Rain water harvesting and slow sand water filters

There are more and more websites and YouTube videos appearing on the internet showing how to “build your own slow sand filter”. Some of the videos have what appear to be very impressive results; but be cautious. The correct functioning of any slow sand filter is totally the responsibility of the owner / operator no matter who designed it, or sold it, or posted it on the internet. There is no way to be 100 percent sure any “do it yourself filter” will work.  There are some things you can do, however, to increase your chances of putting together a functional filter. First of all, do some research on how a slow sand filter works. Any public library will have the information you need. Then look for designs.

The most important question to answer first when deciding which design to use is this: Has the person showing the video, or publishing the website,  provided evidence of epa certified tests on the input and output of the filter that show it is working? Of course, anyone could forge results; but if the name and/or names of the person or persons doing the testing and the name of the lab providing the results are not shown then don’t trust the information. You should be able see the test results with the name of the Laboratory that did the testing and when the testing was done, and who had the testing done.  Another important question to ask is “how long has the filter been in operation?” It takes at least 1 year to adequately test a new slow sand filter design because each situation is different, with different amounts of pollutants in water at different locations. Look for successful tests done in the summer, winter, fall and spring.

It is entirely possible to put a bunch of sand and gravel in a bucket and make it look like polluted water is going in and clean water is coming out. Look for the test results; if there are none, then the video or website is just good entertainment at best. Water can be very clear, and still be highly polluted and dangerously full of disease causing organisms. Combine that with the fact that every situation is different and what works in one environment may not work in another.

Having tests done on the input and output of any filter is the only way to be reasonably sure of the condition of the water output.

The temps have been 28 degrees F at night for the past 4 days now. All the filters have frozen up at night, and thawed out during the day. Last night, Jan 2/3 was the coldest so far, 25 degrees here. (It gets colder here than near Seattle, we are in the foothills at 600 ft. elevation). An attempt will be made today to do a coliform test on at least one of the filters in order to look at how well it works at freezing temps.

Still waiting for the winter to “cut loose on us”. If it does like it did last year, we’re in for it. Last year: 14 inches of snow, and 1/2 inch of ice on top of that. Power out for 3 days. 4 huge trees down here, hundreds of large branches too, much worse in other parts of this area. State of emergency declared by the Gov. .

Slow sand filters are totally sustainable, will work on gravity feed only, take out hydrocarbons, bacteria, viruses, and do not require addeed chemicals to operate. Check out the websites below. And check out the post just before this one if you want to see just how well a slow sand filter removes bacteria.

There are lots of websites, and YouTube videos showing detailed rain water harvesting systems, but very few include any mention of a slow sand filter. For the most part, websites for state governments fail to show any references to slow sand filters. They are so incredibly simple to set up. A container, some sand and drain pipes and mother nature does the rest. This has been happening since life began on earth in natural settings where water is filtered as it flows through sand beds, and more slowly through rock.

Set up a small slow sand filter on just one of your downspouts and let it run for 3 or 4 months and you will have very clean water. Check the websites for details. We have been doing this here for 7 years now – the websites have documentation. At the very least let the water that runs into your rain barrel come through a slow sand filter first, then the water will not go stagnant or be full of nasty bacteria. Check out the websites.

Here are three websites to check out with information about rain water harvesting with a slow sand filter:

Slow sand filter

Roof water harvesting

Biological sand water filters

And here are more:

This is a fantastic site that really shows how effective small slow sand filters can be

And yet another site here

And one more here

Just Google slow sand water filter, or biosand water filter, or biological sand water filter. All you civil engineers out there designing rain water harvesting systems, why not take a look at small slow sand water filters? They work.

Filter 5, the most recent filter set up here in May of this year is now working.  The drainpipes inside the filter at the bottom of the barrel are covered by 4 inches of pea gravel. The sand directly on top of the pea gravel,  on the bottom, is .25 mm effective size with a U.C. of less than 2; and  700 lbs was used. The top sand is .15 mm effective size with a U.C. of less than 2; and 150 lbs was used. The supernatent water depth is approximately 8 inches. The white bucket on top is a 5 gallon reservoir that feeds 1/2 inch pvc drain pipes with small holes drilled in them. The drain pipes are inside the barrel at the very top just under the lid.
This filters roof water from a composition roof. The sand surface area inside at the top of the filter sand is: .2452 sq. m. .  The container depth 40 inches. The filter was started 2012-05-01.

