the HEPA filter supplyvent was measured, and we found the difference wasminimal. Although the velocity is not expected to beuniform due to viscous shearing at the edges of theopenings, we determined that the simulation results in theparticle dispersion were insensitive to the details of thevelocity profile, and we decided to use the average velocityas the flat velocity profile of supply vent. In other

literatures people’s thermal boundary condition wasusually set as constant heat flux. But we think that thesurface area of the simplified people model was far lessthan the actual area, so a constant heat flux boundarycondition would produce a local high temperature aroundpeople, which is caused by less contact surface area for heattransmission. Thus in our model people were treated asconstant temperature boundary, with field measuredtemperature: 27 1C (299K). The medical lamps were setas 150w/m2 uniform heat flux. Other solid surfaces werenon-slip velocity and adiabatic boundary conditions.One difficulty in this simulation was the boundary

condition of bacteria release rate of operating staffs. Weassumed that there was no bacteria colony before thesurgeries. Previous research result showed that bacteriacolony’s equivalent diameter in filter’s filtering efficiency is5 mm (that is to say the filter efficiency of bacteria colony isequivalent to that of 5 mm particles). HEPA filter’sefficiency on particles larger than 3 mm is 100%. So wetreated HEPA filter covered supply air vent as no bacteriacolony penetrated. Patient was different from operatingstaff, they went through several sterilizing processes beforeoperation and during the simulation they maintainedunconscious all through, their physical function remaineda low rate, all of which would reduce their bacteriaemission. At the same time, colony sent out by patient washard to reach the surgical site, because flow around patientis mainly downwards. On the other hand, doctors operateover the patient body, and bacteria colonies sent out bythem were the major threat for bacteria colony deposition.Thus in our simulation we assume that patient did not sendany bacteria colonies. In the work of Tin et al. [11], theyassumed a bacteria emission rate of 100 bacteria colonies/min for each operating staff. In fact bacteria colony releaserate from people would be influenced by many factorsincluding different clothing of people, different motionspeople are doing. Also it varies in different seasons, insummers the release rate is the highest, while in winter therelease rate drops greatly, which is testified by experimentsdone by Tu Guangbei and Hu [17]. Tu did a lot ofexperiments on bacteria colony release from peoplewearing operating coat in several different motions. Whenpeople doing arm lifting, the release rate was 681 bacteriacolonies/min?person in summer and 83 bacteria colonies/min?person in winter. Further more in Bethune’s experi-ment, he tested bacteria release rate, respectively, fromupper part and lower part of the body, and results showthat the lower part released more than upper part, and thelower body emits nearly twice as that of upper body. In ourinvestigation, the field test was carried out in June. Takingall these factors into consideration, we determined to setthe release rate at 200 bacteria colonies/min?person forupper body and 400 bacteria colonies/min?person forlower part of the body, and assume a uniform release fromsurface of the operating staffs. The total number ofparticles injected to the operating room is600� 9� 30 ¼ 162,000CFU in class 100 operating room

ARTICLE IN PRESS

a

Z. Rui et al. / Building and Environment 43 (2008) 793–803 797

and 600� 8� 30 ¼ 144,000CFU in class 1000 operatingroom. Many research [2] showed that the main bacteriacolonies source were squames, or skin scales or particlesand diameter ranged from 5–10 mm. In our model weassumed 6 mm as the diameter of bacteria colonies. Wecalculated several different diameters like 5 and 10 mm andwe found the deposition result difference was minimalwithin that range, so we assumed a uniform diameterdistribution in our simulation.

Further more, in order to evaluate the performance ofeach ventilation strategies, we defined a critical area. Itconsisted of three connected area around the surgical siteon patient’s body, see Fig. 3. We calculated and recordedthe number of bacteria deposited on these three surfaces indifferent ventilation strategies.

5

4

1 3

2

67

z x

y

b

Fig. 1. (a) Plan view and sample points position in OR.A; (b)

computation model of OR.A.

