By Rick Andrew

UV systems are often used for protection against potential microbiologically contaminated drinking water. These systems use UV energy to disrupt the DNA of microorganisms, which prevents them from reproducing. UV systems can, however, become fouled with scale, or their performance can be impacted by turbid water, or eventually the lamps can fail, or there could be other system issues that impact performance. Because of these performance issues and factors and because of the risk of infectious diseases (if water is contaminated and system performance is negatively impacted), it is important for users to understand when their UV systems are not operating effectively.

NSF/ANSI 55 Ultraviolet Microbiological Water Treatment Systems addresses this situation by requiring that Class A UV systems must have a UV sensor and alarm. This alarm must be a visual indicator, an audible alarm, a termination of the discharge of treated water, or some combination of these options. The standard is quite clear and explicit regarding requirements for an alarm. The standard is less explicit, however, regarding requirements for the UV sensor that triggers the alarm. In fact, there is no explicit specification for a UV sensor at all in NSF/ANSI 55. This omission leads to questions regarding requirements for UV sensors and what constitutes an acceptable sensor under the standard.

A performance-based requirement
In reality, the requirement for UV sensors under NSF/ANSI 55 is purely performance-based. This requirement is stated first in the test for UV alarm performance. This test specifies that the UV system alarm must activate 10 consecutive times in response to a decrease in UV intensity. Under this test, the UV intensity is reduced by flowing water dosed with sufficient parahydroxybenzoic acid (PHBA) to activate the alarm system. Once the alarm activates, clean water is flowed through the UV system until the alarm ceases. The test cycle of decreasing the UV intensity and then flowing clean water is then repeated nine times to assure that the alarm activates 10 consecutive times.

Looking deeper into the UV sensor requirement, the biodosimetry test for microbial performance is also conducted based on reduced UV intensity, as detected by the UV sensor. Similar to the UV alarm performance test, the UV intensity for the microbial performance test is decreased by addition of PHBA until the alarm activates. It is under these conditions that the UV system must deliver sufficient intensity to achieve a 40 mJ/cm2 UV dosage required for effective microbial performance.

What about visual light sensors?
The question regarding UV sensors often gets directed along the lines of whether a sensor acting in the visible wavelength range would be considered acceptable as a UV sensor. This type of sensor would be effective in a real-world setting for detection of scale formation inside the quartz tube, or turbid water caused by particulates, or other causes of reduction in the passage of visible light that would also be causes of reduction of UV intensity.

The answer to this question lies in the absorption spectrum of PHBA. PHBA absorbs UV radiation very specifically at 254 nm, very near to optimal germicidal UV wavelengths and right in the same range as the UV emission from low-pressure mercury lamps. PHBA is visually clear, however. It does not absorb much energy at all in the visible light spectrum. As such, it would require the addition of very large amounts of PHBA to cause a reduction of visible light intensity sufficient to activate an alarm connected to a visible light sensor.

With the addition of this much PHBA, the UV intensity would be very greatly reduced (much more so than the visible light intensity), because of PHBA’s strong absorbance at 254 nm. This reality would make it very unlikely that there would be sufficient UV intensity produced by the system to achieve a 40 mJ/cm2 UV dosage. Ultimately, performance based the nature of the test requirements under NSF/ANSI 55 leads to a necessity for a UV sensor that is specific to the 254-nm wavelength range that is relevant to the UV emission of low-pressure mercury lamps and that is also in the range of strong absorbance by the PHBA.

The importance of sensors
The requirement in NSF/ANSI 55 for UV sensors and alarms is highly prescriptive. In many cases, the NSF Joint Committee on Drinking Water Treatment Units has been successful in being less prescriptive regarding requirements, which allows for more options in system design and construction for conformance. In the case of UV systems, however, the highly prescriptive requirements for UV sensors and alarms is justified by the acute health effects associated with system failure and consumption of microbiologically contaminated drinking water.

If a UV system is not providing sufficient UV dosage and the water is contaminated, people can become very seriously ill from a number of infectious diseases in a short period of time. These are serious consequences to a UV system failure. As such, there is much value in providing the fail-safe mode of operation of a UV system with a UV sensor and alarm, which is worth the cost and limitations imposed by such prescriptive requirements.

Despite the fact that specifications for UV sensors are not explicitly stated in NSF/ANSI 55, it is clear from an examination of the test methods that a performance-based requirement for the sensors dictates that they must be focused on UV wavelengths as opposed to visual light sensors. A system with only a visual light sensor would likely not pass the microbial performance test once enough PHBA was added to the water to cause the visual light sensor to trigger the alarm.

About the author
Rick Andrew is NSF’s Director of Global Business Development–Water Systems. Previously, he served as General Manager of NSF’s Drinking Water Treatment Units (POU/POE), ERS (Protocols) and Biosafety Cabinetry Programs. Andrew has a Bachelor’s Degree in chemistry and an MBA from the University of Michigan. He can be reached at (800) NSF-MARK or email: [email protected]


Comments are closed.