AFM, glass media and sand filtration systems, a comparison of technologies and optimisation of performance


Dr.Howard Dryden

Dated July 2014


Media bed mechanical sand filtration systems comprise of; gravity flow, pressure and moving bed continuous backwash filtration systems.  In all cases the most common mechanical filtration media is quartz silica sand.   The quality of quartz sand is a variable depending upon the country and the location of the deposit.  

There is a requirement for a consistent quality of filter media for all industries using media bed filtration in order to standardise and optimise the filtration process. This aspect becomes more important for filters that have a pressure gradient across the bed such as horizontal filters, or filters that have not been installed on a perfectly level base. The performances of seven different types of filtration media were physically evaluated by IFTS (1) one of the leading independent accredited laboratories in Europe for the evaluation of products used by the water industry.



Sand has been used for over 200 years in Europe as a means of filtering Drinking water. A company in Scotland in 1804 was the first documented report of a company using sand in a slow bed sand filter (2).   Slow bed sand filters typically operate at water flow velocity of 0.1m/hr and use a coarse grade of sand and gravel.  The filters depend on maturation of the sand as a biological filter before they provide adequate mechanical water filtration.


Flow rates

Slow bed sand filters provide excellent water quality and are still used for the treatment of drinking water. Approximately 15% of all water supplies in the UK currently use slow bed filters, but they are being phased out in favour of RGF (Rapid Gravity Filters) and pressure sand filters in order to save space.  RGF filters for drinking water operate at water flow velocity of 6m/hr whereas pressure filters typically operate at 12m/hr.  The water flow velocities of RGF and pressure filters are therefore 60 to 120 times faster than slow bed filters.  The higher water velocities change the bio-dynamics of the filtration process which impacts on filter performance leading to bio-instability and transient wormhole channelling of unfiltered water through the filter bed.

The performance of any media bed will be inversely proportional to the flow velocity, which is a function of the filter diameter, its surface area and bed depth (Darcy’s Law)(3).   The bar graph compares the performance of AFM and Leighton Buzzard sand from England. The slower the filter flow velocity the higher the performance, the relationship is exponential but the coefficient depends on the media characteristics and particle size used for performance evaluation.  One of the key issues in the drinking water industry is ability to remove a parasite called Cryptosporidium, which is almost completely resistant to chlorine and only measures 4 microns in size.   If sand filters are operated at water flows in excess of 12m/hr it becomes increasingly difficult to ensure adequate water quality and the removal of the parasite.

Water treatment systems tend to operate at the highest possible water flow rates in order to save space and reduce capital cost. AFM has been shown to provide performance advantages over sand, which permits higher water flow rates and reduced capita cost of installations.  Typically 50% higher water flow rates can be used with AFM over sand while still maintaining good performance.


















Filtration performance also depends upon filter configuration, horizontal filters save space and maximise surface area.   Bed depth is shallower which reduces adsorption capacity for small particles. Also, a differential pressure gradient across the bed reduces performance when compared to vertical filters that have a consistent pressure gradient and a deep bed.  The differential pressure promotes biofouling of sand, biodynamic instability and transient wormhole channelling; the problems are largely resolved by using AFM which does not suffer from biofouling.

Pressure differential

During the run-phase large solids will accumulate on the top of the filter bed and small solids will penetrate the bed.  Small particles attracted by electrical (Van der Waals) forces may become trapped on the surface of the media. Sand and most media carry a negative charge or Zeta Potential.  In water treatment, coagulants and flocculants such as; Lanthanum chloride, aluminium chloride, ferric chloride, PAC (polyaluminium chloride) or polyelectrolytes may be applied to drop the zeta potential, increase coagulation and flocculation as well as increasing electrical attraction. In some industries  including pre-treatment prior to membranes or the aquarium industry, the use of chemicals would not be advisable.  Reduction of zeta potential and coagulation can nevertheless be achieved by the rapid movement of water, cavitating static mixers such as a ZPM (Zeta potential Mixer) or by slightly increasing redox potential by application of ozone.

