- Email: [email protected]
Experimental study on airflow characteristics and temperature distribution in non-unidirectional cleanrooms for electronic industry
Building and Environment 46 (2011) 1235e1242
Contents lists available at ScienceDirect
Building and Environment journal homepage: www.elsevier.com/locate/buildenv
Experimental study on airflow characteristics and temperature distribution in non-unidirectional cleanrooms for electronic industry Ti Lin b, Yun-Chun Tung a, Shih-Cheng Hu b, *, Yen-Jhih Chen b a b
Department of Industrial Education, National Taiwan Normal University, Taiwan, ROC Department of Energy and Refrigerating Air-Conditioning Engineering, National Taipei University of Technology, 1, Sec. 3, Chung-Hsiao E. Rd., Taipei 106, Taiwan, ROC
a r t i c l e i n f o
a b s t r a c t
Article history: Received 10 July 2010 Received in revised form 24 October 2010 Accepted 27 October 2010
A non-unidirectional airflow cleanroom in electronic industries is prone to be challenged by the wide spread of hot air and contaminants dissipated from process tools to surrounding area, resulting from the collision of the uprising hot air current from the tools and the downward cold air from ceilings. To effectively remove the dissipated heat and maintain the required cleanliness level, we proposed an innovative fan dry coil unit (FDCU) return air system (referring to Figs. 1 and 2), consisting of ceilingsupply grilles and ceiling-return fans/coils, and demonstrated that the FDCU-return air system can effectively eliminate sub-micron particles from the cleanroom, compared with a conventional ceilingsupply and wall-return air system [1]. This study further investigated the effect of the heat dissipation from the tools on airflow characteristics and temperature distribution in the FDCU-return and wallreturn airflow type cleanrooms. Comparisons of velocity vector, turbulence intensity, and temperature distribution between the FDCU-return air system and the conventional wall-return air system were presented. The results showed that the FDCU-return air system can significantly provide better air motion characteristics and temperature distribution in a high heat source case in comparison with the wall-return air system. Ó 2010 Elsevier Ltd. All rights reserved.
Keywords: FDCU FFU Cleanroom Non-unidirectional airflow
1. Introduction The advancement in electronic-product manufacturing technology in recent years helps to scale down the devices into the nanometer generation, and consequently increases the sensitivity of electronic-product manufacturing process to the contamination of particles into the sub-micron regions. Most of the electronic cleanrooms are herded with process tools dissipating high-temperature heat and emitting particles, which cause higher airflow resistance than the normal level and reduce the quality of the cleanrooms by reason of the accretion of heat and particles within the cleanrooms. A non-unidirectional airflow cleanroom is one of the most common systems applied in the electronic industries to control the concentration of airborne particles and the relevant temperature and relative humidity ranges. In a traditional arrangement of the airflow pathway in the cleanrooms, the fan filter units (FFUs) are installed to introduce the supply air from ceilings and the return air is extracted from the return air shafts (RASs) and wall-retunr air grilles which are close to and vertical to the floors. Despite of reducing the cost of construction by eliminating the requirement of * Corresponding author. Tel.: þ886 2 27712171x3512; fax: þ886 2 27314949. E-mail address: [email protected] (S.-C. Hu). 0360-1323/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.buildenv.2010.10.028
return air plenum, the wall-return air system exhibits several drawbacks such as (1) the requirement of high external static pressure for FFUs due to the long airflow paths to overcome the resistance of the wall-retunr air grilles, RASs and dry cooling coils (DCCs, only sensible heat exchanged in the coils), (2) the unchangeable positions of the wall-retunr air grilles and DCCs, (3) the uneven temperature distribution due to the interception of the cold air from the ceilings and the hot air from the process tools [1], and (4) the negative pressure in the supply air plenum (SAP) where un-conditioned air, particles, and moisture may be intruded. Several numerical and experimental studies on the conventional wall-return type cleanroom have been conducted to better understand the airflow patterns and contaminant diffusion characteristics in the cleanrooms, and demonstrated that contaminant particle sizes smaller than 4.5 mm in diameter are regarded as having no gravitational sedimentation effect on the diffusion [2e4]. Kato et al. [5] have proposed a locally balanced ceiling-supply/exhaust airflow rate system which allows more efficient exhaust of contaminants and less extensive diffusion of particles in the cleanrooms. No extra exhaust force is introduced on the return air in their system. Meanwhile, Lin et al. [1,6] have proposed an innovative fan dry coil unit (FDCU) return air system, containing ceiling-supply and ceiling-return air grilles, to maintain the cleanliness level within the standard
1236
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
Fig. 1. Schematic diagram of a full-scale cleanroom with the wall- return and the FDCU-return air arrangements.
