Lighting Retrofit Report

We have completed a six year $6 million lighting retrofit program, resulting in a decrease of around $1 million per year in lighting energy cost. This project, and some of our measurements of electronic ballast reliability and optical reflector performance, are described in the following three documents:

Managing a Large Lighting Retrofit Project

Field Experience with High-Frequency Ballasts

Performance of Retrofit Optical Reflectors

If you have questions about Energy Conservation Projects at the University of California at Berkeley, please contact:

Raul Abesamis, Energy Engineer
(510) 642-6254
Fax: 510-643-6588
abesamis@berkeley.edu

The Berkeley Campus of the University of California have over 100 buildings of varying age and function, comprising nearly 9 million square feet. The campus electrical bill has been continually climbing and is approximately one million dollars per month. In 1986 funding was obtained from a State of California bond issue to embark upon a six-year program to improve the energy efficiency of the lighting in around 60 state-supported buildings. We completed the six million dollar project using a novel project management system, which has resulted in the successful installation of over 87,000 high-frequency ballasts, 16,000 optical reflectors in delamped luminaries, and many other energy-saving devices. This has resulted in a 25%-40% decrease in lighting energy use. In order to maximize lighting energy efficiency, survey results must be used to individually analyze spaces and specify retrofits tailored to the requirements of the different spaces.

Managing a Large Lighting Retrofit Project

Jeffrey D. Kessel, P.E.
Physical Plant-Campus Services
University of California
Berkeley, CA 94720

ABSTRACT

The Berkeley Campus of the University of California has over 100 buildings of varying age and function, comprising nearly 9 million square feet. The campus electrical bill has been continually climbing and is approximately one million dollars/month. In 1986 funding was obtained from a State of California bond issue to embark upon a six year program to improve the energy efficiency of the lighting in around 60 state-supported buildings. We completed the six million dollar project using a novel project management system, which has resulted in the successful installation of over 87,000 high-frequency ballasts, 16,000 optical reflectors in delamped luminaires, and many other energy-saving devices. This has resulted in a 25%-40% decrease in lighting energy use. In order to maximize lighting energy efficiency, survey results must be used to individually analyze spaces and specify retrofits tailored to the requirements of the different spaces. This paper describes our approach and methods.

PRELIMINARY ANALYSIS OF LIGHTING RETROFIT OPPORTUNITIES

Before applying for financing we had to convince ourselves that retrofit hardware was available with cost, performance, and reliability that would ensure a cost-effective installation to reduce energy use. We began by collecting data on candidate systems and brands. Whenever possible we relied on data from independent test labs and on information from experienced users of the products. Price quotes were based on realistic quantities. For example, our ballast orders usually exceeded 2,000 units. When we obtained quotes from lighting maintenance contractors who would be changing out the old ballasts we made clear that the unit prices quoted should reflect a future contract that would represent at least six months work.

There are many lighting retrofits that can be used to increase efficiency, and no way to know what mix of retrofits will be used until a detailed analysis of each space is conducted. We were sure, however, that high-frequency ballasts would be used throughout our project. Furthermore, we wanted to install reflectors with delamping in order to reduce lighting levels in overlit spaces. The following two cases show the assumptions we used in our preliminary analysis of these two common retrofits.

Ballast Swap

When we determined, using methods discussed below in the section titled Analysis of the Spaces, that a space was not overlit, then the luminaires were specified to be cleaned and relamped, and the old ballasts removed and replaced with high-frequency ballasts. The charge for labor and lamps came to $13.00 for a 2F40 luminaire, and the cost of the ballast, which was supplied to the lighting maintenance contractor by the University, was $22.00. Thus for $35.00 the old fixture is renovated. We used 34 watt T-12 (3500 K) lamps on campus, and we assumed that the old ballast was a standard magnetic ballast operating at 79 w., and that the new ballast operates at 61 w. The ballast swap reduces power by 18 w = 0.018 kw per luminaire.

