Considerations for determining the performance of ultraviolet light emitting diode fluid disinfection systems

The International Ultraviolet Association (IUVA) Task Force was formed to develop guidelines regarding testing and reporting on performance of UV LED water disinfection systems. The goal was to provide clarity in a guidance document on measuring system performance across the global UV LED water disinfection system market. A review of current performance measurement protocols for mercury lamp based systems shows that the common elements of UV LED system performance measurement protocols should be as follows: specified standard for the amount of pathogen reduction required, the requirement that the validation testing be conducted by a competent facility, and that the system be continually monitored by UV sensors while in use to verify system performance unless pathogen reduction is not claimed. UV LEDs have selectable peak wavelengths, as opposed to mercury lamps that have fixed emission wavelength values. As a result of this difference, the following changes to protocols used to test mercury lamp systems are recommended. First, the use of disinfection benchmarks other than 254 nm dose, such as direct inactivation values, dose benchmarks referenced to 254 nm, and/or dose benchmarks at the UV LED emission wavelength that give the same inactivation as the original 254 nm UV dose benchmark. Second, the use of 254 nm UV water transmittance values as a placeholder, rather than an assumed correct value, for systems under test with LED wavelengths >250 nm and water transmittance values ≥87%. More research is needed for lower wavelengths and UVTs. Third, the recommendation that germicidal response UV sensors be used in UV LED based systems to ensure that the validated disinfection is delivered. Finally, additional LED‐specific considerations were also noted. UV LEDs are also instant‐on devices, making them ideal light sources for systems operated intermittently. Performance testing of systems operated intermittently should include a test to insure that pathogens do not migrate past the UV LEDs while the LEDs are off. UV LED devices have recognized protocols for determining the lifetime of the devices, as well as for measuring other device properties. Caution should be exercised in using these lifetime values for devices in UV disinfection systems, since the thermal environment of the devices may be different for protocol testing and disinfection system operation.


INTRODUCTION
The International Ultraviolet Association (IUVA) Task Force was formed to develop guidelines regarding testing and reporting on performance of UV LED water disinfection systems so that claims could be better understood and benchmarked in rapidly developing markets.The goal of the Task Force was to provide clarity in a guidance document on measuring system performance across the global UV LED water disinfection system market.The UV LED Task force developed these guidelines to complement existing regulatory protocols with input from stakeholders in academia and other researchers, UV LED water treatment equipment manufacturers including those with research and development efforts, UV LED device manufacturers, and validation and regulatory experts across the world.The Task Force started meeting in February 2020 and submitted the guidelines to IUVA for review by the board of directors in January 2023.Various international stakeholders and organizations also reviewed and provided input on the submitted document, followed by IUVA board approval.Stakeholder consensus on appropriate scope of the guidelines was that the document would be a guideline or protocol with the considerations for best practices in the measurement and reporting on of system performance, but not a regulation or approval method for disinfection treatment systems.The document is global in scope and based on disinfection science but is not overly prescriptive such that decisions on system design are left to individual geographical regions and markets.The document references other existing guidelines, protocols, and regulations in the UV disinfection field.The guidelines in the document are meant for disinfection systems rather than for UV LED light sources themselves.However, relevant properties of UV LEDs that affect system performance are discussed in addition to system validation, system monitoring, system cleaning, and other factors.

