Drivers Evaluation of Performance of LED Traffic Signal Modules

Civil Engineering Sudies Transporaion Engineering Series No. 120 Traffic Operaions Lab Series No. 5 UILU-ENG-2002-2010 ISSN-0917-9191 Drivers Evaluaion of Performance of LED Traffic Signal Modules By Rahim F. Benekohal Madhav V. Chiuri A sudy conduced by Traffic Operaions Laboraory Deparmen of Civil and Environmenal Engineering Universiy of Illinois a Urbana-Champaign Prepared for Illinois Deparmen of Transporaion December 2002

Technical Repor Documenaion Page 1. 1.. Repor No, FHWA-IL/UI-TOL-5 2. Governmen Accession No. 3. Recipien's Caalog No 4. 4. Tile and Subile Drivers Evaluaion of Performance of LED Traffic Signal Modules 5. Repor Dae December 11, 2002 6. Performing Organizaion Code 7. 7.Auhor(s) 8. Rahim F. Benekohal and Madhav V. Chiuri 8. Performing Organizaion Repor No. UILU-ENG-2002-2010 9. Performing Organizaion Name and Address 10. Work Uni (TRAIS) Deparmen of Civil and Environmenal Engineering Universiy of Illinois a Urbana-Champaign 205 N. Mahews Ave. Urbana, Illinois 6180 11. Conrac or Gran No. 13. Type of Repor and Period Covered Projec Repor 2001-2002 12. Sponsoring Agency Name and Address The Illinois Deparmen of Transporaion 14. Sponsoring Agency Code 15. 15. Supplemenary 16. Absrac This sudy evaluaed he performance of he LED modules from four manufacurers (Leoek, Gelcore, Dialigh and Precision Solar) who provided he required number of LEDs for esing and evaluaion a he Traffic Operaions Lab. The crieria for evaluaion were drivers percepion of brighness, doiness, and color compared o incandescen lenses. For circular red indicaion, brighness, doiness, and color of Dialigh LED module was similar o incandescen lens, bu he oher hree were brigher, more doier, and darker han incandescen lens. For circular yellow indicaion, brighness of Precision Solar LED was he same while he ohers were brigher han incandescen lens; hey all were doier and darker (orange-yellow) han incandescen lens. For circular green indicaion all he LED modules were brigher, doier, darker (bluish) han incandescen lens. For yellow arrow indicaion, Leoek and Dialigh are he same while Gelcore and Precision Solar were brigher han incandescen lens; all he LED modules were doier and darker han incandescen lens. For green arrow indicaion, Leoek was he same as incandescen while Dialigh, Gelcore and Precision Solar were brigher han incandescen lens; all he LED modules were doier and darker han incandescen lens. The findings, hough limied, indicaed ha here could be a significan difference in he percepion of he LED modules by he older drivers. I is recommended o furher sudy he effec of aging on he percepion of he LED modules. 17. Key Words Ligh Emiing Diodes, LED Traffic Signal Modules, Traffic Signal, Signal Indicaor, Driver Percepion, Inersecion Traffic Lighs 18. Disribuion Saemen 19. Securiy Classif. (of his repor) Unclassified 20. Securiy Classif. (of his page) Unclassified 21. No. of Pages 73 22. Price `Form DOT F 1700.7 (8-72) Reproducion of compleed page auhorized ii

Acknowledgmen and Disclaimer This sudy was conduced by he Traffic Operaions Laboraory (TOL) a he Universiy of Illinois a Urbana-Champaign. The Illinois Deparmen of Transporaion sponsored he sudy. The conens of his repor reflec he views of he auhors who are responsible for he facs and accuracy of he daa presened herein. The conens do no reflec he official views or policies of he Illinois Deparmen of Transporaion. This repor does no consiue a sandard, specificaion, or regulaion. iii

Table of Conens 1. INTRODUCTION... 1 1.1. LEDS VS INCANDESCENT LIGHT SOURCES... 2 1.2. LEDS IN TRAFFIC SIGNALS... 3 2. LITERATURE REVIEW ON BENEFIT - COST ANALYSES... 6 2.1. OREGON S EXPERIENCE... 6 2.2. MISSOURI S EXPERIENCE... 7 2.3. DENVER S EXPERIENCE... 8 2.4. LONDON STUDY... 8 3. OBJECTIVES... 10 4. DATA COLLECTION, REDUCTION & METHODOLOGY... 11 4.1. DATA COLLECTION... 11 4.2. DATA REDUCTION... 13 4.3. METHODOLOGY... 14 5. STUDY FINDINGS... 16 5.1. CIRCULAR RED... 16 5.1.1.Brighness... 16 5.1.2. Doiness... 17 5.1.3. Color... 18 5.2. CIRCULAR YELLOW... 20 5.2.1. Brighness... 20 5.2.2. Doiness... 21 5.2.3. Color... 22 5.3. CIRCULAR GREEN... 23 5.3.1. Brighness... 23 5.3.2. Doiness... 24 5.3.3. Color... 25 5.4. YELLOW LEFT ARROW... 26 5.4.1. Brighness... 26 5.4.2. Doiness... 27 5.4.3. Color... 28 5.5. GREEN LEFT ARROW... 29 5.5.1. Brighness... 29 5.5.2. Doiness... 31 5.5.3. Color... 32 6. CONCLUSIONS AND RECOMMENDATIONS... 34 6.1. CIRCULAR RED... 34 6.2. CIRCULAR YELLOW... 34 6.3. CIRCULAR GREEN... 35 6.4. YELLOW LEFT ARROW... 35 iv

