Performance testing of ultraviolet bulblamps for point-of-use water disinfection
Rachel L. Peletz1,2, Amy J. Pickering1,3, Alicia R. Chakrabarti1,4, and Kara L. Nelson1,*
Abstract
Designs for the disinfection of drinking water at the point-of-use (POU) using ultraviolet (UV) light typically employ a single bulblamp, and thus the treatment performance and operating costs are dependent on the irradiance and lifetime of this single bulblamp. Short-term experiments in a quasi-collimated beam set-up were conducted to characterize the warm-up period after UV bulblamps were turned on, as well as the decrease in bulblamp irradiance over the first 14 d. Long-term experiments were conducted to study the effects of on/off cycling (1-h, 3-h, and 12-h cycles) on bulblamp irradiance and lifetime. All experiments were conducted with bulblamps suspended in air, which is representative of the low-cost POU designs currently available. After being turned on, the UV bulblamps required an average of 5 min to reach 98% of their maximum irradiance. The irradiance of bulblamps decreased continuously throughout their lifetime, falling on average by about 10% during the first 100 h, and another 70% by the end of the bulblamp’s life. The lifetime of bulblamps was significantly affected by the on/off cycling; the mean lifetimes (until failure) were 2100, 3680, and 6760 h for the 1-h, 3-h, and 12-h cycling regimes, respectively. The implications of these results for POU disinfection of drinking water with UV are discussed.
1 Department of Civil and Environmental Engineering, University of California, Berkeley
2 Current address: London School of Hygiene and Tropical Medicine, London, United KingdomCentre for Affordable Water and Sanitation Technology, Calgary, Canada
3 Current address: Stanford University, Palo Alto, CAFundaciόn Cantaro Azul, La Paz, Baja California Sur, Mexico
4 Current address: East Bay Municipal Utility District, Oakland, CA
*Corresponding author: Department of Civil and Environmental Engineering, University of California, Berkeley, CA, 94720-1710; (510) 643-5023; ; www.ce.berkeley.edu/~nelson
Manuscript submitted to XXX Journal of Environmental Engineering, OctoberMarch, 20097
Introduction
The World Health Organization estimates that 1.87 million deaths per year worldwide are attributed to unsafe water, sanitation, and hygiene (WHO 20052). Ninety percent of these deaths are in children and virtually all are in developing countries. Globally, 8001.1 million billion people still lack access to improved drinking water sources (WHO 2010and UNICEF 2006). Thus, one of the Millennium Development Goals is to halve the population without access to safe water and sanitation by the year 2015.
Treating drinking water at the point-of-use (POU) is being promoted worldwide, using low-cost technologies such as chlorination, flocculation plus chlorination, solar disinfection (SODIS), and filtration with granular media or ceramics. Low-cost options for disinfecting drinking water with ultraviolet (UV) light have recently been developed (Brownell et al. 2007; Drescher et al. 2001; Rau 2003). Unlike larger-scale UV applications, POU technologies typically employ a single UV bulblamp, and thus the treatment performance and operating costs are highly dependent on the irradiance and lifetime of this single bulblamp. Although UV bulblamp manufacturers generally provide some information regarding bulblamp performance, the data available are not sufficient for designing reliable water disinfection technologies. The goal of the research reported herein was to characterize the performance of UV bulblamps in terms of: (i) bulblamp irradiance during the warm-up period; (ii) changes in irradiance throughout the life of the bulblamp; and (iii) the effects of on/off cycling on bulblamp lifetime and irradiance. Implications for the use of UV disinfection for household drinking water treatment are also discussed.
Methods
The bulblamps studied were low-pressure, 15-W, mercury bulblamps (G15T8, General Electric, Cleveland, OH). Each bulblamp was connected to an instant-start, electronic ballast (REL-2P32-SC, Advance Transformer Co., Rosemont, IL). All bulblamps were suspended in air, as the low-cost bulblamp technologies of interest use UV bulblamps suspended over (rather than submerged in) the water being treated (Brownell 2007; UV Filtration System, RDI Cambodia, http://www.rdic.org/waterfiltrationsystems.htm; UVeta, http://www.niparaja.org/uveta/uv-eta/main.htm) rather that submerge the lamp in water as in more expensive commercial systems. Both long-term and short-term experiments were conducted, as summarized in Table 1 and described in the following sections.
