Demonstration of Dimmable Electronic HID Lighting with Daylight Harvesting
inside a Chinook Air Hangar at a US Army Base in Fort Lewis, WA
Sr. Project Manager - Lighting
Electric Power Research Institute
942 Corridor Park Blvd.
Knoxville, TN 37932 USA
Abstract – Electronics are still making their way into lighting products to provide energy savings, higher efficiencies, and
intelligent lighting control. Although electronic lighting products can help reduce grid load, it is crucial for them to withstand
common everyday electrical disturbances such as voltage sags and surges. Previous EPRI research in the area of system
compatibility can be used to identify energy, emissions, and immunity performance and the “weakest links” in product design. A
project sponsored by the US Department of Energy (DOE) – Bonneville Power Administration (BPA) in collaboration with EPRI’s
Program 170 “End-Use Energy Efficiency and Demand Response in a Low-Carbon Future” included testing and demonstration of an
advanced lighting technology suitable for use in an aircraft hanger. A 575-watt electronic HID lighting product manufactured by
RomLight International Canada was selected for testing at EPRI and installation at the US Army Joint Base Lewis-McChord site in
Fort Lewis, Washington. Compatibility testing of the product revealed its energy and immunity performance and areas within the
ballast design that could be improved. Installation of this lighting technology to replace a 1,000-watt magnetic technology combined
with Delta’s lighting control technology in a three-bay air hangar revealed significant energy savings without compromising
illumination levels on the work surfaces inside the demonstration area. However, when combined with daylight harvesting, the
achievable energy savings, while considering the everyday business of the hangar, provided 40 % to 70 % times more savings. This
paper provides an overview of the concept of EPRI compatibility testing applied to the ballast, some of the benefits of this testing,
and an overview of the installation and measured energy savings at the hangar.
Index Terms—electronic HID lighting, electronic HID ballast, high-intensity discharge lighting, metal halide lamps, lighting
control, daylight harvesting, power quality, system compatibility, efficiency, efficacy.
The following terms may be new to the reader:
Electronic HID ballast – a ballast that is controlled by active electronic components to convert low-frequency AC energy
into a higher frequency energy in which impedance is provided by series capacitive or inductive reactance.
System compatibility – the ability of a device, equipment or system, generally a load, to function satisfactorily with respect
to its power-supply electrical environment without introducing intolerable electrical disturbances to anything in that
Like many other technologies, magnetic-based lighting technologies are slowly being replaced by electronic-based lighting
technologies. Electronic components used in ballast design can provide higher system and lamp efficiency, increase ballast
efficiency, and promote improved quality of light (illumination and color) and energy savings. The increasing use of lamp
dimming directly through the electronic ballast adds another dimension to energy savings, but also adds another layer of
susceptibility concerns to ballasts.
Today, this dimension is being expanded by the use of microprocessors in electronic ballasts. Microprocessors were first
introduced into ballast design a few years ago to provide advanced dimming control of lamps and control interfaces. Now, they
are being used to provide full ballast control including control for power factor correction. Application of microelectronics is
being expanded into the use of digital signal processors (DSP) devices for ballast control, and advanced dimming and lamp
control. They are also being used to provide proactive control of the lamp-ballast system to monitor internal lamp and ballast
conditions, provide wireless control, power-line carrier control, and decrease lamp dropout in the event of a power quality
event. If advanced ballasts using microelectronic devices are not properly protected from the common everyday electrical
environment, they will suffer frequent damage and early failure. Although utilization of microelectronics in ballast design is
the next major step in reaching new levels of lamp and ballast performance, this growing trend increases the susceptibility of
these designs to electrical disturbances. EPRI is conducting research in these areas to further the maturity of advanced ballast
technologies in the areas of system compatibility.
In addition to ensuring reliable ballast operation in common everyday electrical environments, ballast designers want to
know that the lamps driven by their ballasts are providing well-controlled illumination. EPRI compatibility testing can also
determine if disturbances incident upon the ballast cause any temporary or permanent lamp misoperation. Some voltage
anomalies may cause lamp instabilities to occur. Examples of instabilities include under driven lamps, over driven lamps,
oscillations or fluctuations in lamp power, and improper heating of filaments in fluorescent lamps. In HID ballasts, some
voltage anomalies may cause aberrations or changes in the lamp ignition and lamp regulation process. It is paramount that any
of these susceptibilities be determined prior to ballast production.
Energy research projects like those sponsored by BPA and EPRI involving testing and demonstration of electronic HID
ballast technologies are vital to educating utilities, building owners and operators, lighting designers, and end users trying to
achieve measurable energy savings. Testing and demonstration projects are especially valuable to these groups when using
electronic ballasts with intelligent lighting control technologies. Ballast manufacturers have already evidenced this trend in the
electronic fluorescent ballast industry, as microprocessor-based dimmable ballasts are now more frequently used with lighting
Lighting demonstration projects allow the installation of new advanced lighting technologies in utility and customer
facilities where their operation can be realized, witnessed, and verified. However, it is important to determine the performance
of the lighting product and lighting control system before large groups of products are installed. EPRI developed a concept in
the early 1990s called system compatibility, which is defined as the ability of a device, equipment or system, generally an
electronic load (e.g., an electronic ballast) to function satisfactorily with respect to its power-supply electrical environment
without introducing intolerable electrical disturbances to anything in that environment. This also means that the device (e.g.,
the lamp-ballast system) must be able to withstand common everyday electrical disturbances. These disturbances include those
created by the public power system and/or the electrical system in the customer’s facility. Thus, from the definition and its
extension above, one can see that this concept is a ‘two-way street’—the utility has a responsibility to provide power with as
few electrical disturbances as possible, and the manufacturer has a responsibility to design their product to withstand those
disturbances. The goal of the system compatibility concept is to determine what types of disturbances the device should be able
to withstand and to what level. The series of tests applied to the lighting product follow a well-written set of tests included in
an EPRI test protocol for applying each type of disturbance to the lighting device under test.
For electronic lighting systems, EPRI developed a series of test protocols—one of which is for electronic fluorescent
systems, one for electronic HID systems, and others for different types of lighting products. Because this demonstration project
involved the use of electronic HID ballasts, the SC-415 – Test Protocol for System Compatibility: Electronic HID Ballasts
Used in Indoor and Outdoor HID Lighting Systems was used to determine the performance of the electronic HID product
under demonstration. Details regarding the EPRI tests are not included here, but are provided in the paper, Demonstration of
Dimmable Electronic HID Lighting with Daylight Harvesting inside a Chinook Air Hangar at a US Army Base in Fort Lewis,
WA, published at the Energy Efficiency Expo 2010, October 19, 2010 in Dallas, Texas. However, a summary of the test results
are included below.
III. AN OVERVIEW OF SOME EPRI LABORATORY TEST RESULTS
EPRI’s full battery of system compatibility tests were applied to several selected models of RomLight electronic HID
ballasts—250-, 320- and 400-watt metal halide and high-pressure sodium ballasts in a previous project. The results of that
project revealed several improvements in the ballast design, which were integrated into the design by the ballast engineers at
RomLight. Because their 575-watt metal halide electronic HID ballast product is similar to their 400-watt product, many of the
ballast design improvements were also applicable to the 575-watt product. The 575-watt product was selected as the
demonstration ballast for the one of the customer sites for the BPA project. The original ballast technology at the site was a
1,000-watt metal halide lamps driven by magnetic ballasts with aluminum reflectors and a glass reflector cover. This product
was subjected to Tests 0 through 5 in the Energy Performance Test suite (see Table I on page 2) to determine the energy
efficiency performance of the lamp-ballast system (and fixture) under full lamp brightness and lamp dimming as the product
would be used in an aircraft-related military application on a major US Army Base. The results of these tests are described
A. Efficacies and Efficiency – Full Brightness Conditions
Table I provides the ballast efficiency, lamp efficacy, lamp-ballast efficacy, and input and output power performance for the
575-watt electronic HID ballast used in this demonstration project. A 575-watt metal halide lamp designed for high-frequency
operation with electronic HID ballasts was driven at full brightness at a single input voltage of 277 volts AC (60 hertz) applied
to the ballast. (The voltage of 277 volts AC is the preferred line voltage for operating electronic ballasts above 400 watts. This
minimizes line current flow through the input circuit components (e.g., EMI filter and bridge rectifier, etc.) and helps with
thermal management.) From Table I, one can see that the efficacies and efficiencies are above 90 %, which far exceeds the 88
% minimum limit for metal halide luminaires as prescribed by the Title 20 of the California Energy Commission (CEC) for all
line voltages within the universal range. Efficiency and efficacy values are likely to move in the upwards direction as the
electronic HID industry becomes more mature with digital control and as more electronic HID ballasts become available on the
BALLAST EFFICIENCY, LAMP EFFICACY, AND BALLAST INPUT AND OUTPUT POWER
PERFORMANCE AT A 277-VOLT NOMINAL INPUT VOLTAGE AT FULL LAMP BRIGHTNESS
Input Voltage Input Power Output Power Ballast Losses
(Vrms) (Watts) (Watts) (Watts)
276.8 611.3 593.1 18.2
Input Voltage Light Output Lamp Efficacy Lamp-Ballast Efficacy Ballast Efficiency
(Vrms) (Lumens) (LO/Plamp) (LO/Pin) (Pout\ Pin)
276.8 67,410 110.3 113.7 97.2
B. Efficacies and Efficiency – Dimming Conditions
Table II provides the ballast efficiency, lamp efficacy, lamp-ballast efficacy, and input and output power performance for
the 575-watt electronic HID ballast used in this demonstration project with the lamp at various dimming levels. From Table II,
one can see that both the ballast efficiency maintains a high level at 96 %, but the system (lamp and ballast) efficacy degrades
between dimming voltage levels of 4 and 6 volts DC down to very low values at minimum brightness. As mentioned earlier,
determining energy performance metrics with a dimmable lighting product is vital to industries and facilities planning to use
dimmable lighting in applications where varying illumination is required and in applications where demand response activities
will be used.
