Slow sand filter and charcoal biofilter

Summary

A slow sand filter passes water at a very low velocity (on the order of 0.1 m/h) through a fine sand bed 0.5 to 0.9 m deep. On the surface a biological layer (schmutzdecke) grows that traps and digests most bacteria, protozoa and helminths and leaves the water clear (turbidity below 1 NTU). This article covers the intermittent-flow household version (biosand type, 24 to 72 L/day per filter) and the continuous-flow community version (1 m² of bed gives about 2400 L/day). After the sand there can be a granular charcoal bed that improves taste, odor and color and retains part of the organic contaminants. Neither one replaces final disinfection: the filtered water is boiled, chlorinated or exposed to the sun before drinking.

Prerequisites

• Knowledge: measure volumes and times with a graduated container and by counting seconds; disinfect the filtered water (see Boiling water and the related articles on chemical and solar disinfection). Making charcoal: there is no article yet; section F gives a minimal method.
• Materials (household filter):
  • 1 watertight container 90 to 100 cm tall and 25 to 35 cm inside diameter: unglazed fired-clay jar or one with lead-free glaze, a wooden barrel, a concrete or food-grade plastic bucket.
  • Sand: about 40 L already sieved and washed (start with 60 to 80 L raw: washing and sieving remove part of it).
  • Separating gravel (0.7 to 6 mm): about 4 L. Drainage gravel (6 to 12 mm): about 4 L.
  • 1 outlet pipe 1 to 2 cm inside diameter: bamboo or reed cane with no internal nodes, fired-clay, copper or plastic pipe.
  • Sealant for where the pipe passes through the wall: clay with resin or beeswax (N1), lime mortar (N2) or cement (N3).
  • 1 diffuser: thin board, fired-clay plate or sheet metal, with holes about 3 mm.
  • 1 lid and 1 clean container with lid to collect the filtered water.
• Materials (household charcoal biofilter): 25 to 30 L of already crushed, sieved and washed hardwood charcoal; 1 container of 35 to 40 L with the same type of raised outlet pipe.
• Tools: buckets of 10 to 20 L (N1), woven-fabric or basketry sieves (N1), mortar or stones for crushing (N0), ruler or knotted cord with knots every 1 cm (N1), counting seconds (N0). For the community version: hoe and shovel (N1), clay (N1) or masonry (N2), sluice gate or wooden plug (N1).
• Time and people: household filter, 2 people for 1 to 2 days (mostly sieving and washing the sand), plus 2 to 4 weeks of ripening. Community filter of 2 m², 6 to 10 people for 1 to 3 weeks depending on the terrain, plus ripening.

How it works

What the sand retains. Water crosses the gaps between grains (in fine sand, a few tenths of a millimeter). Particles are caught by straining at the surface, by settling inside the pores and by sticking to the grains. That retains turbidity, helminth eggs and protozoan cysts, but lets through many bacteria and almost all viruses.

The biological layer does the fine work. In the top 1 to 2 cm a sticky film of algae, bacteria, protozoa and organic matter forms over weeks (schmutzdecke, German for “dirty layer”). There pathogens get stuck, are devoured by protozoa or die from lack of food. That layer needs:

1. Contact time: WHO/IRC recommends designing for 0.1 m/h, with an accepted range of 0.1 to 0.4 m/h (Visscher et al. 1987; Global WASH Cluster). Faster than that, pathogens cross the bed.
2. Oxygen: it arrives dissolved in the water or, in the household filter, by diffusion through a layer of standing water 5 cm deep (CAWST 2009). More water on top suffocates the layer; less, and it dries out from heat.
3. Not drying out or being poisoned: if the bed is left without water or chlorinated water is poured in, the layer dies.

Grain size. The effective size (d10) is the sieve opening through which 10 % of the weight passes; the uniformity coefficient is CU = d60/d10. For a slow sand filter: d10 = 0.15 to 0.30 mm and CU < 5, preferably < 3 (Visscher et al. 1987; Pennsylvania gives 0.15 to 0.35 mm and CU 1.5 to 3). Coarser sand: water passes too fast. With many fines (silt, clay): it clogs within days. Too little uniformity: fine grains fill the gaps between coarse ones and the flow rate collapses.

Why the outlet rises above the sand. The discharge point sets the minimum water level inside the filter (communicating vessels). If it is above the sand, the bed never empties and the layer never dries out. But if the pipe drops too far outside it can siphon: it keeps pulling water down below the elbow's level and leaves the sand exposed to air. This is the part most often done wrong.

What the charcoal does. Charcoal has an enormous internal pore surface where organic molecules are adsorbed: those that give taste, odor and color, pesticide residues and other organics. It also destroys free chlorine. The higher the carbonization temperature and the finer the grain, the greater the capacity: charcoal made at ≥ 850 °C in a gasifier worked well, though commercial activated carbon outperformed it by a factor of 2 to 34 (average 15) (Kearns et al. 2021).

