Our Own Page Says the UV Is a Safeguard Against Bacteria. It Says Bacteria on Purpose.

Step twelve of our own water refilling station process flow page says this: the purified water is passed through a UV sterilizer as a final safeguard against bacteria. The summary at the bottom of that page says the final filtration and UV system will make sure the water tastes good and is bacteria free. I wrote that page years ago and I still think it is correct. What I did not appreciate at the time is how much work the word bacteria is doing in that sentence, and that the sentence is accurate precisely because it does not say sterile.

I should be plain about where the rest of this comes from, because it is not a Cordova measurement. Our own work is overwhelmingly sewage treatment, and the UV units we specify sit at the end of refilling station trains and in reuse trains, chosen off a flow rating like everybody else's. The dose figures below come from a supplier's published case note for a marine fish hatchery in the Eastern Mediterranean — sea bass, 150 m³/hr, 35 ppt salinity, turbidity between 0.1 and 1 NTU — which we keep on the site as our reference note on treating marine hatchery water. Every number in it is the supplier's published figure for that duty, not something we have measured. This post is not the same kind of post as the ones where we give you a proven rate we costed ourselves, or a cone reading off a plant we run. It sits with the note we wrote on chromium reduction, which was also worked on paper rather than built. The difference matters and I would rather say so than let you assume otherwise.

Dose is not wattage, and it is not litres per minute either

Ultraviolet disinfection has one governing quantity and it is a dose: the energy actually delivered to a unit area of the water, usually written in millijoules per square centimetre. It is an intensity multiplied by an exposure time. The supplier's note writes it as mJ/cm²/sec, which I think is loose — dose is mJ/cm², and the per-second belongs to the intensity, not the dose — but the figures are the ones the industry quotes and the ladder they describe is the useful part.

What a vendor hands you instead is almost never a dose. It is a lamp wattage, or a flow rating in gallons per minute, or a model number. None of those three things is a dose. A wattage tells you what the lamp draws, not what reaches the far wall of the chamber. A flow rating tells you how fast water can pass through before the exposure time becomes too short, which is one of the two factors, with the other one left out entirely. And a model number tells you nothing at all.

This is the same shape as a problem we have written about twice before on this blog in different clothes. We have written about carbon sold on a bed volume when the figure that decides whether it works is the iodine number, and about a specification built on a figure that was real, authoritative, and measuring something else. A UV unit sold on wattage is a third member of that family. The number on the box is not false. It is just not the number that decides the outcome.

The ladder: sixty, a hundred and twenty, two hundred

Here is what the supplier's hatchery note actually sets out, and the reason it changed how I read our own pages.

For bacteria, following good mechanical filtration, their figure is a dose rate of 60 mJ/cm². For a virus — the case in the note is infectious pancreatic necrosis — 120 mJ/cm² is required to denature the RNA. And because viruses have a very high resistance to UV and nobody wants to find out they were at the edge of the window, systems for that duty are usually rated at 200 mJ/cm² as a safeguard. The note is explicit that parasites come out by good filtration down to 5 microns, bacteria by filtration plus UVc at the 60 figure, and viruses only at the higher doses.

So the step from a bacterial duty to a viral one doubles the dose on the requirement — 60 to 120 — and more than triples it on what actually gets installed, because nobody specifies to the edge. That is not a knob you turn on an installed unit. It is a different machine, and in most cases a different chamber geometry and a different number of lamps. A unit specified against bacteria is not a weaker version of a unit specified against viruses. It is a unit for a different organism.

Which brings me back to our own page. It says bacteria. On a refilling station fed from a utility supply or a deep well, run through multimedia filtration, carbon, softening, a cartridge and a membrane, bacteria is the honest target for the final lamp, and the membrane upstream has already done most of the microbiological work. The page is not underselling the unit. It is describing it correctly, and the reason it does not say sterile is that the unit is not a virus barrier and nobody should buy it believing it is one. We made the same distinction from the other direction in the post on whether you can drink treated wastewater, where the point was that chlorine at the dose we use at the end of a sewage plant handles bacteria and does not deal with viruses or with protozoa like Giardia and Cryptosporidium. Same gap, different barrier.

The number that actually sets the dose is the clarity of the water

This is the line in the supplier's note that I would put at the top of every UV quotation if it were up to me: the irradiation level will be a function of the UVc transmission through the water.

Read that slowly, because it relocates the whole problem. The delivered dose is not a property of the lamp. It is a property of the lamp and the water in front of it. Ultraviolet at the germicidal wavelength is absorbed and scattered by what is dissolved and suspended in the water it is crossing. Turbidity scatters it. Dissolved organics and colour absorb it. Iron absorbs it. So the same lamp, at the same flow, in two different waters, delivers two different doses — and the shortfall is invisible, because there is no reading at the outlet that tells you the dose fell.

Notice the condition the hatchery figures were quoted under. Turbidity 0.1 to 1 NTU. The 60 and the 200 are doses achievable in water that has already been filtered to near-drinking clarity. Quote the same numbers against a secondary effluent, or against a deep well carrying iron, and they are not wrong so much as unreachable.

The design consequence is the useful part. If the delivered dose is short, the first place to look is not a bigger lamp. It is upstream. Better filtration raises the transmission, and raising the transmission raises the dose from the lamp you already have. That is a cheaper correction than a larger UV chamber and it fixes two problems at once. Iron in particular is worth treating properly rather than hoping a membrane or a lamp will carry it, which is one reason we keep a separate reference note on iron and manganese removal, and why our post on refilling station pre-treatment spends most of its length on the three vessels in front of the membrane rather than on the membrane.

Filtration and disinfection are each other's precondition

The hatchery note makes a second point that is easy to skim past. After the coagulation stage and the filtration stage — the mechanical half of that train, which is good filtration by any standard — the organism concentration was down by 50% to 90%, and the note says plainly that this is not sufficient to protect the fish. So the filtration, good as it was, did not substitute for disinfection.

And the disinfection could not have worked without the filtration, because the filtration is what made the water transmissive enough for the dose to arrive. Each stage is the other's precondition. You cannot trade one against the other, and a train that is strong on one and thin on the other is not a compromise design, it is a design with a hole in it.

That is the same argument as the one in our note on tertiary filtration for reuse, where disinfection is listed as the last stage and the multimedia filter ahead of it is described as the stage doing the work. I had the order right in that post without quite saying why the order is non-negotiable. It is non-negotiable because of transmission. The contact chamber attached to that last stage has its own separate way of spoiling a result, which is something we have written about on its own, and if your disinfection route is chlorine rather than a lamp then our honest view of breakpoint chlorination is the companion piece to this one.

The gallons-per-minute rating is measuring something else

Here is where this lands on equipment people in the Philippines actually buy. The UV units in our refilling station packages are rated at six, eight and twelve gallons a minute, and they sit after the product tank and the product booster pump. We have written about that rating before, in the context of a plant having three different flow numbers that are not derived from each other: those units are not sized on production at all, they are sized on dispensing, because dispensing happens in bursts at whatever the faucets can draw.

A flow rating is a real constraint and you should respect it. Exceed it and the exposure time falls and so does the dose. But two units with the same flow rating are not interchangeable, because the rating says nothing about the intensity inside the chamber, nothing about the dose at that flow, and nothing about the organism the dose was chosen for. If you are comparing two offers, the question that separates them is: at my design flow, in water of my transmission, what dose does this chamber deliver, and which organism was that dose chosen to inactivate? If neither supplier will put a figure in millijoules per square centimetre on paper, you are not comparing disinfection performance. You are comparing enclosures. That is the same discipline as reading two competing quotations line by line, and it belongs in the same part of the conversation as what we ask of a piece of equipment before we will specify it.

Peroxide or ozone immediately before the lamp, and the seven times

The last stage in the supplier's hatchery train is the part I find most interesting, because it is the only place in that note where the chemistry does more than the arithmetic suggests.

Before the water reaches the UVc chamber, a small quantity of oxidant is injected: hydrogen peroxide at 1 to 6 mL per cubic metre of water treated, or ozone at about 0.1 mg/L as an alternative. Then the water passes immediately into the chamber. The note's claim is that the presence of that small amount of peroxide or ozone is catalysed by the UVc to generate up to seven times the concentration of hydroxyl radicals, and that the combination is essential to ensure adequate treatment. Not the lamp alone and not the oxidant alone — the two in sequence, with the oxidant still present when the light hits it.

That is advanced oxidation, and it is recognisably the same mechanism we use at the top of our own reuse ladder. For us advanced oxidation means ozone catalysis, with a membrane stage behind it only where the case really calls for one, which is the position set out in the post on choosing a reuse treatment level and in the one on when reuse actually pays. We have also put the combination into a real train before: a treatment system for a thirty-bed hemodialysis centre in Metro Manila, about 30 cubic metres a day, where bringing the effluent to the Class C discharge standard needed ozone with venturi injection, catalytic granular carbon, UV, chemical phosphate removal and forced exhaust for the off-gas, all of it described in the post on drinking treated wastewater that I linked further up. So the ordering principle is not foreign to us. What the hatchery note adds is the magnitude — a factor of up to seven on radical concentration from a dose of oxidant small enough to be a rounding error on a chemical bill — and the insistence that the oxidant must still be in the water when it is irradiated. Dose it too far upstream and you have bought an oxidant and a lamp and not the combination.

