Secondary Treatment Gets You to Thirty. Reuse Needs Ten. That Gap Is the Whole Filter.
The drawing came to us with the treatment train written down the side of the sheet as a list of boxes, and the second-to-last box said tertiary filtration. One line. No size, no media, no chemical, no backwash rate, and no indication that anybody had decided what the box was for. The plant underneath it was a conventional biological plant for a mid-sized building, already drawn and already priced, and the owner's question to me was whether that box was expensive. It is a reasonable question and it has no answer in that form, because that one line is where a plant stops being a compliance plant and starts being a reuse plant, and the cost of it depends entirely on which of those two things he wanted.
This note is about that box. We have written about whether reuse pays for itself at all, and about which level of reuse to treat to, and both of those posts name tertiary filtration in passing as a line in a train without ever saying what the thing does. That is the gap I am filling here. It is also not the media filtration note I wrote a couple of days ago, which was about the vessels standing in front of a membrane on clean well water: there the load is mineral and the enemy is hardness and iron. Here the water arriving at the filter has just come off a clarifier and the load is organic and biological, which changes the media argument completely and is the reason the two notes do not collapse into one.
What the secondary plant has already done, and the twenty milligrams it leaves behind
Start with the numbers we design against. Our own long-standing description of sanitary wastewater puts the raw strength at COD around 600 to 800, BOD around 300 to 400, TSS 100 to 200, TKN 100 to 200 and oil and grease 300 to 400 milligrams per litre. Those are the figures we have used for years for household and commercial sources, and they are the reason we draw at several strengths rather than one whenever nobody can tell us where a number came from — which, as we have said plainly, is most of the time.
A properly run activated sludge process takes that to BOD under 30 and TSS under 20. That is the honest expectation, not a sales figure, and it is what a conventional plant with FBBR-style fixed-film media in the aeration stage will hold when the sludge age and the air are right. Reuse and irrigation, limited or unlimited, need BOD under 10 and TSS under 10.
So the entire job of the tertiary stage, in arithmetic, is thirty down to ten and twenty down to ten. That is all. It is not a treatment stage in the sense that the aeration tank is one. It is a polishing stage that exists to close a gap of about twenty milligrams per litre of organic matter and ten of solids, and the whole of its cost, its chemical bill and its maintenance burden has to be justified against closing that gap.
Compliance does not need this box. Reuse does.
Put the secondary numbers against the discharge standard and the point makes itself. BOD under 30 and TSS under 20 sit inside the general effluent standards for the classes most of our clients discharge to, and for the stricter classes the BOD figure sits close enough to the limit that the margin is worth checking rather than assuming — the limits are on our compliance page and we have gone through what DAO 2021-19 changed and did not change. A secondary plant that is working does not need a filter to pass a BOD or TSS test. The parameters that actually fail are the nutrients, and they fail for reasons a filter cannot touch, which I will come back to.
So when a drawing arrives with a tertiary filter in it, the first thing to establish is whether the building is going to reuse the water or discharge it. If it is discharging to a watercourse and the receiving water class does not demand it, that box is money spent on quality nobody is going to use — which is the same argument as the job where we talked a client down from the cleanest water we could build. If it is going to flush toilets or water a garden or feed a wash-bay, the box is not optional, because at TSS 20 the water will mark tiles and load a storage tank, and the rung of the reuse ladder you are on is what sets the specification rather than any general idea of cleanliness.
That is a design conversation and it takes about ten minutes. It happens far too late far too often, usually because the box was copied off a previous drawing by somebody who correctly assumed a tertiary stage was good practice and never asked what it was for.
The conventional train, in the order the water meets it
The ordinary tertiary stage, the one we would build if nobody specified anything unusual, is three things in series after the clarifier.
- A mechanical pre-filter to around 100 microns. Normally a screen or a disc filter. It is there to catch the occasional slug — the carry-over event rather than the steady load — so that the media filter behind it is not shock-loaded by something the clarifier let past on a morning peak.
- A multimedia filter, gravity or pressure. Conventionally a bed built up of gravel, quartz sand and anthracite in layers. This is the stage doing the work.
- Disinfection — chlorine, ultraviolet, or both. To bring the bacteriological load down. This is the stage people remember, because it is the one with a contact chamber attached to it, and that chamber has its own way of undoing your phosphate result.
If a coagulant is dosed ahead of the media filter — to pull the fine particles together so the bed can catch them — the arrangement has a name: contact filtration. That dosing point is the single biggest decision in the box and almost nobody treats it as one. More on it below.
A filter is sized by area, and it is the backwash that bites
This is the arithmetic worth taking away, because it is the part that surprises people who are used to thinking about tanks.
