Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
9 This chapter reviews the relevant research applicable to SSL and its application to roadway lighting and provides a gap analysis detailing available information, missing information, and strategies for addressing those gaps in this and future research. Differences Between Solid-State Lighting and High-Intensity Discharge Fixtures Until recently, the majority of roadway lighting was HID, specifically, HPS. There are many differences between SSL and HID beyond the physical equipment. Major differences include (1) light distribution, (2) lighting output control (ability to dim the light), (3) spectral power distribution (SPD), and (4) efficiency of lighting production suitable for human perception (luminous efficacy). Research comparing SSL with HID for roadway lighting has identified some benefits of SSL in terms of energy consumption, luminous efficacy, color rendering, and adaptability. SSL uses semiconductors such as LEDs as a source of illumination, rather than filaments or gas plasmas. In the last decade or so, white SSL lighting has exceeded the efficacy of other lighting because less energy is lost in the form of heat. LEDs operate by applying voltage to the leads of a diode, which causes electrons to accelerate. These electrons move through the diode material and recombine with electron holes within the device, which causes a deceleration and in turn releases energy in the form of photons. In broad-spectrum white SSL, the emitted photons create short wavelength light (blueâtypically about 450â480 nm), which is phosphor converted to other colors or shades, the result of which is a white-appearing light. This white SSL process occurs at a lower temperature than that of other light-emitting devices, which allows for a longer life for the light source as compared with gas or filament lights and lower costs for dimming and color. LEDs also emit lighting from a very small area on the diode in a specific direction (as opposed to omnidirectional lighting), which results in more flexibility in terms of aim. HIDs and LEDs differ greatly in terms of luminous efficacy, or how well a light source produces visible light as a ratio of luminous flux to power. A metal halide HID lamp ranges from 65 to 116 lumens per watt, which equates to an efficacy of between 9% and 17% overall. HPS lamps are slightly better, with a range of 85 to 150 lumens per watt, or an efficacy of 12% to 22% (Rodrigues et al. 2011; Stouch Lighting n.d.), and are generally regarded as being highly efficient as compared with fluorescent lights, arc lamps, and incandescent lights, which has led to the prevalence of HPS lamps in existing roadway lighting. Modern LED lamps, however, have been found to produce from 37 to as high as 303 lumens per watt, or a range of 0.66% to 43% (Cree Inc. 2014; Stouch Lighting n.d.). Luminous efficacy can also be weighted by photopic and scotopic response curves because the eye is more sensitive to certain wavelengths, depending on light level (Rodrigues et al. 2011). C H A P T E R 4 Literature Review and Gap Analysis
10 Solid-State Roadway Lighting Design LEDs can be customized for their intended use. They can be acutely dimmed to specific levels (Dyble et al. 2005; Li et al. 2009; Gil-de-Castro et al. 2013; Jin et al. 2015), turned on and off rapidly without a need for a warm-up period (Gaston et al. 2012; Wang and Liu 2007), and fine-tuned for color output during the manufacturing process to achieve a range of CCTs (Liu and Luo 2011). Precise cutoffs can be implemented to increase the control of the lightâs focus (Timinger and Ries 2008). In general, LEDs are better in terms of photometric and economic performance as com- pared with HPS light sources. As of 2012, many organizations focused on roadway lighting have directed their focus to developing guidelines for LED device performance and predictions. These organizations include IES, CIE, Industrial Lighting Products, the U.S. Department of Energy, the American National Standards Institute (ANSI), and the National Electrical Manufacturers Association. LEDs are predicted to offer from 50,000 to 100,000 hours of use before substantial degradation due to overheating or overuse (Evans 1997), although the accuracy of these claims is still undetermined. In terms of life expectancy, LEDs far outperform HIDs (Josefowicz 2012). The spectral outputs of HIDs and LEDs differ greatly in terms of wavelength content but are often identified by the CCT measured in kelvins (K). The CCT values are related in appear- ance to the absolute temperature of a blackbody radiator (incandescent). Figure 2 shows the SPDs for HID and LED luminaires for a variety of CCTs: an HPS luminaire with a 2100K CCT and 3500K and 6000K LED light sources. It is important to note that the CCT of light sources cannot be compared across different technologies; specifically, the CCT of an HPS lamp cannot be compared with that of an LED lamp. The critical aspect of the two methods of generating light is that the phosphor-converted LED generally has a broad-spectrum output, and the HPS relies on the energy band conversion of electrons and has significant spectral gaps in light output. Moreover, the SPD of LEDs can be tailored to achieve specific CCTs. One of the difficulties with the CCT is that it has become (wrongly) a de facto industry standard to describe the light output of the LED. This topic is discussed later in this report. The color-rendering capabilities and luminous efficacy of LEDs give them an edge in visibility performance. The cost of installation is the biggest hurdle for implementation. Because most of the detections in a roadway are foveal (Gibbons et al. 2012) and based on the cone detectors only, particularly at high speeds, the ability of an LED light source to render colors provides color contrast. This additional color contrast has been shown to provide benefits in detection (Terry 2011). Because LEDs render color so well, white lightâemitting LEDs have been consid- ered as direct replacements for HPS lighting on streets and roadways. Because nighttime driving Source: Gibbons et al. (2015b). Figure 2. Spectral power distribution of overhead lighting types.
Literature Review and Gap Analysis 11 inherently involves low-light vision, luminance contrast typically prevails; however, color con- trast can vastly improve an objectâs visibility (Lutkevich et al. 2012), especially when a source such as an LED has strong color-rendering capabilities closer to that of natural light (Gibbons et al. 2015a). Figure 3 shows some of the variations that can occur in detection distances for various types of CCT sources (the x-axis shows source typeâthose shown in wattage are HPS and those shown by CCT in kelvins are LEDâand system output of 25%, 50%, or 100%). An asymmetrical distribution (ASYM) was also assessed. LEDs have some perceived drawbacks. The advent of LED lighting invites the use of more white or blue CCT lighting on roadways because the color-rendering abilities at these levels are much greater than the yellow often associated with HPS and low-pressure sodium (LPS) light sources. However, some research shows that driver vision may be negatively affected when the eye adapts as it moves from darkness into an area lit by a white or blue light source (Boyce 2009; Goldstein 2010). Other concerns include performance in adverse weather conditions, sky glow, and health; these topics are discussed later in this report. From a lighting design and specification standpoint, SSL requires a much different approach than HID technology. The life of the source is rated differently. HID sources are typically rated by the time in operation when 50% of the lamps failed; LED sources are rated according to the point at which the lumen depreciation reaches 70% of the initial output rating. Some of these factors are quantified and included in LED test data acquired in accordance with Projecting Long Term Lumen Maintenance of LED Light Sources (IES TM-21-11) (IES 2011). Other data are usu- ally available from luminaire manufacturers. Figure 4 shows an example of test data used for predicting the life of an LED luminaire. The testing is done for three different temperatures, and a calculated life is estimated and a rated life determined (which can be no more than six times the test duration). This testing, however, only predicts the life of the LED; the lifetime of the driver is related to the case temperature. Figure 5 shows an example of driver lifetime as it relates to case temperature. Because the driver is an elec- tronic device, it can be evaluated by testing of mean time between failures (MTBF) to determine reliability and product failure rates. Established methods of classifying photometric distribution (IES Types II, III, IV, V) are not as relevant with LED luminaires. Distribution classifications that were often used for grouping Source: Clanton and Associates Inc. (2014). M ea n De te cti on D is ta nc e (ft ) Figure 3. Mean detection distance differences for various light sources.
Source: IES (2011). Figure 4. TM-21 inputs for predicting luminance maintenance.
