Subscribe via RSS Feed

Recirculating Sand Filters (RSF)

Authors’ Note: There are significant differences between this section and the newsletters. The differences reflect what we have learned about the various design features. Accordingly those techniques and design features that we found do not work, have not been included in this document.

Recirculating sand filters (RSFs) are perhaps the most common non-proprietary technology in use in Massachusetts as of this writing. In Barnstable County, a few earlier installations were supported through grant moneys in an effort to introduce RSFs again into Massachusetts (the recirculating sand filter or RSF was initially developed in Massachusetts in the earlier 1900s) for the purpose of reducing nitrogen discharge from onsite systems. The first recirculating sand filters in Massachusetts to be monitored extensively were installed in Gloucester. These system installations attempted to demonstrate that overall wastewater management goals of that city could be met by employing advanced onsite treatment. Later installations in the Buzzards Bay watershed (one system in Fairhaven and one in Bourne) focused on the nitrogen removing capability of recirculating sand filters. We estimate that, as of this writing, there are still less than twenty installations in southeastern Massachusetts.

Figure 1. Recirculating Sand Filter

Recirculating sand filters come in a variety of sizes and configurations. The general schematic of three systems installed in Barnstable County is illustrated above, however this illustration shows only one of many ways to achieve the required recirculation. The system is composed of a standard septic tank, a pump chamber, the sand filter, and a soil absorption system or SAS.

The theory behind the recirculating sand filter is simple. Septic tank effluent is pumped from the pump chamber to the top of the sand filter. As the effluent passes through the sand filter, the ammonium-nitrogen is converted to nitrate-nitrogen in a sequence of steps that occur in the presence of air and two genera of bacteria. The first bacterium called Nitrosomonas converts ammonium or NH4+ to nitrite or NO2 and the second bacterium, Nitrobacter, converts NO2 to nitrate or NO3. Following the conversion of ammonium to nitrate in the sand filter, a portion of the effluent is piped back to the pump chamber or the septic tank, while a portion of the effluent passes on to the leachfield. The nitrate contained in the portion that returns to the pump chamber or the septic tank undergoes a further transformation to nitrogen gas (N2). This harmless gas is vented to the atmosphere through the vents in the system. Conditions that must be present for the conversion of nitrate to nitrogen gas to take place are anaerobic conditions and a carbon food source. Both the pump chamber and the septic tank are potential candidates for these conditions, and thus nitrified waste can be returned to either component. It is more common, however, to return nitrified waste from the sand filter to the pump chamber for subsequent denitrification in order to minimize the disruption in the septic tank and promote its function as a primary anaerobic digestion unit in the system.

First Stop – The Septic Tank

Perhaps the most familiar component of the system is the septic tank. The recirculating sand filter, as with most on-site wastewater treatment, must be preceded by a settling chamber such as a septic tank. The revised Title 5 requires that the tank be a minimum capacity of 1500 gallons. For this and other requirements for septic tanks refer to Section 15.223 of the new code. As with all systems having a pump chamber following a septic tank, it is recommended that an effluent filter be installed at the discharge end of the septic tank. This will minimize solids passing through to the leachfield or fouling the pump. As of this writing, there are three effluent filters approved for use in Massachusetts; they are the subject of another chapter in this book.

The Pump Chamber

Following the septic tank, the effluent passes by gravity into a pump chamber. In the recirculating sand filter design shown, the pump chamber serves a dual purpose. First, as a pump chamber, it stores the mixture of septic tank effluent and sand-filter return until it is pumped up to the top of the sand filter. Secondly, facultative anaerobic bacteria located in the pump chamber act on the nitrate in the waste returning from the sand filter to convert it to nitrogen gas. Some designs you may see have a separate chamber, prior to the pump chamber, for this denitrification step. Also, in some proprietary nitrogen removal systems, return effluent from a trickling sand or other filter is returned to the septic tank. The variety of designs are outside the scope of this summary, however you should be aware that there is a wide variety of designs.

