ArticlePDF Available

Energy Saving Lighting Control Systems for Open-Plan Offices: A Field Study

Authors:

Abstract and Figures

This field study in a deep-plan office building assessed energy savings from the use of luminaires using integral occupancy sensors, light sensors (daylight harvesting), and individual dimming control adjusted by office occupants via their computer screens. Data were collected from 86 workstations over a year to examine the energy savings and power reduction attributable to the controls, and how the controls were used. Occupants were encouraged to use the individual lighting control feature by means of e-mail reminders. Energy savings and peak power reductions were determined by comparison to a conventional fluorescent lighting system installed on a neighbouring floor. Dans le cadre de cette étude sur le terrain menée dans un immeuble de bureaux à plan profond, on a évalué les économies d'énergie réalisées lorsqu'on utilisait des appareils d'éclairage fonctionnant avec des capteurs de présence intégrés, des capteurs de lumière (mise en valeur de la lumière naturelle) et des gradateurs à commande individuelle, réglés par les occupants eux-mêmes via leurs écrans d'ordinateur. Des données ont été recueillies à partir de 86 postes de travail sur une période d'un (1) an, afin de permettre d'examiner quelle était la part des économies d'énergie et des réductions de puissance qui était attribuable à ces commandes, et de quelle façon l'on employait celles-ci. On a encouragé les occupants à utiliser les commandes d'éclairage individuelles en leur envoyant des aide-mémoire par courrier électronique. Les économies d'énergie ainsi que les réductions de puissance en périodes de demande de pointe ont été déterminées par comparaison des systèmes étudiés avec un système d'éclairage à appareils fluorescents de type traditionnel. RES
Content may be subject to copyright.
htt
p
://irc.nrc-cnrc.
g
c.ca
Energy saving lighting control systems for
open-plan offices: a field study
NRCC-49498
Galasiu, A.D.; Newsham, G.R.; Suvagau, C.;
Sander, D.M.
A version of this document is published in / Une version de ce document se trouve dans:
Leukos, v. 4, no. 1, July 2007, pp. 7-29
The material in this document is covered by the provisions of the Copyright Act, by Canadian laws, policies, regulations and international
agreements. Such provisions serve to identify the information source and, in specific instances, to prohibit reproduction of materials without
written permission. For more information visit http://laws.justice.gc.ca/en/showtdm/cs/C-42
Les renseignements dans ce document sont protégés par la Loi sur le droit d'auteur, par les lois, les politiques et les règlements du Canada et
des accords internationaux. Ces dispositions permettent d'identifier la source de l'information et, dans certains cas, d'interdire la copie de
documents sans permission écrite. Pour obtenir de plus amples renseignements :
http://lois.justice.gc.ca/fr/showtdm/cs/C-42
ENERGY SAVING LIGHTING CONTROL SYSTEMS FOR OPEN-PLAN OFFICES:
A FIELD STUDY
Anca D. Galasiu, Guy R. Newsham, Cristian Suvagau, Daniel M. Sander
Anca D. Galasiu (corresponding author)
Tel: +1 (613) 993-9670
E-mail: anca.galasiu@nrc-cnrc.gc.ca
National Research Council Canada
Institute for Research in Construction
Indoor Environment Program
Building M-24, 1200 Montreal Road
Ottawa, ON, Canada, K1A 0R6
Guy R. Newsham
National Research Council Canada
Institute for Research in Construction
Indoor Environment Program
Building M-24, 1200 Montreal Road
Ottawa, ON, Canada, K1A 0R6
Cristian Suvagau
BC Hydro
Technology Solutions Power Smart
900-4555 Kingsway
Burnaby, BC, Canada V5H 4T8
Daniel M. Sander
National Research Council Canada (retired)
Abstract - We conducted a field study in a deep-plan office building equipped with
suspended direct-indirect luminaires located centrally in cubicle workstations. In
order to reduce lighting energy use, the luminaires employed integral occupancy
sensors and light sensors (daylight harvesting), as well as individual dimming control
accessed through occupants’ computer screens. Data were collected from 86
workstations over a year to examine the energy savings and power reduction
attributable to the controls, and how the controls were used. An awareness
campaign that used e-mail reminders to encourage the occupants to use the
individual control feature of the lighting system was also conducted. Results indicate
that the lighting system generated substantial energy savings and peak power
reductions compared to a conventional fluorescent lighting system installed on a
neighboring floor. The installed lighting power was 42% lower than that of the
conventional system. The three controls combined saved 42 to 47% in lighting
energy use compared to the same lights used at full power during work-hours; this
2
translated into overall savings of 67 to 69% compared to the conventional lighting
system. If the three lighting controls systems had been installed separately,
occupancy sensors would have saved, on average, 35% if used alone, light sensors
(daylight harvesting) 20%, and individual dimming 11%. The light sensor savings
were, as expected, higher in perimeter workstations, and would have matched the
performance of the occupancy sensors with some modifications to the control
parameters. The average daily peak power demand for lighting was also reduced by
a similar amount, which resulted in an average effective lighting power density of
only 3 W/m2. Although not detailed in this paper, surveys indicated that the studied
lighting system was also associated with higher occupant satisfaction. This was
likely due to the individual dimming control, although use of this control beyond an
initial preferred setting was rare.
Keywords - lighting, daylighting, lighting controls, occupancy sensors, individual
controls, lighting systems, daylight-linked dimming, energy savings, open-plan
offices
1 INTRODUCTION
As part of the effort towards sustainability buildings need to use less energy. In
2004, Canadian office buildings accounted for 33% of the total energy used by the
commercial/institutional sector, with lighting accounting for 10% of the total building
energy use, and 24% of the electricity use (NRCan 2006). Several research studies
have generated promising results suggesting that electrical energy use can be
substantially reduced by using lighting control systems such as daylight-linked
dimming and occupancy sensors (Maniccia and others 1999; Jennings and others
2000; Lee and Selkowitz 2006). Individual (personal) dimming controls have also
been shown to reduce energy use, while increasing occupant satisfaction (Boyce
and others 2003; Newsham and others 2004).
Despite the fact that various energy saving technologies have been available for
some time, their implementation continues to be very slow. This is not surprising,
however, given the scarcity of long-term performance assessments demonstrating
that these systems do work as asserted and justify their higher initial cost. Many
earlier investigations either took place in laboratory settings, or reported failures in
attaining the projected energy savings, revealing significant problems with
commissioning and user acceptability (Bordass and others 1994; Love 1995; Slater
1995, 1996; Heschong Mahone Group 2006). Even fewer studies have surveyed
concurrently the opinions and preferences of the users of these systems. A review of
the scientific literature to date showed that there is almost no information available
on the long-lasting success of energy-saving lighting control technologies when used
in combination in real buildings. This study was designed to partially remedy this gap
and to generate information that could improve the uptake of such lighting controls in
buildings.
3
The study took place in an open-plan office building featuring a lighting control
system equipped with occupancy sensors, daylight-linked dimming, and individual
dimming control accessed through occupants’ computer screens. It included the
monitoring of the energy use of the lighting system over the course of a year, along
with an evaluation of the occupants’ satisfaction with the lighting system and their
work environment, and the occupants’ use of the horizontal blinds. In this paper we
focus on the energy performance of the lighting control system; other aspects of the
study will be reported in future publications. Specifically, we present here:
The overall energy savings and power demand reductions attributable to the
lighting control system compared to (1) the energy used at full power during
work-hours by the installed system, and (2) a static, ceiling-recessed,
conventional fluorescent lighting system on a neighboring floor.
The separate energy-saving contributions from individual control, occupancy
sensors and daylight photosensors, a key factor toward optimizing the
performance of automatic lighting control systems.
The effect on the energy use and power demand of an intervention to the
workplace expected to increase energy savings. The intervention consisted of
an awareness campaign, which used e-mail reminders to encourage the
occupants to use the individual control feature.
The energy-saving potential of four other design/operation options that, in
theory, could have further reduced the energy use of the lighting system.
2. LITERATURE REVIEW
A number of prior research studies relevant to this study are summarized below.
Jennings and others (2000) found that in private offices occupancy sensors that
turned the lights off after a 15-20 minute period of no-occupancy saved between 20-
26% in lighting energy compared to the manual operation of a wall switch alone.
Daylight-linked dimming provided additional savings of about 20%. In the offices
where occupancy was low, energy savings resulted mostly from the occupancy
sensors, while in the offices with high occupancy, savings were mostly attributable to
light sensor dimming. In the offices where manual wall dimmers were available,
savings were in the range of 9% by occupant dimming alone, while savings in areas
with occupant bi-level switching of a 3-lamp fixture were 23%. The authors noted
that “by the time the (dimming) system had been in place for over a year there was
little significant (manual) dimming activity taking place”. When the lights were on,
they were usually used at more than 90% of full power. The authors speculated that
the occupants may have used the manual dimmers more actively had they been
placed closer to their working area (desktop mounted or hand-held remotes).
Comparing the occupants’ switching behavior in the offices that had occupancy
sensors to the ones that did not have them, the authors did not find any evidence to
suggest that people without occupancy sensors would be more likely to manually
switch the lights off when leaving the office for long periods of time than the people
having them.