requirements and effectively remove the dissipated heat from process tools. Lin et al. [1] qualitatively and quantitatively investigated the influences of ventilation arrangements, air changes per hour (ACH), and SAP pressures on the removal of 0.1 mm particles, and reported that the FDCU-return air system can effectively eliminate more than 50% of particles from the cleanrooms, compared with the conventional wall-return air system. The air change rate dominates the average concentrations of particles in the cleanrooms, while the influence of the SAP pressures on the averaged concentrations of particles may be ignored in the FDCU-return cleanrooms. The effect of increasing ACH on the particle removal in the FDCUreturn cleanroom is obviously better than that in the wall-return cleanroom. Knowledge of the air motion characteristics in the cleanroom is significantly useful in the control of the cleanliness level and gaseous contaminants. Air turbulence plays an important role in the diffusion process of gases and particles, and can even be produced at a low speed of 0.2 m/s [7]. With a 10% increase of air turbulence, the resulting total deposition of larger particles (> 0.33 mm) on horizontal and vertical surfaces is increased by 39% and 46%, respectively [8]. Kuehn [9,10] evaluated the air velocities and turbulence characteristics in a full-scale vertical laminar flow (VLF) type cleanroom by using a single-dimensional hot wire anemometer. Fujii et al. [11] reported the two-dimensional turbulence data for airflow turbulence characteristics in a clean booth. Hope and Milholland [12] evaluated the effects of perforated plates on air supply system in a VLF type cleanroom by using a three-dimensional anemometer. Hu et al. [13] employed a three-dimensional ultrasonic anemometer to evaluate the non-uniformity of the airflow under the ultra low penetration air (ULPA) filter of the FFU type cleanroom and the turbulence generated by equipment, ceiling grids, lighting fixtures, blank panels, and clean benches in the cleanroom. There have been many numerical studies on the cleanroom flow fields; however, it is noticed that the effects of heat source have been neglected in most cases. The objective of this study was to
experimentally investigate the effects of the process tools dissipating high-temperature heat in the FDCU-return and wall-return airflow type industrial cleanrooms on the temperature distribution and the air motion characteristics, such as velocity vector and turbulence intensity. 2. Experimental setup Fig. 1 shows a schematic diagram of a full-scale cleanroom, designed at 23 0.5 C and 45% of relative humidity (RH), with dimensions of 4.8 m (length, L) 6.3 m (width, W) 2.8 m (height, H) in the X, Y, and Z directions. Two dummy tools, each size of 1.6 m (length) 1.2 m (width) 2.4 m (height), and a process tool sized of 0.9 m (length) 0.6 m (width) 1 m (height) were arranged in the cleanroom. In order to examine the collision of the uprising hot air current from the process tool and the downward cold air from the FFUs, the process tool was located under one of the FFUs, and the coordinate of the center of the process tool in the X and Y directions was X 2.4 m (X/L 0.5) and Y 1.8 m (Y/W 0.29), respectively. The supply air grilles consisted of 12 FFUs (each size of 1.2mlength 0.6-width 0.275m-height). Moreover, ULPA filters were installed with the FFUs. Fig. 2 shows a schematic diagram of an FDCU sized of 1.17m-length 0.57m-width 0.45m-height. Three sets of FDCUs were arranged parallel with the FFUs on the ceiling.
Fig. 2. Schematic diagram of an FDCU.
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
1237
Fig. 3. Top views of the cleanroom. Left side: sections A and B were examined when the process tool was turned off. Right side: sections C and D were examined when the process tool was turned on.
For both sidewall RASs, three units of sidewall return- air grills (RAGs) and two units of DCCs were respectively installed. In the wall-return air arrangement, all FDCUs in Fig. 1 were tightly sealed by blind plates and turned off; the room air was supplied from the SAP through the FFUs into the cleanroom, and then returned from RAGs through RASs and DCCs into the SAP, while the returned air was cooled by the DCCs. As for the FDCU-return airfow system, all
RAGs in Fig. 1 were tightly sealed by blind plates; the supply air was introduced from the SAP via the FFUs, while the indoor air was returned through the FDCUs. Returned air was cooled by dry coils inside the FDCUs, instead of DDCs, and then forwarded to the SAP. To evaluate the airflow characteristics and temperature distribution, sections A, B, C, and D with measured points, each other apart with a distance of 30 cm, were positioned on the x-z and y-z
Fig. 4. Measured points on the x-z and y-z planes.
1238
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
Table 1 Uncertainties of air velocity and temperature measurements. Measurement uncertainty
Air velocity (m/s)
Air temperature ( C)
Standard uncertainty, SX Resolution uncertainty, Zd Calibration uncertainty, Zs Combined standard uncertainty, Zc
0.0206 0.00144 0.012 0.024
0.001 0.0288 0.2 0.202
planes as shown in Figs. 3 and 4. T-type thermocouples (ISHIKAWA) with temperature ranges of 0e350 C and a tolerance of 0.75% of reading value were used to measure the temperature distribution. A three-dimensional ultrasonic anemometer (KAIJO, WA-590) with velocity ranges of 0e10 m/s, a sampling rate of 140 Hz, and accuracy of 2% was applied to measure the air velocity vector. The sensors of the three-dimensional ultrasonic anemometer consist of three pairs of transmitters and receiver probe heads spaced 5 cm apart. The present experiments applied the quantitative data in describing the velocity vector and turbulence intensity, and all instruments were calibrated before conducting experiments.
2.1. Measurement uncertainty
however, the true values are seldom identified. Therefore, the experimental data are denoted by the measured values and their uncertainties, estimated values for the errors. Because the measured values of a quantity generally show some variability, a standard deviation, SX , is used as a measure of this variability as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PN 2 k ¼ 1 ðXk XÞ Sx ¼ N1
where Xk , k 1, 2,.N (If an experiment is repeated N times) are denoted the individual measured values, and X the mean of the individual measured values. When the mean, X, is the best estimate determined from the repeated measurements, then a standard uncertainty, SX , is written as follows:
S SX ¼ pXffiffiffiffi N
(2)
Moreover, the uncertainty from the resolution of instruments, Zd , is calculated as follows:
Zd ¼
Experimental data are generally represented by measured values and their errors, the measured values minus the true values;
(1)
1 d pffiffiffi 2 3
where d is the resolution of the instruments.
Fig. 5. Velocity vector and turbulence intensity on the x-z plane of Y ¼ 3 m (Section A, Y/W ¼ 0.48).
(3)
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
1239
Fig. 6. Velocity vector and turbulence intensity on the y-z plane of X ¼ 2.4 m (Section B, X/L ¼ 0.5).
A combined standard uncertainty, Zc , is calculated by taking the square root of the sum of the variances as follows:
Zc ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi S2 þ Zd2 þ Zs2 X
(4)
where Zs is the uncertainty from calibration certificates of instruments. Table 1 tabulates the measurement uncertainties in the present study.