Order of the treatment train: (settling if the water is turbid) → sand → charcoal → disinfection → covered storage.

What to expect from a ripened sand filter. A 1-log reduction is 90 %, 2-log is 99 %, 3-log is 99.9 %.

WHO's lower figures correspond to poorly operated filters (unripened, freshly scraped, too fast): retention drops a lot. Even 95 % lets through 50 bacteria out of every 1000: it greatly reduces diarrhea, it does not guarantee safe water. Hence the need for final disinfection.

Inlet limits: recommended turbidity up to 10 NTU, with peaks of up to 50 NTU in the community filter (Global WASH Cluster); the biosand accepts up to 50 NTU (CAWST 2009). Above that, settle it first.

Procedure

A. Sizing

1. Calculate demand. If capacity is scarce, filter only drinking and cooking water: 7.5 L per person per day. If all household water is filtered, basic access is around 20 L per person per day (Howard and Bartram 2003). Success criterion: a liters/day figure written down.

2. Choose the type. A biosand gives 24 to 72 L/day in 4 batches of 12 to 18 L (CAWST 2009).

   • Up to ~70 L/day: 1 biosand (at 7.5 L/person/day, up to 9 people: 9 × 7.5 = 67.5 L).
   • Between 70 and ~290 L/day: 2 to 4 biosands in parallel, one per 70 L/day (4 × 72 = 288 L).
   • More than ~300 L/day or collective use: continuous-flow community filter. Success criterion: type and number of filters decided.

3. Community filter area. Area (m²) = flow rate (m³/h) / velocity (m/h), at 0.1 m/h and 24 h/day. So, 1 m² gives 0.1 m³/h = 2400 L/day: 120 people at 20 L/day or 320 at 7.5 L/day. Worked example: village of 200 people at 20 L/day.

   • Demand: 200 × 20 = 4000 L/day = 4 m³/day = 4 / 24 = 0.167 m³/h. Area: 0.167 / 0.1 = 1.67 m².
   • 2 identical units to clean one without cutting off supply: 2 × 0.9 m² (0.95 × 0.95 m). Under normal operation they run at 0.167 / 1.8 = 0.093 m/h; with one stopped, the other runs at 0.167 / 0.9 = 0.19 m/h, within 0.1 to 0.4 m/h.
   • Flow rate to measure: 167 L/h = 2.8 L/min total, 1.4 L/min per unit: a 10 L bucket in about 7 min.
   • For drinking and cooking only (7.5 L): 1500 L/day = 0.0625 m³/h → 0.63 m² → 2 × 0.32 m² (0.57 × 0.57 m). Visscher et al. (1987) give 5 to 200 m² per unit: the 0.3 to 0.9 m² units are below what's typical. They work, but are scraped from the edge, without stepping inside. Success criterion: area, units and flow rate per unit recorded.

4. Household filter flow rate. Maximum load 600 L/h per m² (CAWST 2009): maximum flow rate (L/h) = area × 600. Container of 30 cm: area = 3.14 × 0.15² = 0.0707 m²; maximum = 42 L/h = 0.7 L/min. The CAWST v10 filter works at 0.4 L/min maximum with a full reservoir. Since 0.7 L/min is 1.75 times that value, aim for 0.4 to 0.5 L/min. Success criterion: target flow rate for your container recorded.

5. Household filter heights (bottom to top; layers according to CAWST 2012 and Global WASH Cluster):

   Layer	Thickness	Material
   Drainage gravel (surrounds the pipe mouth)	5 cm	6 to 12 mm
   Separating gravel	5 cm	0.7 to 6 mm
   Filter sand	45 to 50 cm	< 0.7 mm, no fines
   Standing water layer	5 cm (4 to 6 cm)	—
   Gap to the diffuser	just enough that it doesn't touch the water (5 cm in the calculation)	—
   Reservoir above the diffuser	whatever a 12 L batch occupies	—

   For a 30 cm diameter, 12 L occupies 12,000 cm³ / 707 cm² ≈ 17 cm. Total ≈ 5 + 5 + 50 + 5 + 5 + 17 + 5 cm of freeboard ≈ 92 cm. Success criterion: a scale drawing with the measurements, including the outlet elbow's height = sand surface + 5 cm.

6. Check the pore volume. Each batch must fit within the sand's pores so it stays inside the bed during the pause. Porosity of silica sand = 1 − bulk density / grain density = 1 − 1560 / 2650 ≈ 0.41 (values from Thames Water 2005). Pore volume = 0.0707 m² × 0.50 m × 0.41 = 0.0145 m³ = 14.5 L ≥ 12 L. Success criterion: batch ≤ pore volume; if not, reduce the batch size.