In a space lighted for 3,000 hours per year, at a cost of electricity of $0.08, the simple payback for one ballast swap was calculated to be: SP = [cost] / [annual savings]

= [35.00] / [(.018w)*(3000hrs/yr)*(0.08 $/kwh)]

= 8.1 years. Today's paybacks should be shorter due to lower ballast prices for the T-8 system ballast.

If the luminaire is used 8760 hours per year, at an average cost of electricity of $0.06/kwh, then the SP = 3.7 years.

The difference in rate between $0.08 and $0.06 reflects higher utility charges during periods of peak use.

In cases where spaces are overlit, it may be possible to install partial power ballasts with reduced light output, or to convert from HO to slimline by changing lamp sockets. Table 1 presents the power level for various lamp/ballast systems. The values in parentheses are the system power when energy-saver lamps are used.

TABLE 1. BALLAST SYSTEM POWER, WATTS

Although our campus experienced poor reliability when using first-generation high frequency ballasts in the late 1970s, our present experience with ballast reliability is very satisfactory. Moreover, failures within the three year year warranty period qualify for free replacements as well as a $10 labor rebate.

Reliability results after three years of operation are reported in [Abesamis, 1990]. We have updated our measurement of reliability after the installation of over 87,000 high frequency ballasts (80% T-12 rapid start, 17% slimline, and 3% HO) between December, 1986 and December, 1992. We have recorded 3,900 failed ballasts out of 423,700 ballast-years experience, resulting in an average annual failure rate of 0.9 %, for the ballasts made by the two manufacturers supplying most of the project's ballasts.

Optical Reflectors and Delamping

Some optical reflector vendors claim that their installation will increase light levels, even while removing half the lamps. In order to evaluate reflector performance we conducted a "Reflectoff" that compared the illuminance from a basecase luminaire (cleaned and fully lamped with new stable lamps) to the illuminance from the same fixture using one-half the quantity of the same lamps and an optical reflector. The results are presented in [Kessel, 1990] and are summarized in Table 2.

TABLE 2. OPTICAL REFLECTOR PERFORMANCE

We also measured the distance out from the main axis of the luminaire at which the illuminance dropped to 1/2 its value directly under the luminaire. This "half-max" distance decreased by around 15% for all the reflectors we measured. Although this decrease in uniformity is not significant for most applications, it should be taken into account when tasks are located midway between luminaires. Note that this decrease in high-angle luminance may advantageously decrease glare in some situations.

Spaces that are overlit such that they can tolerate a reduction in illuminance of at least 29% may do well with removal of half the lamps and installation of an optical reflector above the lamps. This mirror-like (specular) piece of sheet metal is designed to increase the efficiency of the luminaire. Modifying a four lamp luminaire produces nearly three lamps worth of light for the energy cost of operating two lamps. Our cost in 1986 of $70.00 included removal of old lamps and ballasts, provision and installation of the reflector and two lamps, and installation of a University-supplied ballast. The decrease in wattage in changing from two standard magnetic ballasts to one high-frequency ballast is 2*79 - 61 = 97 w. Then the simple payback for a luminaire operating 3000 hours per year at $0.08/kwh is calculated to be: SP = [70.00] / [(97)*(3000)*(0.001)*(0.08)] = 3.0 years. And for 8000 hrs/yr at $0.06/kwh, the SP = 1.4 years.

We made calculations such as these before beginning the project, and realized that no matter what the final mix of retrofits eventually specified, the overall project SP would likely be around five years. Obviously this parameter is dependent on many factors, including the cost of electricity, annual hours of use, type of existing equipment and illuminance levels, utility rebates, and the cost of materials and labor.

SURVEY OF THE FACILITY

We developed a form to record survey data for every space in a building. The data included the room geometry and luminaire type which are both required for subsequent cavity ratio method calculation of available illuminance, which is then compared to the illuminance required for the task in the space. The form also holds information on occupancy patterns that indicate potential for motion sensor lighting control. The surveyors also input data on the quantities of luminaires in the daylighting zone (approximately within 15 feet of windows) for possible application of photocell-controlled dimming ballasts.