Every other chamber in a sewage plant is sized by volume — you need so many hours of retention, so you need so many cubic metres, so the walls go here. A filter is not. A filter is sized by bed area, because what matters is how fast the water is travelling downward through the media. The working window we use is a filtration velocity of 5 to 15 cubic metres per hour per square metre of bed. The bed depth is a separate matter: two supporting layers of about 30 and 20 centimetres of the coarser grades, with 80 to 120 centimetres of the main grade above them, running deeper in a gravity filter than in a pressure vessel.
Work it through on a 150 cubic metre a day plant, which is the size of building we see most. That is about 6.3 cubic metres an hour averaged over 24 hours. At 10 cubic metres per hour per square metre, you need about 0.63 square metres of bed. That is a vessel under a metre across. The filter is a small object. People expect it to be a big one, and it is not.
Now the backwash. The backwash velocity range is 30 to 50 cubic metres per hour per square metre, because the point of a backwash is not to pass water through the bed but to lift and tumble it. Take 40 against that same 0.63 square metres and the backwash duty is about 25 cubic metres an hour — four times the entire plant's average hourly flow, from a vessel smaller than a drum.
That number is what sizes the pipework, the pump, the valves and the drain, and it is the number that gets left off a one-line specification. We have made the same point about a softener that had plenty of capacity and could not keep up with two showers: a vessel has a service duty and a wash duty, they are different numbers by a factor of several, and the one on the datasheet is usually the smaller one.
Where the backwash goes is the real design decision
Here is the part that makes a tertiary filter a wastewater problem rather than a filtration problem. The filter does not destroy anything. It collects the last twenty milligrams per litre of solids and holds them in the bed, and then the backwash takes them out again in a concentrated form. Those solids have to go somewhere, which is the same argument we make about everything else that comes out of a plant.
In practice the backwash goes back to the head of the works, usually to the equalisation tank. That is normally right and it is what we draw. But it means you have built a loop, and the loop has to be counted: a slug of 25 cubic metres an hour arriving at the front of a plant whose average flow is 6 is a hydraulic event, not a trickle, and if the equalisation volume was already thin — or given up earlier in the drawing for something else — the backwash will show up as carry-over at the clarifier a couple of hours later. That is a fault that reads as a clarifier problem and is actually a filter washing itself at nine in the morning, and it is worth eliminating before anybody adjusts the clarifier.
Two things follow. Backwash on a timer at a known hour beats backwash on differential pressure at an unknown one, on a small plant, purely so that the loop lands at a time somebody chose. And the backwash volume belongs in the equalisation sizing, not in a footnote.
Coagulant ahead of the filter is a standing order, not a line item
Dosing a coagulant in front of the media is the cheapest way, in capital, to get a conventional bed from TSS 20 down under 10. It works. It is standard practice and we have specified it. What it also is, is a chemical the building now has to buy, store, dose and never run out of, every week, for as long as the plant runs.
Our position on that is long-standing and it is a design rule rather than an opinion: we will not let compliance depend on a chemical that may not be available locally. A plant that passes with the coagulant and fails without it has a supply chain in its compliance path. On a site with a maintenance contract and a storeman that is a manageable risk. On a building where the plant is run by whoever is available — which, as we have written, is most buildings, and the turnover is the design input — it is not a risk, it is a forecast.
Two honest footnotes on the coagulant, both of which cut the other way. A coagulant dosed ahead of a filter will also drop phosphate, often substantially, which means some plants are passing their phosphate limit on a chemical they think they are dosing for solids — and will fail it the month somebody economises. And a metal-salt coagulant consumes alkalinity, which on a plant already nitrifying near its buffer limit is a second effect nobody dosed for. The dosing point is never just about the filter.
There is a third effect, and it only appears on the water treatment side of our work, so I note it here rather than argue it: where a membrane sits downstream, an aluminium-based coagulant has been reported to combine with the free silica a sand bed gives up and scale the membrane — the chemical dosed to help the filter turning on the thing the filter was protecting. It does not arise on a sewage plant discharging to a watercourse. It is worth knowing about before anybody carries a dosing habit from one kind of plant to the other.
The media choice is a fouling argument, and the fouling here is organic
Our own technical note on the tertiary stage states the problem with a conventional bed in this duty plainly, and it has been on our site for years: because wastewater carries a high organic and bacteriological load, there is a high potential for organic and biological fouling to develop on the media surface. That leads to frequent air scouring during regeneration, and eventually the media has to be replaced. That is the honest failure mode of a quartz-and-anthracite bed on secondary effluent, and it is a maintenance and replacement cost rather than a dramatic breakdown.