4
In contrast, based on a study conducted in 63 private offices over 11 months, Pigg
and others (1996) concluded that people are likely to change their behavior in the
presence of controls. In offices equipped with occupancy sensors, the occupants
were “half as likely to turn out the lights” when leaving the space compared to people
without occupancy sensors. The authors noted that if the occupants with occupancy
sensors had switched the lights manually, the savings from that group would have
increased by 30%. The additional energy used was due to the lights remaining on
during the time-delay of the occupancy sensors. Similar to Jennings and others
(1999), people in both groups selected full lighting output from the luminaires when
using wall-mounted dual-level switches: 95% of the time in the offices with
occupancy sensors, and 89% of the time in the control group. The authors
speculated that people who rely on controls to operate the lights are less likely to
choose “a switch setting other than full illumination.” While the occupants did not
adjust their lights very often, they appreciated the ability to do so.
Boyce and others (2000) speculated that given control over lighting, people would
“initially explore the range of illuminances available and then gradually home-in on
the illuminances they like.” This suggests a decrease in the frequency of use of
controls over time. However, the authors did not perceive this as an argument
against the provision of such control. The participants in Boyce’s experiment viewed
the ability to select lighting levels as highly desirable, and the light levels they
selected were linked to the type of work that they did.
In daylit private offices, based on responses to questionnaires, Maniccia and others
(1999) found that occupants did not consciously use their manual light dimmers to
save energy but rather to accommodate the tasks they performed. Nevertheless,
data showed that the selected light levels did not vary with the type of task. Offices
were occupied an average of 4½ hours a day and 74% of the 58 occupants
observed over a 7-week period used their wall-mounted or portable desk-dimmers to
adjust their lights. Over half of the time the lights were either dimmed or turned off.
Energy savings from the manual controls were 15% in addition to savings from
occupancy-based controls, which provided 43% savings on their own. Upon re-
entering the office after the occupancy sensors had extinguished the lights, the lights
remained off unless the occupant used the dimmers to restore them. The occupants
appreciated having the dimmers located on their desks, and removing the desk-
dimmers (so that dimming was possible via a wall switch only) resulted in fewer
dimming adjustments.
Several investigations into various open-plan office buildings in the UK showed that
occupants generally prefer to have the capability to choose their own lighting
environment rather than having to accept lighting levels chosen for them (Slater and
Carter 1998; Slater and others 1998; Carter and others 1999; Moore and others
2001, 2002). Questionnaires from 410 occupants collected over a 3-year period
showed that the occupants viewed the installations that they could control more
positively, even when the measured lighting conditions did not meet the currently
5
recommended lighting levels for offices. The authors noted that individuals purposely
used the controls to set their preferred lighting levels and not only to counteract
discomfort. They reported that “by far the most frequent response was that people
wanted control over an individual luminaire” (Moore and others 2002). In the winter,
20% of the users chose illuminance levels below 100 lux, while over 50% worked at
levels below 300 lux, and less than 25% worked under the recommended 300-500
lux range. All lighting installations were used at less than full power, the average
power demand varying between 50 to 60% all year round. Building depth,
percentage of glazed area, or degree of obstructions had no effect on these outputs.
Conflicts were reported in areas where groups of luminaires were linked together
and controlled by more than one user. While the authors found that individuals
generally worked in a very wide range of illuminances, they also noted “a strong
correlation between luminaire output and distance of the workstation from the
window” (Carter and others 1999). This suggests that people actively changed the
electric light levels in response to the available daylight.
After surveying several open-plan office buildings in the UK incorporating various
types of lighting controls, Bordass and others (1994) also reported that unfamiliarity
with controls or the bad locations of controls discouraged their use. These authors
suggested that the best location for switches while working is at the workstation. The
authors also reported high rates of dissatisfaction with photocontrolled lighting due to
the lights going on/off inappropriately; distracting transitions; incorrect installation
and calibration; and lack of possibility to override them. Occupancy sensing was also
not always perceived positively due to the lights going off inappropriately. The very
few successful installations they found, from both occupant satisfaction and energy
efficiency points-of-view, were installed according to workstation layout and daylight
availability; had local controls with clear user interfaces permitting easy-tuning to
individual requirements; had easy access to blinds; well-informed occupants; and
good building management. Similar observations were made by Slater (1995,
1996), Escuyer and Fontoynont (2001), Roche and others (2001), and Wyon (1999).
Based on 26 case studies that investigated the effectiveness of occupancy sensors
to generate energy savings in various space types, Figueiro (2004) proposed
estimates for expected energy savings from occupancy sensors in private and
shared spaces, with scheduled versus sporadic use. In private offices with sporadic
use, occupancy sensors accounted for an average of about 25% energy savings
during 7.5 to 10 hours of use, while in shared spaces with sporadic use, including
open-plan offices, the average savings were around 40%. In shared spaces with
scheduled use, such as classrooms, the average savings were around 30%. The
larger energy savings related to occupancy sensors installed in shared spaces was
attributed to the fact that in such spaces the occupants generally do not feel as
responsible for manually switching off the lights when leaving a space as they would
when leaving a private office.
In a full-scale open-plan test installation with 1.2 m high workstation partitions, Lee
and Selkowitz (2006) tested two types of daylight-linked lighting system (open-loop
6
dimming system with proportional control, and a DALI dimming system) and found
that the lighting energy savings were still substantial at a depth of 7 meters from a
window wall equipped with automated roller shades. In a side-lit area with an open-
loop dimming system, from mid-February to mid-September, the average savings for
a 7-meter depth zone were 20-23%. At the same distance from the window, in a
bilateral daylit zone featuring the DALI dimming system, the average savings were
52-59%. In the DALI area, the lights were turned off when there was sufficient
daylight (0 light = 4% of full power draw), whereas in the area featuring the open-
loop dimming system, the lights were dimmed only down to a minimum power (5-
10% light = 35% of full power consumption). The authors noted that “without active
shade management” the energy savings would have been significantly lower “due to
non-optimal control by the occupants”.
3 SITE DESCRIPTION
The study was conducted on floors 8 to 11 of a 12-storey rectangular, curtain-wall,
green-tinted glazed structure (Fig. 1) located in Burnaby, British Columbia, Canada,
(latitude 49°11’ N, longitude 123°10’ W). The study floors consisted mostly of open-
plan areas (75% of total floor area) furnished with cubicle-type workstations and no
private offices. A few enclosed areas were located at the core of the building
providing shared spaces for meeting rooms, break rooms, and storage. All perimeter
workstations had two or three window panes equipped with manually operated off-
white standard horizontal blinds. Each floor had an approximate area of 835 m2. The
height of the partitions between the workstations varied from 0.84 m next to the
windows, to 1.25 m between two adjacent workstations, and 1.42 m next to the
aisles. There were few external obstructions to hamper daylight admittance into the
building.
Fig. 1 North-east view of the test-site.
The majority of workstations on the study floors had commercial direct-indirect
luminaires suspended at about 0.3 meters below the ceiling and located centrally in
each workstation (Fig. 2). When fully on, the system provided an average
7
illuminance of 450 lux in the centre of the workstation at 0.85 m above the floor
(desktop height).
Fig. 2 Typical installation of the luminaires and the window shading.
Each luminaire (Fig. 3) consisted of 3x32-W lamps (3500 K) connected by a network
to a central control computer and to each occupant’s desktop computer. The fixture
also included an occupancy sensor and a daylight photosensor. The lamp in the
center of the luminaire was equipped with a static electronic ballast and directed the
light mainly upward, providing constant general lighting around the open-plan space.
During the study, these lamps were controlled centrally based on a daily schedule
that kept them continuously on at full power from 7:30 AM to 5 PM on workdays.
Outside of these hours, the uplight lamps were turned on by an integrated
occupancy sensor when sensing occupancy in the vicinity.
The two lamps at the sides directed the light mainly downward. The downlights were
controlled during the study based on the following three control options:
An integrated occupancy sensor (OS). It consisted of an infrared motion
sensor mounted directly on the luminaire. On detecting vacancy, the sensor
prompted the downlights to gradually dim down to zero and switch-off. When
presence was detected, the downlights were automatically restored to the
previously set lighting level.
An integrated light sensor (LS), used to monitor the surrounding light levels
and dim the downlights when sufficient light (from either daylight or
neighboring electric light) was present to maintain the occupant preset light
level. The light sensor consisted of a photocell mounted directly on the
luminaire.
Individual control (IC), consisting of an on-screen slider located on the
occupants’ desktop computers that allowed both on/off switching or dimming
of the downlights to a preferred level.
8
IC
OS
LS
Fig. 3 Illustration of the three control options of the luminaire downlights
During the study, the field installation comprised a total of 195 luminaires distributed
over floors 8, 9, 10 and half of floor 11. At installation (4 years prior to this study),
these fixtures replaced a total of 530 2’x4’ (60x120 cm) conventional ceiling-
recessed fluorescent louvered luminaires with 2x32-W T8 lamps (3500 K) and
electronic static ballasts, which reduced the lighting power density by almost half
(5.8 W/m2 versus 10 W/m2). This conventional lighting system remained in the other
half of floor 11, and provided a comparison group in the occupant satisfaction survey
conducted as part of the larger project.
4 METHOD
4.1 CHARACTERISTICS OF LIGHTING SYSTEM OPERATION
Data related to the luminaires was continuously collected over a 12-month period in
three phases:
Phase 1 was conducted from January 18 to March 11, 2005 (39 workdays)
with the light sensor disabled. During this time the downlights were controlled
only by the occupancy sensors and the on-screen individual controls. The
occupancy sensors were set to operate with a time-delay of 8 minutes. This
time-delay was followed by a period of 7 minutes of continuous dimming
before the downlights turned off.