This is a picture of filter 5, after the series of rain storms and wind storms that have passed through here Dec. 2012.

The results of the most recent field test on this filter are shown below. Please note: this is a field test and was not done by an epa certified lab. We did this test here using prepared solution from Micrology Laboratories..

The pink dots are coliform bacteria, the blue/purple dots are ecoli bacteria. “Pre-filter” means the test was done using water taken right out of the top of the filter’s supernatent water source, that is water that will be running through the filter on an ongoing basis. “Post-filter” means the test was done using water taken directly from the output pipe of the filter.

These are the field tests done on filter 5 as of December 21, 2012. The input and output water was tested for the presence of coliform and ecoli bacteria.

This is yet another update on what has been learned regarding the construction and operation of a small slow sand water filter.

These are issues that stand out after experience building, testing, and operating small slow sand filters over the past 6 years:

1. Fine sand (.15 mm effective size) allows for high quality water output both chemically and biologically.

2. Fine sand results in slow output flow rate.

3. Fine sand becomes clogged very quickly if the input water is turbid.

4. Coarse sand (between .35 mm and .50 mm effective size) results in high flow, and lower water output quality.

5. Coarse sand does not clog as quickly as fine sand.

6. Sand used in a slow sand filter will work best if the grains of sand are all very close to the same size (a low uniformity coefficient of 2 or less)

7. Use very coarse sand only at the bottom of the filter (the bottom 5 or 6 inches in a 30 inch tall container) the rest of the sand should all be much finer (smaller grains) sand

8. Roof water quality will vary drastically depending on the time of year, the condition of the roof surface, and the physical location of the roof surface with respect to trees, industry, and wildlife.

9. Some sort of pre-filtering is absolutely necessary if roof water is being filtered by a slow sand filter. A wire screen followed by a first flush diverter is one way to pre-filter roof water. A “settling container” is another way to “pre-filter” roof water.

10. Some sort of re-circulation of output water back through the filter is helpful when filtering roof water.

If roof water is being filtered:
11. Know that water quality will vary considerably
12. Have some sort of pre-filtering in place and,
13. For high water quality; use fine sand and recirculate the output water through the filter at least once
14. Use some sort of secondary purification process such as uv sterilization, boiling, or chlorination

If a slow sand filter is cleaned by “wet harrowing” :
15. be absolutely certain to stop the output flow while the filter is being cleaned by wet harrowing. If the output is not stopped, the contaminated water will flow down into the lower areas of the filter and foul the output. This is not good.

Some of the water quality tests done here on filter 4 and 5 are done. These tests were done with the testing kits available from Micrology Labs. The interpretation of the results of our tests are posted here along with images of the actual results we obtained. Please see the Micrology Labs website for a detailed explanation of the interpretation of these results. There is some math involved. We used 5 ml sample sizes, so the results we obtained by counting must be multiplied by 20 to find the cfu’s per 100 ml. The most important issue here is to note that the results for the pre-filter water showing nearly solid color represent super-highly contaminated water where there are so many colonies forming that it is not possible to count them without magnification. Also note that this is AFTER the water has been initially filtered by a first flush diverter. We have not tested water directly off of the roof surface yet. The number of cfu’s per 100 ml in the super-highly contaminated water turns out to be in the hundreds of thousands of cfu’s per 100 ml. Fortunately, both filters show a log reduction of coliform at slightly over 2 (there are 100 times less coliform bacteria in the output water of each filter.) It is possible to see, however, that the output still contains some coliform bacteria. Note that the ouput of fitler 5 contains no ecoli bacteria (dark blue-greenish colored dots with no pink around them – the blueish stain in the filter 4 output cannot be confirmed as an ecoli colony, although it may be). The filters are working, but we would like to see less bacteria in the output. More tests are upcoming as time permits.