The alternative we supply is a glass-based activated medium in place of the sand and gravel. The manufacturer's published figures for it in this duty — and I want to be clear these are the supplier's numbers, not measurements from our own plants — are that it is bio-resistant with a high surface oxidation potential, that it will filter secondary effluent to under 10 milligrams per litre of BOD and TSS with no coagulant or a minimum dose, that air scouring is normally not required, that chlorine use during regeneration is eliminated with total chlorine use down by up to 70 per cent, and that the expected media life is 10 to 20 years. We keep that note and a separate one on the media's published specification on the site, and we have written about how it is layered inside a standard vessel.
I will say what I would say to a client's face, which is the same thing I said about our own fixed-film media holding nitrifiers better than moving media: I believe the no-coagulant claim because of what we have seen on our own jobs, and I cannot hand you a set of side-by-side trials on a Philippine sewage plant to prove it, because we have not run them. What I can tell you is which way the decision breaks. If the bed is going to fail by organic fouling rather than by abrasion, then media that resists biofilm is worth paying for and media life is the figure to interrogate. If the plant has a competent operator and a reliable coagulant supply, a conventional bed is not bad engineering and the cheaper bed may be the right answer. We also put the same question to existing gravity filters we are asked to improve — very often the fix there is the media and the underdrain, not a new filter.
What this box will not do for you
This matters more than any of the above, because it is where money gets spent on the wrong machine. A tertiary filter removes suspended material and the organic matter travelling inside it. It does essentially nothing for what is dissolved.
Look back at the influent figures: TKN 100 to 200 milligrams per litre. Ammonia and nitrate are dissolved. They pass through a sand bed, a glass bed and a disc filter alike, and ammonia and nitrate are precisely the parameters that fail first in the results that reach us — the whole of our nutrient field guide exists because of it. If your lab sheet shows ammonia and phosphate failing, a filter is not the answer; the answer is in the sludge age and the air, or in the anoxic stage and the recycle ratio.
Nor will it rescue a secondary plant that is not finished. A filter on a plant whose aeration is washing its nitrifiers out will simply blind faster and more often, and the operator will conclude the filter is faulty. The order of work is the same as always: get the biology right, get the clarifier settling properly, and only then polish. Filtering a failing plant is paying for a bandage on the wrong limb — and a filtered BOD reading alongside the total will tell you in one lab visit which of the two problems you actually have.
What we ask before we draw one
These are the questions, in the order we ask them, and none of them needs a laboratory.
- Is this water being reused, or discharged? Discharged and inside the class limits, the filter may not be needed at all.
- If reused, for what exactly? Garden, washdown, cooling, flushing. The rung sets the specification; our design generator walks the same ground.
- What is the real hourly flow the filter has to pass, and is it the average or the peak? Bed area comes off that figure and nothing else. The water bill is usually the most reliable source.
- What is the backwash rate, where does the water go, and at what hour? Several times the service rate, into the equalisation volume, at a time somebody chose.
- Is a coagulant in the compliance path? If yes, who buys it, where is it stored, and what happens in the month it does not arrive.
- What is the media, and what is its expected life in this duty? A bed that must be replaced every few years is an operating cost wearing a capital cost's clothing, which belongs in the cost per cubic metre rather than in the purchase price.
If a quotation answers one line of those six, it is not yet a specification. We have gone through how to read two quotations against each other and this box is one of the places the scope quietly differs; the price estimator will give you a sense of the order of magnitude a polishing stage adds before anybody quotes you.
Where this carries, and where it does not
The arithmetic carries anywhere — velocity sets area, backwash is several times service, and dissolved things pass through filters. The rest of it is conditional, and I would rather say so.
The thirty-to-ten framing assumes a secondary plant that is actually reaching 30 and 20. A great many are not, and on those plants everything above is premature. The velocity window of 5 to 15 is a range because the right number inside it depends on the solids load arriving and on how often you are willing to wash; we sit low in it on effluent we do not trust. The media life figures are the supplier's, in a duty they describe generically, and a Philippine building with a kitchen on it is not a generic duty — hard water and kitchen fat change what arrives at any filter. A process plant making most of its flow in three hours is a different filter problem from an office tower with the same daily total, because the peak is what the bed sees. And on a very small plant the honest answer is sometimes that the backwash loop and the dosing habit are more plant than the site can carry, and a simpler train with a lower reuse ambition will serve the owner better than a filter nobody washes.
We keep a separate note on the same filtration decision on a cooling tower side-stream, where the fouling is mineral as well as biological and the arithmetic comes out differently again. Different water, same question: what is the bed actually failing by.
If you have a drawing with a one-line tertiary box in it, or a plant that is passing its discharge test and not getting the reuse quality somebody promised, send us the train and the lab sheet and we will tell you which of the six questions above is unanswered. Book a free 15-minute consultation and we will go through it with you. If you would rather start with a scope, our proposal request asks the same questions in writing, and the wastewater services page sets out what we do and who runs it afterwards.