Literature Review and Gap Analysis 13 Note: Tcase = temperature of the case of the driver. Figure 5. Example of driver life determined by MTBF analysis. Note: FC = footcandles; Typ = typical. Figure 6. Comparison of HPS and LED luminaires according to standardized distribution ratings. similar HID products from the standpoint of optical distribution often do not work well as descriptors for LED products. Figure 6 shows an example of distributions from an HPS lumi- naire and an LED luminaire, both classified as Type II. There is also great variation between available products, and subjective acceptance is different considering the overall change in color and brightness, a wide variety of control options, and different failure modes resulting from temperature and voltage variations. With HID streetlights, defining the wattage and optical distribution (e.g., Type II, cutoff) would produce similar results from different luminaires. LED luminaires, however, require closer review and assessment. Differences Between ssL anD HiD fixtures What We Know ⢠Photometric and output control for SSL is much different from that for HID. ⢠SSL designs often result in less overlighting of the roadway and adjacent areas. ⢠The spectral content of commonly used LED sources may result in different detection distances. ⢠The perceived brightness of LED systems has been reported to be higher than that of HPS sources. (continued on next page)
14 Solid-State Roadway Lighting Design Differences Between ssL anD HiD fixtures (continued) ⢠Commonly used LED sources may result in varied glare responses. ⢠Life, depreciation, and performance vary for LED luminaires. Typical HID or âequalâ options are not as readily available with LED. ⢠Maintenance considerations are quite different for LED luminaires. What We Donât Know ⢠If commonly used, the differences in safety impacts of LED and HID sources. ⢠The true expected life of LED systems with electrical and climate considerations taken into account. ⢠The safety impacts of greater uniformity on roadways, which can often be achieved with LED installations. Results/Recommendations of This Study ⢠Spectral impacts were analyzed, and results are presented in Chapter 5. ⢠Further crash analysis should be performed as more systems are put in place and recommended for future research. ⢠Maintenance considerations are known and included in the SSL Guide. Off-Roadway Lighting In scotopic conditions where adaptation luminance is 0.001 cd/m2 or lower, the choice of a light source based on color is not important because the use of rods in the periphery is more prevalent. In a mesopic scenario, spectral effects are slight. However, in scotopic conditions, where adaptation luminance is very low (0.001 cd/m2 or lower), peripheral vision and off-axis spectral effects become primary (Gibbons et al. 2015b). Research suggests that on lighted road- ways, drivers focus on the lighted portion of the road (Mortimer and Jorgenson 1974), and that the adaptation luminance of the eye remains somewhat steady. However, on rural roads that may not be straight or conventional, drivers tend to vary the direction of their gaze, which causes multiple spatial and temporal levels of eye adaptation. In these conditions, lighting off-axis may benefit rod photoreceptors, which are active in the periphery and operate mainly in low-light conditions (Boyce 2009). Off-road lighting (the area adjacent to the travel lanes) has been a criterion in Europe for several years. In CIE 115, Lighting of Roads for Motor and Pedestrian Traffic, the metric of the sur- round ratio (SR) is used to âensure that light directed on the surrounds is sufficient for objects to be revealedâ (CIE 2010a, p. 18), as shown in Table 1. Essentially, this criterion is used for an illuminated roadway in darker surroundings to help the visibility of pedestrians, cyclists, and hazards. Evidence suggests that increased lighting in areas immediately adjacent to the roadway may help to increase visibility and associated safety. This lighting would not be extended to areas beyond the roadway right-of-way, which needs to be limited, but addresses the area between the edge of the right-of-way and the edge of the travel lane. The approach for determining the visibility of off-road or off-axis objects stems from a paper that shows visibility levels (VL) at different angles from the driver (Gibbons 1993). For a distance of 60 m at an angle of 7°, a VL of 3 is needed. The wider the angle of view, the higher the VL that is needed. The higher the calculated VL of a target, the more visible it is. The VL model is generally used in static observations to determine the ability of a light source to provide sufficient visibility.
Literature Review and Gap Analysis 15 In general, VL values are only comparable to other VLs calculated in the same scenario. No scales display VL benchmarks for visibility; therefore, VL has generalizable limitations. Surface luminance includes the amount of light hitting the surface, the surface reflectance, and the color under the artificial illumination. Adrian and Gibbons (1999) also suggest a procedure for designing a roadway illumination system that considers off-roadway reflectance. This inves- tigation shows the complexity of visibility-based lighting. Off-rOaDway LigHting What We Know ⢠LED systems can often have less off-road spill lighting than HID systems. ⢠On higher-speed roadways, drivers tend to focus on the road. On lower-speed roadways, drivers tend to vary their gaze direction. ⢠Off-road lighting requirements have been a CIE criteria for many years. ⢠Off-axis visibility is different from on-axis visibility. Spectral content is a variable. ⢠Off-road lighting requirements need to be balanced with light trespass limits. What We Donât Know ⢠How critical off-roadway lighting is. ⢠How much area adjacent to the roadway it is necessary to light. ⢠How off-roadway lighting relates to safety and crashes. Results/Recommendations of This Study ⢠Off-roadway lighting was tested, and the results are included in Chapter 5 of this report. Metrics for Lighting Several important metrics define the performance of a light source. Lumens measure the luminous flux of light or the quantity of visible light emitted by a source. The SPD renders the color, or, more specifically, the characteristics of the wavelengths that contribute to the color of a light source. Luminance and illuminance are measures of light incident on a surface or from Lighting Class Road Surface Luminance Threshold Increment Surround Ratio Dry Wet Lav (cd/m2) Uo Ul Uo [TI (%)] [SR] M1 2.0 0.40 0.70 0.15 10 0.5 M2 1.5 0.40 0.70 0.15 10 0.5 M3 1.0 0.40 0.60 0.15 15 0.5 M4 0.75 0.40 0.60 0.15 15 0.5 M5 0.50 0.35 0.40 0.15 15 0.5 M6 0.30 0.35 0.40 0.15 20 0.5 Source: CIE 115 (2010), Table 2, âLighting of Roads for Motor and Pedestrian Traffic,â p. 18 . Note: Lav = average road surface luminance; Uo = overall uniformity of the luminance; Ul = longitudinal uniformity of the luminance. Table 1. Criteria for lighting areas adjacent to the traveled way.