Ten engineering students were hired and trained to use the survey form. Five million square feet of occupied space in sixty campus buildings was surveyed in a period of approximately three months. A building's data base comprised one data form for each room that used electric lighting beyond a minimum number of kwh/yr, and a plan sketch of the room showing workplaces, luminaires, switches, and obstructions. In addition the surveyor developed a luminaire glossary containing sketches of the different lighting fixtures in the building.

The forms for each building were entered into a computer as a data base file. One could use an ultra-sonic "tapeless" tape measure, and a "palm-top" computer containing the survey form as a spread sheet that could be subsequently down-loaded into a data base running on a desk-top computer. Once a building's data base was completed, analysis of the lighting began.

ANALYSIS OF THE SPACES

As an example, Figure 1 shows how a building is divided into three families of identical spaces. The input consists of room dimensions, reflectance, use (task), lamp type, maintenance factor, and IES code number characterizing the luminaires. The software then uses the cavity ratio method to calculate maintained illuminance, and compares this to the stored value recommended by the IES for the indicated task in the space. The result for each family of spaces is a calculated quantity called the fullratio, described below.

We attempted to prescribe a lighting retrofit that was suited to each space in a building. Rather than use a single type of retrofit throughout a building, we tailored each retrofit to the room in which it would be installed. The key to this is to calculate or measure the illuminance in each space and to compare this level to the illuminance recommended for the on-going visual task by the Illuminating Engineering Society of North America (IES). The IES recommendations are for maintained illuminance levels, but the cavity-ratio method calculations assume a new luminaire with fresh lamps. For this reason a suitable light-loss factor must be introduced. This factor was chosen to be 0.70 for our campus, which does not use group relamping. A facility using group relamping would have a higher factor which allows more aggressive retrofits.

We characterize the degree to which each space is over- or under-lit by a calculated quantity F, called the Fullratio, which is defined to be the (required illuminance) / (available illuminance).

F = (maintained illuminance recommended by the IES)

= (cavity-ratio calculated illuminance) * (0.70)

Care must be taken in interpreting F, because its value is based on a calculation of average illuminance throughout a space. Partitions, luminaire/worktable relative position, and other factors can invalidate the calculation. For this reason the surveyor's room sketch is used to detect anomalous conditions.

If F=1, then a space has exactly the recommended maintained illuminance. If F=0.5 a space has twice the recommended level. The value of F calculated from the data base by software we developed was the main indicator used to select an appropriate retrofit for each space. If there are many identical spaces in a building, then one of them can be selected for cleaning and relamping, followed by illuminance measurements (after the lamps have aged for at least 100 hours). This yields greater accuracy than the calculations, but is too costly to do in all spaces. Some retrofits, such as full light output high-frequency ballasts and motion controls, are suitable for spaces that are not overlit. Other retrofits, such as delamping with optical reflectors, or partial light-output ballasts, are suitable for overlit spaces. Figure 2 shows retrofits appropriate to various fullratio ranges.

RETROFIT HARDWARE AND STRATEGIES

In order to reduce lighting energy use it is necessary to focus on the two components of energy:

ENERGY = POWER x TIME

The basic cost of energy is the product of power and time. Table 3 shows some of the many strategies and lighting retrofit devices that decrease power level or duration of use.

TABLE 3. STRATEGIES AND HARDWARE

These measures are not uniformly applicable throughout a building. Those marked with an asterisk (*) should only be used in spaces that can tolerate a decrease in maintained illuminance.

Note that depending on the aesthetic and illuminance requirements, there are several retrofits that will replace an existing incandescent source: replacement with a lower wattage halogen capsule source, replacement with a compact fluorescent lamp (preferably hard-wired to eliminate the incandescent socket), or elimination of the incandescent luminaire and replacement with one using a more efficient source.