In addition to mechanical and electrical attraction, there will also be some degree of molecular sieve filtration. This will be the case with activated carbon, and to a lesser extent with new sand.  The ability of sand to adsorb is a function of the silicon to aluminium ratio and how the molecules are configured.  An example of natural ion exchange molecular sieve sand is the zeolitic sand clinoptilolite (4)(5)

Zeolites are used in water treatment as a mechanical filtration media and also as an ion exchange mineral for the selective removal of ammonium and radioactive nucleotides from freshwater. Zeolites cannot be used for marine systems or water with a high TDS because the competing cations will prevent ion exchange.  In freshwater systems zeolites provide a good substrate for the growth of autotrophic nitrifying bacteria, a characteristic that is likely due to the adsorption of ammonium into the mineral and its availability to be metabolised by autotrophic species such as Nitrosomonas spp. 

Biofouling and worm-hole channelling

The performance of any mechanical filtration media depends upon the passage of water through the filter bed and the ability to remove the collected solids during the back-wash phase.

The zeolite, Clinoptilolite initially provides very good filtration, but the media rapidly biofouls, especially at high water temperatures (above 20oC) in nutrient rich water.   In closed system eel Anguilla anguilla aquaculture at 28oC the clinoptilolite filter bed blocked within 2 hours when operated at 10m/hr filtration velocity (6). Blockage was due to collected solids as well as growth of heterotrophic bacteria on the filter media. The rapid growth rate of bacterium and the production of bacterial alginate exopolysaccharides cause coagulation of the filter bed (6) which leads to transient wormhole channelling.  The alginates are actually advantageous in slow bed filters (7) and can improve filtration performance, in rapid gravity and pressure sand filters the alginates lead to blockage and bio-instability of sand beds. Back-washing will not remove biofilm or prevent biofouling, indeed continuously fluidised sand beds make excellent biofilters for bacterial nitrification (8).

Back-washing, what goes into a filter must come back-out

Proper back-washing is very important. Under German DIN standards the bed should be fluidised and expanded by 20%  for a period of 5 minutes duration. The velocity of the water required to achieve the required bed expansion is a function of the bulk bed density of the media, particle size, shape as well as the temperature and density of the water. Sand with a PSD (particle size distribution) between 0.5 and 1.0mm requires a flow velocity the region of 55m/hr at 28oC for freshwater.  For a marine system due to the higher density of water at 35ppt, the back-wash flow velocity may be reduced to 40m/hr.

Back-washing is critical, any solids or organic matter remaining in the filter bed after a back-wash will simply act as a food source for the growth of more bacterium and production of exopolysaccharides.  However we know that even sand beds fluidised 100% of the time make very good biofilters (8), so back-washing of sand is never 100% effective.  Organic matter and particles will become embedded in the alginate and will remain after a back-wash and will continue to feed heterotrophic bacteria.  Gradually the sand biofilm layer will mineralise with calcium, magnesium, ammonium and phosphate to form calcites or struvite.  The biofilm becomes more stable, alginate production increases and filtration performance gradually decreases until a point when a media change is required.

Glass media and sand

Glass is an aluminosilcate manufactured from silica sand or from the re-melt of glass bottles.  It has a similar chemical composition to sand, but may contain metal oxides such aluminium, or ferric to make amber glass or manganese and chromium for green glass.

Glass as a filter media was used in 1984 by Dr Howard Dryden as an alternative to the zeolite clinoptilolite as a means of filtering water in a RAS (Recirculating Aquaculture System) for eels and Atlantic salmon. The glass was initially used as a feedstock for the manufacture of synthetic zeolites. The glass was subsequently used as a substrate and the surface of the glass was change by a solgel process to give it a hydrophilic high surface area to avoid biofouling while still acting as a molecular sieve similar to clinoptilolite for the adsorption of organics.