16 Solid-State Roadway Lighting Design a source, respectively, and are the most prominent and often the easiest characteristics for measuring aspects of light. Depending on the amount of light in a given scenario, different modes of human vision come into play: photopic, scotopic, and mesopic. Lighting design guidelines often attempt to take into account these basic metrics (CIE 2010a; FHWA 2009; IES 2018); however, other elements such as eccentricity, contrast polarity, speed, and individual differences make it difficult to apply universal models for visibility. Luminous flux (lumens) describes the quantity of visible light emitted by a source. To deter- mine how the human eye evaluates lumens, a response curve V(l) was adopted to define the spectral response that a typical person would experience under photopic conditions. The SPD is weighted by V(l) at each wavelength, and then all values are integrated to determine lumen output. This output is regarded as accurate until there is a change in the viewing condition, which results in a change in the effectiveness of a lampâs output. Spectral metrics, or measures of colors produced by a light source, are important for differen- tiating a sourceâs abilities. The color depends on the SPD of the light source. An HPS lamp, for example, emits wavelengths that spike between 560 nm and 625 nm. The colors associated with this range are yellow and orange, which, unsurprisingly, result in the perceived hue from an HPS lamp. An LPS lamp has a minimal spectral output, except for a doublet emission line at 589 nm and 589.6 nm, which gives the light output a deep orange appearance. White light sources are more evenly balanced across the electromagnetic spectrum and are composed of more blue light, or shorter wavelengths, than HPS or LPS light (Gibbons et al. 2015b; Lewin et al. 2003). Basic measures of luminance such as average (Lavg), minimum (Lmin), and average-to-minimum ratios (Lavg/Lmin) drive design criteria and recommended practices. For example, the IES RP-8-18 (IES 2018), a recommended practice guide for roadway lighting, has specifications for average luminance and average uniformity ratios for certain road classes. A street classified as major with a high pedestrian conflict should have an average luminance of 1.2 cd/m2 and an average unifor- mity ratio of 3.0 Lavg/Lmin. A roadway classified as local with high pedestrian conflict is required to have an average luminance of 0.6 cd/m2 and an average uniformity ratio of 6.0 Lavg/Lmin. The difference in the two scenarios is the road class, which often involves differences in speed, lane width, and identifiable markings. Luminance metrics for road classes and pedestrian zones do not differ by the type of lighting system or the color temperature produced (IES 2018). The AASHTO Roadway Lighting Design Guide (AASHTO 2018b) uses similar luminance design values as IES RP-8-18 but also includes illuminance design values that can be used instead of luminance values. However, the illuminance recommendations do include limits on glare based on the veiling luminance ratio, which requires the calculation of roadway luminance as well as illuminance. Neither of these methods includes spectral modifiers. Similarly, illuminance specifications do not differ by the light source. A vehicle and pedestrian conflict zone, such as a walkway or crosswalk, requires an average vertical illuminance of 1.9 fc (20 lux) at 5 ft (1.5 m) from the ground, and this is true regardless of whether the lighting is an HID or LED installation (IES 2018). In most nighttime roadway applications, light levels fall in the mesopic range (0.001â10 cd/m2), where the CIE photopic luminous efficiency function (V(l)) does not accurately describe the visual performance for off-axis tasks because of the transition of cone photoreceptor vision to a mix of rod and cone photoreceptor vision. Models for mesopic vision attempt to predict the effective luminance (and this visual performance) in the mesopic adaptation range. There are three major mesopic models: ⢠X-model (Rea et al. 2004), ⢠Mesopic optimization of visual efficiency (MOVE) (Eloholma and Halonen 2006), and ⢠Recommended System for Mesopic Photometry Based on Visual Performance (CIE 2010b).
Literature Review and Gap Analysis 17 The X-model and the MOVE model have different photometric limits at which the mesopic range becomes active because the upper luminance limit of the X-model is 0.6 cd/m2, whereas the upper luminance limit of the MOVE model is 10 cd/m2. The X-model was based on achro- matic tasks, while the MOVE model was based on chromatic tasks. Because night driving involves both achromatic and chromatic tasks, CIE recommended an intermediate system that performed better for both achromatic and chromatics tasks (CIE 2010b). The CIE recommended system for mesopic photometry has an upper luminance limit of 5 cd/m2 (which is between those of the X-model and the MOVE model) and a lower luminance limit of 0.005 cd/m2 (CIE 2010b). Research showed that, for foveal target detection, V(l) was more accurate in predicting thresh- old contrast than mesopic models (including the X-model and the MOVE model) but that, for peripheral detection (10° and 14°), the mesopic models were more accurate at predicting threshold contrast (Raphael and Leibenfer 2007). In a recent study that sought to verify the applicability of the CIE-recommended mesopic model to nighttime roadway conditions, the model predicted visibility accurately for LED sources but was unable to achieve the same level of confidence for prediction involving HPS light sources (Gibbons et al. 2015b). Lewin et al. (2003) suggest the current metrics for determining the usefulness of a light source, specifically, those tied to luminance and illuminance, do not directly address the notion of visibility. It is generally agreed that an enhanced system for characterizing nighttime visibility is necessary. An inclusive approach that considers nighttime driving independent of daytime driving for pavement markings, signs, and differences in driver behavior at night should be considered (McInnis and Morton 2015). In addition to undergoing tests involving efficacy and visual performance, LEDs and HIDs should be tested against current standards to determine whether alterations to the standards themselves would benefit visibility and safety, and not just simply whether the light source can achieve the standards currently in place. Specific tests would be a consideration in a future gap analysis. Metrics fOr LigHting What We Know ⢠Illuminance and luminance have limited use in determining visibility. ⢠Visibility is one of several factors that can lead to a crash. ⢠Research has correlated lighting levels with crashes. ⢠The amount of optical control and output regulation that can be achieved with LED lighting will make lighting installations more closely match design criteria. ⢠Several metrics for lighting criteria are being developed and discussed, including visibility, visual performance, and probability metrics. ⢠Uniformity may have an effect on crashes. What We Donât Know ⢠Which design metric will result in lower crash rates for roadways. ⢠Whether spectral weighting of sources can affect crash rates. Results/Recommendations of This Study ⢠Several metrics were used in the research described in Chapter 5. ⢠Different models (e.g., probability models) are discussed in Chapter 5. ⢠Recommendations for additional research of more complete driver and lighting metrics are included under âResearch Roadmapâ in Chapter 5.
18 Solid-State Roadway Lighting Design Dirt Depreciation/Light Loss Factor The construction of SSL luminaires is different from that of HID luminaires. Some optical systems use individual LEDs with individual refractors, others use external lenses, and still others use reflector systems or light guides as part of their design. The amount of dirt depreciation for each type of design varies significantly. IES produced the document RES-1-16, Measure and Report Luminaire Dirt Depreciation (LDD) in LED Luminaires for Street and Roadway Lighting Applications (IES 2016). This docu- ment is based on research conducted by Virginia Tech Transportation Institute (VTTI) that measured the dirt depreciation of various optical systems and the effect of cleaning on those luminaires. The results of this research provide advice on predicting dirt depreciation for dif- ferent optical systems. The research compared the light output from the luminaire at the time of installation with that after 2 years of use (dirty) and that after the luminaire had been cleaned after 2 years of use (Figure 7). These results allowed the researchers to calculate the light loss due to lumen depreciation and due to dirt depreciation (Figure 8). In this calculation, the results from different optical methods (glass and reflector and individual optics) can be compared. Other factors also need to be considered when the light loss factor of SSL luminaires is being determined. LEDs are temperature dependent in terms of output and estimated life. Many LED 0.0 5.0 10.0 15.0 20.0 25.0 Design A (1) Design B (1) Design C (2) Design D (1) Design E (1) HPS 250 (2) Average Illuminance Before Average Illuminance After (Clean) Average Illuminance After (Dirty) Av er ag e Ho riz on ta l I llu m in an ce (L ux ) Figure 7. Effect of lumen and dirt depreciation with cleaning. -10% -9% -8% -7% -6% -5% -4% -3% -2% -1% 0% Design A (1) Design B (1) Design C (2) Design D (1) Design E (1) HPS 250 (2) LED Average (with D (2)) LED Average (w/o D (2)) Lumen Depreciation Dirt Depreciation Overall Figure 8. Lumen and dirt depreciation for different optical assemblies.
Literature Review and Gap Analysis 19 luminaires also have an automatic output adjustment if the temperature of internal fixture gets too high. Failure of an LED luminaire often occurs at the LED driver, the estimated failure rate of which is determined by MTBF calculations. The rated life of an LED luminaire is based on the point in time at which the lumen output of the LED has declined to 70% of its initial lumen rating. Some of these factors are quantified and included in the LED test data performed in accordance with Projecting Long Term Lumen Maintenance of LED Light Sources (IES 2011). Other data are usually available from the luminaire manufacturers. Dirt DepreciatiOn/LigHt LOss factOr What We Know ⢠Many LED luminaires collect dirt differently than HID luminaires because of their different optical systems, heat management, size, and shape. ⢠For LED systems, dirt depreciation affects optical performance and can alter expected life because dirt in heat sinks changes expected heat profiles. What We Donât Know ⢠Changes in heat profile due to dirt accumulation. Results/Recommendations of This Study ⢠Recommendations for light loss factors are included in the SSL Guide. ⢠Future research needs to include development of a test method for determining the expected dirt depreciation factors of various types of luminaire construction. Disability and Discomfort Glare Nonuniformity in the visual field, particularly that caused by bright sources, affects the adaptation level of the eye. Because these sources tend to fluctuate as the driver proceeds, the adaptation level is constantly changing (transient adaptation). Roadway lighting thus aids the eyes in adapting to an increased level of luminance more than headlights alone. Bright sources create other effects, collectively termed âglare,â which should be avoided as much as is practical. Disability glare occurs when light rays passing through the eye are slightly scattered, primarily because of diffusion in the lens and the vitreous humor that fills the anterior chamber of the eye. When a light source of high intensity is present in the field of view, this scattering tends to superimpose a luminous haze over the retina. The effect is similar to looking at the scene through a luminous veil. The luminance of this veil is added to both the task and background luminance and thereby reduces contrast. The effect is termed âdisability glareâ or âveiling lumi- nance,â and it may be numerically evaluated by expressing the luminance of the equivalent luminous veil. A well-known example of this is trying to see beyond oncoming headlights at night. Because of the reduction in contrast caused by disability glare, visibility is decreased. Increasing luminance counteracts this effect by reducing the eyeâs contrast sensitivity. Research has shown that disability glare is not significantly influenced by the SPD of the glare source (Davoudian et al. 2014; Woten and Geri 1987). Thus, roadway lighting from LED sources will have the same effect on disability as other conventional roadway lighting sources.
