by Raj Jaisinghani
As industries and markets grow more competitive, and as the unarguable edict of energy conservation becomes widely accepted in all industries, it is necessary to re-evaluate existing methods of airflow design, since airflow is the biggest factor effecting both the initial start-up and operating costs of cleanrooms.
The basis of cleanroom airflow design has not changed much over the past 20 years. In designing new facilities or upgrading existing ones, most cleanroom companies simply use charts1 (Table 1) showing different air exchange rates and average velocity for various classes of cleanrooms. Another design approach is to simply fix the filter velocity at 90 feet per minute (fpm) and then specify different ceiling coverage percentages for different classification levels.1
Experience has demonstrated that these methods are not efficient and, in many cases, result in over design or in problem installations.
Further, these methods are based on experience garnered from the use of specific types of filtration and air-handling equipment: for example, such methods are based on experience developed when most HEPA filters had a rated filtration efficiency of 99.97 percent at 0.3 µm. Most terminal HEPA filters in use today have a filtration efficiency of 99.99 percent at 0.3 µm. This is not an insignificant difference. A 99.97 percent efficient filter has a fractional penetration of 0.0003, while a 99.99 percent filter's fractional penetration is 0.0001. This means that a 99.99 percent filter is three times more efficient in removing 0.3 µm particles. Thus, when considering the use of ULPA and better than ULPA filtration systems, such charts become even less useful.
Many regulatory agencies and technical societies propagate these charts in terms of guidelines. The FDA, for example, uses 90 fpm filter velocity as a recommended design guide for filter validation. Although such guidelines provide the non-expert end user with some useful information, after a few years of publication such guidelines become unreasonable edicts, especially when considering the fact that they apply to secondary parameters. After all, the primary performance criteria are:
- The cleanroom should maintain required particle and bio burden cleanliness levels.
- The desired effect on the product/process should be attained as determined by Quality Control and Assurance (QCA) testing.
The dilution effect is determined by the cleanliness of the filtered air and the airflow rate. This simplified model helps understand how airflow dilutes generated contaminants. [If the output from each packet is taken as input for each successive packet, then the model can be applied to the whole room. If the elemental volume considered is small enough, this model should accurately describe the dilution effect of the airflow in the cleanroom.] If there is particle generation only in the elemental volume considered in Figure 1, then the diluted air concentration, Cd,can be related to the filter efficiency, E, the concentration of particles in the make-up air, Cm, the fraction of make-up air to total air flow, f, the internal particle generation rate, G, and to the flow velocity/rate, Q, in the elemental volume, V (o unity cross section) by (Refer to reference 2 for details):
Cd = [f (1-E) Cm+ G/Q]/[1 (1-f)(1-E)]
All other design parameters are, in reality, secondary. Guidelines should leave room for innovation in achieving the above primary performance objectives in an energy-efficient and cost-effective manner.
The primary purpose of air handling units (AHUs) in a cleanroom is to provide clean air, adequate for the cleanroom classification, that can efficiently dilute and carry away or transport particles or other contaminants generated within the room. Since the cleanroom classification is affected by the highest contamination areas or zones within the cleanroom, it is important to design the airflow for these dirtier areas.
Figures 2a, 2b, 2c and 2d (above) are based on Equation 1:
Cd = [f (1-E) Cm+ G/Q]/[1 -(1-f)(1-E)]
and show how Cd is affected by the different velocities or airflow rates in the parcel of space with internal generation rate G (in #/min.) as a parameter. In these calculations the make-up air conditions are fixed: 3 percent make-up air (as a percent of total air flow circulation), with make-up air concentration, Cm, fixed at 500,000/ ft3.