Dimming from full lamp brightness at 67,410 lumens requiring 611.3 watts of input power to the ballast to 57,990 lumens
(12.8 % dimmed at a 2-volt DC dimming level) at 533 watts of input power provides a savings of 78.3 watts per fixture with a
lumen reduction of 13.9 %. This drop in luminous flux is typically not noticeable at ceiling heights characteristic of high-bay
ceilings. The same is essentially true for changing the dimming level from 2 volts to 3 or 4 volts DC (conditions for the 4-volt
DC dimming level are shown in Table II). The reduction in input power without substantial reduction in illumination is ideal
for demand response systems of the future. This is especially of benefit in facilities where daylight harvesting can enable
customers to shave lighting load during certain periods of daytime work activity throughout a 12-month period.
A reduction of less than 10,000 lumens is achievable without compromising significant lamp-ballast efficacy while still
providing adequate illumination. This means that the lamp is still operating in an acceptable efficiency region not dipping
below 88 %. The use of high-frequency lamp current improves the ability of these lamps to operate at higher efficacies at
dimming levels below full brightness. Interestingly, at 30 % lamp dimming, the ballast efficiency is about 97 % at an
illumination of nearly 48,000 lumens. This provides a power savings of 132.9 watts, or 22 % above the ballast input power at
full brightness. This is one reason why lighting control combined with dimmable electronic HID ballasts make good energy
An additional benefit of good lighting design using electronic HID ballasts with high-efficiency reflectors is the improved
uniformity in illumination across a floor or long horizontal work surface like those found in manufacturing plants, open-area
large commercial buildings, and military installations. When illumination is more uniform, reducing light output through lamp
dimming will reduce the likelihood of a noticeable change in illumination. The improvement in uniformity is provided by an
improvement in matching the HID lamp to the proper high-efficiency reflector, which provides a more uniform distribution of
light on work surfaces.
In this project, dimmable HID lighting combined with daylight harvesting and use of daylight sensors were used to control
a total of 24 fixtures in two of three work bays in a large open-area facility. At 30 % lamp dimming, this equates to 3,189.6
watts. However, this saving is only the dimming-related savings. The actual savings over the original 1,000-watt magnetic HID
fixtures for 24 fixtures equates to a total of 12,525 watts (a savings of 521.9 watts per fixture). For all 48 fixtures, this equates
to 25,051 watts for the entire hangar. Maintaining full lamp power when sufficient illumination from daylight exists in the bay
work areas wastes electrical energy at the facility. This is a common problem in daylight-assisted facilities where magnetic
HID technologies are used. Demonstrating dimmable electronic HID technologies with daylight-equipped lighting controls in
these facilities will help convince the larger customer base that this approach to energy savings has merit.
BALLAST EFFICIENCY, LAMP EFFICACY, AND BALLAST INPUT AND OUTPUT POWER PERFORMANCE AT 277 VOLTS
Dimming Level Input Power Output Power Percent Dimmed
(Vdc) (Wrms) (Watts) (%)
0 320.6 306.9 47.6 96.0
2 375.2 352.5 38.6 96.6
4 426.6 412.5 30.2 96.8
6 478.4 463.1 21.7 96.8
8 533.0 516.0 12.8 96.8
10 611.3 593.1 0.0 97.2
Dimming Level Light Output Lamp-Ballast Efficacy
(Vdc) (Lumens) (LO/Pin)
0 8,251 25.7
2 11,901 31.7
4 15,377 36.0
6 47,966 100.3
8 57,990 108.8
10 67,410 110.3
C. Total Harmonic Distortion of the Input Current and True Power Factor
Table III provides some of the data resulting from Test 8. From Table III, one can see that the total harmonic distortion or
the input current to the ballast across the dimming range varies from 7.3 % at full brightness to 12.9 % at minimum brightness.
The true power factor for this dimming range varies from 0.99 to 0.97. According to ANSI C82.77 and EPRI’s research on
non-linear loads and harmonics, this ITHD and power factor performance is acceptable for use as an electronic lighting load in
commercial and industrial facilities powered by public power systems. This indicates that dimming these systems for the future
use of demand response will not introduce high levels of ITHD with poor power factor. Presently, no ITHD and power factor
limits exist specifically for electronic HID ballasts. If limits were set in the future standards, these ITHD would be under those
limits. Ballast design efforts should not further reduce ITHD values as the payback for these design and performance efforts
would be little. This information may serve as useful data points for helping to develop future standards in this area.
TOTAL HARMONIC DISTORTION OF THE INPUT CURRENT AND TRUE POWER FACTOR
Input Current Total
Dimming Level (Vdc) True Power Factor
0 1.19 0.97 12.9
2 1.39 0.98 11.8
4 1.57 0.98 10.1
6 1.75 0.99 9.4
8 1.94 0.99 8.4
10 2.23 0.99 7.3
IV. THE FORT LEWIS DEMONSTRATION SITE
Figure 1 shows a wide-angle aerial view of the US Army Joint Base Lewis-McChord facility captured from Google. One
can see from Figure 4 that the base contains many facilities which are adjacent to some of the airstrips. The yellow arrow
points out Building #03273, which is the Aviation Support Facility (ASF) for the Army Reserve.
Figure 2 illustrates a closer aerial view of Building #03273 where the demonstration of the new fixtures fitted with high-
efficiency reflectors and ballasted with 575-watt electronic HID ballasts took place. The area inside the red box in Figure 2
illustrates the ceiling/floor area of Building #03273 where the new lighting was installed.
Fig. 1. Long-Range View of US Army Joint Base Lewis-McChord
In Figure 2, one can also see seven Chinook helicopters on the air pad outside the facility. Military personnel and civilians
work in this hangar on Chinook helicopters to provide scheduled and emergency repairs and regular maintenance and to
prepare these aircraft for missions carried out by Army reservists in Iraq and Afghanistan. The majority of the work is
conducted from 8:00 am to 5:00 pm. However, in the event of unscheduled or emergency repairs or maintenance, work may be
performed at any hour or on any day of the week to ensure aircraft are ready to fly.
Figure 3 illustrates the condition of the present lighting system in the Ground Shop of the air hangar prior to the lighting
demonstration. The existing lighting system is a high-intensity discharge (HID) system operated by magnetic ballasts driving
1,000-watt, metal halide lamps.
From Figure 3, one can see several fixtures with failed lamps. Approximately 17 of 48 fixtures were not illuminated at the
time EPRI arrived to the site. The poor lumen depreciation using this lamp technology combined with the failing lamps had a
significant impact on the lighting levels inside the facility. Discussion with some of the military and civilian workers inside the
hangar verified these poor lighting conditions.
All failed 1,000-watt metal halide lamps operated from the magnetic ballasts were replaced at the time the new 575-watt
electronic HID ballasts were installed to bring the light levels back up in areas where no new fixtures were planned.
Illumination ranged from 70 to 105 foot-candles at the floor level inside the hangar work bays under the 1,000-watt magnetic
HID-ballasted fixtures with all lamps operational. At the same points under the new 575-watt electronic HID-ballasted fixtures
with the faceted reflectors and same fixture heights, the improved illumination ranged from 97 to 123 foot-candles at full lamp
Fig. 2. Aerial View of the Air Hangar (Building #03273)
Fig. 3. Photograph of the Ground Shop Prior to the Lighting Demonstration
Figure 4 illustrates another photograph of the air hangar. From Figure 4, one can see several of the 1,000-watt HID fixture
failures over the middle work bay. The fixtures in the far work bay are the new electronic HID-ballasted fixtures. Daylighting
is provided by the large clerestory windows. The clerestory window designs are used on both sides of the facility at the curved
roofline. In fact, without these clerestory windows, the light levels in the facility would be so low when using aged metal
halide lamps powered by magnetic HID ballasts that there would not be enough light to carryout the work on the Chinook
helicopters. In fact, more frequent relamping of the 1,000-watt lamps would be necessary if the clerestory windows were not
part of the building design. To provide additional light at the work site on the helicopters, the workers would need to use more
high-powered droplights. Managing a larger number of droplights and cords would congest the work area.
An additional problem caused by lamps with low illumination and lamps that have failed is the difficulty involved in
changing these lamps. Lamps operated on these magnetic systems experience early lumen depreciation resulting in frequent
Fig. 4. View of One-Half of One Clerestory Window with a Chinook Helicopter in the Middle Work Bay
The highest ceiling point inside the hangar reaches about 75 feet above the floor (at the center of the middle work bay).
Moreover, using a 60-foot hydraulic lift to reach the fixtures in the center of the hangar requires careful manipulation around
the three bay work areas. From Figure 4, one can see that with the custom-engineered scaffolding placed around the Chinook
helicopter during preventive maintenance and repair, it is near impossible, unsafe, and against policy to allow a high-lift over
the top of the helicopter work area when personnel are present in the work area. In some cases, work on helicopters must be
stopped to move them out of the hangar to provide lift access to fixtures directly above the work areas. Utilization of electronic
HID ballasts and metal halide lamps designed for high-frequency lamp drive significantly reduces these problems through
extension of lamp life and reduced lumen depreciation.
V. THE INSTALLATION
Figure 5 illustrates the fixture and lighting control circuit plan for installing 24 electronic HID fixtures out of 48 fixtures
locations inside the air hangar. This configuration was required because of the original arrangement of the fixtures on each
lighting branch circuit—four fixtures per circuit with 12 circuits inside the facility. This plan for installing 24 fixtures equates
to six groups of fixtures with four fixtures per group (or circuit).
The two redlined regions represent the division made in programming the lighting controller. Set 1 is the West set, and Set
2 is the East set. There is one daylighting sensor in each set, denoted by the orange dot. Group 1 in the West set and Group 2 in
the East set are monitored for power, energy, and power quality. The control circuit for measurement reference is Group 7
(Circuit #9-Phase C) with 1,000-watt magnetic HID fixtures left in place. The two circuits that are monitored for power
performance are Group 1 (Circuit #13-Phase A) and Group 6 (Circuit #12 on Phase B).
Power-Line Location for Daylight
Electrical Panel Providing Power to HID Light Existing (Magnetic)
Location for Lighting New (Electronic) Dimmable
Upper Wall Clerestory Window Panels
Group 5 (Ckt#17) Group 1 (Ckt# 13 (A))
Group 2 (Ckt# 16) Set 1
-ft Group 7 (Ckt# 9 (C)) Group 3 (Ckt# 11)
Group 6 (Ckt# 12 (B)) Group 4 (Ckt# 8)
Upper Wall Clerestory Window Panels
Fig. 5. Plan for Fixtures and Lighting Control Circuits
VI. EXAMPLE OF DIMMABLE ELECTRONIC HID BALLASTED FIXTURE INSTALLED AT SITE
Figure 6 illustrates one of the new electronic HID fixtures hung from the ceiling of the air hangar with the power to this
circuit turned ‘on’. Notice that the plate above the reflector helps to reflect light and heat downward instead of up on the
ceiling. Heat from the lamp reflected away from the ballast helps to reduce heat flow to the ballast. An additional measure,
which can be observed in Figure 10 as well, is the extruded design of the aluminum ballast housing. Extruded housings
improve thermal management of heat generated inside the ballast and improve heat flow to the outside ambient environment.
The extrusions provide additional surface area on the outside of the housing. They also help reduce the cost of ballast
VII. ENERGY SAVINGS DATA
During a demonstration project, it is important to monitor the energy and power usage as well as the voltage quality on both
the existing low-efficiency lighting products (typically a magnetically-ballasted product) and the new higher-efficiency
products installed during the project. Having tested the product in the EPRI Lighting Laboratory prior to demonstration helps
the manufacturer understand its performance, weakest links, and areas where circuit improvements can be made.
Fig. 6. Photograph of One of the New Electronic HID Fixtures (Power to this Circuit On)
This testing also provides the right type of baseline data necessary to verify and compare energy performance of the product
in the field as well as verification that the product can withstand the everyday electrical environment. Presenting laboratory test
data and field data to potential customers goes a long way in helping them to understand how the product will perform in their
field environment. The customer realizes that the performance of the product not only depends upon the product’s ability to
withstand electrical events generated by everyday utility operations but also those that occur inside the customer’s facility from
exercising loads during normal facility operation.
During the demonstration, voltage and current probes from the monitor are connected to the branch circuits that provide
power to part of the existing (reference) lighting products and also to the circuits that provide power to the new electronic HID
fixtures. Through communication links using a cellular modem, EPRI is able to download energy, power and power quality
data from the monitor to the EPRI Power Quality Monitoring Data Center in Knoxville, TN. The data is processed and made
available for observation on any computer connected to the Internet. Having access to energy savings data is a must when
trying to determine how much savings is provide when the new products are first installed. As lamps age, lumen depreciation
effects will become more detectable through measurement of illumination. Ballast input power may also be measured with the
power quality monitor as the lamp-ballast system ages.
Power quality monitoring data can also demonstrate that any lighting control product (including an occupancy sensor or
daylight harvesting system) or demand response product is functioning properly and causes an actual change in lighting load.