6.5. GREEN LEFT ARROW... 35 6.6. EFFECT OF AGE... 36 7. REFERENCES... 37 APPENDIX A... 38 APPENDIX B... 41 CIRCULAR RED... 42 CIRCULAR RED... 43 CIRCULAR RED... 44 CIRCULAR RED... 46 CIRCULAR YELLOW... 48 CIRCULAR YELLOW... 49 CIRCULAR YELLOW... 50 CIRCULAR YELLOW... 52 CIRCULAR GREEN... 54 CIRCULAR GREEN... 56 CIRCULAR GREEN... 58 CIRCULAR GREEN... 59 YELLOW LEFT ARROW... 60 YELLOW LEFT ARROW... 62 YELLOW LEFT ARROW... 64 GREEN LEFT ARROW... 66 GREEN LEFT ARROW... 67 GREEN LEFT ARROW... 68 GREEN LEFT ARROW... 69 GREEN LEFT ARROW... 70 APPENDIX C... 72 v

1. INTRODUCTION Nearly everyone is familiar wih Ligh Emiing Diodes (LEDs) from heir use as indicaor lighs and numeric displays on consumer elecronic devices. Ligh Emiing Diodes, which were developed in he lae 1960s are solid-sae semi-conducor devices ha conver elecrical energy direcly ino ligh. Figure 1[Ref. 1] shows a schemaic of a ypical T 1-3/4 LED (T-1 ¾ is a ype of LED lamp package. Type T-1 3/4= 0.2188 inch lamp diameer or a 5.56mm lamp diameer.). Figure 1. Schemaic of a ypical T 1-3/4 LED [Ref. 1]. The ligh-generaing chip iself is very small ypically 0.25 millimeers square, wih res of he volume being occupied by he plasic encapsulan and he lead frame. Ligh is generaed inside he chip when curren flows across he juncions of differen maerials. The wavelengh of he ligh generaed and herefore he color of he ligh emied depends on he composiion of he maerials in he chip. Ligh emiing chips in LEDs are essenially layers of differen semiconducor maerial grown on op of one anoher. They are manufacured using a process called epiaxy. Advances in epiaxial crysal growh processes have made i possible o 1

manufacure LED maerials wih high enough puriy o produce colors ha were no possible previously. Figure 2[Ref. 2] summarizes he performance of differen LED echnologies as hey have evolved over ime. The dark line saring from mid 1960s summarizes he seady increase in he LED performance and his can be compared wih he performance of several convenional lighing devices, indicaed by arrows, near he verical axis. In he mid-1990s high-performance, niride-based LEDs, were developed and his is indicaed by he line labeled Niries which sprous a branch. The lines labeled Molecular solids and Polymers show he improving performance of Organic LEDs. Figure 2. Growh in performance of LEDs over ime [Ref. 2] Presenly LEDs are available ha can produce ligh spanning he visible specrum. AlInGaP (Aluminum Indium Gallium Phosphide) maerial sysems produce red, orange, and yellow ligh while AlInGaN (Aluminum Indium Gallium Niride) sysems generae green, blue, and near-uv ligh. 1.1. LEDs Vs Incandescen ligh sources Almos 100% of he energy radiaed by LEDs lies wihin he visible specrum. Bu for an incandescen source 2% of he emied energy is in he visible region wih he res 98% being wased as hea. Owing o he high ligh oupu efficiency an LED source can 2

