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.
NCHRP Web-Only Document 146: Replacement Processes for Light Emitting Diode (LED) Traffic Signals 26 CHAPTER 4 CONCLUSIONS, RECOMMENDATIONS, AND SUGGESTED RESEARCH CONCLUSIONS Human Factors Issues With respect to human factors issues, the change in photometric requirements for red, yellow and green LED signal modules in the current ITE (2005) specifications from the earlier, interim ITE (1998) specifications appear to have been consistent with human visual performance and response to colored signal lights. The use of higher luminous intensity for yellow and green relative to red in the older specification (ITE, 1998) was probably originally based on the higher transmittance of the yellow and green cover lens relative to the red one for incandescent signal modules. Nonetheless, these colors do appear to require higher intensities than red to elicit equivalent reaction times, detection percentages, and perceptions of brightness (Fisher and Cole, 1975; Freedman, 2001; Bullough et al., 2000). The revision of the signal color boundaries between the previous (ITE, 1998) and current (ITE, 2005) LED specifications also appear to be sensible based on research describing the ability of color-deficient observers (protan and deutan) to detect and properly identify colored light signals (Huang et al., 2003; Cole, 2004). Evidence exists that LED traffic signals can create measurable discomfort glare when viewed at night (Bullough et al., 2001). Indeed, the current ITE (2005) specification for LED traffic signal modules permits dimming to levels as low as 30% of the daytime luminous intensity. No manufacturers were identified that incorporated this feature, which could also be used during periods of high ambient illumination from the sun to temporarily increase the luminous intensity of a signal module to overcome the sun phantom effect (Starr et al., 2004). Measurement and Monitoring Because the current (ITE, 2005) photometric requirements for LED traffic signal modules is a performance specification, improved ease for transportation agencies to monitor and/or measure signal performance would be beneficial to transportation agencies. One manufacturer developed signal modules with an on-board photosensor that could signal when the signal's luminous intensity dropped below minimum performance levels, in response to Louisiana Department of Transportation and Development (DOTD) requirement for such a sensor. However, the module was not compliant with the ENERGY STAR specifications for maximum power (required by the Energy Policy Act of 2005), and Louisiana DOTD no longer uses them (Urbanik, 2008). Hand-held devices that can be placed over a traffic signal module face to gather luminous flux and provide a relative light output measurement in the field are available, but are not widely used because of their cost and because they require a field technician to have access to the signal module face. Field measurement from off-road locations using a luminance meter can be made, but site conditions including the amount of ambient daylight, the specific measurement geometry, angular position and effects of wind or other factors will compromise the accuracy of such measurements.
NCHRP Web-Only Document 146: Replacement Processes for Light Emitting Diode (LED) Traffic Signals 27 As summarized in the previous chapter, low-precision measurements can be made using an illuminance meter in a dark room if the geometry between the signal module and the illuminance meter are very carefully controlled so repeatable measurements can be made. Such measurements require a long distance of a minimum of 6 m and preferably at least 12 m to ensure the inverse square law (Rea, 2000) can be applied. Such methods may be useful for tracking the performance of a small sample of working modules, but requires substantial effort to retrieve, measure and record data for each module. Nonetheless, this process may have value for transportation agencies working to estimate practical replacement intervals for LED signal modules. Until and unless a simple, reliable field measurement method can be developed, and assuming that self-monitoring products meeting ITE (2005) and ENERGY STAR specifications will not be forthcoming in the near future, guidance for transportation agencies regarding strategies for replacement would be useful. RECOMMENDATIONS FOR REPLACEMENT STRATEGIES As documented by Behura (2007) and by Urbanik (2008), few transportation agencies have a formal replacement strategy for LED traffic signal maintenance; when signal modules fail or are judged (usually based on visual inspection) to produce insufficient luminous intensity, they are replaced on an emergency or "spot" basis. For lighting systems with well-documented failure characteristics, such as incandescent and fluorescent lamps, published S-shaped "mortality curves" describing the percentage of lamps likely to fail under normal operating conditions as a function of rated life. Typically for these curves, there are very few lamp failures (<10%) until about 70% of lamp life is reached, after which lamp failures occur at a relatively linear rate until most of the lamps have failed (>90%) by the time 130% of lamp life has been reached (Rea, 2000). As described in the previous chapter, the failure mechanisms for LED traffic signal modules are quite different from those of conventional lighting systems, and often are unrelated to LED technology per se but rather control electronics and other factors. Failure can be caused by burned-out indications, or by luminous intensity reductions below the ITE (2005) specified minimum values. Documentation of the failure properties of LED traffic signal modules in the field is scarce. Bronson (2005) surveyed transportation agencies in California and respondents reported on average, that about 12% of signal modules failed over a five-year period. In the survey reported by Urbanik (2008), the median value for burned-out signal modules within a period of five years was close to 5%, and the median value for the acceptable percentage of modules that might produce luminous intensities below the ITE (2005) specification within this time frame was also close to 5%. Assuming that such failures would be distributed equally throughout the five-year warranty period, these results suggest that an annual failure rate of around 2% might be a reasonable value to expect in the field, at least as a preliminary estimate.