Each figure is for a different filtration efficiency. Four 0.3 µm efficiencies are considered: a) 99.9 percent, b) 99.97 percent, c) 99.99 percent and d) 99.999999 percent. The last one is an extremely high efficiency filtration system such as achievable at reasonable pressure drop using electrically enhanced filtration as a primary HEPA in series with a terminal HEPA filter. The mid efficiencies, 99.97 percent and 99.99 percent, are common HEPA efficiencies. The 99.9 percent efficiency filter is used to consider a lower HEPA efficiency filter that will have a significantly lower pressure drop.
Figures 3a, 3b, 3c and 3d (see page 36) illustrate Equation 1 in a different manner. Each figure shows the effect of G (#/min) on Cd for all the four different filter efficiencies.
The following are the salient observations from Figures 2 and 3.
1) Clearly, airflow requirements for a particular room class depend on the internal generation rates, flow velocities and filter efficiency for given make-up air conditions. Design charts such as Table 1 do not show such dependencies.
2) Higher efficiency filters can be utilized to achieve the desired cleanroom classification at lower airflow ratesfor the same internal particle generation rates.
3) For higher class (i.e., less clean) cleanrooms, lower (than HEPA) efficiency filters may be utilized for adequate cleanroom performance at significantly lower pressure drops (and thus lower energy consumption) in such cleanrooms.
4) Lower airflow rates are sufficient at lower internal generation rates.
5) At higher internal generation rates, the differences in room performance with different filter efficiency is dramatically reducedonce again indicating that high performance filters are unnecessary for higher class cleanrooms.
6) The differences in room performance with respect to filter efficiency are, however, magnified at lower velocity and at lower particle generation rates.
7) From 5 and 6, it is clear that for low energy consumption in cleanrooms, both high filtration efficiency filters operating at lower flow rates and lower efficiency filters operating at higher flow rates should be considered. The low energy consumption selection depends on both the flow rate and pressure drop.
8) Clearly for ISO Class 2 to 4 (Class 0.1 to 10) performance, an ultra high filtration efficiency, with a reasonable pressure drop (such as an electrically enhanced double HEPA filter system) has advantages. Based on Figures 2c and 2d, a 99.99 percent filtration system would require about 60 fpm velocity, versus only about 20 fpm for a 99.999999 percent filter system. Keep in mind, of course, that other factors such as particle re-entrainment due to turbulence and other factors have not been considered here.
None of these trends are apparent in the airflow charts used currently by most cleanroom designers.
Now consider transport of the generated particles out of the room. In the absence of re-entrainment, with proper airflow distribution, even low airflow velocity (e.g., 10-20 fpm) can easily sweep off or transport generated particles out of the room. The mobility of sub-micron particles due to diffusion mechanisms is still much lower than the convection mobility at such low flow rates. Higher velocities will have higher convection transport of such particlesbut these differences may not always be significant in the time frame of the cleanroom processes. The residence time, Tr, in the above elemental volume is given by:
Tr= Volume/Flow Rate = 1/Q
Thus, at 20 fpm the residence time will be 3s. At 40 fpm, it will be 1.5s and at 60 fpm, it will be ~ 1s. What needs to be determined is the probability of whether the existence of the particle for this time frame, within this volume, will have a detrimental effect on the product. The above mentioned turbulence and particle transport studies,4-6 show that the gain in lower convection residence time at velocities higher than about 65 fpm do not result in lower actual residence time. This is due to turbulent re-entrainment as a result of eddy formation.
Transport can be hampered by re-entrainment. Re-entrainment can be due to turbulence and convection due to gradients in room pressure. These gradients are typically a result of poor distribution of airflow through the ceiling. At flow rates typically used in cleanrooms it is unlikely that particle diffusion plays a role in the re-entrainment process, except in zones of very low velocity.
Many cleanroom standards or guidelines call for 90 fpm velocity through ceiling HEPA filters. There is simply no basis for this. Recent work conducted at Sematech4 and MIT5 and related work at Asyst6 has shown that 90 fpm flow velocity is too high and can, in fact, lead to turbulent re-entrainment. Most of these researchers have recommended that the maximum velocity for a well distributed ceiling discharge should not exceed about 65 fpm.