generae ligh of same inensiy as an incandescen source while consuming significanly less elecrical power. Ligh emied by he LEDs falls in very narrow specral bandwidhs, hereby producing nearly monochromaic ligh and does no need o be filered o produce he required colors. On he conrary, incandescen sources are broadband emiers of whie ligh ha mus be opically filered o achieve he desired color. The ligh oupu of LED modules degrades over ime. Degradaion of LED ligh oupu is a direc funcion of he juncion emperaure. However for any given lengh of ime, he ligh oupu degradaion of for an LED device is significanly less han ha of an incandescen lamp [Ref. 3]. LEDs are durable and provide longer lamp life han incandescen sources. 1.2. LEDs in Traffic Signals As can be seen in Figure 2, advances in ligh emiing diode (LED) echnologies have made i possible o produce LEDs ha emi colors hroughou he visible specrum and have beer performance han convenional echnologies. This has made i possible o use LEDs in a variey of lighing applicaions. Manufacurers have developed raffic signal opical modules using LEDs. According o Sraegies Unlimied, a marke research organizaion ha conducs an annual marke updae of LED lighing producs, abou 10% of all red raffic signals in he Unied Saes used LEDs in 2000. In early signals, more han 600 individual LEDs migh be mouned ogeher in a disk formaion o produce he circular indicaion, bu hese signals did no mee he requiremens for disribuion of luminous inensiy. The addiion of lenses in fron of he LEDs and increases in LED ligh oupu brough heir number from 600 o around 200. More recenly, very high oupu LEDs wih broader inensiy disribuions have been developed, and in many ways, he design of LED raffic signals has begun o mirror convenional incandescen signals, wih a clusered LED ligh source se back ino a chamber and opical elemens divering he ligh from his source in he proper disribuion. [Ref. 4] Figure 3 [Ref. 4] shows a schemaic of he various LED raffic signal modules. 3

Figure 3. Top lef: Incandescen raffic signals have been in use for many decades. Top righ: Early LED signals used hundreds of ighly packed LEDs. Cener lef: Wih improved opics, fewer LEDs could be used. Cener righ: High oupu LEDs resul in signals ha appear similar o incandescen signals. Boom: Head-on view of raffic signals, saring op lef and going couner-clockwise [Ref. 4]. There are approximaely 260,000 inersecions wih raffic signals in he US. Each inersecion has an average of 40 signal indicaions using eiher 69, 135, or 150 was incandescen lamps. Though, he incandescen lamps are very inexpensive, heir use in he raffic signals is a considerable economic burden for he governmenal agencies because of heir high power consumpion and annual prevenaive lamp replacemen. On he conrary, LED based raffic signals offer remendous savings poenial because hey consume much less power han he incandescen lenses for producing he 4

same oupu and do no require he annual prevenive replacemen due o heir long life. According o an esimae [Ref. 5] Ligh-emiing diodes (LEDs) could save nearly 2.5 billion kilowa hours annually, if hey replaced incandescen lenses in raffic signals. Thus he use of LEDs in raffic signals resuls no only in energy cos savings for he governmenal agencies, bu also helps he environmen by using less power for operaing he signals. In he following secion a lieraure review of he benefi-cos analyses ha were published is presened. 5

2. LITERATURE REVIEW ON BENEFIT - COST ANALYSES In his secion a summary of four benefi-cos analyses of he use of LED modules in raffic signals ha have been documened in he lieraure is presened. Summaries of he four sudies followed by our commens are presened in four subsecions o follow. However, i should be noed ha he conclusions reached in hese sudies are valid only for he projecs under consideraion and are no necessarily applicable for any oher projec. This is primarily because of he difference in: The benefis and coss being considered: In each subsecion he facors ha have been considered in he paricular sudies are menioned and i can be seen ha he facors ha have been considered are differen for differen sudies. Cos of he LED modules: The coss of he LED modules have been decreasing over he years and he coss ha have been used in hese sudies would be oo high oday. 2.1. Oregon s Experience The Oregon sudy [Ref. 6] raced he hisory of inroducion of LED raffic signals in Oregon by Oregon Deparmen of Transporaion (ODOT) afer esing red LED modules and finding hem o be reliable. In 1995, ODOT sared an LED raffic signal implemenaion program in conjuncion wih 8 ciies in Oregon (only Red 12 balls, Red 8 balls, Red 12 arrows and Pedesrian hand symbols were replaced). The repor presens an economic analysis comparing LED and incandescen raffic signals. In he economic analysis, he facors considered were Lamp Cos Energy cos I was esimaed ha for he ODOT projec, he annual energy savings (considering only red LEDs) would be 88% and afer facoring in he lamp coss would resul in annual savings of 26%. The esimaed payback period for he projec was 3.6 years. However i should be noed ha he LEDs ha were used in he Oregon sudy were based on Aluminum Gallium Arsenide (AlGaAs) echnology while he presen LEDs use Aluminum Indium Gallium Phosphide (AlInGaP) echnology. This economic analysis 6