Although regular cylindrical or box-shaped pump chambers can be used, you might consider the use of a 1,000 gallon septic tank as a pump chamber. This allows for more than adequate storage volume, and these tanks are readily available at costs comparable to a cylindrical pump chamber. The volume of the chamber should be at least 150% of the design flow for the house. The most important characteristic of the pump chamber is that it be watertight. You might also consider, at least at the pump end of the tank, having a 30-inch manhole for easy access to the pump and wiring. The most important consideration for the pump chamber is that the contents be minimally disturbed or agitated (so that it remains anaerobic). The best arrangement is that return lines from the sand filter and effluent from the septic tank enter the pump chamber enter through sanitary tees that extend below the liquid level. These effluents should enter at the opposite end from the pump that discharges to the sand filter. Allowing for maximum storage volume and installing baffles in the pump chamber are also preferable to encourage anaerobic conditions and longer residence times.

The Sand Filter

The sand filter itself can be constructed a number of ways. In Barnstable County, at least three single family designs have incorporated the bottom half of a 2,000 gallon septic tank with an additional shim to increase the volume and contain the filter media. In Orleans, F.L. Quinn, Inc. used an impervious liner within a constructed wooden box to contain the filter media as part of a RSF “kit” distributed by Orenco Systems® Inc. 814 Airway Avenue, Sutherlin, Oregon 97479-9012). The liner poses somewhat of a problem in sandy soils, since it is difficult to keep the box shape of the filter while backfilling and filling the filter with media, unless a wooden box is constructed to support the liner. The box can deteriorate over time, leaving the liner intact, without affecting the filter. In above-grade filters, timber walls should probably be constructed with treated wood, since they must remain structurally sound for the life of the system.

Distribution of Effluent to the Sand Filter

From the pump chamber, effluent is pressure distributed to the top of the sand filter box. This portion of the system accounts for the majority of removal for Biochemical Oxygen Demand (BOD) and Total Suspended Solids (TSS). The sand filter box itself, as previously stated, can be made from a variety of materials from concrete to wood with an impervious liner. It must be constructed to allow air passage to the top of the filter. It must also allow for the distribution of effluent to the top of the sand filter bed. Distribution to the top of the sand filter can occur in a variety of ways. If there is a structure or top over the filter, the effluent can be sprayed directly on top of the sand surface. Different distribution means are used to spray effluent on top of the filter bed. One design distributes the effluent from a single line of spray nozzles. The spray of effluent goes upward/outward through slots cut on either side of a 1-inch riser pipe. Another design using splash blocks on top of the media bed referenced in an earlier fact sheet, is not recommended due to the difficulty in obtaining even distribution across the media. An installation of a recirculating sand filter in Wellfleet (Figure 2) used still another means to distribute the effluent onto the sand filter bed. That system employs a series of 15 riser pipes spraying effluent upward. The spray from the risers is deflected downward by half-round 8-inch diameter PVC pipe that covers three spray nozzles each. This system of distribution has an advantage of distributing the effluent more evenly over the media, a highly desired objective to obtain optimum treatment. A similar arrangement, simply using a pressure manifold with 3/8 inch holes oriented to spray upward against a concave shield was installed to serve three cottages in Wellfleet.

Figure 2. Distribution system currently used in RSF in Wellfleet. Note that a modification of this system using upward oriented discharge holes instead of pipe risers has also been employed successfully.

The two distribution systems illustrated assume that there is a protective cover over the top of the filter bed to prevent uncontrolled aerosolizing and dispersal of the effluent. This technique allows for easy access to the media for servicing by simply lifting the entire cover or portions of it. These accessible-type designs have the advantage of being easier to monitor and gauge the condition of the filter media and service it, should it be needed. The disadvantages to this open-type design include slight to moderate odor problems, particularly if the filter is located near the house.

In the open type design, a wooden top is most commonly used for protection and aesthetics. In summer of 1996, a different type of design with no structural cover was installed in Orleans. In this design, effluent is distributed with a system of pressure laterals to the top of the sand filter, however, peastone is used to cover the entire distribution system. A layer of bark mulch on top of the peastone totally conceals the distribution network. In the Orleans system, the final elevation of the top of the sand filter is nearly at grade, so the entire system is rather inconspicuous. The illustration below (Figure 3) illustrates the distribution piping system to the top of the open-top system in Orleans. Similar “covered” designs have been installed in Gloucester. One of them appears in the yard to look like raised bed contained by timber walls approximately 30 ” high. Figure 4 shows a picture of a RSF distribution system by Orenco Systems® prior to covering with peastone and bark mulch.