To precisely design for minimal turbulence and re-entrainment, finite element or finite difference analysis methods are typically used to solve the Navier-Stokes fluid mechanics equations. Such methods typically require personnel with significant training and experience in numerical fluid mechanics (computational fluid dynamics) and a detailed layout of the cleanroom equipment to produce accurate results. Another method is to use the solution to the average particle material balance differential equation,3 using a value of G (approximately for the dirtiest portion of the room), to compute the room recovery rate.
How fast a room cleans up after being initially contaminated (Figure 4) is an important design parameter, since there is always periodic step ingression of contamination into the room (e.g., from material or personnel influx) apart from the continuous ingression from make-up air and other process sources. The lower (i.e., cleaner) the class of cleanroom, the faster it should recover to its classification. Keep in mind that the ordinate shows the spatial average concentration and that there will be a local variation about this space average.
Air flow distribution
Airflow distribution is one of the most important aspects of airflow design in cleanrooms. Unfortunately, not enough attention has been paid to this matter, and instead a great deal of emphasis has been placed on baseless criteria such as requiring 90 fpm filter face velocity.
One of the most commonly used filter systems, the fan filter unit (FFU), puts quantum packets of flow in a 2 3 4 ft. space. In such cases, because the cleanroom airflow is not always well distributed, the room may require higher flow rates to achieve the required classification. This is because a system with a higher level of airflow distribution will outperform a system that inputs the same amount of airflow in a smaller percentage of the room cross section. It should be noted, however, that FFUs may be used along with “membrane” ceilings to accomplish better airflow distribution.
Better airflow distribution also ensures the uni-directionality of the airflow. Thus turbulent entrainment of particles is minimized. Consequently, paying attention to airflow distribution can significantly reduce the airflow rate requirements and thus result in significant energy and operating cost savings.
The traditional central air handler includes typically custom rooftop air handlers used with some level of pre-filtration. The air is supplied to the room via terminal HEPAs using “spider leg” ducting. Some older cleanroom designs used a pressurized plenum to distribute the air into the terminal HEPAs. However, this practice is now frowned upon, especially for ISO Class 6 (Class 1,000) and better cleanrooms, due the high probability of leakage of unfiltered air into the cleanroom. The disadvantage with this system is that while most standard air conditioning units are designed for about 400 scfm per ton of refrigeration, cleanrooms require significantly higher flow rates. This therefore requires the use of additional banks of coilsi.e. customized units. Standard air conditioning systems (not for dehumidification) cost about $350 to $400 per ton, while customized units can cost anywhere from $1200 to $2000 per ton.
Additionally, customized rooftop units can be significantly heavier than standard units. Thus, in some cases roof structural reinforcement becomes necessary. Obviously, this adds to the cost of the project. In order to circumvent the higher cost of customized air conditioners standard air conditioners may be used along with a booster fan that mixes in excess re-circulating air.
This figure shows a schematic of a hybrid distributed air handling system. The distributed air handling system consists of standard air conditioning equipment and multiple fan powered in-duct electrically enhanced HEPA filters. The electrically enhanced filters (EEFs) enable high flow rates at lower pressure drop in a small packet. The fans in these units are highly efficient industrial quality three-phase fans with low noise characteristics and are capable of overcoming return and supply resistance, including the use of terminal ceiling mounted HEPA filters. Typically each in-duct filter produces the same flow rate as 4 to 5 ceiling FFUs. This system requires no booster fan.
Another method of air handling involves the use of FFUs. Since FFUs do not develop the pressure head required to overcome ducting flow resistance, either custom air conditioning systems or booster fans are often required in conjunction with the FFUs. A negative pressure plenum is created by the use of the FFUs and the air handling units simply feed air into this negative pressure plenum. This system has the following drawbacks:
1. The plenum can often end up having small zones of positive pressure depending on the ceiling coverage of the FFUs. This can cause leaks into the room.