The manufacture of filter media provides an opportunity to make a filter media with a specific tailored performance.  The performance can then be quantified and compared against other filter media.  Such an investigation has never been conducted for sand. Given that sand is used to treat more than 99% of our drinking water supply, it is rather surprising that there has been no detailed comparison of sand media performance from different deposits or different countries.

IFTS. The Institut de la Filtration et des Techniques Séparatives is recognised as being the leading institute in Europe for the testing of water filter technology. As part of the development of a new International ISO 14034 standard for ETV (Environmental Technology Verification) of product performance in the water industry, glass media from different manufactures in Europe were evaluated by IFTS(1).

The three basic tests conducted include;

  1. Run phase efficiency
  2. Injected mass test
  3. Back-wash performance

Run Phase particle size removal

Seven different types of glass media and one sand media were tested.  The sand was from the Leighton Buzzard deposit in England.  The silica sand was recognised by IFTS to be one of the best in Europe.  The glass media was provided by different manufacturers of glass granules and glass beads in Europe.

  1. AFM, activated filter media, Scotland, grade 1 and 0
  2. Sand, Leighton Buzzard, England
  3. Garofiltre filter media, France
  4. Astral filter media, Spain
  5. Bioma, filter media, Spain
  6. EGFM filter media, England
  7. Vitrosphere filter media, Germany

The run phase performance test involved the injection of particles of a known particle size directly into the water under controlled conditions.  Particle size analysers were fitted to the test rig in order to check the concentrations and confirm the performance. Two grades of AFM were tested, grade 0 is fine grade media with a psd from 0,25 to 0.50. Grade 1 AFM is typical of most filter grade media with a psd of 0.4 to 1.0. The psd of all the media tested approximated to a standard 16 x 30 mesh size.

The test was performed in a 150mm diameter column with a 900mm bed depth at a flow velocity of 20m/hr and temperature 23 deg C.  At 5 micron particle size, AFM grade 1 was removing >97% of all particles and sand, 72%. Vitrosphere filter media is manufactured from glass spheres, showed zero particle removal at 5 microns.

Run phase Injected mass test

The injected mass test was run at the same time as the particle size removal test. The differential pressure across the filter bed was monitored.  The data should be viewed in conjunction with the particle size removal performance. For example, Vitrosphere did not show any increase in differential pressure against injected mass, this was because the solids were simply passing through the filer bed.

The sand produced a smooth curve and predictable performance, AFM tracked the sand curve but at a slightly higher starting differential pressure.  Astral and EGFM both exhibited slippage or discharge of solids back into the water, but in the case of EGFM this was above a differential of 0.7 bar.

Filter beds will remove particles from the water. Under ideal conditions, as solids collect in the bed the differential pressure will increase and the bed will block. An undesirable feature would be for the bed to discharge collected solids back into the water such as in the case of the Astral filter media.

Sand and glass media will mechanically remove large particles from the water, in addition small particles that could pass through the filter are adsorbed by electrostatic attraction.  There will be a finite capacity for adsorption, and when this capacity is reached the filter media may discharge the solids back into the product water.  The capacity of a filter bed to hold onto solids is a function of the filter media, water flow rates and differential pressure.  It is therefore desirable to operate filters at as slow a velocity as possible and not to exceed a differential pressure of 0.5 bar.

It should be noted that the tests were conducted with new sand and new glass filter media. Sand will become a biofilter and both mechanical filtration performance as well as electrostatic attraction will decline as the biofilm develops.  Non-activated glass media will also be subjected to biofouling and will deteriorate albeit at a slower rate than sand.


AFM was the only activated filter media tested. Sand and crushed glass typically has a surface area of 3000m2/tonne with a PSD of 0.5 to 1.0mm. AFM as measured by argon gas adsorption has a surface area close to 1,000,000m2 per metric tonne.  The surface area is 300 times greater than untreated glass.  The higher surface area determines the ability of AFM to remove small particles and its ability to hold on to the particles during the run phase.

