Criteria for Calculating the Efficiency of HEPA Filters During and After Design Basis Accidents

We have reviewed the literature on the performance of high efficiency particulate air (HEPA) filters under normal and abnormal conditions to establish criteria for calculating the efficiency of HEPA fitters in a DOE nonreactor nuclear facility during and after a Design Basis Accident (DBA). The literature review included the performance of new filters and parameters that may cause deterioration in the filter performance such as filter age, radiation, corrosive chemicals, seismic and rough handling, high temperature, moisture, particle clogging, high air flow and pressure pulses. The deterioration of the filter efficiency depends on the exposure parameters; in severe exposure conditions the filter will be structurally damaged and have a residual efficiency of 0%. Despite the many studies on HEPA filter performance under adverse conditions, there are large gaps and limitations in the data that introduce significant error in the estimates of HEPA fitter efficiencies under DBA conditions. Because of this limitation, conservative values of filter efficiency were chosen when there was insufficient data. The estimation of the efficiency of the HEPA filters under DBA conditions involves three steps. In the first step, the filter pressure drop and environmental parameters such as temperature and moisture are determined during and after the DBA. The second step consists of comparing the filter pressure drop to a set of threshold values above which the filter is structurally damaged. There is a different threshold value for each fitter type and different environmental conditions. The filter efficiency is determined in the third step. If the filter pressure drop is greater than the threshold value, the filter is structurally damaged and is assigned 0% efficiency. If the pressure drop is less, then the filter is not structurally damaged and the efficiency determined from literature values of the efficiency at the environmental conditions. The efficiency of the HEPA filters within DOE facilities should be determined on a case-by-case basis. 'This work was performed under the auspices of the US. Department of Energy by Lawrence Livemre National Laboratory under contract No. W-7405-Eng.48. 1. Harvard School of Public Health, 665 Huntington Ave., Boston, MA 02115 2 Consultant, R.R. 4, Box 4172, LaPlata, MD 20646 3. Consultant, P.O. Box 704, McLean, VA 22101 4. Consultant, P.O. Box 29720, Columbus, OH 43229 23rd DOE/NRC NUCLEAR AIR CLEANING AND TREATMENT CONFERENCE The primary standard that governs the use of HEPA filtration systems in DOE facilities is DOE Order 6430.1A [l], which in turn refers to ASME N509 [2]. However, the standards do not provide guidance for determining their efficiency under abnormal or accident conditions. Under normal operating conditions, a HEPA filter will have a minimum efficiency of 99.90% [3]. This is the minimum filter efficiency at the most penetrating particle size of about 0.15 pm diameter for a HEPA filter that has a minimum efficiency of 99.97% for 0.3 pm DOP particles. Previous publications had assigned a single value for the efficiency ot HEPA filtration systems under all accident conditions. However the publications reviewed in this paper demonstrate that the efficiency of HEPA filters will vary greatly depending on the operating conditions, thereby requiring a case-by-case analysis. Elder et al[4] prepared a guide for analyzing the accidental release of radioactive material from nonreactor nuclear facilities. This guide reviews the applicable DOE orders, provides a description of the design basis accidents, and evaluates the consequences of the accidents. In his section on reduction and removal factors, Elder discusses the efficiency of HEPA filters under DBAs. Elder recommended that the first stage HEPA be credited with an efficiency of 99.9% and each subsequent stage with 99.8%. These values were selected from an unofficial HEPA filtration guideline (AEC unpublished) that was established in 1971 during a meeting between officials from the Atomic Energy Agency and Albuquerque Operations Office [5]. These guidelines represented the opinions of the meeting attendees, and were not supported by technical data. Walker [6] tabulated the available data on HEPA filter efficiency as of 1978 and recommended efficiencies of 99.9% for the first stage, 99.8% for the second and third stages, and 83.0% for the fourth stage. Although Walker cited experimental data, the recommended values were based on his opinion. Walker further stated in the Summary that further study of HEPA filter efficiency is needed "to better establish relationships with relative humidity and temperature of sweep gas, service aging, material loading, etc." The Nuclear Regulatory Commission in Regulatory Guide 1.52, recommended an efficiency of 99% for a filtration system consisting of a two stages of HEPA filters with an adsorber in between m. The guide is applicable to light-water-cooled nuclear power plants, where the filtration system is off-line during normal operations and only activated during accident conditions. Table 1 of the code shows that environmental conditions to which the HEPA filters would be exposed are approximately atmospheric pressure, 180°F, 100% relative humidity, and 105 radshr. There is no explanation for setting the HEPA 23rd DOUNRC NUCLEAR AIR CLEANING AND TREATMENT CONFERENCE filter efficiency at 99%. when the new filter is tested at 99.97% for 