4. Field measurement

One field measurement was carried out in a newly builtlocal ISO class 6 operating room (OR.B) during a kneereplacement surgery. Fig. 1 showed a plan view of theoperating room, configuration of HEPA filter supply ventand outlet vents, and positions of sample point ofdeposited microorganism. The other one was carried outin a local ISO class 5 operating room (OR.A) during afractured jaw reduction surgery. Fig. 2 showed the planview, configuration of HEPA filter supply vent and outletvents, and the sample position. Air velocity of supply ventwas tested using a hot-wire anemometer, and we found thatair velocity was basically homogeneous, because HEPAfilter can produce uniform air flow. An average velocity of0.2m/s for OR.B and 0.43m/s for OR.A wasadopted. Room air change rate was about 38ACR/h inoperating room B, and 85ACR/h in operating room A.Bacteria collecting agar dishes positions are shownin Figs. 1 and 2. Dishes, which were placed in supply ventcovered area, were put at the height of 1.0m. Otherswere placed on the floor, which was the same with field test(Fig. 3).

In OR.B, the patient was a 71-year-old woman. Therewere four doctors, two instrument nurses, one circulatingnurse and one anaesthetist, eight operation staffs in total.The surgery started at 9:30 and finished at 12:30. In OR.A,the patient was a young man, the surgery started at 8:30and finished at 13:30. There were four doctors, three nursesand two anaesthetist, nine people in total. In the test,bacteria colony collecting agar dishes were put in differentplaces of the operating room for 30min to get thedeposited bacteria colony numbers. Then we put new agardishes to replace each original ones for another 30min.Then the bacteria colony-carrying-agar dishes were culti-vated in a 32 1C constant-temperature-oven for 48 h, whichwas referenced from NASA standard. After that, thebacteria colonies become colony-forming units (CFU) andcountable with naked eyes, and were recorded. Results arelisted in Table 1 and Figs. 4 and 5.

5. Results

The experiment test data and simulation result for OR.Bwere listed in Table 1 and Fig. 4. For particle trajectorycalculation, 60min’ calculation was carried and the resultwas then divided with 2 to get the 30min result. Despitesimplification of the computational model, there were toomany factors during the surgery might have influenced theresults. In the numerical model, bacteria concentrationchanges caused by people’s movement were not included.In field test, the doctors and equipment nurses stood attheir position and did not move around, but theanaesthetist and circulating nurse moved frequently. Theirmovement’s effect can be seen from the test data, forsample point 1, 2, 5, 9, which were located mainly aroundthe anaesthetist, the simulation result of these points werefar less than field test data. Further more, for points 6, 8,10, their test data changed greatly for each 30min CFUnumber. And the simulation results were more close to thesmallest number, and the smallest CFU numbers weremostly obtained in the same testing group. We can assumethat the smallest number for each point was when people

ARTICLE IN PRESS

10

7

8

91

2

34

5

6

y

xz

zy x

a

b

c

Fig. 2. (a) Plan view and sample points position in OR.B; (b,c)

computation model OR.B.

Lights

Doctors

Critical surface - upper surface

Critical surface - left hand surface

Critical surface - right hand surface

Fig. 3. Critical surface on patient body and the physical model of doctors

and surgical lamps.

Z. Rui et al. / Building and Environment 43 (2008) 793–803798

were relatively more stationary and the CFU number canbe seen as the steady state result, which was more close tothe condition that was assumed in simulation. Furthermore, the total particle number that was sent out byoperating staffs was about 4200 bacteria colonies/min, butthe differences between simulated and field tested resultswas less than 20 for most points. Taken all these factorsinto consideration we can say that the simulation result wasreliable. Fig. 9 showed the bacteria concentration contourmap of horizontal surface that is right above the patient, inOR.A operating room.

In OR.A, we collected 60min bacteria number for eachsample points. 60min particle trajectory was calculated innumerical simulation. In order to compare with the resultof OR.B, both tested and simulated results were dividedwith 2 to get the 30min CFU number, see Fig. 5 fordetailed numbers. Simulation results for points that werelocated out of the HEPA filter covered area (i.e. point 5, 6,7) were less than that tested, which might be influenced bypeople’s movement. For HEPA covered operating area,although there had a high concentration of contaminantssources, the deposited bacteria number was close to zeroboth in field test and simulation. Moreover, the low CFUnumber proved that HEPA filter covered area in thisoperating room has a much stronger stability againstdisturbances from surroundings, caused by circulatingnurses movement of others, and better protected thesurgical area. Fig. 8 showed the bacteria concentrationcontour map of horizontal surface that is right above thepatient, in OR.A.From the comparison above, we know that the