Why validation is necessary: experience with hg lamp technology
The purpose of fluid disinfection systems using ultraviolet (UV) radiation is the reduction of harmful microorganisms that are a risk to human health.Validation of these systems is used to verify the amount of reduction that occurs for a system under test and that this reduction amount meets or exceeds a given standard.Validation for any system claiming to reduce the amount of pathogenic organisms is important to give confidence in the system to protect health.UV water treatment systems have been in use for several decades; primarily using mercury lamp technology with low pressure mercury lamps mainly emitting at 253.7 nm, or with medium pressure mercury lamps which have many emission lines in the germicidal UVC region of the UV spectrum and also emit many other wavelengths.The regulatory framework used to verify operation of these systems is well developed for both small point of use (POU) and point of entry (POE) systems operated directly by users of water, or larger municipal systems operated by an authority such as a government.This framework has some common features that should be present for the testing of any treatment system claiming to reduce pathogenic organisms: a specified standard for the amount of pathogen reduction required, the requirement that the validation testing be conducted by a competent facility, and that the system be continually monitored by UV sensors while in use to verify system performance.In this case, the definition of a competent facility is one accredited for this purpose in accordance with ISO/IEC 17025 or by a specialized third party.Furthermore, the facility needs to be recognized by regulatory authorities where required.The basic features of validation of systems using mercury lamp technology are described below for point of use/point of entry systems and for municipal systems.
For point of entry and point of use systems For these systems, the basic requirements were originally set out in the US Environmental Protection Agency (EPA) Guide Standard and Protocol for Testing Microbiological Water Purifiers (1987), which specifies required reductions of certain bacteria (6 log or 99.9999% reduction of the initial concentration), viruses (4 log or 99.99% reduction), and cysts (3 log or 99.9% reduction).The procedure for testing and specifications for the test water were described.Since it was not economically feasible to mount a routine testing program using all of the target microorganisms, for example, bacteria, viruses, and protozoan cysts, an equivalent "disinfection" set of tests and requirements is represented in the NSF/ANSI Standard 55 (2019) first published in 1991.It was determined that the log reductions specified could be achieved in a low pressure mercury lamp UV disinfection system test by achieving a dose of 40 mJ/cm 2 .The dose for a disinfection system is determined when disinfection (typically expressed as log reduction) in the test organism (MS2 coliphage) passing through the system under test is the same as the log reduction in MS2 exposed to a low pressure mercury lamp collimated beam delivering a dose of 40 mJ/cm 2 .The collimated beam dose is determined following standardized procedures (e.g., Bolton &Linden, 2003, andthe updated Bolton et al., 2015).This equivalent dose that results in the same MS2 log reduction delivered by the system is called the reduction equivalent dose (RED), also known as the reduction equivalent fluence (REF).For Class A systems certified to treat pathogens, UV sensor monitoring is required with a sensor alarm below the required RED of 40 mJ/cm 2 when the system is operated at its rated flow.This type of validation using sensor alarms is called UV sensor setpoint validation and is the most common type for small systems.

For municipal systems
UV technologies are used to provide municipal water disinfection for potable and non-potable reuse applications.
With many of those applications, utilities and regulators select UV reactors that use the delivered UV dose as a means of determining system performance that have been developed and demonstrated through UV validation testing.The established validation protocols include the EPA's UV Disinfection Guidance Manual (UVDGM) (USEPA, 2006), the German and Austrian UV Rules (DVGW W294-2, 2006;ÖNORM M 5873-1, 2020;ÖNORM M 5873-2, 2003), and the protocols developed by the Japan Water Research Center (JWRC, 2018).The NWRI UV Guidelines (NWRI and WRF, 2012) are used for non-potable reuse applications.
There are two approaches for UV dose monitoring and validation of municipal systems: the UV sensor setpoint approach as described above and the calculated dose approach.The UV sensor setpoint approach is specified by the German and Austrian rules.With the calculated dose approach (used by EPA UVDGM), validation test data is used to develop an equation that predicts the RED delivered by the reactor as a function of the flow through the reactor, the UV transmittance (UVT) of the water being treated, and the UV intensity measured by UV sensors.The initial dose-response relationship for the challenge microbe used during validation is determined with the use of a low pressure mercury lamp collimated beam.Validated dose for a system can then be determined by adjusting the RED considering uncertainties in measurement, differences between sensitivity of target pathogen organisms and the challenge organisms and spectral considerations for the use of medium pressure mercury lamps that have a polychromatic UV output.Validation methods can be prescriptive, such as specifying the dimensions of sensor ports used by UV sensors in the German and Austrian rules or non-prescriptive for system design like the UVDGM.All the established protocols have at least the three common elements of set disinfection standards, testing at a competent facility, and system monitoring.There is also an upcoming standard for the use of UV LEDs in public drinking water disinfection from DVGW.A brief summary of the current UV water treatment validation protocols as summarized by the Figawa organization, and for JWRC, is shown in Table 1.