Phase 2 was conducted between March 12 and October 2, 2005 (140
workdays) with all the controls enabled. The occupancy sensors were set to
operate with a time-delay of 12 minutes. This time-delay was followed by a
period of 3 minutes of continuous dimming before the downlights turned off,
which resulted in a total time between the last detected motion and the
downlights off condition of 15 minutes (equivalent to the previous period).
Phase 3 (Awareness Campaign) was conducted from October 3 to December
31, 2005 (61 workdays) similar to Phase 2 with the exception that monthly e-
mails were sent to the employees to: remind them about the lighting control
9
system; provide them with information on how to use it; and encourage them
to save energy by using the on-screen individual lighting controls. The
following wording was used in the e-mail regarding how the individual control
might be used: ” Remember that you can reduce the energy use substantially
by setting the brightness of the light fixture in your workstation to the lowest
level that is comfortable for you, and by turning the lights out completely when
not needed”.
All lighting fixtures were preset at installation to restrict downlight dimming to 50%
light output when controlled by the light sensor. This was done to prevent large
variations in light levels, which the design team believed might inconvenience the
occupants. The occupants could still dim the downlights using the on-screen control
to any levels below this limit, if desired. When turned on, but dimmed to minimum
output, the power demand of the downlights was 19 W (or 51 W/luminaire, including
the 32 W uplight), whereas dimmed at 50% output, the power demand of the
downlights was about 41 W (or 73 W/luminaire).
4.2 MONITORING OF LIGHTING SYSTEM ENERGY USE AND POWER DEMAND
Each individual luminaire was monitored using a modified version of the
communication software provided by the manufacturer as part of the standard
installation of the lighting system. The software was adapted to automatically log the
energy use of each luminaire every 15 minutes, and record the occupant use of the
on-screen control slider, and the status of the occupancy and daylight sensors. The
field installation reported the energy use of each luminaire with all available controls
in operation simultaneously. However, since we also wanted to derive the separate
saving contribution of each control, we developed a mathematical model to calculate
the energy use and power demand of each luminaire if only one control, or two
controls combined, had been in operation. The mathematical model used
correlations between the dimming level, electric power demand, and light sensor,
occupancy sensor and on-screen individual control setpoints, determined based on
the field-collected data and measurements from a similar system installed in a
laboratory setting.
To calculate the savings associated with the controls, we considered the following
three basic cases, which assumed full lighting energy use during work-hours:
energy use and power demand in the absence of controls during the basic
daily work-schedule of the lighting system (7:30 AM to 5 PM).
energy use and power demand in the absence of controls during the total
work-hours (basic work-schedule, 7:30 AM to 5 PM, plus the additional time
that the lights were reported to have been on outside the scheduled hours).
10
energy use and power demand of a conventional lighting system consisting of
2’x4’ (60x120 cm) parabolic louvered luminaires with 2x32-W T8 fluorescent
lamps, during the total work-hours.
For each of the three phases of the study, we calculated the percentage in energy
savings and the power demand reductions with the downlights controlled by:
occupancy sensors only (os); In this case it was considered that no individual
control or light sensor control were available, therefore, the downlights would
have been used continuously at full power when the occupants were present
in their workstations (actual occupancy), and off at other times.
individual controls only (ic); In this case it was considered that no occupancy
or light sensor control were available, therefore, the downlights would have
been used at the occupant-selected level during the total work-hours.
light sensor controls only (ls); In this case it was considered that no
occupancy or individual control were available, therefore, the downlights
would have been used continuously at the light sensor selected dimming level
during the total work-hours.
occupancy sensors and individual controls combined (os+ic); In this case the
downlights would have been used at the dimming level set by the occupants
during the workstation actual occupancy, and off at other times.
occupancy sensors and light sensor controls combined (os+ls); In this case
the downlights would have been used at the dimming level set by the light
sensors during the workstation actual occupancy, and off at other times.
individual controls and light sensor controls combined (ic+ls); In this case the
downlights would have been used at a dimming level selected to be the
minimum between the dimming level dictated by the light sensor setting and
the dimming level dictated by the individual control setting during the total
work-hours.
all available controls combined; (os+ic) for Phase 1; (os+ic+ls) for Phases 2
and 3; In this case the downlights would have been used at a dimming level
selected to be the minimum between the dimming level dictated by the light
sensor setting and the dimming level dictated by the individual control setting
during the workstation actual occupancy, and off at other times. These
values, calculated with the same mathematical model used for the one-control
and two-control scenarios described above, were subsequently compared to
the real energy use reported by the lighting system monitoring software, being
indicative of the accuracy of our theoretical model.
11
All calculations included the energy used by the uplights, which were continuously
on at full power during scheduled hours, and on outside these hours when
occupancy was detected in the workstations.
In order to identify the effect of the downlight dimming restriction on the energy use,
we also calculated the energy savings if the downlights had been allowed to drop to
zero on light sensor control. Furthermore, we also calculated the energy savings
associated with three other design/operation options that could have, theoretically,
further reduced the energy use, as follows:
Option 1 = lighting system equipped with 32-W static uplights and 2x32-W dimmable
downlights (as installed), but with the downlights allowed to dim to zero on light
sensor control;
Option 2 = lighting system with 25-W static uplights and 2x32-W dimmable
downlights allowed to dim to zero on light sensor control;
Option 3 = system with 3x32-W dimmable uplights and downlights, both restricted at
50% output on light sensor control;
Option 4 = system with 3x32-W dimmable uplights and downlights, both allowed to
dim to zero on light sensor control.
The above calculations apply only to the data collected during Phases 2 and 3 when
the lighting system operated with all three controls enabled.
4.3 DATA ACCESS AND SAMPLE SIZE
Because the study included the examination of the lighting control usage data at the
individual level, formal consent to analyze the lighting system data was sought from
each occupant, in accordance with the requirements of our Research Ethics Board
(similar requirements were met for the survey aspect of the larger study). Only the
records logged in the workstations where the occupants gave specific permission to
release their data for analysis were included. This reduced the sample size from 195
to 86 luminaires, of which 57 were in workstations located at the perimeter of the
building with direct access to windows; 18 were located in 2nd row workstations
adjacent to the perimeter workstations at distances between 2.5 and 4.5 meters from
the windows; and 11 were located at the core of the building at distances greater
than 5.0 meters from the closest window. The sample size was not big enough to
permit analysis by façade orientation.
5 RESULTS
5.1 LIGHTING SYSTEM USED AT FULL POWER DURING WORK-HOURS
12
Data collected throughout the year showed that during all three phases of the study,
the actual average daily time-of-use of the lighting system was higher than the 9.5
hours used by the lighting system schedule (7:30 AM to 5 PM), as shown in Fig. 4.
Whereas, due to the occupancy sensors, the average daily on-time of the downlights
was between 4-6 hours/day, the uplights were used at full power for an average of
10-11 hours/day.
0
2
4
6
8
10
12
Jan 18 - Mar 11 (L S OFF) Mar 12 - Oct 2 (LS ON) Oct 3 - Dec 31 (Awaren ess
campaing)
Daily average time-of-use [hrs]
Downlights
Uplights
Weekdays only (86 workstations)
Fig. 4 Average daily time-of-use of the lighting system during all three phases of the study.
The average daily energy that would have been used by the downlights if maintained
at full power during the actual work-hours was 0.69 kWh/workstation/day for Phase1;
0.72 kWh/workstation/day for Phase 2; and 0.74 kWh/workstation/day for Phase 3.
These values are about 10-17% greater than the energy used by the downlights at
full power during scheduled hours (0.65 kWh/workstation/day from 7:30 AM to 5
PM). Therefore, we calculated the percentage in energy savings for the various
control scenarios mentioned above relative to the energy used by the fixtures at full
power during the total daily work-hours, which we considered to be a more realistic
comparison baseline than the 9.5 hours used by the lighting system’s daily schedule.
The average daily energy used by the uplights was about 0.33 kWh/workstation/day.
5.2 PERFORMANCE OF THE LIGHTING SYSTEM AS INSTALLED
Table 1 presents a summary of the luminaire daily average percentage energy
savings and power demand reductions associated with all the control scenarios,
compared to full light output from the studied system. Additionally, the table also
includes the theoretical energy savings associated with the other four
operation/design options mentioned previously.
13
Table 1 Summary of luminaire daily average energy savings and power demand reductions for various control scenarios compared to full light
output from the studied system
As Installed Option 1 Option 2 Option 3 Option 4
Phase 1 Phase 2 Phase 3 Phase 2 Phase 3 Phase 2 Phase 3 Phase 2 Phase 3 Phase 2 Phase 3
Jan18-Mar11 Mar12-Oct2 Oct3-Dec 31 Mar12-Oct2 Oct3-Dec 31 Mar12-Oct2 Oct3-Dec 31 Mar12-Oct2 Oct3-Dec 31 Mar12-Oct2 Oct3-Dec 31
Energy Savings % % % % % % % % % % %
occupancy sensors (os) 29 35 38 35 38 38 40 52 54 52 54
individual controls (ic) 20 11 5 11 5 11 5 15 7 15 7
light sensors (ls) - 20 11 32 16 34 18 29 16 47 24
occupancy sensors + individual controls 40 40 39 40 39 43 42 59 56 59 56
occupancy sensors + light sensors - 45 44 51 46 55 49 66 62 75 66
individual controls + light sensors - 24 14 34 19 37 21 35 20 51 27
all available controls (estimated) 40 47 44 52 47 56 50 69 64 76 67
all available controls (real) 39 47 42 - - - - - - - -
Power Demand Reductions % % % % % % % % % % %
occupancy sensors (os) 31 36 38 36 38 39 41 54 57 54 57
individual controls (ic) 21 12 5 12 5 13 5 17 7 17 7
light sensors (ls) - 23 15 39 24 42 25 34 23 59 35
occupancy sensors + individual controls 41 41 40 41 40 45 43 62 59 62 59
occupancy sensors + light sensors - 47 46 55 50 59 54 70 68 82 74
individual controls + light sensors - 26 18 41 26 44 28 39 26 61 38
all available controls (estimated) 41 48 46 55 50 59 54 72 69 82 75
all available controls (real) 40 49 43 - - - - - - - -
Phase 1 = Installed system with the light sensor disabled (January 18 to March 11, 2005);
Phase 2 = Installed system with all three controls enabled (March 12 to October 2, 2005);
Phase 3 = Installed system with all three controls enabled during the Awareness Campaign (October 3 to December 31, 2005);
Option 1 = Installed system if downlights were allowed to dim to zero on LS (maximum as installed saving potential);
Option 2 = System with static 25 Watt uplights and downlights allowed maximum dimming on LS;
Option 3 = System with dimmable uplights and downlights restricted at 50% output;
Option 4 = System with dimmable uplights and downlights allowed maximum dimming on LS.