The dosing schedule of effluent to the RSF is also an important design feature. In general, many and small applications of effluent to the top of the sand filter will result in a better chance for waste nitrification (the important first conversion of ammonium to nitrate). Accordingly, the sand filter pump must be activated by a timer. A preferred dosing schedule is 3-5 minutes on (when the filter receives fresh effluent from the pump chamber), 25-27 minutes off (when the RSF is draining and resting). To achieve the proper recirculation rate of 3:1 to 5:1, a designer should size the pump to deliver three to five times the volume of wastewater generated by the house in equal doses over 48 dosing periods per day. For example, to achieve a 5:1 recirculation rate for a three bedroom home (330 gpd), approximately 1650 gallons of effluent must be distributed to the sand filter over 24 h in 48 equal doses. This equates to just over 34 gallons per dose. For this application, a small pump can be used to deliver 11 gallons/minute over 3 minutes. A note worth mentioning here is that earlier experiments by the Barnstable County Department of Health and the Environment tried using a “demand” rather than timed dose. Under these earlier experiments, demand from the house (approximately 20 gallons) caused 100 gallons or so to be pumped to the top of the filter. 80 or so gallons returned to the pump chamber and 20 discharged to the SAS. This eliminated the need for timers. However, we found that, only if the house has a more evenly spaced flow pattern, will this system work reasonably well. As you will see in the section below, we achieved good results from one system operated in such a fashion, and poor results in another, where the majority of flow to the system was in a very short period of time.

Figure 3. Schemata of a RSF with no exposed sand surface. The final grade may be at or above existing and surrounding grade. If above grade, the containing liner is supported by a concrete or timber wall.

Figure 4. Pressure distribution system of a RSF in Orleans, Ma. prior to the installation of protective caps for the spray orifices, peastone, filter fabric, and a top coating of bark mulch. The finished sand filter is only slightly above the surrounding grade.

Specifications for the sand filter media are given in a guidance document issued by DEP. In essence, the sand must have an effective size of 1-2 mm, have a uniformity coefficient of less than or equal to 3.0, and exhibit little fine material (less than 1% by weight shall pass though a # 200 sieve). One of the two recirculating sand filters installed in Barnstable County using native sand, near the tolerances of that specified in the guidance document did experience filter clogging after nine months. The situations were easily remedied, however, since the clogged system was an open type design, and simple raking of the media restored its hydraulic function. We highly recommend that this detail of the system receive the highest level of scrutiny. One source for sand, that has been used in at least four sand filters in southeastern Massachusetts is Holliston Sand and Gravel in Slatersville, Rhode Island (Phone 401-766-5010), however, there are likely many such places to obtain the sand in Massachusetts. An important thing to remember is that deviating from the recommended specifications of the sand must be avoided, if you don’t want that midnight call from a homeowner. A final note on media. Recently, while cruising the web, I visited the site of David Venhuizen. The site contains many design features of recirculating sand filters, and many experiences he has had. David is a regulator in Minnesota and has some opinions on the benefits of RSFs. David feels that the media size has very little effect on treatment, and that peastone works as well as 1-2 mm sand. The advantage to larger diameter media is lower maintenance problems. A review of the literature suggests that this is likely true, as long as frequent small doses of effluent are sent to the sand filter.

Achieving Recirculation

Perhaps the widest variation one can find in RSFs is the method of splitting the flow from the sand filter to achieve the desired recirculation. Three methods will be discussed below:

splitting the flow in the sand filter itself; using a redirecting valve (ball valve, “Mickey mouse” valve or the like) inside the pump chamber, and; using a splitter valve outside of the pump chamber

Splitting the flow within the sand filter

In our first illustration, we show that flow from the sand filter is split in the bottom of the sand filter box. Approximately 80% of what is sprayed on top of the sand filter is returned to the pump chamber for denitrification, while approximately 20 % is released to the SAS. The proportioning of filtrate is set by the location of a dam constructed on the bottom of the sand filter that directs the larger portion of the effluent back to the pump chamber, while allowing the smaller portion to discharge to the SAS. This is perhaps the simplest design, however it is crucial in this design to uniformly distribute effluent to the top of the sand filter. For instance, if the effluent was all distributed over the return portion of filter, effluent would never discharge to the SAS. Conversely, if the effluent is distributed more over the top of the discharge side of the dam, more discharge to the SAS will occur than is desired.

