In all water-treatment applications intended to mitigate microbial or chemical contaminants through the addition of chemicals, the concept of dose is paramount in determining treatment efficacy and process recommendations for target levels of contaminant treatment.

Dose is defined as “a portion of a substance added during a process,” according to the Merriam-Webster dictionary. It is embodied in the CT concept (representing the concentration of disinfectant, or “C,” multiplied by exposure time, or “T”) used by the U.S. Environmental Protection Agency (EPA) to provide adequate microbial inactivation to meet treatment technique requirements. Likewise, when using a physical treatment technique, such as UV irradiation, the dose is typically represented as an IT value (irradiance of the UV light multiplied by the exposure time). Notably, UV dose is formally described as fluence, a term that conveys the incident UV energy (i.e., energy supplied) rather than the absorbed UV energy (i.e., energy responsible for contaminant transformation).¹

While the “I” is sometimes described as intensity, irradiance is technically more appropriate given its definition as the total radiant power from all directions on a surface (W/m², or commonly mW/cm² in UV disinfection applications).¹ Unlike the “C” part of CT for chemical disinfection, which can be directly measured using a range of analytical techniques, determination of the “I” part of UV fluence is not a straightforward process and can be quite complicated in full-scale applications.²

Initially, quantifying the UV fluence needed to achieve a target level of contaminant mitigation is often done by bench-scale testing. For example, bench-scale tests can be used to determine the UV fluence required to achieve 4-log inactivation of adenovirus. James Bolton and Karl Linden (2003)¹ developed a landmark standard method for quantifying UV fluence in bench-scale experiments, including specifications for the testing apparatus and determination of irradiance. Prior to this development, the process of determining UV irradiance and the corresponding fluence was not always clear.

Example of a collimated beam used for bench-scale UV testing. Modified after Bolton and Linden developed a standard method for quantifying UV fluence in bench-scale experiments (2003).¹ The exact design may vary, but important components are noted here.

Using Bolton and Linden’s approach, a collimated beam is used to standardize the delivery of UV fluence to the water sample. (See illustration.) Using a collimated beam, dispersion of the UV output is minimized as the light is directed down a long collimator to irradiate a horizontal surface, with the intent that the light is collimated into quasi-parallel beams normal to the point of application.

Irradiance is typically measured directly using a radiometer, with the detector head placed at the same height as the surface of the water sample to be treated using the bench-scale system. Radiometers and their detectors must be periodically calibrated, typically by a third party. The radiometer quantifies the irradiance incident to the water at the center of the beam (E0). Several correction factors are employed to account for the average irradiance in the water, which is the best estimate of the average fluence rate to which the sample is exposed. (See list of correction factors.) The adjusted irradiance is used to calculate the actual fluence applied to the sample (IT). For a low-pressure UV lamp (nearly monochromatic emission at 254 nanometers, or 254 nm), the correction factors can be incorporated to calculate the average germicidal fluence rate, Eₐᵥₑ using the equation shown. Multiplying the adjusted irradiance (fluence rate) by the exposure time in seconds gives the UV fluence (mJ/cm²).¹

Equation: E’ₐᵥₑ=E*RF*PF*WF*DF, where E0 = radiometer meter reading at the center of the dish at the height of the surface of the sample (mW/cm²). The correction factors are described in the correction factors list. For polychromatic emissions, multiply the equation by SF and GF, as well.

Correction factors for calculating UV irradiance using monochromatic and polychromatic UV sources.¹


  • Reflection Factor (RF): As the UV light passes from the air into the water, a fraction of the light is reflected off the water surface. The RF accounts for the fraction of the light that enters the water, often represented as 0.975 for air-water interfaces.
  • Petri Factor (PF): The irradiance can vary across the surface area of the sample. The PF accounts for variation by taking the ratio of the average of the incident irradiance over the area of the sample, often in a petri dish, measured by systematically taking measurements at five-millimeter intervals over the area of the petri dish, dividing by the irradiance at the center, and averaging these ratios. Well-designed beams target a PF greater than 0.9.
  • Water Factor (WF): Water matrices may contain species that absorb UV at the emitted wavelength, in which case a WF should be applied to account for the change in irradiance. For completely mixed samples: WF=(1-10⁻ᵃˡ)/(al*ln(10)), where a = decadic absorption coefficient (cm⁻¹) and l = vertical path length of the water (cm).
  • Divergence Factor (DF): Average divergence of the UV beam: DF=L/(L+l’), where L = distance between the UV lamp and the water surface and l = vertical path length of the water (cm).


  • Sensor Factor (SF): Sensitivity of the detector at 254 nm light wavelength divided by the weighted average sensitivity of the detector for 200 to 300 nm wavelengths.
  • Germicidal Factor (GF): The germicidal effect of different light wavelengths varies and is accounted for by the GF, which is the absorbance spectra over the range of wavelengths for the target contaminant (or the absorbance spectra of DNA is often used as an indicator for microbial inactivation studies).

Use of a collimated beam and the commonly associated radiometer-based techniques for determining the UV fluence for a given application is quite effective in tightly controlled laboratory settings. However, in practical applications (e.g., drinking water, wastewater, secondary water supply, small-scale water supply settings),³ alternative approaches may be needed to determine fluence and establish UV system reliability. Toward this aim, different versions of standards have been established to regulate fluence in different waters and serve as references for the design, maintenance, and long-term operation of UV systems, including the Ultraviolet Disinfection Guidance Manual for the Long Term 2 Enhanced Surface Water Treatment Rule (U.S. EPA), Ultraviolet Disinfection Guidelines for Drinking Water and Water Reuse (National Water Research Institute), UV Disinfection Devices for Drinking Water Supply (German Gas and Water Management Union), and Plants for the Disinfection of Water Using Ultraviolet Radiation (ӦNORM).

