Whitepaper HVAC Pre-filtration In Pharmaceutical Facilities I N D U S T R Y I N S I G H T S Norman A. Goldschmidt September 11, 2011 Principal, Engineering www.geieng.com
I N D U S T R Y I N S I G H T S Genesis Engineers periodically publishes white papers and reports about topics of special interest to the industries we serve. As veteran advisors for major corporate infrastructure, energy management, facilities, technology, manufacturing and building systems of every type, our leaders share their perspectives to help both clients and the public at large make high value decisions by having the best available information. All information contained herein is copyrighted and cannot be reproduced without permission. For academic uses, please contact us. Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 2
Whitepaper HVAC Pre-filtration In Pharmaceutical Facilities Introduction A common challenge in the design of sterile filling and biotech facilities is selection of the appropriate type of pre-filtration to employ ahead of supply HEPA or ULPA filters serving processing areas. This document suggest an approach intended to assure compliance with applicable regulation and industry guidance - specifically, FDA 2004 Guidance for Industry, Sterile Drug Products, Produced by Aseptic Processing, Current Good Manufacturing Practice, EudraLex Vol.4 Annex 1, PIC/S and WHO TRS 957 Annex 4. Hypothesis The scientific rationale for non-hepa filters as pre-filters is based on the following: 1. A single HEPA or ULPA filter can provide a 4-12 log reduction in airborne particulates, depending on the particle size - This makes a typical HEPA suitable for the GMP job of excluding bioburden. 2. At all particle sizes (above about 0.001µ) the filter exceeds its rated efficiency, peaking (leakage too small to measure) at around 1µ. 3. Typical installed HEPA failures are pinhole leaks that do not result in a measurable increase in particle count or risk. 4. High Efficiency (MERV 14-18) filters also reach peak efficiency (leakage too small to measure) at around 1µ and certainly before 5µ. 5. High Efficiency filters reduce the challenge presented to HEPA filters and extend the life of the HEPA nearly as well as a HEPA. Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 3
Background and Rationale Regulations and guidance documents do call for the use of HEPA filters in air systems serving Aseptic processing areas; however, they do not mandate the location of the HEPA nor do they usually mandate the type of pre-filtration used (WHO guides being the exception). In our benchmarking, we find that a significant number of pharmaceutical manufacturers have moved away from double HEPA filtration (one in the AHU, one terminal). The rationale for moving away from this practice has been that the old practice does not yield a scientifically supportable reduction in risk. Further the change to a non-hepa pre-filter gives enhanced simplicity of testing and record keeping and reduces operating costs. Interestingly, bioburden containing particles tend to be well above 1 micron in size. We note that European Good manufacturing Practice (GMP) regulations acknowledge that the 5µ particle size range is of interest as most indicative of bioburden (EudraLex Annex 1). This supposition is supported both by data on the sizes of spores and other microorganisms, as shown in figures 1-3 and by studies of contaminants in hospital rooms which suggest that the average size of airborne viable containing particles (virus containing droplets, shed skin, spores, etc.) may be around 12 microns. s in this size range are stopped nearly absolutely by HEPA or ULPA filters and are nearly as efficiently filtered by lower efficiency (MERV 13-15) filters. This fact would give the HVAC designer significant leeway in the design of pre-filtration. Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 4
