Technical ReportPDF Available

Seattle LED Adaptive Lighting Study - NEEA

Authors:
  • Clanton & Associates, Inc.

Abstract and Figures

The Northwest Energy Efficiency Alliance (NEEA) and the City of Seattle partnered to evaluate the future of solid state street lighting in the Pacific Northwest with a two-night demonstration in Seattle’s Ballard neighborhood in March 2012. The study evaluates the effectiveness of LED streetlights on nighttime driver object detection visibility as function of light source spectral distribution (color temperature in degrees K) and light distribution. Clanton & Associates and VTTI also evaluated adaptive lighting (tuning of streetlights during periods of reduced vehicular and pedestrian activity) at three levels: one hundred percent of full light output, fifty percent of full light output, and twenty-five percent of full light output. The study, led by Clanton & Associates, Continuum Industries, and the VTTI, built upon previous visual performance studies conducted in Anchorage, Alaska; San Diego, California; and San Jose, California. The use of LED technology for city street lighting is becoming more widespread. While these lights are primarily touted for their energy efficiency, the combination of LEDs with advanced control technology, changes to lighting criteria, and a better understanding of human mesopic (low light level) visibility creates an enormous potential for energy savings and improved motorist and pedestrian visibility and safety. Data from these tests support the following statements: • LED luminaires with a correlated color temperature of 4100K provide the highest detection distance, including statistically significantly better detection distance when compared to HPS luminaires of higher wattage. • The non-uniformity of the lighting on the roadway surface provides a visibility enhancement and greater contrast for visibility. • Contrast of objects, both positive and negative, is a better indicator of visibility than is average luminance level. • Dimming the LED luminaires to fifty percent of IES RP-8 levels did not significantly reduce object detection distance in dry pavement conditions. • Participants perceived dimming of sidewalks as less acceptable than dimming to the same level on the roadway. • Asymmetric lighting did reduce glare and performed similarly to the symmetric lighting at the same color temperature (4100K).
Content may be subject to copyright.
May 29, 2014
REPORT #E14-286
Seattle LED Adaptive
Lighting Study
Prepared by:
Clanton & Associates, Inc.
4699 Nautilus Ct. So. #102
Boulder, CO 80301
Northwest Energy Efficiency Alliance
PHONE
503-688-5400
FAX
503-688-5447
EMAIL
info@neea.org
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Table of Contents
1 Executive Summary.......................................................................................................... i
1.1 Project Background ............................................................................................................................ i
1.2 Study Description ............................................................................................................................... i
1.3 Research Results .............................................................................................................................. ii
1.4 Industry Implications......................................................................................................................... iv
2 Introduction ........................................................................................................................1
2.1 Background ........................................................................................................................................ 1
2.2 Technology and Market Overview .................................................................................................... 2
2.3 Project Objectives.............................................................................................................................. 3
2.4 Project Hypotheses ........................................................................................................................... 3
3 Methodology ......................................................................................................................6
3.1 Overall Project Setup ........................................................................................................................ 6
3.2 Site Selection ..................................................................................................................................... 7
3.3 Public Outreach ................................................................................................................................. 8
3.4 Lighting Criteria.................................................................................................................................. 9
3.5 Luminaire Selection ......................................................................................................................... 10
3.6 Controls Selection ........................................................................................................................... 10
3.7 Asymmetric Luminaire Design ........................................................................................................ 11
3.8 Road Conditions .............................................................................................................................. 13
3.9 Light Output Level ........................................................................................................................... 14
3.10 Participant Recruitment ................................................................................................................... 14
3.11 Written Evaluation ........................................................................................................................... 15
3.12 User Field Test ................................................................................................................................ 16
4 Procedure ........................................................................................................................ 19
4.1 Equipment Pretesting ...................................................................................................................... 19
4.2 Equipment Installation ..................................................................................................................... 22
4.3 Setup of Visibility Targets ............................................................................................................... 23
4.4 Written Evaluations and User Field Tests ..................................................................................... 24
4.5 Experimental Protocol ..................................................................................................................... 25
4.6 Dry Pavement .................................................................................................................................. 25
4.7 Wet Pavement ................................................................................................................................. 26
4.8 Luminance Measurements ............................................................................................................. 27
5 Findings ........................................................................................................................... 28
5.1 Written Evaluation ........................................................................................................................... 28
5.2 User Field Test ................................................................................................................................ 32
5.3 Contrast ............................................................................................................................................ 37
5.4 Illuminance and Detection Distance ............................................................................................... 41
5.5 Lighting Metrics................................................................................................................................ 43
5.7 Sidewalk Lighting Characteristics................................................................................................... 49
5.8 Light Trespass ................................................................................................................................. 51
5.9 Glare ................................................................................................................................................. 52
5.10 Spectral Power Distribution ............................................................................................................ 55
6 Discussion ....................................................................................................................... 58
6.1 Comparison to Previous Studies .................................................................................................... 58
6.2 Adaptive Lighting Opportunities ..................................................................................................... 59
6.3 Future Design Standards ................................................................................................................ 60
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
6.4 Economic Analysis .......................................................................................................................... 60
7 Conclusions ..................................................................................................................... 68
7.1 Written Evaluation ........................................................................................................................... 68
7.2 User Field Test ................................................................................................................................ 69
7.3 Color Temperature .......................................................................................................................... 71
7.4 Pavement Conditions ...................................................................................................................... 72
7.5 Asymmetric Design ......................................................................................................................... 73
7.6 Control Systems .............................................................................................................................. 73
7.7 Lessons Learned ............................................................................................................................. 74
8 References ...................................................................................................................... 75
Appendix A: Prior Work ......................................................................................................... 76
Appendix B: Written Evaluation Form .................................................................................. 78
Appendix C: Written Evaluation Comments ......................................................................... 80
Appendix D: Product Specifications ................................................................................... 110
Appendix E: Preliminary Luminaire Testing ....................................................................... 115
Appendix F: Luminance Calculations ................................................................................. 133
Appendix G: Procedure ....................................................................................................... 166
Appendix H: Written Evaluation Findings ........................................................................... 170
Appendix I: Written Evaluation Results Duplicate Participant Analysis ....................... 172
Appendix J: User Field Test Results .................................................................................. 177
Appendix K: Public Outreach .............................................................................................. 183
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
List of Tables
Table 1. Potential Cumulative Energy Savings ............................................................................2
Table 2. Summary of Test Areas .................................................................................................7
Table 3. IES RP-8 Luminance Criteria for New LED Streetlights ................................................9
Table 4. Percent Reflectance by Target Color ............................................................................ 17
Table 5. Summary of Lab Light Trespass .................................................................................. 21
Table 6. Dry Pavement Written EvaluationTest Numbers .......................................................... 25
Table 7. Dry Pavement Participants User FieldTest Numbers and Computer Conditions ........... 25
Table 8. Wet Pavement Written EvaluationTest Numbers ......................................................... 26
Table 9. Wet Pavement Participants User Field Test Numbers and Computer Conditions .......... 26
Table 10. Recorded Luminance Data ......................................................................................... 27
Table 11. Written Evaluation Results ........................................................................................ 28
Table 12. Luminaire Type, Target Color, Pavement, and Light Level ANOVA Results ............. 32
Table 13. Light System Calculations, One Hundred percent Light Level ................................... 47
Table 14. Light System Calculations, Fifty percent Light Level ................................................ 48
Table 15. Light System Calculations, Twenty-Five percent Light Level .................................... 48
Table 16. Veiling Luminance, Dry Pavement ............................................................................ 54
Table 17. Veiling Luminance, Wet Pavement ............................................................................ 55
Table 18. Scenario 1A Economic Analysis Summary ................................................................ 62
Table 19. Scenario 1B Economic Analysis Summary ................................................................ 62
Table 20. Scenario 1C Economic Analysis Summary ................................................................ 63
Table 21. Scenario 2A Economic Analysis Summary ................................................................ 63
Table 22. Scenario 2B Economic Analysis Summary ................................................................ 64
Table 23. Scenario 2C Economic Analysis Summary ................................................................ 64
Table 24: Summary of Lab Illuminance .................................................................................. 121
Table 25: IES Light Trespass Limitations ................................................................................ 121
Table 26. Summary of Dimming Pretesting ............................................................................. 124
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
List of Figures
Figure 1. Demonstration Site Layout ...........................................................................................6
Figure 2. Preliminary Site Visit to Observe Other Sources of Illumination ..................................8
Figure 3. Elevation of Conceptual Design of Asymmetric Luminaire ........................................ 11
Figure 4. Preliminary Asymmetric Luminaire Design by Philips Lumec.................................... 12
Figure 5. Second Asymmetric Luminaire Design ...................................................................... 13
Figure 6. Visibility Targets Used within Test Areas .................................................................. 17
Figure 7. Laboratory Evaluation Grid ........................................................................................ 19
Figure 8. Dimming Voltage by % Light Output ......................................................................... 20
Figure 9. Dimming Voltage by Illuminance (lx) ........................................................................ 21
Figure 10. Demonstration Site Layout ....................................................................................... 22
Figure 11. Example of Handwritten Description ........................................................................ 23
Figure 12. Flusher Truck Wetting the Roads ............................................................................. 26
Figure 13. Survey Question 11: “I like the color of the light”– Dry Pavement ........................... 31
Figure 14. Survey Question 11: “I like the color of the light”– Wet Pavement ........................... 31
Figure 15. Luminaire Type and Light Level by Detection Distance (Wet and ............................ 33
Figure 16. Luminaire Type and Pavement Condition by Detection Distance .............................. 34
Figure 17. Pavement Condition and Light Level by Detection Distance (All ............................. 34
Figure 18. Luminaire Type and Target Color by Detection Distance (All Light ......................... 36
Figure 19. Luminaire Type, Pavement, and Light Level by Detection Distance ......................... 37
Figure 20. Example Luminance Image ...................................................................................... 38
Figure 21. Target Contrast of One Hundred percent Lighting Level, Dry ................................... 39
Figure 22. Target Contrast of Fifty percent Lighting Level, Dry Pavement ................................ 40
Figure 23. Target Contrast of Twenty-Five percent Lighting Level, Dry.................................... 40
Figure 24. Impact of Headlamps by Distance from Vehicle ....................................................... 41
Figure 25. Luminaire Type and Light Level by Detection Distance for Both Wet and Dry
Conditions ................................................................................................................................. 42
Figure 26. Luminaire Type by Detection Distance across Both Pavement.................................. 43
Figure 27. Illuminance per Section at One Hundred percent, Northbound .................................. 44
Figure 28. Illuminance per Section at One Hundred percent, Southbound .................................. 44
Figure 29. Illuminance per Section at Fifty percent, Northbound ............................................... 45
Figure 30. Illuminance per Section at Fifty percent, Southbound ............................................... 45
Figure 31. Illuminance per Section at Twenty-Five percent, Northbound ................................... 46
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Figure 32. Illuminance per Section at Twenty-Five percent, Southbound ................................... 46
Figure 33. Average Roadway Luminance: Dry Pavement Condition ......................................... 49
Figure 34. Vertical Illuminance by Dim Level and Luminaire Type .......................................... 50
Figure 35. Average Sidewalk Luminance by Lighting System and Brightness Level ................. 51
Figure 36. Light Trespass at Property Line by Lighting System and Brightness Level ............... 52
Figure 37. Test Area and Light Level by Mean Vertical Illuminance (lx) of Glare ..................... 53
Figure 38. Spectral Power Distributions of LED Luminaires ..................................................... 57
Figure 39. 400 W HPS Replaced with 105 W LED (Scenario 1A) ............................................. 65
Figure 40. 250 W HPS Replaced with 105 W LED (Scenario 2A) ............................................. 66
Figure 41. Survey Question 8: “It would be safe to walk on the sidewalk here at night.” ........... 69
Figure 42: Simplified Response to a Signal by a Human ........................................................... 70
Figure 43. Survey Question 11: “I like the color of the light” – Dry Pavement .......................... 72
Figure 44. Survey Question 11: “I like the color of the light” – Wet Pavement .......................... 72
Figure 45: Vertical illuminance meter on cart .......................................................................... 116
Figure 46: 4100K Type II Grid - Vertical Illuminance ............................................................. 117
Figure 47. ASYM Grid - Vertical Illuminance ......................................................................... 118
Figure 48. 4100K Type II Grid - Horizontal Illuminance ......................................................... 119
Figure 49. Asymmetric Grid - Horizontal Illuminance ............................................................. 120
Figure 50: 4100K - Light Trespass - House Side ..................................................................... 122
Figure 51. 4100K - Light Trespass - Street Side ...................................................................... 123
Figure 52. 3500K Grid - Vertical Illuminance ......................................................................... 125
Figure 53. 5000K Grid - Vertical Illuminance ......................................................................... 126
Figure 54. 3500K Grid - Horizontal Illuminance ..................................................................... 127
Figure 55. 5000K Grid - Horizontal Illuminance ..................................................................... 128
Figure 56. ASYM - Light Trespass - House Side ..................................................................... 129
Figure 57. ASYM - Light Trespass - Street Side ...................................................................... 130
Figure 58. 3500K - Light Trespass - House Side ..................................................................... 130
Figure 59. 3500K - Light Trespass - Street Side ...................................................................... 131
Figure 60. 5000K - Light Trespass - House Side ..................................................................... 132
Figure 61. 5000K - Light Trespass - Street Side ...................................................................... 132
Figure 62. Question 3 Wet Pavement.................................................................................... 172
Figure 63. Question 5 Wet Pavement.................................................................................... 173
Figure 64. Question 6 Dry Pavement .................................................................................... 173
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Figure 65. Question 7 Dry Pavement .................................................................................... 174
Figure 66. Question 10 Wet Pavement .................................................................................. 175
Figure 67. Question 12 Wet Pavement .................................................................................. 176
Figure 68. Question 13 Wet Pavement .................................................................................. 176
Figure 69. Dry Pavement at One Hundred percent Light Output .............................................. 177
Figure 70. Dry Pavement at Fifty percent Light Output ........................................................... 177
Figure 71. Dry Pavement at Twenty-five percent Light Output ................................................ 178
Figure 72. Wet Pavement at One Hundred percent Light Output ............................................. 179
Figure 73. Wet Pavement at Fifty percent Light Output ........................................................... 179
Figure 74. Wet Pavement at Twenty-five percent Light Output ............................................... 180
Figure 75. Wet vs. Dry Pavement at One Hundred percent Light Output ................................. 180
Figure 76. Wet vs. Dry Pavement at Fifty percent Light Output .............................................. 181
Figure 77. Wet vs. Dry Pavement at Twenty-five percent Light Output ................................... 181
Figure 78. Dry Pavement at Each Light Level ......................................................................... 182
Figure 79. Wet Pavement at Each Light Level ......................................................................... 182
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - i
1 Executive Summary
1.1 Project Background
The Northwest region operates approximately 1.7 million streetlights consuming an approximate
average 150 MW. The City of Seattle has been actively converting its existing High Pressure
Sodium (HPS) street lighting to light emitting diode (LED) lighting. The City reduced energy
consumption by more than forty percent through this process (Smalley 2012). LED streetlights,
coupled with controls enabling adaptive lighting, can save an additional twenty-five percent of
energy.
The City can realize street light energy savings from both a reduction in wattage and from
dimming. LED lamps are approaching the efficacy (lumens/watt) of HPS. Clanton & Associates
and Virginia Tech Transportation Institute (VTTI) designed this study to test the idea that a
lower quantity of better-quality light provides equal or better detection distance. This would
create an opportunity for savings from luminaire lumen reductions and from dimming.
The Northwest Energy Efficiency Alliance (NEEA) and the City of Seattle partnered to evaluate
the future of solid state street lighting in the Pacific Northwest with a two-night demonstration in
Seattle’s Ballard neighborhood in March 2012. The study evaluates the effectiveness of LED
streetlights on nighttime driver object detection visibility as function of light source spectral
distribution (color temperature in degrees K) and light distribution. Clanton & Associates and
VTTI also evaluated adaptive lighting (tuning of streetlights during periods of reduced vehicular
and pedestrian activity) at three levels: one hundred percent of full light output, fifty percent of
full light output, and twenty-five percent of full light output.
The study, led by Clanton & Associates, Continuum Industries, and the VTTI, built upon
previous visual performance studies conducted in Anchorage, Alaska; San Diego, California; and
San Jose, California.
1.2 Study Description
Clanton & Associates and VTTI conducted the demonstration in Seattle’s Ballard neighborhood
along 15th Avenue NW, between NW 65th Street and NW 80th Street. They divided the fifteen-
block stretch into six evaluation test areas with approximately one test area per two blocks. The
demonstration used Philips Lumec LED luminaires equipped with the Schreder Owlet lighting
control system.
Clanton & Associates and VTTI conducted the data collection demonstration over two evenings.
Following an initial evening without data collection to allow representatives from the City and
media to view the demonstration and the capabilities of the system, researchers conducted
qualitative and quantitative testing the following two evenings. Each evening, three groups of
participants evaluated the entire test site. The first group evaluated all of the lights at one
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - ii
hundred percent of full light output, the second group evaluated all of the LED lights at fifty
percent of full light output (HPS remained at one hundred percent), and the final group evaluated
all of the LED lights at twenty-five percent of full light output (HPS remained at one hundred
percent). Full output of the luminaires represented the maximum output of the specified
luminaires, which Clanton & Associates had selected to meet the Illuminating Engineering
Society of North America (IES) Recommended Practice for Roadway Lighting (RP-8) criteria for
a collector road with medium pedestrian conflict. The first night of general participant testing
took place on dry pavement. On the second night, flusher trucks wetted the pavement and the
process was repeated.
