Leds) to Traffic Control Signals

NCHRP National Cooperative Highway Research Pro~ram RESEARCH REsULTS DIGEST April 1995 Number 205

These Digests are issued in the interest of providing an early awareness of the research resulis emanating from projects in the NCHRP. By making these results known as they are developed, it is hoped tha1 the potential user.; of the research findings will be encouraged loward their early implementation in operating practices. Persons wanting co pursue the project subject maner in grealcr depth may do so through contact with the Cooperative Research ProgrJms Scaf!, Transponacion Research Board, 2101 Constirution Al'e., N.W, Washingion, D.C, 20418.

Subject Area: IV A Highway Operations, Capacity, and Responsible Senior Program Officer: B. Ray Derr Traffic Control

Requirements for Application of Light Emitting Diodes (LEDs) to Traffic Control Signals

"-..,.1· ' )

An NCHRP digest of the findings from the final report on NCHRP Project 5-12, "Requirements for Application of Light Emitting Diodes (LEDs) to Traffic Control Signals," conducted by Lighting Sciences, Inc. Mr. John Corbin, Mr. Michael Janoff. and Dr. Ian Lewin served as Principal Investigators.

INTRODUCTION LED technology is rapidly evolving, and the research project evaluated the technology available Incandescent lamps are presently the major at the time. Some of the findings and recommend illumination source for traffic signals. The incan ations will undoubtedly become obsolete as the descent lamps used for traffic signals are, however, technology progresses. inefficient compared with other light sources. Maintenance costs are high for these lamps, and the light output degrades as the lamp ages. These ADHERENCE TO COLOR STANDARDS problems-plus the ever-increasing cost of energy justify considering other light sources. One The Institute of Transportation Engineers (ITE) alternative is light emitting diodes (LEDs). standard, "Vehicle Traffic Control Signal Heads," is Before existing incandescent lamps are replaced commonly used throughout the United States. This by LEDs, it is necessary to show that LED signals standard defines the required color for new traffic will meet applicable standards for color and signals. intensity, will not adversely affect the safety or An LED's color is usually referred to by the operation of the roadway, and will be economically peak wavelength of the spectral distribution. This advantageous. NCHRP Project 5-12 was initiated in can be misleading. The other colors in the response to these needs. The objective of the distribution cause the dominant wavelength (the research was to determine the feasibility and apparent color) to differ from the peak wavelength. implementation potential of LEDs. The project For this reason, LEDs should be plotted on a found that red and Portland Orange (pedestrian) chromaticity diagram (Figure 1) to verify whether LED signals are currently feasible. they meet the color requirements. This digest summarizes the project findings and Red. At this time, the most common red LEDs includes testing results, a discussion of economic consist of Gallium-Aluminum-Arsenide (known as analysis, specification guidelines, recommendations GaAlAs or AlGaAs) and emit a 660-nm peak for LED use, and suggested future research. The wavelength. These fall around 640-nm dominant digest will be of interest to those involved in wavelength on the chromaticity diagram. Generally designing and specifying traffic signals. all of the 660-nm LEDs tested will fall within the

TRANSPORTATION Rli5EARCH BoARD NA"nONAL RF.SEARCH COUNCIL 6

-..

0-+-----1------1------t----;------t------t -30 -20 -10 0 10 20 30 Horizontal Angle

Type of Signal Head • ITE Req. u 30° LEDs + 8° LEDs * 8° LEDs w/ Prism. Lens

Figure 4. -12.5° vertical viewing angle.

140 --·--· ------·

120 ------1---

'ii, lil 100 ----- ···-··- "U !a -9. >, 1i m -C

04------+-----+-----+------,,-----+----I -30 -20 -10 0 10 20 30 Horizontal Angle

Type of Signal Head • ITE Req. a 30° LEDs + 8° LEDs * 8° LEDs w/ Prism. Lens

Figure 5. -17.5° vertical viewing angle. 7

• 30/30 1.00 - Ill D 40/20

t Do •• • 0 D • 0.85 0 ~ D ~-. •••• •• • D •••• ~ 0.80 - C/1 " Do •••••••••• • C •0 D • GI 0 • ... an • • • •i ~ .5 0.75 • GI -o - c.t. Dao ~ ooDoooooo ai C. n D 0::: 0.70 u ... o t•~t.o - - 0 a ·~·

0.55

I I I I I I I 0.50 I I I I 'I I I I I 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 minutes