In the images below of the tests of water samples shown please note that these tests are not done by an epa certified laboratory. They are field tests done by a non-professional with some lab experience. Some error is highly likely. What these tests demonstrate is whether or not the filters are starting to remove bacteria from contaminated water. These tests are only a general indication of how well the filters are working. To get a specific accurate number, a test must be done by an epa certified lab, and the sample must be taken by someone with some lab experience and some experience in taking water samples.

The next test will use a 1 ml sample for the pre-filter, and a 5 ml sample for the post filter. We have used the 5 ml sample on both for this test so the images can be compared with validity. If we showed the 1 ml sample size compared to the 5 ml sample size for a visual evaluation, it would be inaccurate and misleading.

More notes on roof water quality required for a slow sand filter:

Experience here in the art of roof water harvesting and slow sand water filter construction is a continual learning experience. The newest roof water filter here (filter 5 – more info on filter 5 is upcoming as soon as I get time) has had some issues with adequate filtering. Three issues stand out:

1. First flush diverter operation

2. Gutters and downspout debris

3. The conditions surrounding the roof surface

We have discovered that a functioning first flush diverter, or some sort of pre-filter device is absolutely necessary for the proper operation of a slow sand filter used to purify roof water. The filter will plug up rapidly (within hours) if too much solid material flows into the filter. The filter 5 system here requires at the very least 50 gallons of diversion after a long dry spell. This is because of the location. This part of the roof is directly under a massive old growth fir tree. The tree is huge – more than 4 feet in diameter and about 125 feet tall some of the lower branches are as big as small trees. Huge amounts of organic material accumulate on this part of the roof yearly.  This past month, the diverter failed to prevent muddy water from entering the filter. The result was catastrophic. The filter’s flow rate slowed to nearly non existent and the water was no longer clear, and was foaming up in the container – all bad signs. We cleaned the filter, drained the diverter, and re-cleaned the gutters and downspouts. It took over 50 gallons of runoff before the water was acceptable for input to the filter. Part of the problem was due to the fact that the automatic slow drain was not set so as to allow the diverter storage to gradually drain, and the diverter was totally full when the rainy season started. This was my error. Had I set the slow drain before the rain started, the worst water would have been diverted. This is an extreme example, but not for this area of the country. Trees are common around houses and rooftops, and we often have 2 month dry spells during the summer here – lots and lots of stuff builds up on the roof – dust, air pollution, organic material and pine needles.

If a slow sand filter is used to purify your roof water , it is critical that the downspouts and gutters be kept free of solid material. In most cases, normal maintainance will keep the system clean enough. Some debris is unavoidable, and the diverter should handle that ok.

The surroundings will have a huge effect on the water quality. If there are no trees near the roof surface and the roof is not near any highways, or subject to frequent poor air quality episodes, the runoff water may not contain as much solid material.

From this we have learned that slow sand filters are not perfect. The water input must be fairly clear to start with.

Another way to handle this situation is to have one or two settling barrels with an overflow on the last one before the water enters the filter. This is the way filter 1 is set up. (in the image at this preceeding link to filter 1, 2/3 of the way down the page, the settling barrels can be seen on the right, with the pitcher pump in the background on the 3rd barrel used as storage). Filter 1 has been in operation for 6 years, and there have not been any problems with too much debris entering the filter. It is necessary to clean out the settling barrels every 1 or 2 years.

This is an update of the previous post on slow sand filter sand sources. Instead of trying to find specialized sand sizes from a manufacturer, it can be more convenient to sift easily available sand to get the right grain size distribution (uniformity coefficient) that will work most effectively in a slow sand filter. Below is an image of six different sand products we compared here. Three of them (the .15 mm effective size, the .25 mm effective size,  and the Target filter sand)  are sand that is already working in successful tested slow sand filters we have in operation here.  The other three are readily available sand products that can be found just about anywhere. We ran 1 liter of each type of sand through successively smaller wire mesh sizes to determine, more precisely,  the breakdown on the variation in size of the individual grains of sand in each type of sand. Each mound of sand shows how much of a specific size group is in each type of sand. Of particular interest are the first mounds and the last mounds. The wire mesh sizes and a manufacturer source are here:

McNichols has the wire mesh available from an online catalog
At their page on the left side of the page, click the “see more” link in red below the “mesh size 1″ link in red, to expand the list and show all the available sizes.