14
5.2.1 ENERGY USE
Figure 5 shows the daily average energy used per luminaire during Phase 2 for the
various control scenarios. Also shown in Fig. 5 is the calculated energy use of the
conventional lighting system (providing a similar target desktop illuminance) at 1.83
kWh/workstation/day. Due to its reduced lighting power density, the installed system
would have saved 42% in electric energy if used at full power during work-hours
compared to the conventional system.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
no controls, conventional system
no controls, scheduled occupancy
no controls, real occupancy
occupancy sensors (os)
individual controls (ic)
light sensors (ls)
os+ic
os+ls
ic+ls
os+ic+ls, estimated
os+ic+ls, real
Fixture average daily energy use [kWh]
Interior WS
2nd Row WS
Perimeter WS
Mar 12 - Oct 11, 2005
(Phase 2)
Fig. 5 Luminaire daily average energy use for various control scenarios from March 12 to October 2,
2005 (Phase 2); the energy use of a conventional lighting system is shown for comparison (data
shown by luminaire proximity to windows).
Three-controls scenario
As shown in Fig. 6 (for Phase 2) and Table 1 (for all three phases), the three
controls combined (os+ic+ls, real) saved 42-47% compared to the energy used by
the same luminaires at full power. This translated into energy savings of 67-69%
compared to the static conventional system (Fig. 7)
The energy use reported by the system was also remarkably close to the estimated
energy use (os+is+ls, estimated), obtained using the theoretical model we used to
separate the saving contributions from each control feature if used individually,
which provides confidence in the accuracy of the model (Fig.6).
15
35%
11%
20%
40%
45%
24%
47%
47%
0% 20% 40% 60% 80% 100%
occupancy s ensors (os)
individual controls (ic)
light sensors (ls)
os+ic
os+ls
ic+ls
os+ic+ls, estimated
os+ic+ls, real
Fixture (Downlight + Uplight) average energy savings
Mar 12 - Oct 2, 2005
(Phase 2)
Fig. 6 Luminaire average energy savings for various control scenarios from March 12 to October 2,
2005 (Phase 2) compared to full lighting use of the installed system during total work-hours (data
shown averaged across all locations; downlight restricted to 50% output on light sensor).
42%
63%
48%
54%
65%
68%
56%
69%
0% 20% 40% 60% 80% 100%
full power (no controls)
occupancy sensors (os)
individual controls (ic)
light sensors (ls)
os+ic
os+ls
ic+ls
os+ic+ls, real
Fixture (Downlight + Uplight) average energy savings
(Phase 2)
Mar 12 - Oct 2, 2005
Fig. 7 Luminaire average energy savings for various control scenarios from March 12 to October 2,
2005 (Phase 2) compared to the energy use of a conventional static lighting system during total work-
hours (downlight restricted to 50% output on light sensor).
Occupancy sensor control scenario
Calculations of the energy use of the lighting system had it been controlled by
occupancy sensors only, revealed that the occupancy sensors were the single
control option with the highest potential for energy savings. As shown in Table 1, the
fixture average daily savings across all three phases of the study were between 29-
38% compared to lights fully on during work-hours. These values are slightly higher
than those reported for private offices by Jennings and others (2000) and Figueiro
(2004), but close to Maniccia and others (1999). A small difference of 4-8% in
occupancy sensor savings was observed in our study based on workstation
proximity to windows (as shown in Fig. 8 for Phase 2), which was most likely linked
to the occupants’ type of work rather than to their window/daylight exposure.
16
37%
13%
24%
44%
48%
28%
50%
52%
34%
10%
20%
38%
44%
24%
46%
45%
35%
8%
16%
39%
43%
20%
45%
44%
0% 20% 40% 60% 80% 100%
occupancy sensors (os)
individual controls (ic)
light sensors (ls)
os+ic
os+ls
ic+ls
os+ic+ls, estimated
os+ic+ls, real
Fixture (Downlight + Uplight) average energy savings
Interior WS
2nd Row WS
Perimeter WS
Mar 12 - Oct 2, 2005
(Phase 2)
Fig. 8 Luminaire average energy savings for various control scenarios from March 12 to October 2,
2005 (Phase 2) compared to full lighting use of the installed system during total work-hours (data
shown by luminaire proximity to windows; downlight restricted to 50% output on light sensor).
Light sensor control scenario
Had they been used as the only control option of the lighting system, the light
sensors showed an overall potential for energy savings in the range of 11 to 20%
compared to the energy use at full power (Table 1). As shown in Fig. 8 for Phase 2,
the savings were, as expected, higher in the perimeter workstations (24%) versus
the interior workstations (16%). Across both Phases 2 and 3, the light sensor saved
between 17-24% in the perimeter workstations; 9-20% in the 2nd row workstations,
and 9-16% in the interior workstations. These findings are consistent with Jennings
and others (2000) and Lee and Selkowitz (2006). However, in Lee and Selkowitz’s
study the windows were equipped with automated roller shades, whereas in the
present study the windows were covered by manual horizontal blinds. In addition,
the height of the partitions between the workstations was 1.2 meters in Lee and
Selkowitz compared with varying partition heights of 0.84, 1.25 and 1.42 meters in
this study.
The energy savings from light sensor control beyond the perimeter workstations (16-
20%) were a direct result of the relatively high daylighting levels available in the
interior workstations. Illuminance data collected on the top of 12 partitions separating
the workstations and measured at 10-minute intervals from 6 AM to 6 PM during all
weekends of the study, in the absence of electric lighting, showed that the average
daylight illuminance was between 200 to 400 lux at distances between 2.5 and 4
meters from the nearest window (average of 4 locations), and about 100 to 150 lux
beyond 5 meters from the nearest window (average of 8 locations).
Individual control scenario
As shown in Table 1, if used as the only control option, the on-screen individual
controls showed the lowest potential for energy savings, ranging from 5 to 11%
during Phases 2 and 3, when the light sensor was enabled, to 20% during Phase 1,
when the light sensor control was disabled. These values are generally consistent
17
with Jennings and others (2000), Maniccia and others (1999), and Veitch and
Newsham (2000). Note that our calculation is likely an underestimate of the true
energy savings in this scenario, as we cannot account for the additional switching
that some occupants would enact on leaving the space.
Table 2 presents the number of workstations with dimming and on/off occupant-
requested adjustments for each phase of the study. During Phase 1 (39 workdays),
user light level adjustments occurred in 81 out of the 86 workstations considered.
However, most of these adjustments occurred at the beginning of Phase 1, in the
two days following an unannounced lighting system reset that deactivated the light
sensor at the start of the project. During these two days only, there were a total of 55
on/off and 71 dimming user-requests. Throughout the rest of Phase 1, however, the
occupants used the on-screen individual controls only occasionally, and the number
of workstations where user adjustments occurred (active workstations) was similar to
that shown for Phases 2 and 3. During Phase 2 (140 workdays), on-off user-
requested adjustments were observed in only 40% of the 86 workstations
considered, and user-requested dimming occurred in 60% of these workstations.
Similarly, throughout Phase 3 (61 workdays), on-off user-requested adjustments
were observed in 25% of the workstations, while user-requested dimming occurred
in 50% of the workstations.
Even among the occupants who used the on-screen controls, half used them only
once or twice during each phase of the study, the average number of user-requested
light level adjustments per active workstation being 1.8 on/off and 2.5 dimming
adjustments for the whole of Phase 1; 2.5 on/off and 2.7 dimming adjustments for
Phase 2; and 5.2 on/off and 3.5 dimming adjustments for Phase 3. The apparent
higher average rate-of-use of the individual control during Phase 3 was, however,
due to one single occupant who used the system very actively (50-57 adjustments).
The average number of light level adjustments for Phase 3 drops to 2.6 on/off and
2.4 dimming adjustments if this outlier user is excluded. The maximum number of
either on/off or dimming adjustments occurring in any other workstation during each
period was eight.