does no consider he iniial cos of replacing incandescen lenses wih LEDs and mainenance savings due o he use of LED modules. Also, he daa used for esimaing he energy coss is based on echnical daa shees and no on acual meered energy usage. For he Oregon projec, hrough he Federal Highway Adminisraion (FHWA) Prioriy Technology Program, he paricipaing agencies were reimbursed 30% of he cos of he lenses and in addiion, elecric uiliy companies provided energy rebaes o he paricipaing agencies. However, in he Benefi-Cos analysis performed, he acual cos of he lenses was used and hese rebaes/reimbursemens were no considered. 2.2. Missouri s Experience The prime objecive of he Missouri sudy [Ref. 7] was o develop an evaluaion mehodology ha would provide daa comparing LED signal heads o incandescen signal heads. Towards his objecive, all he red and green incandescen indicaions (a oal of 41) a he inersecion of US Roue 50-63 and Missouri Boulevard in Jefferson Ciy, Missouri were replaced wih ligh emiing diode (LED) signal indicaions. The evaluaion procedure consised of collecing daa ha included insallaion and maerial coss for he LED signals as compared o similar incandescen bulb insallaions, energy usage and cos comparisons of each ype of insallaion, and mainenance and repair coss of he LED insallaion compared o he previous incandescen insallaions. From his daa signal life cycle coss were compued and life cycle cos analysis was performed. The Life Cycle Cos Analysis (LCCA) gave an Incandescen Bulb/LED raio of 0.97 based on 1997 coss of LED signal heads. However he raio was 1.36 when esimaed coss of he LED signals in 1999 were used in he analysis. I is also repored ha he raio would be much higher for oher inersecions because his inersecion is locaed close o he Cenral Disric Headquarers, here by having a smaller mainenance cos. This sudy clearly highlighs he economic benefis of he use of LED signals in lieu of incandescen signals by using acual field daa for a paricular inersecion. For compuing he energy savings, acual meered usage of he inersecion over several monhs of operaion wih LED signals was used. During he course of he sudy, one of 7

he lef urn LED signals was seriously damaged in a lighning sorm and had o be replaced. The resuling repair cos was included in he repair coss for LEDs. However, he cos of he replacing he LED signal was no included as ha cos was borne by he vendor under warranee. The life cycle of he LED lamp was considered o be 7 years and he cash flows were discouned while performing he analysis. 2.3. Denver s Experience This sudy [Ref. 8] provides an esimae of he payback period for rerofiing all he 1200 inersecions of Denver wih LED raffic signals (red 8 and 12 ball, arrow, pedesrian hand only). The facors considered in he analysis were acquisiion cos, insallaion cos, ancillary maerials cos (cos of incandescens for yellow, green and whie which were group relamped in he rerofi projec), energy savings, reduced replacemen maerial savings, labor savings due o reduced mainenance. The average simple payback period for he projec was esimaed o be 4.28 years. The ne acquisiion cos for he complee program was $ 1,333,225 and represens 17,036 unis. Denver realized a oal rebae of $405,900 for he insallaion of 14,036 unis before he expiraion of he Public Service Company of Colorado s (PSCO) Demand Side energy Managemen (DSM) program. The supplier gave a 1% credi for paymen wihin 31 days, which resuled in an addiional decrease of $ 17,556. This sudy does no discoun he cash flows. Bu, i should be noed ha of he hree sudies discussed so far, his sudy accouns for all he benefis and coss and is more realisic and could possibly serve as a model for benefi-cos analyses of fuure projecs. 2.4. London Sudy This sudy [Ref. 9] provides esimaes on he expeced savings if all raffic signals in London were replaced by LED modules. In his sudy, he auhors considered cos of he LED modules, insallaion cos of LED modules, elecriciy and mainenance savings. Based on he hen prices of LED modules and elecriciy coss, he auhors esimae ha 8

he payback period for ha projec would be 8 years for Red, 28 years for Yellow and 18 years for Green LED modules. I is ineresing o noe ha while he Oregon sudy indicaed a payback period of 3.6 years for heir projec ha involves red signals mosly, he London sudy indicaes a payback period of 8 years for red signals. This is because he incandescen lenses being used in London consume 50 was while he lenses ha are used in Oregon have a consumpion of 158 was and he LED modules in London and Oregon use 13 and 20 was, respecively. So he payback period for red lenses for London should be approximaely hree imes he payback period for Oregon. Bu he repored payback period is significanly lower han 10.8(3*3.6 years) because in he London sudy, he mainenance savings due o he use of LED modules are also considered while compuing he payback period. However, i should be noed ha he auhors assume ha he enire annual mainenance coss, which are being currenly incurred wih incandescen lenses will be saved once he swich o LED modules akes place. I is imperaive here will be some mainenance, no necessarily involving relamping, even when he raffic signals are convered o LED modules. I should also be noed ha he auhors do no consider any replacemen cos for LED modules over he payback period of he projec, which is quie high a 8 years for red, 28 years for Yellow and 18 years for green. 9