20 Solid-State Roadway Lighting Design Discomfort glare is a further result of overly bright light sources in the field of view and causes a sense of pain or annoyance. While its exact cause is not known, it may result from pain in the muscles that cause the pupil to close. The SPD of the light source could affect the perception of discomfort glare, with vehicle headlamps that have a higher blue content caus- ing more discomfort glare (Bullough et al. 2003). Moreover, recent research into perception of discomfort glare from LED luminaires has shown that higher discomfort glare ratings are due to the luminaire design rather than the SPD (van Bommel 2014). LED roadway luminaires are composed of arrays of small LEDs, which result in a nonuniform luminance distribution of the LED as compared with a conventional luminaire. Research has shown that this non uniform distribution of luminance results in a higher discomfort glare rating (Tashiro et al. 2015). Therefore, discomfort glare from LED sources is a result of the optical design of the luminaires rather than the SPD. Disability glare and discomfort glare normally accompany one another, and beneficial lumi- naire light control that reduces one form of glare is likely to reduce the other. Discomfort glare, which can cause effects from an increased blink rate to tears and pain, does not reduce visibility. It is also generally accepted that reducing disability glare will reduce discomfort glare; however, it is possible to reduce discomfort glare and increase disability glare. North American roadway lighting standards do not specify numerical limits for discomfort glare. Methods of quantifying discomfort glare from roadway lighting exist but are mostly subjective. In addition, no instru- ments have been developed to measure discomfort glare. The presence of extraneous light in the field of view may cause a nuisance or distraction to the driver, independent of the effects of disability and discomfort. Bright light sources tend to cause distraction, and the eye may be drawn to them. Lighting that is used for advertising, for example, may cause visual clutter and add complexity to the scene, making the driving task more challenging. Beyond this definition, there is no measure or method of assessing nuisance glare. DisaBiLity anD DiscOMfOrt gLare What We Know ⢠Disability glare from roadway lighting systems is predicted by several models, including veiling luminance ratio, threshold increment, and glare mark. IES and AASHTO currently use the luminance ratio. ⢠Discomfort glare is most commonly measured using the deBoer rating scale. Other models that have been developed look at source luminance, background luminance, source size, and viewing position. ⢠Different spectral content sources may be perceived differently. ⢠The impact of disability glare may be much more pronounced in the aged eye. What We Donât Know ⢠How discomfort glare is different for different spectral content sources. ⢠How different SPD affects user comfort. Results/Recommendations of This Study ⢠Disability glare was experimentally controlled as part of the work of this study and is described in Chapter 5. ⢠Discomfort glare was included in the results obtained and is described in Chapter 5.
Literature Review and Gap Analysis 21 Considerations for Pedestrians, Cyclists, and Motorists The volume of road users at night is far less than during the day; even so, areas where traffic volumes and the rate of conflict between drivers and pedestrians are higher are typically lighted with artificial lighting to improve safety. In addition, locations where pedestrians and cyclists are common are typically lighted to specifications regarding their visibility and safety (IES 2018). In 2015, there were 32,166 fatal crashes in the United States, and 28% of those occurred at night on unlit roadways. Considering the significant decrease in nighttime traffic exposure, 28% represents a significant number of crashes (NHTSA 2017). These statistics boil down to a significant need for research in lighting to improve motorist, pedestrian, and cyclist safety. While exposure is not directly accounted for in this data, lighted areas are known to have a higher volume of traffic and more frequent conflicts with pedestrians and cyclists. These fatal- ity statistics show a benefit in lighting, although not all crashes can be attributed to an absence of lighting. The purpose of roadway lighting is to increase the visibility of the roadway and the road environment and to increase visual awareness of potential conflicts. While there is no direct link between visibility and safety, roadway lighting has been shown to improve safety (Donnell et al. 2009; Rea et al. 2009). Roadway lighting augments the amount of lighting provided by motor vehicle headlamps and provides context to the roadway by illuminating the surround- ings. By improving visibility, roadway lighting serves to reduce accidents and increase safety and the feeling of safety. A meta-analysis performed by Elvik (1995) aggregated the results from 37 studies and determined that introducing roadway lighting should reduce fatal nighttime crashes by up to 65%. There were some issues with the study because a wide variety of approaches and measures were used in the meta-analysis. As a result, further studies have been undertaken to build the link between safety and roadway lighting (Bruneau et al. 2001; Senadheera et al. 1997; Oya et al. 2002). One such study (Gibbons et al. 2014) investigated the link between lighting level and the night-to-day crash rate ratio. This study was the largest consideration of lighting and crash reduction ever undertaken, having measured 1,000 centerline miles of lighting and considered more than 88,000 crashes. This study examined lighting measured in situ and was not based on estimates from design approaches or typical luminaire layouts from DOT specifications. The study investigated lighting impact and found a limit to safety improvements, namely, while additional lighting on a roadway provides benefits, it exhibits diminishing returns. Increasing light levels, as shown on the x-axis of Figure 9, decrease the night-to-day crash rate ratio. The impact of the decrease differs on the basis of road class, as Figure 10 shows. The dashed lines indicate where additional lighting does not improve road safety for those roadway types. To better understand the link between lighting, visibility, and safety, several metrics have been investigated, including small target visibility, relative visual performance, and cumulative detection probability (Adrian 1989; Bhagavathula and Gibbons 2015; Bullough et al. 2013); however, none of these metrics has fully described the relationship. Many experiments have been performed in which time to collision was measured at the moment a participant detected a con- flict. Time to collision is a standard safety surrogate (Gettman and Head 2003) but is often not explicitly calculated from the separation distance at detection (i.e., detection distance) because experiments are performed at fixed speeds; therefore, detection distance is used often as a safety surrogate. These test track studies have shown increased detection distances for pedestrians and cyclists when roadway lighting is used (Gibbons et al. 2012). When compared with headlamps alone,
22 Solid-State Roadway Lighting Design Source: Gibbons et al. (2014). Figure 9. Relationship between night-to-day crash rate ratio and lighting level. Source: Gibbons et al. (2014). Figure 10. Relationship between night-to-day crash rate ratio and lighting level, by road class.