NCHRP Web-Only Document 146: Replacement Processes for Light Emitting Diode (LED) Traffic Signals 28 Group Replacement Cost Sensitivity Analysis The primary value of group replacement strategies in comparison with spot replacement strategies is the potential for reduced cost of labor for the replacement process. This is because time to travel to and from an intersection, to set up temporary traffic control, can be needed only once if all of the signal modules at that intersection will be replaced. This can also be performed during regular work hours when field technicians have access to vehicles and equipment for replacement. Spot replacement can require greater costs because many replacements are needed outside of regular work hours, and workers may have to travel to a maintenance facility to get needed equipment and vehicles. In the present series of analyses, the effect of group replacement is considered using different assumptions about failure rates, in comparison to a spot replacement strategy. The following assumptions are made: ⢠A transportation agency is responsible for 100 signalized intersections, each with an average of 10 LED signal modules (for a total of 1000 LED signal modules). ⢠The material cost of an LED signal module is assumed to be $75. ⢠During work hours, it is assumed that 15 minutes of travel to and from the intersection and 5 minutes to set up traffic control is required, per trip to an intersection. ⢠Outside work hours, it is assumed that 30 minutes of travel to and from the intersection and 5 minutes to set up traffic control is required, per trip to an intersection. ⢠Removal and replacement of a single LED signal module is assumed to require 9 minutes. ⢠Labor costs for a two-person crew during work hours are $100/hour and outside work hours are $150/hour. Using the assumptions listed above, the cost to replace all 1000 of the LED signal modules on a spot basis, where workers would need to travel and set up traffic control for each signal module, would be $110,500 for labor and $75,000 for materials, or $185.50 per module replaced. In comparison, the cost to replace all 1000 LED modules on a group replacement basis (where 10 modules could be replaced during a single trip to an intersection) would be $18,700 for labor and $75,000 for materials, or $93.70 per module replaced. Of course, even with a group replacement strategy, some LED signal modules will fail before they would otherwise be replaced and therefore, spot replacement of some modules will always be necessary even with a group replacement strategy. If the planned replacement period is too long, the bulk of replacement costs may be spot replacements with little savings associated with group replacement. If the planned replacement period is too short, costs will increase
NCHRP Web-Only Document 146: Replacement Processes for Light Emitting Diode (LED) Traffic Signals 29 because of the lost opportunity created by disposing of many otherwise functional modules that could have operated for longer periods. To assess appropriate periods for group replacement, the total long-term cost of replacing LED signal modules was assessed using two different assumptions for the expected life of LED signal modules: 7 years (City of Little Rock, 2003; SEDA-COG, 2008) and 10 years (Urbanik, 2008); and for several different failure rates: an S-shaped curve matching the failure profile for both incandescent and fluorescent lamps (Rea, 2000), and constant failure rates corresponding to 1%, 3% and 5% failures per year (some higher and lower than the 2% per year figure described above). Spot and group replacement costs were assessed, using replacement periods of 4, 5 and 6 years when a 7-year expected life is assumed, and replacement periods of 6, 7, 8 and 9 years when a 10-year expected life is assumed. Tables 6 and 7 show the long-term annual replacement costs under each scenario. Only labor and material costs for LED signal module replacement are included; it is assumed that the energy costs would be the same for spot and group replacement since all assume LED signal modules. Table 6. Long term annual costs for spot and group replacement strategies for an agency responsible for maintaining 100 signalized intersections, assuming a 7-year expected life. Shaded cells indicate when group replacement is estimated to be more costly than spot replacement. Expected Life (years) Long-Term Annual Spot Replacement Cost Planned Replacement Period (years) Expected Failure Rate Long-Term Group Replacement Cost 7 years $26,500 4 years S-shaped $25,400 1%/year $25,200 3%/year $28,900 5%/year $32,500 5 years S-shaped $23,400 1%/year $20,700 3%/year $24,400 5%/year $28,100 6 years S-shaped $24,300 1%/year $17,400 3%/year $21,100 5%/year $24,785
NCHRP Web-Only Document 146: Replacement Processes for Light Emitting Diode (LED) Traffic Signals 30 Table 7. Long term annual costs for spot and group replacement strategies for an agency responsible for maintaining 100 signalized intersections, assuming a 10-year expected life. Shaded cells indicate when group replacement is estimated to be more costly than spot replacement. Expected Life (years) Long-Term Annual Spot Replacement Cost Planned Replacement Period (years) Expected Failure Rate Long-Term Group Replacement Cost 10 years $18,600 6 years S-shaped $17,300 1%/year $17,500 3%/year $21,200 5%/year $24,900 7 years S-shaped $16,300 1%/year $15,200 3%/year $18,900 5%/year $22,700 8 years S-shaped $16,500 1%/year $13,600 3%/year $17,300 5%/year $21,000 9 years S-shaped $17,600 1%/year $12,300 3%/year $16,000 5%/year $19,700 In general, as expected, the estimated costs are lower for signal modules with a longer expected life. For spot replacement strategies, the cost of replacement is inversely related to the expected life. And for both expected lifetimes, if the expected number of failures per year is less than 3%, a planned replacement period of about 80% of expected life will reduce replacement costs relative to spot replacement. If the observed failure rate exceeds 3%, an agency should consider group replacement only very cautiously, as costs might exceed those from a simple spot replacement strategy. Further cost savings might be achieved in a group replacement program if a photometric or visual assessment of modules removed from service could be performed. Under such a program, modules could be measured using a technique such as the illuminance test method summarized in the previous chapter, or simply assessed visually in comparison to a new module of the same color and type. Modules still exceeding the ITE (2005) recommendations, or those appearing similar in brightness appearance to a new signal module, could be set aside and used for spot replacement for signal modules, especially those that would undergo group replacement within two or three years. An agency might set aside the top 10% of measured or visually assessed modules for this purpose.