B. Obtaining, sieving and washing the sand and gravel

1. Choose the source. In order of preference (CAWST): crushed rock, dry quarry sand and river sand taken high on the bank. River sand usually carries organic matter and pathogens from excrement. Look for hard quartz grains that don't crumble when rubbed. Avoid beach sand (salt and shells), sand near latrines, dumps or mine tailings, and clayey sand. Success criterion: a squeezed handful of damp sand does not form a ball that keeps the imprint of the fingers (if it does, it carries too much clay).
2. Sieve out the coarse material through a sieve of about 0.7 mm. Without proper sieves: cloth, mosquito netting or regularly woven basketry; opening ≈ (10 mm / threads per cm) − thread thickness (10 threads/cm with 0.3 mm thread → 0.7 mm). Also a thin board or leather perforated with hot wire of known thickness. Keep what is retained for the gravels. Success criterion: the sieved material passes the measurement in step 3 and, at the end, the flow-rate test (C.7).
3. Measure the grain size without sieves. Line up grains touching each other along 1 cm and count them: average diameter ≈ 10 mm / number of grains. Look for typical grains of 0.2 to 0.5 mm (20 to 50 grains per cm), consistent with d10 = 0.15 to 0.30 mm and CU < 3. Do this with 3 samples from different spots in the pile. Success criterion: all 3 give 20 to 50 grains/cm, without two very different populations (dust plus coarse grains).
4. Remove fines by decantation. In water at 20 °C, a natural quartz grain falls at about 7.5 mm/s if it measures 0.1 mm and at about 10 mm/s if it measures 0.12 mm (Ferguson and Church 2004). Practical rule, valid for any container: after stirring, wait 1 s for every cm of water above the sand (30 cm → 30 s) and pour off the turbid water. What settles at less than 10 mm/s is removed: grains smaller than about 0.12 mm, below the target d10. In water at 10 °C the viscosity is 30 % higher and grains fall more slowly: wait 36 to 40 s per 30 cm. Procedure: fill the container with one third sand and the rest water (Thames Water 2005); measure the height of water above the sand with the knotted cord; stir vigorously for 10 s; wait according to the rule; pour off without dragging the sand off the bottom; repeat. Success criterion: after stirring and waiting, the water lets you see the sand at the bottom and no longer carries dust when poured off.
5. Check the silt content without over-washing. Bottle test (Pyper and Logsdon 1991, in Thames Water 2005): 100 mL of sand in a clear jar, water up to 200 mL, shake hard and let it settle for 20 min. Turbid water, a silt band and the sand form. Silt (%) = silt height / sand height × 100; wash it if it exceeds 1 % (Visscher et al. 1987). In the biosand, don't wash more than necessary: sand that is too clean lets water through too fast, and the final call is made by the flow-rate test (C.7). Success criterion: silt band barely visible, ≤ 1 % of the sand height.
6. Prepare the gravels. Sieve what was retained in step 2 into two fractions: 0.7 to 6 mm (separating) and 6 to 12 mm (drainage), checking the largest grains in each pile with the ruler. Wash both until the water runs clear. Success criterion: wash water clear on the first pour; no flat or elongated stone larger than 12 mm.
7. Store the fractions separately and labeled, covered, on a clean surface. Success criterion: 3 identified piles, free of soil, leaves or animal droppings.

C. Building the household filter (biosand, N1)

1. Prepare the container. Fired clay: fill it with water for 24 h to check for leaks. Wood: untreated and non-toxic, let it swell with water for 2 to 3 days until it stops seeping. Success criterion: no visible leaks and the water level does not drop more than 1 cm in 24 h.
2. Place the outlet pipe. Two ways:
   • Internal (preferable): a vertical pipe inside the container, with its lower mouth 1 to 2 cm from the bottom, running up against the wall, passing through it at the height “sand surface + 5 cm” and exiting through an elbow.
   • External: a hole near the bottom and a pipe that runs up outside to that height and then exits. Avoid siphoning: after the elbow, the spout must drop only a little (5 cm at most) or have a small hole (2 to 3 mm) at the top of the elbow to let air in. Seal the pass-through in the wall with clay and resin or wax (N1), lime mortar (N2) or cement (N3), inside and out. Cover the lower mouth with cloth or mesh so gravel cannot get in. Success criterion: with the container full of water, water flows out of the spout and stops when the inside level reaches exactly the elbow's height, without dropping further over the next 10 min; no leaks from the seal after 12 h.
3. Fill with water first and pour the media into the water, so as not to leave air pockets. Mark the levels on the wall beforehand. Pour in 10 to 15 cm of clean water, add the drainage gravel up to 5 cm, level it, and add the separating gravel on top up to 5 cm. Success criterion: flat layers (± 1 cm) with water always above them.
4. Add the sand by handfuls inside the water, adding water as needed, up to 45 to 50 cm. Level it with your flat hand, without pressing. Success criterion: flat surface 5 cm (4 to 6 cm) below the elbow.
5. Place the diffuser, with holes about 3 mm spread over its whole surface, resting on a ledge, on wedges or on the wall, above the standing water layer without touching it. Success criterion: when a 10 L bucket is poured all at once, the sand shows no crater or channels.
6. Put on the lid and, under the spout, the collection container on a stand, covered around it with a clean cloth. Success criterion: the stream falls inside without splashing or touching the rim.
7. Flow-rate test. Fill the reservoir with the full batch (12 L). As soon as it starts to come out, time how long it takes to fill a glass of known volume. Flow rate (L/min) = volume (L) × 60 / seconds. Example: 1 L in 150 s → 0.4 L/min. Success criterion: 0.4 to 0.5 L/min in a 30 cm container (≤ 0.4 in a CAWST v10), and no less than half that value so the family doesn't tire of waiting.
   • Too fast: sand is coarse or over-washed. Replace the top 5 to 10 cm with finer or less-washed sand and repeat.
   • Too slow: too many fines. Remove the top 5 to 10 cm, wash them more (B.4) and repeat.
8. Check the standing water layer. When water stops coming out, measure the height of the water above the sand. Success criterion: 5 cm (4 to 6 cm; CAWST 2009). If not, correct the sand's height (not the pipe's); if it drops below 4 cm with the sand well leveled, the pipe is siphoning (see Common mistakes).