VALIDATION CONSIDERATIONS FOR UV LED SYSTEMS
The properties of UV LEDs in comparison to those of mercury lamps will be considered, and then the implications of these device properties for validation of disinfection systems using UV LEDs will be discussed.

UVC light emitting diodes as disinfection light sources
Emission spectrum properties UVC light emitting diodes (LEDs) are semiconductor devices capable of emitting germicidal UV radiation at a selectable wavelength depending on the composition of the semiconductor materials in a particular device.The emission wavelengths are selectable from 254 nm or less, to >300 nm where germicidal action on microorganisms is significantly reduced (Bolton, 2017).Higher power LEDs are currently in the wavelength range of $265-290 nm and are being continuously developed.Unlike UV LEDs, the emission wavelengths for mercury lamps are fixed and determined by the radiative emission lines of the mercury atom and vapor pressure in the lamp envelope.For germicidal low pressure mercury lamps, mainly the single 253.7 nm transition is emitted, while many additional transition lines are excited for medium pressure mercury lamps.Small differences in the manufacturing process will cause UV emission intensity differences from the rated output at a fixed electrical input for both UV LEDs and mercury lamps, but for UV LEDs, there are also UV wavelength differences from the rated wavelength that are also noted.Wavelength shifts can have an important effect on disinfection system performance, and this will be discussed in a later portion of this report.Manufacturers of UV LEDs frequently state this range as an uncertainty in the rated wavelength.The Abbreviations: EoL, end of life; LP, low pressure mercury lamps; MP, medium pressure mercury lamps.
emission from LEDs typically has a spectral width at half-maximum intensity of $10-15 nm, so they are not strictly monochromatic UV sources like low pressure mercury lamps.However, they are also not broadly polychromatic as medium pressure mercury lamps which have emission lines throughout the 200-300 nm region of the spectrum and beyond.In a recent study, no significant differences were observed between monochromatic peak wavelength analysis and polychromatic analysis for spectroradiometer and ferrioxalate actinometer measurements (Pousty et al., 2022).However, testing using other criteria may show more polychromatic behavior, which would increase with the full width at half-maximum (FWHM) of the LED source tested, spectral response of the radiometer and actinometer, and could have varying impacts based on the action spectrum of microorganisms.
Microbiocidal action and responsivity of germicidal response sensors tends to drastically decreases at wavelengths above $280 nm.The contribution to microbiocidal action or sensor signal will therefore be significantly reduced at wavelengths >$280 nm, and this impact will increase with both the LED FWHM and any shifts in peak wavelength toward longer wavelengths.