3. OBJECTIVES The Illinois Deparmen of Transporaion (IDOT) is considering replacing he incandescen raffic signal lamps on he sae highway sysem wih LED modules by requiring LED modules in he fuure raffic signal consrucion projecs and rerofiing he exising signals. The objecive of his sudy is o evaluae he performance of he LED modules from four manufacurers. The manufacurers are Leoek, Gelcore, Dialigh and Precision Solar. These manufacurers provided he required number of LEDs for esing and evaluaion a he Traffic Operaions Lab. The purposes of he evaluaion are: Idenify hose manufacurers whose LED modules are close o he incandescen lens in heir performance. Idenify subse(s) of manufacurers whose LED modules are saisically he same in heir performance. The crieria used for performance evaluaion includes: 1. Brighness: Brighness could have been measured phoomerically, bu he specral conen and spaial deail may differ beween differen echnologies and ha may influence he percepion of drivers. So, drivers were asked o compare he brighness of LEDs o incandescen indicaions. 2. Doiness: As he LED modules consis of srings of LEDs, hey end o appear like a cluser of dos which is no he case wih incandescen lens. Therefore, i is imperaive o know how drivers perceive he LED modules compared o incandescen lenses. 3. Color: Since he specral disribuions of differen lenses differ, i is required o ensure ha any lenses ha are used in he raffic signal heads resemble he incandescen lenses in color. So, drivers were asked o compare he color of LED modules o he corresponding colors in incandescen lenses. 10

4. DATA COLLECTION, REDUCTION & METHODOLOGY 4.1. Daa Collecion A quesionnaire was developed and daa collecion was conduced a he Traffic Operaions Laboraory a Universiy of Illinois. Four manufacurers lenses were mouned on signal heads. On a given signal head all he lenses were from he same manufacurer. Three signal heads of each manufacurer were used. The model numbers of he lenses used in he survey and heir power raings are shown in Appendix C. Two of he signal heads had only Red, Yellow and Green lenses while he hird one included a Yellow lef arrow and a Green lef arrow. Four frames were assembled and each frame had 3 signal heads. The signal heads of he differen manufacurers were arranged in such a way ha he signal heads of each manufacurer occupied lef, middle and righ posiions on differen frames. A ypical frame is shown in Figure 4. Figure 4. Frame D used in he Survey. 11

Observers were asked o si a four differen viewing saions and a each saion view one of he frames. The disance beween he viewing saion and frame was abou 70 fee. This represens approximaely he disance beween a sopped vehicle and he signal head a an inersecion wih 2 hrough lanes per direcion and a lef urn pocke. The viewing saions and he frames were clearly marked o avoid any confusion. Observers moved from one saion o anoher saion afer compleing he survey for ha frame. The survey quesionnaire requesed he drivers o rae he performance of he LED signal lenses relaive o incandescen signal lenses in erms of brighness, solidness (as opposed o being a series of dos) and color on a scale of 1 o 5, and 0 referring o No Opinion. In order o avoid ambiguiy abou wha he raings 1 hrough 5 indicaed, descripions of each raing were also provided. For example, in he case of brighness, 1 indicaed much dimmer and 5 indicaed much brigher compared o incandescen lenses and 3 indicaed he same as incandescen lenses. The observers had valid driving licenses and were asked o idenify heir age group and indicae if hey needed o wear glasses or had any difficuly in disinguishing differen shades of a color. A copy of he Survey Quesionnaire is given in Appendix A. Thiry eigh drivers paricipaed in he survey. The demographics of he drivers are shown in Table 3.1. Group Age Group Frequency A < 25 0 B 25 40 14 C 40 50 12 D 50 60 11 E > 60 1 Table 3.1. Demographics of he survey paricipans. The LED signal heads were operaed using a NEMA Conroller. The cycle lengh was abou 93 seconds and here were four phases. Yellow imes were made longer han normal so ha he observers had enough ime o view hem. Some observers reviewed he lenses more han once, before filling in heir responses. 12