Figure 6. Output vs. time for 12" red signal-three duty cycles. 8

5. The use of visors on LED signals normally an orange pedestrian-hand symbol. Typical costs for decreases their visibility while the opposite is true some types of heads are 12" red ball-$230; 8" red for incandescent signals. ball-$190; 12" red arrow-$115; and orange 6. Visibility and conspicuity are directly related pedestrian-hand symbol-$250. On the other hand, to the signal intensity.. The required intensity for a an incandescent krypton bulb costs about $2. As a red ball LED signal seems to be in the range of 300 market is established for LED signals, the initial cost to 500 candelas. Red arrows appear to require about will undoubtedly decrease. 44 additional candelas of intensity. Yell ow ball LED signals probably require about twice the Power Savings. LED signals consume much .. intensity as a red ball LED signal and green ball less power than incandescent signals. The reduced LED signals require about 2 1h times the red ball cost of power can be estimated using the following signal intensity. equation: 7. Improving the light distribution of an LED signal can significantly reduce the intensity A = 8760*D*(PrPL)*C, where requirements. The intensity required for red ball A is the annual power cost savings ($/yr), LED signals can be reduced by about 26 percent to 8760 is the number of hours in a year (hr/yr), 52 percent if the distribution is doubled (from a D is the duty cycle (percent time on),

viewing angle of 7° to 15°). Yellow and green LED P1 is the incandescent power consumed (kW), signals do not show as dramatic an improvement. PL is the LED power consumed (kW), and 8. The results for red signals are believed to be C is the cost per kilowatt-hour ($/kWhr). much more accurate and valid than for the other two colors. Similarly, the results for ball signals are Some typical examples of annual power savings are believed to be more accurate and more valid than for shown in Table 1. arrow signals. This is partly attributed to the larger amount of data available for the red ball signals. Life. LED signals will last longer than 9. Color-vision deficient drivers and normal incandescent signals though good estimates of vision drivers seem to have similar error patterns expected life are not yet available. Incandescent and reaction times, but the small number of color bulbs are replaced every 6 months to 1 year. The vision deficient drivers makes generalizations life of the LED signal is a critical factor because somewhat tenuous. calculations using a 5-year life can justify LED 10. Performance at night appears to be similar to signals not justified using a 3-year life. daytime performance. Overall, nighttime If an entire intersection could be converted to performance was better than daytime performance. LED signals, there would be significant savings 11. The red ball, red arrow, and Portland because semiannual relampings would not need to be Orange pedestrian signals appear to perform as well done. This would reduce the amount of preventive as incandescent signals, both during the day and at maintenance required and the number of trips. The night. lack of a satisfactory green LED makes this unlikely in the near future.

Analysis. One method of analysis is to ECONOMIC ANALYSIS determine the payback period. This is the amount of time over which the savings in operating costs will Justification for use of LED signals must depend equal the increased initial cost. Figure 7 shows a on an economic analysis that demonstrates a cost payback-period analysis for the four most promising savings. This analysis should consider the following LED applications. factors. While the payback-period analysis method is useful, it fails to consider the time value of money. Initial Cost. LED signals are very expensive Another approach is to convert the annual cost relative to incandescent signals. The cost of a single savings over the life of the signal into present day lens can range from $115 for an arrow to $430 for dollars and then compare that to the initial cost. 9

TABLE 1 Annual power cost savings of LED signals

Duty Cycle 12" Red Ball 8" Red Ball 12" Red Arrow Orange Ped Signal

50% $54.3 $24.8 $57.5 $42.8

60% $65.1 $29.8 $69.0 $51.3

70% $76.0 $34.8 $80.5 $59.9

80% $86.8 $39.7 $92.0 $68.4

LED Power 17 W 15 W lOW 23 W Consumption

Incandescent Power 135 W 69W 135 W 116 W Consumption

Power is assumed to cost $0.105/kWhr.

en -

0.45 0.5 0.6 0.65 0.75 0.8 0.85 Duty Cycle

Type of Signal Head • 12" red ball o 8" red ball + 12" red arrow * orange ped. signal