2. Although individual FFUs are now very quiet, the use of multiple fans (in some cases hundreds of them) can result in significant noise levels.
3. When many FFUs are used, it is essential to have a monitoring system to check the status of each motor for assured proper operation of the cleanrooms.
a) Double HEPA filtration can be used. Due to double HEPA filtration, as shown by the model (Figure 5), the required cleanroom performance for ISO Class 3 to 5 (Class 1 to 100) is attainable at lower air flow rate and lower energy costs.
b) The system requires no plenum over the cleanroom. Flow distribution is independent of the flow rate. This better air flow distribution also enables achieving desired ISO Class 6 to 8 (Class 1,000 to 100,000) cleanroom performance at significantly lower total air flow a compared to systems using FFUs.
c) The system enables the use of standard air conditioners, at significantly lower initial costs.
d) The in-duct units enable convenient servicing of the filters outside the roomdramatically reducing the downtime, risk of potential damage to cleanroom equipment and materials and costs for filter replacement. The room is not contaminated during filter change outs. The terminal HEPA filters in use do not require replacement, except in the case of accidental damage.
e) The use of EEF filters (as opposed to conventional in-duct HEPA filters) results in significantly lower bio burden, due to the bactericidal properties of the EEF.7
f) The system produces less room noise than FFU based systems since there are no fan/motors in the ceilingthere are less of these and these are located at a distance from the ceiling. Also the lower air flow rate also substantially reduces noise levels.
In summary, this paper has presented a basis for cleanroom airflow design and discussed various types of air handling systems. Considering the high energy consumption and cost of cleanrooms and the need to reduce this, it is about time that various regulatory and standards setting organizations improve their standards and guidelines by basing them on sound fundamentals rather than on oversimplified irrational edicts. These guidelines should allow flexibility and innovation in order to enable efficient cleanroom air handling. These guidelines should concentrate on the fundamental cleanroom requirements vs. room classification under operation and bio burden, rather than on secondary and unconnected criteria such as filter velocity.
Rajan Jaisinghani is President of Technovation (Midlothian, VA). He is a chemical engineer with a bachelor of science degree from BHU, India, and a master of science degree from the University of Wisconsin. He has 30 years of R&D experience in aerosol and colloid science, filtration and fluid mechanics. He is widely published in these fields, including chapters in handbooks, and holds 11 patents. Raj has played a pioneering role in the research, development and commercialization of electrically enhanced filtration technology.
- Hansz, T., “Cleanroom Programming and Planning,” Proc. CleanRooms 96 East, 11th Int. Conference On Advanced Technology and Practices for Contamination Control, p. 233, 1996.
- Jaisinghani, R.A., ” New Ways of Thinking About Cleanroom Airflow Design,” A2C2, Vol. 3, #11, Dec. 2000.
- Jaisinghani, R.A., T.J. Inzana and G. Glindermann “High Filtration/Biocidal Performance Cleanroom System,” CleanRooms West '96, Proc. Conference on Advanced Microcontamination and Ultrapure Manufacturing, 1996.
- Sematech Reference – unpublished research.
- Vazquez, M. “The Study of Altering Air Velocities in Operational Cleanrooms,” Proc. CleanRooms West '99, San Jose, CA,1999.
- Tannous, A.G. and K.H. Compton, “Studies Conclude Low Air Velocity Increases Effectiveness of Minienvironment Design,” CleanRooms, March 1998.
- Jaisinghani, R.A., G. Smith and G. Macedo, “Control and Monitoring of Bioburden in a Biotech/Pharmaceutical Cleanrooms,” J. Institute of Validation Technologies, Vol. 6, # 4, pg 686, Aug 2000.
Jaisinghani will expand on this topic in a conference presentation at CleanRooms East 2001 in Boston in March. For more information or to register, call (603) 891-9267 or visit www.cleanrooms.com to view the conference brochure.