4.2. Daa Reducion In an Excel Workbook, hree spreadshees were made corresponding o he hree quesions in he survey quesionnaire. The daa from he 38 surveys was coded in spreadshees. However, afer ransferring he survey responses o he Excel file, i was observed ha no all drivers responded o all he quesions in he survey. Therefore, for each quesion in he survey (peraining o he hree aspecs i.e., brighness, doiness and color) he subse of drivers who gave hree raings for he lenses of all he LED manufacurers were idenified. The average raing for each manufacurer s LED modules was calculaed. These average raings of he drivers were used o perform saisical ess and conclude abou he performance of he LED modules. To check for he normaliy of he daa, hisograms of he average raings of he drivers were ploed and visually inspeced for normaliy. As he daa is caegorical, we could only check o see ha here exiss a symmeric curve wih a peak a he cener. To ensure ha drivers were consisen in he responses hey gave o he hree lenses of he same color from a given manufacurer, he range of each driver s responses o each of he manufacurer s LED modules was calculaed. I was assumed ha a range of one was reasonable and did no consiue an inconsisency in he raing. Therefore, all drivers who had ranges greaer han or equal o wo for any of heir raings were idenified. For each such driver, he sum of he ranges for each of he quesions in he survey was compued. The drivers who had a sum of ranges greaer han 12 were idenified and are lised in Table 3.2. Those drivers who had a sum of ranges greaer han weny four were no included in he analysis because a sum of weny four implies ha on he average heir range was wo for all he raings hey gave. Therefore drivers 28 and 34 were no included in he analysis for color. From Table 3.2 we can also observe ha all he drivers who had higher ranges, excep for driver 10 are aged above 50. This migh indicae ha he older people are more inconsisen han he younger people in heir raings of he LED modules. 13

Driver # Age Group Color Problems Glasses 9 50 60 No Yes Aspec Sum of Ranges Brighness 16 Doiness 13 10 40 50 No Yes Doiness 18 26 50 60 No Yes Brighness 15 27 > 60 No Yes Brighness 12 Doiness 15 28 50 60 No No Color 25 34 50 60 No No Color 29 38 50 60 No No Brighness 12 Color 13 Table 3.2. Drivers wih inconsisen responses. 4.3. Mehodology The mehodology used in his sudy was asking drivers opinions on he performance of LED modules compared o incandescen lens. To achieve he objecives of his sudy, ess (one ailed) were conduced o compare he performance of each manufacurer s LED module wih incandescen lens and paired -ess were conduced o compare each manufacurer s LED modules wih every oher manufacurer s LED module. Also, in order o verify, if he age of he drivers affecs he raing of he LED modules, he responses were classified ino wo age groups viz., below 50 years and above 50 years. Considering hese wo daa ses as independen samples, independen samples -es was performed. Equaliy of variances of he wo groups was firs esed and if hey were saisically equal, regular es wih pooled variance was used oherwise Welch s es was performed. Furhermore, o check for he effec of wearing glasses, he responses were classified ino wo groups viz., drivers wih glasses and wihou glasses. Considering hese wo daa ses as independen samples, independen samples es was performed. 14

Equaliy of variances of he wo groups was firs esed and if hey were saisically equal, regular es wih pooled variance was used oherwise Welch s es was performed. An α value of 0.10 was used in all he saisical ess. 15

5. STUDY FINDINGS Five ypes of LED raffic signals (Circular Red, Circular Yellow, Circular Green, Yellow lef arrow and Green lef arrow) were evaluaed. The resuls for hem are presened in five separae secions o follow. Each of hese secions is furher divided ino hree subsecions corresponding o he hree quesions in he survey quesionnaire (brighness, doiness, and color). Each subsecion conains a discussion on he following iems: Performance of each manufacurer s lens wih respec o incandescen lens Relaive performance of he manufacurer s lenses Effec of age on he responses Effec of wearing glasses on he responses 5.1. Circular Red 5.1.1.Brighness Thiry hree drivers gave hree raings for brighness of all he manufacurer s LED modules and he sample saisics are presened in Table 5.1. T ess were performed o find ou if he average raings were saisically differen from 3.00. A raing of 3.00 indicaes ha he LED modules have he same brighness as he incandescen lenses and a higher raing indicaes ha LED modules are brigher. The resuls of he ess, shown in Table 5.1, indicae ha Dialigh is saisically same as incandescen lens in brighness and all he ohers are brigher han incandescen lens. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 33 3.75 0.63 6.82 2.733E-08 Rejec 33 4.15 0.57 11.69 1.241E-16 Rejec Brigher han Inc Brigher han Inc 33 2.92 0.79-0.59 2.795E-01 Don' Rejec Saisically same 33 4.01 0.59 9.90 2.787E-16 Rejec Table 5.1.Resuls of -ess for brighness of Circular Red Brigher han Inc 16