TI ¼
V
(5)
while (sX =V), (sY =V), and (sZ =V) are defined as the turbulence intensities along X, Y, and Z directions, respectively. In Eq. (5),
PN Vi ¼
2.2. Turbulence intensity Turbulence intensity (TI) of airflow is a very efficient parameter describing the degree of air turbulence in the flow fields and useful in comparing the level of flow disturbance with respect to the local mean velocity. In the study, the resultant turbulence intensity at any locations is defined as follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 s2 þ s2 þ s2 Y Z 3 X
V ¼
S¼1 Vi
N qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi V x2 þ V y2 þ V z2
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PN 2 S¼1 ðVi ’Þ si ¼ N
(6)
(7)
(8)
1240
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
Fig. 7. Velocity vector and turbulence intensity on the x-z plane of Y ¼ 1.8 m (Section C, Y/W ¼ 0.29).
where V i is the mean velocity along any direction i, m/s; Vi the instantaneous velocity along i direction, m/s; N the total sampling number; s the sampling data number; V the resultant velocity, m/s; si the standard deviation of Vi, m/s; and Vi ’ the fluctuating velocity along i direction, m/s. The magnitude of turbulence intensity was denoted by the diameter of a circle, and a larger circle represented higher turbulence intensity.
2.3. Local air temperature index This study employed a local air temperature index to describe the temperature influence of hot plume generated from the process tool on the room air at any locations where the local air temperature was measured. The local air temperature index, q, is defined as follows:
Fig. 8. Temperature distribution on the x-z plane of Y ¼ 1.8 m (Section C, Y/W ¼ 0.29).
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
1241
Fig. 9. Temperature distribution on the y-z plane of X ¼ 2.4 m (Section D, X/L ¼ 0.5).
q ¼
Tlocal Tin Treturn Tin
(9)
where Tin, Treturn, and Tlocal refer to the temperature of supply air, return air, and room air at any locations ( C), respectively. The q > 1 (i.e., Tlocal Treturn) indicates the spread of hot air in the cleanroom, whereas the q < 1 (i.e., Tlocal Treturn) indicates the extraction of most hot air from the cleanroom. The q value was used to represent the temperature gradient in the cleanroom, and implicitly provided information on the development of the hot plume dissipated from the process tool. 3. Results and discussion The measured air velocity and TI were normalized by the supply air velocity of the FFU and the maximum TI, TImax 1, respectively. The supply air velocity and temperature of the FFUs in the wall-return air system were measured at 0.36 m/s and 23.1 C, respectively, while the ones in the FDCU-return air system were measured at 0.38 m/s and 22.6 C, respectively. This supply air velocity of FFUs corresponded to the supply airflow rates of 120 and 130 ACH in the the wall-return and FDCU-return air systems, respectively. The pressure in the SAP in the wall-return and the FDCU-return air systems was measured at 20 Pa and 0 Pa, respectively. When the process tool was turned on, the temperature and velocity of supply air from the tool (2.6 kW of heat dissipation) were measured at 45 C and 1.2 m/s, respectively; then the temperature of return air in the wall-return and the FDCUreturn air systems was measured at 25.6 C and 24.6 C, respectively. For hospital operation rooms, the recommended air exchange rates are generally 20 to 25, which are mainly to provide required cleanliness level and thermal comfort conditions in the vicinity of an operation table. As to the industrial cleanrooms, the nonunidirectional airflow cleanrooms of FS cleanliness class 1000 (ISO cleanliness class 6) is likely to have supply airflow rates ranged from 120 to 150 ACH. This additional air supply is mainly provided to extract any contaminants, compensate the large amount of process exhaust gas, and cool the heat dissipated from the tools for maintaining the required cleanliness level, positive pressure, and design temperature and humidity, respectivey. 3.1. Process tool turned off Figs. 5 and 6 show the distribution of velocity vector and TI in the wall-return and FDCU-return air systems on the x-z and the y-z
planes, while the process tool was at rest. In the wall-return air system, the satisfactory uniform downstream airflow pattern was observed underneath the FFUs in Fig. 5(a) and Fig. 6(a). Then, the TI was slightly increased as approaching towards the wall-return air grills. It was noted that the TI in the area of Y/W 0.4e0.7 in Fig. 6(b) reached to 0.4, and could cause particles to accumulate in this region. Above a height of Z/H 0.42 in Fig. 5(b) and Fig. 6(b), the averaged TI reached to 0.34. Examining the measured velocity vector in the FDCU-return air system, one could observe that a vortex under the right FFU was formed at about Z/H 0.5 in Fig. 5(c). The turbulence region resulting from the recirculation zone extended up to Z/H 0.6. The size of the recirculation zone should be minimized in the cleanroom design because it could cause accumulation of contaminant particles. Based on Fig. 5(d), the least TI of 0.1 was read underneath the FFUs and the averaged TI of the measured x-z plane was 0.33, while a maximum TI of 0.59 was found inside the recirculation flow region downstream of the FFU. In addition, it was found that the TI gradually increased in the distance, even though the uniform downstream airflow was blown from the FFUs. Above the height of Z/H 0.42 in Fig. 5(d), the averaged TI was 0.29. As shown in Fig. 5(c) and Fig. 6(c), it was noted that the airflow in the FDCU-return air system moved from the supply air grilles of the FFUs towards the return air grilles of the FDCUs above the height of Z/H 0.42. By assuming Z/H 0.42 as the height of working zone, it could be an effective air-cleaning strategy for extracting particles and heat, generated from the working zone. Such characteristics of air motion were induced by the FDCUs and could efficiently remove hot air and fine particles from the working zones of high heat dissipation and particles generation. 3.2. Process tool turned on Fig. 7 shows the distribution of velocity vector and TI in the wallreturn and FDCU-return air systems on the x-z plane of Y 1.8 m, while the process tool was operated. The downward airflow from the FFUs uniformly moved towards the floor in the wall-return air system as shown in Fig. 7(a), whereas the one inclined towards the process tool in the FDCU-return air system as shown in Fig. 7(c). In Fig. 7(a) and (c), the downward airflow was expelled from the FFUs and affected by the upward airflow of 1.2 m/s dissipated from the process tool. Significant lugged recirculation airflow was formed at the both sides above the process tool in the wall-return and FDCUreturn air systems, and the TI of 0.48 and 0.3 occurred in the