Figure 7. Payback period for LED traffic signal heads. 10

SPECIFICATION GUIDELINES specified (e.g., "signal output at nominal voltage and room temperature conditions shall not fall below If a jurisdiction chooses to use LED signals, it 80 % of intensity requirements before 1000 hours of should consider the following points in developing field operation"). It is reasonable to define signal specifications. end-of-life as the point when the signal reaches some percentage of the stated intensity requirements. 1. Minimum intensity requirements over a range of horizontal and vertical angles should be specified. Stating a particular quantity or type of LED should RECOMMENDATIONS be avoided. The specification should clearly state the conditions under which the signal must meet the Based on the research performed, the following requirements (e.g., "initial intensity of new signal at recommendations are made. room temperature conditions (24 °C) "). Additionally, instructions should be included regarding how the 1. Ball LED signals can be used under the same signal is to be aligned and whether any multiplying conditions as incandescent signals. The intensity of factor will be used to increase signal output tested at the red LED signal should be in the range of 300 to steady state operating conditions to initial conditions. 500 candelas with a narrow distribution (7 ° viewing 2. The minimum level of performance over the angle). If a wider distribution is used, then intensity operating voltage range should be defined (e.g., can be reduced significantly. Color can be either "new signal output at room temperature conditions 660 or 630 nm, but the 630-nm signal is shall not fall below 85 % of intensity requirements recommended because its color is more similar to over the operating voltage range"). To require a the incandescent signal and will pose fewer detection signal to meet 100 percent of the intensity problems for many color vision deficient drivers. requirements over the operating voltage range is 2. Red arrow LED signals can be used under the unrealistic. The amount of loss allowed may (within same conditions as incandescent signals. The reason) be based on what the signal manufacturers intensity of the red arrow LED should be about 40 can achieve with regulatory circuitry. to 50 candelas more than the red ball LED. 3. The minimum level of performance over the 3. Portland Orange pedestrian LED signals can operating temperature range should be defined (e.g., be used under the same conditions as incandescent "new signal output at nominal voltage conditions signals. The intensity of the pedestrian LED signal shall not fall below 70 % of intensity requirements should be at least 20 to 25 candelas. over the operating temperature range"). Also, it 4. Present yellow ball LED signals could should be stated whether the operating temperature probably be used if the intensity of the signals were refers to ambient, printed circuit board, or perhaps, increased. It is difficult to justify their use, internal signal head temperature. It is unrealistic to however, because of the short duty cycle. Yellow require a signal's intensity to remain unchanged at arrow LED signals did not perform well and are not above room temperature. The amount of loss recommended. allowed may be based on environmental testing of a 5. Present green LED signals should not be used variety of discrete LEDs that may be used in signal because of their poor color and very low intensity. applications, or a number of available signals could 6. Present red LED lane control signals were be tested. Instructions on how a signal's found to perform well in visibility but poorly in temperature response will be verified may also be conspicuity. They are not recommended for use. included in the specification. 4. A maximum signal output loss (as percentage of intensity requirements) resulting from a single SUGGESTED RESEARCH LED failure should be specified. Stating a specific number of LED strings should be avoided. Large-scale field testing . in a variety of 5. When some level of output degradation over environments needs to be performed. Red signals of time is specified, the conditions under which the different types using different LEDs and different amount of degradation would be measured should be duty cycles should be included in a multistate 11 cooperative study. Long-term testing should be done period. Intersections in more complex visual for at least 3 years. Signal output as a function of environments with higher approach speeds and time, environment, and duty cycle is needed for the higher traffic densities should also be evaluated. different LEDs. A larger installation base of LED signals will Research into possible methods to reduce the allow a better analysis of the safety impacts of LED temperature sensitivity of LED traffic signals is signals. needed. The solutions should not significantly As new types of LEDs become available, they increase the power consumption or total signal price. must be evaluated in comparison to incandescent Complex retrofit work on the signal heads should be signals. avoided. The effect of stabilizing the temperature of the LEDs on the usable signal life should also be investigated. AVAILABILITY OF REPORTS The LED manufacturers should investigate the development of an asymmetric LED package, which The final report for this project will not be would direct most of the LED output to the areas published but is available for purchase on microfiche defined by the traffic signal intensity standards. A from the Transportation Research Board (202/334- more efficient use of the light would result in a 3214). lower signal cost with reduced power consumption. The human factors and safety performance experiments, in particular the field test, only REFERENCES addressed a limited number of types of intersections and only for a very brief time period. A fuller 1. ITE, "Vehicle Traffic Control Signal Heads," evaluation would necessarily include a broader range Publication Number ST-008B, Institute of of types of intersections over a much longer time Transportation Engineers (1985).

GLOSSARY OF TERMS chromaticity diagram: This two dimensional diagram allows for quantifying colors without using the spectral output data. Colors are located on the chart by using their associated x and y values. A z value is also used such that the sum of x, y, and z equals one.

AlGaAs (or GaAIAs): A family of LEDs that contains the materials Aluminum, Gallium, and Arsenide in the structure of the chip. These LEDs emit red light usually at 660-nm peak wavelength.

AllnGaP (or InGaAIP): A family of LEDs that contains the materials Aluminum, Indium, Gallium, and Phosphor in the structure of the chip. These LEDs are produced to emit light in the range of 590-nm to 630-nm peak wavelength (depending on the relative amounts of the constituent materials). candela (cd): Unit of luminous intensity describing how much light is going in a particular direction. lumens (Im): Unit of luminous flux describing how much total light is emitted from a source. nanometer (nm): Common unit of measure of light wavelength. In the case of LEDs, often used to describe the color of light emitted.

"string" of LEDs: A number of LEDs that are placed in series on a printed circuit board are referred to as a string. The string is usually designed such that the total voltage drop across the string is close (or equal) to the input voltage. If one LED in a string fails, the entire string will fail-similar to most Christmas tree lights. The printed circuit board will include more than one string. Failure of one string will not affect the other strings.