To idenify he LED modules which are saisically no differen from each oher in heir brighness, paired -ess of he six ( 4 C 2 ) possible combinaions were performed. The resuls are shown in Table B.1 in Appendix B. The resuls indicae ha he decreasing order of brighness is Gelcore, Precision Solar, Leoek and Dialigh. However, i should be noed ha Gelcore would no be brigher han Precision Solar if he confidence level were raised o 95%. Of he 33 drivers who gave hree raings for all he manufacurer s LED modules, 21 had glasses and 10 did no. Two oher drivers did no indicae if hey used glasses or no. I was found ha he variances of he wo groups were saisically he same for all he LED modules. Saisically (resuls shown in Table B.2 in Appendix B) no difference was found beween he raings given by he wo groups of drivers for any of he manufacurer s lenses. When he raings of he older drivers (over 50 years) were compared (resuls shown in Table B.3 in Appendix B) wih he raings of he younger drivers (under 50 years), i was found ha for all he LED modules he average raings of he wo groups were saisically he same. Also for Leoek he variances for he wo groups were found o be saisically differen and consequenly Welch s es was used for Leoek. 5.1.2. Doiness Thiry four drivers gave hree raings for doiness of all he manufacurer s LED modules and he sample saisics are presened in Table 5.2. T ess were performed o find ou if he average raings were saisically differen from 3.00. A raing of 3 indicaes ha he LED module has he same doiness as he incandescen lens and a higher raing indicaes ha LED module is doier han incandescen lens. The resuls shown in Table 5.2 indicae ha Dialigh is less doier han an incandescen lens while all he ohers are more doier han incandescen lens. However, i should be noed ha for Dialigh comparison, he confidence level is 96% while he confidence level for he oher comparisons is almos 100%. This means ha, if we increase our confidence level o 97%, we canno say ha Dialigh is less doier han incandescen lens. 17

Manufacurer N Sd. Deviaion (1-ailed) Inferenc e Conclusion LEOTEK 34 3.94 0.55 10.03 1.067E-16 Rejec 34 4.46 0.52 16.23 1.067E-16 Rejec 34 2.80 0.62-1.84 3.731E-02 Rejec 34 3.61 0.52 6.88 1.906E-08 Rejec Table 5.2.Resuls of -ess for doiness of Circular Red More Doier More Doier Less Doier More Doier The resuls of he six paired -ess are shown in Table B.4 in Appendix B. The resuls indicae ha he descending order of doiness is Gelcore, Leoek, Precision Solar and Dialigh. Saisically no difference was found (resuls shown in Table B.5 in Appendix B) beween he raings of drivers wih glasses and wihou glasses for all he four brands. The raings of he wo age groups for Gelcore were saisically differen a 95% confidence level, wih he older age group giving a higher raing han he younger age group indicaing ha older age group found i more doier han he younger. The Gelcore raings of he wo groups were individually compared wih 3.00 (incandescen raing) and i was found ha boh he groups found i o be saisically doier han incandescen. For he oher hree brands he raings of he wo groups were saisically he same (resuls shown in Table B.6 in Appendix B). 5.1.3. Color Thiry hree drivers gave hree raings for color of all he manufacurer s LED modules and he sample saisics are presened in Table 5.3. T ess were performed o find ou if he average raings were saisically differen from 3.00. A raing of 3 indicaes ha he LED module has he same color as he incandescen lens and a higher raing indicaes 18

ha LED module is darker han incandescen. The resuls are shown in Table 5.3. The p values srongly indicae ha Dialigh is saisically he same as incandescen lens while all he ohers are darker han incandescen lens. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 33 3.35 0.52 3.91 2.267E-04 Rejec 33 3.37 0.45 4.73 2.205E-05 Rejec Darker han Inc Darker han Inc 33 2.98 0.53-0.22 4.141E-01 Don' Rejec Sa same 33 3.33 0.41 4.68 2.496E-05 Rejec Table 5.3.Resuls of -ess for color of Circular Red Darker han Inc The resuls of he six paired -ess are shown in Table B.7 in Appendix B. The resuls indicae ha he Leoek, Gelcore and Precision Solar are saisically same in color and all of hem are darker han Dialigh. However, i should be noed ha Gelcore and Precision Solar would no be saisically he same if he confidence level were lowered o 85%. Saisically no difference was found beween he raings of drivers wih glasses and wihou glasses (resuls shown in Table B.8 in Appendix B) for Leoek, Gelcore and Precision Solar. Bu for Dialigh he wo groups raings were differen. Also i should be noed ha on he average drivers wih glasses said Dialigh is paler han incandescen while he drivers wihou glasses said ha i is darker han incandescen. The raings of he wo groups were individually compared wih 3.00 (incandescen raing) and i was found ha hey were saisically he same as incandescen, bu he associaed confidence levels are very close o he confidence level of 90%. The raings of he wo age groups were compared (resuls shown in Table B.9 in Appendix B) and i was found ha for Dialigh and Precision Solar hey are saisically same. However, for Leoek and Gelcore hey were differen. In boh he cases, he older group gave a higher raing han he younger group indicaing ha older group found hem 19