3860753610 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 304,
Woven Construction, 60 Mesh, 0.0075″ Wire, .0092″ Opening, 36.0000″ Width

3860754810 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 304,
Woven Construction, 60 Mesh, 0.0075″ Wire, .0092″ Opening, 48.0000″ width

3880554810 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 304,
Plain Weave Construction, 80 Mesh, 0.0055″ Wire, .007″ Opening, 48.0000″ width

3850093610 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 304,
Woven Construction, 50 Mesh, 0.0090″ Wire, .011″ Opening, 36.0000″ Width

3850094810 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 304,
Woven Construction, 50 Mesh, 0.0090″ Wire, .011″ Opening, 48.0000″ Width

3140104810 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 316,
Woven Construction, 40 Mesh, 0.0100″ Wire, .015″ Opening, 48.0000″ Width

3840103610 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 304,
Woven Construction, 40 Mesh, 0.0100″ Wire, .015″ Opening, 36.0000″ Width

3840104810 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 304,
Woven Construction, 40 Mesh, 0.0100″ Wire, .015″ Opening, 48.0000″ Width

3830123610 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 304,
Woven Construction, 30 Mesh, 0.0120″ Wire, .0213″ Opening, 36.0000″

3830124810 McNICHOLS Quality Wire Mesh, Square Weave, Stainless Steel Type 304,
Woven Construction, 30 Mesh, 0.0120″ Wire, .0213″ Opening, 48.0000” Width

The first mounds, in row 1 of the play sand, mortar sand and mason’s sand columns; are the very coarse pieces mixed with smaller pieces that really prevent a slow sand filter from working. This is because too many large pieces of material mixed with small pieces, prevent the biofilm from forming in, and on the top layer. The biofilm at the top layer is where most of the biological purification action takes place.

The last mounds of sand, in row 6 of all but the “Target filter sand” column, are all the very small powder like pieces of sand that will make the output water highly turbid (cloudy) and undesirable.

This shows sand sifted into specific particle sizes

A SUMMARY OF THIS POST:

For a slow sand filter to work we need to use “uniformly graded” sand. “Uniformly Graded” sand means sand that has already been processed (sifted) so that the grains of sand it contains are mostly the same size. “Un-graded” sand means un-sifted sand that has many different sizes of grains of sand. Here, we try to make our own “uniformly graded” sand from “un-graded” “mortar” sand. First, we sift “uniformly graded” sand and “un-graded” sand through a series of wire screens to determine if we can duplicate the sizes of the “uniformly graded” sand product that we already know will work in a slow sand filter. We have found that it is possible to very closely duplicate the grain sizes we need by sifting the “un-graded” “mortar” sand. The image directly below is the result of comparing the “uniformly graded” sand to sifted “un-graded sand” (in this case “mortar” sand) after sifting each through a series of wire screens.

graded and ungraded sand that has been sifted using wire mesh

MORE ABOUT SAND GRAIN SIZE (THE PROCEDURE WE USED TO COMPARE TWO PRODUCTS):

We compared two sand products (shown in the large image above). One is a “graded sand” product from Unimin, specified as .15mm effective size for filtration (the bottom row in the above image). This is the sand we have used successfully in 3 of the filters running here.
The other product is sand from a local sand and gravel supplier. They call it “mortar sand”. It contains very little dust. The less dust, the better.

To sift the sand for comparison purposes, we used 3 different stainless steel wire mesh sizes available from Grainger and McNichols (both suppliers have extensive online catalogs) the links I have here may change slightly so here are the size specifications of the wire mesh I have used. If you order, make sure it is stainless steel wire mesh. It may be necessary to search the website online catalogs or actually call the suppliers:

30 mesh stainless steel wire screen: opening size: .54 mm (.0213 inches) plain weave

40 mesh stainless steel wire screen: opening size:   .381 mm (.015 inches)  plain weave

60 mesh stainless steel wire screen: opening size:  .259 mm (.0102 inches) plain weave

We took random samples in equal quantities of each type of sand and ran them through the 30 mesh screen. Be aware that both types of sand are angular in shape (not round). We took what went through the 30 mesh screen and put it through the 40 mesh screen. Then we took that which went through the 40 mesh screen and put it through the 60 mesh screen.