numerical method can be used as a reliable reference inevaluation and optimization of operating room HVACsystem design. HVAC system in clean operating roomserves one ultimate goal that is to control the surgical sitebacteria concentration. We could not get the actualsurgical site bacteria concentration during the surgery, sonumerical method becomes a powerful tool for us. We alsocompared CFU number for the critical area, which wedefined before, in the two rooms. Great difference wasfound; see Table 2 for detailed numbers. The upper surfacehad a very high risk of bacteria contamination in the threesurfaces for each room. OR.A had obviously betterbacteria control efficiency, with 86.9% less bacteria lessthan that of OR.B. From field test data of the twooperating rooms, HVAC system in OR.A was much betterin protecting the surgical area from disturbs of circulatingnurses’ movement, since nearly zero bacteria collected byagar dishes around the surgical bed, although with a high

ARTICLE IN PRESS

Table 1

Field test and simulated result for OR.B

Sample 1 2 3 4 5 6 7 8 9 10

1st 30min test result 36 37 6 8 34 10.5 — 16.5 6.5 5

2nd 30min test result — 20 6 7 44 5.5 13.5 7.5 — 16.5

3rd 30min test result — — 8 12 42 10.5 23.5 18.5 7.5 7.5

Average CFU 36 28.5 6.7 9 40 8.8 18.5 14.2 7.0 9.7

Smallest CFU for each sample point 36 20 6 7 34 5.5 13.5 7.5 6.5 5

Simulation result for 30min 0 0 3.5 6 1 7 6 1 1 1

05

1015202530354045

0 2 4 6 8 10 12Sample Points

Subs

ided

Bac

teri

a N

umbe

r

Averagedeposited CFUSmallest CFU foreach sample pointSimulation resultfor 30 minutes

Fig. 4. Comparison of field tested deposited CFU between computational

results in OR.B.

0

2

4

6

8

10

12

0 1 2 3 4 5 6 7 8Sample Points

Subs

ided

Bac

teri

a N

umbe

r Field Measured DepositedBacteria Number

Numberical simulatedDeposited Bacteria Number

Fig. 5. Comparison of field tested deposited CFU between computational

results in OR.A.

Table 2

Comparison of particle deposition between class100 operating room and

OR.B

Surface Upper surface Left hand surface Right hand surface

OR.B (CFU) 7 1 1

OR.B (CFU) 48 12 9

0

10

20

30

40

50

60

70

80

A B C

Air change rate (ACH)Total CFU numberCFU number of Upper surfaceCFU number of Left hand surfaceCFU number of Right hand surface

Fig. 6. Comparison of particle deposition on critical surface between

strategies A B,C, (under the same air change rate with A).

0

10

20

30

40

50

60

70

80

0.15 0.20 0.25 0.30 0.35 0.40

Supply Air Velocity (m/s)

Air change rate (ACH)Total CFU numberCFU number of Upper surfaceCFU number of Left hand surfaceCFU number of Right hand surface

Fig. 7. Comparison of particle deposition on critical surface between

strategies A D, E, F, G (under the same supply air vent with A).


concentration of surgical staffs and with circulating nursesmoving around.

Further more, we did some changes in the OR.B in thenumerical model. The different strategies (different combi-

nation of supply air velocity and HEPA filled supply ventarea) and simulation result were listed in Figs. 6 and 7. InFig. 6 each strategy was under the same air change rate. Wecan see that strategies B and C have both improved thebacteria control capability. In strategy B we reduced thewidth of the HEPA filled supply vent by 200mm, and

ARTICLE IN PRESS

y

z x

7.00e-08

6.65e-08

6.30e-08

5.95e-08

5.60e-08

5.25e-08

4.90e-08

4.55e-08

4.20e-08

3.85e-08

3.50e-08

3.15e-08

2.80e-08

2.45e-08

2.10e-08

1.75e-08

1.40e-08

1.05e-08

7.00e-09

3.50e-09

0.00e-00

Fig. 8. Bacteria concentration contour map of horizontal surface that is right above the patient, in OR.A.

