UV Dose Explained: mJ/cm², UVT and Lamp Life

Microorganisms are inactivated when UV light damages their nucleic acids. The absorption peak of DNA and RNA lies near 260 nm, and low-pressure mercury lamps emit strongly at 254 nm, close enough to be effective. The energy breaks bonds in the genetic material and prevents replication; it does not necessarily kill the cell outright, which is why the correct term is inactivation rather than sterilization.

UV at 254 nm adds no chemical to the water and leaves no residual. That is an advantage for taste and for processes that cannot tolerate oxidants, but it also means protection ends at the reactor. Any contamination introduced downstream, in the storage tank or in a dead leg, is not neutralised by the UV that treated the water earlier. Medium-pressure lamps emit a broader spectrum and are used in some high-flow or reuse duties, but low-pressure and amalgam lamps dominate drinking-water applications.

UV transmittance is the percentage of 254 nm light that passes through a 1 cm water sample. Filtered drinking water commonly runs above 90% UVT. Dissolved organics, iron, manganese and colour absorb UV and lower the value; a water at 70% UVT receives roughly three-quarters of the irradiance that the same reactor delivers at 90%, so the dose falls unless flow is reduced or lamp output increased.

This is why a reactor should be quoted with the UVT used in the calculation, not only with a flow and a dose. A unit selected for 10 m³/h at 30 mJ/cm² and 90% UVT needs more lamps or a larger chamber to reach 40 mJ/cm², or to treat water at 80% UVT. Ask for the design UVT, the fouling factor and the end-of-life lamp factor, and compare quotations on the same basis.

Particles shield microorganisms. A bacterium attached to a clay particle or embedded in a floc may sit in the shadow of the particle and receive a fraction of the calculated dose, even though the water around it is clear enough for the sensor. Turbidity also scatters light and reduces penetration. For drinking-water UV, the usual minimum is below 1 NTU before the reactor, and many designers target below 0.5 NTU.

A 5 µm cartridge upstream is the standard protection. It removes particles that would shield microorganisms and foul the quartz sleeve, and it protects the reactor after maintenance. It is not a substitute for clarification: if turbidity stays above 1 NTU, the UV dose is unreliable regardless of the reactor rating. Fix the filtration before increasing the lamp count.

Low-pressure lamps lose output gradually and are usually rated for 8,000–9,000 hours. A lamp at 70% of initial output delivers 70% of the dose at the same flow, which can take a 40 mJ/cm² reactor below 30 mJ/cm² without any alarm if the sensor is not calibrated or the lamp hours are not tracked. Size the reactor at end-of-life output, not at the output of a new lamp.

As a rule of thumb, one 40 W low-pressure lamp treats about 1 m³/h at 30–40 mJ/cm² with UVT above 90%. Higher flow needs more lamps or a larger chamber; colder water, lower UVT and a higher target dose all push the lamp count up. Replace lamps on hours or on measured output, whichever comes first, and replace the quartz sleeve when cleaning no longer restores performance.

Different organisms respond differently. Bacteria are relatively sensitive, viruses generally need more, and protozoan cysts such as Cryptosporidium are inactivated at low UV doses but still require a validated reactor and the log credit specified by the authority. Adenovirus is the outlier among viruses and may need a much higher dose than the ranges below.

The table gives typical ranges for 3–4 log inactivation, not regulatory limits. The correct approach is to state the target organism, the required log credit and the validation protocol, then select a reactor with a certified dose-flow relationship at the design UVT. Where no validation is required, 30–40 mJ/cm² remains a sound general design target for drinking water.

A UV sensor measures lamp intensity at the sleeve, not the dose in the water. It is still the most useful continuous indicator: if intensity falls while flow and UVT are stable, the lamp or sleeve is the cause. Record intensity, flow, UVT and lamp hours together. A sensor reading that stays normal while the water quality changes is a warning that the instrument, not the reactor, is being monitored.

At commissioning, verify the reactor against its validation report at the design flow and UVT, confirm the sensor calibration and set the low-dose alarm. In service, clean the sleeve, check the sensor window and compare the reading with a reference meter. Where the local authority requires validation, use a reactor with a report covering the exact model, dose and flow; a general certificate for the product family is not the same thing.

Give the peak flow and the minimum flow, the design UVT, the maximum turbidity, the water temperature, the target dose and the required log credit. State whether the dose must be met at end of lamp life and what fouling factor to assume. Add the power supply, the installation orientation, the pipe size and the alarm or interlock requirements. A quotation without those inputs is a guess dressed as a specification.

Finally, confirm who validates the dose. The reactor supplier validates the hardware; the system integrator validates the installed train; the local authority accepts the finished water. Those three responsibilities are different, and a drinking-water system needs all three to be clear before the first lamp is energised.