Ultraviolet disinfection
UV dose chart: Class A, Class B and how dose falls as flow rises
Quick answer
NSF/ANSI 55 Class A requires 40 mJ/cm2 for water not known to be safe, while Class B delivers only 16, supplemental use only. Delivered dose falls as flow rises, so a unit rated for 40 mJ/cm2 at its tested flow can drop below that figure at a higher household flow rate, and it fails silently: there is no taste, smell or warning when it does.
Ultraviolet disinfection is one of the cleaner standards in this entire category because NSF/ANSI 55 defines exactly two classes by delivered dose, and the numbers do not move. What does move, constantly, in real installations, is the delivered dose itself, because dose depends on flow rate, water clarity and lamp age all at once, and a unit that meets its rated dose in a lab test can fall short of it in a real house running more flow than the rating point.
UV disinfection also does something categorically different from filtration, and confusing the two is a real safety gap: UV inactivates organisms so they cannot reproduce, but it removes nothing at all from the water. Everything that was in the water before the lamp is still in the water after it, just no longer able to multiply. That makes UV a poor substitute for finding out why a well is contaminated in the first place, and it is never a treatment decided without a certified laboratory test first.
Because UV addresses bacteria and coliform directly, and a positive coliform result is never a do it yourself situation, sizing a UV system correctly starts with a certified laboratory test, not with a product listing. A Class B unit installed on a well that has tested positive for coliform is a serious and common mistake, since the two classes look nearly identical on a retail page.
NSF/ANSI 55 Class A versus Class B
Class A delivers 40 mJ/cm2 for water not known to be safe; Class B delivers only 16 mJ/cm2 and is supplemental treatment only.
Published figure A published standard, regulatory limit, unit conversion or manufacturer specification. It does not change because somebody disagrees with it.
| Class | Delivered dose | Intended use |
|---|---|---|
| NSF/ANSI 55 Class A | 40 mJ/cm2 | Disinfection of water that is not known to be safe |
| NSF/ANSI 55 Class B | 16 mJ/cm2 | Supplemental treatment of water already considered safe |
A Class B unit is not a treatment for a well with a positive coliform result. The two classes frequently look nearly identical on a retail listing, and the difference only shows up in the certification documentation.
The prerequisites a UV system needs to actually deliver its rated dose
Ultraviolet light has to physically reach the organism to inactivate it, so the water feeding a UV system has to meet several published conditions before the rated dose means anything at all.
Turbidity has to stay below 1 NTU and ultraviolet transmittance at 75 percent or higher for a UV system to deliver its rated dose at all.
Published figure A published standard, regulatory limit, unit conversion or manufacturer specification. It does not change because somebody disagrees with it.
| Condition | Required limit |
|---|---|
| Ultraviolet transmittance | 75% or higher |
| Turbidity | Below 1 NTU |
| Iron | Below 0.3 mg/L |
| Manganese | Below 0.05 mg/L |
| Hardness | Below 7 gpg, or the sleeve scales |
| Tannins | Below 0.1 mg/L |
These are published manufacturer prerequisites essentially universal across the category. A UV system installed outside any one of them does not deliver its rated dose regardless of what the lamp itself is doing, which is why pretreatment, sediment filtration, iron removal, or a water softener ahead of a UV stage, is standard practice rather than optional on well water.
Dose falls as flow rises: a worked example
Delivered dose is inversely proportional to flow, because the faster water moves through the chamber, the less time each organism spends under the lamp. The table below illustrates that relationship for a unit rated 40 mJ/cm2 at its tested flow of 10 GPM; a specific manufacturer's own curve will differ in its exact numbers but not in its direction.
The same lamp delivers only 20 mJ/cm2, half its rated dose, once flow doubles to 20 GPM.
Published figure A published standard, regulatory limit, unit conversion or manufacturer specification. It does not change because somebody disagrees with it.
| Actual flow | Delivered dose |
|---|---|
| 5 GPM | 80.0 mJ/cm2 |
| 7 GPM | 57.1 mJ/cm2 |
| 10 GPM (rated point) | 40.0 mJ/cm2 |
| 12 GPM | 33.3 mJ/cm2 |
| 15 GPM | 26.7 mJ/cm2 |
| 20 GPM | 20.0 mJ/cm2 |
This table applies the published inverse relationship between dose and flow to one illustrative rated point; it is not a specific manufacturer's tested curve. Use the manufacturer's own flow rating, and the household's actual peak flow from the whole house flow rate calculator, to check whether a specific unit stays above its certified Class A or Class B threshold at real household demand.
