There are two main types of sand filters that are used for water treatment: rapid sand filters (also known as Rapid Gravity Filters or RGFs) and slow sand filters. Sand filters should not be confused with Roughing Filters, which tend to be horizontal-flow gravel filters used as a roughing treatment for turbid raw waters prior to sand filtration.
The two sand filtration processes differ from each other in several ways. In general however, rapid sand filters are usually fully automated, complex and costly, forming part of a wider treatment process in industrialized countries. They are not known to be suitable for household-level use, as is the case with versions of the slow sand filter, for example the intermittently operated slow sand filter (or IOSFF). For the sake of clarity though, a brief comparison is laid out below. For a fuller explanation of rapid sand filtration, please refer to Schulz and Okun (1984 [ref_01]).
||Rapid sand filters
||Slow sand filters|
|Improvement of water quality
||With pre-treated raw water, a filtrate quality is possible that has less than 1 NTU, 90% removal of coliforms, 50 – 90% removal of Cryptosporidium and Giardia cysts, 10% removal of colour, 5% removal of Total Organic Content (WEDC, 1999 [ref_02]). As a single process, rapid sand filtration was ranked third most effective of all treatment processes, given a range of pathogenic, chemical and aesthetic factors.
||With raw water, a filtrate quality is possible that has less than 1 NTU, 95% removal of coliforms, 99% removal of Cryptosporidium and Giardia cysts, 75% removal of colour, 10% removal of Total Organic Content (WEDC, 1999 [ref_02]). As a single process, slow sand filtration was ranked second most effective of all treatment processes, given a range of pathogenic, chemical and aesthetic factors.|
|Rate of flow
||A flow of between 4 – 21 m/h can be expected from a rapid sand filter, which is somewhere between 20 and 50 times faster than the range of slow sand filtration.
||Flow rates are usually around 0.1 m/h but can increase up to 0.4 m/h. Check out Flow Rates for more information.|
||Rapid sand filters are made using graded sand, sometimes with an additional coarser layer of material on top of the sand to increase the flow rate (for example, anthracite), in which case they become known as dual-media filters. The effective size for rapid filters is usually greater than 0.55mm with a Uniformity Coefficient of less than 1.5.
||Slow sand filters on the other hand, should ideally have an effective size of between 0.15 – 0.35mm, and a Uniformity Coefficient of between 1.5 – 3, but preferably less than 2. |
|Penetration of solid matter
||Penetration of suspended matter into the sand bed is deeper for rapid sand filters, which are usually cleaned by backwashing.
||Solids only tend to penetrate slow sand filter beds by between 0.5 and 2 cm by comparison, allowing more manual methods to be employed for cleaning.|
||Pre-treatment is usually necessary for rapid sand filtration. Such treatment could include coagulation and flocculation, followed by sedimentation.
||No pre-treatment is usually necessary for raw waters with turbidities of less than 50 NTU. Huisman and Wood (1974 [ref_03]) suggest that the best filtration occurs when turbidity is less than 10 NTU. Note that coagulation and flocculation are not appropriate pre-treatments for slow sand filters because floc carryover is possible, which rapidly blocks the sand bed.|
Note that the unit metres/hour can be confusing, and is in fact a compaction of m3/m2/hour; that is, how much water volume can go through a certain area in one hour. Also sometimes this unit is expressed per day rather than per hour. For a fuller explanation of the relationship of water traveling through sand, check out Darcy’s Law explained.
References: (jump back)
Ref 01: Schulz, C. R.; Okun, D. A. (1984). Surface water treatment for communities in Developing Countries. IT, London
Ref 02: Unpublished information supplied by WEDC, 1999.
Ref 03: Huisman, L; Wood, W.E. (1974). Slow Sand Filtration. WHO, Geneva, Switzerland. p.44. Available from IRC