Statistically Significant Efficiency Testing of HEPA Filters

摘要

High Efficiency Paniculate Air (HEPA) filter manufacturers must ensure the accuracy of efficiency testing for these filters and, at the same time, keep productivity at a reasonable level. This study describes tests conducted to determine the minimum time required to provide statistically significant efficiencies of HEPA filters using a test system currently in use for HEPA filter production. The test results of over 300 HEPA filters were analyzed in this study. The approaches in IEST RP-CC-007.1: Testing ULPA Filters for statistical analysis of particle counts were followed. A Poisson distribution was used to describe the variations and to calculate the lower and upper limits of measured counts at 95% confidence level. This study shows that 90 seconds is sufficient for this test system to determine whether a HEPA filter has an efficiency of 99.990% or higher at 95% confidence. Longer tests will cause the mean measured efficiency to approach the filter's true efficiency and also narrow the confidence interval. However, challenging HEPA filters made of glass fiber media with oil aerosols for up to one hour does not significantly change the filter's true efficiency performance. The preferred approach for a new test system is to use software that calculates the confidence limits of filter efficiency after each sampling cycle and determines whether to continue testing. For an existing system without this type of program (such as the one evaluated in this study), analysis is necessary to make sure the HEPA filters are challenged with enough aerosol particles.