FIG 1. PARTICLE TYPE / SIZE Source AAF Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 5
FIG 4. BACTERIA PARTICLE SIZE FIG 5. FUNGUS PARTICLE SIZE Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 6
What's more, given the size of typical bioburden, an intact HEPA or ULPA filter is virtually absolute with respect to its ability to fulfill the GMP role of excluding airborne organisms from a zone of concern (e.g. exposed product or product contact surfaces). This information, coupled with the fact that the typical failure mode of terminal HEPA and ULPA filters is a Pinhole leak (a leak of greater than 0.01% penetration) which does not impact the room classification; we suggest that the role of pre-filters is to fulfill Good Engineering Practice (GEP) and/or Good Business Practice (GBP). Based on the preceding, the GEP or GBP role of pre-filters ahead of terminal HEPA or ULPA filters is two fold: 1. Extend the life of the GMP filter by minimizing filter loading 2. Minimize the challenge handled by the GMP filter to reduce risk in the case of failure Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 7
Evaluation The rationale most often used to justify the use of HEPA filters as pre-filters unit is that, in the case of a leak in the terminal filter, the upstream HEPA would protect the product. This logic has two critical flaws: 1. Pinholes leaks don t raise the particle count in Grade A or B in any measurable way. 2. Contamination downstream of the pre-filter is typically not known or managed. (If the HEPA pre-filter in the AHU is providing GMP protection of the product then anything downstream of the AHU [e.g. VAV boxes, etc] is part of the room system. This is cumbersome and not usually how the system is maintained.) To explain why pinholes in HEPA or ULPA filters don t raise particle counts measurably in a grade A or B area, it must be understood that, under normal conditions, the terminal filters reduce the contribution from the HVAC to a tiny fraction of the allowable particle count. As a first approximation, we can calculate the impact of intact and flawed HEPA filters on recirculated ISO 7 return air: Note: These evaluations neglect filter frame leakage, which is assumed to be independent of filter media type. Example Count Reduction from HEPA Filtration Size Return Air Count* Fractional Filter Efficiency Resultant Count ISO 5 Standard Counts 0.1µ 1.0E+07 part/m 3 x 99.9990% = 1.0E+02 part/m 3 1.00E+05 part/m 3 0.5µ 3.5E+05 part/m 3 x 99.990% = 3.5E+01 part/m 3 3.52E+03 part/m 3 1.0µ 8.3E+04 part/m 3 x 99.9990% = 8.3E 01 part/m 3 8.32E+02 part/m 3 5µ 2.9E+02 part/m 3 x 99.9999% = 2.9E 04 part/m 3 2.90E+01 part/m 3 Total 1.1E+07 2.4E+02 part/m 3 1.1E+05 part/m 3 *Based on ISO 7 Maximum Conditions PASS Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 8
Example Count Reduction from leaking HEPA Filtration Size Return Air Count* Fractional Filter Efficiency Pinhole Effect** Resultant Count ISO 5 Standard Counts 0.1µ 1.0E+07 part/m 3 x 99.9990% 0.0199% = 1.9E+03 part/m 3 1.00E+05 part/m 3 0.5µ 3.5E+05 part/m 3 x 99.990% 0.0199% = 3.5E+01 part/m 3 3.52E+03 part/m 3 1.0µ 8.3E+04 part/m 3 x 99.9990% 0.0199% = 1.6E+01 part/m 3 8.32E+02 part/m 3 5µ 2.9E+02 part/m 3 x 99.9999% 0.0199% = 5.8E 02 part/m 3 2.90E+01 part/m 3 Total 1.1E+07 2.0E+03 part/m 3 1.1E+05 part/m 3 *Based on ISO 7 Maximum Conditions PASS ** Based on leak of.01.02% total efficiency loss across filter at all particle sizes (conservative) Example Count Reduction from failed HEPA Filtration Size Return Air Count* Fractional Filter Efficiency Hole Effect** Resultant Count ISO 5 Standard Counts 0.1µ 1.0E+07 part/m 3 x 99.9990% 0.9999% = 1.0E+05 part/m 3 1.00E+05 part/m 3 0.5µ 3.5E+05 part/m 3 x 99.990% 0.9999% = 3.5E+03 part/m 3 3.52E+03 part/m 3 1.0µ 8.3E+04 part/m 3 x 99.9990% 0.9999% = 