In a recent paper entitled “Evaluating the Sensitivity of the Mass-Particle Removal Calculations for HVAC Filters in ISO 16890 to Assumptions for Aerosol Distributions” the author, Dr. Brent Stephens of the Illinois Institute of Technology, found that the particle distribution data used in ISO 16890 was outdated and inaccurate. The paper was published in the peer reviewed Journal Atmosphere 2018, 9(3), 85.
The particle distribution data is an important component of ISO 16890 in that it is used in conjunction with tested air filter efficiency by particle size to determine the air filter classification. In the Standard, the efficiencies by particle size ranges are multiplied by the density of the particles in those ranges as defined by “urban” and “rural” particle distributions. These “urban” and “rural” particle distributions are supplied in the Standard.
The particle distributions used in ISO 16890 come from a textbook on atmospheric chemistry and physics by Seinfeld and Pandis (2006). Dr. Stephens traced the origins of this data and found that it came from very old studies. In the case of the “urban” distribution it appears to have originated in textbooks dating back to the 1960’s and 1970’s. In the case of the “rural” distribution it comes from studies conducted in 1975 and 1976 measuring particle counts in “the lower troposphere over the high plains of North America.”
(So if someone defends the particle distributions in ISO 16890 by saying: “Do you think we pulled this data out of thin air?” The answer is: Yes.)
Dr. Stephens and his Built Environment Group at IIT have done extensive work in analyzing outdoor particle distributions. Their paper, Azimi et al. (2014), identified a total of 194 long-term average air distributions primarily from North America (the US and Canada) and Europe (Sweden, Finland, Norway, Denmark, Germany, France, Italy, Ireland, England, the Netherlands, Switzerland, Lithuania, Hungary, the Czech Republic and Bulgaria).
The current paper compared that data to the “urban” and “rural” distributions used in ISO 16890. Much has changed in outdoor air in the last 40 years. Even the “urban” data used in ISO 16890 reflects substantially higher particle numbers than current particle distribution measurements. However, the curve of the “urban” distribution is similar in shape to the 194 other outdoor air distributions.
Unfortunately, the same is not true for the “rural” distribution. It is a stunning outlier when compared to all of the other distributions. It is not similar in shape or number. In fact, one look at the chart in the study that overlays all of the distributions should be enough to convince anyone that the “rural” distribution should not be used to classify filters.
Dr. Stephens then calculated estimated filter removal efficiencies (called ePM1, ePM2.5 and ePM10 in the Standard) using all of the distributions. Typical filter efficiency “curves” were used for MERV 6, MERV 8, MERV 10, and MERV 14 filters. For MERV 14 filters these calculations showed that the removal efficiencies using the ISO 16890 “urban” particle distribution and the efficiencies using the Azimi et al distributions gave similar results for ePM1. The results for ePM 2.5 varied by about 5% in the same comparison. (This is enough to result in a different filter classification in ISO 16890.) On the other hand, the calculations of PM2.5 on a MERV 10 filter efficiency curve showed a 23% variance between the “rural” distribution and the Azimi et al average distributions.
A major problem with ISO 16890 is that the filter classifications are based on calculations using 100% outdoor air. Very few HVAC systems in the United States use 100% outdoor air. The article addresses this issue by using particle distributions that include indoor particles of outdoor origin and indoor generated particles.
When using particle distributions adjusted for indoor particles of outdoor origin, there are substantial differences in estimates of ePM 2.5 and ePM 10. For example, for a MERV 10 filter the calculated ePM 10 varies by 11.5% when using the “rural” distribution verses the Azimi et al average distribution. This would mean a difference of three filter classifications in the ePM10 group. In other words, an ePM10 – 65 filter could become an ePM10 – 50 just because of inaccuracies in the data.
The differences become even more pronounced when using “indoor” particle distributions from several sources. When this data is used, ePM 10 calculations are as much as 29% off between the “rural” particle distributions in ISO 16890 and the indoor particle distributions. Basically what this means is that in US residential applications where filters are primarily exposed to recirculated indoor air that the ISO 16890 calculations would be 6 filter classification levels away from where they should be. This is not the type of information that instills confidence in air filter purchasers.
The article closes with five recommendations to improve ISO 16890:
- The particle distributions need to be updated.
- The “urban” and “rural” distributions should be reduced to one distribution used to calculate ePM1, ePM2.5 and ePM10.
- The Standard should be modified to include indoor particles instead of just 100% outdoor air.
- More research needs to be conducted to determine indoor particle distributions for a variety of different building types and conditions.
- It would be helpful to include particle distributions and efficiencies on particles smaller than 0.3um.
To summarize, this article does an excellent job of analyzing the particle distribution data used in ISO 16890. The data is both outdated and, at least in it’s use of “rural” particle distributions, inaccurate. Since this data is required as a part of the calculations in the Standard to determine air filter classifications, the use of ISO 16890 is questionable. For a standard to be widely accepted it must be unbiased, accurate and repeatable. ISO 16890, as it currently stands, is none of these.
The article reviewed in this blog post can be found at: