UV Water Disinfection Sizing Calculator — the 40 mJ Dose and the Transmittance That Decides It

NSF 55 Class A Versus Class B

The first decision on a UV system is not a size. It is which of two certifications applies, and they describe different devices for different water.

NSF/ANSI 55 Class A systems are certified to deliver a dose at least equivalent to 40 millijoules per square centimetre at the alarm set point. They are intended for disinfecting water that is microbiologically unsafe, which on a private supply means a positive coliform result or a source with reason to suspect one.

NSF/ANSI 55 Class B systems are certified to 16 millijoules per square centimetre. They are intended for supplemental treatment of water that has already been tested and deemed acceptable for human consumption, as additional confidence rather than as the barrier the supply depends on.

Published regulatory guidance draws the line without qualification: Class B certified systems should not be used for the production of potable water.

So a Class B unit is not a smaller Class A unit. It is a different device with a different job, certified against a dose less than half as large, and choosing one for an unsafe supply is a category error rather than an under-specification. This calculator treats that combination as a hard stop rather than a warning.

Untested water sits outside both. The class depends on whether the UV system is producing potable water or adding confidence to water already known to be acceptable, and that is a laboratory question rather than an engineering one. A microbiological test comes before the system selection, not after it.

Where a project or an authority sets a target dose above the Class A figure, that figure governs. Where anyone proposes a target below it on an unsafe supply, nothing does.

UV Dose and Rated Flow

Ultraviolet dose is the intensity of the light multiplied by the time the water spends in it, expressed in millijoules per square centimetre at a germicidal wavelength of 254 nanometres. Both terms are outside the lamp control.

The flow sets the time. Water moving through the chamber twice as fast spends half as long in it and receives half the dose, so dose is inversely proportional to flow. That relationship is why the rated flow on a system is a ceiling rather than a target: at the rated flow the certified dose is delivered, and above it, it is not.

This is the reason an NSF 55 Class A system is required to carry a built-in flow restrictor or automatic fixed flow rate control. The certification is not a claim about the reactor alone; it is a claim about the reactor operating at or below a stated flow, and the restrictor is what makes that claim enforceable in a house where somebody opens three taps at once.

It also means the flow figure to use is the peak instantaneous flow through the reactor, not an average and not a daily total. A house served by a pressure tank and pump does not draw water smoothly. It draws in bursts, and during a burst the dose is whatever the burst flow produces.

Read the other way, the relationship is a design lever. Running a system at half its rated flow doubles the dose, which is exactly what the manufacturer guidance for low transmittance water is doing when it recommends a system with more capacity than the duty requires.

What the rated flow does not tell you on its own is what water it was rated on. That is the next section.

UV Transmittance and Flow Derating

Ultraviolet transmittance is the percentage of light at 254 nanometres that passes through one centimetre of water. It is measured in a laboratory on a sample, and it is the single most important water quality figure for a UV system.

The reason it matters more than it sounds is that the light has to cross the whole annular gap between the lamp sleeve and the chamber wall, and the loss is taken at every centimetre. The fraction of intensity reaching a point is the transmittance raised to the power of the path length in centimetres, so the effect compounds rather than scaling.

Across an illustrative five centimetre path: 95 percent transmittance delivers about 77 percent of the intensity, 85 percent delivers 44, 80 percent delivers 33, 75 percent delivers 24, and 55 percent delivers 5. Read against the 95 percent case that most ratings assume, those are 57, 42, 31 and 6 percent.

Because the dose must be held at the certified figure, the flow has to fall in the same proportion. That is what derating means here, and it is why a system rated comfortably above a household peak can still fail to deliver its dose.

The path length is a reactor dimension that manufacturers do not publish, so those figures show the shape of the relationship and are not a capacity for any product.

What the manufacturers publish instead is a rule of thumb. Where the transmittance falls between 75 and 95 percent, published guidance recommends choosing a system with about 50 percent more flow capacity than the duty requires. That works out to running at about 67 percent of the rated flow, which is generous at the top of the band and short at the bottom, because the underlying effect varies by roughly a factor of two across it. It is offered as a rule of thumb rather than a curve, and the same guidance invites a call for a specific case.

Below 75 percent transmittance after pre-treatment, the published route is activated carbon filtration to improve the transmittance itself rather than a larger system. Where the transmittance is unknown, the published advice is to choose a model with more flow capacity than the duty requires, which is the same rule applied to an unmeasured case.

Rated Flow Basis, 95 Percent Versus 70 Percent UVT

Two systems can carry flow ratings that look comparable and are not, and the reason is published by the manufacturers themselves.

Published guidance states that flow rates for one major range are based on water at 95 percent transmittance, except for NSF validated systems, which use 70 percent transmittance as the basis for the flow rate calculation.

That is a large difference in the condition being described. Seventy percent transmittance is a much harder water to treat than 95, so a system rated at a given flow on 70 percent water is delivering its dose under conditions where a 95 percent rated system would be well short.

The practical consequence: a 15 gallon per minute rating on a 70 percent basis can describe a more capable system than a 20 gallon per minute rating on a 95 percent basis. Comparing the two numbers alone gets the answer backwards.

So a rated flow without its transmittance basis is not a capacity, in the same way that a valve flow range without its pressure differential is not one. This calculator asks for both and reports them together.

There is a second condition on those ratings that is easy to miss. Published flow rates are quoted at the end of the stated lamp life, and one series states the basis explicitly as after 9,000 hours of bulb operation. The rating therefore already accounts for the lamp aging across its rated life.

That matters in a specific way: adding an arbitrary extra margin for lamp aging on top of such a rating is double counting. The correct response to lamp aging is to replace the lamp on schedule, not to oversize the reactor to survive an old one.

UVT by Water Source

Before a sample goes to a laboratory, the source type narrows the answer considerably, and it costs nothing.

Published manufacturer guidance gives two bands. Water drawn from a drilled well typically has a transmittance of 85 to 97 percent. Water drawn from a dug well or a surface source such as a lake or river typically has 55 to 80 percent.

Those bands do not overlap, and the entire surface water band sits below the 85 percent that several manufacturer manuals state as their requirement. So a drilled well is likely to be inside the window and worth measuring to confirm, while a dug well or surface intake is likely to need treatment before a UV system is a candidate at all.

Two source types sit outside those published bands and both need care.

A spring is shallow and frequently surface influenced. It behaves more like a dug well than a drilled one, so this calculator treats it as a shallow source requiring measurement and does not assign it the drilled well band. Assuming otherwise gives a favourable band to a source with no claim to it.

A municipal supply usually has a high transmittance and still needs measuring, because organics and the disinfectant residual both affect the figure. A municipal installation is also usually a supplemental one rather than the primary barrier, which changes the class question as well as the transmittance one.

The same guidance recommends a transmittance test specifically for surface or shallow well sources, and for water that is even slightly coloured. Colour is the clue that absorption is happening, and it is visible long before it is measurable by eye as a number.

UV Pre-Treatment Requirements

A UV system works on water that is already clean enough for the light to get through. The limits that define that are consistent across four independent manufacturer installation manuals, which carries more weight than any single data sheet.

Hardness below 7 grains per gallon, which is 120 milligrams per litre. At or slightly below that figure the quartz sleeve must be cleaned periodically; above it the water should be softened.

Iron below 0.3 parts per million. Manganese below 0.05. Turbidity below 1 nephelometric turbidity unit. Tannins and organics below 0.1 parts per million.

Each has a mechanism, and none of them is solved by a larger reactor.

Hardness scales the quartz sleeve, which blocks the light before it enters the water at all. The lamp is working perfectly and the light is not getting out.

Iron does both things at once. It absorbs ultraviolet in solution, and it scales and stains the sleeve. Dissolved ferrous iron is invisible until it oxidises, at which point the particles scatter the light as well.

Manganese behaves the same way at lower concentrations, oxidising into dark particles that absorb ultraviolet wavelengths.

Tannins absorb the light directly. They are the usual reason for a low transmittance in water that looks completely clear, which is why clarity is not a substitute for a measurement.

Turbidity fails by a different mechanism entirely, and that has its own section below.

Pre-filtration of at least 5 microns ahead of the system is recommended by several manufacturers as a baseline regardless of the measured turbidity.

One coincidence is worth knowing. The iron and manganese limits of 0.3 and 0.05 parts per million are the same figures as the secondary drinking water standards for staining. That is a coincidence of numbers rather than a shared basis, since the standards exist for staining and these limits exist for light transmission, and it is useful: a water only just acceptable aesthetically for iron is already at the boundary for ultraviolet, and the iron filter that fixes one fixes the other.

Why Turbidity Fails Differently

Every other pre-treatment parameter reduces the amount of light. Turbidity does something else, and the distinction changes what the remedy is.

Suspended particles do not primarily absorb ultraviolet. They shield organisms from it. A bacterium sitting behind a grain of silt is in a shadow, and the light passing on either side of that grain does it no harm at all.

That is why turbidity cannot be answered with a larger system. A bigger reactor or a stronger lamp delivers more light to the same shadow, and the organism behind the particle receives the same protection it did before. The dose is delivered to the water and not to the target.

Published guidance puts the scale of it plainly: turbidity above 5 nephelometric turbidity units reduces effectiveness by 50 percent or more, because suspended particles create shadows that protect bacteria from exposure. The manufacturer limit is 1 nephelometric turbidity unit, well below that.

The remedy is filtration. A 5 micron sediment filter ahead of the system is the standard baseline, and where the turbidity is high a backwashing multimedia filter is the usual answer. Both work by removing the particle rather than by out-powering it.