8.3E+02 part/m 3 8.32E+02 part/m 3 5µ 2.9E+02 part/m 3 x 99.9999% 0.9999% = 2.9E+00 part/m 3 2.90E+01 part/m 3 Total 1.1E+07 1.1E+05 part/m 3 1.1E+05 part/m 3 *Based on ISO 7 Maximum Conditions AT LIMIT ** Based on leak of 1% total efficiency loss across filter at all particle sizes (large failure) In summary, a HEPA filtering ISO 7 air into an ISO 5 area would need to have a hole (or holes) of sufficient size, when diluted by the air going through the balance of the filter, to drop the overall filter efficiency below 99% before the air would cause a classification failure. Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 9
To better understand the impact of true pinhole leaks (individual photometer readings of greater than 0.01% penetration, per FDA guidance) on overall particle count, we can calculate the impact of individual and groups of leaks on the output of a single HEPA filter: Size HEPA Filter Leak Effect Pinhole Return Air Count Penetration Resultant Count ISO 5 Standard Counts Area of Reading flow Balance of Filter 624 cfm/filter 0.5µ 3.52E+05 part/m3 0.0099% = 3.4848E+01 part/m3 Leak 1 cfm/filter 0.5µ 3.52E+05 part/m3 0.0100% = 3.5200E+01 part/m3 1x leak Total 625 cfm/filter Average 3.4849E+01 part/m3 3.52E+03 part/m 3 PASS Size HEPA Filter Leak Effect Larger Pinhole Return Air Count Penetration Resultant Count ISO 5 Standard Counts Area of Reading flow Balance of Filter 624 cfm/filter 0.5µ 3.52E+05 part/m3 x 0.0099% = 3.4848E+01 part/m 3 Leak 1 cfm/filter 0.5µ 3.52E+05 part/m3 x 1.0000% = 3.5200E+03 part/m3 100 x leak Total 625 cfm/filter Average 4.0424E+01 part/m3 3.52E+03 part/m 3 PASS HEPA Filter Leak Effect Limit Determination Size Return Air Count Penetration Resultant Count ISO 5 Standard Counts Area of Reading flow Balance of Filter 563 cfm/filter 0.5µ 3.52E+05 part/m3 x 0.0099% = 3.4848E+01 part/m 3 Leak 62 cfm/filter 0.5µ 3.52E+05 part/m3 x 10.0000% = 3.5200E+04 part/m3 1000 x leaks Total 625 cfm/filter Average 3.5232E+03 part/m3 3.52E+03 part/m 3 At Limit From this analysis we conclude that it would take over 60 leaks, with a penetration of 10% each (1,000 times greater than the lower acceptance limit) for the discharge of a filter to supply air that does not meet classification. This analysis is very conservative as it doesn t account for the impact of any pre-filtration that might be in place and neglects the fact that a scanned filter (repaired or leak free) has a much higher overall efficiency than it s rated efficiency (the act of assuring that a HEPA has no leaks exceeding 0.01% penetration means that the entire filter must have greater than 99.99% efficiency.) Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 10
The use of in-duct or in-ahu HEPA s as pre-filters for terminal HEPA or ULPA filters have other disadvantages: 1. HEPA filters have a low dirt holding capacity, this necessitates frequent change out, or the use of a high dirt holding capacity filters ahead of the HEPA (typically a MERV 11-15), this adds unnecessary pressure drop and energy consumption. 2. There is difficulty in discriminating between GMP HEPA filters that require efficiency or integrity testing, and those employed as prefiltration, which do not require testing. 3. If testing is employed for HEPA filters used as pre-filters, there is no clear guidance on the appropriate acceptance criteria for such filters. 4. When integrity scan testing is employed for in AHU or in duct HEPA filters, leakage in the filter holding frame or between the frame and the filter are often found and require significant intervention to reduce penetration below 0.01%. By contrast, MERV 13-15 filters remove the vast majority of dirt and virtually all particles in the size range of concern ahead of terminal filters, providing them with a 10-15 year life. The next two calculations demonstrate the effectiveness of MERV 15 prefiltration in reducing the particle count and mass of contaminants from ISO 7 return air: Example Count Reduction from MERV 15 Filtration Initial Size Count (ISO7) Fractional Filter Efficiency Resultant Count 0.1µ 1.0E+09 part/m 3 x 50.000000% = 5.0E+08 part/m 3 0.3µ 1.0E+08 part/m 3 x 78.600000% = 2.2E+07 part/m 3 0.5µ 3.5E+07 part/m 3 x 92.500000% = 2.6E+06 part/m 3 1.0µ 8.3E+06 part/m 3 x 97.400000% = 2.2E+05 part/m 3 5µ 2.9E+05 part/m 3 x 99.500000% = 1.5E+03 part/m 3 Total 1.2E+09 5.3E+08 part/m 3 Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 11