Figure 5. Schemata of recirculation achieved by splitting the flow at the bottom of the sand filter.

While the advantage of this type of design is simplicity, as usual simplicity has its cost. Under this design, a portion of effluent is always discharged to the SAS, and chances for further treatment of this effluent in the RSF is lost. In addition, since the dam at the bottom of the system is permanently installed, changing the proportion of return to discharge would extremely difficult. This is a popular design among some regulators, notably Richard Piluk, a County Health Department Official in Ann Arundel County Maryland, who sees this as a low maintenance feature that can still achieve 60+% nitrogen reduction. A note here, however, is worth repeating. The flow to the RSF should be on a timed cycle, ideally with many small doses and long resting periods which will encourage more complete nitrification. As mentioned, our earlier experiments with demand doses gave inconsistent results. Also worth mentioning here is the fact that, under this arrangement, when there is no water use in the residence (such as through the night or when a vacation is taken) eventually the low-water shutoff of the pump chamber will stop any distribution of effluent to the sand filter. While some researchers feel that this might be an advantage, many feel that it will result in the sand filter becoming anoxic. In any event, when using the bottom of the sand filter to split the flow, it is desirable to adjust the pump cycles so as to minimize the inactive periods where the filter will not be fed effluent.

Splitting the flow inside the pump chamber

Another way to achieve the desired recirculation in an RSF system is to collect all of the filtrate from the sand filter and split (a portion going each to the pump chamber and the SAS) somewhere else in the system. In the majority of systems installed in Massachusetts using this technique the flow is split within the pump chamber. Simply put, all of the effluent from the RSF is returned by gravity to the pump chamber, where it either empties back into the pump chamber (for denitrification) or passes through the pump chamber into the SAS. This feat is achieved by use of a device variously called a ball valve, buoyant ball valve, or “Mickey Mouse” valve. The concept is illustrated below. The control of flow (either to the pump chamber or through to the SAS) is dependent on the volume of liquid in the pump chamber.

Figure 6. Schemata of a RSF using a buoyant ball valve for recirculation control.

The buoyant ball valve illustrated here (Figs. 7a and 7b) consists of an inlet from the sand filter, an outlet to the leach field, a downward outlet to the pump chamber, and a buoyant ball which seals the downward outlet. Sand filter-treated effluent returns to the pump chamber via the ball valve. As the level of liquid in the pump chamber rises, the ball rises and exerts enough pressure to make a firm seal on the downward outlet of the pipe. When the ball seals the downward outlet the remainder of the effluent passes to the leach facility. Use of a buoyant ball valve has the advantage of being reliable, inexpensive, and simple to maintain. More importantly, the buoyant ball valve allows recirculation of pump chamber contents at times of no water usage in the building without voiding any volume to the leaching field. Theoretically, this allows for better treatment of the waste during times of lower flow. The buoyant ball valve only discharges to the field if there is adequate volume in the pump chamber.

Figure 7a. All flow returning from the sand filter is returned to the pump chamber where denitrification takes place. Figure7 b. Liquid level in the pump chamber rises to the point where the buoyant ball rises and seats seals the drainage port into the pump chamber and causes effluent to pass through to the leaching facility.

An attempt to improve treatment even more was developed and is in use in the Orleans RSF. This addition to the standard buoyant ball valve prohibits the direct discharge of all of the returning effluent from the sand filter, even when the buoyant ball valve seals the downward path to the pump chamber. It does this by using the scheme shown below in Figures 8a-d.

Figure 8a. All flow returning from the sand filter is returned to the pump chamber where denitrification takes place. Figure 8b. and Figure 8c. Liquid level in the pump chamber rises to the point where the buoyant ball rises and seats seals the drainage port into the pump chamber and causes effluent to pass through the five "fingers", one of which connects with the pipe exiting to the leachfield, and the remaining four spill back into the pump chamber. Figure 8d. Picture of the valve device as installed in a RSF in Orleans.