Considerations and measurements for output over lifetime
UV LEDs are capable of reaching full UV intensity within a microsecond of power application, unlike mercury lamps that require time on the order of minutes to reach full UV intensity.LEDs can also be cycled on and off rapidly and for prolonged periods of time without affecting the lifetime of the device.This is opposed to the situation with mercury lamps where on/off cycling will cause wear to filaments or electrodes and shorten the life of the lamp.One factor that is common to both UV LEDs and mercury lamps is a decrease in UV output with age.Mercury lamps are frequently assigned a lifetime value, which represents the amount of time a lamp operates until the UV output intensity reaches a specified percentage of its initial output.UV LEDs are rated in a similar manner with an L xx value, where L is the lifetime when the device output decreases to XX% of its original output.A common rated percentage of the initial output for UV LEDs is 70%, so an L 70 value would be specified in this case.For LEDs, the lifetime is significantly affected by the device temperature, where higher temperatures cause faster degradation of the device and correspondingly lower UV output and shorter life.Thermal management of the device is therefore important, where increased drive current to the device and poor thermal contact to conduct heat away from the device will both cause the device lifetime to decrease.For this reason, L xx values should be specified at a given drive current and case temperature.Heat sinking technology used in the final system should therefore be considered to minimize the case temperature which affects lifetime.
Current protocols exist for the measurement of the device lifetime, such as the Illuminating Engineering Society LM-80 (2021a), although terms such as luminous flux and luminous flux maintenance apply to visible LEDs only.This protocol specifies that the lifetime of a LED must be measured at two device case temperatures or more, at a specified electrical drive condition.In this way, different thermal conditions of the LEDs under test are used at the same drive level.The Illuminating Engineering Society TM-21 (2021b) protocol provides guidance on how to use the LM-80 data to extrapolate to L XX values for visible LED's.While initial studies show these formulas in the TM 21 protocol may also apply to UV LED radiant flux maintenance, there still remains work to be done to find a consensus on its suitability for all UV LEDs.The Illuminating Engineering Society LM-82 (2020) protocol then relates the case temperature with the device junction temperature through forward voltage measurements for visible LEDs.The new Illuminating Engineering Society LM-92 (2022) protocol extends the use of LM-82 for UV LED's by taking transients in forward voltage into account for pulsed operation.This information can then be used by UV LED disinfection system manufacturers to estimate the lifetime of devices in the system.
Caution should be exercised in applying these protocols directly for determining device lifetime in UV disinfection systems, as factors such as the use of multiple LEDs in an assembly and the system environment will affect the lifetime of the devices.For this reason, it is recommended that performance testing of actual UV disinfection systems be conducted to determine the lifetime of UV LED based assemblies within the system.However, these ANSI/IES and other accepted protocols are valuable for UV LED manufacturers to use to ensure that devices with the same performance are used in disinfection systems when it becomes necessary to replace aged devices.
The LM-80 protocol also specifies methods of measuring device peak and centroid wavelength, as well as radiant and photon flux.Radiant flux measurement of LEDs used in disinfection systems may not be needed for systems with UV sensors, since sensor alarms will notify the user that the UV flux in the system, combined with other system factors, is below the validated limit.However, use of a protocol such as LM-80 by LED manufacturers would be a good quality control measure to ensure that devices with the same performance are used.For systems without sensors, providing a uniform standard of LED device UV flux for systems will be more critical, since these systems do not have a direct monitoring system.
In summary, disinfection system performance should be the main determinant for device output and lifetime within a disinfection system.However, established protocols for device property measurement must be used to ensure that devices with consistent properties are made by device manufacturers and specified by UV disinfection system manufacturers.

Unique system design considerations
Since LEDs are effectively "instant-on" devices, they can be turned off when not needed for disinfection, which is advantageous for systems with intermittent water flow.Intermittent water flow may be desirable for certain POU/POE applications, and certain municipal applications such as intermittently operated ground-water wells and post-Sequence Batch Reactors.The duty cycle of the device, which is the percentage of the total time that the device is on, can be as low as 5%-10% for some disinfection applications.Since the device only ages when it is on, the effective lifetime of the device will be the rated lifetime divided by the duty cycle, which would be 10-20 times the rated lifetime of the device for the example of 5%-10% duty cycle.Simple timers that log the total time when the device is on can be used to estimate the useful lifetime of the LED.Also, more sophisticated systems can also analyze information from current-voltage characteristics, and device case temperature as discussed in this section, to give a more accurate estimate of the end of the useful life.
UV LEDs are very compact devices, with the size of an individual semiconductor chip being 1 mm 2 or less and individual device packages that house the chip being $0.25-1 cm 2 .Arrays of chips can be contained in a single device package, and arrays of packages can be mounted on an external circuit board to form an assembly of variable size and shape.Therefore, there is great flexibility for the design of the UVC light source in a LED-based disinfection system, as opposed to systems with mercury lamps where most of the UV lamps are cylindrical in construction.The UV output from LED packages can also be adjusted from wide-angle output where the irradiance varies approximately as the cosine of the output angle (a Lambertian source), to narrow angle collimated output using lenses or collimating mirrors.This allows an additional degree of flexibility in the disinfection system design.