Table 2 Frequency-of-use of the on-screen individual controls (dimming and on/off occupant-
requested adjustments) Phase 1 Phase 2 Phase 3
Workdays in period 39 140 61
No.of workstations with manual on/off adjustments (out of 86) 81 (37)* 34 21
No.of workstations with manual dimming adjustments (out of 86) 82 (50)* 52 44
Total number of manual on/off adjustments for all workstations and days 145 (68)* 86 109 (52)**
Total number of manual dimming adjustments for all workstations and days 205 (108)* 138 152 (102)**
Average manual control adjustments/workstation/day (across 86 workstations) 0.10 (0.05)* 0.02 0.05 (0.03)**
*excluding the first week after the initial lighting system reset (Jan 18-25, 2005)
**excluding one outlier user
During Phase 1, there was a daily average of 3.72 on/off adjustments/day across the
86 workstations and 5.26 dimming adjustments/day. However, if we exclude the
18
adjustments which occurred during the first week after the lighting system’s reset,
these numbers drop to 1.74 on/off and 2.77 dimming adjustments/day. During Phase
2, the number of individual control adjustments dropped even more to an average of
0.99 dimming adjustments/day and 0.61 on/off adjustments/day, which shows that
the active occupants used the individual control less frequently when the lighting
system was controlled by the light sensor. This suggests that once the occupants
selected a light level, if that light level was reasonably well regulated and maintained
by the light sensor, the users were satisfied with the selected level for long periods
of time. Jennings and others (2000) also noted very little dimming activity after wall
dimmers in their study had been in place for more than a year.
Data presented in Table 2 also shows that the awareness campaign (Phase 3) did
increase slightly the daily rate-of-use of the individual controls to 0.85 on/off
adjustments/day and 1.67 dimming adjustments/day (when excluding the outlier
user). However, the average daily energy savings from the individual controls
dropped from 11% during Phase 2 to only 5% during Phase 3. This was a direct
result of the fact that in 20 out of the 44 active workstations in Phase 3, the light
levels selected were higher than those recorded during Phase 2, whereas 12
settings were the same as Phase 2, and another 12 lower. We speculate that this
was a result of both the occupants being reminded periodically about the lighting
controls available to them, combined with a seasonal effect that reduced indoor
daylight availability.
Calculated on a per workstation per day basis, the frequency-of-use of the individual
controls was notably low, averaging only 0.02-0.05 control actions (on/off and
dimming together) per workstation per day (Table 2).
There was a 5% difference in downlight energy savings from individual control based
on workstation proximity to windows (as exemplified in Fig. 8 for Phase 2), with the
savings being a little higher closer to the window. This is, generally, consistent with
an earlier finding that people manually reduced electric light levels in response to the
available daylight (Moore and others 2003). However, in our study this phenomenon
was observed only for the period with longer daylight hours (Phase 2, March 12 to
October 2).
Two controls scenarios
Data collected during Phases 2 and 3 show that the occupancy control combined
with light sensor control would have generated energy savings almost as high as
those generated by the system operating with all three controls (estimated average
savings of 44-45%; Table 1). The next best two-control scenario was the
combination between occupancy and individual control (average savings of 39-
40%). The light sensor and the individual control used together would have saved
only 14-24% in energy for lighting, which is 11-24% lower than the energy savings
generated by the occupancy sensors used alone. Especially for Phase 2, the period
with longer daylight hours, this was mainly because of the downlight’s restriction to
19
50% output when controlled by the light sensor. Nevertheless, during Phase 3, the
occupancy sensor remained the single best energy-saving strategy.
5.2.2 POWER DEMAND
Data showed that the lighting system also generated significant reductions in the
peak power demand for lighting. Table 1 shows the daily average reductions in peak
power demand for all phases of the study, for the various control scenarios
investigated. When used together, the three controls reduced the daily peak power
demand during work-hours on average by 43-49% compared to the same fixtures
used at full power (97 W/workstation), and by 65-70% compared to a conventional
lighting system (174 W/workstation). This is an important benefit for electric utilities
seeking to balance supply and demand at reasonable cost on high electricity
demand days (Newsham 2006). Because the controls ensure that not all the
installed lighting power is used simultaneously, in this study the installed lighting
power density of the studied system (5.8 W/m2) was reduced to an effective average
lighting power density of only 3 W/m2. This compares very favorably with the 12
W/m2 installed allowed for open-plan offices in the widely used ASHRAE 90.1
Energy Standard (2004).
Figure 9 shows the average daily power demand profile of the luminaires during
Phase 2. The average peak load was about 53 W/luminaire and occurred from 9 AM
to 12 AM, and from 2 PM to 4 PM. The average power demand dropped to about 45
W/luminaire at 1 PM due to the lunchtime break. The controls also reduced the
power demand during non-scheduled hours and during the rounds of the cleaning
and security staff.
0
20
40
60
80
100
120
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time of Day [hour]
Fixture Average Power Demand [Watt]
no controls (scheduled occupancy)
no controls (real occupancy)
os
ic
ls
os+ic
os+ls
ic+ls
os+ic+ls (estimated)
os+ic+ls (real)
Mar 12- Oct 2, 2005
Fig. 9 Luminaire average daily power demand for various control scenarios compared to full lighting
use from March 12 to October 2, 2005 (Phase 2) during scheduled and total work-hours (downlight
restricted to 50% output on light sensor).
20
5.3 LIGHTING SYSTEM ENERGY SAVINGS POTENTIAL
Below we present the potential for energy savings and power demand reductions of
four system and control alternatives, compared to the one installed.
5.3.1 OPTION 1: SYSTEM WITH STATIC UPLIGHTS (AS INSTALLED) AND
DIMMABLE DOWNLIGHTS ALLOWED TO DIM TO ZERO ON LIGHT SENSOR
CONTROL
As shown in Table 1, if the downlights had been allowed to dim to zero on light
sensor control, then all the control scenarios incorporating light sensor control would
have reduced the fixture average peak power demand by an additional 7 to 16% for
Phase 2, and by an additional 4 to 9% for Phase 3, compared to the actual case.
It is interesting to note that in terms of demand reduction, during Phase 2, the light
sensors alone would have performed slightly better than the occupancy sensors
alone for most of the workday (39% power reduction compared to 36% from
occupancy control alone). Nevertheless, during Phase 3, because of the reduced
daylight availability, the occupancy sensors would have still been the single control
option with the greatest power reduction.
The average energy savings of the lighting system with all three controls in operation
would have been by only 3-5% higher during both periods. Data sorted by fixture
proximity to windows showed that in terms of energy use, the light sensor savings
would have exceeded those of the occupancy sensors only in the perimeter
workstations and only during Phase 2, the period with longer daylight hours. On
average, under light sensor only control during Phase 2, the energy savings would
have been 12% higher had the downlight been allowed to dim to zero. The savings
would have been by only 5% higher during Phase 3.
5.3.2 OPTION 2: SYSTEM WITH REDUCED POWER STATIC UPLIGHTS AND
DIMMABLE DOWNLIGHTS ALLOWED TO DIM TO ZERO ON LIGHT SENSOR
CONTROL
In this case the luminaire average energy savings with the three controls in
operation would have increased by an additional 6 to 9% compared to the actual
case (Table 1). The luminaire average peak power demand would have been
reduced on average by an additional 8-11%, the peak power demand being in this
case about 40-46 W/fixture. Of course, reducing the wattage of the uplights would
have slightly reduced the overall light levels.
21
5.3.3 OPTION 3: SYSTEM WITH DIMMABLE UPLIGHTS AND DOWNLIGHTS,
BOTH RESTRICTED TO 50% OUTPUT ON LIGHT SENSOR CONTROL
In this case the average energy savings with all three controls in operation would
have increased by an additional 20-22% compared to the actual case (see Table 1).
The luminaire average peak power demand would have also been reduced by an
additional 24%, the peak power demand dropping in this case to about 33-36
W/fixture.
5.3.4 OPTION 4: SYSTEM WITH DIMMABLE UPLIGHTS AND DOWNLIGHTS,
BOTH ALLOWED TO DIM TO ZERO ON LIGHT SENSOR CONTROL
In this case the average energy savings with all three controls in operation would
have increased by an additional 23 to 29% compared to the actual case (see Table
1). The luminaire average peak power demand would have also been reduced by an
additional 29-34%, the peak power demand being in this case between 20-32
W/luminaire depending on the season. Uplights, however, are often not switched off
with direct-indirect luminaires because this would create an uneven light distribution
on the ceiling.
6 FURTHER DISCUSSION
One of the goals of this project was to weigh the relative contribution of the three
different control systems to the energy savings. Data indicated that if only one type
of control were to be installed in this building, and the energy savings and power
demand reductions were the principal performance criteria, then the occupancy
sensors would be the best choice. As installed, they provided savings in lighting
energy use in the range of 30 to 40% compared to full lighting use. Note, however,
that these savings were a result of a period of 12 minutes between the moment the
last motion was detected and the start of gradual dimming, followed by a period of 3
minutes of continuous dimming before the lights turned off. Changes to either of
these two intervals would have affected the energy savings.
Light sensor control (“daylight harvesting”) would have provided similar power
demand reductions and energy savings to the occupancy sensor control only in the
perimeter workstations, only seasonally during periods with long daylight hours, and
only if downlight dimming to zero were permitted. On average, the light sensors
saved about 10-20% energy compared to full lighting use of the installed system,
and 16-32% had the dimming to zero been allowed. Note that data on blind use
collected throughout this study showed a high average blind occlusion of the
facades of this building of 55%; clearly blind use may strongly affect energy savings
with such controls.