The written evaluation asked participants a series of general questions based upon where they
live, in addition to demographic questions and site condition questions. For each test area, a
participant next rated twelve statements on a five-point scale (strongly disagree to strongly
agree).
VTTI conducted the user field test on both nights of general participant testing. On each evening,
three participants at a time participated in the user field test. Two participants sat in the back seat
of the test vehicle, while one participant sat in the front passenger seat. A representative from
VTTI drove the car. The driver instructed each participant to depress a push button device when
he or she identified a wooden visibility target through the front windshield. A GPS device
recorded the detection distance between the vehicle and the target, thus creating data for
quantitative comparison among luminaire types and light levels.
1.3 Research Results
The use of LED technology for city street lighting is becoming more widespread. While these
lights are primarily touted for their energy efficiency, the combination of LEDs with advanced
control technology, changes to lighting criteria, and a better understanding of human mesopic
(low light level) visibility creates an enormous potential for energy savings and improved
motorist and pedestrian visibility and safety.
Data from these tests support the following statements:
LED luminaires with a correlated color temperature of 4100K provide the highest
detection distance, including statistically significantly better detection distance when
compared to HPS luminaires of higher wattage.
The non-uniformity of the lighting on the roadway surface provides a visibility
enhancement and greater contrast for visibility.
Contrast of objects, both positive and negative, is a better indicator of visibility than is
average luminance level.
Dimming the LED luminaires to fifty percent of IES RP-8 levels did not significantly
reduce object detection distance in dry pavement conditions.
Participants perceived dimming of sidewalks as less acceptable than dimming to the same
level on the roadway.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - iii
Asymmetric lighting did reduce glare and performed similarly to the symmetric lighting
at the same color temperature (4100K).
The results indicate that the 105-watt LED luminaire, with a correlated color temperature (CCT)
of 4100K (symmetric and asymmetric), has the highest detection distance of all of the test areas,
with a value of approximately 130 feet. This luminaire outperformed even the 250-watt (280
system watts) and 400-watt (450 system watts) HPS, with over two and four times the wattage
respectively. Even when reduced to twenty-five percent of full light output during the dry
pavement test, the LED 4100K luminaires did not have a significantly different detection
distance compared to the same luminaires at one hundred percent of full light output. The wet
roadway conditions did show a decrease in detection distance when the lighting system was
dimmed, especially at twenty-five percent of full light output.
The illuminance uniformity ratio of the 4100K LED luminaire is the highest (least uniform) of
all of the LED luminaires, yet this luminaire also has the greatest detection distance.
The contrasts of targets for all colors increased as the light levels dimmed. Participants assessed
the contrast from 200 or more feet away from the vehicle, or beyond the reach of headlamp
lighting, whereas most detections occurred within 200 feet, or within the headlamp span. A
greater contrast ratio typically results in greater visibility; however, based on the average
detection distances and the test vehicle’s headlamp assessment, VTTI concluded that headlamps
are not the primary source of detection.
Luminance does not exhibit a correlation to detection distance; the two HPS luminaires and the
5000K LED provided greater levels of luminance than did either of the 4100K LEDs (symmetric
or asymmetric) but did not show a related increase in the object visibility. This demonstrates that
the primary indication of visibility is contrast and that a reduced luminance level with equivalent
contrast may provide equivalent or better detection distances.
The user field test results indicate that the implementation of adaptive lighting does not
significantly affect object detection distance for dry roads. However, coupling this data with the
written evaluation results indicates that reducing the light level to twenty-five percent of full
light output for all hours of the night raises concerns for the public, especially on the sidewalks.
Tuning the light to a point such as twenty-five percent of full light output may be justified at low
vehicular and pedestrian volumes and under dry pavement conditions, but not for all hours of the
evening.
The asymmetric luminaires recorded the lowest glare values of all of the test areas, as the light
was intended to be directed away from the driver. While the asymmetric luminaires performed
on par with the symmetric 4100K LED luminaire, participants did not rate the asymmetric test
area very high, especially at the lower light levels. Participants deemed the distribution to be
patchy and claimed that signage was difficult to view.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - iv
1.4 Industry Implications
Standards
The user field test data findings demonstrate that less uniformity trends toward greater detection
distance. The importance of uniformity in target detection constitutes another aspect to consider.
The 4100K luminaire exhibited the highest illuminance uniformity ratio, indicating the most
non-uniform appearance; it also showed the highest visual performance. While industry
standards list maximum uniformity ratios, they do not address a lower limit. The IES research
committee should explore both the maximum and minimum ratios to account for higher contrast
with less uniform pavements. Visibility level (VL) concepts are addressed in the standards, but
researchers should refine the values to replicate field data.
The data gathered from the user field test in this study also supports the use of mesopic
corrections under IES RP-8. The contrast of objects illuminated by shorter wavelength light
(from LEDs) at low light levels is a valid reason for reducing light levels while not affecting
visual detection distance.
Future Product Designs
LED technology and the design of luminaires can address the sidewalk lighting issue. Ideally, the
luminaire controller would maintain higher light levels on the sidewalks for pedestrians during
conditions when roadway illumination is reduced. The authors believe that current LED drivers
could be adapted to support this methodology. Such design changes would ensure that sidewalks
remained illuminated and that pedestrians would likely feel safer than they did in this test, in
which some participants expressed concern about under-lighted sidewalks with the luminaires at
twenty-five percent of rated output. More uniform sidewalks may also play a greater role in
pedestrians’ perception of security.
Economic Analysis
The analysis indicates that the implementation of LEDs and controls can pay back in just over
three years when replacing 400 W HPS luminaires with 105 W LED luminaires, and within six
years when replacing 250 W HPS luminaires with 105 W LED luminaires. The payback values
improve with more aggressive adaptive lighting.
1.5 Future Work
This project used a test site along a roadway with a speed limit of thirty-five miles per hour. In
order to fully understand the magnitude of mesopic benefits, researchers should expand this
study to take place on a roadway with a higher speed limit (fifty-five mph) to verify the existence
of similar results. Future detection tasks may focus on foveal (line-of-sight) versus non-foveal
(peripheral) vision; researchers may develop new adjustment factors to account for color contrast
in foveal vision.
Given that researchers found contrast to be a strong indicator for detection distance, future
research should delve further into Visibility Level (VL). The weighted average VL comprises the
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - v
Small Target Visibility (STV) within RP-8. Researchers should refine these values to more
accurately predict STV based upon visibility research.
The test site was located in an urban environment with light contribution from several adjacent
businesses. As researchers reduced the light from the LED luminaires, the influences of non-
uniformity and contrast as strong indicators of visibility became evident. However, the light
contribution from the adjacent businesses remained at full brightness throughout the
demonstration, thus contributing to the contrast values. Subsequent studies would benefit from
having adjacent businesses extinguish their lights for the duration of the demonstration to
determine whether or not non-uniformity and contrast remain strong indicators of visibility
without their contribution.
Researchers added the condition of wet pavement to this demonstration to test the impacts on
visibility when pavement conditions change. Future research should look into other weather
conditions such as fog and snow to determine visibility variance when introducing these
common weather elements.
Future work should also include a more thorough look at sidewalk visibility: when pedestrians
are navigating detached and attached sidewalks, what type and quantity of light achieves the
highest level of visibility? Future studies should include both static and dynamic objects in-situ
in a city to measure detection distance.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 1
2 Introduction
2.1 Background
The use of light emitting diode (LED) technology for city street lighting is becoming more
widespread. While these lights are primarily touted for their energy efficiency, the combination
of LEDs with advanced control technology, changes to lighting criteria, and a better
understanding of human mesopic (low light level) visibility creates an enormous potential for
energy savings and improved motorist and pedestrian visibility and safety.
LED efficacy (the amount of light generated per watt of electricity) continues to increase
substantially as the technology and deployment improve. However, efficacy varies dramatically
with the spectral distribution (color temperature) of the LED. Due to the current manufacturing
process, cooler colors (higher Correlated Color Temperature, or CCT) result in higher efficacies
than do warmer colors (lower CCTs). In fact, as of 2013, the Department of Energy (DOE)
reported cool white LED packages (CCT=4746K to 7040K) with an average of 164 lumens per
watt and warm white LED packages (CCT=2580K to 3710K) with an average of 129 lumens per
watt (DOE 2013). However, public preference typically favors warm white light, or the lower
color temperatures, such as 3500K. Ignoring this in favor of higher efficacies, manufacturers’
marketing media push higher-color-temperature 5000K and 6000K light sources to gain a
competitive edge. This results in installations that produce light very efficiently, but within a
spectrum that can affect brightness perception, color rendering, discomfort glare, circadian
rhythm, and other possible health issues (IES 2013).
Network control of exterior lighting provides another layer of energy savings potential to street
lighting systems. Most street luminaires use photo sensors that detect a drop in ambient daylight,
causing the luminaire to switch on. At dawn, the same sensor turns off the luminaire. Although a
straightforward solution, this strategy results in the light source being activated all night and
during periods of dusk and dawn. Additionally, photo sensors are typically the most likely
component of the system to fail. Networked lighting controls link groups of luminaires together
with either radio frequency or by power line carrier. When dimmable sources (such as LED) are
controlled in this mesh network, the luminaires can be dimmed or turned off as a group, or
individually.
The Illuminating Engineering Society (IES) roadway lighting criteria (RP-8) outlines decreasing
light level requirements for decreasing levels of pedestrian and motorist conflict. However,
without dimming and control technology available, changing light levels after installation has
been impractical. A roadway lighting design provided the appropriate amount of light for the
worst set of design conditions. Additionally, because light output diminishes with time (lumen
maintenance), traditional design practice puts the initial light output well over the requirement so
that the light level will still be met when light output has decreased at the end of the light
source’s life. Implementing dimming control, or “tuning” these light sources, can now provide
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 2
the design level of light at all times, accounting for decreasing late night pedestrian activity and
for lumen depreciation over the life of the light source.
LEDs also produce a white light with high color rendering ability as opposed to the yellow light
produced by the high pressure sodium (HPS) sources commonly used in North American cities.
With the revision of Technical Memorandum (TM 12-12) Spectral Effects of Lighting on Visual
Performance at Mesopic Lighting Levels, the IES recognized that some spectral distributions
provide better visibility under mesopic (low light level) conditions than do others. Although
pinpointing it is difficult, TM 12-12 documents a method for calculating the effective luminance
of a roadway lighting design. This means that designers can design for lower light levels when
using a white light source and still achieve the same level of visibility provided by higher light
levels produced by an HPS light source.
The energy savings from this combination of opportunities far exceeds the potential of the
efficient technology alone. However, most pilot studies and tests evaluate the solid state
technology alone without analyzing the potential synergies of these other opportunities.
Table 1. Potential Cumulative Energy Savings
Potential Cumulative Energy Savings
Energy Savings Potential
Luminaire Replacement
15-40%*
Lumen Maintenance
5-15% (at beginning of life)
Mesopic Multipliers
5-10% (4000K source compared to HPS at
2000K)
Adaptive Lighting
25%**
TOTAL ENERGY SAVINGS
50-90%
Notes: *This savings is in addition to the application of adaptive lighting, and greatly depends on whether the incumbent
technology was currently meeting performance or prescriptive criteria.
**Assumes fifty percent light level reduction during fifty percent of the operating hours.
2.2 Technology and Market Overview
The cost effectiveness of LED street luminaires varies dramatically depending on site-specific
factors and on the variables considered in the economic analysis. Using a Return on Investment
or Net Present Value analysis captures not only energy costs but maintenance savings as well.
Additionally, light source efficacy continues to improve each year. The DOE predicts that by
2020, even warm white LED packages will exceed 200 lumens per watt (DOE 2013). While
efficacy continues to increase, the growing number of luminaire manufacturers in the market
intensifies competition and reduces costs. However, quality also varies significantly among these
manufacturers; the DOE reports payback periods from its gateway projects as short as three years
and as long as twenty years(DOE 2013). The following variables affect the cost analysis of any
project:
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 3
Cost of current energy and future predicted escalation
Maintenance costs (light source, driver/ballast, photocell replacements) and whether or
not these are considered in the analysis
Use of adaptive lighting and controls to reduce light output after hours
Use of lumen maintenance tuning to keep lighting at design levels throughout the course
of the system’s life
Change from current lighting levels that have over-lighted streets
2.3 Project Objectives
The large number of previous pilot and study projects has failed to sufficiently address the
following objectives:
1. Test different color temperatures for the existence of preferences for a specific color and
for performance advantages to a specific color temperature (spectral distribution).
2. Evaluate opinions of citizens toward various light sources that may be suitable candidates
for selection of replacement luminaires.
3. Evaluate the performance of luminaires with an asymmetric distribution that maximizes
the vertical brightness along a roadway.
4. Test three different light output levels (one hundred percent, fifty percent and twenty-five
percent) to see the point at which the lower levels become undesirable.
5. Test different road conditions (wet and dry pavement) to identify potential luminaire and
light source performance advantages for one condition over the other.
6. Evaluate object detection performance and community acceptance of white (broad-
spectrum) LED lighting.
The results of this study will also support the development of LED streetlight design guidelines
for the Northwest. With the rapid pace of change in this industry, this design guidebook will help
municipalities and utilities to cost-justify and confidently select luminaires and control systems
to meet their individual needs.
2.4 Project Hypotheses
Researchers developed the following series of hypotheses prior to beginning this project to help
define the study parameters.
Luminance versus Illuminance
Illuminance has constituted the basis of most lighting design criteria, yet it does not
address visual adaptation. Luminance will more accurately predict object detection versus
illuminance, since it best represents what one “sees”: visual adaptation and object
contrast. This distinction between illuminance and luminance will be emphasized under
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 4
wet conditions where the difference between illuminance and luminance will be the
highest.
Since the earlier-mentioned experiments were illuminance-based, this experiment should
perform both illuminance and luminance setups. Since the test areas will be identical, the
luminance detection will be compared with the illuminance test in order to establish the
relationship between the two, and also to confirm the hypothesis that luminance is the
best object detector predictor.
Broad Spectrum versus HPS Lighting
Broad spectrum lighting will yield object detection higher than that of HPS lighting under
the same illuminance or luminance level. This occurred in three other streetlight
experiments conducted by Clanton & Associates and Virginia Tech Transportation
Institute and is predicted to have the same results with this experiment. See Work for
additional information.
Adaptive Lighting
Even under lower luminance or illuminance levels, broad spectrum lighting will have
greater detection distances than HPS.
Asymmetric Lighting
Asymmetric distribution will increase detection distances, especially under wet
conditions. Since pavement reflectance and viewing angle both affect luminance values,
wet pavement luminance increases when light is directed toward the driver. This
increases veiling luminance and low contrast. Asymmetric distribution directs light away
from the driver, similar to an extension of headlamps. These effects may also appear
under dry conditions, since small target visibility increases with asymmetric distribution.
Energy Savings
Asymmetric broad-spectrum lighting at dimmed levels will result in maximum energy
savings due to increased detection distances even at dimmed lighting levels.
Vertical Illuminance and Luminance
The asymmetric luminaire design approach relies on the propagation of light at higher
angles, resulting in a greater ratio of vertical illuminance to horizontal illuminance. This
will likely result in higher luminance conditions for vertical targets as well, which should
result in greater contrast between vertical and horizontal surfaces (target versus
background/foreground). This shift in the light propagation in a roadway lighting system
may result in considerable improvements in visibility.
CCT Importance
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 5
Changing CCT for broad-spectrum lighting will not lead to a linear decrease in detection
differences and will not follow the lumens-per-watt efficacy. While lower CCT for broad
spectrum lighting will have lower detection distances, the difference will be insignificant
compared to higher CCT broad spectrum lighting within the range that is considered
reasonable for roadway lighting. This will give communities the ability to choose CCT
based on community preference without the penalty of reduced energy savings potential.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 6
3 Methodology
3.1 Overall Project Setup
This demonstration project targeted two audiences over the course of three evenings. On the first
evening, industry professionals, media, and political representatives received an overview of the
lighting system and the goals of the demonstration project. On the two succeeding evenings,
participants from the general population of Seattle, Washington viewed the lighting with both a
written evaluation and a user field test.
Researchers conducted the two surveys, a written evaluation and a user field test, to gain an
understanding of the public’s views on solid state street lighting using LED technology and to
quantify the difference in visibility under the new technology and under traditional high pressure
sodium (HPS) light sources.
Researchers conducted the demonstration along a portion of 15th Avenue Northwest in the
Ballard neighborhood in Seattle, Washington. The 15-block stretch of the demonstration test site
contained six test areas, as shown in Figure 1 below.
Figure 1. Demonstration Site Layout
While the test site contained two 400 W HPS sections, the study surveyed only the larger of the
two. The integrity of the study required equal numbers of LED luminaires in each test area.
Given the existing layout of the lighting along this street, the intersection at NW 70th Street has
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 7
too few luminaires for its own test area, so it remained with the Seattle City standard 400 W HPS
streetlights.
Two of the test areas used HPS, one with 400 W light sources and one with 250 W light sources.
Both of these test areas used cobra-head-style luminaires. The other test areas used LED
luminaires of varying distributions and color temperatures, as outlined in Table 2.
Table 2. Summary of Test Areas
Summary of Test Areas
Test Area 1
Test Area 2
Test Area 3
Test Area 4
Test Area 5
Test Area 6
Light Source
LED
LED
HPS
LED
LED
HPS
Color Temperature
3500K
4100K
2000K
4100K
5000K
2000K
Distribution
Type II
Type II
Type II
Type II
Asymmetric
Type II
Note: Color temperature values are as stated by the manufacturer; they were not measured.