1242
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
recirculation zones in Fig. 7(b) and Fig. 7(d), respectively. The colliding of the downward airflow from the central FFU with the upward high-velocity plume caused the increase in the TI with the reduction of the plume velocity from Z/H 0.4 to Z/H 0.9 at X/ L 0.4e0.6. Above the working zone (Z/H 0.42) in Fig. 7(b) and Fig. 7 (d), the averaged TI in the FDCU-return and the wall-return air systems was 0.27 and 0.3, respectively; hence, the FDCU-return air system exhibited better airflow characteristics than the wall-return air system. Figs. 8 and 9 show the local air temperature index on the x-z and y-z planes. Higher q values were found above the process tool, X/ L 0.4e0.6 and Y/W 0.45e0.55, and reduced with a distance out of the process tool because of the cooling of supply air from the central FFU. In these regions above the process tool, the hot plume had higher airflow velocity than room air, and thus formed forced convection to enlarge the temperature gradient. The colliding of the downward airflow from the central FFU with the upward highvelocity plume caused the hot plume to spread along the ceiling; the q values, hence, were more than 1. As revealed in Fig. 8(a) and Fig. 8(b), the q values in the FDCU-return and wall-return air systems were insignificantly different, and the q values in most regions ranged from 1 to 1.5. The q ¼ 1 indicated that Tlocal and Treturn were almost at same temperature and the hot air dissipated from the process tool has been well-mixed with the cold supply air in the cleanroom before it was exhausted. On the contrary, Fig. 9(b) shows that the both side FDCUs effectively limited the spread of hot plume to Y/W 0.15e0.45, and revealed smaller q values than Fig. 9 (a); moreover, the regions in the both sides of the process tool in Fig. 9(b) had large-scale q values smaller than 1, i.e., no wide spread of hot air to these regions. 4. Conclusions According to the definition of Airborne Particulate Cleanliness Classes in ISO Standard 14644-1 (1999), air cleanliness class in industrial cleanrooms is associated with the concentration of airborne particles along with diameter of the particles of concern. The air cleanliness class of cleanrooms in the present study is ISO cleanliness class 6. This study experimentally employs two indexes of turbulence intensity and local air temperature to quantitatively examine the airflow characteristics and temperature distribution in the industrial cleanrooms, herded with process tools dissipating
high-temperature heat and emitting particles. The variations in the turbulence intensity are in agreement with those in the airflow streamlines; the recirculation and votex airflow result in bigger turbulence intensity, and where the high-velocity air is, e.g. the supply air outlets, results in smaller turbulence intensity. No matter how the process tool operates, i.e., turned on or off, the FDCUreturn air system has a lower turbulence intensity above the working zone than the wall-return air system. Moreover, when the process is turned on, the hot plume spreads and raises the room air temperature, but the FDCU-return air system effectively limits the affected regions. References [1] Lin T, Hu SC, Lin CY, Chang A, Hwang J. An innovative ventilation system for cleanrooms with high cooling loads. ASHRAE Transactions 2010;116 (pt 1):293e7. [2] Murakami S, Kato S, Suyama Y. Numerical and experimental study on turbulent diffusion fields in conventional-flow-type clean rooms. ASHRAE Transactions 1988;94(2):469e93. [3] Murakami S, Kato S, Suyama Y. Numerical study on diffusion field as affected by arrangement of supply and exhaust openings in conventional flow type clean room. ASHRAE Transactions 1989;95(2):113e27. [4] Murakami S, Kato S, Nagano S, Tanaka Y. Diffusion characteristics of airborne particles with gravitational settling in a conventional-dominant indoor flow field. ASHRAE Transactions: Research 1992;98:82e97. [5] Kato S, Murakami S, Nagano S. Numerical study on diffusion in a room with a locally balanced supply-exhaust airflow rate system. ASHRAE Transactions 1992;98(pt 1):218e38. [6] Lin T, Tung YC, Hu SC, Lin CY. Effects of the removal of 0.1 mm particles in industrial cleanrooms with fan dry coil unit (FDCU) return systems. Aerosol and Air Quality Research 2010;10(6):571e80. [7] Ljungqvist B. Some observations on the interaction between air movements and the dispersion of pollution. Stockholm, Sweden: Document D8: Swedish Council for Building Research; 1979. [8] Milberg J, Fischbacher J, Engel A. Fluid integration of equipment in clean rooms. Solid State Technology 1991;34(8):43e9. [9] Kuehn TH. Computer simulation of air flow and particle transport in clean rooms. Journal of Environment Science 1988;XXXI(September/October):21e7. [10] Kuehn TH, Marple VA. Comparison of measured and predicted airflow patterns in a clean room. In: 34th annual technical meeting. King of Prussia, PA: Proceeding of Institute of Environment Sciences; 3e5 May, 1988. p. 331e6. [11] Fujii S, Yuasa K, Arai Y, Watanabe T. Characteristics of airflow turbulence behind HEPA filter. In: 40th annual technical meeting. Chicago, IL: Proceeding of Institute of Environmental Sciences; 1e6 May, 1994. p. 344e9. [12] Hope D, Milholland D. The use of a three dimensional ultrasonic anemometer to measure to measure the performance of clean zone air delivery systems. In: 39th annual technical meeting. Las Vegas, NV: Proceeding of Institute of Environmental Sciences; 2e7 May, 1993. p. 516e28. [13] Hu SC, Wu YY, Liu CJ. Measurements of air flow characteristics in a full-scale clean room. Building and Environment 1996;31(2):119e28.