The degree of change data in lighting load when using lighting controls is used to verify energy savings performance for utility
rebate or incentive programs. Continuous capture of energy and power data is necessary in order to determine the maximum,
average, and minimal energy savings for a varying lighting load. With daylighting systems, variations in load will occur as
outdoor light levels vary. With these systems, daylight may be used to offset electric illumination, as is the case at the Fort
Lewis and potentially many other military and aircraft facilities where daylighting could used to provide natural light to indoor
and even outdoor areas.
A. Winter-Spring Data: 2010
Figure 7 illustrates the power profile for Phase A (Circuit #13) where one group of new electronic dimmable HID ballasts are
installed. This represents the power profile for four fixtures under active dimming conditions with the day light harvesting
system active. From this compressed view, one can see that not all periods (five working days – Monday through Friday) have
the same power profile. This is because of the difficulty in reading day profiles over a 10-day period with overcast during the
last week of January and the first few weeks of February. One will also notice that as spring approaches, the power reduction
gets wider and deeper (more white areas show up in each day power cycle). The average peak power in this graph is
approximately 2,400 watts, which represents four fixtures (2,400 / 4 = 600; representing a 575-watt fixture (by design) with
approximately 25 watts loss by graphical analysis).
Fig. 7. Phase A (New Circuit #13), January 22nd through April 7th, 2010
Figure 8 illustrates the power profile for Phase B (Circuit #12) where the new electronic dimmable HID ballasts are also
installed. This represents the power profile for four fixtures under dimming conditions with the daylight harvesting system also
active. From this compressed view, one can see that not all periods (five working days – Monday through Friday) have the
same power profile. This is because of the overcast during the last week of January and the first few weeks of February. One
will also notice that again as spring approaches, the power reduction gets deeper. Additionally, one will notice that the power
reduction from lamp dimming (through day lighting) is again wider and deeper. Group 6 (Circuit #12) is part of the center row
of fixtures nearer to the middle of the hangar where more day lighting is available. (The clerestory windows near the center of
the hangar are taller, and thus more day light enters the facility in this area.) The average peak power in this graph is
approximately 2,400 watts, which represents four fixtures (approximately 2,400 / 4 = 600; representing a 575-watt fixture (by
design) with approximately 25 watts loss by graphical analysis).
Fig. 8. Phase B (New Circuit #12), January 22nd through April 7th, 2010
Figure 9 illustrates the power profile for Phase C (Circuit #9) where the existing magnetic HID ballasts are installed. From
the graph, one will notice that there are no reductions in power during these periods. In fact, there are increases in power
representing higher power during lamp turn-on. The average peak power here is approximately 3,800 watts (3,800 / 4 = 950
watts per fixture with approximately 100 watts losses). The aged lamps removed from the facility also showed signs of
significant aging; thus, light reduction from poor lumen maintenance.
Fig. 9. Phase C (New Circuit #9, January 22nd through April 7th, 2010
Figure 10 illustrates five, one-day cycles for the week beginning Monday, January 22nd through Friday, January 26th on
Phase A for the electronic HID ballasts. There are several interesting points about these cycles:
• Power reduction due to lamp dimming is evident in every cycle.
• No dimming exists at the beginning and ending of each day cycle (as expected), although it can occur on a clear day with
early and strong day lighting conditions.
• Maximum power reduction due to lamp dimming only occurs once during each day.
• Maximum power reduction due to lamp dimming is closets to the center of the workday when the sun is brightest (even
given the average cloud cover each day).
• Minimum circuit power (first one day cycle) reaches 1.4 kW (1.4 / 4 = 350 watts) indicating that the four fixtures (on
average) never reached minimum brightness (575 / 2 [for 50 % dimming] = 287.5 watts) for this weeks cycle.
Fig. 10. Phase A (New Circuit #13), January 22nd through 26th, 2010
Figure 11 illustrates five, one-day cycles for the week beginning Monday, February 1st through Friday, February 5th on
Phase B for the electronic HID ballasts. There are several interesting points about these cycles:
• Power reduction due to lamp dimming is evident in each cycle.
• Enough sunlight entered the clerestory windows at the end of Day 2 and Day 5 to allow some lamp dimming even at the
end of the day when the fixtures were manually turned off.
• Power reduction was deepest and widest during Day 2 when it did reach a minimum of approximately 480 watts for four
fixtures (480 / 4 = 208 watts). 208-watts (34 % dimming) is close to the minimum (180 watts, 30 %); there is some variation
among fixtures (expected variation is 5 %) in minimum dimming.
• This five-day period contains more power reduction cycles (lamp dimming cycles) because of more broken cloud cover
passing over the facility during the week, especially on Day 5.