While ideal UV reactors (approached using the collimated beam testing system) deliver the same fluence to all contaminants in the water, realistic UV reactors provide a distribution of fluences caused by non-uniform optical fields and water flow patterns. In these cases, a reduction equivalent dose (RED) is used to represent the equivalent fluence delivered to all contaminants. Accordingly, the determination of fluence distribution, as well as the RED, is important in reactor design and third-party UV system validation. Approaches including biodosimetry, model simulation, dyed microspheres, and model-detector can be applied for this purpose (as reviewed by Sun et al., 2022).²

Biodosimetry is often used to validate UV disinfection given that it directly assesses microbial inactivation; however, this technique cannot measure the fluence distribution of a UV system. Oftentimes, nonpathogenic surrogate microorganisms are tested in lieu of more dangerous waterborne pathogens that are difficult to work with, such as Cryptosporidium, Giardia, and adenovirus. Biodosimetry requires initial testing in well controlled settings, for example using a bench-scale collimated beam to assess the dose-response behavior of the microbe in response to a range in UV fluences. Using the same conditions as the bench-scale tests (e.g., microbial concentration, flow rate, UV transmittance), full-scale tests are then performed to assess microbial inactivation relationships. Comparison of the bench- and full-scale tests enables estimation of the RED at the operating conditions but cannot provide real-time fluence determination or long-term site monitoring. This is useful for determining disinfection efficacy but is limited to the specific microorganism and conditions tested. This underscores the need for the selection of appropriate microbial surrogates, with common choices including E. coli, B. subtilis spores, and MS2. Cultivation of these microbes to fulfill full-scale testing needs can be cost- and labor-intensive.²

The U.S. EPA recommends model simulation to evaluate UV performance and changes in operational conditions.²,⁴ Such simulations include predictions of both the optical field, as well as the fluid field, the latter of which is modeled using computational fluid dynamics. A number of techniques have been developed to model optical fields, including the point source summation, line source integration, multiple segment source summation, extensive source volumetric emission, and superficial diffuse emission models.

Regardless of the model selection, subsequent experimental validation is needed, which can be performed using approaches such as spherical actinometers, fluorescence microspheres, a silicon carbide detector, or a micro fluorescent silica detector. Commercial software has been developed to integrate optical and fluid modeling in UV reactor design. Using such approaches can be more economical compared to biodosimetry; however, models cannot easily account for uncertainties in input parameters such as flow fields, UV lamp output, fouling, and water temperature.²

Dyed microspheres offer one approach to assess the combined effect of optics and fluids in a UV reactor, wherein a UV-sensitive compound is attached to a synthetic microsphere of size and concentration similar to the target microorganism. Once the compound is activated by UV, it fluoresces, enabling rapid detection and development of relationships used to determine UV fluence. Results from this type of test compare favorably with biodosimetry assessments of the RED, with the added benefit of measuring the UV system’s optical field and posing less microbial safety concerns. However, caution must be exercised to evaluate the water source’s reactivity with the fluorescing compound, since autofluorescence can lead to inaccuracies. The dyed-microsphere method was initially proposed to address some of the limitations associated with the common chemical actinometry approach for radiation measurements.

An actinometer is “any of various instruments for measuring the intensity of incident radiation,” per Meriam-Webster. Chemical actinometers such as potassium ferrioxalate, potassium peroxodisulphate, potassium iodide, and uridine have been widely employed to provide stable and accurate absolute radiation measurements. Yet, they cannot be used to determine fluence distribution or the RED, which can impede UV testing in real systems.²

Using the model-detector method to determine real-time fluence in a UV reactor consists of computational fluid dynamic modeling incorporating parameters such as flow rate, UV transmittance, lamp power output, and sleeve fouling, with actual detector measurements. However, both actual lamp power output and sleeve fouling are difficult to assess in real time, and assumed values are commonly used in the model. This method enables on-site fluence estimation, but its reliability depends on optimized positioning of the detector to facilitate accurate estimations of real-time fluence.²

Each of the approaches described here offers benefits in different applications. Specifically, bench-scale collimated beam testing is useful for establishing initial dose-response models, including kinetic rates of contaminant degradation. Model simulation is useful for assisting with UV reactor design. The biodosimetry and dyed microsphere approaches are well suited for validation of UV reactors. For on-site, long-term monitoring of reactor performance, the model-detector method is appropriate, although its relatively recent development necessitates more thorough validation in realistic scenarios.


  1. J.R. Bolton and K.G. Linden. “Standardization of Methods for Fluence (UV Dose) Determination in Bench-Scale UV Experiments,” Journal of Environmental Engineering 129, no. 3 (March 2003): 209-15.
  2. Z. Sun et al. “A Review of the Fluence Determination Methods for UV Reactors: Ensuring the Reliability of UV Disinfection,” Chemosphere 286, Part 1 (January 2022), 131488.
  3. K. Song et al. “Application of Ultraviolet Light-Emitting Diodes (UV-LEDs) for Water Disinfection: A Review,” Water Research 94 (May 2016): 341-49.
  4. U.S. Environmental Protection Agency. Ultraviolet Disinfection Guidance Manual for the Final Long Term 2 Enhanced Surface Water Treatment Rule, Washington, D.C., 2006; Vol. EPA 815-R-.

About the author
Dr. Brooke K. Mayer is an associate professor in the Department of Civil, Construction and Environmental Engineering as part of the Opus College of Engineering at Marquette University. She holds bachelors, masters and doctorate degrees in civil engineering with an emphasis in environmental engineering from Arizona State University. She is a registered Professional Engineer in the state of Arizona.


Comments are closed.