Contents lists available at ScienceDirect
Building and Environment journal homepage: www.elsevier.com/locate/buildenv
Experimental study on airflow characteristics and temperature distribution in non-unidirectional cleanrooms for electronic industry Ti Lin b, Yun-Chun Tung a, Shih-Cheng Hu b, *, Yen-Jhih Chen b a b
Department of Industrial Education, National Taiwan Normal University, Taiwan, ROC Department of Energy and Refrigerating Air-Conditioning Engineering, National Taipei University of Technology, 1, Sec. 3, Chung-Hsiao E. Rd., Taipei 106, Taiwan, ROC
a r t i c l e i n f o
a b s t r a c t
Article history: Received 10 July 2010 Received in revised form 24 October 2010 Accepted 27 October 2010
A non-unidirectional airflow cleanroom in electronic industries is prone to be challenged by the wide spread of hot air and contaminants dissipated from process tools to surrounding area, resulting from the collision of the uprising hot air current from the tools and the downward cold air from ceilings. To effectively remove the dissipated heat and maintain the required cleanliness level, we proposed an innovative fan dry coil unit (FDCU) return air system (referring to Figs. 1 and 2), consisting of ceilingsupply grilles and ceiling-return fans/coils, and demonstrated that the FDCU-return air system can effectively eliminate sub-micron particles from the cleanroom, compared with a conventional ceilingsupply and wall-return air system [1]. This study further investigated the effect of the heat dissipation from the tools on airflow characteristics and temperature distribution in the FDCU-return and wallreturn airflow type cleanrooms. Comparisons of velocity vector, turbulence intensity, and temperature distribution between the FDCU-return air system and the conventional wall-return air system were presented. The results showed that the FDCU-return air system can significantly provide better air motion characteristics and temperature distribution in a high heat source case in comparison with the wall-return air system. Ó 2010 Elsevier Ltd. All rights reserved.
Keywords: FDCU FFU Cleanroom Non-unidirectional airflow
1. Introduction The advancement in electronic-product manufacturing technology in recent years helps to scale down the devices into the nanometer generation, and consequently increases the sensitivity of electronic-product manufacturing process to the contamination of particles into the sub-micron regions. Most of the electronic cleanrooms are herded with process tools dissipating high-temperature heat and emitting particles, which cause higher airflow resistance than the normal level and reduce the quality of the cleanrooms by reason of the accretion of heat and particles within the cleanrooms. A non-unidirectional airflow cleanroom is one of the most common systems applied in the electronic industries to control the concentration of airborne particles and the relevant temperature and relative humidity ranges. In a traditional arrangement of the airflow pathway in the cleanrooms, the fan filter units (FFUs) are installed to introduce the supply air from ceilings and the return air is extracted from the return air shafts (RASs) and wall-retunr air grilles which are close to and vertical to the floors. Despite of reducing the cost of construction by eliminating the requirement of * Corresponding author. Tel.: þ886 2 27712171x3512; fax: þ886 2 27314949. E-mail address: [email protected] (S.-C. Hu). 0360-1323/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.buildenv.2010.10.028
return air plenum, the wall-return air system exhibits several drawbacks such as (1) the requirement of high external static pressure for FFUs due to the long airflow paths to overcome the resistance of the wall-retunr air grilles, RASs and dry cooling coils (DCCs, only sensible heat exchanged in the coils), (2) the unchangeable positions of the wall-retunr air grilles and DCCs, (3) the uneven temperature distribution due to the interception of the cold air from the ceilings and the hot air from the process tools [1], and (4) the negative pressure in the supply air plenum (SAP) where un-conditioned air, particles, and moisture may be intruded. Several numerical and experimental studies on the conventional wall-return type cleanroom have been conducted to better understand the airflow patterns and contaminant diffusion characteristics in the cleanrooms, and demonstrated that contaminant particle sizes smaller than 4.5 mm in diameter are regarded as having no gravitational sedimentation effect on the diffusion [2e4]. Kato et al. [5] have proposed a locally balanced ceiling-supply/exhaust airflow rate system which allows more efficient exhaust of contaminants and less extensive diffusion of particles in the cleanrooms. No extra exhaust force is introduced on the return air in their system. Meanwhile, Lin et al. [1,6] have proposed an innovative fan dry coil unit (FDCU) return air system, containing ceiling-supply and ceiling-return air grilles, to maintain the cleanliness level within the standard
1236
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
Fig. 1. Schematic diagram of a full-scale cleanroom with the wall- return and the FDCU-return air arrangements.