D. Ripening and daily use

1. Use the filter every day with the same water source from day one; changing sources reduces performance (CAWST 2009). Success criterion: daily log of batches (one mark per batch).
2. Ripen the biological layer: up to 30 days in the biosand (CAWST 2009); in the community filter, up to 3 weeks with new sand (Visscher et al. 1987), longer in cold water. During that time the water comes out clear but is not treated: always disinfect it. Success criterion: the flow rate drops somewhat and stabilizes, and a slightly sticky brown or greenish film appears on the sand.
3. Respect the pause between batches: at least 1 h from when water stops flowing; 6 to 12 h is recommended and 48 h is the maximum (CAWST 2009). Typical pattern: 4 batches per day. Success criterion: it's never refilled while still dripping; never more than 2 days pass without a batch.
4. Pretreat turbid water. Above 50 NTU, strain through a folded cloth or let it settle (CAWST 2009). Without a turbidimeter, CAWST's bottle test (2009): fill a clear 2 L plastic bottle with the water and set it over large printed letters; if they can be read looking down through the bottle from above, it's probably below 50 NTU. Success criterion: the letters can be read; if not, settle it first.
5. Never pour in chlorinated water, soap or hot water: they kill the biological layer. Success criterion: the incoming water does not smell of chlorine.

E. Maintenance

Household filter: “swirl and dump” (CAWST). Only when the flow rate no longer meets the household's needs or drops below 0.1 L/min (CAWST manual), not on a schedule.

1. If the filter is empty, pour in about 4 L of water (CAWST). Remove the lid and the diffuser. Wash your hands.
2. Stir in circles with your palm, only the sand surface, without pushing your fingers down toward the gravel.
3. Scoop out the dirty water with a cup and dump it away from the source and the filter. Repeat until the flow rate recovers.
4. Level the sand, put the diffuser and lid back on. Clean the spout on the outside with water and soap or chlorinated water, without letting it get into the filter. Success criterion: the flow rate returns close to that of the initial test. For a few days the filter retains less well while the layer rebuilds (CAWST 2009): disinfect as always.

Community filter: scraping

1. When: when, with the outlet fully open, the flow rate falls below the design value or the water reaches the overflow. This ranges from several weeks to a year depending on the water (Global WASH Cluster).
2. Partially drain it: close the inlet and lower the level to about 10 cm below the sand surface (Global WASH Cluster). Success criterion: the visible top sand is moist, not dry.
3. Scrape the top 1 to 3 cm with a flat shovel or board (Thames Water 2005), in strips and evenly, removing it in baskets. On units smaller than about 2 m², work from the edge; on larger ones, step only on boards. Do it in a single day. Success criterion: clear, uniform sand appears underneath.
4. Refill from below with already filtered water: open the connecting pipe to the twin unit (H.6) and let the water rise through the drain. If there is no filtered water, do not fill from below with settled water (it seeds pathogens into the drain): fill from the top instead, slowly, pouring onto the flat stone (Thames Water 2005). Success criterion: the level rises a few cm per minute, without bubbles or craters, until it covers the sand.
5. Re-ripen: after scraping, allow at least a few days, and more after adding new sand (Global WASH Cluster; Visscher et al. 1987 gives 1 to 2 days after scraping and up to 3 weeks with new sand). Meanwhile, dump the water or disinfect it. Success criterion: outlet turbidity and flow rate back to their usual values.
6. Wash and store the scraped sand (B.4) in a covered pile.
7. Recharge the bed when it drops to 0.5 to 0.6 m (Visscher et al. 1987; Global WASH Cluster), every few years. Trench technique (Thames Water 2005, from Huisman and Wood 1974): remove the old sand down to the gravel in a strip, put washed sand at the bottom and the old sand on top, and continue strip by strip. The old sand, already colonized, ends up on top and shortens re-ripening. Success criterion: bed back to 0.8 to 0.9 m, level, and design flow rate recovered.