Literature Review and Gap Analysis 23 roadway lighting doubles the distance at which a pedestrian is visible (Gibbons et al. 2015b). This is true even when a lighting system is at 40% of its full power (Gibbons et al. 2015b) (Fig- ure 11), likely because of the impact of increased surface luminance that reduced crash rates by 35% when augmented by an average of 1 cd/m2 (Scott 1980). The Scott study shows that an increase in road surface luminance and a smaller increase in the headlamp power or the vertical illuminance on an object in the roadway increase the negative contrast. An increase in negative contrast makes an object more visible through its silhouette and therefore increases its visibility. Aside from its visibility aspects, lighting can also be a modifier of driver behavior. A study conducted by Li et al. (2017) related the naturalistic time series data on driver behavior from the second Strategic Highway Research Program (SHRP 2) to field lighting measurements conducted by VTTI to explore the correlations between roadway lighting parameters (average illuminance and uniformity) and several safety surrogate variables related to driver behavior (e.g., speed, acceleration, time to collision, etc.) at entrances and exits of highway ramps. Effects of lighting were more significant for entrance than for exit ramps. Results indicated that higher right-lane illuminances were correlated with lower speeds and lower and more gradual lane changes. Overall, illuminance seemed to correlate more with driver behavior than uniformity. Recommendations and guidelines for lighting implementations can be found in publications from IES, CIE, the Transportation Association of Canada, and AASHTO. These organizations have recommended minimum lighting levels, placements, luminaire types, and lighting trajectory guides to aid in implementations. While the addition of lighting is stated to have a net positive effect on safety, there are special considerations in terms of economy, glare, and distribution that are important and addressed in these guides. As an example, IESâs RP-8 suggests recommended lighting levels that are based on luminance at the roadway surface for several road classifica- tions to maximize uniformity and provides details of the best methods for measuring lighting distribution over a given area (IES 2018). Note: Same letters indicate no significant differences. Figure 11. Overhead lighting level experiment: StudentâNewmanâKeuls groupings for pedestrian detection distance with headlamps on by overhead lighting level.
24 Solid-State Roadway Lighting Design cOnsiDeratiOns fOr peDestrians, cycLists, anD MOtOrists What We Know ⢠Research has shown the impact of lighting on safety. ⢠Nighttime risks are greatest for pedestrians and cyclists. ⢠Lighting affects safety differently for different types of roadway facilities. ⢠Headlights impact safety and are currently not quantified as part of lighting recommendations. What We Donât Know ⢠The most effective means for assessing safety with regard to roadway lighting at night. ⢠The interaction of various roadway geometries and features with lighting with regard to safety. Results/Recommendations of This Study ⢠Research regarding headlamps, pedestrians, and small objects in the roadway was performed as part of this study and is described in Chapter 5. CostâBenefit Methodologies Life-cycle costs are mainly used by designers to evaluate and compare various lighting sys- tems (e.g., conventional versus high-mast lighting, LED versus HID lighting systems). Often, a lighting system may have a less-expensive capital cost, but when life-cycle costs are considered, a lighting system with a higher capital cost may result in significant cost savings over the life of the system. This is mostly the case when LED and HID systems are compared; however, it can also be a major factor in the selection of one LED luminaire over another. The use of adaptive lighting systems also has benefits from a cost and maintenance perspective. Life-cycle costs include the capital cost, as defined above, as well as operating costs over the estimated life of the system. Operating costs include power and preventive maintenance costs, which are also calculated over the life of the system (typically 30 years, although this will vary depending on the grade of the equipment used). It is doubtful that, after 30 years of operation, existing equipment would be reused; therefore, no residual value should be considered. When life-cycle costs are used to compare lighting systems, current operating costs (defined at the time of the estimate) can be used as the basis over the operating period. Because costs are most likely to increase over time, inflation may be factored in to provide a more-accurate estimate of total costs. For baseline comparisons, the following provide a good comparative tool for system selection: selecting a period of analysis (e.g., 20 years); estimating the capital, energy, and main- tenance costs of the system; and developing a present worth value. The photometric performance and overall quality and design of LEDs vary greatly from product to productâmore so than for HID luminaires. When an LED is being selected, the life cycle of the product should also be considered because it is often more significant over time than using product cost alone. To achieve the best value, it is therefore critical to use life-cycle costs that include power costs and maintenance costs (such as cleaning of the luminaire optics) for the life cycle of the lighting system.
Literature Review and Gap Analysis 25 Several other factors can be considered in a costâbenefit analysis, depending on power costs and utility agreements, such as the use of adaptive lighting technologies and asset management tools and the availability of energy rebates, if applicable. The contracting methodology, which would include things like asset value, lease opportunities, and the cost of money, also can be considered in the analysis. The value of increased safety as a result of decreased nighttime crash rates can also be assessed, if that information is available. In assessments of the financial elements involved in selecting a lighting system, two main methods are commonly used: ⢠Return on investment (ROI): ROI is the profit generated by the money an owner puts into an investment. ROI is typically expressed as a percentage of return. ⢠Payback: The simple payback period refers to the period of time needed to pay for the sum of the original investment. For example, a $100,000 investment that returned $10,000 per year would have a 10-year payback period. The time value of money is not taken into account. cOstâBenefit MetHODOLOgies What We Know ⢠Current costâbenefit models show energy and maintenance savings. ⢠Current models do not factor in the cost for crash reduction in the costâbenefit model. ⢠LED products vary greatly in photometric efficiency (watts/area being lit) and energy consumption over the product life cycle. What We Donât Know ⢠Potential crash reduction from the use of LED lighting and how that translates to costâbenefit. Results/Recommendations of This Study ⢠Future research and evaluation of crash data and benefits are included under âResearch Roadmapâ in Chapter 5. Applications and Obstacles to Adaptive Lighting The ability to easily control and dim LED luminaires is one of their biggest advantages over other sources of roadway lighting. Dimming LED luminaires can result in additional energy savings of up to 30% (Avrenli et al. 2012). The concept of dimming roadway lighting during periods of low vehicular and pedestrian activity is called adaptive lighting. Adaptive lighting control can be key to reducing the impact of an exterior lighting system on sky glow, glare, and light trespass. It is an emerging technology but has already been effectively applied in a few cities. The purpose of an adaptive lighting control system is to reduce the impact of the lighting system and save a considerable amount of energy and maintenance costs. Adaptive street and roadway lighting systems have already been installed in many cities and towns in the United States and have resulted in energy savings. In California, the city of San Jose and the University of California, Davis, converted to LEDs and installed adaptive
26 Solid-State Roadway Lighting Design lighting controls. Energy savings between 84% and 87% were observed. Installation of an adaptive lighting control system can result in energy savings between 30% and 46% (Escolar et al. 2014; Lau et al. 2013; U.S. Department of Energy, 2013). Cambridge, Massachusetts, converted to LEDs and installed an adaptive lighting system, which resulted in a total energy reduction of 80% with about $500,000 in energy savings per year (Echelon 2016). The payback time for this type of adaptive lighting control system is relatively short (3â5 years) and will only improve with advances in adaptive lighting technologies and increases in the efficacy of LED luminaires (Clanton and Associates Inc. 2014; Echelon 2016). One of the issues affecting the adoption of adaptive lighting is the absence of standardized guidelines for reduced light levels. The current AASHTO standards do not have a method for determining dimmed values. IES uses pedestrian volumes to determine changes in lighting requirements, but those pedestrian criteria apply only to streets and not to limited-access road- ways (Table 2). FHWAâs Guidelines for the Implementation of Reduced Lighting on Roadways (Gibbons et al. 2014) offers a method of classifying highways that allows for the dimming of lighting on high- ways on the basis of various factors, such as traffic volumes. This method can be reviewed and considered as part of the SSL Guide, as shown in Figure 12. Another potential impediment to adoption of an adaptive lighting system arises when street- lights are powered from a nonmetered service and paid for by an established utility rate structure. Most utilities do not offer rate structures that address adaptive lighting technologies (Gibbons et al. 2014). A third issue with adaptive lighting control systems is security. The lighting control network must include the appropriate security protocols to avoid unauthorized use by hackers. Finally, locations where adaptive lighting control systems will be installed should be carefully selected, because improper selection and installation can result in decreased traffic safety. Locations with highly variable and mixed traffic patterns (like signalized intersections and roundabouts) should be given special attention before an adaptive lighting control system is installed. Street Classification Pedestrian Area Classification Average Luminance (cd/m2) (Lavg Lavg) Average Uniformity Ratio ( /Lmin) Maximum Uniformity Ratio (Lmax/Lmin) Maximum Veiling Luminance Ratio (Lv,max /Lavg) Major High 1.2 3.0 5.0 0.3 Medium 0.9 3.0 5.0 0.3 Low 0.6 3.5 6.0 0.3 Collector High 0.8 3.0 5.0 0.4 Medium 0.6 3.5 6.0 0.4 Low 0.4 4.0 8.0 0.4 Local High 0.6 6.0 10.0 0.4 Medium 0.5 6.0 10.0 0.4 Low 0.3 6.0 10.0 0.4 Source: IES RP-8-14 (IES 2018). Note: Lmin = minimum luminance; Lmax = maximum luminance; Lv,max = maximum veiling luminance Table 2. Street lighting levels by pedestrian volume.