Back-wash, what goes in must come out

A mass balance was conducted on the data, it is very important to achieve as close as possible to a 100% back-wash efficiency.    Solids and organic matter remaining in the filter after a back-wash will act a food source for the growth of bacteria and production of alginates leading the bed coagulation and channelling of unfiltered water through the bed.

Sand and AFM demonstrated close to 100% back-wash efficiency and almost identical back-wash profile curves.  All of the non-activated glass filter media showed a reduced back-wash performance with 10% to 50% of solids remaining in the filter bed.

Summary and conclusions

The performance of a mechanical filtration system will depend on the quality of the media, design of the filter and on operating criteria.  For best performance and water clarity with AFM, vertical pressure or RGF filters should be used. The run phase should be less than 20m/hr and differential pressure should never exceed 0.4 bar.  It is also best to back-wash the media at least once a week at a water flow that fluidises the bed by more than 20% for a period of 5 minutes, or until the back-wash water runs clear.

There is a wide choice of filter media available; the sand tested was the best sand available, and the best performing filter media as shown by IFTS data was AFM activated filter media.  The results reflect the performance of new sand, as sand ages, it will gradually become a biofilter and mechanical filtration performance will deteriorate over the following months.

The clarity of water is a function of the Zeta potential of all particles in suspension.  As redox potential increases, zeta potential decreases, when zeta potential is zero you have the lowest turbidity.  Prior to filtration the water should have a redox potential over 300mv, this may be achieved by aeration of the water. For manganese removal the potential should be 500mv, in addition to aeration, chemicals such as ozone or chlorine dioxide may be required to remove manganese.

We recommend the use of a ZPM prior to filtration; the cavitating ZPM static mixer will drop the zeta potential and can raise the redox potential. The injection of cationic coagulants and flocculants such as APF (All-Polyfloc) will drop the zeta potential, clarify the water, increase redox potential and allow AFM to remove micron and sub-micron particles and even chemicals from solution.  AFM grade 0 will remove >99.7% of all particles down to 3microns, AFM grade 1 will remove >97% down to 5 microns, however when AFM grade 1 is combined with a ZPM, coagulation and flocculation it will nominally remove most particles down to 0.1microns as well as much smaller particles and even chemicals from solution.

80% of all disease in the developing world is from drinking water, 60% of the infections are from parasites.  AFM grade 0 with no coagulation, flocculation or even chlorination has the capacity to remove the parasites using a very simple gravity flow filter.


1. IFTS (2014)Institut de la Filtration et des Techniques Séparatives, Sièges Social, Adresse de livraison, Rue Marcel Pagnol 47510 FOULAYRONNES, France

2. WHO.  World Health Organization

3. H. Darcy (1856), Les Fontaines Publiques de la Ville de Dijon, Dalmont, Paris.

4. Dryden, H. T. and L. R. Weatherley (1987). "Aquaculture treatment by ion-exchange: II. Selectivity studies with clinoptilolite at 0.01N." Agricultural Engineering 6: 51-68.

5. Dryden, H. T. and L. R. Weatherley (1987). "Aquaculture water treatment by ion-exchange: I. Capacity of Hector clinoptilolite at 0.01-0.05N." Agricultural Engineering 6: 39-50.

6. Dryden H.T. (1984). The removal of ammonium by selective ion exchange filtration using the natural zeolite Clinoptilote. PhD 1984 Heriot Watt University.  Dept of Chemical Engineering

7. Visualisation of the establishment of a heterotrophic biofilm within the schmutzdecke of a slow sand filter using scanning electron microscopy. Biofilm, Volume 6, Paper 1 (BF01001) 2001

8. Thomas M. Losordo1 (April 1999), Michael P. Masser2 and James E. Rakocy. Recirculating Aquaculture Tank Production Systems A Review of Component Options.  SRAC Publication No. 453 Southern Region Aquaculture Centre.

Latest Articles