The same shadowing logic applies to anything particulate in the water, which is why oxidised iron is a UV problem as well as a staining one. Iron that has come out of solution is a particle, and particles cast shadows.

This is also the reason a transmittance measurement and a turbidity measurement are both worth having. Transmittance captures absorption well and shadowing imperfectly, because a laboratory sample and a flowing reactor do not see particles the same way.

Lamp Life, Glow and Intensity Monitoring

This is the part of a UV system that has no honest signal, and everything in the certification exists because of it.

Published manufacturer guidance states the position without hedging: the lamp lasts 9,000 hours of use, approximately one year, after which it must be replaced even though it may continue to light, because after 9,000 hours its ability to kill bacteria drops below safe limits.

The visible glow comes from the mercury discharge, and the germicidal output is a specific wavelength within it. The two decline at different rates, so the part a person can see outlasts the part that does the work. A lamp at the end of its germicidal life looks exactly like a new one.

Lamp life is product specific. Some lines are rated at 9,000 hours or one year, others at 18,000 hours or two years, and one source gives 9 to 12 months of effective germicidal output. The rated life comes from the lamp manual rather than from a general figure, which is why this calculator takes it as an input rather than assuming one.

The monitor distinction is where this becomes a purchasing decision. Published guidance states that all systems in one range are equipped with lamp-out monitors, which indicate lamp status. An ultraviolet intensity monitor is a separate feature and appears on specific models within the same families.

A lamp-out monitor detects the lamp failing to strike, which is the failure that announces itself: the light is off and the display says so. It does not detect the lamp aging, the sleeve fouling, or the transmittance dropping, which are the three ways the dose actually falls short in service.

An intensity monitor measures the light that reaches the far side of the water, which is the quantity that matters. That is the sensor an NSF 55 Class A system is required to have, alongside a visual and audible alarm and warning devices or an automatic shutoff at a fail-safe set point below the certified dose. Regulatory guidance commonly requires or recommends the shutoff, and local approval governs.

The reasoning behind the shutoff is worth stating: on an unattended residential system an alarm nobody hears is not a control. Someone drawing a glass of water at two in the morning has no way to know the dose has fallen, and the only mechanism that protects them is the one that stops the water.

The quartz sleeve belongs in the same maintenance discussion. It must be kept free of hardness and iron deposits, and where the hardness sits near its limit or the iron is above its own, the cleaning interval shortens accordingly. A fouled sleeve produces exactly the same invisible failure as an aged lamp.

UV Removes Nothing and Leaves No Residual

Two properties define where a UV system belongs in a treatment train, and both are absences.

It removes nothing. No dissolved metals, no nitrates, no hardness, no chemicals of any kind. Water leaving the reactor has exactly the composition it entered with, minus the viability of the organisms in it. Every figure on the water quality analysis is unchanged, and everything that needed removing still needs removing.

It also adds nothing, which is the corresponding advantage: no chemicals go in, and it generally creates no disinfection by-products.

And it leaves no residual. Treated water carries no disinfectant onward, so a UV system protects the water passing through the chamber at that moment and nothing beyond it. Published guidance states this directly: treated water is susceptible to recontamination if the distribution lines are compromised.

A chlorinated supply carries its protection along the pipe. A UV treated supply does not, and that difference decides two practical things.

Where plumbing downstream of the unit has been contaminated, running the supply through UV does not disinfect it. Shock chlorination of the distribution system is a separate operation on the pipe itself.

On a municipal supply, the incoming water may already carry a residual maintained for the distribution system. A UV unit at the point of entry does not replace that residual, extend it, or remove it. It adds a barrier at one location and changes nothing about what the pipe downstream relies on.

The placement that follows from both properties is the same one every manufacturer specifies. UV goes after the sediment filtration, the iron and manganese removal, the softening where hardness requires it, and the carbon or tannin treatment where the transmittance requires it, so that the water arriving at the reactor is already inside the limits. It goes before distribution, at or near the point of entry, with isolation and bypass arranged per the manufacturer.

What Is UV Water Disinfection

Ultraviolet disinfection passes water through a chamber containing a lamp that emits at 254 nanometres, a wavelength that damages the genetic material of bacteria, viruses and protozoa so they cannot reproduce. It does not remove them; it makes them unable to cause infection.

The measure of the treatment is the dose, and it is the intensity of the light multiplied by the time the water spends in it. Both terms are controlled by things outside the lamp: the flow rate sets the time, and the clarity of the water sets how much of the intensity actually reaches the organisms.

NSF/ANSI 55 is the certification framework, with Class A for microbiologically unsafe water and Class B for supplemental treatment of water already considered acceptable.

What makes UV different from every other treatment on a water system is that it sits at the end of a treatment train rather than at the start of one. The water arriving at it has to be clean enough for the light to get through, which means the equipment that makes UV possible is usually equipment installed before it.