Literature Review and Gap Analysis 27 Source: Gibbons et al. (2014). Figure 12. Classification method for streets. appLicatiOns anD OBstacLes tO aDaptive LigHting What We Know ⢠IES and FHWA have included recommendations for adaptive lighting in their guides. ⢠Adaptive lighting systems require equipment capable of using this type of control. ⢠Various types of adaptive lighting systems exist, including cell based, wireless, and power line carrier systems. ⢠Communication protocols and system security are key considerations in using these systems. ⢠There are several successful implementations of adaptive lighting systems. What We Donât Know ⢠The future of adaptive lighting systems and integration with Smart city technologies. ⢠What the integration of adaptive lighting with autonomous and connected vehicles will be. ⢠What other systems should be an input into adaptive lighting control. (continued on next page)
28 Solid-State Roadway Lighting Design appLicatiOns anD OBstacLes tO aDaptive LigHting (continued) Results/Recommendations of This Study ⢠Guidance is included in the SSL Guide. ⢠Spectral modifications are included as part of adaptive lighting systems discussed as potential future research. Impacts of Weather and Climate on Lighting Effectiveness Weather can dramatically change the way light behaves and affect the human perception of brightness, glare, and depth. Roadway lighting can occasionally be an even greater hindrance in the presence of certain inclement weather such as dense fog, rain, or heavy snow. The United Kingdom has roadway lighting recommendations for when the roadway is wet versus dry. The United Kingdom is more accustomed to rainfall than most parts of the United States, so specific guidelines regarding wet pavement make sense. In general, these guidelines specify that the ratio of uniformity be lower for wet pavement than for dry; that is, lighting on wet pavement should be more uniform (IES 2018). Areas of dense fog are often the most difficult to design lighting for because fog is dynamic and consists of pockets of varying thickness. In general, more intense or brighter light is discouraged during times of heavy fog, because fog particles scatter and reflect light, which results in greater difficulty in seeing (Boyce 2009). The effect is most pronounced with headlamps but also occurs with roadway lighting. This scattering of light results in veiling luminance, or glare, that restricts the visibility of a motorist (Boyce 2009). The use of high beams in fog is not recommended for this reason. Conventional roadway lighting is not âsmartâ enough to allow dimming during periods of fog and can result in light scatter and driving situations in which visibility is more difficult because of the presence of light. In a recent study investigating the detection of pedestrians in clear weather, rain, and fog, researchers found that the spectral impact of the light source was diminished in conditions of rain and fog and that the more critical impact was that of the intensity of the light source (Gibbons 2016). The mean detection distance of objects in the various weather conditions is shown in Figure 13. It should be noted that the weather also affected vehicle speed and object contrast, so these results require significant analysis and caution. Conventional roadway lighting is mounted between 8 and 9 m from the ground, pointing downward. Computer simulations by Girasole et al. (1998) found that a lower mounting height of about 1 m illuminated the forward roadway, and he postulated that, at this height, the light did not have to permeate as much fog, so that there was less light scatter and therefore less glare to the driver. Snow presents unique challenges to visibility that rain does not. First, snowfall obstructs vis- ibility and can obscure the roadway, depending on the size of the falling snowflakes and the rate of snowfall. In general, lighting cannot improve the visual conditions caused by heavy snowfall. Additionally, standing snow on the roadway and on the edges of the roadway is highly reflec- tive and can cover important lane markings. In general, the reflective nature of standing snow is more likely to be beneficial for visibility (Boyce 2009) because the environment is illuminated by the reflectance of ambient light.
Literature Review and Gap Analysis 29 With the advent of adaptive lighting technologies, it may soon be possible for lighting to change in response to weather conditions. In cases of severe inclement weather, such as heavy rain or fog, it may be best for lighting to be dimmed or for the spacing to be increased (Boyce 2009). Owing to the difficulty in replicating fog-like or heavy rain-like conditions in a road setting, more research is needed on the affordances different lighting systems can provide for visibility. iMpacts Of weatHer anD cLiMate On LigHting effectiveness What We Know ⢠Various weather conditions alter the visibility on the roadway. ⢠Different weather conditions (e.g., rain, snow, fog) alter different things. ⢠Different sources of spectral content perform differently under different weather conditions. ⢠Adaptive lighting systems can help adjust roadway lighting in response to weather. What We Donât Know ⢠Specific lighting metrics to alter for differing weather conditions. Results/Recommendations of This Study ⢠Weather is included as a topic under âResearch Roadmapâ in Chapter 5. Source: Gibbons (2016). Figure 13. Detection distance by weather conditions and light source.
30 Solid-State Roadway Lighting Design Application to Freeways, Arterials, Intersections, and Interchanges SSL lighting is applicable to all forms of outdoor lighting, including freeways, arterials, inter- sections, and most modern architecture. LEDs have already replaced many incandescent systems in the transportation industry, including signalized lights and some headlamps. SSL is less suscep- tible to mechanical failure and therefore is a more efficient and longer-lasting option than non-SSL (Boyce 2009). Earlier issues with SSL, such as sufficient lumen output, optical systems, and long- term viability, have been addressed in currently available products. Earlier issues with manufacturer availability have also been much improved. In fact, from a design perspective, almost all design work in the past 2 years or more has used LED technologies. The application of the technology is, how- ever, being done in the same way and using the same AASHTO and IES recommendations that are used for HID technologies. The amount of optical control offered by LED technologies has raised some questions regarding the applicability of those recommendations. Further research is needed to determine measurable reductions in sky glow resulting from LED optics; considerations of sidewalk- specific lighting, which has previously taken advantage of spilled light from the roadway; and the impact of accommodation and transient-dark adaptation that may result from modern optics. Testing that compared LED with HPS in a roadway environment with targets matched for vertical illuminance determined that LED sources provided a benefit to visibility (Table 3). These tests were completed at different locations with different environments and geometries, Luminaire Type CCT and System Type ~Average Target Detection Distance (ft) Anchorage, Alaska LED 4100K 213 LED 4300K 210 Induction 4000K 174 LED 3500K 167 HPS 2000K 141 San Diego, California LED 3500K 135 Induction 3000K 131 HPS 2100K 128 Induction 3000K 125 LED 3500K 105 San Jose, California LED 5000K 233 LED 4000K 223 Induction 4000K 197 HPS 2100K 193 LPS 1700K 190 LED 3500K 157 Seattle, Washington LED 4100K 145 LED 4000K 138 LED 5000K 122 HPS 2000K 103 LED 3500K 100 HPS 2000K 68 Table 3. Detection distances of luminaires from evaluations comparing different lighting systems.
