Field Experience with High-Frequency Ballasts

Raul Abesamis, Paul Black, and Jeffrey Kessel

Physical Plant-Campus Services, University of California at Berkeley

Reprinted from: IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 26, NO. 5, SEPTEMBER/OCTOBER 1990

Related Reports:

Managing a Large Lighting Retrofit Project

Performance of Retrofit Optical Reflectors

ABSTRACT

The Energy Conservation Office of the University of California at Berkeley has installed over 35,000 high frequency ballasts, produced by three manufacturers, in the period 1986 to the present. The quantities of defective ballasts, which are returnedto manufacturers for replacement, are used to estimate the failure rate of these ballasts under field conditions.

UPDATE FOR EXPERIENCE THROUGH 1996

We have installed over 87,000 high frequency ballasts (80% T-12 rapid start, 17% slimline, and 3% high output) between December, 1986 and December 1992. We have recorded 3,900 failed ballasts out of 423,700 ballast-years experience. This results in an overall failure rate of 0.9% for ballasts made by the two manufacturers supplying most of our ballasts.

INTRODUCTION

Utilizing state bond issue funding, the University of Californiaat Berkeley is increasing energy efficiency by spending approximately $1,200,000 per year in the period 1987-1991 on lighting retrofits in campus buildings. The program is based on a detailed survey of all interior spaces in 67 buildings comprising over 6 million sq. ft. The survey results are used as input by a computer-based analysis program which uses IES standard methods to calculate maintained illuminance, which is compared to recommended levels for the taskwithin the space. The ratio of (required maintained illuminance)/(calculated maintained illuminance) is used in eachspace to determine the degree to which the space is underlit or overlit. All existing rapid start, slimline, and high output fixtures receive a cleaning, new lamps (generally energy-saver), high-frequency ballasts and, depending on predicted light levels,may also receive additional measures such as delamping, optical reflectors, daylight sensors, and occupancy sensors. Whenever feasible, incandescent fixtures are replaced by fluorescent fixtures, and high-output fixtures are converted to slimlineoperation.

The installation of high-frequency ballasts is currently the basis of the retrofit program. Our prior experience with such ballasts was disappointing due to very high failure rates (>10%). Nevertheless, we believed that the technology had matured so thatthrough careful selection a satisfactory product could be obtained. Most of our recent ballast purchases have been from three manufacturers, as summarized in Table 1. Typically severalthousand ballasts were ordered at a time, in both 120v. and 277v. versions. The rapid start ballasts were specified to be true rapid start, because campus maintenance procedures benefit from maximizing lamp life. For installations using group relamping on a regular basis, the choice of a modified rapid start high-frequency ballast, assuming its reliability was acceptable, will result in even greater energy savings.

Table 1. High-Frequency Ballast Purchases, 1986-1988, Three Manufacturers

Notes to Table 1: a) These ballasts are specially ordered from manufacturer to produce the indicated percentage of light output relative to the manufacturer's standard model. b) These ballasts may be operated at partial light output by installing special dimming resistors supplied by the manufacturer. c) Includes cleaning, relamping, and disposal of old lamps and ballasts. The energy-saving potential of high-frequency ballasts has been reported by independent laboratories and by manufacturers [1].

Table 2 presents a comparison between typical high-frequency units and their energy-efficient magnetic counterparts.

Table 2. Ballast System Power , Watts

Note to Table 2: Data compiled from manufacturer's literature and independent laboratory reports. The first value refers to operation with standard lamps. The value in parentheses refers to operation with low-wattage lamps.

BALLAST MORTALITY

Contractors performing luminaire reballasting under our program, as well as campus electrical maintenance workers replacing failed ballasts, were instructed to return defective ballasts to our office. We inventoried these ballasts in order to return them to their manufacturers for replacement. The results of our inventory are presented in Table 3. Ballast manufacturers have reported to us in private communications that 20-30% of the ballasts returned to them as defective are, in fact, not defective. This factor tends to inflate the percentage of reported defective ballasts. On the other hand, we believe that a small but unknown percentage of defective ballasts were thrown away and not returned to our office forinventory. Another possible confounding factor is the fact theballasts from manufacturer C will not start lamps when the ballast has been rapidly switched off and on, because a 15 second pause in the off position is required. Although we trained our contractors and maintenance personnel to be aware of this fact,some of these ballasts may have erroneously been removed as defective. Nevertheless, we believe our statistics to provide a good indication of the ballast mortality that will be experienced under diverse interior field conditions, both in terms of absolute ranges of failures, as well as in relative terms between models and manufacturers.

Table 3. High-Frequency Ballast Mortality in Field Operations (After three years of operation)

Notes to Table 3: a. The maximum time since installation is around 3.5 years. b. Approximately 75% of the ballasts operate low-wattage lamps. c. Rapid start ballasts from manufacturer A operate either one or two lamps. d. Ballasts characterized by 70%, 80%, or 85% were specially modified by manufacturer to provide the stated percentage of full light output, relative to the unmodified version. e. Ballasts from manufacturer C were mostly operated at partial light output by means of special dimming resistors supplied by the manufacturer.

CONCLUSION

Careful selection of high-frequency ballasts will result in not only 20-30% energy savings, but also reliable, quiet, flicker-free operation. They provide a rapid payback whenspecified for new construction, or when used as the replacement for failed magnetic ballasts. Depending on operating hours andthe cost of electricity, the payback may be acceptable even when high-frequency ballasts are used to replace operating energy-efficient magnetic ballasts (in conjunction with cleaningand lamp replacement). The availability of partial light-output ballasts makes it feasible to fine tune light levels in spaces that are overlit. gaining a proportional decrease in energy use. Although ballasts from manufacturer C had an apparent failure rate considerably higher than the other ballasts, it must be remembered that these "C" ballasts provided some unique features, such as the ability to operate at various partial light-output levels depending on choice of a resistor wired to the ballast. These ballasts were also available with both a manual and a photocell-controlled low-voltage dimming system. These ballasts are no longer in production.

REFERENCES

(1) Performance of Electronic Ballasts and Other New Lighting Equipment, Electric Power Research Institute, Palo Alto, CA., EPRI EM-4510, March 1986.