If they had been installed independently, the individual controls in this installation
would have been the worst single control choice in terms of energy savings (average
savings of less than 10% compared to full lighting use), and adding individual control
22
to the system already controlled by occupancy and light sensors provided very little
additional energy saving benefit. Nonetheless, the ability of the occupants to choose
their own preferred light level remains an important benefit not offered by the other
two control types. Individual control has been linked to improved occupant mood,
satisfaction and comfort (Newsham and others 2004), and improved environmental
satisfaction has been linked to improved job satisfaction (Charles and others 2003).
The occupant surveys conducted as part of our larger study supported these earlier
findings, demonstrating significantly higher satisfaction for the occupants with the
multi-control direct-indirect system compared to those with conventional lighting.
Given previous research, it seems likely that this benefit can be attributed to the
individual control. The survey results will be discussed in more detail in a future
publication.
Table 1 illustrates that the energy savings from two or three controls used together
(os+ic, os+ls, ic+ls, os+ic+ls) are not equal to the sum of the savings of the controls
when used separately (os, ic, ls). This has a straightforward explanation, but one
which is often not appreciated. For example, if an occupancy sensor would save
40% on its own, and a light sensor 20% on its own, the saving with both sensors is
not 60% (40% + 20%), but somewhat less. The light sensor cannot contribute
energy savings during the period when the lighting has already been switched off
due to occupancy. Similarly, the occupancy sensor could not claim all of the savings
during absence if daylight harvesting would have reduced the lighting load during
that period anyway.
Note that we did not attempt to analyze the effect of the lighting energy savings on
the thermal energy use in the building. Reducing lighting energy use means
reducing the internal heat gains. During the cooling season, the lower internal gains
will reduce the cooling load, which is also an electrical end use. However, in the
heating season, the internal gains would have to be made up by the heating system,
which was fuelled by natural gas in this building. In general, the overall thermal
effect will depend strongly on the local climate, the building design and properties,
and the characteristics of the building HVAC system (Newsham and others 1998).
It is also important to note that the lighting system in the studied building was
maintained and operated by a highly qualified and motivated employee. This likely
played an important role in realizing energy savings of the magnitude reported here.
7 CONCLUSIONS
The results of this study are drawn from long-term data of a single building using a
particular lighting system. This lighting system was a workstation-specific, 3-lamp
direct-indirect system with integral occupancy and light sensors, and individual
dimming control. The three controls affected the light output of the two lamps
directed downwards, while the single lamp directed upwards was always on during
scheduled occupancy. The sample of luminaires investigated was smaller than the
one originally targeted and it is possible that the performance of the luminaires
23
assigned to people who did not volunteer their data for study was different from
those that were studied. Nevertheless, despite the specifics of this field study, we
believe the following findings are likely helpful to general office lighting practice:
Due to its reduced lighting power density alone, the direct-indirect,
workstation-specific, lighting system saved 42% in lighting energy use
compared to the static ceiling-recessed system it replaced. With all three
controls in operation, the lighting system saved an additional 42-47%, which
translates into energy savings of 70% compared to the conventional lighting
system.
If the three lighting controls systems had been installed separately,
occupancy sensors would have saved, on average, 35% if used alone, light
sensors (daylight harvesting) 20%, and individual dimming 11%. The light
sensor savings were, of course, higher in perimeter workstations, and would
have matched the performance of occupancy sensors on the perimeter if the
downlights had been permitted to dim to zero under light sensor control,
rather than to only half output.
There were concomitant reductions in peak power demand for lighting. The
three controls reduced the average daily peak power demand by 65-70%
compared to a conventional lighting system, and by 40-50% compared to the
installed fixtures used at full power. This is an important benefit for electric
utilities seeking to balance supply and demand at reasonable cost on high
demand days.
The peak power reductions due to the controls meant that the installed
lighting power density of the system (5.8 W/m2) was reduced to an effective
average lighting power density of only 3 W/m2.
Several other design/operation options that would have further reduced
lighting energy use were identified: allowing the downlights to dim to zero
instead of only 50% output when under light sensor control; reducing the
wattage of the uplights; or controlling the uplights identically to the downlights.
However, design options including the uplights risk greater ceiling non-
uniformity.
Occupant surveys conducted as part of our larger study demonstrated
significantly higher satisfaction for the occupants with the multi-control direct-
indirect system, compared to those with conventional lighting. Given previous
research, it seems likely that this benefit can be attributed, at least in part, to
the individual dimming control.
Although previous research has addressed various aspects of lighting control, this
study is unique in addressing three different control options simultaneously, over a
long-term field study, in an open-plan office, and with detailed occupant surveys (to
be reported in subsequent publications). Overall, the results of this field-study show
that lighting systems incorporating both automatic and personal controls have the
potential to realize considerable energy savings and peak power reductions in open-
24
plan environments, while being at the same time positively perceived by the
occupants. Demonstrating these findings in an open-plan setting is particularly
important because this is the most common office environment in North America,
and because recommended practice has often assumed that controls such as
occupancy sensors and individual dimming cannot be successfully implemented in
spaces without full-height walls.
ACKNOWLEDGEMENT
This project was a collaboration between the Institute for Research in Construction
(IRC) of the National Research Council of Canada (NRC), Program on Energy
Research and Development (PERD), Public Works and Government Services
Canada (PWGSC), BC Hydro Power Smart, and Ledalite Architectural Products.
The authors are grateful to Roy Hughes, Sunny Dhannu, Tyler Nitsch, Caroline
Ngan, Kevin White, Rico Luk, and Zorawar Bhatia, all of BC Hydro Power Smart, for
their support in the field implementation of the data acquisition systems and their
regular interaction with the building occupants. Special thanks are also due to Jindra
Ryvola and Ron Scott, of Ledalite Architectural Products, for customizing the lighting
system monitoring software for this project and for their valuable assistance with the
data interpretation. We would also like to acknowledge Cara Donnelly, Jennifer
Veitch, Chantal Arsenault, Roger Marchand and Christoph Reinhart, all of NRC-IRC,
for their contributions in all of the aspects of the project.
REFERENCES
ASHRAE. ANSI/ASHRAE/IESNA Standard 90.1-2004. Energy Standard for
Buildings Except Low-Rise Residential Buildings. American Society of Heating,
Refrigerating and Air-Conditioning Engineers, Atlanta, USA.
Bordass W, Heasman T, Leaman A, Perry MJ.1994. Daylight use in open plan
offices: The opportunities and the fantasies. Proceedings of National Lighting
Conference and Daylighting Colloquium. Cambridge. Robison College. London.
UK:CIBSE; p 243-256.
Boyce PR, Eklund NH, Simpson SN. Winter 2000. Individual lighting control: Task
performance, mood, and illuminance. Journal of the Illuminating Engineering
Society; p 131-142.
Boyce PR, Veitch JA, Newsham GR, Myer M, Hunter C. 2003. Lighting quality and
office work: A field simulation study. Richland. WA. USA. Pacific Northwest National
Laboratory. <http://irc.nrc-cnrc.gc.ca/pubs/fulltext/b3214.1/> last accessed April
2007.
Carter D, Slater A., Moore T. 1999. A study of occupier controlled lighting systems.
Proceedings of the 24th Session of the CIE. Warsaw. Poland.1(2): 108-110.
25
Charles KE, Veitch JA, Farley KMJ, Newsham GR. 2003. Environmental satisfaction
in open-plan environments: 3. Further scale validation (IRC-RR-152), Ottawa.
Canada. National Research Council of Canada. Institute for Research in
Construction. <http://irc.nrc-cnrc.gc.ca/pubs/fulltext/nrcc47630/> last accessed April
2007.
Escuyer S, Fontoynont M. 2001. Lighting controls: A field study of office workers’
reactions. Lighting Research and Technology. 33(2): 77-96.
Figueiro MG. January 2004. Occupancy Sensors: Are there reliable estimates of the
energy savings?. Lighting Design + Application (LD+A).
Heschong Mahone Group. 2006. Sidelighting photocontrols field study. Report
#06-152, prepared for the Northwest Energy Efficiency Alliance.
<http://www.nwalliance.org/research/reports/152.pdf> last accessed May 2007.
Jennings JD, Rubinstein FM, DiBartolomeo D, Blanc SL. 1999. Comparison of
control options in private offices in an advanced controls testbed. Ernest Orlando
Lawrence Berkeley National Laboratory. Environmental Energy Technologies
Division. LBNL-43096.
Jennings JD, Rubinstein FM, DiBartolomeo D, Blanc SL. 2000. Comparison of
control options in private offices in an advanced controls testbed. Journal of the
Illuminating Engineering Society. 29(2):39-60.
Lee ES, Selkowitz SE. 2006. The New York Times Headquarters daylighting mock-
up: Monitored performance of the daylighting control system. Energy and Buildings.
38(7): 914-929.
Love J, 1995. Field performance of daylighting systems with photoelectric controls.
Proceedings of the 3rd European Conference on Energy-Efficient Lighting – Right
Light 3. Newcastle. England. p 75-82.
Maniccia D, Rutledge B, Rea MS, Morrow W. 1999. Occupant use of manual lighting
controls in private offices. Journal of the Illuminating Engineering Society. 28(2): 42-
56.
Moore T, Carter DJ, Slater AI. 2001. A comparative study of user opinion in offices
with and without individually controlled lighting. Proceedings of the 9th European
Lighting Conference (Lux Europa). Reykjavik. Iceland. p 234-241.
Moore T, Carter DJ, Slater AI. 2002. A field study of occupant controlled lighting in
offices. Lighting Research and Technology. 34(3):191-205.
26
Moore T, Carter DJ, Slater AI. 2002. User attitudes toward occupant controlled office
lighting. Lighting Research and Technology. 34(3): 207-219.
