Ultraviolet-C systems installed inside ventilation ducts are often evaluated by lamp rating alone. A new study suggests that this can miss a central engineering factor: airflow can cool the lamps, changing their ultraviolet output while also reducing the time airborne particles spend in the irradiated section.
Researchers at the Singapore Institute of Technology experimentally measured how low-pressure mercury UV-C lamps respond to changing duct airflow and inlet-air temperature. They then used the measured ultraviolet fields to calculate comparative particle doses and estimated microbial reductions within a test duct.
The findings, published in the Journal of Building Design and Environment , show that a higher-power 95 W lamp maintained a more stable UV-C output across the conditions tested than a 60 W lamp. The lower-power lamp was more sensitive to convective cooling caused by faster airflow and lower inlet-air temperatures.
In-duct ultraviolet germicidal irradiation, or UVGI, is used in heating, ventilation, and air-conditioning systems to expose moving air to germicidal ultraviolet radiation. Because the UV dose received by airborne particles depends on both irradiance and exposure time, airflow has two connected effects: it can cool a lamp and reduce its output, while also shortening the time particles remain in the duct.
The researchers tested 95 W and 60 W low-pressure mercury lamps in an approximately 2.7 m-long square duct. Air velocities ranged from 1.1 to 2.5 m/s, and inlet-air temperatures ranged from 14 to 22 degrees Celsius. The team measured lamp-surface temperature and 254 nm irradiance, mapped irradiance from arrays containing two to six lamps, and calculated particle-specific UV dose from the measured radiation fields.
The 95 W lamp produced mean 254 nm irradiance values of approximately 273 to 316 W/m² across the tested conditions. By comparison, the 60 W lamp ranged from about 151 to 262 W/m² and showed stronger dependence on airflow and ambient temperature.
At lower temperature conditions, the 60 W lamp operated at approximately 58% to 74% of its maximum measured output, depending on airflow velocity. At 22 degrees Celsius, its relative output increased to approximately 86% to 100%. The 95 W lamp maintained approximately 97% to 100% of its maximum measured output at the two lower airflow conditions studied.
The study also found that convective heat loss became increasingly important as airflow rose. The calculated ratio of convective to radiative external-surface heat loss increased from approximately 4.6–5.0 at the low-flow condition to approximately 7.1–7.6 at 2.5 m/s.
For six-lamp arrays operating at 1.1 m/s, the mean modeled dose was approximately 191 J/m² for the 95 W lamps and 146 J/m² for the 60 W lamps. Increasing lamp count generally increased measured irradiance and calculated dose, while lamp arrangement affected the spatial distribution of lower- and higher-dose regions.
Using published UV-susceptibility constants, the authors calculated reductions exceeding 6 log for several susceptible viral and bacterial species under selected configurations. In contrast, predicted reductions for Aspergillus spores were much smaller: below approximately 0.05 log for A. niger and generally below approximately 0.4 log for A. flavus and A. glaucus.
The microbial inactivation findings are comparative calculations, not direct measurements of viable bioaerosols. The study did not expose live airborne viruses, bacteria, or fungal spores to the tested system. Instead, it combined measured irradiance fields with published species-specific susceptibility constants in a simplified particle-transport model.
The authors caution that the calculated reductions should not be interpreted as guaranteed real-world performance. The model assumed plug flow and fixed cross-sectional particle locations, and it did not account for turbulent cross-stream mixing, deposition, particle aggregation, or other complexities of operational ventilation systems. Lamp ageing, fouling, ballast performance, duct geometry, and microbial conditions can also affect practical UVGI performance.
The findings indicate that, for the tested system, the favorable combined conditions for irradiance and particle residence time occurred at nominal airflow velocities of 1.1–1.5 m/s. This range should not be treated as a universal operating recommendation, since practical HVAC design must also account for required ventilation rates, duct size, pressure losses, fan energy, acoustic requirements, and the intended microbial-control target.
The work provides a measurement-informed framework for comparing lamp rating, airflow condition, and lamp arrangement in in-duct UVGI systems. Future studies using controlled bioaerosol experiments and more detailed particle-transport modeling will be needed to validate predicted microbial performance.