3.2 Site Selection
Researchers began selection of the demonstration site in June 2011. The Seattle Department of
Transportation (SDOT) provided a list of possible streets and vetted them for consistency of
streetlight arrangement and spacing. Researchers also considered ease of closing the road to
through traffic. Possible demonstration site locations included:
8th Avenue NW between NW 51st Street and NW 45th Street
4th Avenue South between South Industrial Way and South Michigan Street
Winona Avenue North between N 76th Street and North 66th Street
West Nickerson Street between 13th Avenue West and Warren Avenue North
Greenwood Avenue North between North 95th Street and North 74th Street
1st Avenue between the West Seattle Bridge and East Marginal Way
35th Avenue SW between SW Thistle Street and SW Roxbury Street
California Avenue SW between SW Edmunds and SW Myrtle Street
15th Avenue NW between NW 85th Street and NW 65th Street
Researchers finalized the demonstration site at 15th Avenue NW because it is long enough to
accommodate six test areas, it has a fairly uniform opposite pole arrangement, it does not cross
major streets, and it is straight with a uniform width throughout. The demonstration site was
originally intended to start at NW 85th Street and continue to NW 65th Street. A paving project in
that area with a detour route along NW 83rd Street moved the demonstration site to the stretch
from NW 80th Street to NW 65th Street.
Once researchers had selected the demonstration site, the SDOT conducted a truck turning
movement study and developed a detour route. Coordination with the Metro bus service began to
design a detour route that would minimize the impact on ridership.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 8
While SDOT coordinated the road closure along 15th Avenue NW, Clanton & Associates and
VTTI developed the characteristics of the lighting. Seattle City Light (SCL) conducted a site
visit to observe the luminaire mounting height, existing light source wattage, light source type,
arm length, pole spacing, and other sources of illumination along the site aside from street
lighting. The characteristics of the roadway include:
Seventy feet curb to curb width
Ten feet center lane no median
Three lanes of traffic in each direction
Single head luminaire
Six foot arm length
Thirty foot luminaire mounting height
Wooden poles
Overhead electric feeds
~130-foot spacing pole to pole
Figure 2. Preliminary Site Visit to Observe Other Sources of Illumination
3.3 Public Outreach
In advance of the demonstration, SDOT notified business owners and residents in the vicinity of
15th Avenue NW. Each business owner received notice of the demonstration and street closure in
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 9
a one-page handout. SDOT mailed residents in the area near the demonstration site a postcard
notifying them of the street closure and detour routes. SDOT also contacted the local
neighborhood blog, The Ballard Blog, to inform residents as well as to recruit participants for the
demonstration. SDOT also provided a press release with details of the demonstration and street
closure. Appendix J: Public Outreach shows all of the outreach documents.
3.4 Lighting Criteria
The existing HPS luminaires along the stretch of 15th Avenue NW consist of 400 W light sources
meeting the current City of Seattle prescriptive standards. To provide a baseline for comparison,
Clanton & Associates designed the new LED streetlights around performance criteria
recommended by the Illuminating Engineering Society of North America (IES), Recommended
Practice for Roadway Lighting (RP-8). The roadway and pedestrian classification determined the
luminance criteria to use for the design. The area surrounding the demonstration site is mainly
residential with some commercial buildings along 15th Avenue NW. During the daytime, 15th
Avenue NW is heavily-used with three lanes of traffic in each direction with a center turn lane.
At night, the street reduces to two lanes of traffic in each direction, with parallel parking in the
third lane. These conditions classify the site as a collector roadway.
The IES defines a collector road as:
A roadway servicing traffic between major and local streets. These are streets used
mainly for traffic movements within residential, commercial, and industrial areas. They
do not handle long, through trips. Collector streets may be used for truck and bus
movements and give direct service to abutting properties.
Two schools are located within two blocks of the demonstration site. Pedestrian activity is
moderate during the nighttime. Many residents in the nearby area walk their pets along 15th
Avenue NW. Given these conditions, the demonstration is classified as having medium
pedestrian conflict.
IES RP-8 (2005) defines a medium pedestrian conflict as:
An area where lesser numbers of pedestrians utilize streets at night. Typical are
downtown office areas, blocks with libraries, apartments, neighborhood shopping,
industrial, older city streets, and streets with transit lines.
Clanton & Associates selected the wattage and distribution for the new LED streetlights based
upon their performance in meeting the criteria in the following table.
Table 3. IES RP-8 Luminance Criteria for New LED Streetlights
Luminance Criteria for New LED Streetlights
Roadway
Classification
Pedestrian
Conflict Area
Average
Luminance Lavg
(cd/m2)
Uniformity
Ratio Lavg/Lmin
(max)
Uniformity
Ratio Lmax/Lmin
(max)
Veiling
Luminance Ratio
Lvmax/Lavg (max)
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 10
Collector
Medium
0.6
3.5
6.0
0.4
Clanton & Associates used no luminance criteria for the HPS luminaires. The 400 W HPS
luminaires represent the City of Seattle’s standard luminaire on this type of roadway. The 250 W
HPS luminaires compare the visibility performance to that observed in other studies and to the
higher 400 W luminaire.
3.5 Luminaire Selection
Philips Lumec RoadStar provided the LED luminaires for the demonstration. Selecting one
manufacturer streamlined the procurement and installation processes while eliminating any
variables due to distribution types and reflector components. Clanton & Associates also
considered the IES TM-15-11 Backlight Uplight Glare (BUG) ratings in luminaire selection.
Clanton & Associates selected no luminaires with uplight ratings above 0 (U0) or glare ratings
above 2 (G2) to limit options appropriate for a mixed-use neighborhood.
In order to meet the luminance criteria outlined in IES RP-8, Clanton & Associates selected the
Lumec GPLM 105 W Type II luminaire. This luminaire comes standard with a correlated color
temperature (CCT) of 4100K and consumes 105 W. Lumec provided the other, non-standard
luminaires with CCTs of 3500K and 5000K for this demonstration. Lumec also provided a
custom-designed asymmetric luminaire: CCT 4100K. Clanton & Associates used the lighting
software AGi32 and a 0.765 light loss factor for all lighting calculations; this value is based upon
Philips Lumec’s 0.85 lamp lumen depreciation value and a 0.9 luminaire direct depreciation
factor. Appendix E: Preliminary Luminaire Testing shows all calculations.
The four LED test areas each contained ten luminaires. Because the research included the effect
of color temperature on participant opinions, the test areas also represented three different color
temperatures: 3500K, 4100K, and 5000K. All ten luminaires have a standard Type II distribution
for each of these color temperatures. The fourth test area also has a color temperature of 4100K
with an additional custom asymmetric distribution designed specifically for this study.
3.6 Controls Selection
Owlet’s Nightshift system controls the LED luminaires. Each luminaire contains one 2mW
luminaire controller (LuCo) at 120 volts. One segment controller (SeCo) receives signals from
each of the LuCos. This system can remotely meter energy consumption, provide two-way
communication with status updates, and dim the lights on command or on schedule.
At the time of specification, the City of Seattle was also implementing the Owlet control system
on another project.
To ensure that the signal would propagate along the entire demonstration site, SCL installed two
10W LuCos within the 400 W HPS test area. These LuCos bypass the HPS luminaires and only
act as repeaters to the next LED LuCos so the entire system receives a signal.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 11
3.7 Asymmetric Luminaire Design
As mentioned above, the luminaires in test area 4 utilize a non-traditional distribution. Instead of
offering even light distribution both toward and away from the driver, this design directs the
majority of the light away from the oncoming car, effectively “leading” the driver (pro-beam)
down the road, in a manner similar to extending the headlight range. This design is intended to
compare the measured detection distance under the asymmetric luminaire to a standard Type II
distribution luminaire. The reflected glare from wet pavement can cause discomfort and can
impair a driver’s ability to see; however, if the majority of the light leaving the luminaire is pro-
beam, reflected glare will be decreased and will, in theory, increase the visibility of the driver.
Coordination with Philips Lumec began in June of 2011. Because the demonstration site was not
yet finalized and the luminaire design depended on the specific road dimensions, Lumec did not
begin luminaire design until October 2011. Figure 3 shows one of the initial conceptual drawings
for the demonstration.
Figure 3. Elevation of Conceptual Design of Asymmetric Luminaire
After the finalizing the demonstration site, Clanton refined and optimized the design to meet the
exact conditions of the roadway. It provided Philips Lumec with the roadway characteristics as
well as design parameters, which included:
Maximize small target visibility (STV)
Have a maximum to minimum luminance uniformity ratio of no greater than 10:1
Maintain the same BUG glare rating of G2 as the symmetric luminaires
Do not allow pro-beam (the forward distribution of the light) to cross into oncoming
traffic, where it would become a glare source for motorists on the opposite side of the
street
This new distribution design does not meet the average luminance values in IES RP-8. However,
while the average luminance levels would be less, maximizing the STV may result in
uncompromised visibility.
The initial design provided by Philips Lumec met these requirements, as illustrated in Figure 4. It
shows minimal light crossing into oncoming traffic and the majority of the light from the
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 12
luminaire leading the driver. A sharp light cutoff to the left of the luminaire creates a unique
distribution, noticeable in the field.
Figure 4. Preliminary Asymmetric Luminaire Design by Philips Lumec
While the first design met all of the design requirements, Clanton & Associates directed Philips
Lumec to revise the original design to an IES BUG uplight rating of 0, to further reduce the light
crossing over onto oncoming traffic and to maintain or increase small target visibility. The
second asymmetric luminaire design (Figure 5) increased the STV from 4.6 to 5.4, and lowered
the glare threshold along with the average luminance.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 13
Figure 5. Second Asymmetric Luminaire Design
Philips Lumec revised the second design to exclude the backlight mask and created a third and
final design that was completed in early November 2011. Philips Lumec began production on
this design in early 2012. It delivered one luminaire of each type to VTTI for preliminary testing,
and shipped the remaining nine of each type to Seattle for installation.
3.8 Road Conditions
Because of the Pacific Northwest’s rainy climate, the demonstration evaluated the streetlight
performance with two different pavement conditions: dry and wet. Fog is also prevalent in this
area but is quite difficult to generate on command. VTTI made plans to accommodate for fog in
the calculation results should it be experienced during the demonstration, but it made no plans to
generate artificial fog.
Clanton & Associates scheduled the demonstration for the first week of March 2012. Since
wetting a dry road is easier than drying a wet road, Clanton & Associates scheduled the
demonstration for a time when the probability of rain was small, but also prior to the onset of
Daylight Saving Time. The demonstration had to begin no sooner than an hour after sunset. Later
sunsets as summer approaches increases the difficulty of getting participants to attend a post-
sunset demonstration later in the night. Given all these considerations, Clanton & Associates
decided that the ideal time for the demonstration would be March 6, March 7 and March 8 from
8:00 p.m. to 1:00 a.m. each night.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 14
Since the demonstration required one evening with dry roads and one evening with wet roads,
Clanton & Associates along with SDOT had several contingency plans in place to address
potential weather issues. If the roads were dry all three evenings, they would be artificially
wetted one evening to simulate rain conditions. If the roads were wet all three evenings, Clanton
& Associates and VTTI would conduct only two evenings (one with media and industry
professionals and another for participants) of surveys, with a makeup date scheduled later in the
year when the probability of dry roads would be higher. Continuum Industries printed all of the
written evaluations on waterproof paper in the event of rain while participants completed their
surveys.
3.9 Light Output Level
The study evaluated three light output levels for the LED luminaires: one hundred percent, fifty
percent, and twenty-five percent of full light output. This effort evaluated the concept of adaptive
lighting standards and determined the effect of reduced light levels at lower traffic volume
conditions at nighttime.
Adjustments to the voltage inputs provided to the luminaire via the Owlet control system tunes
the light output of the luminaires. Prior to the demonstration, Philips Lumec sent a sample of
each luminaire to VTTI, where preliminary testing determined the corresponding voltage inputs
for each light output value (one hundred percent, fifty percent, and twenty-five percent). See
Appendix E: Preliminary Luminaire Testing for additional information.
3.10 Participant Recruitment
Continuum Industries recruited approximately 180 participants for each night of the
demonstration, sixty for each of the three light levels. VTTI determined the number of recruits
for the user field test based on the required number of eighteen participants for each light level.
The fact that historically many participants choose not to ride in the user field test vehicle
dictated an over-recruit of approximately three times the required number of participants for each
light level.
SDOT provided assistance for recruitment outlets throughout the City. Continuum offered each
potential recruit a forty-dollar gift card as an incentive for participation.
Continuum recruited participants from across Seattle through colleges and universities, as well as
through employment offices and non-profits. It encouraged non-profits to alert their constituents
about the survey, and several used the forty-dollar gift cards as fund-raising opportunities by
having members donate their gift card directly to the organization. Continuum also targeted
participants from sources closer to the test site, including neighborhood associations and assisted
living facilities. It also informed businesses affected by the street closures of the opportunity to
participate in the test as means of generating community interest and goodwill.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 15
Individuals interested in participating signed up online via a recruitment website that collected
contact information, availability by date and time, gender, and age. The recruitment website also
educated participants on what the demonstration would require of them. As the survey dates
drew near, Continuum Industries contacted participants via email and phone with priority based
on the dates of their online registration and on their age groups. To account for changes to the
human eye over a lifetime, Continuum made every effort to evenly represent each age group for
each light level of the test. Given the weather contingency plans and the intent to have a
representative sample of age groups, some participants were recruited for both nights of testing,
at different light levels.
3.11 Written Evaluation
On each of the two nights of general participant testing, participants completed subjective
surveys for each test area. Clanton & Associates modified a survey originally developed for
parking lot lighting surveys by Dr. Peter Boyce, formerly of the Lighting Research Center, to
render it suitable for use in street lighting demonstrations. The team used this same survey to
evaluate the street lighting in Anchorage, Alaska; San Diego, California; San Jose, California;
and Roseville, California.
The written evaluation asked participants a series of general questions based upon where they
live, in addition to demographic questions and site condition questions. For each test area, a
participant next rated twelve statements on a five-point scale (strongly disagree to strongly
agree). Following are the statements that comprised the written evaluation:
1. It would be safe to walk here, alone, during daylight hours.
2. It would be safe to walk here, alone, during darkness hours.
3. The lighting is comfortable.
4. There is too much light on the street.
5. There is not enough light on the street.
6. The light is uneven (patchy).
7. The light sources are glaring.
8. It would safe to walk on the sidewalk here at night.
9. I cannot tell the colors of things due to the lighting.
10. The lighting enables safe vehicular navigation.
11. I like the color of the light.
12. I would like this style of lighting on my city streets.
Participants also answered an additional question on the scale of much worse to much better
13. How does the lighting in this area compare with the lighting of similar Seattle city streets
at night?
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 16
3.12 User Field Test
Equipment
The data collection equipment used during the experiment consisted of a variety of components
for collecting illuminance, luminance, color temperature, and participant response data. Most of
these are part of the Roadway Lighting Mobile Measurement System (RLMMS), a device
created by the Center for Infrastructure Based Safety Systems (CIBSS) at VTTI as a method for
collecting roadway lighting data in addition to participant response data.
VTTI mounted a specially-designed “Spider” apparatus containing four waterproof Minolta
illuminance detector heads horizontally onto the vehicle roof in a manner that positioned two
illuminance detector heads over the right and left wheel paths and positioned the other two
illuminance detector heads along the centerline of the vehicle. VTTI positioned an additional
vertically-mounted illuminance meter in the vehicle windshield as a method to measure glare
from the lighting installations. VTTI connected the waterproof detector heads and the
windshield-mounted Minolta head to separate Minolta T-10 bodies that sent data to the data
collection PC positioned in the trunk of the vehicle.
VTTI positioned a NovAtel Global Positioning Device (GPS) at the center of the four roof-
mounted illuminance meters and attached it to the “Spider” apparatus. It connected the GPS
device to the data collection computer via USB so the vehicle latitude and longitude data was
incorporated into the overall data file.
VTTI mounted two separate video cameras on the vehicle windshield. One collected color
images of the forward-driving luminous scene and the second collected luminance information
for the forward-driving luminous scene. VTTI connected each camera to a standalone computer
that was in turn connected to the data collection computer. The data collection computer
recorded illuminance, human response (reaction times), and GPS data and synchronized the
camera computer images with a common timestamp. Additional equipment inside the vehicle
consisted of individual input boxes for participant-entered responses and a Controller Area
Network (CAN) reader for collecting vehicle network information.
A specialized software program created in LabVIEW™ controls each component of the
RLMMS. The software synchronizes the entire hardware suite and sets data collection rates at
20Hz. VTTI set the video image capture rate for this demonstration at 3.75 frames per second
(fps). The final output file used during the analysis contained a synchronization stamp, GPS
information (such as latitude and longitude), input box button presses, individual images from
each of the cameras inside the vehicle, vehicle speed, vehicle distance, and the illuminance meter
data from each of the four Minolta T-10s. VTTI incorporated the vehicle’s latitude and longitude
data into the overall data file via USB connection to the data collection computer.
Visibility Targets
Research has established a relationship between certain visibility metrics and the detection and
avoidance of a small object on a roadway. Research has also established a correlation between
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 17
these visibility metrics and the frequency of vehicular accidents at night. Small Target Visibility
(STV) is a method to calculate this relationship.
The STV method (as defined by IES RP-8) is used to determine the visibility level of an array of
targets along the roadway when considering certain factors such as the luminance of the targets,
the luminance of the immediate background, the adaptation level of the adjacent surroundings,
and the disability glare. The weighted average of the visibility level of these targets results in the
STV value.