F. Preparing the charcoal for the filter

This also serves for the charcoal cartridge of household ceramic filters.

1. Wood. Hard, dense and dry (holm oak, oak, beech, fruit trees or coconut shell). Never treated, painted, particleboard or varnished. Success criterion: firewood with no rotten bark that sounds dry when struck.
2. Carbonize it. The best charcoal for water comes from a TLUD gasifier (top-lit updraft: lit from the top, with rising draft; sheet-metal drum, N2-N3): the one made at ≥ 850 °C adsorbed much better than low-temperature charcoal (Kearns et al. 2021). Without sheet metal, minimal N1 method in a pit or mound (FAO 1983, chap. 5 and 6):
   • Stack the firewood tightly, filling gaps with thin sticks, in a pit or around a central pole. Cover with leaves and 10 to 20 cm of sandy soil; leave a gap at the top and air inlets at the base (6 to 10 in a 4 m diameter mound).
   • Light it by dropping embers through the top gap. Thick white smoke: the fire has caught. Cover with soil any crack from which flame escapes.
   • When the smoke decreases and turns bluish and then almost transparent (days, in a 4 m mound), carbonization is finished and what is burning is now the charcoal itself: seal all inlets and the top gap with soil or clay.
   • Let it cool sealed (2 to 3 days in that mound). Good-practice yield: 1 kg of charcoal per 4 kg of air-dried firewood. This charcoal is made at a lower temperature (about 550 °C in an improved mound, according to FAO) and unevenly: it adsorbs less than gasifier charcoal and will remove less taste and odor. Set aside the under-carbonized pieces from near the smoke outlet. Success criterion: pieces black inside, with no brown core, that ring metallically when struck, leave little soot, and break with a shiny fracture.
3. Cool without air. If you put it out with water, do so outdoors and away from the steam (see Safety). Spread the charcoal in the shade. Success criterion: plunging a hand into the pile shows no heat anywhere, and none the next day either; only then bag it.
4. Crush it. Dampen the charcoal so it doesn't raise dust and crush it in a mortar or between stones inside a sack. Success criterion: almost all of it in grains the size of a pea or smaller.
5. Sieve it. Aim for grains of 4 to 8 mm. Kearns et al. (2021) model a full-scale adsorber (bed 58 cm in diameter and 30 cm deep) with 4.5 mm grains and 2.5 h of contact time; Thailand's community system, in use since 2008, uses 7.5 mm grains with about 18 h of contact time. Finer grain adsorbs more (going from 0.165 to 0.059 mm multiplied capacity by 3.5 to 3.9 in the laboratory), but it clogs and is lost during washing. Crush again what doesn't pass through 8 mm and discard the dust that passes through a mosquito net. Success criterion: grains of similar size, with no dust when a dry handful is rubbed.
6. Wash and saturate. In a bucket with water, stir, wait 30 s and pour off the black water along with whatever floats. Repeat until it comes out nearly clear. Leave it soaking at least overnight so water fills the pores. Success criterion: rinse water light gray and almost all the charcoal at the bottom; discard whatever keeps floating.
7. Decolorization test (useful for comparing charcoals and for monitoring the bed, G.6): in two identical glasses put 200 mL of water with 1 drop of vegetable ink or strong tea; add to one a handful of the charcoal being tested and to the other an equal handful of reference charcoal; shake for 10 min and let it settle. Success criterion: the good charcoal leaves the water noticeably clearer than the reference, or the same if the reference is new charcoal.
8. Steam activation (optional; when in doubt, skip it). Industrial activation works between 700 and 1000 °C and opens more pores by burning part of the charcoal. There is no reliable data on how much it improves charcoal already made at high temperature: if it comes from a gasifier, it probably isn't worth it. If you do it (N2), read Safety first:
   • Use an iron container, never clay, filled with grains up to 2/3, with a lid pierced by a short vent and by an iron pipe at least 1 m long with a funnel on top to add water far from the fire.
   • Heat until the container is red-hot (cherry red to orange, ≈ 800 to 900 °C, verificar) and keep it there for 1 to 2 h (verificar). Light the gas coming out of the vent.
   • Pour very little water through the funnel each time (about 50 mL every 5 to 10 min, verificar), with your face turned away from the vent and the funnel, because steam comes out in bursts.
   • Stop adding water at least 15 min before finishing and keep heating until there is no flame at the vent (gas exhausted). If you sealed it with gas still inside, air entering as it cooled could ignite the mixture, blow off the lid and throw embers.
   • Only then remove it from the fire, seal the vent and the funnel with clay or damp sand and let it cool.
   • Do not open it until you can rest your hand on the lid. Success criterion: lighter, more porous grains (weigh before and after) that, once washed, outperform the same unactivated charcoal in the decolorization test (step 7).