Literature Review and Gap Analysis 31 so the locations themselves are not directly comparable. It is also important to note that the light sources used in those comparisons differed by manufacturer and experimental location; however, at the same testing sites, the LEDs were the best performers in terms of visibility. Direct comparisons with HPS sources indicated that LED lamps, typically between 4000K and 5000K, significantly outperformed HPS and LPS sources in terms of visibility distance (Mutmansky et al. 2010a, 2010b; Gibbons et al. 2010; Clanton and Associates Inc. 2014). appLicatiOn tO freeways, arteriaLs, intersectiOns, anD intercHanges What We Know ⢠Design approaches are different for various roadway types. For SSL, the benefits from certain spectral content are different. ⢠Design metrics most likely will be different for roadways and intersections/ interchanges because the driving task is different. ⢠Driver speed and behavior can be modified according to the availability of light. What We Donât Know ⢠Which metric is best. ⢠Impact on overall safety. Results/Recommendations of This Study ⢠Different SPD sources were included in the testing, and the results are described in Chapter 5. High-Mast Lighting High-mast and conventional roadway designs have been effectively executed with SSL tech- nologies. The largest problem in the past has been finding luminaires with adequate output to replace HID high-mast fixtures and higher-wattage highway luminaires, but a small number of SSL options have been developed. The control of the light from the LED luminaire is better and therefore allows better limits on the light trespass from the installation. Figure 14 provides an Figure 14. Example of high-mast lighting and light trespass.
32 Solid-State Roadway Lighting Design example of typical light trespass issues often seen with high-mast lighting. Shielding LED lumi- naires is sometimes more difficult because, instead of a single lamp with a reflector, luminaires include several point sources of light, and conventional side shielding is not as effective. The use of LEDs in high-mast interchange lighting is small but growing, likely due to costs associated with the installation and replacement of HID counterparts, although the lifelong maintenance costs of LEDs are considerably lower. Research has determined that the light dis- tribution of high-mast LED fixtures is better than that of conventional HPS fixtures common to high-mast applications (Cai 2015). To date, the selection of LED high-mast products is much smaller than that of conventional cobra head luminaires. HigH-Mast LigHting What We Know ⢠High-mast systems are currently designed to meet the same requirements as conventional lighting. ⢠In the past, lower lighting levels were allowed for high-mast systems. ⢠Light trespass and obtrusive light complaints occur more often with high-mast systems. What We Donât Know ⢠The effect of high-mast lighting on safety. ⢠Whether the reduction in glare produced by these systems increases visibility. ⢠How this approach relates to the off-road findings. Results/Recommendations of This Study ⢠Additional research is included under âResearch Roadmapâ in Chapter 5. ⢠Shielding of LED high-mast luminaires is different from that of HID/reflector luminaires and is discussed in the SSL Guide. Potential Environmental and Health Effects Environmental and health concerns have been raised regarding the use of LED sources and their spectral content. Research in human circadian rhythms has shown that exposure to blue light from LEDs in light sources and devices in the evening can disturb circadian rhythms and cause sleep loss by suppressing melatonin production (Cajochen et al. 2011; Chang et al. 2015; West et al. 2011). The shorter wavelengths from LEDs can also result in higher discomfort glare as compared with other light sources for the same photopic illuminance measured at the eye of the observer (Rea 2017). Another drawback of LED roadway lighting pertains to maintaining dark skies and limit- ing light trespass related to astronomical and ecological considerations. With the added blue content from the SPD of LEDs, Rayleigh scattering of light in the atmosphere increases sky glow (Luginbuhl et al. 2009). LED light sources can also cause sky glow because they emit more energy in the shorter wavelengths. Shorter wavelengths scatter more than longer wavelengths and can affect the visibility of stars from the earth (Kinzey et al. 2017). Sky glow is also affected by the total light output from the light source and the distribution of light from the luminaire (uplight, in particular) (Kinzey et al. 2017). Recent research shows that sky glow is affected by aerosol content and the operating characteristics of the detector used to assess sky glow (Rea
Literature Review and Gap Analysis 33 and Bierman 2014). The effects of sky glow can be mitigated with the use of adaptive lighting technology, which would minimize the total amount of flux generated from the source. Because of the above-mentioned concerns, a recent report by the American Medical Asso- ciation, Human and Environmental Effects of Light Emitting Diode (LED) Community Lighting (Kraus 2016), recommends the use of lower CCTs (possibly because of lower blue content in the SPD) to minimize potential ill effects on health and the environment. However, using CCT as a defining metric for determining harmful health effects from LED lighting is not accurate because it is not the only factor involved in defining light exposure (Rea and Figueiro 2016). CCT is not a definitive metric for determining the blue content of a light source. Moreover, the spectral content of the LED light source determines the amount of blue light emitted by the source. There is no metric for determining the blue content in an LED light source other than measuring the energies of each of the wavelengths that compose the spectrum. The lack of an easy metric for determining the blue content of LED light sources has led to the adoption of CCT in the lighting industry, but CCT is not a metric of the SPD; rather, CCT gives information on the color appearance of the light source. Dosage, which is the duration and intensity (or level) of light exposure, also plays an impor- tant role in the disruption of circadian rhythms. The majority of the research that has exam- ined the impact of light on circadian rhythms was conducted on shift workers or animals in controlled environments who were exposed to very high light levels (Borugian et al. 2005; Cissé et al. 2016, 2017; Dauchy et al. 2014; Lewy et al. 1980; Schernhammer et al. 2003). For street lighting, though, the duration of exposure to light is much lower than the duration of exposure for shift workers. Light levels from street and outdoor lighting are also much lower than those in a lighted nighttime work environment (McLean and Lutkevich 2016). Furthermore, expo- sure to light levels from inside a personâs home could be much higher than exposure to light levels from street lighting (Kinzey 2016). Another study that assessed the effect of melatonin suppression induced by indirect blue light from inside a stationary automobile reported that there was no melatonin suppression with exposure to a light level of up to 1.25 lux (Lerchl et al. 2009). Some epidemiological studies that examined the effect of artificial light at night on the incidence of breast cancers found that higher levels of outdoor artificial light at night were correlated with higher incidences of breast cancer (Hurley et al. 2014; Kloog et al. 2008, 2010; Portnov et al. 2016). In all of these studies, outdoor nighttime light levels were estimated by using satellite imagery data from the U.S. Defense Meteorological Satellite Program; light expo- sure was not measured at an individual level. Research also showed no relationship between light levels estimated from satellite imagery and light levels experienced by individuals (Rea et al. 2011). However, there is no existing research that shows that LED roadway lighting (of any SPD) at light trespass levels and exposure durations under realistic road conditions disrupts human circadian rhythms. Research shows that using LED roadway lighting results in longer detection distances. A human factors study that compared various types of LED and HPS sources in an urban roadway environment concluded that the CCT of an LED source could play an important role in detec- tion distance and relative safety (Clanton and Associates Inc., 2014). Because CCT was the only metric available during the testing, it was used as the metric of interest. This study incorporated a variety of colored targets located on the roadway under matched LEDs with different CCTs that were dimmed to match the road surface luminance. The targets were located in the road- way at points of equivalent vertical illuminance to control the object contrast. As a foveal task, participants were asked to search for and indicate when they could perceive the target in the roadway. The targets were of different colors; therefore, the benefit of the lighting is believed to be from color contrast. It is important to note that CCT was the only metric available during the time period of this testing; therefore, it is the metric used for describing the results. The results