Instant-on operation
Considering the "instant-on" property of UV LEDs, it is possible that the LEDs within the disinfection system are off when water does not flow in the system to preserve device life.Similarly, there are mercury lamp based systems that can be turned off for prolonged periods of time, but for these systems, the lamp warm-up time also needs to be considered when system operation is required.Under these conditions for any type of system, it will be necessary to include additional sampling in the system validation to test water that is produced directly after water flow resumes.The log reduction of challenge microbes can be determined to ensure that microbes do not bypass disinfection while the light sources are off.The inclusion of this type of test should also be conducted for systems that are specifically designed to disinfect water in batches, where a system is filled with fluid, the light sources are then turned on for a specified time to achieve the desired disinfection, and then the light sources are turned off before the fluid is drained from the system.

System geometry
For disinfection systems with varying geometries, these should not affect the validation of the disinfection systems, since the validation will determine the level of performance of these systems.Current validation protocols also routinely include testing scenarios such as performance with certain lamp banks on and off and/or with designated lamps on and off.This can easily be extended to testing of UV LED disinfection systems to determine performance with certain LEDs or groups of LEDs in operation during the validation test.

Validation benchmarks
The validation benchmark of a specified UV dose to achieve the desired inactivation of certain organisms or group of organisms needs to be carefully considered for validation of UV LED disinfection systems.Since inactivation will vary with UV wavelength at a given dose due to the action spectrum of the microbe, simply using an accepted dose benchmark such as 40 mJ/cm 2 is not acceptable for wavelengths other than 254 nm.For example, the dose benchmark to achieve the same inactivation as 40 mJ/cm 2 at 254 nm for a system with LEDs with a peak wavelength of >280 nm will likely be greater than 40 mJ/cm 2 because the microbiocidal action for most microbes is less for wavelengths above $280 nm.For wavelengths near the peak of the action spectrum at $265 nm, the dose benchmark would likely be less than 40 mJ/cm 2 .Since the ultimate goal of the validation is to provide a standard for inactivation of organisms, the following validation benchmarks are acceptable to use: Here, a collimated beam using UV LEDs at the same wavelength as those in the system would be used to determine the dose-response relationships during the validation process and set the target dose.Collimated beams using UV LEDs typically have short distances between the radiation source and exposed organisms, and have different UV radiation distribution patterns than standard mercury lamps.Therefore, modified protocols to determine the UV fluence delivered to organisms in a collimated beam apparatus will need to be employed.A modified protocol is described by Kheyrandish et al. (2018).The challenge with using this method will be the large variety of UV LED wavelength emission spectra available, necessitating buildup of a large enough database of dose-response curves to give confidence in inactivation performance at relevant LED emission wavelengths.3. A dose benchmark referenced to 254 nm.For this type of validation, the disinfection system operates at the UV LED wavelength, but the collimated beam used to determine the dose-response curves uses a 254 nm wavelength collimated beam.A standard low pressure mercury lamp collimated beam can be used, or any other source with an emission wavelength of 254 ± 1 nm.In this way, standard 254 nm dose benchmarks such as 40 mJ/cm 2 can be used, and the large body of disinfection data at 254 nm can be referenced.This technique has already been used for validation of medium pressure mercury lamp systems using the municipal EPA UVDGM protocols, and most recently for UVDGM validation of a municipal UV LED disinfection system (Brooks et al., 2021).