G. Assembling and using the charcoal biofilter

1. Place it after the sand, never before. Success criterion: the water entering the charcoal is already clear.
2. Size it by contact time. The beds described (model and field) work with 2.5 to 18 h of contact (step F.5). In the intermittent filter, make sure each full batch stays inside the charcoal during the pause (6 to 12 h): bed pore volume ≥ batch size. Measure the porosity between grains: fill a 1 L jar with already-saturated charcoal, add water to the brim and measure how much went in; porosity = liters of water / 1 L. Example with 0.45: 12 / 0.45 ≈ 27 L of charcoal, which in a 30 cm container occupies 27,000 cm³ / 707 cm² ≈ 38 cm. Success criterion: porosity measured, bed calculated and container chosen.
3. Assemble the container like the biosand, but simpler: 5 cm of drainage gravel around the pipe mouth, the charcoal on top and the outlet elbow above the surface of the charcoal, with the same precaution against siphoning (C.2). On top, a cloth or perforated plate. Fill with water first and pour the charcoal in. Success criterion: with the filter at rest, a visible layer of water remains above the charcoal and nothing floats.
4. Rinse before use: run through 2 or 3 batches and discard them. Success criterion: it comes out with no gray color or particles.
5. Always disinfect after the charcoal: boil, chlorinate or SODIS. If you chlorinate, do it at the outlet of the charcoal, which removes chlorine. Success criterion: the stored water smells slightly of chlorine, or was boiled.
6. Replace the charcoal when the taste or odor it used to remove comes back, or when, in the monthly decolorization test (F.7) with a handful taken from the top layer of the bed, it decolorizes clearly less than freshly prepared new charcoal. There is no published interval: Thailand's system has gone through several replacement cycles, but Kearns et al. (2021) do not give months. Do not reuse charcoal used for water without reactivating it; burn it outdoors, away from people. Success criterion: loading date and result of each monthly test recorded on the container.

H. Building the continuous-flow community filter (N1-N2)

1. Choose the site: above the flood zone, downhill from the intake and uphill from the reservoir, and at least 30 m from latrines and cesspits (Sphere 2018), farther from animal pens if the terrain drains toward the filter. Success criterion: water flows from the intake to the filter and to the reservoir by gravity.
2. Excavate or build the tank. Height: 0.3 to 0.5 m of drain and gravel + 0.8 to 0.9 m of sand
   • supernatant water + 0.2 to 0.3 m of freeboard (Visscher et al. 1987; Thames Water 2005). WHO/IRC recommends 1 m of supernatant water; with less it can still work, but there is less head to push the water through and it must be scraped more often. Waterproof it with puddled and compacted clay in layers (about 5 cm per layer, verificar) or with masonry set in lime mortar (N2). Success criterion: full of water, the level does not drop more than 1 cm/day aside from evaporation; if it drops more, add another layer of clay.
3. Drain: pipes 6 cm in diameter spaced 1 m apart (Visscher et al. 1987), of fired clay laid end to end with the joints slightly narrower than the finest stones of the gravel layer covering them, or a channel of flat stones, with a slight slope toward the outlet. Success criterion: water poured in at the high end reaches the outlet without pooling.
4. Support gravel, bottom to top (Visscher et al. 1987, in Thames Water 2005): 16 to 23 mm, 15 cm; 4 to 5.6 mm, 10 cm; coarse sand 1 to 1.4 mm, 10 cm. Wash them until the water runs clear. Success criterion: flat layers; a handful of filter sand poured with water onto the top layer does not disappear between the grains.
5. Sand: 0.8 to 0.9 m initial thickness (Visscher et al. 1987), poured into the water as in C.3-C.4. Mark the initial and minimum level (0.5 to 0.6 m) on the wall. Success criterion: flat surface (± 2 cm).
6. Outlet and connection between units: the drain leads to a chamber whose weir sits slightly above the sand surface (Huisman and Wood 1974, in Thames Water 2005), so the bed never empties; in front, a wooden plug or gate regulates the flow. Connect the drains of the two units with a 6 cm pipe with a plug so one can be filled from below with filtered water from the other. Success criterion: with the inlet closed, water stops coming out and stays above the sand; with the plug closed, no water passes from one unit to the other.
7. Inlet: an overflow in the wall at the maximum supernatant water level returns the excess to the stream; the inlet pours onto a slab or flat stone. Success criterion: with the inlet fully open, the level stabilizes at the overflow and the sand does not move.
8. Initial filling from below with already-filtered water (from the twin unit through the connecting pipe, or from another filter poured into the chamber with the gate open), until it covers the sand; if only unfiltered water is available, fill from the top over the flat stone (Thames Water 2005). Afterward open the normal inlet. Success criterion: the level rises a few cm per minute, with no bubbles.
9. Set the design flow rate with the gate, measuring with a bucket and a clock (A.3: 10 L in about 7 min per unit). Check it every day (Visscher et al. 1987) and open it up little by little as the layer slows the flow; a large sudden adjustment worsens the water quality (Thames Water 2005). Success criterion: the bucket fills in the design time at the daily check.
10. 24 h operation toward a covered reservoir of at least 40 % of daily production (Visscher et al. 1987; in the example, 1.6 m³), from which water is drawn with a tap. During ripening, dump the water or disinfect it. Success criterion: the reservoir never runs empty and the filter never stops.
11. Cover the filter with a small roof of branches, reed matting or boards: light favors algae, which clog the filter sooner (Pennsylvania manual). Success criterion: no floating algae or excrement in the water.