34 Solid-State Roadway Lighting Design of this research found that objects lighted with a 4100K CCT had as much as a 20% higher detection distance over other color temperatures. Detection distances of color targets were also the highest for the 4100K CCT LED. The results of the study also indicated that, to achieve the same level of visibility with a 3000K light source, higher levels of light were required than with a 4000K light source. The higher light levels needed for the 3000K LEDs could increase power consumption by about 8% to 10% as compared with the 4000K LED (McLean and Lutkevich 2016). Therefore, reducing the blue spectral component in the light source may increase energy consumption. Researchers have argued that, to correctly understand the effect of light on the disruption of circadian rhythms, light stimulus must be measured in terms of circadian response rather than in terms of the conventional visual response (Figueiro 2017). One proposed mathemati- cal model allows the response of acute melatonin suppression to be predicted after 1 hour of exposure to a specific light level and light spectrum (Rea et al. 2005). The circadian light is com- parable to photopic lux measured at the eye (but takes into account the circadian response of the eye) and can then be used to determine the circadian stimulus, which is the effectiveness of the incident light at suppressing melatonin. The circadian stimulus ranges from 0 to 0.7, where 0.1 is the threshold and 0.7 is the point of saturation. Comparison of circadian light values for 4000K and 3000K LEDs showed that the 4000K LED was less effective at melatonin suppression than the 3000K LED (Rea 2017). The biological effects of light on humans can also be measured in terms of the response of the five potential photoreceptors in the eye that can affect circa- dian responses: short-wavelength cones, medium-wavelength cones, long-wavelength cones, intrinsically photosensitive retinal ganglion cells (ipRGCs), and rods (Lucas et al. 2014). The five sensitivity functions (one for reach receptor) can be used to calculate the activation of each of the receptors, which can help in comparing the light sources of different SPDs. Research that evaluated the effect of time of exposure and light level from 3000K and 4000K LEDs on melatonin suppression showed that melatonin suppression was significantly affected by time of exposure and light level (Lighting Research Center 2016). CCT (or the light source spectrum) did not significantly affect melatonin suppression. These results show that there are no discern- able differences between the CCTs of 3000K and 4000K LEDs in terms of health effects; however, a higher light level is required for the 3000K LED to maintain a level of visibility similar to the 4000K LED, which results in higher energy consumption. Other environmental impacts include the effect of roadway lighting significantly delay- ing the maturity of plants, such as soybeans, that need a night cycle to mature. Research that evaluated the effect of HPS road lighting on the maturity of the soybean plants in Illinois showed that soybeans exposed to light trespass were delayed from two to seven weeks (Palmer et al. 2017). The effect of LED roadway lighting on the maturity of soybean plants is yet to be reported. However, a study conducted in a lab examined the effect of LEDs with increas- ing amounts of blue content on the growth and development of soybeans and reported that LEDs with higher blue content resulted in shorter stems (Cope and Bugbee 2013). Overall, LEDs with lower blue content resulted in stem elongation and leaf expansion, and LEDs with higher blue content resulted in plants that were more compact. These results show that blue content on the spectrum of LED roadway lighting could potentially influence the maturity of soybeans and that proper precautions must be taken to limit light trespass into fields adjacent to lighted roadways. The presence of artificial light also affects the activities of wildlife. For example, the regular movements of hatchling marine turtles toward the sea occur primarily at night, but the presence of artificial light disrupts their photic cues and can cause them to move away from the sea, which often leads to mortal consequences (Peters and Verhoeven 1994). Artificial light also adversely affects specific breeds of bats (Rydell 1991) and mice (Bird et al. 2004) that forage predominantly
Literature Review and Gap Analysis 35 in areas of darkness. Because the health, growth, behavior, and maturity of some plants and wildlife can be affected by roadway lighting, it is important to engage in research that leads to favorable outcomes for both ecology and highway safety. pOtentiaL envirOnMentaL anD HeaLtH effects What We Know ⢠Artificial light at night has a physiological effect that is generally affected by light level, spectral content, and duration of exposure. ⢠Current research does not appear to show a significant human health impact from well-designed and -controlled streetlights. ⢠Current common methods of classifying the spectral content of luminaires (CCT) do not adequately describe spectral content. ⢠Exterior lighting at night has an impact on the environment in terms of sky glow and changes in the natural cycles of flora and fauna. This effect is generally affected by level and spectral content. What We Donât Know ⢠Whether further research will show impacts of melatonin suppression at light levels encountered on roadways. Results/Recommendations of This Study ⢠Future research is identified under âResearch Roadmapâ in Chapter 5. Aging Drivers Older drivers (those aged 65 years and older) are a growing population that makes up about 15% of the population. As of 2015, older drivers made up 18% of all traffic fatalities and 10% of all crash injuries, an increase of 8% for both statistics from 2014. Over the 10-year period between 2006 and 2015, which saw a steady 33% increase of older drivers, fatalities climbed 3%, a significant number considering the population growth. Almost three-quarters of the total fatal crashes involving older drivers occur during the day (NHTSA 2015); however, the exposure of older drivers to nighttime driving is much lower because fewer seniors opt to drive in dark conditions. The fact that one-quarter of fatalities of older drivers occurs at night points to a noteworthy research and design target. Some physiological factors associated with aging affect visibility. As the eye ages, the optical density of the crystalline lens is affected. The crystalline lens of the eye is responsible for focusing light onto the retina. The yellowing lens filters certain wavelengths of light from reaching the retina and alters color perception and contrast sensitivity. This is important in the differentia- tion of the visual experience of older drivers under HPS lighting, which produces a yellow hue, and LED lighting, which is typically whiter or bluer (Boyce 2009). The changing density of the crystalline lens by age in years is shown in Figure 15. A notable decline in contrast sensitivity begins between ages 40 and 50 (Ross et al. 1985). In addition to filtered wavelengths and poorer contrast sensitivity, the yellowing of the lens is also said to increase susceptibility to glare (DMD and Associates Ltd. 2013). Aging factors, or multipliers, have been recently factored into many design guidelines, which are by default based on a 25-year-old observer (DMD and Associates Ltd. 2013; IES 2018).
36 Solid-State Roadway Lighting Design Yellowing of the lens is not the most dire or prevalent physiological issue facing older drivers. Medical conditions such as diabetic retinopathy (National Eye Institute 2015b), glaucoma (National Eye Institute 2010), and macular degeneration (National Eye Institute 2015a) have more impactful and serious effects on vision. While issues stemming from the yellowing of the crystalline lens may be mitigated through the spectral content of light, the effects of the afore- mentioned conditions cannot generally be improved through changes in roadway lighting. As expected, the mechanics of the eye also deteriorate with age. Consequently, the range of the pupil area narrows specifically on the maximum end, making it difficult for older drivers to absorb more light in low-light levels. It is estimated that the wavelengths filtered by the clouded lens and the reduced abilities of the pupil result in a 70-year-old eye receiving only one-third of the light a 30-year-old eye can receive in a similar setting. This vast difference in visual abilities in a driving population should be considered in lighting design, specifically, in terms of intensity and wavelength (Adrian 1995). In 1998, FHWA published the Older Driver Highway Design Handbook, seeking to provide highway designers and engineers with a practical information source linking the declining func- tional capabilities of older road users to the need for design, operational, and traffic engineering enhancements keyed to specific roadway features (see Staplin et al. 2001). An update, Handbook for Designing Roadways for the Aging Population (FHWA-SA-14-015), which considers recent research on the aging population and modern advancements in lighting and technology was published in 2014 (Brewer et al. 2014). Ultimately, visibility as it relates to age does not depend on lamp type, either gas discharge or solid-state, but the intensity, uniformity, and spectral properties of the lighting can affect how well an older driver can see at night. LED lamps are more malleable in how they can be fine-tuned for color, aiming, and intensity than are HIDs, which may benefit an older population. Source: Coren and Girgus (1972). Figure 15. Yellowing of the human lens with age.
Literature Review and Gap Analysis 37 aging Drivers What We Know ⢠Age-related changes in the eye have the potential to cause issues for the older driver. ⢠The spectral responsivity of the eye with age (yellowing) can change the impact of a light source. What We Donât Know ⢠How the spectral changes in the eye change the visual performance and safety of an older driver in an LED-based lighting design. Results/Recommendations of This Study ⢠Age was considered and quantified in the research and is discussed in Chapter 5. Summary The data collected as part of this literature review were supplemented by the research per- formed as part of this project and described in this report. Recommendations for the implemen- tation of this analysis are included in the SSL Guide.