System sensing
Disinfection systems use UV photo-optical sensors to monitor performance during operation, and are required for recognized pathogen reduction by current regulatory authorities.Since UV LEDs can be manufactured in a variety of wavelengths as well as intensity, it is important that UV sensors be able to detect changes in both of these properties in order to maintain system performance.For this reason, it is important that the sensors used in UV LED disinfection systems have germicidal wavelength response.A germicidal wavelength sensor response detects different wavelengths with different sensitivities similar to a microbial action spectrum (which resembles the spectral absorbance of DNA).As an example for sensor operation, if a system uses a 280.0 nm LED, and the replacement LED had a wavelength of 285.0 nm at the same peak wavelength UV flux, the system performance would be degraded since the action spectrum of most microbes decrease either side of a 265 nm peak.Without knowing the peak wavelength of the initial or replacement LED and therefore making any adjustments for sensor spectral response, a germicidal wavelength response sensor would detect less irradiance and give an alarm if the dose delivery drops below the acceptable threshold, since the sensor responds less to the longer wavelength radiation of the replacement LED.Validation of these systems should require germicidal wavelength response sensors to address variability in LED characteristics.An alternative would be the requirement that any replacement LEDs have the same spectral characteristics (wavelength, FWHM, and UV output) as the original devices, and that the wavelength and FWHM of the original and replacement LEDs remain stable with age within an acceptable tolerance limit.Since UV LEDs are small compact radiation sources, disinfection systems using UV LEDs could contain many LEDs.It will be impractical to have a UV sensor for every LED source, and even for systems containing mercury lamps, there are frequently fewer UV sensors than lamps.Requirements of current validation protocols vary for specification of the number of sensors required.Some validation protocols such as the EPA UVDGM require one sensor for every distinct grouping or "bank" of lamps, which can be individually controlled in a UV disinfection system, and this would be a recommended practice for systems containing UV LEDs where they are arranged in a bank or grouping.It will also be advantageous to use sensors that can view multiple LEDs in an individual bank to detect overall UV output changes in the bank.Since there can be a great number of individual LEDs within a disinfection system, it is possible that one or more of the individual LED devices fail during operation of the system.UV optical sensors that monitor multiple LEDs should be able to detect the overall output change with the failed LED and give an alarm if the sensor signal drops below a validated level.In addition, the delivered current could be monitored in a precise manner such that LED device failures are detected and used in combination with optical UV monitoring.In case of an LED device failure, the validation protocol should test the system for disinfection performance where failure of individual devices resulting in "dark spots" is simulated during testing to ensure that minimum disinfection performance is maintained.Criteria for the maximum number of allowed failed devices should be set during validation testing.
It is an economic reality that there will be a significant number of UVC disinfection systems that will not have UV sensors, since there is a large low-cost market for POE and POU systems.These will not have the same recognized pathogen reduction for protocols such as the NSF/ANSI Class A systems, but would still be considered for supplemental disinfection in Class B operation where achievement of certain dose benchmarks is required for specified system operating conditions.Protocols other than the NSF/ANSI Standard 55 could be developed for validation of systems without UV sensors.It is recommended here that a quality management system also be employed by UV LED device manufacturers and disinfection system manufacturers so that devices of consistent quality are used to maintain system disinfection performance as determined in the validation test, in the absence of available UV sensing.