The visibility targets for this demonstration are wooden squares seven inches on each side, with a
tab measuring 2.375 inches by 2.375 inches on one side (pictured in Figure 6). The targets came
in four colors: red, green, gray, and blue. VTTI painted the target bases to be similar to the road
surface. VTTI placed these objects along the roadway as the objects of interest in the
performance portion of the project.
Figure 6. Visibility Targets Used within Test Areas
VTTI positioned targets of each color within each of the test areas to achieve a consistent level
of vertical illuminance for all luminaire types. Each target location had fourteen lux of vertical
illuminance except for the 400 W HPS section, where twenty lux was the lowest achievable
vertical illuminance.
VTTI’s goals in setting up the visibility targets consisted of exposing each luminaire type to each
target color and matching each location by vertical illuminance. VTTI paired the target colors
(green/gray and red/blue) and intermittently shifted them among luminaires during breaks when
the luminaires were dimmed. The percent reflectance by each target color is shown in Table 4.
Table 4. Percent Reflectance by Target Color
Color
Reflectance
Gray
17%
Green
17%
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 18
Blue
15%
Red
12%
Illuminance more directly characterizes a luminaire’s output, whereas luminance more directly
characterizes the amount of light perceived. Matching the targets for illuminance isolates the
lighting output, thus making the luminaires comparable on that basis. Matching the targets for
luminance would require considerations of target surface reflectance, road surface reflectance,
and target color.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 19
4 Procedure
4.1 Equipment Pretesting
Prior to the demonstration, VTTI received four luminaires (one of each of the four types of LED
luminaires) and all equipment necessary to operate the control system. This preliminary testing
ensured that the luminaires were operating as specified and determined the driver voltage inputs
corresponding to the desired light output levels of one hundred percent, fifty percent, and twenty-
five percent. While the testing also included correlated color temperature measurement, the
results of the color temperature test were inconclusive.
The testing process entailed mounting each luminaire in the VTTI test facility for a “burning-in”
time of approximately one hundred hours. Next, VTTI individually mounted the luminaires in
the outdoor environment at the VTTI test facility (Figure 7). In this location, VTTI measured the
output of the luminaire in terms of horizontal and vertical illuminance along a grid of test
locations beneath and to the side of the luminaire.
Figure 7. Laboratory Evaluation Grid
Dimming Data
VTTI also measured the effects of dimming each LED luminaire. Figure 8 shows the relationship
between the dimming voltage and the percent of light output for each luminaire. The results
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 20
indicate that as dimming voltage increases, the percent of light output also increases similarly for
each luminaire.
Figure 8. Dimming Voltage by % Light Output
VTTI used the 1-10 dimming voltage inputs collected through this preliminary testing to allow
the control system to tune the lights at the desired light levels. The values calculated are:
One hundred percent light output 8V
Fifty percent light output 3.1V
Twenty-five percent light output 1.2V
VTTI modified the final voltage inputs from the above values because the control system needs
to have a linear curve between the low and high value in order to function properly. Therefore,
the low voltage inputs used during the demonstration are:
One hundred percent light output 8V
Fifty percent light output 3.3V
Twenty-five percent light output 1V
The relationship in Figure 9 shows the illuminance for each luminaire as it is dimmed or
brightened. Illuminance differs among the luminaires, but the relationship is similar for each.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 21
Figure 9. Dimming Voltage by Illuminance (lx)
Table 5. Summary of Lab Light Trespass
Summary of Lab Light Trespass
4100K
ASYM
3500K
5000K
HS
SS
HS
SS
HS
SS
HS
SS
Average Lab Measured
Light Trespass (lux)
4.21*
2.23
1.68*
1.99
2.37*
4.67
2.22*
4.74
Max (lux)
4.88
4.75
4.13
4.42
4.64
5.51
4.42
5.36
Min (lux)
3.37
0.53
0.37
0.22
0.69
3.26
0.61
4.05
Note: * Meets pre-curfew and post-curfew limit
Based on the IES recommendation, the four luminaires evaluated are below the light trespass
limit. Appendix E: Preliminary Luminaire Testing contains additional collected data.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 22
4.2 Equipment Installation
SCL installed all new light sources and cleaned luminaires to ensure that all 400W HPS
luminaires along the demonstration site were not subject to dirt or luminaire depreciation. SCL
also purchased and installed ten 250 W HPS cobra head luminaires. Lumec shipped forty LED
luminaires to the City in late January and early February for installation. Figure 10 illustrates the
sequence of each of the test areas.
Figure 10. Demonstration Site Layout
Seattle City Light began installing the HPS luminaires first. No luminaire controllers were
installed within the housings of the either the 250 W or the 400 W HPS luminaires. Two of the
400 W HPS luminaires have externally-mounted LuCos on their poles; however, the LuCos do
not control the light output of the HPS luminaire but instead propagate the signal to the next
LED series of luminaires.
Seattle City Light next installed the LED luminaires after completing all of the HPS luminaire
installations. Because Philips Lumec had installed the LuCos in the luminaires at the factory,
they did not require field installation. SCL staff correlated each LED luminaire with a badge
number matching the pole displayed on an aerial image of the demonstration site. A description
of the luminaire along with the pole badge number is hand-written on the inside of each housing
to help properly identify the placement of the luminaire, as indicated in Figure 11.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 23
Figure 11. Example of Handwritten Description
and Badge ID within Housing
Each LuCo has a unique ZigBee label on the module describing the LuCo, luminaire, and GPS
location. The installation process also used three other identical labels for each LuCo: one is
placed in the luminaire housing, one on the pole that can be read by someone standing on the
ground, and the last label is placed on a commissioning spreadsheet.
4.3 Setup of Visibility Targets
The team set up the small visibility targets the night before the demonstration began. The team
placed signs along 15th Avenue NW restricting residents from parking on the street.
The team placed each of the four colors of targets (red, blue, green and gray) under each test area
and measured the vertical illuminance, with the goal of finding locations within each of the test
areas where each target location measured an equal vertical illuminance value under each of the
test areas. The team marked the location of the target on the pavement with spray chalk. This
demonstration did not use yellow targets because the researchers could not find a yellow paint
color with a reflectance comparable to the other target colors.
The researchers found target locations with achieved vertical illuminance values of fourteen lux
for five of the six test areas. The 400 W HPS test area was unable to achieve a fourteen-lux
vertical illuminance value; this area achieved a lowest vertical illuminance value of twenty lux.
Researchers do not expect this higher illuminance level to affect the results, as visibility is based
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 24
on contrast and illuminance. While this area exhibited a higher target illuminance, the roadway
illuminance was higher as well, and the contrast should remain unaffected. The contrast is
explored in more detail later in this investigation.
To simulate rain on the roads, SDOT flusher trucks artificially wetted the roads for the March 8
demonstration. They sprayed the entire width of 15th Avenue NW from NW 80th Street to NW
65th Street. The flusher trucks sprayed down the road three times: once immediately after the
road closure, and then again before both the first and second dimming tests while the lights were
being dimmed. The sidewalks where participants were completing the written evaluations
received no water from the flusher trucks.
4.4 Written Evaluations and User Field Tests
Three groups of participants took part in the written evaluations on each of the two nights of
general participant testing. Each of the three groups arrived at different times throughout the
evening. The first group arrived at 7:00 p.m.; the second group at 8:30 p.m.; the third group
at10:00 p.m. At the arrival of each group, the participants where then divided into two subgroups
based upon assigned color of their written evaluations.
A bus dropped off the first subgroup at the intersection of 15th Avenue NW and NW 80th Street.
This group of participants made its evaluations on the west side of the street and took advantage
of the slight downhill topography. The second subgroup started at the intersection of 15th Avenue
NW and NW 65th Street. This group walked slightly uphill and made its evaluations from the
east side of the street.
Beginning at 8:00 p.m. each night, SDOT and the Seattle Police Department began closing 15th
Avenue NW between NW 65th Street and NW 80th Street. The road remained closed until 1:00
a.m. One police officer stayed at each end of the demonstration site throughout the duration of
the survey, and one additional police officer roamed the site throughout the evening. Once the
police officers gave clearance to begin the surveys, a team member gave instructions to each
group of participants.
Two project team members led each survey group and answered questions. At the beginning of
the survey, team members instructed participants not to look up at the streetlights, but rather to
evaluate the whole field of view into the street. Team members also instructed participants not to
talk to one another to avoid influencing other participants’ opinions. In addition to the two team
members per group, SCL staff members helped on site to manage traffic and to keep the
visibility targets in the correct standing positions. Two private security officers also remained on
site for the duration of the demonstration.
The bus picked up each subgroup of participants approximately an hour after they had been
dropped off. All participants returned to Salmon Bay middle school and turned in their surveys in
exchange for the forty-dollar gift card incentive.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 25
4.5 Experimental Protocol
A member from the VTTI team drove the experimental vehicle at a maximum speed of thirty-
five mph. Researchers selected the thirty-five mph speed as a typical speed for a commercial
lighting roadway, and simulated stopping distances outside of headlamp range. The experimental
vehicle drove in the middle of the three lanes, as the far right lane was used for target placement.
Along the route, participants pressed buttons when they were confident they had detected the
targets located along the roadway. The total testing time for the detection task lasted
approximately five minutes.
4.6 Dry Pavement
The dry pavement demonstration ran a total of eighty-three participants in the user field tests. As
many as three participants could take part in the driving test at one time. Some runs contained
fewer than three participants, depending on the number of volunteers from each subgroup.
Table 7 details the number of runs and the number of participants by light level. The light levels
are divided into number of runs (or laps) and the number of participants.
Due to a failed video card, VTTI could not record the luminance data concurrently with the
participant target detections. It conducted a second data collection effort approximately four
months later (July 2012) to gather the necessary luminance data. This data is included in the
analysis.
Table 6. Dry Pavement Written Evaluation Test Numbers
Pavement
Condition
Light
Level
Number of
Participants
Dry
100%
62
Dry
50%
54
Dry
25%
49
Table 7. Dry Pavement Participants User Field Test Numbers and Computer Conditions
Pavement Condition
Light Level
Number of
Runs (Laps)
Number of
Participants
Computer
Condition
Dry
100%
9
24
No Video Card
Dry
50%
12
35
No Video Card
Dry
25%
11
24
No Video Card
TOTAL
32
83
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 26
4.7 Wet Pavement
As mentioned earlier, SDOT flusher trucks wet the roads for the second evening of general
testing. Two trucks arrived on site shortly after the road barricades were up and 15th Avenue NW
was officially closed. Each truck held 3,000 gallons of water. Starting full, both trucks began
wetting the road near the intersection of NW 15th Avenue and NW 80th Street.
Figure 12. Flusher Truck Wetting the Roads
A total of fifty-one participants participated during the wet pavement portion, thirty-two fewer
than the dry pavement portion. Researchers reduced the time for each evaluation group due to the
allowance of time for flusher trucks to wet the pavement, thus reducing the number of
participants per run. Table 9 details the number of runs, number of participants, and computer
conditions for the wet pavement portion of the user field test.
Table 8. Wet Pavement Written Evaluation Test Numbers
Pavement
Condition
Light
Level
Number of
Participants
Wet
100%
59
Wet
50%
59
Wet
25%
49
Table 9. Wet Pavement Participants User Field Test Numbers and Computer Conditions
Pavement Condition
Light Level
Number of
Runs (Laps)
Number of
Participants
Computer
Condition
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 27
Wet
100%
10
24
Normal
Wet
50%
7
21
Normal
Wet
25%
2
6
System Error
TOTAL
19
51
Only six participants were able to be tested for the twenty-five percent light level condition as
the mobile computer system encountered further communication errors. However, complete
luminance data exists for the wet portion, given its ability to record during participant testing.
4.8 Luminance Measurements
As mentioned earlier, a video card failure meant the inability to measure all of the luminance
conditions on the first day of the experiment. Luminance data for the dry condition was recorded
prior to the wet pavement test after repair of the video card the following day. Unfortunately,
time restrictions led to recording only of the one-hundred-percent-lighted scenario; no luminance
data of the visibility targets was recorded for the fifty percent or twenty-five percent dim
conditions. Table 10 details the available luminance data. In the analysis, the one hundred
percent dry data was scaled by the illuminance measurements to provide estimates for the
background luminance and the target luminance. Researchers undertook a secondary effort in
July 2012 to re-collect luminance data to ensure quality. Unfortunately, wet pavement was not
available on this attempt. A discussion of the user field test measurements can be found in
Section 5: Discussion.
Table 10. Recorded Luminance Data
Recorded Luminance Data
March 2012
Pavement
Light Level
Luminance Data
Dry
100%
Recorded Luminance
Dry
50%
Unable to Record - Estimated
based on the illuminance ratio
Dry
25%
Unable to Record - Estimated
based on the illuminance ratio
Wet
100%
Recorded Luminance
Wet
50%
Recorded Luminance
Wet
25%
Recorded Luminance
July 2012
Pavement
Light Level
Luminance Data
Dry
100%
Recorded Luminance
Dry
50%
Recorded Luminance
Dry
25%
Recorded Luminance
Wet
100%
Wet Pavement not available
Wet
50%
Wet Pavement not available
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 28
Wet
25%
Wet Pavement not available
5 Findings
5.1 Written Evaluation
Researchers administered the written evaluation to participants on each night of the experiment.
Each participant rated twelve statements on a scale from “strongly disagree” to strongly agree.
The statements address perceptions of safety, comfort, glare, preference for light color and color
rendering, and overall style. Overall, 332 participants surveyed the test area, roughly split
between the wet and dry nights.
Table 11. Written Evaluation Results
Written Evaluation Results
Topic
Result
Statement 1
Safe, daylight
hours
Participants considered the street a safe area
Statement 2
Safe, darkness
hours
Participants considered the street a safe area, at
night, even with dimming.
Statement 3
Comfortable
No statistical difference between responses
within a given test area when the light output is
reduced to 25% light level.
Statement 4
Too much light
Participants did not rate the HPS luminaires as
having too much light.
Statement 5
Not enough light
Asymmetric luminaire showed agreement at all
light levels.
Statement 6
Uneven (patchy)
Asymmetric luminaire showed agreement at all
light levels.
Statement 7
Glare
Asymmetric had the best glare rating (lowest),
followed by the 3500K and 4100K.
Statement 8
Safe on sidewalk
HPS 400 W received highest rating, followed by
HPS 250 W.
Statement 9
Cannot tell colors
No statistical differences across color
temperatures, including HPS, across all dim
levels.
Statement 10
Safe vehicular
navigation
HPS luminaires received the highest ratings,
while the asymmetric received the lowest
ratings.
Statement 11
Color of light
Both HPS sources showed nearly the same
neutral preference as the LED sources.
Statement 12
Style of light
With the exception of the asymmetric luminaire,
participants preferred the LED luminaires as
much as the current 400 W and 250 W HPS
standards.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 29
In an effort to recruit enough survey participants, some participants came to both nights of
testing. The duplicate participants evaluated different road conditions and light levels each night.
The participants had a full twenty-four hours and bright daylight between the two nights of
testing, and researchers did not expect their participation on both evenings to significantly affect
the results. However, researchers analyzed the duplicate surveys as a group and compared their
responses to the larger population to identify any differences.
Statements S3, S5, S10, and S12 (comfortable, not enough light, vehicular navigation, and style)
exhibited different agreement ratings among those who saw the test areas twice. The participants
who saw the test areas twice agreed more strongly with these statements for the 250 watt HPS
under the wet condition than did those who went through the test areas once.
Statement S7, the light sources are glaring, participants viewing the asymmetric luminaire test
area for the second time, under dry conditions and at 25% of full light output, rated their
agreement with this statement higher than those who saw the asymmetric luminaire test area for
the first time.
When responding to statement S10, the lighting provides for safe vehicular navigation, for the
asymmetric and 3500K LEDs, participants seeing these areas, under wet conditions, for the
second time disagreed with this statement more strongly than did the first-time viewers.
Second-time viewers agreed with statement S12, I would like this style of lighting, more strongly
than did first-time viewers when evaluating the 250 watt and 400 watt HPS on the wet road
conditions.
When responding to statement S13, comparison of lighting to other City of Seattle streets, for the
3500K LED, participants seeing these areas for the second time disagreed with this statement
more strongly that did the first-time viewers.
These variations between the responses of the repeat participants and the larger study population
do not change the overall conclusions drawn from the written evaluation. While most of the
variations occurred when participants were viewing the existing HPS luminaires, it is unclear
why participants might have rated these more highly on the second night of viewing. Under wet
conditions, the standard LED products may have produced more reflected glare on the pavement.
The relatively lower glare of the more diffuse HPS sources may have been more noticeable to
individuals who had seen a less dramatic difference on the previous night.
To address potential bias in the experiment, researchers divided each of the three groups of
participants into two subgroups; half of the participants walked down one side of the street and
the other half walked down the other side of the street. For instance, the participants who
traveled from south to north generally viewed the LED test areas lower than those who traveled
north to south. Although evaluation ratings between the two groups exhibited some trends, the
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 30
majority of the values are not significantly different from one another. Researchers combined the
entire dataset, essentially averaging the two different vantage points.
When researchers analyzed this question by gender, women generally preferred the warmer color
temperatures while men tended to prefer cooler color temperatures, as shown in Figure 41 and
Figure 42.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 31
Figure 13. Survey Question 11: “I like the color of the light”– Dry Pavement
at One Hundred percent (Light Level by Gender)
Figure 14. Survey Question 11: “I like the color of the light”– Wet Pavement
at One Hundred percent (Light Level by Gender)
Research explains the reasoning behind women’s preference for warmer color temperatures (.