How this gets worse quietly

A UV unit has no honest outlet reading. The lamp still glows while its germicidal output falls with hours run, and the quartz sleeve fouls with whatever the water leaves on it, and both of those reduce the delivered dose without changing anything the operator can see. There is no equivalent of a differential pressure climbing on a filter or a hardness breakthrough on a softener.

So the dose that matters for design is not the dose on the certificate on day one. It is the dose at the end of lamp life, through a sleeve that has not been cleaned as often as the manual asks, in water that is a little less transmissive than the day the system was commissioned. If that number is still above the dose the organism needs, you have a disinfection barrier. If it is not, you have a lamp.

That makes it a maintenance item with a date on it rather than a thing you look at, which is exactly the category of task that a written daily and monthly log exists to catch, and exactly the kind of failure that is not the operator's fault to find. We have written about what a good operator can and cannot fix, and an invisible dose shortfall is firmly in the cannot column unless somebody wrote the sleeve-cleaning interval and the lamp replacement hours into a schedule at handover. It also sits beside the slow filter failure we wrote about that takes about six months to show up — the two failures have nothing in common chemically and everything in common in how they are discovered, which is late.

What we would put in writing instead of a wattage

If you are specifying or buying UV for a refilling station, a reuse train or a tertiary stage, these are the things we would want on paper, in the order you would actually chase them.

  1. The organism the dose is chosen against. Bacteria, or viruses, or protozoa. Not "disinfection".
  2. The required dose for that organism, in mJ/cm², with the source of the figure named.
  3. The delivered dose of the proposed chamber at your design flow, not at its rated maximum.
  4. The UVc transmission the delivered dose was calculated at, and a measurement or a defensible estimate of the transmission of your actual water.
  5. What the delivered dose becomes at end of lamp life, which is the only version of the figure you will live with.
  6. The sleeve cleaning interval and the lamp replacement hours, written into the maintenance schedule at handover rather than left in a manual.

Those six items are all paperwork and phone calls. Only the third and fourth look like engineering, and the fourth is the one most likely to come back as a shrug. If it does, that is information too.

If you are at the earlier stage of working out what treatment level you need at all, our wastewater treatment services page sets out how we approach a plant, the compliance page covers what DAO 2016-08 and DAO 2021-19 require of the effluent, and the price estimator and the design generator will give you an indicative shape before anyone visits. For drinking water trains specifically, the refilling stations page, the process flow note this post started from, our note on the proposed method of treatment and the comparison of refilling systems are the four pages worth reading together. Hatchery and aquarium work sits on the aquaria and aquaculture page, and the filtration side of pre-treatment ahead of a membrane is covered in our reference note on pre-treatment ahead of reverse osmosis and in the tertiary treatment note. If what you have is an existing plant that is failing a bacteriological parameter, our consultations page and the maintenance and operation page are the right starting points, and we have written about why a coliform failure and an ammonia failure often have the same cause, which is worth ruling out before anybody buys a lamp. If it is a new plant you are after rather than a repair, the page on getting a sewage treatment plant explains how we start, our Cebu page covers where we work from, the completed projects page shows the kind of plants we build, and the significant parameters list will tell you which parameters your industry is actually measured on, since a bacteriological limit is not on every establishment's list.

What travels from this, and what does not

The structure travels. A disinfection barrier is specified against an organism at a dose, the dose is delivered only to the extent the water lets the light through, and the figure you must design to is the one at end of lamp life rather than the one on the certificate. Ask which organism, ask for millijoules per square centimetre, ask at what transmission. Those three questions cost nothing and they separate a specification from a catalogue page.

The ordering principle travels too, and it is the part I would keep if I had to keep one thing: filtration and disinfection are not alternatives at different price points. Each is the condition under which the other works.

The numbers do not travel. Sixty, a hundred and twenty and two hundred are a supplier's published figures for a marine hatchery at 35 ppt salinity, 25 degrees, and a turbidity between 0.1 and 1 NTU, in the Eastern Mediterranean. Philippine deep well water is not that water and a secondary effluent in Cebu is much further from it again. The peroxide and ozone doses are theirs as well, as is the factor of seven, and none of it is a measurement we have taken in Cordova. Our own pages are honest about the narrower thing we have actually done: a final lamp as a safeguard against bacteria, behind filtration, carbon, softening and a membrane. If your duty is wider than that, the equipment has to be chosen on a number nobody has yet shown you, and the first useful step is to ask for it.

If you have a bacteriological result in front of you, or a UV quotation with a wattage on it and no dose, send it over and we will read it with you. Book a free 15-minute consultation.