UV water transmittance
The UV transmittance of water for a system undergoing validation testing needs to be determined in order to establish the validated range of UVT values for a system operating at users' sites.The UV transmittance of natural waters, and for waters with specific UVT modifiers added during validation, will have water transmittance that varies with wavelength.Typically, UVT increases with increasing wavelength.Since UV LEDs can have selectable emission wavelengths, the UVT of the same water will change depending on the wavelength of the LEDs for a particular system.The standard wavelength for monitoring UV transmittance (UVT) is at 254 nm, which coincides with the emission wavelength of monochromatic low pressure mercury lamps.This will measure the actual UV transmittance of water being treated by LP lamp disinfection systems.For polychromatic medium pressure lamp systems, the UVT used for measurement and validation is still the 254 nm UVT, although monitoring both low and high wavelengths has been recommended (Wright et al., 2020).It is common practice and currently most economically feasible to measure UVT at 254 nm only, which introduces possible uncertainty in the validation results since UVT at the LED wavelengths will differ from UVT at 254 nm.For UV LED disinfection systems, the situation is different than medium pressure mercury lamp systems, since LED emission wavelengths are currently normally greater than 254 nm, with common popular choices of $265 nm and $280 nm and a spectral full width half maximum of $10-15 nm.As a result, UVT at the LED emission wavelengths will likely be greater than that of an LP lamp.However, it would be possible to use the 254 nm UVT value as a placeholder for the UVT at the LED wavelength as long as there is a consistent relationship between these UVT values.A study conducted at 156 water treatment sites with UVT at 254 nm ranging from 87% to 98% for 10 mm water layer (Figure 1) shows that this is indeed the case for LED wavelengths >250 nm, but not for ≤250 nm, further supporting the recommendation of both high and low wavelength UVT monitoring for wavelengths <250 nm.The absorbance spectra of natural waters are all in close agreement from the 5th to 95th percentile for wavelengths >250 nm, but have much more variation for the shorter wavelengths.The figure also shows that the natural waters will have a lower absorbance (higher UVT) at the LED wavelength when compared to the spectra of the commonly used UVT modifiers Superhume and Superhume + vanillin.Validation results using these UVT modifiers will therefore be more conservative, since the system under test will be operating under more stringent conditions.It should be noted that the absorbance of other UVT modifiers should be higher than that of the waters being treated to verify that the validation results are more conservative.An example for sample UV absorbance spectra normalized to 91.2% UVT at 254 nm, is shown below.
UV transmittance of Superhume at various concentrations (Figure 2) also demonstrates a consistent trend of higher transmittance with longer wavelength.However, there is a growing percent difference between UVT at 254 nm versus UVT at higher wavelengths, and this effect increases as the 254 nm UVT decreases, as quantified from Figure 2 in Table 2.
Since UV-C LED devices are quasi-monochromatic, the interaction between spectral UVT variations, spectral sensor response, and spectral dose delivery (based on the action spectrum of the treatment target) may follow inconsistent trends across wavelengths.This may cause validation testing to be either more or less conservative depending on the spectral interactions if changes in UVT are not consistent between the absorber and the target action spectrum or between the spectral response of the sensor.Further, sensor readings will vary depending on placement of the sensor.In general, UV-C LED systems should be treated in a more similar manner to polychromatic UV sources (Medium Pressure Lamps) than monochromatic UV sources (Low Pressure Lamps) when considering these factors.More studies are needed to determine these spectral impacts to adequately characterize that extent that spectral changes have for validation and monitoring.
Although monitoring UVT at LED wavelengths would be most accurate, it appears feasible to use 254 nm UVT values in validations of UV LED water treatment systems for wavelengths >250 nm and high UVT, but the UVT for systems with LED emission wavelengths ≤250 nm will need to be measured at the LED wavelength, and at low UVT may need to be measured at the LED wavelength of the system.A continuation of the spectral study of natural waters and other relevant liquid matrices for UVT values <87% is recommended to extend the range of the 254 nm UVT placeholder.
F I G U R E 1 UV Absorbance spectra for natural waters at 156 water treatment sites ranging from the 5th to 95th percentile, along with the spectra for Superhume and Superhume + vanillin (data courtesy of H. Wright, Carollo Engineers, adapted from Wright & Linden, 2018).Data normalized to 91.2% UVT at 254 nm.