Some studies on gender and color indicate that women can match colors more accurately and
quickly than men. Only the X chromosome contains genes for pigments in red and green cones;
since women have two X chromosomes, they have a potential advantage over men for superior
vision if their two X chromosomes align in such a way that activates two red and two green
cones.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
LED
3500K LED
4100K 400W
HPS LED
Asym. LED
5000K 250W
HPS
Female
Male
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
LED
3500K LED
4100K 400W
HPS LED
Asym. LED
5000K 250W
HPS
Female
Male
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 32
5.2 User Field Test
Data Analysis Approach
Researchers performed analyses for visibility data and for illuminance sensor data. For the user
field test visibility analysis, researchers conducted an initial data cleaning in which they located
targets via GPS coordinates, verified responses and matched them to each target section, and
removed additional data anomalies (outliers) from the data. For example, researchers excluded
all data that exceeded three standard deviations away from the mean. Researchers performed an
additional data check to look for any other outliers and to check the images associated with the
data file. They did so by checking the data in Arc Map and verifying the image information.
Next, the entire data file, including the button input box, latitude and longitude information, and
respective images from the color and luminance cameras, was imported into a Statistical
Analysis Software (SAS) program for review and analysis. As an example, researchers obtained
the detection distance calculation by calculating the distance using latitude and longitude
coordinates for each button press. The coordinates for the target locations were registered
separately and also integrated to determine the distance from the button press to the target
location. Researchers rechecked these calculations using the distance calculation obtained from
the vehicle network data. When they had completed the distance calculations, the dataset
underwent additional data checking for outliers, and researchers made necessary corrections
(including deletions for false button presses and frame corrections, and deletions of anomalous
data). Researchers used Analysis of Variance (ANOVA) as the statistical tool to investigate
differences among lighting type, lighting location, target color, target location, travel direction,
and vertical illuminance level. Findings are shown in Table 12.
Table 12. Luminaire Type, Target Color, Pavement, and Light Level ANOVA Results
ANOVA Results
Source
F
value
Pr>F
Significant
Luminaire Type
38.13
<0.0001
*
Target Color
39.6
<0.0001
*
Light Level
0.01
0.9289
Pavement Condition
2.22
1.438
Luminaire Type * Light Level
0.24
0.8704
Luminaire Type * Pavement
0.83
0.5104
Luminaire Type * Light Level * Pavement
1.08
0.3675
Luminaire Type * Target Color
10.42
<0.0001
*
The illuminance data for the lighting sections underwent the same data cleaning process as the
visibility (or detection distance) data. Researchers checked the entire data file for anomalies and
verified sections with GPS information. They conducted additional spot checks using the color
images collected during the drive to verify the section location and starting/ending points for
each run.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 33
Researchers next imported the cleaned data file into SAS for review and analysis. The
illuminance data gave an approximation of the light intensity reaching the road surface, which
provided further understanding of the performance of the different lighting sections.
Researchers conducted an additional analysis using detection distance, illuminance, and
luminance in a linear regression model. This model allowed better visualization of the linear
relationship among the three variables.
Detection Distance
Researchers conducted an Analysis of Co-Variance (ANCOVA) on the detection distance and
illuminance data to identify any differences among the lighting sections. They used the Student
Newman-Keuls (SNK) test to identify where the significant differences occurred. Figure 15
below highlights the results.
Figure 15. Luminaire Type and Light Level by Detection Distance (Wet and
Dry Pavement Combined)
As Figure 15 shows, detection distance is not predictable based on the luminaire’s light level.
Note that the 250 W and 400 W HPS luminaires were not dimmed for the experiment; the
lighting level of the LED luminaires surrounding the HPS luminaires likely affected the contrast
of the targets in these sections.
Figure 16 shows comparisons of luminaire types and the pavement conditions by mean detection
distance. Dry and wet conditions alone did not exhibit statistically significant differences;
however, this relationship shows that the difference in pavement wetness condition did affect
some luminaires. The HPS luminaire types (250 W and 400 W) shared a similar trend with the
effect of the wet conditions. The differences for the LED luminaire types were more muted.
0
20
40
60
80
100
120
140
160
180
100 50 25 100 50 25 100 50 25 100 50 25 100 50 25 100 50 25
250W 3500K 400W 4100K 5000K ASYM
Mean Detection Distance (ft)
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 34
Figure 16. Luminaire Type and Pavement Condition by Detection Distance
(All Light Levels Combined)
Figure 17 shows a comparison of wet and dry conditions by light level. Again, dry and wet
comparisons alone yielded no statistically significant difference for the LED luminaires;
however, in this case the wet twenty-five percent condition is noticeably lower than the other
combinations. The fact that the twenty-five percent of full light output scenario for the wet
condition had significantly fewer trials than did the fifty and one hundred percent scenarios may
have contributed to its lower average, due to a larger margin of error.
Figure 17. Pavement Condition and Light Level by Detection Distance (All
LED Luminaires)
0
20
40
60
80
100
120
140
160
Dry Wet Dry Wet Dry Wet Dry Wet Dry Wet Dry Wet
250W 3500K 400W 4100K 5000K ASYM
Mean Detection Distance (ft)
0
20
40
60
80
100
120
140
100 50 25 100 50 25
Dry Wet
Mean Detection Distance (ft)
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 35
The other contributing factor is the potential for an overall reduction in glare from the dry road.
Wet pavement has a significantly higher specularity than dry pavement, thus causing greater
impacts of glare of the light source.
A Student Newman-Keuls (SNK) Test for both nested error terms (pavement condition and light
level) found target colors to be significantly different from one another. Participants detected
blue and red targets approximately twenty to thirty feet sooner than either gray or green targets.
Participants detected green targets with the shortest average distance of any of the four colors,
meaning participants took longer to identify the green targets than any of the other target colors.
Figure 18 illustrates the differences in the comparisons between luminaire type and target color
by detection distance. The varying spectral distributions due to the different CCTs of the LED
luminaires contributed to the range of detection distances by target color.
The 3500K luminaires have more red and green color content than do the other sources; this
explains the substantial drop-offs in blue detection for this light source, as the blue targets are
less activated than the other target colors. While the 5000K luminaires have the highest CCT and
have more blue content than the other sources, they did not outperform the 4100K luminaire in
blue target detection distance, suggesting either a difference in contrast or a wash-out of color.
The 5000K luminaires maintain a relationship similar to the 4100K luminaires across all target
colors except for the neutral gray, where the two performed nearly equally. The asymmetrical
LED luminaires performed on par with the 4100K luminaires with no statistical difference across
the target types.
The HPS luminaires performed well for the colors red and blue while dropping significantly for
gray and green. Neither HPS luminaire outperformed the 4100K luminaires or the asymmetrical
LED luminaires for any target color.
Colors gray and green exhibited significantly lower detection distances for the 250 W compared
to other luminaire types. The researchers interchanged gray and green targets by location, as they
did for red and blue targets. The location of some of the targets may have played a role in these
low averages, given researchers placed these targets at the start of the uphill portion. The
distance of approach to the targets may have been less than that of other test areas with targets of
the same color after the test vehicle changed direction. The 400 W HPS test area resulted in
lower gray and green detection distances suggesting that the yellowish hue provided by HPS
lamps negatively affects the visibility of green and neutral gray.
The results show that on average, the 4100K test area performed among the best for each target
color. Based on these results, the 4100K luminaire appears to provide a sufficient balance
between the red and blue extremes in the target color and is the most appropriate color
temperature for color detection for all of the targets.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 36
Figure 18. Luminaire Type and Target Color by Detection Distance (All Light
Levels)
Figure 19 illustrates the differences between test areas by pavement condition and light level.
The 4100K test area and the LED asymmetrical test area performed best overall with regard to
color detection distance. While Clanton & Associates did not dim the 400 W and 250 W HPS
luminaires for the twenty-five and fifty percent conditions, the HPS detection distances varied
for those conditions; the contrast of neighboring luminaires may have been a factor.
Interestingly, the one hundred percent light output condition did not always result in the best
detection distances; in scenarios such as LED asymmetrical dry conditions, detection distances
were higher at the dimmed state. As light level tends to affect detection distance in an
unpredictable way, researchers cannot form conclusions here; however, the results suggest a
possibility that dimming a luminaire as low as twenty-five percent of full light output and
reducing its energy use may not have a negative impact on detection distance. Notably, even
though researchers did not dim the HPS luminaires, the dimming of the surrounding luminaires
may have slightly affected the light levels in these test areas. Extraneous light sources such as
lights from businesses or neighboring parking lots may have slightly affected these results as
well.
The lack of a predictable trend between dry and wet conditions constitutes another noteworthy
finding. This suggests that the presence of a wet road surface does affect detection distance in
some form, perhaps due to higher spectral reflectance off of the roadway.
0
20
40
60
80
100
120
140
160
180
250W
3500K
400W
4100K
5000K
ASYM
250W
3500K
400W
4100K
5000K
ASYM
250W
3500K
400W
4100K
5000K
ASYM
250W
3500K
400W
4100K
5000K
ASYM
BLUE GRAY GREEN RED
Mean Detection Distance (ft)
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 37
Figure 19. Luminaire Type, Pavement, and Light Level by Detection Distance
Note: 250 W and 400 W operated at one hundred percent light level and were not dimmed
5.3 Contrast
Contrast is defined as the difference in luminance that renders an object visible. The contrast
metric used for these analyses is a formulation called Weber contrast, which is advantageous for
these types of analyses due to its consideration of negative contrast. Values above zero are
positive contrast, or the point at which an object is made visible by a dark background. Values
below zero are negative contrast, or the point at which an object is made visible by a lighter
background. Both negative and positive contrasts are represented here.
Equation 1. Weber Contrast Equation
     

The researchers assessed the contrast and luminance of the targets using a program created in
MATLAB® as part of a National Surface Transportation Safety Center of Excellence (NSTSCE)
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 38
endeavor. This data reduction used the still images, as shown in Figure 20, recorded by the
luminance cameras of the RLMMS. Data reductionists verified the validity of each image and
traced the outline of the visible target using a tool within the software. The luminance and
contrast of the outlined target are calculated by the program while considering the inside of the
trace and its surrounding elements.
Figure 20. Example Luminance Image
Figures 21 through 23 show the results of target contrast for each target color across all light
types for the dry pavement condition. Because these data came from the second data collection
effort, now wet pavement condition was available to be recorded for this analysis. A comparison
of the contrast results for each dim level shows a slight trend from negative contrast to positive
contrast as the illuminance is decreased (one hundred percent output versus twenty-five percent
output). VTTI did not expect this finding, because as the illuminance on the roadway decreases,
the roadway luminance and the target luminance would also decrease and therefore the contrast
would remain the same. This increasing trend toward positive contrast indicates that the
luminance of the face of the target does not drop as significantly as the luminance of the roadway
surface. This implies that the ambient lighting from the areas off of the roadway provides some
illumination on the target face and as the roadway dims, the ambient lighting becomes a more
significant component of the target luminance.
These results notably demonstrate that although the contrast changes, the detection distance from
the one hundred percent light level to the twenty-five percent light level did not change.
Researchers expected to see this finding, given that as the driver adaptation is reduced, the
threshold luminance difference required for visibility is also reduced, resulting in an equivalent
visibility distance.
In order to remove the impact of headlamps, VTTI recorded the contrast in these figures at least
200 feet from the target locations, while they recorded the average detection distances for the
targets all within 200 feet.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 39
Figure 21. Target Contrast of One Hundred percent Lighting Level, Dry
Pavement Condition, per Target Color across All Light Sources
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 40
Figure 22. Target Contrast of Fifty percent Lighting Level, Dry Pavement
Condition, per Target Color across All Light Sources
Note: 250 W and 400 W remained at one hundred percent
Figure 23. Target Contrast of Twenty-Five percent Lighting Level, Dry
Pavement Condition, per Target Color across All Light Sources
Note: 250 W and 400 W remained at one hundred percent
Figure 24 illustrates the impact of headlamps from the test vehicle. VTTI took a vertical
illuminance measurement at target height every twenty-five feet, beginning at twenty-five feet
from the vehicle to 300 feet ahead of the vehicle. The figure shows the calculated difference of
the measurements with headlamps on versus off. The greatest headlamp impact occurs at a
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 41
distance of fifty feet from the vehicle, where headlamps contribute up to eighty-five lux of light.
The impact is reduced at twenty-five feet from the vehicle, where the light of the headlamps goes
over top of the target. At 200 feet from the vehicle to the target, the test vehicle headlamps have
little-to-no impact on small target visibility.
For detection distances of one hundred feet or less, assuming that headlamps provide a
substantial contribution to visibility would be accurate. As dimming occurs, the headlamp
becomes the dominant cause of the detection. As a headlight becomes the dominant detection
mechanism, the vertical illuminance on the face of the target increases and the contrast of the
target increases. As the roadway becomes dimmer, detection is limited by headlamp distance
rather than by illuminance provided by overhead lighting. Managing this effect is crucial as the
potential exists for a target to go through an invisibility period during the transition from
negative to positive contrast.
Figure 24. Impact of Headlamps by Distance from Vehicle
5.4 Illuminance and Detection Distance
Figure 25 illustrates the relationship between the average horizontal illuminance of each test area
and the corresponding differences in detection distance for each pavement condition. The red
line represents the average horizontal illuminance for the luminaire’s test area; the bars represent
the average detection distance. Analyses found no significant differences among the light levels
for detection distance. No other relative trends with light level existed for dry pavement;
however, the average detection distances on wet pavement do trend similarly to the horizontal
illuminance. This suggests that horizontal illuminance has a greater impact on wet pavement than
dry, again likely due to the specularity of the pavement surfaces and the potential for increased
glare. Note that the illuminance figures are representative of the horizontal illuminance within
the entire luminaire’s test area and not just at the target’s location.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 42
Figure 25. Luminaire Type and Light Level by Detection Distance for Both Wet and Dry
Conditions
Results indicate significant advantages to the LED 4100K test area and to the LED asymmetrical
test area when crossed with light level and pavement condition. Light levels and pavement
conditions resulted in no significant differences; however, significant differences did exist by
luminaire type, as illustrated by the SNK results in Figure 26 (between each luminaire by
groups). Each column labeled with a different letter (such as A or B) signifies significance
between the test areas. The 4100K test area and the LED asymmetric test areas are both in group
A and thus provided a significantly better detection distance than the other groups. The 400 W
HPS is within both groups B and C, suggesting it does not significantly differ from either B or C
but is significantly different from groups A and D.
100
100
100
100
100
100
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 43
Figure 26. Luminaire Type by Detection Distance across Both Pavement
Conditions and Light Levels
5.5 Lighting Metrics
The following six figures (Figures 27 through 32) represent the average horizontal illuminance
gathered as the experimental vehicle traveled northbound and southbound with the RLMMS
equipment. These data average the readings from each sensor (left, right, rear and front) of the
spider apparatus mounted atop the vehicle. The spikes represent the peak output of an individual
luminaire. Each test area is divided and labeled in the figures below.
Northbound and southbound readings are similar within each luminaire’s light level, except for
the 250 W HPS, which produced a higher illuminance on the northbound side.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 44
Figure 27. Illuminance per Section at One Hundred percent, Northbound
Figure 28. Illuminance per Section at One Hundred percent, Southbound
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 45
Figure 29. Illuminance per Section at Fifty percent, Northbound
Figure 30. Illuminance per Section at Fifty percent, Southbound
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 46
Figure 31. Illuminance per Section at Twenty-Five percent, Northbound
Figure 32. Illuminance per Section at Twenty-Five percent, Southbound
The profile of the asymmetric test area is noteworthy as it does not provide the evenly balanced
light distribution. Again, the 400 W HPS and 250 W HPS test areas were not dimmed.
Tables 13 through 15 show metrics for all of the test areas. The relationship between each
luminaire’s horizontal illuminance and pavement luminance is shown in the following tables for
each light level. Uniformity ratios are also included and are noted in the tables as Avg/Min and
Max/Min.” The illuminance method uses only the “Avg/Min” value, while the luminance
method uses both.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 47
The maximum Avg/Min uniformity ratio for the horizontal illuminance in RP-8 is 4.0 for a
collector roadway with a medium pedestrian conflict, but because of the elevated plane of the
RLMMS (approximately seventy-six inches), VTTI did not expect the values here to meet those
criteria.
The maximum uniformity ratio for pavement luminance in RP-8 is 3.5 for Avg/Min and 6.0
for Max/Min. The asymmetrical luminaire exceeded the maximum “Max/Min” uniformity
ratio of 6.0 only at 50 percent of full light output level.
The RP-8 recommended average luminance for a collector roadway with a medium pedestrian
conflict is 0.6 cd/m2. At the one hundred percent lighting level, only the 400 W HPS was able to
achieve this recommended value. Although simulated values indicated that all of the Type II
distribution LED luminaires exceed 0.6 cd/m2, environmental conditions resulted in lower actual
average luminance levels. This kind of variance between simulated values and measured values
is not uncommon.
These data also demonstrate the inaccuracy of the current uniformity metric to adequately
represent the lighting distribution. The asymmetrical design has an average uniformity ratio that
is similar to the full distribution luminaires. However, the characterization measurements and the
known distribution both indicate that uniformity is lower with the asymmetrical luminaire. These
findings suggest a need for additional consideration to fully characterize the roadway appearance
with special luminaire types.
Uniformity is also important in target detection. The 4100K luminaire exhibited the highest
“Avg/Min” uniformity ratio for horizontal illuminance at both one hundred percent and fifty
percent, indicating the most non-uniform appearance; it also had the highest visual performance.