7.00e-08

6.65e-08

6.30e-08

5.95e-08

5.60e-08

5.25e-08

4.90e-08

4.55e-08

4.20e-08

3.85e-08

3.50e-08

3.15e-08

2.80e-08

2.45e-08

2.10e-08

1.75e-08

1.40e-08

1.05e-08

7.00e-09

3.50e-09

0.00e-00z

y

x

Fig. 9. Bacteria concentration contour map of horizontal surface that is right above the patient, in OR.B (ventilation strategy A).


increased the supply air velocity accordingly (Fig. 8). Thenumber decreased by 9, about 13% compared with theoriginal design strategy A. In strategy C we raised thewidth of the HEPA filled supply vent by 200mm, anddecreased the supply air velocity accordingly. The numberdecreased by18, about 26% compared with the originaldesign strategy A. Obviously increasing the HEPA filtercovered area is more effective in bacteria controlling forsurgical area in steady state. Each bacteria concentrationprofile of the surface right above the patient is shown inFigs. 9–11. We can see that strategy B has thelowest concentration around the doctors, and in accor-dance, the number of particles deposited on the two side

surfaces in strategy B was lower than that of the other twostrategies. But due to high velocity the upper surface didnot receive fewer particles than that in the original design,and in most surgeries upper surface was the main surgicalsurface, then increasing the HEPA covered supply ventarea showed a favorable bacteria control result in oursimulation.Fig. 7 lists particle deposition in strategies with different

air velocity and same supply vent. We can see that therewas a valley of CFU number of upper surface and TotalCFU number. Thus there was a best supply air velocity.The deposited CFU number would increase, if air velocityincreases beyond a certain velocity. Fig. 12 shows the

ARTICLE IN PRESS

7.00e-08

6.65e-08

6.30e-08

5.95e-08

5.60e-08

5.25e-08

4.90e-08

4.55e-08

4.20e-08

3.85e-08

3.50e-08

3.15e-08

2.80e-08

2.45e-08

2.10e-08

1.75e-08

1.40e-08

1.05e-08

7.00e-09

3.50e-09

0.00e-00 z x

y

Fig. 10. Bacteria concentration contour map of horizontal surface which is right above the patient, in ventilation strategy B in OR.B.

7.00e-08

6.65e-08

6.30e-08

5.95e-08

5.60e-08

5.25e-08

4.90e-08

4.55e-08

4.20e-08

3.85e-08

3.50e-08

3.15e-08

2.80e-08

2.45e-08

2.10e-08

1.75e-08

1.40e-08

1.05e-08

7.00e-09

3.50e-09

0.00e-00z x

y

Fig. 11. Bacteria concentration contour map of horizontal surface which is right above the patient, in ventilation strategy C in OR.B.: (a) 0.2m/s; (b)

0.23m/s; (c) 0.25m/s; (d) 0.3m/s; (e) 0.35m/s.


velocity field over the patient for each supply velocity. Wecan see that vortex developed above the patient aftersupply velocity exceeds 3.0m/s, which is the main reasonthat further higher supply velocity does not reduce theparticle deposition on patient. The buoyancy effect on flowfield around operating staff and patient is minimal. In thisoperating room, with the currant configuration of supplyvent and return air vents, the best supply air velocity is0.25m/s.

6. Conclusion and future work

This paper use Standard K–e model and Lagrangianapproach simulated particle dispersion in operating rooms.

Simulated results agree with field test data acceptably.Then we defined a critical area, which consisted of threeconnected areas around the surgical site on patient body.We calculated the number of bacteria deposited on thesethree surfaces. Great difference of that number between thetwo different ISO class operating rooms was found, and wechanged the computational model in the OR.B. We foundthat under the same air change rate by improving air flowpattern can reduce particle deposition on critical surface inthe static state, but its effect is less evident by increasing theair change rate in a certain amount. Different air velocitieswere simulated, with the same supply vent area. Resultshowed that CFU deposition would increase, if air velocityincreases beyond a certain velocity. In this operating room,