UV removes nothing, and undersized systems fail silently
It is worth repeating precisely because it gets confused so often: ultraviolet treatment inactivates organisms so they cannot reproduce, and it removes absolutely nothing from the water. Anything present before the lamp chamber, dissolved minerals, chemical contaminants, sediment, remains present after it at the same concentration. UV is not a filtration stage and was never meant to be one; it is exclusively a microbiological control measure, and it works alongside pretreatment rather than replacing it.
The more dangerous property of an undersized or degraded UV system is that it fails silently. There is no taste change, no smell, no visible sign that delivered dose has fallen below a safe threshold, whether from oversized household flow, a dirty quartz sleeve, low ultraviolet transmittance, or a lamp reaching the end of its rated life. A household can run an underperforming UV system for months with no feedback at all, which is exactly why annual lamp replacement on a fixed schedule, rather than waiting for a symptom that will never appear, is the standard published maintenance interval, alongside periodic sleeve cleaning and retesting the treated water with a certified laboratory test to confirm the system is actually working.
Frequently asked questions
What is the difference between NSF/ANSI 55 Class A and Class B UV systems?
Class A delivers a minimum of 40 mJ/cm2 and is intended for water that is not known to be safe. Class B delivers only 16 mJ/cm2 and is labeled supplemental treatment for water already considered safe. Installing Class B on a well with a confirmed bacteria or coliform result is a serious mistake, since the two classes can look nearly identical on a retail listing.
Does UV treatment remove contaminants from water?
No. Ultraviolet light inactivates microorganisms so they cannot reproduce, but it removes nothing at all from the water. Anything present before the UV chamber, sediment, dissolved minerals or chemical contaminants, remains present afterward at the same concentration. UV is a microbiological control measure only and works alongside filtration rather than replacing it.
Why does UV dose fall as flow rate increases?
Delivered dose depends on how long water spends exposed to the lamp as it passes through the chamber. Higher flow means less exposure time for the same lamp output, so dose falls in direct proportion as flow rises. A system rated for 40 mJ/cm2 at a specific tested flow delivers proportionally less once household flow exceeds that rated point.
How do I know if my UV system is undersized?
There is often no symptom at all: an undersized or underperforming UV system fails silently, with no change in taste, smell or appearance. Check the manufacturer's rated flow against your household's actual peak flow, keep to the annual lamp replacement schedule regardless of apparent performance, and confirm the treated water periodically with a certified laboratory test rather than relying on the absence of symptoms.
What water conditions does a UV system need to work properly?
Ultraviolet transmittance at 75 percent or higher, turbidity below 1 NTU, iron below 0.3 mg/L, manganese below 0.05 mg/L, hardness below 7 grains per gallon, and tannins below 0.1 mg/L. A UV system installed outside any of these published limits will not deliver its rated dose regardless of lamp condition, which is why pretreatment ahead of a UV stage is standard practice on well water.
How often does a UV lamp need to be replaced?
Annually, on a fixed calendar schedule, regardless of whether the system shows any apparent problem. UV lamp output degrades gradually over its service life even while the lamp still glows visibly, and because an underperforming system fails silently with no taste or smell warning, waiting for a symptom before replacing the lamp defeats the entire purpose of scheduled maintenance.
Get the complete water treatment shopping list
Every item in the starter, complete and premium systems for both city water and well water, with running totals, in one printable list. It opens with the test kit, because that is genuinely the first purchase. No charge.
One list, no sequence of daily emails, and you can leave at any time.
Researched, not professional advice. This page is compiled from published standards, regulatory limits, manufacturer specifications and owner-review consensus, not hands-on testing. Nothing here diagnoses a water problem or says any product makes water safe to drink. Figures described as a rule of thumb are water treatment trade convention rather than published standards, and they are labeled that way wherever they appear. Test your water before you buy equipment: on city water start with the Consumer Confidence Report your utility publishes each year, and on a private well only a certified laboratory test tells you what you have. Bacteria, nitrate, lead, arsenic and PFAS are not do it yourself topics, they are invisible and tasteless, and they need a laboratory and usually a licensed professional rather than a product. Remember that a certification covers a specific claim under a specific standard, so "NSF certified" with no number attached tells you very little.