The non-uniformity of the lighting on the roadway surface seems to provide a visibility
enhancement and greater contrast for visibility; however, further efforts to more fully define
uniformity requirements and their importance in visibility are necessary.
Table 13. Light System Calculations, One Hundred percent Light Level
100%
Horizontal Illuminance
at Grade (lux)
Dry Pavement Luminance
(cd/m2)
Test
Area
Avg
Avg/Min
Avg
Avg/Min
Max/Min
250 W
36.93
3.98
0.54
1.66
2.39
3500K
21.83
3.80
0.45
1.55
3.03
400 W
54.88
4.45
0.67
1.53
2.36
4100K
20.44
8.63
0.43
1.95
4.46
5000K
21.97
2.87
0.49
1.48
3.47
ASYM
18.89
6.79
0.40
1.97
5.66
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 48
Table 14. Light System Calculations, Fifty percent Light Level
50%
Horizontal Illuminance
at Grade (lux)
Dry Pavement Luminance
(cd/m2)
Test
Area
Avg
Avg/Min
Avg
Avg/Min
Max/Min
250 W
36.93
3.98
0.53
1.81
2.62
3500K
17.45
5.68
0.38
1.50
3.08
400 W
54.88
4.45
0.65
1.54
2.49
4100K
16.27
8.41
0.35
1.78
4.60
5000K
17.00
2.97
0.44
1.59
3.62
ASYM
14.85
6.88
0.34
2.12
6.46
Table 15. Light System Calculations, Twenty-Five percent Light Level
25%
Horizontal Illuminance
at Grade (lux)
Dry Pavement Luminance
(cd/m2)
Test
Area
Avg
Avg/Min
Avg
Avg/Min
Max/Min
250 W
36.93
3.98
0.52
1.61
2.39
3500K
13.60
5.78
0.32
1.79
4.19
400 W
54.88
4.45
0.66
1.56
2.43
4100K
11.01
8.08
0.33
1.55
4.09
5000K
10.54
8.25
0.38
1.71
4.21
ASYM
9.83
7.21
0.32
1.96
5.25
Figure 33 illustrates the data from the preceding three tables: dry pavement luminance per
luminaire section. However, because the follow-up data collection did not collect wet pavement
data, that difference is not illustrated here.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 49
Figure 33. Average Roadway Luminance: Dry Pavement Condition
5.7 Sidewalk Lighting Characteristics
The second luminance data collection effort also considered sidewalks as part of the
investigation. Researchers used a Minolta T-10 illuminance meter and a hand-held Minolta LS-
110 luminance meter to measure the vertical illuminance on a pedestrian, the sidewalk
luminance, and the light trespass from the roadway. VTTI used the meters to recorded vertical
illuminance at pedestrian height, or five feet from the ground.
Figure 34 shows that at the one hundred percent light level, the 400 W HPS exhibited by far the
greatest vertical illuminance, almost two times that of the next-brightest measurement. The
5000K luminaire demonstrated unpredictable results, as it showed a vertical illuminance at the
fifty percent lighting level less than that observed at twenty-five percent. However, the 5000K’s
vertical illuminance at the one hundred percent light level is surprisingly more than two times
greater than that of the other LED luminaires; however, this anomaly may be attributable to
contributions of businesses and other off-roadway lighting in the area of the 5000K installation.
The 3500K, 4100K, and asymmetrical LED luminaires demonstrated similar sidewalk vertical
illuminance for each light level.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 50
Figure 34. Vertical Illuminance on Sidewalks by Dim Level and Luminaire Type
An IES publication, “Lighting for Exterior Environments,” recommends a vertical illuminance
level of five to twenty lux based on the type of area surrounding the installation (IES RP-33
1999). Among the new technologies considered, notably only a single LED (5000K at one
hundred percent light level) met this criterion, with a vertical illuminance of 5.5 lux.
As shown in the recorded average sidewalk luminance results in Figure 35, the HPS light sources
produced the highest luminance at approximately 2.0 cd/m2 each. The 5000K luminaire produced
unexpected results as it achieved greater luminance at the fifty percent light level than at the one
hundred percent light level. Off-site lighting from fuel stations, restaurants, and nightclub
signage may have contributed to some of the higher values in certain areas. The 3500K, 4100K,
and asymmetric LED luminaire results behave predictably, as the luminance increases with light
level. No luminance level among these three types of luminaires exceeded 0.5 cd/m2.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 51
Figure 35. Average Sidewalk Luminance by Lighting System and Brightness Level
5.8 Light Trespass
Researchers used a hand-held illuminance meter to record light trespass at the property line.
They took this measurement along the side of the roadway with the illuminance meter facing into
the road; measurements represent the amount of light leaving the roadway onto the adjacent
properties, as Figure 36 shows. The 400 W HPS produced nearly 120 lux of light trespass, while
the 250 W HPS provided approximately thirty lux. All LED luminaires provided light trespass of
ten lux or below, which meets the IES criteria from TM-11 as shown in Table 25 in Appendix E.
Clanton & Associates selected these luminaires specifically for this application. The light
trespass is a characteristic of the luminaire and is not a reflection of the light source technology.
It does, however, highlight the potential for improved lighting designs based on the optical
design controllability of the LED light sources.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 52
Figure 36. Light Trespass at Property Line by Lighting System and Brightness Level
5.9 Glare
Glare comes in two forms: discomfort and disability. Discomfort glare is measured using a
subjective rating scale; disability glare, or veiling luminance, can be measured in the field. IES
RP-8 offers a formula for calculating veiling luminance (IES RP-8 2005):
Equation 2. Veiling Luminance
            

 
        
  
Using a glare meter placed vertically at eye level, a researcher can calculate the amount of
veiling luminance and assess the quantity of glare. Higher lux values reaching the meter typically
result in more glare or veiling luminance.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 53
Figure 37 shows the average vertical illuminance recorded by the glare meter placed inside the
windshield of the experimental vehicle for each test area and light level. The results indicate that
the HPS test areas produce more glare than do the LED test areas, as evidenced by their higher
average vertical illuminance values. The glare meter’s vertical illuminance output values can be
affected by neighboring luminaires, billboard lights, or business lights.
Figure 37. Test Area and Light Level by Mean Vertical Illuminance (lx) of Glare
The asymmetrical design generates particular interest here; this luminaire was designed to
project the light in the direction away from the driver. It appears that the glare level experienced
in the asymmetrical test area represents the glare from the environment around the roadway and
not from the luminaires themselves. Environmental glare is location-dependent and can vary
based on proximity to lighted businesses, stadiums, billboards, and campuses. Road geometry
also plays an important role as vehicle orientation can determine the angle at which light enters
the windshield. Environmental glare is defined as the overall impression of the ambient light in
and around the street.
The overall results suggest that since the LED luminaires tested here have a maximum glare
rating of IES G2, these luminaires have less glare consequences than current lighting
technologies. Again, given that researchers selected these luminaires for their low glare
characteristics, this statement relates more to the luminaires, rather than to the LED lighting
technology.
Table 16 and Table 17 detail the average veiling luminance for each luminaire by direction of
travel. These values are typically compared to the IES RP-8 recommended maximum veiling
luminance ratio of 0.4 for a collector roadway grade. However, because RP-8 requires
consideration only of the roadway luminaires, these results are not directly comparable to the
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 54
RP-8 guidelines. The glare meter placed in the experimental vehicle records vertical illuminance
from multiple sources of light, including those not placed on the roadway and reflections, thus
resulting in a ratio much greater than the 0.4 guideline. Luminaires placed at intersections not
included in the study, lights from maintenance vehicles, and business signs factor into the large
Lv values. However, the LED designs have lower glare ratios than those of the HPS luminaires.
Inter-comparisons of the technologies are important. In this demonstration, the HPS luminaires
both had glare ratings higher than the LED replacements, particularly compared to the
asymmetrical system. These values show the superior beam control of the LED systems. This
lower ratio to the average luminance also indicates that streetlight designers and engineers can
use a lower overall average while still providing the same visual performance, as the lighting
system does not need to overcome the detrimental impact of the glare.
Table 16. Veiling Luminance, Dry Pavement
Veiling Luminance, Dry Pavement
Light Type
Light Level
Northbound
Lv
Southbound
Lv
Lvmax/Lavg
Veiling
Luminance
Ratio
North
South
250 W HPS
100%
4.491
3.387
1.850
1.399
100%
4.415
3.911
1.954
1.499
100%
3.969
3.898
2.242
2.202
LED 5000K
100%
3.016
3.391
1.268
1.429
50%
2.467
2.893
1.115
1.307
25%
1.859
2.531
1.158
1.458
LED ASYM
100%
1.567
1.939
0.713
0.905
50%
1.526
1.988
0.746
0.972
25%
1.349
2.034
0.841
1.268
400 W HPS
100%
6.232
5.402
2.701
2.341
100%
6.561
5.378
3.056
2.505
100%
6.131
5.583
3.641
3.316
LED 4100K
100%
2.493
3.299
1.099
1.455
50%
2.145
2.630
1.017
1.247
25%
1.737
2.261
1.049
1.366
LED 3500K
100%
3.006
2.616
1.385
1.206
50%
2.677
2.587
1.332
1.287
25%
1.986
1.525
1.256
0.965
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 55
Table 17. Veiling Luminance, Wet Pavement
Veiling Luminance, Wet Pavement
Light Type
Light Level
Northbound
Lv
Southbound
Lv
Lvmax/Lavg
Veiling
Luminance
Ratio
North
South
250 W HPS
100%
5.339
5.717
2.518
2.696
100%
5.265
5.261
2.842
2.658
100%
5.155
4.879
3.325
3.147
LED 5000K
100%
3.281
3.861
1.575
1.853
50%
4.031
3.434
2.079
1.772
25%
2.890
2.952
1.901
1.942
LED ASYM
100%
4.416
4.421
2.294
2.296
50%
4.453
4.670
2.487
2.607
25%
3.221
3.993
2.292
2.841
400 W HPS
100%
4.901
4.998
2.426
2.474
100%
4.965
5.300
2.641
2.819
100%
4.818
5.173
3.266
3.507
LED 4100K
100%
4.684
4.751
2.358
2.392
50%
4.462
4.244
2.414
2.296
25%
4.203
3.552
2.898
2.449
LED 3500K
100%
3.460
3.786
1.821
1.992
50%
3.079
2.999
1.741
1.696
25%
3.196
2.351
2.304
1.695
The results indicate that glare is higher in the wet environment for all luminaires except for the
400 W HPS. Light from businesses along the side of the roadway likely contributed to the
differences in glare directionality. The southbound side of the road beyond the shoulder
consisted mainly of residential blocks or closed businesses, while the northbound side consisted
of multiple open or well-lighted business locations.
5.10 Spectral Power Distribution
The spectral distribution of the available light constitutes an important factor in color detection.
The spectral power distributions (SPDs) for each luminaire are shown below in Figure 38. All of
the luminaires show the typical sharp spike at 440 nm, which represents the amount of blue
power needed to drive the other light production through the LED phosphor. The 5000K CCT
shows comparatively little output in the higher wavelength regions. The 3500K produces less
blue than the other luminaires but more yellow-orange spectral power. On the CCT spectrum, the
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 56
3500K luminaire is in the yellow-to-white transition and is the closest relation to the HPS
luminaires in terms of CCT among the LED luminaires tested. The asymmetric LED and 4100K
LED are very similar in spectral power distribution due to their comparable color temperatures.
The 4000K-4500K CCT range is considered neutral white light.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 57
Figure 38. Spectral Power Distributions of LED Luminaires
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 58
6 Discussion
6.1 Comparison to Previous Studies
The previous streetlight studies performed by Clanton & Associates in Anchorage, San Diego,
and San Jose all had different study parameters, luminaire wattages, color temperatures, light
sources, and existing pole layouts. Therefore, making direct comparisons to these findings would
be difficult and likely valueless. For example, the City of San Diego had already decided to
implement relatively low color temperature street lighting and did not test a large range of CCTs.
Test conditions in Anchorage included snow (dramatically different contrast than the other cities)
and car headlights, because that study did not include a road closure. San Jose street lighting
forms a staggered pole arrangement, while Seattle’s forms an opposite pattern.
However, a few trends do appear among the studies in terms of preference for color and light
source. In both San Jose and San Diego, the participants preferred the 3500K LED luminaire
when responding to the overall style of the lighting. In Anchorage and Seattle, respondents
preferred the white light LEDs over the existing high pressure sodium lights, but no specific
color temperature stood out as most preferable. In all studies, survey participants considered
white light LED and induction sources to be acceptable.
Findings from the user field test data contain similarities to other streetlight studies. For example,
in San Jose the 4000K LED performed better compared to the HPS and 3500K LED tested in the
study. However, in that study, the 5000K luminaire performed the best which was not the case
in Seattle. Researchers cannot compare the relative detection distances between the two studies
due to differences in road geometry, but the studies did show that the 3500K luminaire is not
optimal for visibility among the LEDs tested in these two locations.
Three LED luminaires tested in Anchorage had color temperatures of 3500K, 4100K, and
4300K. All three luminaires came from different manufacturers; however, when comparing the
performance of their CCTs, the 4100K outperformed the 3500K by approximately seven meters
(twenty-three feet). Although target color, pavement type, and the presence of snow distinguish
the Anchorage study from the Seattle study, the twenty-three foot detection distance difference is
similar to that observed in Seattle. The Anchorage testing included no LEDs with CCTs above
4300K; however, the 4100K outperformed the 4300K, which exhibited performance results
similar to that of the 3500K. These differences could be attributed to road geometry at the test
site, pavement type, or the differences in manufacturers.
In San Diego, the test included two LED luminaires in the roadway portion of the study. Both
luminaires had manufacturer-stated color temperatures of 3500K, although on-site measurements
showed them to be different; one measured at 3475K and the other at 4560K. Counter to findings
in the other three test locations, the luminaire with a CCT closer to 3500K significantly
outperformed the one measured at 4560K, by approximately thirty feet.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 59
Based on the findings of the previous streetlight studies together with the data collected in
Seattle, CCT clearly affects visibility. These results indicate that 4000K and 4100K luminaires
regularly outperform CCTs of 3500K and below, and perform just as well (San Jose) or better
(Seattle) than CCTs of 4300K and above.
This research also illuminates inaccuracies in current uniformity rates. Uniformity is important in
target detection, as a greater uniformity ratio indicates a non-uniform appearance. The 4100K
luminaire provided the highest visual performance and the highest uniformity ratio. The
uniformity ratio of the asymmetric design is lower than that seen with the full distribution
designs, warranting consideration of how to fully characterize roadway appearance with special
luminaire types. These results call for future efforts to fully define uniformity requirements as
they relate to visibility.
Light trespass constitutes another interesting result. Only the LED luminaires met the IES criteria
from TM-11. Although Clanton & Associates selected these luminaires specifically for this
application, the results indicate that the light trespass is a characteristic of the luminaire and is
not indicative of the light source technology. It does, however, highlight the potential for
improved lighting designs based on the optical design controllability of the LED light sources.
6.2 Adaptive Lighting Opportunities
Traditionally, engineers design street lighting around the worst set of conditions that can exist for
a particular street based upon vehicular volume, pedestrian volume, and ambient luminance. In
reality, these worst-case conditions occur only part of the time. The rest of the time, traffic and
pedestrian volumes are reduced. With the advancement of network controls for exterior lighting,
streetlight designers and engineers can tune light output via adaptive lighting to deliver the
appropriate amount of light based upon the corresponding vehicular and pedestrian volume
present at a particular time. The implementation of adaptive lighting not only reduces the overall
energy consumption of the streetlights, it also prevents over-lighting, reduces glare, and
minimizes light pollution. Both the IES and the International Commission on Illumination (CIE)
provide for implementation of adaptive lighting in different forms.
The user field test results from this study indicate that the implementation of adaptive lighting
does not significantly affect object detection distance under dry conditions. However, when
coupled with the written evaluation results, pedestrians consider reducing the light level to
twenty-five percent of full light output for all hours of the night to be unacceptable. This result is
not surprising. Tuning the light to a point such as twenty-five percent of full light output (when
roads are dry) is justified at low vehicular and pedestrian volumes, but not for all hours of the
night. While some industry metrics allow the implementation of adaptive lighting, each city can
determine how to best apply adaptive lighting to its particular traffic conditions and community.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 60
6.3 Future Design Standards
Traditional streetlights have long been a primary cause of light trespass. While sometimes the
light leaving the back of the luminaire illuminates the sidewalk behind the luminaire, frequently
the light leaving the back of the luminaire trespasses into residential windows, causing
discomfort to the occupants. To combat the latter issue, luminaire manufacturers began offering
house side shields for specification. These shields, which can be mounted either internally or
externally, allow better optical control and reduce the amount of backlight leaving the luminaire.
LEDs by nature are directional sources and do not disperse light toward the back of the luminaire
unless purposely designed to do so. This technology advancement decreases the amount of light
trespass potential, but consequently leaves the sidewalks behind the luminaires with less light.
Controls coupled with LED streetlights provide valuable energy savings for end users. The
integration of these two products also provides a unique opportunity for innovative design. While
reducing the light level by seventy-five percent minimally affected detection distances, some
participants expressed concern that the sidewalks were difficult to navigate through the written
evaluation comments. Although this study was not designed for participants to evaluate the
sidewalk separately from the roadway, conducting such a study in the future may be beneficial.
Higher-wattage LED luminaires often contain two LED boards and two drivers. Right now, the
luminaire controllers send signals to which both drivers respond, and dim accordingly. However,
manufacturers could design controllers to send unique signals to each driver, in which case one
LED board could dim to a different light level than the other. This dual control capability might
alleviate residents’ apprehensions about reducing light levels on the roadway but maintaining the
light level on the sidewalks simultaneously. Municipalities could still realize energy savings
because half or more of the lights that contribute to the roadway could be dimmed.