Short-term Experiments
The short-term experiments included intensive monitoring of new UV bulblamps for the initial 30 min of operation, the following 14 d, and a subsequent 30-min start-up period (Table 1). One bulblamp was studied at a time, by placing it in a quasi-collimated beam apparatus (Brownell et al. 2007); the irradiance was measured 38 cm below the bulblamp using a hand-held digital ultraviolet radiometer (IL1400A, International Light, Newburyport, MA) positioned by a fixed metal stand. Because each bulblamp was placed in the same apparatus, any differences in bulblamp irradiance were attributed to the bulblamp itself (the effect of temperature is discussed later). The bulblamp temperature (approximately 1 cm from bulblamp surface) and ambient temperature were measured using graduated glass thermometers inserted into the bulblamp housing and in the lab room, respectively. A power supply regulator was used to eliminate voltage fluctuations (Tripp Lite Voltage Regulator and Conditioner, Model LS 604, Chicago, IL). Nine bulblamps were monitored for the entire 14-d period; at the same time each day, the irradiance, bulblamp temperature, and room temperature were recorded. Six of the bulblamps were monitored more frequently during the first 30 min that they were turned on as new bulblamps. At the end of each 14-d study, the bulblamp was turned off for a few days, and then another 30-min experiment was performed on six of the bulblamps. The frequency of irradiance measurements is shown in Table 1. The temperature of the room and bulblamp were recorded only at the end of the 30-min studies.
Long-term experiments
The long-term experiments were designed to characterize bulblamp performance during the entire lifetime of the bulblamps. Nine bulblamps were studied simultaneously; each bulblamp was mounted horizontally within a metal tube encasing that had a 2-cm hole in the bottom below which irradiance measurements were taken. Because nine different bulblamp set-ups were used, we cannot rule out the possibility that slight differences between them (size of opening in metal encasing, distance from bulblamp to surface of radiometer) may have caused the irradiance measurements to differ slightly (unlike the short-term studies, in which the same collimated beam apparatus was used for all bulblamps). Thus, the emphasis of the data analysis was on changes in irradiance over the lifetime of each bulblamp. The on/off cycling of the bulblamps was regulated by automatic digital timers in three categories: 1-h, 3-h, and 12-h cycles. For each cycle, the bulblamps were off for the same amount of time that they were on, so that each bulblamp was on 12 h per day regardless of its cycling category. Measurements were taken at the same time once a week, after all bulblamps had been on for 30 min (after being off for a minimum of 1 hour and a maximum of 12 h, depending on cycling category). Irradiance was recorded using a hand-held ultraviolet digital radiometer (Spectroline DM-254XA, Spectronics Corporation, Westbury, NY) and bulblamp temperature (approximately 1 cm from bulblamp surface) was measured using graduated glass thermometers inserted into each metal encasing. Room temperature near the bulblamps was also recorded during the weekly measurements. Originally, three bulblamps of each cycle category were included in the set-up, and after burn-out, bulblamps were replaced, so that testing included 6 bulblamps for 1-h, 4 bulblamps for 3-h, and 4 bulblamps for 12-h cycles, for a total of 14 bulblamps.
Results
Short-Term Experiments
30-min study with new bulblamps. The mean relative irradiance of six new bulblamps during their first 30 min of operation is shown in Figure 1. Relative irradiance was calculated as a percentage of the maximum irradiance recorded during the 30-min operating period. This maximum irradiance was reached after an average of 15.4 min, though the irradiance was within 98% of the maximum from 5 to 30 minutes (Figure 1). On average, the bulblamps reached about 70% of their maximum irradiance after 5 s, 84% after 1 min, 92% after 2 min, 96% after 3 min, and 98% after 5 min (Table 2).
30-min study with 14-d old bulblamps. The mean relative irradiance of six 14-d old bulblamps during the first 30 min of operation (after being turned back on) is shown in Figure 2. As with the new bulblamps, relative irradiance was calculated as a percentage of the maximum irradiance recorded during the 30-min operating period; this maximum occurred after an average of 10.9 minutes. The relative irradiance was similar to the new bulblamps, except that the initial relative irradiance was lower by approximately 10%. The bulblamps reached about 60% of their maximum irradiance after 5 s, 80% after 1 min, 90% after 2 min, 95% after 3 min, and 98% after 5 min (Table 3).
14-d study. The decrease in bulblamp irradiance, as a percentage of the initial irradiance, during the first 14 d of operation is shown in Figure 3. Note that the initial irradiance was defined as the irradiance after 30 min, which was 1.4% lower than the maximum irradiance determined from our 30-min studies, on average (Table 2). The bulblamp irradiance decreased rapidly for approximately the first 24 h and then continued to decrease for the entire 14 d, but at a slower rate. The mean initial irradiance (38 cm from bulblamp) was 85.2 μW/cm2, which decreased to 78.3 μW/cm2 (91.9%) after 100 h and to 74.2 μW/cm2 (87.4%) after 14 d (Table 3).
Long-term Experiments
The relative irradiance (% of initial) of bulblamps over their entire lifetime is shown in Figure 4. In the long-term study, the initial irradiance was measured after approximately 100 h of operation; thus, the initial irradiance for these experiments was about 10% lower than the maximum irradiance measured during the 30-min studies. At the end of the bulblamps’ lifetimes, the average relative irradiances were 70.5, 71.5, and 67.7% of the initial irradiance for the 1-h, 3-h, and 12-h cycling regimes, respectively. The decrease in irradiance for bulblamps in all three on/off cycling categories was similar; no significant difference in irradiance decrease was observed (Figure 4). However, the bulblamp lifetime decreased with more frequent on/off cycling. The mean lifetimes for bulblamps with 1-h, 3-h, and 12-h on/off cycles were 2100 h, 3680 h, and 6760 h, respectively (Figure 5 and Table 4).