Example Total Mass Reduction from MERV 15 Filtration Size Mass Fractional Filter Efficiency Fractional Mass 0.1µ 5.3E 01 mcg/m 3 x 50.000% = 2.6E 01 mcg/m 3 0.3µ 1.5E+00 mcg/m 3 x 78.600% = 3.1E 01 mcg/m 3 0.5µ 2.3E+00 mcg/m 3 x 92.500% = 1.7E 01 mcg/m 3 1.0µ 4.4E+00 mcg/m 3 x 97.400% = 1.1E 01 mcg/m 3 5µ 1.9E+01 mcg/m 3 x 99.500% = 9.6E 02 mcg/m 3 10µ 3.6E+01 mcg/m 3 x 99.999999% = 3.6E 07 mcg/m 3 Total 6.4E+01 mcg/m 3 0.96 mcg/m 3 Mass Filter Flow Mass Loading 0.96 mcg/m3 x 2045 m 3 /hr = 1.95E 03 g/hr 1.95E 03 g/hr x 8000 hr/yr = 15.6 g/yr From the preceding, we conclude that a MERV 15 filter is adequate to satisfy both the reduction of biological challenge and the reduction of loading on downstream terminal filters. Note: The above analysis assumes a spherical particle with a specific gravity of 1.0 Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 12
Economics At the manufacturer s recommended airflow, the pressure drop for a MERV15 filter is significantly lower than for an equivalently sized conventional HEPA. The expected energy savings from this the use of the non-hepa filters is greater than $100/unit. Filter Type Bin Hours Average Supply Air TSP ("w.c.) Fan Power per Filter (W) Annual Energy Usage per Filter (KWh) Annual Fan Energy Cost per Filter HEPA 8000 1.8 402.41 3219.3 $321.93 MERV 15 8000 1.1 251.51 2012.0 $201.20 Energy Savings 1207.2 $120.72 /filter year Inputs and Assumptions: Starting DP Ending DP Fan Efficiency: 70% HEPA 1.2 2.4 "WC Motor Efficiency 90% MERV 15 0.75 1.5 "WC Electric Power Cost: $0.100 /kwh Airflow 1200 cfm Conclusion The most common rationale we see for using HEPA filters as GMP prefilters is " to protect against terminal filter leaks". This argument is not persuasive for three reasons: 1. The leaks commonly found in terminal filters aren t sufficient to impact room class. 2. There is normally little or no control of contamination between the AHU and terminal. 3. High efficiency filters (below HEPA rating) are more energy efficient, while still providing excellent control of loading and upstream biological challenge to terminal filters. Switching to MERV 14/15 pre-filters should save over $100/filter in annual energy cost, as well as a significant amount in annual filter testing. Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 13
Recommendations 1. We recommend using 95% efficiency (MERV 14/15) pre-filters on AHU(s) serving aseptic areas to extend the life of the terminal filters and reduce the biological challenge that these filters see. 2. We recommend mounting only MERV 15 filters, without pre-filters, at the AHU discharge in recirculated air streams for aseptic processing, including Grade A and B spaces, with a high capacity bag MERV12/13 pre-filter on any outside air. 3. We don t recommend having two GMP filters in series, unless you are focused on cross-contamination protection, this adds cost and distracts from the critical control point (the terminal HEPA) 4. The added cost and complexity of HEPA s in the AHU don t add significant product protection benefits due to the relatively lower filter efficiency required to meet Grade A or B using predominantly recirculated air. Copyright Genesis Engineers 2011 - All rights reserved - Do not reproduce without written permission. 14