By Rick Andrew
NSF/ANSI 53 defines a mechanical filtration system (section 3.35) as: A system that mechanically separates particulate matter from water. What this definition means is that mechanical filtration uses a size-exclusion mechanism, as opposed to absorption, adsorption, or electrostatic effects, to separate suspended solids from water.
Mechanical filtration is a key type of performance not only for residential applications but also for commercial and industrial water treatment systems. There are many technologies that may have mechanical filtration capabilities of varying degrees, including melt-blown filters, pleated filters using various media, carbon blocks, ceramic filters and others.
This column will examine some of the common elements of this testing. It will also discuss why results on one type of mechanical filtration test are not necessarily predictive of results on another type.
Common elements of testing
Mechanical filtration claims include nominal particulate reduction (85 percent) under NSF/ANSI 42 and asbestos reduction, cyst reduction and turbidity reduction under NSF/ANSI 53. All of the tests include using test dust to clog the filters and reduce flow rate, with samples collected at various points in the test based on reduction in flow rate.
Because of this comparability in test protocols, manufacturers often expect results of one of these tests to be predictive of results of one or more of the others. It is not unusual for a manufacturer to ask if passing a cyst reduction test is sufficient to justify a nominal particulate reduction Class I claim or a turbidity reduction claim.
Although the tests have certain similarities, they also have differences that are significant enough to render them non-predictive of each other. It is possible for a specific filter system to pass any one or two tests but fail others.
Differences among the tests
Direct comparison of test procedures for nominal particulate reduction (85 percent), cyst reduction and turbidity reduction can easily be noticed (Figure 1). There are significant differences in these procedures.
Nominal particulate reduction (85 percent) under NSF/ANSI 42 discounts particles that originate from the filter itself. That is, any particulate shedding of the filter media is analyzed, flushed and accounted for, such that particulate counts in the effluents can be directly correlated to influents.
NSF’s procedure is to flush the filters prior to beginning the test until the particle counts in the size range of interest are stabilized or five sets of flush samples have been collected and analyzed. This can involve running a few gallons of water through the test filters or it may mean running tens of gallons.
When effluent samples are analyzed, the background particle counts established through the flushing samples are subtracted from actual effluent counts to arrive at the reported effluent values. Through this procedure, effluent particle counts are tied directly back to the influent challenge.
This procedure can allow a product with significant shedding of particles to achieve conforming results under the nominal particulate reduction (85 percent) test. This test also uses fine test dust, which has a very different particle size distribution with much fewer small particles than the 0-5 micron nominal test dust used in the turbidity reduction test.
Turbidity reduction under NSF/ANSI 53 requires measuring the actual turbidity of effluent samples regardless the source of the turbidity. Turbidity is a measure of the scattering of light passing through the water.
All sizes of particles in the water, including sub-micron and colloidal particles, influence the turbidity measured in filter effluents. Unlike nominal particulate reduction under NSF/ANSI 42, there is no adjustment to the measured effluent turbidity resulting from any carbon fines shedding of the filters.
Any turbidity resulting from fines shed by the filters counts against the filters in terms of the maximum allowable effluent turbidity. Analysis of turbidity values is different from the laser particle counting analysis of particles in specific size ranges used for nominal particulate reduction (85 percent).
Cyst reduction testing under NSF/ANSI 53 is very different from both nominal particulate reduction and turbidity reduction testing. Live Cryptosporidium oocysts or microspheres must be reduced by 99.95 percent. This is a much higher reduction percentage than is required under either of the other two tests.
Particles themselves are in a very specific size range: 95 percent of microspheres are 3.00 µm ± 0.15 µm. Cryptosporidium oocysts range in size from about three to seven µm, with most being about five µm (although they can deform to ‘football’ shapes which make their profile a bit smaller as they pass through filters).
It is possible that filters could allow particles in the Class I size range or small particles contributing to turbidity to pass through, while retaining particles in this size range. Also, enumeration of Cryptosporidium oocysts requires staining and counting under an epiflourescent microscope. Microspheres are fluorescent to begin with, so they are simply counted under the microscope.
This epifluorescent microscopy is highly specific, with no possibility that particles shed from the filters themselves could be contributing to values detected in the effluents. Cyst testing the analysis is very specific with no doubt as to the source of the contribution to the effluent—the only way for Cryptosporidium oocysts or microspheres to show up in the effluent is to go through the filters.
No predictive relationships
NSF has not been able to use any one of these tests to predict results for either of the others. Because of this lack of ability to predict, there is no potential to establish any of these tests as a surrogate for any of the others. They all have different measurement techniques. They all measure particles of different sizes.
Contributions of filter media to particulate matter in the effluent are addressed differently. Percent reduction requirements are different. Because of these and other differences, the tests do not measure the same characteristics of the filter.
The nominal particulate reduction test measures only the ability of a filter to nominally remove particulate matter of a very narrow size range from the influent when introduced as part of a broader distribution of sizes of particulates. The turbidity reduction test measures the amount of light scattering resulting from the presence of particles in the water.
All of the particles present in the effluent contribute to this scattering of light. Cyst reduction requires extremely efficient removal of a specific sized particle that is larger than those measured in the nominal particulate reduction Class I test, with a highly specific analysis focused only the particles of interest.
Important and distinct performance measurements
The differences cited above are part of the reason why there are three distinct and separate test protocols to make these three distinct mechanical filtration claims. Whether testing melt-blown filters, pleated filters using various media, carbon blocks, ceramic filters or other mechanical filter types, different characteristics of the filter are being evaluated under different mechanical filtration tests.
This knowledge is relevant because for various residential, commercial or industrial mechanical filtration applications, certain of these performance measurements may be more or less important.
In residential drinking water treatment, cyst reduction may be a key performance requirement, but it may be much less important for industrial applications. It is critical for manufacturers addressing these markets to understand the specific requirements of the market and note the specific criteria against which the filter will be tested to demonstrate conformance to these requirements in order to have the greatest likelihood of success.
About the author
Rick Andrew is the Operations Manager of the NSF Drinking Water Treatment Units Program for certification of POU/POE systems and components. He enjoys leveraging his more than ten years of experience in this area to help explain the complexities and details of the NSF/ANSI DWTU Standards. Andrew has a Bachelor’s Degree in chemistry and an MBA from the University of Michigan. He can be reached at 1 (800) NSF-MARK or email: Andrew@nsf.org.