In this paper the possibility of UV-reactor validation based on computational fluid dynamics will be discussed and related to biodosimetry and actinometry. Microbial inactivation depends on the UV-C dose that is described as UV intensity multiplied by exposure time. As a microbe enters a chamber containing UV lamps, it will receive varying irradiance levels from lamps depending on its distance from the lamp and the exposure time will depend on the specific path of the microbe through the reactor. It is necessary for UV-C dose calculation to determine the exposure time of a particular particle (microbe) and the UV intensity as function of position in the irradiation chamber based on the assumed UV-C power emission of the lamp. We can determine UV-C dose of a particular particle as function of position using the powerful software (3D Intensity calculation) supported by computational fluid dynamics. In order to calibrate CFD model, biodosimetric tests with the Bacillus subtillis spore were carried out in the four different reactors, each reactor equipped with 3, 4, 6 and 8 lumps respectively. It was founded that CFD model for UV reactor validation was in excellent agreement with the biodosimetric results. The actinometric tests with free chlorine were also undertaken to verify its possibility as alternative to the biodosimetry and the obtained results showed that the actinometry with free chlorine was a useful tool for determination of the average UV intensity in UV reactor.
UV-disinfection Reactor Validation by Computational Fluid Dynamics and Relation to Biodosimetry and Actinometry Kenichiro Deguchi 1) , Satoshi Yamaguchi 2) and Hiroshi Ishida 3) *1 UV System Sales Engineering Department, Cyiyodakohan Co.Ltd. 5-2-1 Ginza Chuo-ku, Tokyo 104-8115, Japan (E-mail: kenichiro.deguchi@chiyodakohan.co.jp) *2 UV System Research Center, Cyiyodakohan Co.Ltd. 6-111-12 Onuma Kasukabe, Saitama 344-0038, Japan (E-mail: satoshi.yamaguchi@chiyodakohan.co.jp) *3 Process Development and engineering Department, Cyiyodakohan Co.Ltd. 5-2-1 Ginza Chuo-ku, Tokyo 104-8115, Japan (E-mail: hiroshi.ishida@chiyodakohan.co.jp) ABSTRACT In this paper the possibility of UV-reactor validation based on computational fluid dynamics will be discussed and related to biodosimetry and actinometry. Microbial inactivation depends on the UV-C dose that is described as UV intensity multiplied by exposure time. As a microbe enters a chamber containing UV lamps, it will receive varying irradiance levels from lamps depending on its distance from the lamp and the exposure time will depend on the specific path of the microbe through the reactor. It is necessary for UV-C dose calculation to determine the exposure time of a particular particle (microbe) and the UV intensity as function of position in the irradiation chamber based on the assumed UV-C power emission of the lamp. We can determine UV-C dose of a particular particle as function of position using the powerful software (3D Intensity calculation) supported by computational fluid dynamics. In order to calibrate CFD model, biodosimetric tests with the Bacillus subtillis spore were carried out in the four different reactors, each reactor equipped with 3, 4, 6 and 8 lumps respectively. It was founded that CFD model for UV reactor validation was in excellent agreement with the biodosimetric results. The actinometric tests with free chlorine were also undertaken to verify its possibility as alternative to the biodosimetry and the obtained results showed that the actinometry with free chlorine was a useful tool for determination of the average UV intensity in UV reactor. KEYWARDS UV Inactivation Kinetics, Computational Fluid Dynamics, UV Dose, Point Source Summation, Bacillus subtillis spore. INTRODUCTION UV inactivation of bacteria, in the ideal case of uniform UV intensity and piston flow, can be approximated by the first order expression. N/N 0 = exp(-kφ) (1) Where: N=bacterial density after exposure to UV; N 0 =the initial bacterial density; k=inactivation rate constant(m 2 /J); φ=UV Dose(J/m 2 ). UV Dose is described as UV intensity multiplied by exposure time. UV Dose (J/m 2 ) = Intensity(W/m 2 )×Exposure Time(sec) (2) The rate constant, k, is the slope of the relationship on ln(N/N 0 ) as a function of the dose. Journal of Water and Environment Technology, Vol.3, No.1, 2005 - 77 - In Equation 2, the use of a single exposure time presumes the ideal case of piston flow in the reactor, with no axial dispersion. However, under actual conditions, ideal piston flow does not exist. Axial dispersion, lack of radial turbulence and UV intensity gradient will cause a distribution of UV doses. The kinetics of inactivation in the actual reactor can be developed from Equation 1 as: Ne/N 0 =∫exp(-kφ)・E(φ)dφ (3) Where: Ne =Average bacterial density after exposure to UV; E(φ)=UV Dose distribution function. In order to determine UV Dose distribution function, it is necessary to know the path of the particle (microbes) in the reactor and the UV intensity in the reactor as a function of position. The particle trajectory calculations can be performed by computational fluid dynamics. UV intensity at any point P (r, z) within the reactor is developed by modifying AKEHATA Equation 1) as: I(r，z)＝Σψ 1 ・ψ 2 ・S Ｌ ・ΔL ｎ ／(π ２ ・ρ ３ )exp(－[ε Ｌ (r－r 0 )＋ε Ｑ ・δ Q ]ρ/r) (4) Where: ｒ = cylindrical radial coordinate; r 0 = radius of sleeve; ρ= spherical radial coordinate; S L = linear source strength; ΔＬ ｎ = linear source length coordinate; ε Ｌ =absorption coefficient of water; ε Ｑ =absorption coefficient of quartz sleeve; δ Q =wall thickness of quartz sleeve: ψ 1 = reflection factor ψ 2 = dirt factor of quartz sleeve surface. The reflection factor of quartz sleeve is given by Ψ 1 = (1-R 1 )(1-R 2 ) (5) Where: R 1 is reflectance on inside surface of quartz sleeve and R 2 is reflectance on outside surface of quartz sleeve. The reflectance for one side of quartz sleeve surface is given by R＝ (1/2)[sin ２ (ｉ- r)/sin 2 (i + r)+tan 2 (i-r)/tan 2 (i + r)] (6) r=sin -1 (sin(i)・(n 1 /n 2 )) (7) Where: i =incident angle; r= refraction angle, n 1 =refractive index of incident side medium, n 2 =refractive index of refraction side medium. UV DOSE RESPONSE CURVE OF BACILLUS SUBTILLIS SPORE IAM 1145 Biodosimetry is a very reliable method available at present for measurement of UV dose in a UV reactor. It involves employing a certain microorganism (e.g. the Bacillus subtillis spore or the virus MS2-phage), for which the UV dose-response curve can be determined accurately in a collimated beam apparatus. After measuring the log inactivation achieved between influent and effluent samples, the reduction equivalent dose (RED) can be obtained by reading off the UV dose corresponding to that log inactivation from the UV dose-response curve. The UV dose response curve was prepared using a collimated beam apparatus whenever the stocked strain of Bacillus subtillis spore IAM1145 was grown up overnight in nutrient broth at 37 ℃. Any sample to be tested was placed in the petri dish (sample depths of 0.5 cm). The measured amount of UV light was collimated down to the sample. The ratio of survivors to initial numbers then was plotted against UV dose. Figure 1 shows the UV dose response curve for the Bacillus subtillis spore IAM 1145 determined in a 254 nm collimated beam apparatus. - 78 - Journal of Water and Environment Technology, Vol.3, No.1, 2005 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 UV Dose(J/m 2 ) N/N 0 4 Apr.'99 18 Mar.'99 24 Mar.'99 31 Mar.'99 2 Jun.'99 7 Jun.'99 8 Jun.'99 （Eqー １） （Eqー ２） Fig. 1 UV Dose Response Curve of the Spore Bacillus Sbutillis IAM 1145 Data from doses of 95-375 J/m 2 are appeared linear and fit the regression line ” Equation (8)”. N/N 0 =5 exp(-3.387φ) (8) Equation (8) was used to determine the reduction equivalent dose (RED) in off-site validation of UV reactor and computational validation by CFD model. OFF-SITE BIODOSIMETRIC TESTING OF UV REACTOR The off - site biodosimetric tests with the Bacillus subtillis spore IAM 1145 were carried out in the four different reactors. The each reactor was quipped with 3, 4, 6 and 8 germicidal lumps (100W) respectively. The reactors equipped with 3 and 4 lamps were made from a 159.2 mm (ID) stainless Fig. 2. Schematic of off-site