requirements and effectively remove the dissipated heat from process tools. Lin et al. [1] qualitatively and quantitatively investigated the influences of ventilation arrangements, air changes per hour (ACH), and SAP pressures on the removal of 0.1 mm particles, and reported that the FDCU-return air system can effectively eliminate more than 50% of particles from the cleanrooms, compared with the conventional wall-return air system. The air change rate dominates the average concentrations of particles in the cleanrooms, while the influence of the SAP pressures on the averaged concentrations of particles may be ignored in the FDCU-return cleanrooms. The effect of increasing ACH on the particle removal in the FDCUreturn cleanroom is obviously better than that in the wall-return cleanroom. Knowledge of the air motion characteristics in the cleanroom is significantly useful in the control of the cleanliness level and gaseous contaminants. Air turbulence plays an important role in the diffusion process of gases and particles, and can even be produced at a low speed of 0.2 m/s [7]. With a 10% increase of air turbulence, the resulting total deposition of larger particles (> 0.33 mm) on horizontal and vertical surfaces is increased by 39% and 46%, respectively [8]. Kuehn [9,10] evaluated the air velocities and turbulence characteristics in a full-scale vertical laminar flow (VLF) type cleanroom by using a single-dimensional hot wire anemometer. Fujii et al. [11] reported the two-dimensional turbulence data for airflow turbulence characteristics in a clean booth. Hope and Milholland [12] evaluated the effects of perforated plates on air supply system in a VLF type cleanroom by using a three-dimensional anemometer. Hu et al. [13] employed a three-dimensional ultrasonic anemometer to evaluate the non-uniformity of the airflow under the ultra low penetration air (ULPA) filter of the FFU type cleanroom and the turbulence generated by equipment, ceiling grids, lighting fixtures, blank panels, and clean benches in the cleanroom. There have been many numerical studies on the cleanroom flow fields; however, it is noticed that the effects of heat source have been neglected in most cases. The objective of this study was to
experimentally investigate the effects of the process tools dissipating high-temperature heat in the FDCU-return and wall-return airflow type industrial cleanrooms on the temperature distribution and the air motion characteristics, such as velocity vector and turbulence intensity. 2. Experimental setup Fig. 1 shows a schematic diagram of a full-scale cleanroom, designed at 23 0.5 C and 45% of relative humidity (RH), with dimensions of 4.8 m (length, L) 6.3 m (width, W) 2.8 m (height, H) in the X, Y, and Z directions. Two dummy tools, each size of 1.6 m (length) 1.2 m (width) 2.4 m (height), and a process tool sized of 0.9 m (length) 0.6 m (width) 1 m (height) were arranged in the cleanroom. In order to examine the collision of the uprising hot air current from the process tool and the downward cold air from the FFUs, the process tool was located under one of the FFUs, and the coordinate of the center of the process tool in the X and Y directions was X 2.4 m (X/L 0.5) and Y 1.8 m (Y/W 0.29), respectively. The supply air grilles consisted of 12 FFUs (each size of 1.2mlength 0.6-width 0.275m-height). Moreover, ULPA filters were installed with the FFUs. Fig. 2 shows a schematic diagram of an FDCU sized of 1.17m-length 0.57m-width 0.45m-height. Three sets of FDCUs were arranged parallel with the FFUs on the ceiling.
Fig. 2. Schematic diagram of an FDCU.
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
1237
Fig. 3. Top views of the cleanroom. Left side: sections A and B were examined when the process tool was turned off. Right side: sections C and D were examined when the process tool was turned on.
For both sidewall RASs, three units of sidewall return- air grills (RAGs) and two units of DCCs were respectively installed. In the wall-return air arrangement, all FDCUs in Fig. 1 were tightly sealed by blind plates and turned off; the room air was supplied from the SAP through the FFUs into the cleanroom, and then returned from RAGs through RASs and DCCs into the SAP, while the returned air was cooled by the DCCs. As for the FDCU-return airfow system, all
RAGs in Fig. 1 were tightly sealed by blind plates; the supply air was introduced from the SAP via the FFUs, while the indoor air was returned through the FDCUs. Returned air was cooled by dry coils inside the FDCUs, instead of DDCs, and then forwarded to the SAP. To evaluate the airflow characteristics and temperature distribution, sections A, B, C, and D with measured points, each other apart with a distance of 30 cm, were positioned on the x-z and y-z
Fig. 4. Measured points on the x-z and y-z planes.
1238
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
Table 1 Uncertainties of air velocity and temperature measurements. Measurement uncertainty
Air velocity (m/s)
Air temperature ( C)
Standard uncertainty, SX Resolution uncertainty, Zd Calibration uncertainty, Zs Combined standard uncertainty, Zc
0.0206 0.00144 0.012 0.024
0.001 0.0288 0.2 0.202
planes as shown in Figs. 3 and 4. T-type thermocouples (ISHIKAWA) with temperature ranges of 0e350 C and a tolerance of 0.75% of reading value were used to measure the temperature distribution. A three-dimensional ultrasonic anemometer (KAIJO, WA-590) with velocity ranges of 0e10 m/s, a sampling rate of 140 Hz, and accuracy of 2% was applied to measure the air velocity vector. The sensors of the three-dimensional ultrasonic anemometer consist of three pairs of transmitters and receiver probe heads spaced 5 cm apart. The present experiments applied the quantitative data in describing the velocity vector and turbulence intensity, and all instruments were calibrated before conducting experiments.
2.1. Measurement uncertainty
however, the true values are seldom identified. Therefore, the experimental data are denoted by the measured values and their uncertainties, estimated values for the errors. Because the measured values of a quantity generally show some variability, a standard deviation, SX , is used as a measure of this variability as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PN 2 k ¼ 1 ðXk XÞ Sx ¼ N1
where Xk , k 1, 2,.N (If an experiment is repeated N times) are denoted the individual measured values, and X the mean of the individual measured values. When the mean, X, is the best estimate determined from the repeated measurements, then a standard uncertainty, SX , is written as follows:
S SX ¼ pXffiffiffiffi N
(2)
Moreover, the uncertainty from the resolution of instruments, Zd , is calculated as follows:
Zd ¼
Experimental data are generally represented by measured values and their errors, the measured values minus the true values;
(1)
1 d pffiffiffi 2 3
where d is the resolution of the instruments.
Fig. 5. Velocity vector and turbulence intensity on the x-z plane of Y ¼ 3 m (Section A, Y/W ¼ 0.48).
(3)
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
1239
Fig. 6. Velocity vector and turbulence intensity on the y-z plane of X ¼ 2.4 m (Section B, X/L ¼ 0.5).
A combined standard uncertainty, Zc , is calculated by taking the square root of the sum of the variances as follows:
Zc ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi S2 þ Zd2 þ Zs2 X
(4)
where Zs is the uncertainty from calibration certificates of instruments. Table 1 tabulates the measurement uncertainties in the present study.