Splitting the flow outside of the pump chamber using a splitter valve

The final way we will discuss splitting the flow returning from the recirculating sand filter is by use of a splitter valve located outside of the pump chamber. Various valves and devices have been suggested for this method, however, we have not seen any, as of yet, used in Massachusetts. The simplest splitting device is a distribution box with multiple exit ports. If four exits are present, three may be piped to return to the pump chamber and one to the SAS. By using adjustable inverts in the distribution box, a wide adjustment of forward (to the SAS) to return (to the pump chamber) flow can be achieved (Figure 9). Since we have not seen any of these in use yet, we can not comment on their merit.

Figure 9. Schemata of a recirculation strategy using a splitter box or valve outside the pump chamber.

A Final Word About RSF Design

Before designing a RSF, you should consult the DEP Guidance Document regarding various aspects of the design. Some of the lessons we have learned have already been stated in this and the following text where we give the results from BCHED’s first installation. In general, designers should make sure that all parts of the RSF are easily serviceable.

One frequently asked question relates to how long the filter media can go without replacement.. While the media never really has to be replaced, the filter may eventually clog due to the unavoidable buildup of suspended solids which will not break down. The service interval will depend highly on the amount to TSS allowed on top of the filter. Accordingly, designers should incorporate effluent filters in all designs. When the unit must be serviced, the sand may either be replaced outright or cleaned and replaced. Richard Piluk, that County Health Department Official in Ann Arundel County Maryland, makes allowances for shutting the underdrain from his filter off. He then floods the filter with water and blows compressed air into the filter bed. As the water boils to the top, carrying the collected solids, he pumps them off and disposes of them similar to septage.

This non-proprietary technology can be configured many different ways and still comply with the required criteria. What is important for the designer is that you think ahead to the time when this pump or that float valve malfunctions. Safeguards and alarms are the first line of defense against system failure.

On the following pages are some of the results from a year’s worth of sampling at an RSF we installed in Bourne. As you will see, some of the results are quite variable, particularly in regard to nitrogen removal. We feel that the reason for this is the fact that for the first year of operation, we operated this, and one other system, on a demand cycle. In the coming year, we will be operating our systems closer to the state guideline mode of operation and expect better removal efficiencies. At the time of installation, no official state guidelines were available.


As many of you may know, our Department assisted in installing and monitoring a recirculating sand filter at the house of Paul Montague, who now serves as the Shellfish Constable in Falmouth. Paul has a two-bedroom home on the water in Bourne, and in 1994 decided to upgrade his cesspools to at least a Title 5. With some funding for design through the Buzzards Bay Project, Paul and his wife Edna bit the bullet and installed a recirculating sand filter. The system consists of a 1000-gallon septic tank, a 1000 gallon watertight pump chamber, a 2000 gallon recirculating sand filter and a leaching facility comprised of two leaching chambers surrounded by three feet of stone. Since July, 1994 on at least 20 occasions, we have been monitoring untreated effluent from the septic tank, water in the anaerobic pump chamber, sand filter effluent and groundwater from a well and suction lysimeters directly below the leachfield. A schematic of the system is presented in figure 1 at the beginning of this chapter.


Removal of fecal coliform exceeded 95% on all but five sampling dates (Figure 10). Fecal coliform densities in septic tank effluent ranged from 10,000/100 ml to 4,300,000/100 ml. Effluent from the sand filter ranged from 50-50,000 fecal coliform/100 ml. Occasionally, there is significant passage of fecal coliform through the sand filter, however the monitoring well placed directly beneath the leaching facility had fecal coliform levels reaching only 0-100 FC/100 ml indicating that the soil beneath the flow diffusors is acting to efficiently filter fecal coliform prior to reaching the groundwater.

Reduction in Biological Oxygen Demand (BOD) consistently exceeds 90% after treatment by the sand filter, and ranges from 97-98% efficiency when the water temperatures exceed 10 C (Figure 11). Septic tank BODs ranged from 151-344 mg/l with an average of 223 mg/l. Treatment through the sand filter reduced this to 2.3-18.0 mg/l and averaged 8.8 mg/l. The data show a clear seasonal trend related to water temperature. In March, a clogging layer was observed on the top of the sand filter that appears related to the reduction in BOD removal efficiency. After raking the top of the filter to break up the clogging layer, the filter returned to normal operation and has not clogged since.