Biodosimetric Test Apparatus - 79 - Journal of Water and Environment Technology, Vol.3, No.1, 2005 steel column 110 cm high and the reactors equipped with 6 and 8 lamps were made from a 210.3 mm (ID) stainless steel column 110 cm high, respectively. Tap water was used as the feed to the UV reactors. The tap water was spiked with the Bacillus subtillis Spore IAM 1145 in a feed tank, which was well mixed with mixer during each run. Samples were taken before and after passing through the UV reactor at a range of measured flow rates. The ratio of survivors to initial numbers was compared to Equation (8) and RED for each flow rates were determined. Fig.3 shows the relationship between RED and the theoretical exposure time. When mixing characteristic coefficientβ is introduced, RED is given by Equation (9). RED(J/m 2 )= β×Average intensity (W/m 2 )×Theoretical exposure time (s) (9) The average intensity can be determined using the point source summation method (PSS) based on Equation (4)~(7). β can be expressed as: β = V 0 /V (10) Where: V= required volume of reactor to achieve same N/N 0 as piston flow reactor; V 0 =volume of a piston flow reactor. Fig.3. Relationship between RED and Theoretical Exposure time The slope of each regression line in Fig.3 is equal to β×the average intensity determined by PSS using Equation (4)~(7). Therefore, β can be given by: β= the slope of regression line / the average intensity determined by PSS (11) Table 1 shows the obtained β for the each reactor. The mixing characteristic coefficient β depends on axial dispersion characteristic, lack of radial turbulence and UV intensity gradient of reactor. y = 172.8x y = 229.79x y = 333.47x y = 313.57x 0 50 100 150 200 250 300 350 400 450 0.00 0.50 1.00 1.50 Theoretical Exposure Time sec RED J/m 2 3Lamps5306W/m 3 4Lamps8783W/m 3 6Lamps6648W/m 3 ８Lamps9316W/m 3 - 80 - Journal of Water and Environment Technology, Vol.3, No.1, 2005 Table 1. the mixing characteristic coefficient β Reactor with 3 lamps (ID=108.3) Reactor with 4 lamps (ID=108.3) Reactor with 6 lamps (ID=210.3) Reactor with 8 lamps (ID=210.3) Measured Average Intensity w/m 2 172.8 313.57 229.79 333.47 Average Intensity by PSS W/m 2 251.98 379.65 335.04 403.49 β= V 0 /V 0.686 0.826 0.686 0.826 UV density W/m 3 5306 8783 6648 9316 The UV density is defined as: UV density = 254nm UV output×Number of lamps/Volume of reactor (12) It was founded that the higher UV density leads to higher β values . VALIDATION OF UV REACTOR BY COMPUTATIONAL FLUID DYNAMICS MODEL To know UV dose distribution function within a reactor is important in calculation of Equation (3). The following figure visualizes the exposure path of a particle passing through the reactor mounted with 8 lamps. Fig. 4 Particle tracks in the reactor mounted with 8 lamps Fig. 5 shows UV dose distribution function of the reactor with 8 lamps at flow rate 230 m3/h calculated by using the particle tracks data and Equation (4) ~(7), where refractive index:1.5 for quartz, 1.3 for water, assumed dirt factor of quartz sleeve surface ψ 2 =1. - 81 - Journal of Water and Environment Technology, Vol.3, No.1, 2005 0 0.05 0.1 0.15 0.2 0.25 0 100 200 300 400 500 UV Dose J/m 2 E(φ) Fig.5 E(φ) curve of the reactor with 8 lamps at flow rate 230 m3/h and 20 ℃ Fig.6 shows the comparison of RED between the biodosimetry and CFD model which were simulated by using Equation (3). Fig.6 Comparison between biodosimetric RED and CFD model’s RED ACTINOMETRIC VALIDATION OF UV REACTOR WITH FREE CHLORINE A combination of UV dose-response curve and chemical actinometry were used in this experiment. A UV dose-response curve of free chlorine actinometery was prepared by replacing the microorganism and then the UV dose was obtained by reading the differences of the free chlorine concentrations between the influent and effluent. The photolysis rate constant kp of free chlorine is given by presuming first - order kinetics as described by the following equation: kp= -ln(C/C 0 )/t (13) 0 50 100 150 200 250 300 0 100 200 300 RED dete rmined by CFD model J/m 2 Biodosimetric RED J/m 2 3-Lamps 4-Lamps 6-Lamps 8-Lamps ideal line ** 線形 (**) - 82 - Journal of Water and Environment Technology, Vol.3, No.1, 2005 Fig.7. The product φ cl ・ε of free chlorine in tap water where: C=free chlorine concentrations of effluent; C 0 =free chlorine concentrations of influent; t=exposure time. In the present case, ln(C/C 0 ) is plotted against t. Then the slope is equal to kp, which is given by: kp=φ cl ・ε・I av (14) where: φ cl =quantum yield of free chlorine: ε= molar natural absorption coefficient; I av =average UV intensity. In order to determine the product φ cl ・ε, the collimated beam tests were conducted using tap water containing 0.8~1.0 mg/L chlorine. The measured amount of UV light was collimated down to sample that was placed in the petri dish (sample depths of 0.5 cm). The average UV intensity I av within sample is given by: I av = I 0 (1-10 - α・ d )/2.303/log(10 α・ d ) (15) Where : I 0 =intensity at surface of sample in petri dish; α = absorbance coefficient of sample water; d= sample depth . The intensity at surface of sample in petri dish was measured with a radiometer. The product φ cl ・ε (m 2 /Wh) is equal to the slope of the regression line in Fig.7. The off - site actinometric tests with free chlorine were carried out in a reactor equipped with one germicidal lump (100W) mounted longitudinally in the 3 cm diameter quartz sleeve. The reactor was made from a 108.3 mm (ID) stainless steel column 133 cm high. Tap water was used as the feed to the UV reactor. The tap water in a feed tank (refer to Fig.2) was not spiked with additional free chlorine. In order to calibrate the actinometry, the biodosimetric test and validation by the CFD model were also carried out. Fig.8 shows the results of the actinometric testing with free chlorine. The slope of each regression line in Fig.8 is equal to the effective average intensity (=β×the average intensity determined by PSS method using Equation (4)~(7)). y = 0.3795x R 2 = 0.979 0 0.5 1 1.5 2 2.5 3 3.5 0 2 4 6 8 10 UV Dose Wh/m 2 -ln(C/C 0 ) - 83 - Journal of Water and Environment Technology, Vol.3, No.1, 2005 Fig.8 The results of actinometric testing with free chlorine. Table 2. Comparison of the mixing characteristic coefficient Actinometry Biodosimetry CFD model Measured average intensity W/m2 179.5 121.75 115.53 Average intensity by PSS W/m2 180.5 180.5 180.5 Mixing coefficient β 0.96 0.67 0. 64 All values of C/C 0 in the actinometric test were more than 0.9. Therefore, the mixing characteristic coefficient β =1 is expected 2) . The measured average intensity in the actinometric test was approximately equal to the average intensity determined using the point source summation method. CONCLUSION The CFD model for UV reactor validation has been developed and shown capable of predicting the reduction equivalent dose (RED) close to the biodosimetric RED. The mixing characteristic coefficient has been introduced into evaluation of UV reactor and it has been shown that higher UV density leads to higher mixing characteristic coefficient. It has been shown that the actinometry with free chlorine is useful tool for both of on-site and off-site measurement method of average UV intensity in the case of C/C 0 >0.9. The combination of the actinometry with free chlorine and tracer test is suggested as useful tool for the on-site validation of UV reactor. REFERENCES 1) Takashi Akehata, Takashi Shirai (1972) Effect of Light-Source Characteristics on the Performance of Circular Annular Photochemical Reactor, Vol.5, No.4, 385 -391. Jour. of Chem. Eng. of Japan 2) (1961) Reaction Rate Constant May Modify the Effect of Backmixing, Vol.53,No.4,313 –314.Ind.Eng.Chem. y = 179.5x y = 115.53x y = 121.75x 0 200 400 600 800 1000 1200 0 2 4 6 Theoretical Exposure Time sec RED J/m 2 ＲＥＤ (Biodosimetry) ＲＥＤ (Actinometry) ＲＥＤ (CFD) - 84 - Journal of Water and Environment Technology, Vol.3, No.1, 2005 . 172.8 31 3.57 229.79 33 3.47 Average Intensity by PSS W/m 2 251.98 37 9.65 33 5.04 4 03. 49 β= V 0 /V 0.686 0.826 0.686 0.826 UV density W/m 3 530 6 87 83 6648 931 6. = 33 3.47x y = 31 3.57x 0 50 100 150 200 250 30 0 35 0 400 450 0.00 0.50 1.00 1.50 Theoretical Exposure Time sec RED J/m 2 3Lamps 530 6W/m 3 4Lamps8783W/m 3