How can filter cleanliness be determined?

Abstract: It is general believed that a sand filter starts its life with new, perfectly clean media, which becomes gradually clogged with each filtration cycle, eventually getting to a point where either head loss or filtrate quality starts to deteriorate. At this point the backwash cycle is initiated and, through the combined action of air and water, returns the media to its original perfectly clean state. Reality, however, dictates otherwise. Many treatment plants visited a decade or more after commissioning are found to have unacceptably dirty filter sand and backwash systems incapable of returning the filter media to a desired state of cleanliness. In some cases, these problems are common ones encountered in filtration plants but many reasons for media deterioration remain elusive, falling outside of these common problems. The South African conditions of highly eutrophic surface waters at high temperatures exacerbate the problems with dirty filter media as such conditions lead to the formation of biofilm in the filter media, shown to inhibit the effective backwashing of sand and carbon filters A systematic investigation into filter media cleanliness was therefore started in 2002 at the University of Johannesburg (the then Rand Afrikaans University), involving media from eight South African Water Treatment Plants, varying between sand and sand-anthracite combinations and raw water types from eutrophic through turbid to low-turbidity waters. Five states of cleanliness and four fractions of specific deposit were identified relating to in-situ washing, column washing, cylinder inversion and acid-immersion techniques. These were measured and the results compared to acceptable limits for specific deposit, as determined in previous studies, though expressed in kg/m3. These values were used to determine the state of the filters. In order to gain greater insight into the composition of the specific deposits stripped from the media, a four-point characterisation step was introduced for the resultant suspensions based on acid-solubility and volatility. Results showed that a reasonably effective backwash removed a median specific deposit of 0.89 kg/m3. Further washing in a laboratory column removed a median specific deposit of 1.34 kg/m3. Media subjected to a standardised cylinder inversion procedure removed a median specific deposit of 2.41 kg/m3. Immersion in a strong acid removed a median specific deposit of 35.2 kg/m3. The four-point characterization step showed that the soluble-volatile fraction was consistently small in relation to the other fractions. The organic fraction was quite high at the RG treatment plant and the soluble-non-volatile fraction was particularly high at the BK treatment plant