Discussion
BulbLamp Irradiance
The irradiance of UV bulblamps increased during an initial warm-up period, which lasted about 5 min. The relative initial irradiance (% of maximum) of used (14-d old) bulblamps was slightly lower than new bulblamps, but within 1 min the differences were minimal. (Note, however, that the absolute irradiance was lower for the old bulblamps; Table 3). During this warm-up period, the liquid mercury is vaporized, and the current that flows between the electrodes excites the mercury vapor, which discharges UV radiation (IESNA 2000; NSF Joint Committee on Drinking Water Treatment Units 2002). The bulblamp warm-up time we observed is consistent with information published by the manufacturer, which states that typical warm-up periods are 4 to 8 min, depending on the specific bulblamp design, ballast type, ambient temperature, and the voltage applied (Weitz 1956). The relative irradiance during the warm-up period and the length of the warm-up period have direct implications for the design and operation of POU disinfection devices, as discussed in a later section.
After the initial warm-up period, the irradiance decreased with time throughout the life of the bulblamp. The main reason for the decrease in irradiance is believed to be the accumulation of light-absorbing deposits within the lamp (IESNA 2000; U.S. EPA 2006). The lamp aging factor, defined as the ratio of the output at the end-of-life over the output at 100 h, is typically between 0.5 and 0.8 for germicidal UV bulblamps (U.S. EPA 2006). For the bulblamps cycled at 1, 3, and 12-h intervals in this experiment, the average lamp age factors were 0.71, 0.72, and 0.68, respectively. The shape of our irradiance curves (i.e., Figures 3 and 4) is similar to the curve published by the manufacturer for the G15T8 bulblamp (General Electric 2005), except that we observed a much faster rate of decrease in irradiance. One factor affecting the lifetime in our experiments compared to the manufacturers is the bulblamp cycling, as discussed below. (Presumably the value of 8000 h is for a continuously burning bulb; note, however, that on 2/21/03 a similar data sheet was accessed on the same website that reported a decrease to 80% of initial irradiance after 7500 h for a bulb operated with 3-h on/off cycles).
In addition to the bulblamp irradiance decreasing more rapidly than reported by General Electric (GE), the bulblamps in our study also produced a lower initial irradiance. The data available from GE is the irradiance (254 nm) at 1.0 m from the lamp, after 100 h of continuous burning. We converted the irradiance measured in our experiments at 100 h at a distance of 0.38 m (78.3 μW/cm2) to the irradiance at 1.0 m using the equation (Power) = (Area) x (Irradiance). The calculated irradiance from our experiments (for continuous burning) was 30 μW/cm2 (95% CI 28.9-30.5 μW/cm2), compared to a value of 49 μW/cm2 reported by GE for continuous burning (General Electric 2005) and 38 μW/cm2 for 3-h cycling (Weitz 1956). (Note that these two references from General Electric were published many years apart, and other factors may have changed in addition to the bulblamp cycling regime.) The lower bulblamp irradiance in our study may be partially due to the ballast used, the on/off cycling, the ambient temperature (discussed below), or the fact that our measurements did not include the contribution of scattered light.
The first 100 h of bulblamp operation are typically referred to as the “burn-in period”; during this time the irradiance is reported to be potentially unstable (Benya et al. 2003; Weitz 1956). Such instability was not observed in any of our bulblamps, though it is acknowledged that this study examined a limited number of lamps from a single manufacturer. Thus, our results imply that it appears safe from a performance perspective to utilize new bulblamps for water disinfection.
BulbLamp Lifetime
Under normal operating conditions, germicidal UV bulblamps are typically reported to be operational for about one year, or 8640 h (Masschelein 2002; NWRI 2003; U.S. EPA 2006). The mean bulblamp lifetimes (until failure) observed in our research were much shorter, ranging from 2100 h to 6760 h for the different bulblamp cycling regimes (Figure 5 and Table 4). GE reported the specific G15T8 bulblamp to have a useful life of 8000 h with continuous burning (General Electric 2005). (Presumably the value of 8000 h is for a continuously burning lamp; note, however, that on 2/21/03 a similar data sheet was accessed on the same website that reported a decrease to 80% of initial irradiance after 7500 h for a lamp operated with 3-h on/off cycles). GE defines the end of a bulblamp’s useful life as when the lamp light irradiance drops below 80% of the initial irradiance, with initial defined as the irradiance at 100 h (Benya et al. 2003; General Electric 2005). Using this definition, tThe average useful lifetimes of the bulblamps we tested were 1470, 1610, and 1200 h for the 1-h, 3-h, and 12-h cycling regimes, respectively (Table 4). Only two of the 14our bulblamps examined even lasted for 8000 h.