We ended up with six samples of sand:

The first sample is what is retained on the 30 mesh screen (R30). This sample consists of grains that will not pass through the .54 mm openings in the wire mesh.

The second sample is what passed through the 30 mesh screen (T30). This sample consists of sand grains that fit through the .54 mm openings and anything smaller.

The third sample is what was retained on the 40 mesh screen (R40).This sample consists of sand grains that will not fit through the .381 mm openings in the wire mesh.

The fourth sample is what passed through the 40 mesh screen (T40). This sample consists of sand grains that will fit through the .381 mm openings in the wire mesh and anything smaller.

The fifth sample is what was retained on the 60 mesh screen (R60).This sample consists of sand grains that will not fit through the .259 mm openings in the wire mesh.

The sixth sample is what passed through the 60 mesh screen (T60).This sample consists of sand grains that will fit through the .259 mm openings in the wire mesh screen and anything smaller.

Note (Sept. 10, 2012) We have tried using a 50 mesh stainless steel wire screen in place of the 60 mesh screen. The openings in the 50 mesh are square weave at .2794 mm (.011 inches) using .2286 mm (.0090 inches) wire diameter supplied by McNichols. The slightly larger openings in the 50 mesh screen allow more of the excessively fine particles and dust to be sorted out, and make the sifting faster and easier. This, of course depends on the nature of the sand being used.

We are, in essence, sorting the grains of sand in a given sample to get a better idea of the sizes of the grains. As it turns out, the most significant difference between the “uniformly graded” sand and the “mortar” sand is that the mortar sand has more of the much larger pieces throughout a non-sifted sample than does the “uniformly graded” sand. These large pieces are the ones you DON’T want mixed up in the sand you use in the upper regions (and on the top region) of your filter. Both samples show a significant amount of very fine powder-like residual material after passing through the 60 mesh screen. This fine powder-like substance is what becomes a problem. It takes a long time (weeks to months) for it to wash out as the filter ripens. This means the output water does not “clear” up for a significant time period. Sifting out this dust requires wearing a dust mask and is quite labor intensive. The other alternative is to use lots of water to wash this fine material out at the last stage of sifting on the 60 mesh screen.

CONCLUSIONS:

It does appear that by sifting the “mortar sand” through the 30 mesh wire and then washing what has passed through the 30 mesh (use the 60 mesh wire to hold the sand while you wash it) it is possible to “manufacture” your own “uniformly graded” sand without having to search for a “commercially manufactured” “brand”. It is important to note that both sand products are randomly shaped angular sand (not round) so the way the grains fit through the openings may not be perfectly consistant in each instance of sifting. The “mortar sand” we used here is 24 dollars for a half yard – more than enough to put together a small slow sand filter. A layer of two or 3 inches of the coarse sand that does not pass through the 30 mesh wire can be used on the very bottom above the pea gravel for added insurance that the fine sand won’t overwhelm the drain pipes. The stainless steel wire screen material costs anywhere from about 15 dollars to 35 dollars depending on the size you get (the number of lineal feet). The screen material is available as 3 feet wide or 4 feet wide, and you can specify the length you want. It comes as just the wire screen, you will need to build a frame for the screen, at a cost of about 5 or 6 dollars. We used 2X4’s .

These are the frames we made for the wire screen

use small pieces of trim to hold the wire screens in place

Although we have not tried all possible types of sand, it is very likely that this procedure will work with any sand because the sizes of the screens stay the same. The amount of sand you end up with that does not pass through the 30 mesh, and the amount you end up with that does not pass through the 60 mesh may vary depending on the various sizes of the grains of sand you have on hand. “Mortar” sand usually is already somewhat “uniformly graded”. “Builders” sand is less likely to be “uniformly graded” and will probably have bigger grains, and lots of dust. The best sand is that which is dug from a quarry or directly out of the ground in an area where there is not likely to be contamination. The “masons” sand from Lowes will work, but it is very dusty. The “play sand” from Lowes also may work but it, too, is very dusty. When you sift sand; wear a dust mask, ALWAYS!!!!!!