Verification

With N1 means, no laboratory:

• Clarity: the filtered water looks clearly cleaner than the inlet water. Sharply seeing a black mark through 30 cm of water indicates improvement, not that 1 NTU has been reached; a turbidimeter is needed for that.
• Odor and taste: no rotten or moldy smell. If it smells rotten, the layer is oxygen-starved: check the standing water layer (5 cm) and the maximum pause (48 h).
• Flow rate: record it every week. A slow decline is normal; a sudden increase indicates a disturbed bed, a crack or a channel next to the wall (water is not being filtered).
• Standing water layer of 5 cm (4 to 6 cm) in the biosand; sand always covered in the community filter.
• Charcoal: monthly decolorization test (F.7).
• Health: diarrhea decreasing in the household or village compared with before (a delayed but real indicator).

With more resources (N3-N4):

• Turbidity of the outlet < 1 NTU with a turbidimeter.
• Hydrogen sulfide test (paper strip in a 20 mL jar): incubate at 25 to 35 °C and read at 24 and 48 h; if it turns black, fecal contamination is likely. Below about 20 °C it takes much longer and can give false negatives (Pillai et al. 1999).
• E. coli by membrane filtration or chromogenic substrate: WHO's target is 0 in 100 mL (WHO 2011, table 7.10), which in practice is only achieved with final disinfection.
• Charcoal exhaustion: pour water that smells of chlorine directly onto the charcoal bed, never through the biosand (D.5). If the outlet of the charcoal smells of chlorine, it is exhausted.

Common mistakes

Safety

• False sense of safety (the main risk). A ripened filter retains most pathogens, not all of them; during ripening and after every cleaning, much less. Charcoal does not retain pathogens and can grow them. Always disinfect the filtered water. With water from an obviously fecal source (a river below a settlement, a pond with livestock), an undisinfected batch can transmit cholera, typhoid fever, hepatitis A or rotavirus, which kill small children through dehydration.
• Chemical contaminants. Neither sand nor charcoal removes salt, nitrates (dangerous for infants under 6 months), fluoride, arsenic or most metals. If the source is near mines, tailings, heavily fertilized fields or industries, or there are tooth stains or frequent skin lesions in the area, look for another source. Exhausted charcoal can release what it has adsorbed: replace it.
• Containers. Old or traditional glazed pottery can leach lead: use unglazed clay or lead-free glazes. Do not use drums that held fuel, pesticides or chemical products, nor treated wood.
• Dust. Sieving dry sand raises silica dust, which causes silicosis over time: sieve it wet or outdoors, with a damp cloth over nose and mouth. Charcoal dust irritates eyes and lungs: crush it damp.
• Carbonization and freshly made charcoal. The mound and the pit give off smoke with carbon monoxide: work upwind (with the wind at your back) and never near homes. Stepping on a lit mound can sink you into embers: don't climb on it. Putting out red-hot charcoal with water releases steam and some carbon monoxide and hydrogen: do it outdoors and upwind. Charcoal can reignite hours later and burn down storage: don't bag it hot. Natural-fiber clothing, water on hand and an area clear of dry leaves.
• Steam activation (optional: when in doubt, don't do it). Risks: steam bursting out and burning the face; combustible and toxic gas (carbon monoxide and hydrogen); flame flashback if air enters with gas still inside, blowing off the lid and throwing embers. Therefore: only outdoors; water through a pipe at least 1 m long and little at a time, never with your face above the vent or the funnel; stop adding water 15 min before and heat until there is no flame before sealing (F.8); do not open until you can touch the lid; never in clay containers: they burst. Keep children and animals away.
• Carbon monoxide poisoning. It has no smell and cannot be seen. Symptoms: headache, dizziness, nausea, fatigue, confusion. At the first sign, move upwind, out of the smoke, and move others away too. If someone is unconscious but breathing, place them in the recovery position outdoors; if not breathing, give rescue breathing (and chest compressions if you know how) and call for help. No one returns to the area until the fire is out and it has been ventilated.
• Community filters and wells. A tank with water or a pit near the river is a drowning risk for children: fence it or cover it. In sandy ground, pit walls collapse: don't go into a pit deeper than waist height without shoring or sloping it. To scrape it, use boards and never work alone.
• Maintenance hygiene: the water from “swirl and dump” and the scraped sand carry pathogens. Dump that water and that sand away from sources, wells and raw-vegetable gardens, and wash your hands with soap.
• Frost. Frozen water kills the biological layer and cracks clay containers: protect or bury the filter.