0.3 um DOP particles and is also tested in-place for leaks at 99.95% as prescribed in ASME N510. [SI The Nuclear Air Cleaning Handbook 191 recommended 99.8% efficiency for the first stage and 99.9% to 99.95% for each of the remaining stages. The Handbook presumably recommends a lower efficiency for the first HEPA filter and a higher efficiency for the second stage HEPA than the AEC 1971 draft because of the assumption that the first filter wouM take the brunt of any adverse effect from accidents. In discussing multistage HEPA filtration, the Handbook stated on page 38, The purpose is to increase the reliability of the system by providing backup filters in the event of damage, deterioration, or failure of the first filters." The major deficiency in the previous guidelines on HEPA filter efficiencies under DBA conditions is that the HEPA filter efficiency in DOE facilities can vary from 99.9% to 0% depending on which of the large number of different DBA conditions is applicable. In contrast, for Ight water reactors, there is a single DBA condition for which the HEPA filter efficiency is assigned 99% [q We have proposed a method for computing the efficiency of HEPA filters on a case-bycase basis using the available data reported in the literature. In this paper we describe the development of the criteria for calculating the efficiency of HEPA filters during and after a DBA. This criteria is intended to be used in a future DOE Standard. The goal of this paper is to provide guidance for computing the efficiency of HEPA filters during and after a DBA. The computed filter efficiency can be used in determining off-site doses from postulated releases of airborne radioactivity for both existing and future facilities. However, this study is not intended to define criteria for HEPA filter survival or provide guidelines on how to construct a HEPA filter that survives a DBA. Since there are a large number of different DBAs that are applicable to the many different DOE facilities, the HEPA filter efficiency will be computed on a case-by-case basis. . . -roaches for C o r n P w HFPA Fitter Fffcrw Under= lue for All DRA Cond itions We considered various approaches for developing the criteria for computing HEPA filter efficiency under DBA conditions. The previous approach was to specify a single efficiency value for all conditions. We considered this to be unrealistic since the efficiencies of the available HEPA filters vary widely with different environmental parameters. Since there are no commercially available HEPA filters certified for 23rd DOffNRC NUCLEAR AIR CLEANING AND TREATMENT CONFERENCE use in DOE facilities that can survive all of the postulated DBA conditions it is not possible to assign a sing\e efficiency value to HEPA filters. For the currently available HEPA filters, the only efficiency value that can be applied for all DBA conditions is 0%. However, since the HEPA filters in most DOE facilities will survive the applicable DBA, assigning 0% efficiency to these filters is unrealistic. We concluded that until a high-strength HEPA finer is developed, it will not be possible to assign a single efficiency to HEPA filters that will apply for all DBA conditions. From Matrix of Three Severitv I evels of Fire. FxDlosions.and Torn&~s We then reviewed various matrix approaches in which there are multiple filter stresses and multiple responses. A filter efficiency would be assigned for each of the filter stresses. In our initial review, there appeared to be only a few realistic DBA scenarios that include combinations of anticipated stresses that reflect graduated levels of severity. For example, a remote fire usually represents a filter loading episode without a very high temperature or a pressure pulse; an explosion usually represents a temperature and pressure pulse, and it may be the pressure resulting from rapidly increased airflow that does the damage and rapidly dies away to be followed by a fire without excessive pressure or airflow volume increase; a tornado results in a rapid pressure increase and/or decrease but no fire or particle/droplet loading. Each scenario can occur in varying degrees of severity that can probably be classified as 'below concern", "moderate stress", and 'extreme stress". This represents a 3 x 3 matrix and should not be excessively complex. Although the concept is great, we were not able to assign an efficiency value for each of the proposed nine elements of the matrix because the three stresses of fire, tornado, and explosions were not single stress quantities, but could very widely in terms of more fundamental parameters such as temperature, pressure, and aerosol quantity and composition. For example estimaling the efficiency of a HEPA filtration system after a fire cannot be derived from categories such as below concern, moderate stress or extreme stress. In order to assess the response to the fire, it will be necessary to estimate the temperature. quantity of smoke, type of smoke, water spray from a fire suppression system, the system flow rate, and the pressure drop across the HEPA filter. The pressure drop across the filer is, in turn, a function of the flow rate, water contact, and the quantity and type of smoke. There is not a single DBA sequence that represents a fire. For example the remote fire scenario only involves filter clogging with little temperature increase or pressure pulse. In reality, there are a large 23rd DOE/NRC NUCLEAR AIR CLEANING AND TREATMENT CONFERENCE number of different fire scenarios: a well