Moore T, Carter DJ, Slater AI. 2003. Long-term patterns of use of occupant
controlled office lighting. Lighting Research and Technology. 35(1):43-59.
Natural Resources Canada (NRCan). August 2006. Energy Use Data Handbook,
1990 and 1998 to 2004.
<http://oee.nrcan.gc.ca/corporate/statistics/neud/dpa/data_e/handbook05/Datahand
book2005.pdf> last accessed April 2007
Newsham GR, Mahdavi A, Mathew P, Cornick SM, Sander DM, Brahme R. 1998.
Impact of the adoption of efficient electrical products and control technologies on
office building energy use. ASHRAE Transactions. 104(2): 286-298, 1998.
Newsham GR, Veitch JA, Arsenault C, Duval C. 2004. Effect of dimming control on
office worker satisfaction and performance. Proceedings of the IESNA Annual
Conference. Tampa. FL. USA. p 19-41. <http://irc.nrc-cnrc.gc.ca/fulltext/nrcc47069/>
last accessed April 2007
Newsham GR, Mancini S. 2006. The potential for demand-responsive lighting in
non-daylit offices, Leukos, 3(2):105-120.
Pigg S, Eilers M, Reed J. 1996. Behavioral aspects of lighting and occupancy
sensors in private offices: A case study of a university office building. Proceedings of
ACEEE Summer Study on Energy Efficiency in Buildings. Pacific Grove. USA. p
8.161-8.170.
Roche L, Dewey E, Littlefair P. 2000. Occupant reactions to daylight in offices.
Lighting Research and Technology. 32(3):119-126.
Slater A. 1995. Occupant use of lighting controls: A review of current practice,
problems, and how to avoid them. Proceedings of the CIBSE National Conference.
Eastbourne. London. UK. p 204-209.
Slater A. 1996. Lighting controls in offices: How to improve occupant comfort and
energy efficiency. Proceedings of the CIBSE National Lighting Conference. Bath.
UK. p 178-184.
Slater A, Carter D. 1998. A field study of lighting levels in offices. Proceedings of the
CIBSE National Lighting Conference. London. UK. p 23-33.
Slater A, Carter D, Moore T. 1998. A study of lighting in offices equipped with
occupant controlled systems. Proceedings of the 1st CIE Symposium on Lighting
Quality. Ottawa. Canada. p 219-227.
27
Veitch JA, Newsham GR. 2000. Preferred luminous conditions in open-plan offices:
Research and practice recommendations. Lighting Research and Technology,
32:199-212.
Wyon DP. 1999. Enhancing productivity while reducing energy use in buildings.
RAND Conference Proceedings.
<http://www.rand.org/pubs/conf_proceedings/CF170.1-1/CF170.1.wyon.pdf> last
accessed April 2007
28
... It had been argued that this approach is not only the most prevalent, but is also the most successful control strategy used to reduce energy from electric lighting [7]. Harnessing daylight to displace electrical light is another approach, but has not yet received the same widespread implementation seen with solid-state lights [8,9]. Nonetheless, standards [10][11][12] specify its implementation so that buildings are capable of reducing electric lighting loads through effective fenestration, building design, and management of daylight admittance. ...
... These reasons could explain why ceiling sensors tend to save significantly less energy compared to other sensor positions (i.e., facing either window, ceiling, sideways, or outside the building) [44]. However, in spaces with relatively high daylight availability, Galasiu et al. [9] office cubicles with sensors mounted to light fixtures generated up to 20 % energy savings. Therefore, location alone may not always be the only significant determiner for sensor performance, and other conditions should also be considered. ...
Article
Full-text available
Daylight can reduce electric lighting in buildings. This is facilitated by sensors that relay real-time illuminance data to a light controller. When daylight provides greater than required or desired levels of illuminance, control actions enable electric lights to reduce their output and save energy. Occupant behaviours can block desk sensors and this reduces the amount of energy saved. However, no method exists that can be used to continuously monitor sensors to ensure they operate as intended (e.g. remain unblocked). We carried out a study in an open-plan office building in Singapore, consisting of 39 workstations each fitted with desk illuminance sensors independently controlling a dedicated ceiling light. Power over Ethernet was used to collect individual data signals for both illuminance and power from each workstation. Data were collected across a one month period, sampling signals at every 2-minute interval. A linear support vector machine model accurately classified 99% of the data points using our sensor blocking algorithm. From 447445 data points analysed, 12% of dataset showed that sensors were blocked and this had an estimated energy penalty of 24%. We do not recommend installing illuminance sensors at the desk. Our study highlights the usefulness of Power over Ethernet for closed-loop daylight harvesting. The data collected can be used to monitor the health of the sensors’ performance to help minimise energy use.
... These design scenarios are based on the literature, building codes and standards, and real-time field observations of IoT-based smart building systems. There are nine design scenarios in total (Table 9), with the following characteristics: a) Lighting system integration: Savings of 10-40% arise through the integration of the sensor to the lighting systems and are an estimation based on values reported by the literature [98][99][100][101][102][103][104][105]. USGBC LEED [105], and ASHRAE 90. ...
Article
Wireless Sensing Systems based on the Internet of Things (IoT) are increasingly being deployed in smart buildings to reduce their energy consumption and environmental impacts. However, the actual environmental costs of integrating IoT-based sensing systems into smart buildings have rarely been considered when reporting on their overall sustainability footprint. This paper presents an Emergy-based methodological approach to ascertain the effectiveness of integrating IoT-based sensing systems into smart buildings. The proposed method employs three novel Emergy Neutrality Indices (ENIs): The Sensing System to Building Emergy (SBEm) ratio, the Emergy Neutrality Score (EmN), and the Emergy Payback Time (EmPBT). The Solar House at Gainesville, Florida, retrofitted with an IoT-based sensing system, is used as an example building for computing these indices. A Building Energy Model (BEM) is created using the EnergyPlusTM simulation engine available in DesignBuilder V7 to determine the operational Emergy of the building. A custom-built CO2 sensor is used to collect the real-time occupancy data for incorporation into the BEM. The Emergy Neutrality computation demonstrates that the integration of the given sensing system is effective in the Solar House, with SBEm, EmN, and EmPBT values of 2.06E-03, 0.28, and 3.61 years, respectively. By varying the sensors’ energy-savings potential and spatio-functional configurations, nine design scenarios are generated and evaluated for their environmental performance using Emergy ratios and the proposed ENIs. Finally, the Solar House is compared with the Rinker Hall to emphasise the significance of reporting these indices. In the future, designers, architects, and other stakeholders could use these ENIs as useful sustainability assessment tools to evaluate the environmental effectiveness of integrating these smart sensing systems into buildings.
... With the development of digital analysis, camera-based occupant behavior recognition is also used for lighting control (Pham, Nguyen, & Kwon, 2019). Furthermore, each lighting circuit can be operated separately in office buildings, according to the occupant behavior (Galasiu, Newsham, Suvagau, & Sander, 2007). The lighting control system based on the occupant behavior is expected to save energy by approximately 60% (de Bakker, Aries, Kort, & Rosemann, 2017). ...
Chapter
Full-text available
The lighting system accounts for 8% of the total electricity consumption in commercial buildings in the United States and 12% of the total electricity consumption in public buildings globally. This consumption share can be effectively reduced using the demand-response control. The traditional lighting system control method commonly depends on the real-time occupancy data collected using the passive infrared (PIR) sensor. However, the detection inaccuracy of the PIR sensor usually results in false-offs. To diminish the false-error frequency, the existing lighting system control simply deploys a delayed reaction period (e.g., 5–20 min), which is not sufficiently accurate for the demand-response operation. Therefore, in this research, a novel data-driven model predictive control (MPC) method that is based on the temporal sequential-based artificial neural network (TS-ANN) is proposed to overcome this challenge using an updated historical occupancy status. Using an office as case study, the proposed model is also compared with the traditional lighting system control method. In the proposed model, the occupancy data was trained to predict the occupancy pattern to improve the control. It was found that the occupancy prediction mainly correlates with the historical occupancy ratio and the time sequential feature. The simulation results indicated that the proposed method achieved higher accuracy (97.4%) and fewer false-offs (from 79.5 with traditional time delay method to 0.6 times per day) are achieved by the MPC model. The proposed TS-ANN-MPC method integrates the analysis of the occupant behaviour routine into on-site control and has the potential to further enhance the control performance practice for maximum energy conservation.
Chapter
The chapter describes the authors’ view on how to create an effective educational environment that inspires architectural students to design a sustainable architecture with a special focus on zero net energy (ZNE) building design. Such an environment is known as the integrated design studio (IDS). It is meant to train students in an environment that imitates closely the actual process that takes place in the practice of building design. The main educational challenge is how to prepare the student to become proficient in a highly engaged comprehensive design process that integrates creative architectural thinking with technical design of building energy and environmental systems, structural system, construction system, and building materials. This challenge becomes much harder when all these requirements are applied in the design of a complex multifunction ZNE building located under the extreme hot climate conditions of Dubai, UAE. The established model by the authors, which is believed to have high potentials in improving architectural design education, is demonstrated in this chapter by a real case of students’ project that won the third prize of the reputable international MultiComfort House Student Contest, Edition 2018, by Saint Gobain.KeywordsCooperative learningCreative thinkingDeduction discovery method (DDM)DubaiGreen buildingInduction discovery method (IDM)Integrated design studio (IDS)Performative designSustainable architectureZero net energy (ZNE)
Article
Designing a facade that is able to keep the balance between the daylight, view out and energy load is a challenge. The present paper aims to use the sun’s radiant energy utilizing the facade as the outer skin, which is in contact with the surroundings to reduce nonrenewable energy consumption and provide comfort to occupants. Using bionic science, a kinetic facade prototype was designed and fabricated with Lupinus Succulentus plants’ movement mechanism, a type of sun-tracking plant. This kinetic façade has been simulated by Grasshopper on the facade of the Southern side of an open office in Tehran. Accordingly, this study is to investigate a bio-kinetic facade, using the integration among kinetic architecture, biomimicry and occupants’ visual and thermal comfort. Bio-kinetic façade not only causes cooling load reduction by approximately 7% but also prevents the direct light entry, resulting in electric lighting load’s reduction by about 48%, which was calculated by Honey bee and Ladybug plugins.