SPECIFICATION

Quality

Based on the calculated or measured Fullratio, and on the existing types of luminaires, it is not difficult to select a room-specific retrofit that will provide adequate maintained illuminance. But attention must be given to questions of lighting quality. Most existing luminaires that pre-date the mid-1980s were not designed to maximize the VDT user's comfort and productivity. Financial constraints limited us to replacing prismatic lenses with similar lenses, and then only if the existing lens was in poor condition. However, a retrofit parabolic diffuser costs around $25.00 per 2' x 4' fixture for a 1/2" cell sheet that is 1/2" thick, and around $35 for a 1.5" cell sheet that is 1" thick. The salary of the VDT user inconvenienced by glare from the conventional prismatic lens greatly exceeds this cost. A small improvement in comfort and productivity may well be worth the initial cost of the better diffuser.

When improving lighting quality has higher priority, small cell parabolic louver diffusers can often be used to replace the existing flat prismatic lens. This results in a low-brightness luminaire that interferes less with VDT use. A better solution involves the use of new indirect fixtures, or of direct fixtures with deep-cell parabolic diffusers.

Specifying Room-specific Retrofits

The data base software returns to the specifier the list of rooms, each of which having a Fullratio assigned to it either from calculation or measurement. By taking into account this measure of adequate illuminance, along with the type of existing lighting system, the occupancy patterns of the room's users, and the availability of daylight, the experienced specifier can choose a suitable combination of retrofits from Figure 2 and Table 3.

IMPLEMENTATION

The data base assembled for analysis of the spaces in a building can readily incorporate the specified retrofits, and be used to generate room lists for contractors. These lists present both the existing lighting system and the specified retrofit for each space in a building. These lists also provide the means for contractors to verify the assumed existing conditions, and to note survey errors and unanticipated field conditions. Invoices may be readily verified against the room lists, particularly when contractors have provided quotes for various unit operations that can be coded into the retrofit description. The software developed during this project is available under license from the University of California at Berkeley.

COMMUNICATION

There are a great number of information resources and people with whom the project manager needs to communicate during the various stages of the project.

Planning

During the planning stage information can be obtained from vendors, independent testing laboratories, national laboratories, and other facility managers. Excellent publications are available on lighting issues [Eley Associates, 1993]. Contractors can be contacted to obtain price quotes on the anticipated spectrum of work. Proposed retrofits should be discussed and/or demonstrated to occupants and maintenance personnel.

Construction

Building managers and occupants must be alerted to impending construction, and contractor scheduling provided for limited access areas. Contractors will need training on the required work methods, record keeping, and on the rules for disposal of hazardous materials generated by the project.

Follow-up and Maintenance

The facility electricians need to be familiar with the new technologies being introduced. Hands-on training sessions should be provided, along with an as-built list of all locations of retrofit lighting controls.

The storehouse must be given notice to phase out superceeded equipment, and to stock the new retrofit devices for maintenance use.

ENVIRONMENTAL CONCERNS

In general energy conservation greatly benefits the environment. As Amory Lovins has stated, "Reducing energy use is the quickest, safest, most cost-effective way to reduce the emission of greenhouse gases that accompanies the generation of electricity from fossil fuels." Reducing energy use also reduces the dependence of the U.S.A. on imported fuel. Lighting energy conservation has enormous potential: in the U.S.A. fluorescent lighting consumes around 200 billion kwh/yr, at an annual cost of $14 billion, or 1/10 of the nation's electric bill. If high-frequency ballasts were universally adopted the savings would amount to nearly $3 billion per year.

There are, however, some environmental concerns that accompany a lighting retrofit. Contract documents should specify that both ballasts and lamps receive proper disposal. Under the Toxic Substances Control Act of 1976, old ballasts that are not labeled "No PCBs" must be handled as toxic waste. Contractors must pack the ballasts in storage drums at a designated site on the premises, for shipment by a licensed hauler to ultimate disposal. In addition the contractors must be trained to properly deal with leaking ballasts that may contain PCBs. Additional information can be obtained from the U.S.E.P.A. We currently are paying around $4.00 per ballast for disposal.