ventilated fire generating slow plugging aerosols, an under ventilated fire generating rapidly plugging aerosols, a fire with high temperature flames, use or non-use of fire suppression system based on water sprays and demisters, etc.. if the fire is remote and well ventilated, then the only consequence would be a small or moderate increase in pressure drop and no loss in efficiency. However a remote fire that is under ventilated (oxygen starved) would cause rapid filter plugging. If the air blower does not exceed the breaking point of new filters (37 inches ) then the effect would be a near or total shut down of the ventilation system. The efficiency of the clogged filter would be higher than the efficiency of the clean filter. If the smoke is diluted prior to reaching the HEPA filter, then the filter plugging will be reduced proportionately to the air dilution. As another example dealing with fire, assume the fire is localized at the filter and reaches 400°C. If no water suppression system is activated and the filter pressure drop is below 15 inches then the filter will be structurally undamaged but have a penetration about of about 3% . However if the water spray system is activated, then the fitter will rapidly plug due to water accumulation. If the air blower can pull greater than 10 inches of water then the filter will be structurally damaged and have dramatically reduced efficiency (conservatively set at 0% ). However if the air blower cannot pull greater than 10 inches of water, then the plugged filter will shut down the ventilation system but still have the high efficiency. Thus from these few examples we have shown that estimating the efficiency of the HEPA filters requires a more complex approach than categorizing a stress as below concern, moderate stress or extreme stress. There are also many other potential stresses than fire, tornado and explosion: for example, steam from a ruptured steam line, fire suppression system that sprays water, chemical effluents, and seismic stress. In addition, the DBA stresses cannot be limited to a small number of different conditions because new DBA conditions will be established for new facilities and operations as DOE missions and directions change. New programs in weapons dismantlement, waste clean up, decontamination and decommission operations will bring their own set of stress factors. m Gene r a l i 7 W t r i x of ODe ratina Paramers and P m r a l DWe then evaluated a generalized matrix in which the efficiency is established at each level of stress for a series of fundamental parameters at increasing parameter level. Table 1 illustrates the 23rd DOUNRC NUCLEAR AIR CLEANING AND TREATMENT CONFERENCE generalized matrix for tabulating the filter efficiency at each parameter level for a series of different parameters. These are the known parameters that have an effect on filter efficiency. Table 1. Generalized matrix for establishing filter efficiency for operating parameters Ffficiewv at Different P w e t e r I ey& Lnry Moderate m temperature so l i particle loading waterllquid particle loading air flow Although Table 1 shows only three parameter levels (low, moderate, and high), the complete table couM consist of many parameter levels, each level defined by specific values of the operating parameter. The primary advantage of the generalized parameter matrix over the three-stress matrix is that the parameters are fundamental parameters that can be uniquely defined. It is also possible to use semi-empirical equations for computing the efficiency due Po the different parameters. Since the four parameters can be considered to act independently on the filter, the filter efficiency for combination of parameters can be computed from the efficiency for individual parameters. The effect of the four operating parameters on the HEPA filter efficiency can be explained in terms of particle capture and particle loading theory and experiments. For HEPA filters, the maximum particle penetration is determined by the Brownian motion and interception capture mechanisms [lo] As the temperature is increased, the Brownian motion will increase while the interception mechanism remains constant. Thus higher temperatures will result in higher filter efficiencies. As the air flow increases, the capture by Brownian motion will decrease because of the decreased particle residence time in the filter. Since air flow has no effect on the interception mechanism, the net effect of increased air flow will be a lowering of filter efficiency. For solid particle loading, the particles form dendritic structures that capture additional particles, thereby increasing filter efficiency (1 11 For liquid particles, the particles coalesces withiti the filter and decrease the void volume within the filter medium. The decreased void volume results in higher internal velocity and thus less residence time for Brownian motion [12] Thus, as liquid deposits form in the filter, the aerosol penetration will increase. 23rd DOUNRC NUCLEAR AIR CLEANING AND TREATMENT CONFERENCE Unfortunately the direct approach for computing the efficiency of HEPA filters under DBA conditions described above is not possible because the HEPA filters are frequently structurally damaged, which lowers the filter efficiency. For example as the temperature is increased, the efficiency increases. Figure 1 shows the decontamination factor (DF = Vpenetration) as a function of particle diameter for a deep-pleated HEPA fitter with aluminum separators at increasing temperatures up to 200 C.