Conference Paper
Full-text available
The aim of the study is to determine the effects of tea type, tea concentration and infusion time on the Aluminum (Al) content in tea infusions. Ceylon and Turkish black tea, Ceylon and Turkish green tea infusions were prepared by adding 100 mL of boiling distilled water to 1, 2 and 3 g of tea leaves. The tea infusion was mixed using a glass rod, covered by watch glasses to extract the components of tea leaves and allowed to brew for 2, 5, 10, 20, 30, 45 and 60 minutes. The infusion was filtered through filter paper (Whatman 42, 125 mm) into test tube, cooled, acidified and analyzed immediately for Al. The results showed that the Al content in tea infusion were significantly affected by tea type, tea concentration, infusion time, interactions of tea type x tea concentration, tea type x infusion time, tea concentration x infusion time and tea type x tea concentration x infusion time (P<0.001).
Conference Paper
Full-text available
The aim of the study is to determine the effects of tea type, tea concentration and infusion time on the manganese (Mn) content in tea infusions. Ceylon and Turkish black tea, Ceylon and Turkish green tea infusions were prepared by adding 100 mL of boiling distilled water to 1, 2 and 3 g of tea leaves. The tea infusion was mixed using a glass rod, covered by watch glasses to extract the components of tea leaves and allowed to brew for 2, 5, 10, 20, 30, 45 and 60 minutes. The infusion was filtered through filter paper (Whatman 42, 125 mm) into test tube, cooled, acidified and analyzed immediately for Mn. The results showed that the manganese content in tea infusion were significantly affected by tea type, tea concentration, infusion time, interactions of tea type x tea concentration, tea type x infusion time, tea concentration x infusion time (P<0.001) and tea type x tea concentration x infusion time (P<0.01).
Article
In the promotion of energy efficiency in educational facilities, preschool children play an important role since simple energy-saving behaviour can be developed in early childhood. This study aimed to investigate the effects of behaviour interventions, with a particular focus on preschool children’s turn-off behaviour in shared environments. The intervention tools i.e. digital game, paper-board game and visual prompt were developed for 3-6 years old children. The study took place in a preschool in which energy-efficient lighting and combined automatic and manual controls were installed. The data were collected by means of participatory observations of the children’s light-off behaviour (N = 42), measurements of lighting use using data loggers, and interviews with the teachers (N = 11). The overall results point to associations between the interventions and the behaviour specifically, the visual prompts were found to have a significantly positive effect. No significant effect of the games on the behaviour could be identified however, the games were found to have the potential to promote learning about lighting and lighting use. Effects of the interventions on the long-term behaviour and energy saving remain unclear. Design improvements of the interventions are thus, needed considering other related factors (e.g. social norms, habits) together with lighting quality for young children and activities in preschool environments.
Article
Full-text available
Technological advances have increased the energy performance of integrated daylighting and lighting systems to unprecedented levels. The shift to LED technologies, innovative shadings, and advanced lighting and shading controls promise a dramatic reduction in energy use for lighting. Work will continue to save more energy with technological solutions, but additional savings can be attained by understanding the complexity of human behaviour and by designing systems in such a way that they can accommodate energy-saving behaviour. This article reviews behaviour aimed at saving energy used for lighting. Four categories of behaviour are identified, concerning extent and speed of dimming, heuristics, design of the interface, and information/feedback strategies. The review suggests that lighting use can be halved by introducing simple strategies to existing and future systems, but the energy saving potential comes with low ecological validity. A multidisciplinary approach is required to increase the robustness of findings.
Article
In this paper, we aim to clarify the effectiveness of a thermal‐based sensor, which applies image recognition technique to thermal image to measure occupancy state. Our experiment, which was conducted at a single private office over 8 days, evaluated both energy‐saving rate of occupancy control and F‐measure, which is the comprehensive metric of measurement accuracy. The thermal‐based sensor increased the F‐measure and energy‐saving rate by 2.23% and 27.91%, respectively, compared with the passive infrared sensor, which measures the occupancy state on the basis of change in infrared temperature. Compared with an image‐based sensor, which applies image recognition technique to visible image, it can provide high privacy and security preserving capability and detect occupants under completely dark environment. However, the F‐measure and energy‐saving rate were decreased by 3.13% and 19.07%, respectively. Further results and discussion are included herein toward analysis and improvement of the thermal‐based sensor. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.
Article
In a major test of different lighting control technologies in a typical office building, we present analyses of seven months' results from five control scenarios in private offices. We compare the energy savings and effectiveness of various combinations of occupant detection, daylight dimming, and switching techniques. Comparing measured energy use with occupant sensors against baseline energy use calculated using wall switch operation only, we found occupant sensors saved 20-26 percent lighting energy compared to manual switching alone. In offices where light sensor controls were: installed and properly commissioned, additional savings up to 27 percent for a total of 46 percent were obtained over a seven-month period, even in an area with unusually high minimum lighting requirements. Dimming the lighting system to desired task levels (task tuning) also resulted in significant (23 percent additional, 43 percent total) energy savings in over lighted areas. On the base case floor, where only bi-level switches were installed, we found significant usage of only one switch resulting in an additional 23 percent savings over single-level switches, an unexpected result with implications for building code requirements. We found the energy savings due to occupant sensing vs. dimming depended on the behavior of occupants. In offices whose occupants tended to stay at their desks all day, dimming controls saved more energy and vice versa. The lighting requirements of occupants appear to depend on their type of work.
Article
This study examines the way people choose to use individual control of lighting, the effect of having such control on the illuminance selected for different activities and the performance of those activities and how people feel about having such control. The experiment took place in three small, windowless, private offices, one lit to a maximum of 1240 lx, another to a maximum of 680 lx, both these offices having a handheld dimming system. The third office was lit to a fixed illuminance of 490 lx. Eighteen subjects worked for one eight-hour day in each of the offices. During the day they carried out an alphanumeric verification task in forms which were visually difficult and visually easy, on paper and on a computer, as well as doing some casual reading. The subjects in the offices equipped with individual lighting control could use the control whenever they liked. The data collected showed that the subjects did use the individual lighting control to adjust the illuminance for different tasks, but there were wide differences between individuals. Use of the lighting controls saved energy. The ability to dim the lighting was considered highly desirable and made the tasks seem less difficult. However, the ability to control the lighting did not put the subjects in a more positive mood, did not make them more alert, and did not affect the performance on the task.
Article
This paper reports daily and seasonal patterns of lighting use in offices where occupants were able to vary the level of illuminance falling on their working areas. The results show that occupants set a wide range of illuminances, many of which are significantly below CIBSE Code recommendations leading to significant energy savings throughout the year. Although users report use of controls for a variety of reasons, the results of the long-term monitoring of switching behaviour reveals that very few in fact use them for anything other than to switch on upon arrival at work, with further use of systems during the day being rare. Evidence suggests that the way in which systems are configured exerts influence over the level of electric light occupants work under. There is some evidence of an effect of daylight availability on levels set but little evidence to suggest pervasive and consistent user preferences for levels of electric light.
Article
In a major test of different lighting control technologies in a typical office building, we present analyses of seven months' results from five control scenarios in private offices. We compare the energy savings and effectiveness of various combinations of occupant detection, daylight dimming, and switching techniques. Comparing measured energy use with occupant sensors against baseline energy use calculated using wall switch operation only, we found occupant sensors saved 20-26 percent lighting energy compared to manual switching alone. In offices where light sensor controls were installed and properly commissioned, additional savings up to 27 percent for a total of 46 percent were obtained over a sevenmonth period, even in an area roth unusually high minimum lighting requirements. Dimming the lighting system to desired task levels (task tuning) also resulted in significant (23 percent additional, 43 percent total) energy savings in over lighted areas. On the base case floor, where only bi-level switches were installed, we found significant usage of only one switch resulting in an additional 23 percent savings over single-level switches, an unexpected result with implications for building code requirements. We found the energy savings due to occupant sensing vs. dimming depended on the behavior of occupants. In offices whose occupants tended to stay at their desks all day, dimming controls saved more energy, and vice versa. The lighting requirements of occupants appear to depend on their type of work.
Article
,!This paper describes a survey of user attitudes toward control systems and the luminous conditions they produce in 14 similar UK office buildings—seven sites with user controlled general lighting and seven sites without. The installations without user control had better conditions when judged in terms of conventional recommendations for task illuminance, and luminance ratios and levels. User opinion in installations with user control were more satisfied with planar illuminance and had more positive attitudes toward control, whilst other aspects relating to lighting quality were assessed similarly in the two types of installation. Given that the controllable systems were typically operated at 50% of maximum output it appears these systems offer considerable scope for energy saving without compromising occupants’ perceived lighting quality.