Additional dimming data will be included in the addendum to this report at the one-year anniversary date of the monitoring
period for this site.
Fig. 11. Phase B (New Circuit #12), February 1st through 5th, 2010
Figure 12 illustrates the power profile on Phase A (Circuit #13) for a one-day cycle that powers four new dimmable
electronic HID fixtures at a ballast design lamp power of 575 watts each. The total non-dimmed load is approximately 2,400
watts total: (2,400 / 4 = 600; representing a 575-watt fixture (by design) with approximately 25 watts loss by graphical
analysis). From the graph, one can denote the following:
• Lamp dimming begins at approximately Point A around 8:35 am with a small drop in lamp power.
• Little change in lamp power continued until about 9:45 am (Point B) when significant day light began to enter the facility
through the large clerestory windows.
• Lamp dimming ended at Point C when day light was reduced by passing clouds over the facility, and lamp power
increased briefly to maintain constant illumination in the facility.
• When day light began to increase again (between Point C and D), lamp dimming continued until Point D when it
remained constant until a second brief cloud cover passed (hump between Point D and E) over the facility.
• At Point E, day light in the facility began to decrease until Point F. Shortly after Point F, the lighting system was turned
off at the end of the workday.
• The ramping up and down of the lamp power is also shown to be gradual (evidenced by gradual changes in lamp power)
as expected; thus not causing sudden changes in light output that end users might notice.
Fig. 12. Phase A (New Circuit #13), February 2nd, 2010
Figure 13 illustrates the power profile for Phase C (Circuit #9) for a five-day cycle during a work week where the existing
magnetic HID ballasts are installed. From the graph, one will notice that there are no reductions in power during these periods
that are due to lamp dimming. In fact, there are increases in power representing higher power during lamp turn-on. The
average peak power here is approximately 3,800 watts (3,800 / 4 = 950 watts per fixture). One interesting point can be made
• Peak power occurred at the beginning of each day cycle except for Day 1.
• Lamp power settled after the initial start up.
Fig. 13. Phase C (New Circuit #9, January 25th through January 29th, 2010
B. Excerpt from Summer Data: 2010
Figure 14 illustrates a seven-day period from August 1st through 8th, 2010 on Phase A (Circuit #13) powering electronic
HID ballasts. From the graph, one can see that the daily power waveform is significantly different than the ones shown in
Figures 10 and 11 recorded during the winter/spring months of 2010. In Figure 14, one can see that the daylighting occurs
much sooner during the day evidenced by the earlier decrease in ballast input power.
Fig. 14. Phase A (New Circuit #13), August 1st through 8th, 2010
Figure 15 illustrates a similar power waveform (scales are different) for Phase B (Circuit #12) also powering electronic HID
ballasts. This waveform has similar characteristics.
Fig. 15. Phase B (New Circuit #12), August 1st through 8th, 2010
Figure 16 illustrates a one-day power cycle for August 3rd for the electronic HID ballasts on Phase A (Circuit #13). From
the graph, one can also see that the characteristics are again significantly different from those shown in Figure 12 captured
during the winter/spring period of 2010. This proves the values of daylight harvesting using active lighting control systems
with photo sensors in facilities where daylight is available to contribute to the needs of indoor illumination.
Fig. 16. Phase A (New Circuit #13), One-Day Cycle on August 3rd, 2010
VIII. THREE TIERS OF ENERGY SAVINGS
Figure 17 illustrates the three tiers of energy savings that were achieved at the aircraft hangar. First, the 1,000-watt,
magnetically-based, metal halide fixtures were replaced with 575-watt, electronically-based, dimmable, metal halide fixtures
for a first-tier savings of 425 watts per fixture. BPA and EPRI measurements both show that on the average the illuminance is
either at least as good or better in all measured locations on the hangar floor. The use of the faceted reflectors has improved the
fixture efficacy and distribution of light output such that less light is wasted on the walls and more light is provided for
horizontal and vertical workspaces within the working area of the hangar. The second tier of savings is provided by the
increased ballast efficiency (65 % for the magnetic and 96 % for the electronic) at a savings of 60 watts per fixture. The third
tier of savings is provided by the dimmable lighting ballast combined with the day-light-sensor controlled lighting control
system for a savings of up to 295 watts per fixture, depending upon daylighting conditions. The numbers at the bottom of the
figure below represent the kWh savings for three conditions: no dimming, average dimming, and maximum dimming for the
standard hours of operation inside the hangar.
Fig. 17. Three-Tiers of Energy Savings
CHARACTERISTICS OF THE DEMONSTRATION INSTALLATION
BPA Demonstration Site - US Army Joint Base Lewis-McChord – Building #03273 Army Reserve Air Support Facility (Chinook Hangar); Fort Lewis,
Twisted-pair (0 to 10 volt DC) Electronic Dimmable High-Intensity Discharge (HID)
Type of System Installed
Installed one-lamp, 575-watt, high-efficiency dimmable ballasts in self-contained
Description of Dimmable Lighting Technology & Fixture fixtures manufactured by RomLight International Canada with high-efficiency faceted
Open protocol, manufactured by Delta Controls, 0 to 10 volt analog DC dimming
Description of Lighting Control System system with one Ethernet-connected touch-screen lighting control panel and two wired
day light harvesting sensors
Static energy savings of 425 watts (1,000-watt magnetic HID ballast replaced with
575-watt high-efficiency (96 %) electronic dimmable ballast) with 60 watts of reduced
Energy Savings ballast losses per fixture; dynamic energy savings realized from utilization of day light
harvesting; energy savings 40 to 70 % depending upon amount of available day light
through clerestory windows
Illumination on floor with 1,000-watt magnetically-ballasted fixtures: 70 to 100 foot-
Illumination on floor with 575-watt electronically-ballasted fixtures with high-
efficiency reflectors: 97 to 123 foot-candles
Monitoring two dimming circuits with one control circuit; monitor communicating
Verification Monitoring with EPRI Power-Line Monitor Center in Knoxville, Tennessee via cellular modem
and downloading of power usage and power quality data daily
System Status Complete, data being collected by monitor and downloaded by EPRI
Energy savings opportunities can be realized from implementing a number of electronic lighting technologies. Lumen
packages provided by the use of high-efficiency HID lamps in high-efficiency fixtures are improving the overall performance
(e.g., efficiency, color, illumination, and power quality) for spaces where HID technologies are more suitable. Dimmable
technologies offer the additional savings over that offered by ballast efficiency, lamp efficacy, and fixture efficacy
improvements, with the opportunity of providing the customer and utility with active control over the lighting load and
management of space illumination. The data presented in this paper clearly illustrates the benefits of lamp dimming with
daylight harvesting to enable sunlight as an energy saving mechanism for lighting the aircraft hangar. Facilities similar to the
Chinook hangar are prime spaces for dimming and daylighting applications with dimmable lighting technologies. Verification
of energy savings can be accomplished through utilization of smart monitoring (e.g., data downloading through cellular
modem) without penetrating the customers hard-wired network.
Table IV lists seven characteristics of the demonstration project which provides a summary of what was installed and a
description of the electronic HID ballast and lighting control technology. Of particular importance are the energy savings
improvements and the levels of illumination on the floor and on the work surfaces inside the Fort Lewis Chinook hangar have
been improved without compromising occupant satisfaction.
EPRI and RomLight International Canada gratefully acknowledge the sponsorship of this research support from Bonneville
Power Administration in Portland, Oregon, Sylvania Lighting Services, and support staff from the US Lewis-McChord Army
Base in Fort Lewis, Washington.
The following provides a bibliography related to various subtopics related to efficiency, technology development,
performance improvement, and failure analysis of electronic HID lighting technologies and applications.
 Assessment of Electrical, Efficiency, and Photometric Performance of Advanced Lighting Technologies: Dimmable Advanced Lighting Technologies –
Electronic High-Intensity Discharge Ballasts. EPRI, Palo Alto, CA: 2008. 1018479.
 Electronic High-Intensity Discharge Lighting. EPRI, Palo Alto: 2007. 1016200.
 Failure Analysis of a Flexible Drop-Cord Lighting Cable Used with Electronic HID Ballasts in a Warehouse Merchandiser. EPRI, Palo Alto, CA: 2005.
 Keebler, Philip F., “New Electronic HID Ballast Technologies to Reduce Power Quality Problems in Commercial and Industrial Environments” EPRI
Power Quality Applications Conference, Vancouver, 2004.
 On-Site Power Quality and Wiring and Grounding Investigation at Dickinson College: Electronic HID Ballast Failures. EPRI, Palo Alto, CA: 2007.
 Phipps, Kermit O., Keebler, Philip F., and Nastasi, Doni., “Distinguishing between surge- and temporary overvoltage-related failures of metal oxide
varistors in end-use equipment designs” Interference Technology EMC Directory & Design Guide, 2006.
 Power Quality Study and Failure Analysis Report of Failed Electronic HID Lighting Ballasts in a Rubber Roofing Materials Plant. EPRI, Palo Alto, CA:
 Power Quality Study and Forensic Analysis Report of Failed Electronic HID Ballasts in a US Navy Warehouse. EPRI, Palo Alto, CA: 2007.
 Shedding More Light on Surge Protection for Electronic Ballasts. EPRI, Palo Alto, CA: 2008.