TI ¼
V
(5)
while (sX =V), (sY =V), and (sZ =V) are defined as the turbulence intensities along X, Y, and Z directions, respectively. In Eq. (5),
PN Vi ¼
2.2. Turbulence intensity Turbulence intensity (TI) of airflow is a very efficient parameter describing the degree of air turbulence in the flow fields and useful in comparing the level of flow disturbance with respect to the local mean velocity. In the study, the resultant turbulence intensity at any locations is defined as follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 s2 þ s2 þ s2 Y Z 3 X
V ¼
S¼1 Vi
N qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi V x2 þ V y2 þ V z2
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PN 2 S¼1 ðVi ’Þ si ¼ N
(6)
(7)
(8)
1240
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
Fig. 7. Velocity vector and turbulence intensity on the x-z plane of Y ¼ 1.8 m (Section C, Y/W ¼ 0.29).
where V i is the mean velocity along any direction i, m/s; Vi the instantaneous velocity along i direction, m/s; N the total sampling number; s the sampling data number; V the resultant velocity, m/s; si the standard deviation of Vi, m/s; and Vi ’ the fluctuating velocity along i direction, m/s. The magnitude of turbulence intensity was denoted by the diameter of a circle, and a larger circle represented higher turbulence intensity.
2.3. Local air temperature index This study employed a local air temperature index to describe the temperature influence of hot plume generated from the process tool on the room air at any locations where the local air temperature was measured. The local air temperature index, q, is defined as follows:
Fig. 8. Temperature distribution on the x-z plane of Y ¼ 1.8 m (Section C, Y/W ¼ 0.29).
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
1241
Fig. 9. Temperature distribution on the y-z plane of X ¼ 2.4 m (Section D, X/L ¼ 0.5).
q ¼
Tlocal Tin Treturn Tin
(9)
where Tin, Treturn, and Tlocal refer to the temperature of supply air, return air, and room air at any locations ( C), respectively. The q > 1 (i.e., Tlocal Treturn) indicates the spread of hot air in the cleanroom, whereas the q < 1 (i.e., Tlocal Treturn) indicates the extraction of most hot air from the cleanroom. The q value was used to represent the temperature gradient in the cleanroom, and implicitly provided information on the development of the hot plume dissipated from the process tool. 3. Results and discussion The measured air velocity and TI were normalized by the supply air velocity of the FFU and the maximum TI, TImax 1, respectively. The supply air velocity and temperature of the FFUs in the wall-return air system were measured at 0.36 m/s and 23.1 C, respectively, while the ones in the FDCU-return air system were measured at 0.38 m/s and 22.6 C, respectively. This supply air velocity of FFUs corresponded to the supply airflow rates of 120 and 130 ACH in the the wall-return and FDCU-return air systems, respectively. The pressure in the SAP in the wall-return and the FDCU-return air systems was measured at 20 Pa and 0 Pa, respectively. When the process tool was turned on, the temperature and velocity of supply air from the tool (2.6 kW of heat dissipation) were measured at 45 C and 1.2 m/s, respectively; then the temperature of return air in the wall-return and the FDCUreturn air systems was measured at 25.6 C and 24.6 C, respectively. For hospital operation rooms, the recommended air exchange rates are generally 20 to 25, which are mainly to provide required cleanliness level and thermal comfort conditions in the vicinity of an operation table. As to the industrial cleanrooms, the nonunidirectional airflow cleanrooms of FS cleanliness class 1000 (ISO cleanliness class 6) is likely to have supply airflow rates ranged from 120 to 150 ACH. This additional air supply is mainly provided to extract any contaminants, compensate the large amount of process exhaust gas, and cool the heat dissipated from the tools for maintaining the required cleanliness level, positive pressure, and design temperature and humidity, respectivey. 3.1. Process tool turned off Figs. 5 and 6 show the distribution of velocity vector and TI in the wall-return and FDCU-return air systems on the x-z and the y-z
planes, while the process tool was at rest. In the wall-return air system, the satisfactory uniform downstream airflow pattern was observed underneath the FFUs in Fig. 5(a) and Fig. 6(a). Then, the TI was slightly increased as approaching towards the wall-return air grills. It was noted that the TI in the area of Y/W 0.4e0.7 in Fig. 6(b) reached to 0.4, and could cause particles to accumulate in this region. Above a height of Z/H 0.42 in Fig. 5(b) and Fig. 6(b), the averaged TI reached to 0.34. Examining the measured velocity vector in the FDCU-return air system, one could observe that a vortex under the right FFU was formed at about Z/H 0.5 in Fig. 5(c). The turbulence region resulting from the recirculation zone extended up to Z/H 0.6. The size of the recirculation zone should be minimized in the cleanroom design because it could cause accumulation of contaminant particles. Based on Fig. 5(d), the least TI of 0.1 was read underneath the FFUs and the averaged TI of the measured x-z plane was 0.33, while a maximum TI of 0.59 was found inside the recirculation flow region downstream of the FFU. In addition, it was found that the TI gradually increased in the distance, even though the uniform downstream airflow was blown from the FFUs. Above the height of Z/H 0.42 in Fig. 5(d), the averaged TI was 0.29. As shown in Fig. 5(c) and Fig. 6(c), it was noted that the airflow in the FDCU-return air system moved from the supply air grilles of the FFUs towards the return air grilles of the FDCUs above the height of Z/H 0.42. By assuming Z/H 0.42 as the height of working zone, it could be an effective air-cleaning strategy for extracting particles and heat, generated from the working zone. Such characteristics of air motion were induced by the FDCUs and could efficiently remove hot air and fine particles from the working zones of high heat dissipation and particles generation. 3.2. Process tool turned on Fig. 7 shows the distribution of velocity vector and TI in the wallreturn and FDCU-return air systems on the x-z plane of Y 1.8 m, while the process tool was operated. The downward airflow from the FFUs uniformly moved towards the floor in the wall-return air system as shown in Fig. 7(a), whereas the one inclined towards the process tool in the FDCU-return air system as shown in Fig. 7(c). In Fig. 7(a) and (c), the downward airflow was expelled from the FFUs and affected by the upward airflow of 1.2 m/s dissipated from the process tool. Significant lugged recirculation airflow was formed at the both sides above the process tool in the wall-return and FDCUreturn air systems, and the TI of 0.48 and 0.3 occurred in the