Variants

• With fewer resources (N0):
  • Riverbank well (bank filtration): dig a shallow pit in sandy ground 5 to 10 m from the channel (verificar) and draw the water that seeps in; the ground acts as a filter. It greatly improves turbidity, but flow velocity is not controlled: always disinfect. Fence it or cover it and watch for collapse (see Safety).
  • Bark containers, basketry waterproofed with clay, or leather: work the same if the 5 cm water layer, the raised siphon-free outlet and the maximum flow rate are respected; they don't last as long.
  • Without charcoal: the improvement in taste and odor is lost, not the removal of pathogens (which the charcoal never provided anyway).
• With more resources:
  • N2: masonry tank with lime mortar, glazed clay pipes, perforated sheet-metal diffuser, sheet-metal TLUD gasifier for the charcoal, iron container for activating it.
  • N3: concrete filter with a steel mold (CAWST v10, 70 to 75 kg, service life of more than 30 years according to CAWST 2009); standardized sieves for d10 and CU; gravel prefilters for more turbid water.
  • N4: commercial granular activated carbon (on the order of 15 times the capacity of good charcoal; Kearns et al. 2021); turbidity and E. coli monitoring; molded plastic filters (about 3.5 kg).
• Scaling up:
  • Family: 1 biosand (24 to 72 L/day) + 1 charcoal bed of ~27 L + disinfection + container with a tap.
  • Village (100 to 1000 people): 2 or more continuous-flow units; 5 to 200 m² per unit is typical (Visscher et al. 1987), though small ones under 1 m² serve small groups if scraped from the edge. Covered reservoir, a trained caretaker who measures the flow rate daily and staggers the scrapings.
  • City: dozens of beds (the municipal slow sand filter has been used since 1829 in London), with prior settling, mechanical sand washing, final chlorination and a laboratory. The slow sand filter takes up up to 10 times more area than a rapid one (Global WASH Cluster): beyond a certain size, coagulation with rapid filtration takes over.

Sources

• Huisman, L. y Wood, W.E. (1974). Slow Sand Filtration. OMS, Ginebra. Capa biológica, vertedero de salida, raspado y recarga por zanjas.
• Visscher, J.T. et al. (1987). Slow Sand Filtration for Community Water Supply. IRC Technical Paper 24. 0,1 m/h, d10 0,15 a 0,30 mm, CU < 5, lecho 0,8 a 0,9 m (mínimo 0,5 a 0,6 m), 1 m de agua sobrenadante, 5 a 200 m² por unidad, gravas de soporte, dren, depósito del 40 %, re-maduración, control diario.
• Thames Water y University of Surrey (2005). Slow Sand Filters (water-e). sswm.info. Tabla de criterios, apdo. 4.2.11 (gravas), densidades de la arena, prueba de limo, raspado de 1 a 3 cm, zanjas, llenado inverso.
• CAWST (2009). Biosand Filter (ficha y manual v10). 0,4 L/min, 600 L/h/m², lámina de 5 cm, pausa de 1 a 48 h, maduración de hasta 30 días, eficacias, 50 NTU y prueba de la botella de 2 L, «remover y tirar» (4 L, umbral de 0,1 L/min), orden de preferencia de la arena.
• CAWST (2012). Biosand Filter Construction Manual. Arena < 0,7 mm, gravas de 0,7 a 6 y 6 a 12 mm, prueba de caudal.
• Global WASH Cluster. Compendium of Water Supply Technologies in Emergencies, fichas T.9 y H.5. Velocidad de 0,1 a 0,4 m/h, turbidez, eficacias, recarga a 0,6 m, vaciado a 10 cm bajo la arena, re-maduración, superficie frente a filtro rápido.
• Pennsylvania DEP. Drinking Water Operator Certification Training, Module 17, unidad 2. d10, CU, cubierta.
• Kearns, J.P. et al. (2021). Environmental Engineering Science, PMC8165476. Carbón de gasificador ≥ 850 °C, adsorbedor modelo de 4,5 mm y 2,5 h, sistema de Tailandia de 7,5 mm y 18 h, tamaño de grano, carbón activado.
• FAO (1983). Simple Technologies for Charcoal Making. Forestry Paper 41, cap. 5 y 6. Montículo y fosa: color del humo, sellado, enfriado, rendimiento, temperatura.
• Ferguson, R.I. y Church, M. (2004). Journal of Sedimentary Research 74: 933-937. Velocidad de caída.
• Pillai, J. et al. (1999). Water Science and Technology 40(2): 85-90. Prueba de H₂S según temperatura.
• OMS (2011). Guidelines for Drinking-water Quality, 4.ª ed., cap. 7: tabla 7.7 (reducción por filtración lenta) y tabla 7.10 (E. coli 0 en 100 mL).
• Howard, G. y Bartram, J. (2003). Domestic Water Quantity, Service Level and Health. OMS. 7,5 L/persona/día para beber y cocinar; unos 20 L como acceso básico.
• Sphere Association (2018). The Sphere Handbook, 4.ª ed. Distancia mínima de 30 m entre letrinas y fuentes.