Fluorescent lamps also contain harmful substances. California Administrative Code, Sections 66699(b) & 66680(d), discusses proper disposed of the used lamps. Our campus currently is paying $0.40 per four foot tube for this service, which requires that contractors store the used lamps in cartons at designated transfer points for pickup by the licensed lamp recycling firm.

RESULTS

The lighting portion of a building's total electrical use varies greatly depending on the other energy consuming activities in the building. Although we typically reduced lighting energy in each building by 20% - 40%, this amounted to 5% - 36% of the building total use.

Because lighting in campus buildings is not separately metered, the reduction caused by our retrofit program was inferred from the drop in a building's total electrical use following completion of the lighting retrofit.

Figure 3 is based on one-year moving averages of a building's monthly total electrical use. The figures shows how we fit a straight line to the building's consumption for several years prior to the retrofit. The difference between this straight line projection and the actual consumption is taken to represent the energy savings attributable to the lighting retrofit.

OTHER ISSUES

Power Quality

Power quality describes distortions of voltage waveform and changes in the voltage-current phase relation. Table 4 presents measured values [Wolsey, 1995] that characterize the performance of some office equipment.

TABLE 4. POWER QUALITY CHARACTERISTICS

In Table 4, AP=active power, PF=power factor, CTHD=current total harmonic distortion, and THC=total harmonic current. THC is calculated from THC = {(CTHD/100)(AP)}/{(PF) (120)}, following [Wolsey, p.6]. The table presents measurements on specific equipment. Specifiers can obtain similar data from manufacturers or independent compilations [Ji, 1996].

The table shows that electronic ballasts do not necessarily produce more harmful harmonic current than do magnetic ballasts or common office equipment. Even if the measured T-8 system had CTHD=20%, it would then have THC= 0.10, still less than the T-12 system with efficient magnetic ballast. An electronic device affects power quality in a building in proportion to its share of the total building load. Concerns at a facility about the effects of a lighting retrofit on power quality and excess neutral wire currents should be evaluated by a qualified electrical engineer.

Electromagnetic Interference (EMI)

Under certain circumstances electronic devices emit conducted or radiated electromagnetic waves that can interfere with proper operation of theft detection systems, infrared controls, cordless telephones, radios, and power line carrier controls. Descriptions of these potential problems and possible solutions are discussed in [Buddenberg & Fowler, 1995]. The only problem we had on campus, after installing electronic ballasts throughout classroom and laboratory buildings, was interference with a library theft detection system. This was easily corrected by replacing ballasts near the detector with magnetic ballasts.

REFERENCES

Abesamis, R., Black, P., and Kessel, J., Field Experience with High-Frequency Ballasts, IEEE Transactions On Industry Applications, Vol. 26, No. 5, September/October 1990.

Buddenberg, A., Fowler, A., Electromagnetic Interference Involving Fluorescent Lighting Systems, Lighting Answers, Vol. 2, No. 1, March, 1995, Lighting Research Center, Troy, N.Y. 12180-3590

Eley Associates, Advanced Lighting Guidelines: 1993 (Revision 1), Electric Power Research Institute, Publication TR-101022s Rev.1, May 1993. For information on ordering this and other EPRI lighting publications, telephone 510-934-4212.

Ji, Y. , Power Quality, Lighting Answers, Vol. 2, No.2, February, 1995, Lighting Research Center, Troy, N.Y. 12180-3590.

Kessel, J., Performance of Retrofit Optical Reflectors, Strategic Planning for Energy and the Environment, Vol. 10, No. 2, Fall 1990.

Wolsey, R. , Power Quality, Lighting Answers, Vol. 2, No.2, February, 1995, Lighting Research Center, Troy, N.Y. 12180-3590.

FIGURES

FIGURE 1. CLONE FAMILIES AND THE FULLRATIO CALCULATION

FIGURE 2. LIGHTING RETROFITS VERSUS FULLRATIO

FIGURE 3. LIGHTING RETROFIT RESULTS: LATIMER HALL