System fouling
During operation, UV disinfection systems will accumulate material on lamp sleeves, which will decrease the UV flux entering the water for treatment.This accumulation will vary from site to site, depending on the presence and concentration of fouling agents such as hardness, manganese and iron, and biological activity.It is common practice to specify the maximum allowable decrease in UV intensity for a UV system in a validation caused by lamp aging and fouling, determined by multiplying a fouling factor times an aging factor.The system is tested at this reduced UV output, and if the minimum disinfection is achieved during validation, then the system is allowed to operate at this UV output value at a user's site.Sensors will detect decreases in UV output due to fouling as well as lower lamp output and lower UVT and will alarm if the sensor signal drops below the validated level.This will be true for both UV LED and mercury lamp based water treatment systems.Water treatment systems using UV LEDs should be validated in a similar manner to systems using mercury lamps, as fouling will decrease the UV flux of LED sources as well.Although there is no intrinsic difference in validation of LED systems when considering fouling, LED systems may be able to extend the time between system cleaning intervals to remove fouling if the heat produced by the LEDs can be conducted away from the water being treated.This is possible since the rate of fouling frequently increases with water temperature due to the decreased solubility of some fouling agents at these higher temperatures.
For systems without sensors, which are noted for some small system applications, there will be no warning if the UV intensity drops below the minimum allowable level established during validation due to fouling.Unlike UV lamp aging which can be characterized and regulated using a quality control program by the UV LED manufacturer, fouling is highly dependent on water quality at the user's site.It is therefore important for the user to know the rate of fouling at the user's site, and adhere to a cleaning program where required at the specified interval in order to maintain a degree of disinfection.

SUMMARY AND CONCLUSIONS
The International Ultraviolet Association (IUVA) Task Force has produced the above considerations on measuring water disinfection system performance across the global UV LED disinfection system market.A review of current performance measurement protocols for mercury lamp based systems shows that the common elements of these protocols should be: a specified standard for the amount of pathogen reduction required, the requirement that the validation testing be conducted by a competent facility, and that the system be continually monitored by UV sensors while in use to verify system performance.These same elements should exist in performance testing of systems using UV LEDs.The final element can be relaxed if the system under test does not claim to reduce pathogens, but instead is used as a disinfection aid.
UV LEDs have selectable peak wavelengths, as opposed to mercury lamps that have fixed emission wavelength values.As a result of this difference, the 254 nm UV dose benchmarks developed for disinfection systems that use mercury lamps cannot be used at the UV LED wavelength, since the degree of inactivation of pathogens is wavelength dependent.Suitable benchmarks for UV LED systems are direct inactivation values, dose benchmarks referenced to 254 nm, and dose benchmarks at the UV LED emission wavelength that give the same inactivation as the original 254 nm UV dose benchmark.Another result of the selectable wavelength for UV LEDs is the difference between the UV transmittance of water determined at the UV LED wavelength, and the 254 nm transmittance measured by commercially available systems.This work shows that the 254 nm transmittance value can be used as a placeholder for the transmittance at the UV LED wavelength for wavelengths >250 nm and UV transmittance values ≥87%, but further work is required to verify if the transmittance range can be extended below 87% and for other wavelengths.The final result is the selectable wavelengths that will be detected by sensors for disinfection systems with UV LEDs.In order to ensure that the validated disinfection is delivered by the system, the sensor should have germicidal wavelength response.If the sensor does not have germicidal wavelength response, then it must be verified that any UV LEDs used in a validated system have the same peak wavelength and spectral width as the LED device(s) used during validation.Δ UVT (%, from 254 nm) 0.0 1.6 3.7 0.0 0.9 1.9 0.0 0.5 0.9 UV LED devices have recognized protocols for determining the lifetime of the devices, as well as for measuring other device properties.Caution should be exercised in using these lifetime values for devices in UV disinfection systems, since the thermal environment of the devices may be different for protocol testing and disinfection system operation.However, these device protocols are an excellent method of ensuring that devices of consistent quality are used in validated systems.UV LEDs are also instant-on devices, which makes them ideal for disinfection systems that can benefit from intermittent operation.Performance testing of these systems should include a check to insure that pathogens do not bypass disinfection while the UV LEDs are off.Considering system fouling and the geometry of UV LED disinfection systems, there were no significant changes recommended for system performance testing, when compared to testing for mercury lamp based systems.
Brief summary of current UV disinfection protocols as prepared for Figawa Europe by Aquisense Technologies and as provided by JWRC.Different font colors in columns denote large-scale (black, left) and small-scale (blue, right) applications.