1242
T. Lin et al. / Building and Environment 46 (2011) 1235e1242
recirculation zones in Fig. 7(b) and Fig. 7(d), respectively. The colliding of the downward airflow from the central FFU with the upward high-velocity plume caused the increase in the TI with the reduction of the plume velocity from Z/H 0.4 to Z/H 0.9 at X/ L 0.4e0.6. Above the working zone (Z/H 0.42) in Fig. 7(b) and Fig. 7 (d), the averaged TI in the FDCU-return and the wall-return air systems was 0.27 and 0.3, respectively; hence, the FDCU-return air system exhibited better airflow characteristics than the wall-return air system. Figs. 8 and 9 show the local air temperature index on the x-z and y-z planes. Higher q values were found above the process tool, X/ L 0.4e0.6 and Y/W 0.45e0.55, and reduced with a distance out of the process tool because of the cooling of supply air from the central FFU. In these regions above the process tool, the hot plume had higher airflow velocity than room air, and thus formed forced convection to enlarge the temperature gradient. The colliding of the downward airflow from the central FFU with the upward highvelocity plume caused the hot plume to spread along the ceiling; the q values, hence, were more than 1. As revealed in Fig. 8(a) and Fig. 8(b), the q values in the FDCU-return and wall-return air systems were insignificantly different, and the q values in most regions ranged from 1 to 1.5. The q ¼ 1 indicated that Tlocal and Treturn were almost at same temperature and the hot air dissipated from the process tool has been well-mixed with the cold supply air in the cleanroom before it was exhausted. On the contrary, Fig. 9(b) shows that the both side FDCUs effectively limited the spread of hot plume to Y/W 0.15e0.45, and revealed smaller q values than Fig. 9 (a); moreover, the regions in the both sides of the process tool in Fig. 9(b) had large-scale q values smaller than 1, i.e., no wide spread of hot air to these regions. 4. Conclusions According to the definition of Airborne Particulate Cleanliness Classes in ISO Standard 14644-1 (1999), air cleanliness class in industrial cleanrooms is associated with the concentration of airborne particles along with diameter of the particles of concern. The air cleanliness class of cleanrooms in the present study is ISO cleanliness class 6. This study experimentally employs two indexes of turbulence intensity and local air temperature to quantitatively examine the airflow characteristics and temperature distribution in the industrial cleanrooms, herded with process tools dissipating
high-temperature heat and emitting particles. The variations in the turbulence intensity are in agreement with those in the airflow streamlines; the recirculation and votex airflow result in bigger turbulence intensity, and where the high-velocity air is, e.g. the supply air outlets, results in smaller turbulence intensity. No matter how the process tool operates, i.e., turned on or off, the FDCUreturn air system has a lower turbulence intensity above the working zone than the wall-return air system. Moreover, when the process is turned on, the hot plume spreads and raises the room air temperature, but the FDCU-return air system effectively limits the affected regions. References [1] Lin T, Hu SC, Lin CY, Chang A, Hwang J. An innovative ventilation system for cleanrooms with high cooling loads. ASHRAE Transactions 2010;116 (pt 1):293e7. [2] Murakami S, Kato S, Suyama Y. Numerical and experimental study on turbulent diffusion fields in conventional-flow-type clean rooms. ASHRAE Transactions 1988;94(2):469e93. [3] Murakami S, Kato S, Suyama Y. Numerical study on diffusion field as affected by arrangement of supply and exhaust openings in conventional flow type clean room. ASHRAE Transactions 1989;95(2):113e27. [4] Murakami S, Kato S, Nagano S, Tanaka Y. Diffusion characteristics of airborne particles with gravitational settling in a conventional-dominant indoor flow field. ASHRAE Transactions: Research 1992;98:82e97. [5] Kato S, Murakami S, Nagano S. Numerical study on diffusion in a room with a locally balanced supply-exhaust airflow rate system. ASHRAE Transactions 1992;98(pt 1):218e38. [6] Lin T, Tung YC, Hu SC, Lin CY. Effects of the removal of 0.1 mm particles in industrial cleanrooms with fan dry coil unit (FDCU) return systems. Aerosol and Air Quality Research 2010;10(6):571e80. [7] Ljungqvist B. Some observations on the interaction between air movements and the dispersion of pollution. Stockholm, Sweden: Document D8: Swedish Council for Building Research; 1979. [8] Milberg J, Fischbacher J, Engel A. Fluid integration of equipment in clean rooms. Solid State Technology 1991;34(8):43e9. [9] Kuehn TH. Computer simulation of air flow and particle transport in clean rooms. Journal of Environment Science 1988;XXXI(September/October):21e7. [10] Kuehn TH, Marple VA. Comparison of measured and predicted airflow patterns in a clean room. In: 34th annual technical meeting. King of Prussia, PA: Proceeding of Institute of Environment Sciences; 3e5 May, 1988. p. 331e6. [11] Fujii S, Yuasa K, Arai Y, Watanabe T. Characteristics of airflow turbulence behind HEPA filter. In: 40th annual technical meeting. Chicago, IL: Proceeding of Institute of Environmental Sciences; 1e6 May, 1994. p. 344e9. [12] Hope D, Milholland D. The use of a three dimensional ultrasonic anemometer to measure to measure the performance of clean zone air delivery systems. In: 39th annual technical meeting. Las Vegas, NV: Proceeding of Institute of Environmental Sciences; 2e7 May, 1993. p. 516e28. [13] Hu SC, Wu YY, Liu CJ. Measurements of air flow characteristics in a full-scale clean room. Building and Environment 1996;31(2):119e28.