Update November 25, 2013:
An important observation regarding 5 gallon slow sand filters: They are highly susceptible to disruption from freezing temperature. For the past week all of the 5 gallon filters have been frozen. The temps here, in the forest, have been in the mid 20’s (deg. F) at night and barely above freezing during the day. All work on these 5 gallon filters must now start over, as freezing expands the the surface of the sand and effectively disrupts any biolayer that has formed. The larger filters, with unrestricted continuous flow, did not freeze. They are still running.
Update November 15, 2013:
I’ve made another critical mistake on these small filters. I forgot to “rough up” the inside surface of the buckets with coarse sandpaper. The inside walls of the buckets i am using are smooth, very smooth – in fact far too smooth, because they are new. This is not good, for filters. (this likely would not happen with some recycled containers, as they may already have been “roughed up” on the inside.)The sides of the container must be rough so the biofilm has a randomly rough surface, similar to sand surface, to cling to. The problem with the smoothness is that its “slipperiness”allows untreated water to slip down the inside of the bucket between the sand and the smooth plastic surface. As the schmutzdecke matures on the sand surface it slows down the water flow, but between the sand and the bucket inside wall the schmutzdecke can’t form as fast, or as well, because of the lack of surface area on the smooth surface of the bucket. So the microbes and silt that should remain on the top surface of the sand slip through around the sides of the bucket. This would explain why the output water has been showing more coliform instead of less. This will continue to happen until the biofilm is able to seal up the space between the sand and the smooth inside of the bucket. The seal will most likely be so fragile, that if the bucket is accidentally bumped, the seal will be broken and untreated water will flow past the sand and into the output. The irony here is that I had this same problem with the larger filters. I forgot to rough up the inside of the containers. Since I had fixed that problem in the larger filters, they are working much better.
On September 16, 2013 we described an experimental ( very small ) slow sand filter setup that was started running here at that time. We have used 5 gallon buckets as the filter container in both cases (filter 10a and filter 10b) with a sand depth of 9.5 inches. Filter 10b was started October 16.
To summarize the results so far ( this is my opinion based on observation and preliminary field tests): Some water quality improvement has been noted, both biological and physical. Ecoli are removed. Coliform is reduced slightly, but not eliminated. Initially, water clarity and color was improved slightly. The degree of improvement has not increased.
After 7 weeks of running time, it appears that a slow sand filter setup in a 5 gallon bucket using .25mm effective size sand with a flow rate of 1 cup per minute is, at the very best, only marginally effective at filtering roof water from the roof at this location. A better situation may be to use fine sand (.15mm effective size) as an absolute must, and a flow rate of less than 1/2 cup per minute. However do look at the results (scroll down at the following link to see the image) of the most recent field test on filter 10a. in the original post. The output water quality has become exceedingly more contaminated as time goes on. Compare this to the 38 inch deep filter 5 tests shown. The filter 5 output is nearly perfect. Fiter 5 uses the same water source as filter 10a.