NCHRP Web-Only Document 146: Replacement Processes for Light Emitting Diode (LED) Traffic Signals 31 SUGGESTED RESEARCH The review of research and the results of the activities conducted for the present project lead to several suggestions for research activities that could be undertaken to help transportation agencies develop maintenance programs: ⢠Refinement of sensor systems for reliably detecting when a traffic signal module does not produce the luminous intensity recommended by the ITE (2005) performance specifications. The field and laboratory measurement procedures described in the previous chapter of this report have focused mainly on the light output near the direction of maximum luminous intensity. Of course, the ITE (2005) recommendations require a luminous intensity distribution at wider angles in order to ensure its visibility at more oblique angles of view. Some LED signal modules with visible LED arrays can display a partial indication if some of the LEDs fail, and this could affect luminous intensity in one direction while having little influence in another direction. Thus, multiple sensors, each able to characterize light output in a different directional region, might be appropriate for implementation in a traffic signal module, if the power requirements of such systems do not conflict with ENERGY STAR requirements. ⢠Mechanisms for appropriate dimming of LED traffic signal modules to reduce glare and provide modest energy savings could be implemented, and perhaps even to temporarily increase signal intensity during conditions of reduced visibility such as fog or sun phantom. These could use similar sensors as those that might monitor useful life of LED signal modules. ⢠LED traffic signal module manufacturers might consider modifying product designs to address the several common failure modes that were identified in the present study: simplifying start-up circuitry, selecting robust circuitry components, different grouping of LEDs in arrays, and relocating high-power electrical components likely to produce excess heat. Indeed, the analysis of similar products with different dates of manufacture suggests that such improvements are already underway. ⢠Improved characterization of long-term performance and reliability of LED traffic signal modules is necessary in order to determine with precision the optimum replacement strategies. The present analyses suggest that group replacement does appear to have cost benefits over spot replacement under many circumstances, but identifying the optimal replacement intervals will depend upon the life and reliability characteristics of LED traffic signal modules. Transportation agencies should consider developing programs to collect long-term performance data for their LED traffic signal modules, even on a limited sample basis. ⢠While not within the scope of the present project, compatibility between LED traffic signal modules and the control equipment used to operate them, primarily based on operation of incandescent traffic signals, should be improved, as identified by Urbanik (2008). Retrofit solutions, which attempt to make LED systems behave like
NCHRP Web-Only Document 146: Replacement Processes for Light Emitting Diode (LED) Traffic Signals 32 incandescent systems so that they will be compatible with older control equipment, have been demonstrated to have compatibility and reliability problems and should be addressed in future research. In summary, LED traffic signal modules have largely lived up to their promise for increasing the energy-efficiency of the nation's traffic signal system and have most likely reduced the costs of labor associated with traffic signal maintenance. Generally speaking, the current specifications from the ITE (2005) for photometric and colorimetric performance of traffic signals are sufficient to ensure sufficient visual performance in terms of reaction times, detection probability, and color identification for color-normal and many color-deficient observers. A number of failure modes for LED traffic signal modules have been identified and can be addressed by improvements in designs. As described in the previous chapter, LED signal module manufacturers have modified product designs in the past to solve some of the problems identified in the failure analysis conducted for this study. Solutions to many of the issues identified presently may be well on their way to being implemented, but a large base of installed signals using several of the earlier product designs exists and will need to be dealt with by transportation agencies. Finally, the cost analysis of group versus spot replacement strategies suggests that, subject to some uncertainty in the failure rates for LED traffic signal modules, group replacement, combined with some recovery of replaced modules for the necessary spot replacements that might occur even with a group replacement program, can reduce agency costs. A replacement period of around 8 years if a 10-year operating life is expected, or of around 6 years if a 7-year operating life is expected, will probably reduce overall long-term replacement costs. In the former case, about 12% of signal modules could be replaced each year; in the latter, about 17% would be replaced each year.