ARTICLE IN PRESS

2.37e-012.26e-012.14e-012.03e-011.91e-011.80e-011.68e-011.57e-011.45e-011.34e-011.22e-011.11e-019.96e-028.81e-027.67e-026.52e-025.38e-024.23e-023.09e-021.94e-028.00e-03

x

z

y x

z

y

x

z

yx

z

y

y x

z




6.56-016.24e-015.91e-015.59e-015.26e-014.94e-014.61e-014.29e-013.97e-013.64e-013.32e-012.99e-012.67e-012.34e-012.02e-011.69e-011.37e-011.05e-017.21e-023.96e-027.18e-03

a b

c d

e

Fig. 12. Velocity field above the patient under each supply velocity: (a) 0.2m/s; (b) 0.23m/s; (c) 0.25m/s; (d) 0.3m/s; (e) 0.35m/s.


with the currant supply vent size and return air position,the best velocity is 0.25m/s.

Our investigation was the static state simulation. Wethink that future work should be done on a dynamicsimulation to investigate staff movement’s effect on CFUdeposition risk on patient body. And if computation

condition permits, a Large Eddy simulation should becarried out to compare the difference.Furthermore, inspired by this investigation, we think

that none virulent bacterial particles can be used experi-mentally in investigating particle deposition numericalmodel in future researches.

ARTICLE IN PRESSZ. Rui et al. / Building and Environment 43 (2008) 793–803 803

Reference

[1] Alicia J, et al. Guideline for prevention of surgical site infection.

Infection Control and Hospital Epidemiology 1999;20(4).

[2] Memarzadeh F, Manning A. Reducing risks of surgery. ASHRAE

Journal 2003:28–33.

[3] Lidwell OM, Elson RA, Lowbury EJ, et al. Ultra clean air and

antibiotics for prevention of postoperative infection, A multi-center

study of 8052 joint replacement operations. Acta Orthopaedica

Scandinavica 1987;58:4–13.

[4] Li Chih-Shan, Hou Po-An. Bioaerosol characteristics in hospital

clean rooms. The Science of the Total Environment 2003;305:169–76.

[5] Lidwell OM. Air, antibiotics and sepsis in replacement joints. Journal

of Hospital Infection 1988;11(Suppl. C):18–40.

[6] Schmidt P. Air control in operating theaters. Heizung Luftung Haus

Technik 1987;38(3):145–53.

[7] Weizhen L, Howarth AT, Nor A, Riffat SB. Modeling and

measurement of airflow and aero particle distribution in a ventilated

two-zone chamber. Building and Environment 1996;5:417–23.

[8] Zhao B, Zhang Z, Li X. Computational investigation of human

body’s foam dispersion in indoor environment. Journal of HV&AC

2003;6(33) (in Chinese).

[9] Siegel J, Nazaroff WW. Predicting particle deposition on HVAC heat

exchangers. Atmospheric Environment 2003;37:5587–96.

[10] Chen Q, Jiang Z, Moser A. Control of air borne particle

concentration and draught risk in an operating room. Indoor Air

1992;2:154–67.

[11] Chow Tin-Tai, Yang Xiao-Yu. Performance of ventilation system in

a non-standard operating room. Building and Environment

2003;38:1401–11.

[12] Woloszyn M, Virgone J, St1ephane M. Diagonal air distribution

system for operating rooms experiment and modeling. Building and

Environment 2004;39:1171–8.

[13] Woods JE, et al. Ventilation requirements in hospital operating

rooms—part I: control of airborne particles (RP-202). ASHRAE

Transactions 1986;92(2A):396–426.

[14] Xia J, Leung DYC. A concentration correction scheme for

Lagrangian particle model and its application in street canyon

air dispersion modeling. Atmospheric Environment 2001;35:

5779–88.

[15] Ibsen CH, et al. Comparison of multi fluid and discrete particle

modeling in numerical predictions of gas particle flow in circulating

fluidized beds. Powder Technology 2004;149:29–41.

[16] Beghein C, Jiang Y, Chen Q. Using large eddy simulation to study

particle motions in a room. Indoor Air 2005;15:281–90.

[17] Hu Z. Study on bacteria emission rate of human body and relation

ship of deposited bacteria and suspending bacteria in air. Master

degree dissertation, Tianjin University, 1988, In Chinese.

study on biological contaminant control strategies under different ventilation models in hospital...

Documents