6.4 Economic Analysis
The economic analysis below illustrates the economic viability of replacing the 400 W HPS
luminaires with 105 W LED luminaires with adaptive controls. This analysis included sixty
luminaires, similar to that of the demonstration test site.
Assumptions include real costs from the City of Seattle for the existing light sources, wattages,
hardware, and maintenance. The LED luminaire prices are based on Lumec RoadStar luminaires
used in the actual demonstration. The analysis used the Municipal Solid State Street Lighting
Consortium’s (MSSLC) calculator, released in 2011 by the Department of Energy and the Pacific
Northwest National Laboratory (PNNL).
Since this economic analysis included only a limited quantity of luminaires and only one
manufacturer, actual pricing for larger quantities and multiple manufacturers would be lower.
Scenarios
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 61
The six economic analysis scenarios listed below illustrate a variety of opportunities, including
different luminaires and the use of adaptive lighting. This analysis examines the economics of
sixty LED luminaires installed in Seattle, Washington with technologies available as of March
2012.
400 W HPS no adaptive lighting
o This is typical street lighting for many cities.
250 W HPS no adaptive lighting
o This is typical street lighting for many cities.
105 W LED no adaptive lighting
o This approach allows for immediate energy savings and the light level remains
constant throughout the night.
105 W LED with adaptive lighting (fifty percent light output for six hours per night)
o This approach allows for immediate energy savings. The light level changes
throughout the evening when traffic and pedestrian volumes are reduced.
105 W LED with adaptive lighting (fifty percent light output for three hours per night,
twenty-five percent light output for three hours per night)
o This approach allows for immediate energy savings. The light level changes twice
throughout the evening as traffic and pedestrian volumes are reduced.
Assumptions and Limitations
Each lighting scenario reveals the economic costs and potential benefits to utilities and cities in
the Northwest for investing in LED street lighting and controls. Inputs such as rebates, cost of
labor, cost of power, and greenhouse gas emissions are based on information provided by Seattle
City Light and provide a representative magnitude of cost and returns when implementing LED
street lighting in the Northwest region.
The MSSLC calculator gathers project inputs to generate scenarios that a project manager uses to
understand implementation options such as size, period of installation, cash flows, and a
project’s overall return on investment. While each entity in the Northwest will have different
costs (materials and labor) and rebate programs to consider, this analysis uses SCL’s rebate
incentives and cost information as a representation of what other utilities and consumer
ratepayers may find within the region.
The different street lighting scenarios use the same quantity of poles and luminaires, wattages,
and control systems scenarios, as described within this study. Even scenarios without dimming
include the cost of the control system due to several other non-energy benefits of using the
control system, such as asset management and maintenance alerts. The analysis considers the
cost of maintenance and captures the financial benefit of the longer source life of LEDs over
HPS. The fact that dimming LEDs extends the life of the source is well-understood; however,
given the lack of a precise quantification of this change in life, the analysis did not include this
benefit.
Application
Based on the results of the visual acuity study, the economic analysis assumes that smart-
controlled LEDs can be used as viable replacements for standard HPS luminaires in the
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 62
Northwest region. Since the detection distances in the dimmed LED scenarios compare well to
the baseline non-dimmed HPS scenarios, energy and maintenance savings can be achieved with
lower-wattage LED luminaires controlled by adaptive lighting with reasonable payback periods.
Analysis
The following tables describe the luminaires and control system components used in the analysis.
Scenario 1A: 400 watt HPS (no dimming) replaced with 105 watt LED (no dimming)
Table 18. Scenario 1A Economic Analysis Summary
Economic Analysis Summary
Luminaires Installed
60
Implementation Period (years)
1
Simple Payback (years)
3.3
Annual kWh Savings
94,802
Annual Energy Cost Savings ($)
$11,760
Baseline 400 W HPS (no dimming) Baseline Annual kWh Use
125,356
Baseline 400 W HPS (no dimming) Annual Energy Cost ($)
$15,669
105 W LED (no dimming) Annual kWh Use
31,273
105 W LED (no dimming) Annual Energy Cost ($)
$3,909
Scenario 1B: 400 watt HPS (no dimming) replaced with 105 watt LED (with fifty percent
dimming for six hours per night)
Table 19. Scenario 1B Economic Analysis Summary
Economic Analysis Summary
Luminaires Installed
60
Implementation Period (years)
1
Simple Payback (years)
3.1
Annual kWh Savings
101,901
Annual Energy Cost Savings ($)
$12,738
Baseline 400 W HPS (no dimming) Annual kWh Use
125,356
Baseline 400 W HPS (no dimming) Annual Energy Cost ($)
$15,669
105 W LED (fifty percent dimming) Annual kWh Use
23,455
105 W LED (fifty percent dimming) Annual Energy Cost ($)
$2,932
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 63
Scenario 1C: 400 watt HPS (no dimming) replaced with 105 watt LED (fifty percent dimming
for three hours per night, and twenty-five percent output dimming for three hours per night)
Table 20. Scenario 1C Economic Analysis Summary
Economic Analysis Summary
Luminaire Installed
60
Simple Payback (years)
2.9
Annual kWh Savings
105,810
Annual Energy Cost Savings ($)
$13,226
Baseline 400 W HPS (no dimming) Annual kWh Use
125,356
Baseline 400 W HPS (no dimming) Annual Energy Cost ($)
$15,669
105 W LED (fifty percent dim for three hours; twenty-five percent
output dim for three hours) Annual kWh Use
19,546
New Baseline Annual Energy Cost ($)
$2,443
Scenario 2A: 250 watt HPS (no dimming) replaced with LED 105 watt (no dimming)
Table 21. Scenario 2A Economic Analysis Summary
Economic Analysis Summary
Luminaires Installed
60
Simple Payback (years)
6.7
Annual kWh Savings
46,778
Annual Energy Cost Savings ($)
$5,847
250 W (no dimming) Annual kWh Use
77,526
250 W (no dimming) Annual Energy Cost ($)
$9,691
105 W LED (no dimming) Annual kWh Use
30,748
105 W LED (no dimming) Annual Energy Cost ($)
$3,843
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 64
Scenario 2B: 250 watt HPS (no dimming) replaced with 105 watt LED (fifty percent dimming
for six hours)
Table 22. Scenario 2B Economic Analysis Summary
Economic Analysis Summary
Luminaire Installed
60
Simple Payback (years)
5.8
Annual kWh Savings
54,465
Annual Energy Cost Savings ($)
$6,808
250 W HPS (no dimming) Annual kWh Use
77,526
250 W HPS (no dimming) Annual Energy Cost ($)
$9,691
105 W LED (fifty percent dimming) Annual kWh Use
23,061
105 W LED (fifty percent dimming) Annual Energy Cost ($)
$2,883
Scenario 2C: 250 watt HPS (no dimming) replaced with 105 watt LED dimmed to fifty percent
for three hours per night and twenty-five percent dim output for three hours per night
Table 23. Scenario 2C Economic Analysis Summary
Economic Analysis Summary
Luminaire Installed
60
Simple Payback (years)
5.3
Annual kWh Savings
58,309
Annual Energy Cost Savings ($)
$7,289
250 W HPS (no dimming) Annual kWh Use
77,526
250 W HPS (no dimming) Annual Energy Cost ($)
$9,691
105 W LED (fifty percent dimming) Annual kWh Use
19,217
105 W LED (fifty percent dimming) Annual Energy Cost ($)
$2,402
The sensitivity tables below show the effects on an LED project’s payback when certain cost
factors are changed. The central point of convergence on the graphs represents the baseline
payback for the HPS baseline scenarios converted to LED luminaires: one graph for a 400 watt
HPS to 105 watt LED conversion (Scenario 1C) and the other for a 250 watt HPS to 105 watt
LED conversion (Scenario 2C). The graphs show different potential project costs, altering the
project’s payback, so each cost sensitivity curve shows its impact on the payback.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 65
The central point of intersection, or the baseline scenario, is at a 3.3-year payback for a 400 watt
HPS conversion to 105 watt LED with adaptive dimming controls. The cost factors varied from
this baseline for each scenario are:
Rebate value
Equipment cost
Labor rate
Effective hours of operation (based upon dimming levels)
Cost of energy
Figure 39. 400 W HPS Replaced with 105 W LED (Scenario 1A)
The spider graphs show how the base case changes when a single cost variable changes. Points
on the graph represent an MSSLC calculation with the single variable either increasing or
decreasing. The X-axis represents a percent change in the cost variables; the Y-axis shows the
resulting change in the payback time in years. For example, looking at the cost of power curve, a
fifty percent increase in the cost of power causes the base case payback to drop from 3.3 years to
roughly 2.1 years. As another example, a fifty percent decrease in the labor rate (of maintenance)
has very little impact on the payback, reducing payback only about 0.1 years to roughly 3.2
years. A variable with a steep curve, such as the cost of power, shows that a small change in this
variable has a large impact on the payback. A shallow curve, such as the labor rate, reveals that
even a large change in this variable creates very little impact on the payback.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 66
Interestingly, control systems that allow for dimming lumen output may not have as substantial
an impact on payback as other costs do. “Hours of Operation” represents the use of dimming
with adaptive lighting and its impact on payback period. Note that this curve remains relatively
flat compared to the curves representing “Cost of Power,” “Rebates,” and “Equipment Costs.”
The single largest economic benefit comes from converting the 400 watt HPS to 105 watt LED.
Rebating the initial cost of the LED luminaire has a larger impact than dimming the LEDs.
Dimming may offer other benefits such as lower energy use (carbon footprint) and less light
pollution.
The 250 watt HPS replacement scenario has a steeper and more sensitive curve than the 400 watt
HPS “Hours of Operation” curve. The dimming savings represent a larger percentage of the
overall cost savings than that for the 400 watt HPS replacement scenario.
Figure 40. 250 W HPS Replaced with 105 W LED (Scenario 2A)
The more dramatic slopes defining the curves for “Cost of Energy,” “Rebates,” and “Equipment
Costs” have some notable similarities. Rebates act to effectively reduce the cost of equipment,
reducing the payback period, as reflected in the slope steepness. Likewise, if the cost of
equipment decreases, the payback also occurs in fewer years. Ideally, lower equipment costs and
rebates will greatly reduce the payback period.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 67
The “Cost of Power” curve acts as expected. As cost of power increases, kWh savings are more
valuable, reducing the payback period. Again, rebates are a valuable incentive, especially in
regions with low energy costs.
However, the “Cost of Power” curve is also unique among the cost curves because it is ongoing
and typically rising (typically a three percent increase per year). If a manager implements an
LED project and a year later an unexpected jump in the cost of power occurs, the project will pay
back sooner than originally forecasted. The chart below shows that a forty percent change in the
cost of power reduces the project’s payback by one year.
Another way to understand the current situation regarding control systems is that they increase
the “Cost of Equipment” curve (a very sensitive curve) while not increasing the kWh savings
enough to justify their current expense in all cases. Control systems need not be valued only as
energy-saving tools; streetlight designers and engineers should also consider their other value-
added services that go well beyond energy benefits.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 68
7 Conclusions
7.1 Written Evaluation
The written evaluation results provide valuable feedback on how well the different lighting
systems meet community expectations. The results of the written evaluation indicate a
preference for the incumbent HPS streetlights, while still showing acceptance of the LED
luminaires. At the lower light levels, some participants considered the lighting too dark on the
sidewalks. As seen in the sidewalk evaluations, vertical illuminance falls significantly below the
requirements for all of the LED-based conditions, a condition further exacerbated by the impact
of dimming.
Additionally, the lower light trespass from the LED luminaires, while desirable, fails to create
the same bright surroundings found under the wider distribution of the HPS luminaires. This is
valuable information that suggests separating the sidewalk and roadway lighting systems;
however, many designs across the country intend that sidewalk lighting to be covered by the
roadway lighting. The results of this demonstration suggest that separate dimming control, under
which the backlight illuminating the sidewalk is dimmed separately from the roadway, may be
valuable.
As mentioned previously, researchers assigned the participants to specific evaluation groups in
part based on age to ensure the sample adequately encompassed a wide range of ages. Older
individuals (those over sixty-six) often experience yellowing of the lenses of their eyes, which
can make it more difficult for them to see. In fact, the IES 10th Edition Handbook has developed
lighting criteria based upon three age groups for each application: twenty-four years of age and
younger, twenty-five to sixty-five, and sixty-six years of age and older. The criteria (for light
level) increases with the increasing age groups. Not surprisingly, participants sixty-six and older
rated all of the LED test areas low for the survey question “It would be safe to walk on the
sidewalk here at night” with the lights dimmed to twenty-five percent output.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 69
Figure 41. Survey Question 8: “It would be safe to walk on the sidewalk here at night.”
Note: Mean ratings on a scale from 1 (Strongly Disagree) to 5 (Strongly Agree)
The complete list of open-ended participant comments is available in Appendix C: Written
Evaluation Comments.
7.2 User Field Test
The results of the user field test indicate that the 4100K LED luminaire provides the greatest
balance in the visibility of all of the target colors, which indicates that the 4100K may be the best
choice with regard to luminaire CCT. The 4100K and the asymmetric LED luminaires performed
more impressively than the 3500K and 5000K LED luminaires. However, the other light sources
also demonstrated benefits; thus regional preferences may still play a key role in CCT selection.
Careful consideration should be given to the CCT of a given luminaire upon selection.
The asymmetrical design demonstrated a reduction in glare combined to other distribution
luminaires, but no increased performance on visibility; thus its anticipated higher performance
failed to be substantiated, although drivers under this design may be more comfortable.
In the past, standard making bodies have generally determined lighting levels based on
consensus values and some crash analyses. Previous investigations, including those performed
by Clanton & Associates’ research team in San Jose and San Diego, failed to show significant
impacts on the dimming levels of the lighting system for the object detection task. This is likely a
result of the human response to lighting, which usually follows an exponential function called
Stevens’ Law. In general, this law indicates that the response of a human to a physical stimulus
is as follows:
Equation 3. Stevens’ Law
R = k (S-S0)α
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 70
Where R is the response, S is the stimulus, S0 is a base condition to which the stimulus is
compared, and α is the response exponent.
The exponent for light ranges between 0.33 and 0.5, depending on the lighting parameters (flash
versus steady state). Figure42 illustrates a simplified response to a light signal.
Figure 42: Simplified Response to a Signal by a Human
As a lighting system is dimmed, the non-linear response of the human eye has limited impact on
the detection performance. This means that the dimming might remain on the plateau of the
function and that it has not yet reached the knee in the function, where performance significantly
falls off. The researchers in this investigation took an additional step to attempt to affect the
vision system by dimming to twenty-five percent of full light output. The visual detection
performance in the dry pavement condition appears unaffected; however, in the wet pavement
condition, the light level did significantly affect detection at fifty percent and twenty-five percent
of full light output. This result indicates that while streetlight designers and engineers have an
opportunity for dimming under dry conditions, they must exercise caution in wet conditions.
The impact of headlamps constitutes another noteworthy consideration in dimming. As a lighting
system is dimmed, the use of vehicle headlamps may limit the reduction in visibility. While this
is a realistic condition, researchers doubt it affected this investigation as the results indicate a
performance decrement in the wet condition that would likely not be evident had the headlamp
limit been reached.
The three LED luminaires with standard Type II distribution met IES luminance criteria for a
collector road with medium pedestrian conflict using the standard industry practice of lighting
calculation software with light loss factors. Researchers completed the calculations using the
industry standard software AGi32 and found actual measured average luminance values lower
than the simulated values. Such variation is typical and expected given that simulations assume
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 71
ideal conditions, while in reality, pavement type and road conditions play important roles in the
resulting lighting system performance.
The wet pavement condition presents a dynamic environmental change in which the roadway
reflection is reduced and becomes much more specular. These results indicate the importance of
maintaining the lighting level in adverse weather conditions.
7.3 Color Temperature
Industry representatives have long debated the ideal white light color temperature of exterior
luminaires. Some entities claim that exterior luminaires should not exceed the color of the moon,
or 4100K. Some believe that warmer color temperatures are more ideal (3500K or less). Several
factors influence these debates, including color rendition, quantity of wavelengths below 500 nm
(particularly near observatories), and energy efficiency. Many white light LED luminaires
originate from a blue diode; energy is required to make that blue diode into warm white light.
The more efficient white light LED luminaires have cooler color temperatures (5000K and
above).
The user field test demonstrated that the 4100K test areas, including the asymmetric test area,
outperformed all of the other test areas in terms of detection distance. This finding is not
surprising, given the industry-wide recognition of white light multipliers (IES 2012, CIE 2010).
White light sources receive a calculated benefit based upon the luminaire’s scotopic to photopic
(S/P) ratio, then the calculated multiplier increases the effective luminance values. The 4100K
color temperature effectively represents all colors in the spectrum as a neutral color; it is warm
enough that reds are rendered well and cool enough that blues are rendered well.
From the written evaluation, participants showed no preference for any particular color
temperature among the four tested: 2100K, 3500K, 4100K, and 5000K. With dry pavement,
agreement with the survey question “I like the color of the lighting” is statistically on par across
all test areas (see Figure 43). With wet pavement, participants rated the asymmetric test area
lower than all of the other test areas for this same survey question (see Figure 44). Given the
asymmetric luminaire’s color temperature of 4100K, participants may have rated it lower than
the other test areas due to another factor such as contrast or glare.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 72
Figure 43. Survey Question 11: “I like the color of the light Dry Pavement
Figure 44. Survey Question 11: “I like the color of the light” Wet Pavement
7.4 Pavement Conditions
This study used both wet and dry pavement in its evaluations to determine the existence of any
luminaire or light source advantages for one condition over another. The detection distance
results from the wet and dry pavement tests yielded no predictable trend. While detection
distance is somewhat affected by the higher specular reflectance off of the wet roadway, the two
conditions are not significantly different from one another. This study indicates that light levels
should be not be dimmed at all during adverse weather conditions.