Initially, the phosphorus removal from the system looked good (90% removal from July-September, 1994). From February, 1995- early May, the removal efficiency for phosphorus dropped to 30%. Since May, the efficiency has further dropped to 0-15% (Figure 12). We hypothesize that the removal of phosphorus is governed by the chemisorption of phosphorus onto surfaces of iron minerals. In time, it would be expected that adsorption sites would become saturated and soluble phosphorus would pass unattenuated through the filter. In the coming year, we may be experimenting with coating the top of the filter with iron-rich sand to determine whether we can again increase the phosphorus removal capability of the system.

Nitrogen removal efficiency, measured as loss of total dissolved nitrogen (TDN= nitrate + ammonium + dissolved organic nitrogen) has been variable. Average nitrogen loss over the year has been 32% (Table 1). TDN in septic tank effluent averages 70.6 mgN/l and TDN in the sand-filter effluent averages 48.2 mgN/l.

TDN Septic Tank Effluent TDN D-Box Finished Effluent
Mean 70.6 48.1
Standard Deviation 8.28 9.2
n= 21 21

Table 1.

Nitrification (or the conversion of ammonium from the septic tank to nitrate) appears to be complete, even during the winter months when water temperatures were low (Figure 13). Nitrification approaches 100 % when the water temperatures measured at the pump chamber exceed 10C. Despite what the conventional wisdom says about nitrification, it appears that this is not the limiting process in the overall denitrification, even during winter months.

Since nitrification does not appear to be the limiting step in loss of total nitrogen to the system, we can assume that denitrification (the conversion of nitrate to nitrogen gas) is. Denitrification rates do not appear related to temperature since levels are variable throughout the winter months. The availability of carbon (as measured by BOD in the anaerobic pump chamber) similarly does not appear to limit denitrification since levels there range from 8.4-65 mg/l which should be sufficient to support the growth of denitrifying bacteria. Presently, we hypothesize that the higher dissolved oxygen levels in the pump chamber (average 1.5 ppm from December-March), prevalent in colder months may have limited the denitrification step. In the coming months, however, we will be measuring pH and alkalinity in this pump chamber to see if these parameters may be affecting this vital step in the nitrogen-removal process.


The recirculating sand filter had only few operational problems. During late winter, the surface of the sand filter clogged with a fine organic layer and caused ponding of effluent. The filter resumed normal operation following a raking of the sand filter surface. Based on this experience, we recommend that there be quarterly servicing of the top of the sand filter. It should be visually inspected monthly during winter months. Experience with the sand filter in Fairhaven, however suggests that no clogging would have occurred if more uniform sand was used. In addition to this problem, the opening on the spay nozzles should be periodically inspected for clogging. In our case, leaves inadvertently entered the pump chamber during our sampling and clogged some of the ports.

Septic Tank vs. Sand Filter Effluent Results:

Firgure 10. Fecal coliform levels.

Figure 11. BOD removal.

Figure 12. Phosphorus removal.

Figure 13. Total dissolved nitrogen levels.

System Update

This past winter, we have retrofitted the Montague system with a timed dosed arrangement. We will be taking samples over the next year to see whether , in this instance, we increase the performance of the system and decrease both the slight odor and maintenance requirements. The dose setting is 5 minutes on, 25 minutes off – more in accordance with the state guidelines for recirculation rates. In addition, we replaced the distribution system on top of the sand filter with a 4-pipe lateral pressure distribution network that could be covered with peastone. The pressure distribution network was purchased from ORENCO, however, it could have been made from non-propriatary parts quite simply. Another change we made to Paul’s system has been the volume of liquid we maintain in the pump chamber. Initially, we only maintained one foot or so of liquid in the pump chamber (which, as you remember serves also as the denitrification chamber). We now understand that maintaining a higher volume in the pump chamber will help stabilize the temperatures in the winter and provide for more anaerobic conditions necessary for denitrification. Dissolved oxygen levels in the pump chamber are more affected by returning sand filter effluent when small volumes are in the pump chamber. Accordingly, and to maintain the required anaerobic conditions in the pump chamber, we decided to maintain larger volumes of liquid. Keep an eye out in future newsletters to get the results of these modifications.