Filter to Waste Advice Needed

[aitch2oh]

Our utility is planning to install filter-to-waste piping, valves and appurtenant controls in existing drinking water filters, even though we are a ground water source water system (not under the direct influence of surface water), and we are under no requirement by our state health agency to do so. I am researching industry publications and live resources for controlling criteria in that design. The only references I've found so far discuss the duration of the FTW. Is the volume of FTW controlling, or is the rate of FTW controlling, or a combination of both? In other words, is 400 gpm for 25 minutes just as effective as 2,000 gpm for 5 minutes? Since both our backwash waste and FTW streams are treated and returned to the plant headworks, the duration of FTW and volume of water generated are not major considerations. Thanks in advance for your comments.

 

[bimr]

The rate of flow is dependent on the type of filter media that you have. Assuming that you have sand pressure filters, the minimum backwashing rate is 10 gpm per sq. ft. of bed area and higher rates up to 15 gpm per sq. ft. are preferable. Use of surface washers and air scour greatly aids in the backwashing. Backwashing is generally carried out for 8 to 10 minutes and is usually not carried out to completion as that means a somewhat longer filter-to-waste period. The minimum backwash rate really means minimum. If the minimum rate is 10 gpm per sq. ft. of bed area, backwashing twice as long at a 5 gpm per sq. ft. rate will not clean the bed for the entrained dirt does not respond to any such specious reasoning. To high a concentration of suspended matter, too high a filtration rate, and too low a backwash rate are all factors, any one of which can cause packing of the bed, gravel hills, and eventual overturning off the bed, thus necessitating unloading of the filter units and re-laying of the filter medium.

 

[aitch2oh]

Thanks for your reply, bimr, but I may need to clarify my request. Please indulge me.

Our filters are 42" of anthracite over a graded gravel support. We backwash using air scour alone (5 SCFM/SF), followed by a concurrent air and water stage (air at 5 SCFM/SF and water at 8 gpm/SF), and finished up with a water only backwash at 20 gpm/SF for the transport phase. At some point after returning the filter back to filtration mode, the initial filtrate will usually present a turbidity spike after some time, typically between 5 and 20 minutes (of course, that time is plant-specific). That initial quantity of filtered water is discarded to reduce the turbidity of the filtered water going into the plant clearwell. The questions in my original posting relate to the wasting of filtered water after the backwash is complete, commonly referred to as "filter to waste" (FTW) or "rewash".

Sorry for my lack of clarity or detail.

 

[stanier]

I can recall working on a tender in Australia some years ago where Anglian Water used there US operations experience to demonstrate that they could achieve very high quality water by using this technique. From memory they did this on a time basis to avoid the turbidity spike. 45 minutes seems to ring a bell. Now they had 67 plants in the USA but they dont trade there as Anglian Water I dont think. I am sure there have been papers on this operational technique prented at AWWA.

From memory the turbidity achieved or aimed for was 0.1NTU.

There filters were a mixture of sand, anthracite over sand and anthracite. They used water followed by water + air scour for the washing cycle

 

[bimr]

It is further assumed that you are using clean water for backwashing. When a filter is to be placed in operation for the filtering run, it is first "filtered to waste" for a few minutes. This is because a good filtering action will not be secured until the bed has settled and a small amount of floc has accumulated. Taking successive samples of the effluent and reviewing them for clarity will allow the filtration efficiency to be determined. As soon as the effluent runs clear, the filter-to-waste operation is discontinued and the filter unit is put into service.

Since you are trying to settle the bed, your filter-to-waste flow rate should at the same rate as the normal operating rate. Doing the filter-to-waste operation at a lesser rate will probably contribute to excessive filter-to-waste volume as well as lower effluent quality as suspended matter travels down through the unsettled filter bed

 

[semo]

I haven't seen references on this. You might contact some filter manufacturers to see if they have any studies; but, here is what I know and think (For What Its Worth).

The filter-to-waste or "ripening" stage is done for two purposes. It settles the media back into place after the backwash. This reduces the space between media granuals and allows the filter to capture smaller particles. The ripening also washes any turbid water in the media away and allows the clean media to capture larger particles which also reduces the spaces between media/particles reducing the blow through.

I don't think a slower filtration rate in this stage will settle the media into the same density as the normal filtration rate and therefore could allow some turbidity to still pass through the media once the filter is put into filtration mode.

As you mentioned most references discuss the time. This varies dependent upon many things that are site specific. You will have to do a pilot study at your site to see what each filter requires if you want an exact time. An on-line turbidimeter will show you your spike and its length. Otherwise you could set the length of time to something like 30 minutes and get most if not all of the spike.

I'll note that many plants (particularly groundwater) I've seen do not even utilize this stage. I think at these facilities the spike is small or does little to harm the overall water quality.

 

[stanier]

The idea of ripening the cell at normal flow rate is sound advice.Yes you use more energy through the filter to waste cycle but you can achieve the 0.1NTU consistently.

With on line turbidity metering aim to have one per cell. Some companies have one per filter bank. That way they dont know of a cell has ripened or they have breakthrough in another cell and needs backwashing.

 

[aitch2oh]

Thanks for your comments. Based on my own research, I agree that there is not much literature on the FTW rate. I have contacted AWWA for assistance in finding JAWWA articles where FTW was addressed, hoping to find some discussion of FTW rate setting.

Lacking any "official" criteria, we are looking at an empirical determination of the timing of the post-backwash turbidity spike, and setting the FTW duration to at least that duration. We are establishing the pipe sizing based on the filters' design flow rate. We are also looking at terminating the backwash before the filter is fully cleaned, to "seed" the media. I have found this to be a practice at other water treatment plants to both minimize the volume of backwash water used and to increase the productive filtering time. Finally, we are considering providing sample taps on the FTW line, and installing sample piping to the filter's effluent turbidity meter to signal the "achievement" of the spike, and using that as the determinant for terminating the FTW stage. Honestly, though, since the backwash water and FTW streams are returned to the plant headworks for retreatment, the use of the empirical timing will likely be the method chosen.

Thanks again for your comments. Hopefully, I can be a donor and not just a recipient to other advice-seekers on this site.

 

[semo]

You should be able to set the filter-to-waste time as an input variable in your controls.

We typically set the air scour, concurrent air/water, water wash, filter run, and filter-to-waste times as user inputs that can be modified in the PLC as needed.

Don't know your piping layout; but, if you put your filter-to-waste piping as an offshoot (tee) of the effluent pipe and locate it after the turbidimeter, you can measure the spike and measure the full filter run too.

As stanier said, you want a turbidimeter for each filter and a sample tap for each filter.