0.0
1.0
2.0
3.0
4.0
5.0
LED 3500K LED 4100K 400W HPS LED Asym. LED 5000K 250W HPS
100%
50%
25%
0.0
1.0
2.0
3.0
4.0
5.0
LED 3500K LED 4100K 400W HPS LED Asym. LED 5000K 250W HPS
100%
50%
25%
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 73
7.5 Asymmetric Design
The research team hypothesized a performance advantage for the asymmetric luminaire over the
other standard Type II distribution LED luminaires. With regard to detection distance, the
asymmetric luminaires actually provided the second-highest detection distances, behind the
standard Type II distribution 4100K LED luminaires. The fact that the asymmetric luminaires are
also at 4100K may suggest that CCT is more important than distribution with regard to detection
distance.
The asymmetric luminaires recorded the lowest glare values of all of the test areas, as the light
was intended to be directed away from the driver. While the asymmetric luminaires performed
well in the user field test, participants did not rate the asymmetric test area very high, especially
at the lower light levels. Participants found its distribution patchy and signage difficult to view.
7.6 Control Systems
End users (such as cities, utilities, and departments of transportation) could use a control system
with nodes in every streetlight for reading home energy meters, identifying emergency locations
from 9-1-1 calls, or even for supporting police actions by turning off lights to allow for night
vision goggles. Adding such features to a demand response control system that exists in a
network every 160 to 250 feet across a metropolitan area could fulfill many purposes beyond
dimming for energy savings. Cities should consider such changes and potentially justify them in
a broader context, more like a typical capital expenditure with an energy savings benefit, and less
as a cost-saving project that must justify itself with an acceptable payback. Some non-energy
benefits include lumen depreciation dimming, health and well-being in a darker nighttime
environment, inventory maintenance, and asset management.
Some end users are looking at control systems to keep their lighting levels uniform to reduce
liability over the life of the luminaire. Historically, some end users have purposefully over-
lighted their streets beyond IES recommendations to account for the lumen depreciation. LED
luminaires operated with a control system could maintain lighting levels and their spectral
distribution. Lumen depreciation dimming (not considered in this economic analysis) would
offer additional energy and maintenance savings.
Dimming capability also addresses the potential health and wellness components of exterior
lighting. Much lighting research now explores the impact of exterior lighting on people, plants,
and animals. With design problems such as light trespass, over-lighting, and the new spectrum of
blue light that LED luminaires add to the nighttime world, dimming the lighting can decrease
these potential and currently unknown impacts. For example, controls add the ability to dim or
turn off outdoor beach lighting that draws hatching sea turtles away from the ocean and toward
hotels or housing units. Nighttime neighborhoods with minimal traffic after 9:00 p.m. but large
amounts of light coming through windows and bedrooms would see a marked decrease in the
amount of nighttime light trespass into bedrooms.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 74
Control systems also manage the maintenance of streetlights. With two-way communication, the
control node within the luminaire can send a signal to report a maintenance problem. These
signals can then be aggregated into one email sent daily to the streetlight manager. This feature
reduces maintenance needs by allowing staff to target maintenance efforts to the exact luminaire
that needs to be repaired without sending out crews to identify “day-burners.”
Control systems also provide the valuable benefit of an electronic asset management system..
Traditional street lighting databases are often antiquated and may contain conflicting
information. A control system electronically collects the GPS coordinates of the luminaire along
with its wattage consumption, hours of operation, and other characteristics. The configuration of
the system can permit streets or geographical areas to be grouped together. This comprehensive
database then provides the user with immediate control over a large load. Street lighting could
potentially be used to shed load very quickly.
7.7 Lessons Learned
The research team identified a few lessons learned to apply to similar future studies.
Move the traditional technology sources to the ends of the demonstration site. The layout
of this study located the 400 W HPS test area in the middle of the demonstration site.
With the LED luminaires dimmed to twenty-five percent of full light output, adaptation
between the test areas proved to be difficult.
Allow more time for user field test measurements. The tight coordination of bus and
participant schedules limited time for additional user field test measurements. An extra
thirty to sixty minutes each night for user field testing would have been ideal.
Allow more time for user field tests to increase the sample size and thus decrease the
variance in the results of the study. Keeping written evaluations going on at the same
time as the user field tests requires coordination.
Track duplicate participants through user field test data. If the study necessitates
duplicate participants, analyze their responses separately from the non-duplicate
participants.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 75
8 References
Department of Energy. 2013. “Solid State Lighting Research and Development: Multi-Year
Program Plan.” April 2013.
http://apps1.eere.energy.gov/buildings/publications/pdfs/ssl/ssl_mypp2013_web.pdf
Illuminating Engineering Society. 1999. IES Recommended Practice. 33-99 (RP-33-99) Lighting
for Exterior Environments. Illuminating Engineering Society, NY.
Illuminating Engineering Society. 2000. IES Recommended Practice. 8-00 (RP-8-00) Roadway
Lighting. June 2000, Illuminating Engineering Society, NY.
Illuminating Engineering Society. 2000. IES Technical Memorandum 11-00 (TM-11-11) Light
Trespass: Research Results and Recommendations. December 2000, Illuminating
Engineering Society, NY.
Illuminating Engineering Society. 2012. IES Technical Memorandum12-12 (TM-12-12) Spectral
Effects of Lighting on Visual Performance at Mesopic Light Levels. March 2012,
Illuminating Engineering Society, NY.
Illuminating Engineering Society. 2011. IES Technical Memorandum 15-11 (TM-15-11)
Luminaire Classification System for Outdoor Luminaires. May 2011, Illuminating
Engineering Society, NY.
Illuminating Engineering Society. 2013. IES TM-24-13. “An Optional Method for Adjusting the
Recommended Illuminance for Visually Demanding Tasks Within IES Illuminance
Categories P through Y Based on Light Source Spectrum.” The Illuminating Engineering
Society of North America. New York, NY.
Illuminating Engineering Society. 2012.Technical Memorandum 12 Spectral Effects of Lighting
on Visual Performance at Mesopic Light Levels.
International Commission on Illumination. CIE 191 Recommended System for Mesopic
Photometry Based on Visual Performance. 2010.
Smalley, Edward. “Transformations in Lighting.” 2012 DOE Solid State Lighting Research and
Development Workshop. 2012.
The Illuminating Engineering Institute of Japan, The Influence of Dimming in Road Lighting on
the Visibility of Drivers, Volume 29 Number 1, April, 2005.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 76
Appendix A: Prior Work
A significant number of studies and pilot projects have preceded this work; a partial list follows.
NEEA Study: LED Lighting Technologies and Potential for Near-Term Applications.
This assessment evaluated LED technology and its likely uses in lighting applications.
NEEA Study: Technology and Market Assessment of Networked Outdoor Lighting
Controls. This study evaluated and compared the many manufacturers of outdoor lighting
controls available for street lighting.
Seattle Belltown: For this installation, the City of Seattle will respond to a high crime
area by dimming the lights to seventy percent of full output for general nighttime use. At
1:00 a.m. the lighting will be raised to full output in an effort to increase a sense of
security.
Seattle Residential: The city is halfway through an LED conversion of 41,000 residential
streetlights.
Anchorage: The consulting team for this project performed the first of these studies in
Anchorage, Alaska. With high electricity prices, Anchorage sought to reduce luminaire
wattage and power consumption. Written evaluation test results in both residential and
commercial areas and user field test results in commercial areas provided enough
confidence in LED lighting from HPS to move to a full change-out, city-wide.
San Diego: After a similar test in San Diego, the city adopted induction technology
(another white light source) for its standard luminaire. Its decision was based on better
visibility, a more maintenance-friendly light source compared to low pressure sodium
LPS and high pressure sodium HPS, lower energy use and accepted light source
spectral distribution range for the observatory at Mount Palomar.
San Jose: In San Jose, the team expanded the test to include the dimming capabilities of
LEDs and controls. After showing comparative visibility and community acceptance of
LED sources and reduced light levels, the City developed an Adaptive Lighting Guide
and Luminaire Replacement Guide for use in both retrofits and new installations of street
lighting systems. These documents provide guidance for dimming streetlights to a lower
level at night, when traffic and pedestrian conditions have changed significantly from the
high levels during rush hour, and for replacing existing LPS lighting with new LED
luminaires.
The DOE gateway projects demonstrated LED street lighting in New York, Portland,
Sacramento, Palo Alto, Minneapolis, San Francisco, and Oakland. Most of these
monitored energy and light output performance compared to existing HPS lighting.
California Lighting Technology Center: The CLTC has conducted research and
demonstration projects that evaluate LED and induction white light sources with
networked controls and bi-level dimming. Its research also includes an in-depth study of
California’s existing streetlight infrastructure.
Los Angeles: A city-wide conversion program had replaced nearly 77,000 luminaires
with LEDs as of May 2012.
BC Hydro: Numerous Canadian cities have partnered with the utility for LED and
adaptive control demonstrations and pilot projects including Vancouver, Port Coquitlam,
and Prince George.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 77
BC Hydro: Adaptive Lighting Feasibility Studies performed for cities within British
Columbia found average dimming ranges for each type of roadway (local, collector, and
arterial).
BC Hydro Power Smart Program: The Adaptive Lighting Guide outlines a process for
municipalities to develop adaptive lighting standards based on IES criteria for pedestrian
and vehicular conflict.
A 2005 study by the Illuminating Engineering Institute of Japan found very little effect on
driver visibility with changes in light levels. (Illuminating Engineering Institute of Japan
2005).
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 78
Appendix B: Written Evaluation Form
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 79
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 80
Appendix C: Written Evaluation Comments
Dry Pavement 100% Light Level, Test Area 1
This is OK. Does the job. Discovered one patch of dark sidewalk though.
One streetlight is out on the opposite of the street.
It is a yellow, but not too yellow. Not terrible.
When trees have leaves, the light will be different. Light out at NW of 77th and 15th.
As with some of the other test areas, I like this lighting a lot. Without referencing my
other notes, this seems like the best combo of all.
Dark Alleys
I like this light. It is not patchy and provides sufficient light. It is warm, but bright at the
same time. Lots of shadows though.
Sidewalks are kind of dark. Road okay but could be a little brighter.
More trees.
One lamp isn’t working. The lighting isn’t overpowering. I like the color of the lights.
Lots of shadows on sidewalks.
Add light pollution covers.
My fave so far.
The lights are very bright and glarey, but not enough light reaches the street level. All 4
sets of bright glare lights would be distracting and annoying when driving at night.
Budget Rent A Car has lot spot lights that effected the overall street lighting in a negative
way to bright and too much glare.
Appears lighter, but not too bright (although I like it bright). Sidewalk are better lit, more
shops also have contributed to light source.
Good light on street, sidewalk was a bit dim.
There are some areas I would worry about due to darkness, but where the light covers it is
nice.
Too much time was allotted for the walk. We could have easily done it in half the time
(and not gotten so cold).
My street is residential and I would prefer less of the lights and less brightness. A little
unsafe because of neighborhood, not lights.
I like the white color of the lights vs. the standard orange tinted lights that are common.
Lights could be brighter. Not enough for sidewalk.
The moon is very full tonight.
The nearly full moon should be accounted for. It is brighter tonight than usual for Seattle.
Hard to judge because of the full moon.
Glare when looking up at light.
Light is just a little dim.
Sidewalk park strip tree here.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 81
Dry Pavement 100% Light Level, Test Area 2
Too dim on sidewalks.
Sidewalks are darker.
Cozy. Like being in the moonlight.
Seems like sidewalks has better light on the West sidewalk.
Seems patchy in some places, but not as bad as patchiness in test area south of here (3). I
like the non-glaring “safe softness” of the light.
The lighting is better than some of the other white lights but still patchy.
Good for a side street, but light on road seems dim for Main Street. Sidewalk patchy and
uneven. Road also patchy.
Trees make light patchy on sidewalk.
Lights do not light up sidewalk completely.
This section has better lighting than the other of the same type of lighting source. I like
the softness of the source. They really penetrate into the side yards well.
Add light pollution covers.
Sidewalk area okay.
Again these are too bright and glaring at the light.
It really is dark and does not light up sidewalk or parking lots or shops. It was an obvious
change from previous street.
Darker, less even light.
Not too dim and not too glaring, a little patchy.
Some dark areas but gray, not black. Feels easier to see in shadow then I used to.
Lighting feels softer here maybe because there aren’t as many business signs/lights.
The lighting would make it easy to see bicyclists/pedestrians while driving.
Lights too bright to look at directly, but impact on street lighting is great everything is
illuminated.
Much brighter and cleaner than area 1. Does sort of contrast with house lights looks
weird, but I like the brightness.
Looking directly at the lights is too bright for my eyes. But otherwise, they are nice and
bright.
Better than area 1.
I haven’t really paid attention to other streets lighting.
This amplitude of light is very strong. Even blinds would not stop a bedroom from being
in perpetual daylight.
Sidewalk feels safer because of porch lights.
Glaring when looking up toward/at light.
Good depth of coverage.
Nice solid street lighting.
Brightness of lighting varies depending upon distance and location between streetlights.
Dry Pavement 100% Light Level, Test Area 3
It is nice and bright. Good visibility on sidewalks and streets.
Sidewalk and environs are well lit.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 82
Great for driving in. Would not want to live on a street with these lights.
Just too much light. Plus it is yellow.
Lighting is yellow. Really bright against the apartments. Bright on sidewalk.
Seems bright, yet not glaring I like it!
Alleys and shrubs are lit.
The lights were pleasant and gave enough light. I would feel safe driving and walking at
night.
A little orangey, but good coverage on both street and sidewalk.
Aside from the lights leaving spots in your eyes when you stare too long, it is probably
the best lit block so far. Less shadows on the sidewalk.
Add light pollution covers.
Very even lighting with few shadows.
Very few shadows, especially at edges of sidewalks. Fully lighted building facades at
street side.
The lighting improves visibility tremendously; however, it is too bright. It seems that the
LED lights provide much more comfortable lighting than the current streetlights.
Too bright.
A bit too bright, but way better than 4 or 5.
Too orange.
Sidewalks are better illuminated. I noticed more people talked too.
That tree looks awful.
Trees look awful.
Good light on sidewalk, still a yellow light but not as bad.
The brightness hurt my eyes a bit.
Way too bright for residential. Seems overly bright for commercial too.
Glaring! Have to squint. Very patchy. Hate it.
Seems like the lighting is more even than the other test areas without areas of darkness.
I definitely prefer these lights. Well lit and the lighting is warm. The sidewalk was very
well lit.
Too much light for people in houses along street.
Nice bright glowy yellow light.
The yellow light creates less lighting pollution in the sky.
And by better, I mean brighter to a slight degree. I consider myself as a traditionalist, so
favor the amber orange hue.
Seems brighter than previous 2 sections.
Brighter of all so far.
I like the yellow tone vs. the blue.
Old lights along this area.
Looking right at lights is difficult. Light on street and sidewalk seems to have more
coverage on street and sidewalk than test area 1 and 2.
Dry Pavement 100% Light Level, Test Area 4
Sidewalk has a lot of uneven places and settling and the light does [not] illuminate it at
all in many places. Bad for pedestrians. Unsafe.
Visual Quality, Acuity, Community Acceptance - LED Streetlight Sources
Clanton & Associates - 83
Insurance agency really lights the sidewalk. Trees block light on uneven sidewalk
surface. In places sidewalk is very dark.
Low glare at light source is nice.
REALLY dark on sidewalks. Dark spot at 7040, 7037, 7052
Thinking about bike riders here, patchy lighting makes navigation much tougher!
Retailers sign is brighter than street
I did not feel safe on this road, especially if I was alone. Too many shadows. It seems if I
was driving, I would have a hard time reading road signs.
Too dark on sidewalk. Patchy on street, style of light is a lit distracting
Significant dark areas before lights.
I feel like the color is nice. But it's not intense/bright enough for me to see what I am
writing.
The LED lights here are good for driving, but visibility on sidewalk is not good.
Large areas of shadow on sidewalks and between some lights. Hard to see to avoid
uneven sidewalk in places. Can't read this form well in places.
Darken edges/shadows-borderline almost too dark. Building facades are not illuminated
very well. In the rain, this would be too dark. Sidewalk area too dark but this block has
more residential - no much light from businesses.
A little dimmer.
Light is similar to Area 5, a bit more uneven, but way too much glare.
Patchiness on sidewalks unacceptable.
Can’t see people are tripping.
Patchier, less light than 5.
The owls scared me.
Much too dim, not light reaches the sidewalk.
Something about the light makes my eyes hurt and there are too many dark areas.
This section is a little patchy, which makes the lighting noticeably worse. Compare to the
really bright. section, a burnt out bulb here would be very noticeable. The road is bright,
sidewalk is dark.
Too little lighting on much of the sidewalk.
Horrid. Can’t see a thing. Had to stand under a light to fill this out.
The houses and sidewalks along the street are too dark. There are a lot of patchy areas.
Absolutely not enough light. Tripped on the sidewalk because of the lack of lighting
Patchier.
Not nearly enough dispersion, the patches are quite stark.
I wouldn't want my daughter driving here at night.
Style would be okay if some consistent light.