darker han he younger. The Leoek and Gelcore raings of he wo groups were individually compared wih 3.00 (incandescen raing) and i was found ha all he groups found i o be saisically darker han incandescen. I should also be noed ha for Leoek he wo groups were found o have saisically differen variances. Consequenly for Leoek Welch s es was used. 5.2. Circular Yellow 5.2.1. Brighness Tweny hree drivers gave hree raings for brighness of all he manufacurer s LED modules and he sample saisics are presened in Table 5.4. T ess were performed o find ou if he average raings were saisically differen from 3.00. The resuls shown in Table 5.4 indicae ha Precision Solar is saisically same as incandescen lens in brighness while all he ohers are brigher han incandescen lens. However, i should be noed ha Precision Solar would no be same as incandescen lens if he confidence level were lowered o 87%. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 23 3.81 0.65 5.99 2.505E-06 Rejec 23 3.43 0.62 3.35 1.460E-03 Rejec 23 3.26 0.47 2.66 7.207E-03 Rejec 23 3.16 0.64 1.19 1.234E-01 Don' Rejec Table 5.4.Resuls of -ess for brighness of Circular Yellow. Brigher han Inc Brigher han Inc Brigher han Inc Sa. Same The resuls of he paired -ess are shown in Table B.10 in Appendix B. The resuls indicae ha Leoek is brigher han Gelcore, which is brigher han Dialigh which is saisically same as Precision Solar. Bu, i should be noed ha Dialigh and Precision Solar would no be saisically same if he confidence level were 89%. 20

The raings of he drivers wih glasses were compared wih he raings of he drivers wihou glasses (resuls shown in Table B.11 in Appendix B) and he resuls indicae ha for all he four brands he wo raings were saisically he same. I should be noed ha for Gelcore he raings would be differen if he confidence level were lowered o 85%. The raings of he wo age groups were compared (resuls shown in Table B.12 in Appendix B) and he resuls indicae ha hey are saisically he same for all he four brands. I should be noed ha hese resuls hold even if he confidence level of he es were lowered o as low as 30%. 5.2.2. Doiness Tweny six drivers gave hree raings for doiness of all he manufacurer s LED modules and he sample saisics are presened in Table 5.5. T-ess were performed o find ou if he average raings were saisically differen from 3.00. The resuls (shown in Table 5.5) srongly indicae ha all he LED modules are doier han incandescen lens. Of he four brands, saisically Dialigh is he leas doies. The resuls of he six paired -ess are shown in Table B.13 in Appendix B. The resuls indicae ha Gelcore is doier han Precision Solar which is doier han Dialigh which is saisically he same as Leoek. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 26 3.43 0.56 3.94 2.918E-04 Rejec 26 4.06 0.47 11.50 8.887E-12 Rejec 26 3.36 0.49 3.73 4.898E-04 Rejec 26 3.71 0.49 7.29 6.163E-08 Rejec More Doier More Doier More Doier More Doier Table 5.5.Resuls of -ess for doiness of Circular Yellow. 21

The raings of drivers wih and wihou glasses were compared (resuls shown in Table B.14 in Appendix B) and he resuls indicae ha saisically here is no difference beween he raings of he wo groups for any of he four brands. The raings of he wo age groups were compared (resuls shown in Table B.15 in Appendix B) and he resuls indicae ha for Leoek, Gelcore and Dialigh he wo raings are same. Bu for Precision Solar hey are differen saisically. Boh age groups indicaed ha Precision Solar is (saisically) doier han incandescen lens. Of he wo groups, he older group found Precision Solar o be more doier han he younger group. 5.2.3. Color Tweny nine drivers gave hree raings for color of all he manufacurer s LED modules and he sample saisics are presened in Table 5.6. T-ess were performed o find ou if he average raings were saisically differen from 3.00. The resuls (shown in Table 5.6) srongly indicae ha all he LED modules are darker han incandescen lens. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 29 3.30 0.59 2.71 5.680E-03 Rejec 29 3.34 0.51 3.61 5.943E-04 Rejec 29 3.43 0.54 4.24 1.104E-04 Rejec 29 3.41 0.62 3.59 6.257E-04 Rejec Table 5.6.Resuls of -ess for color of Circular Yellow. Darker han Inc Darker han Inc Darker han Inc Darker han Inc The resuls of he six paired -ess are shown in Table B.16 in Appendix B. The resuls indicae ha, Dialigh is saisically same as Precision Solar bu is darker han ohers and Precision Solar is saisically same as Gelcore bu darker han Leoek and Gelcore and Leoek are saisically same. However i should be noed ha he confidence 22

levels associaed wih some of hese ess are close o 90% and herefore hese indicaions should be acceped wih cauion. The comparison (resuls shown in Table B.17 in Appendix B) of he raings of he drivers wih glasses and wihou glasses indicaes ha he wo raings are saisically same for all he four brands and his conclusion holds even if he confidence level were lowered o 50%. When he raings of he wo age groups were compared saisically (resuls shown in Table B.18 in Appendix B) he resuls indicae ha for Gelcore and Dialigh he wo raings are same while for Leoek and Precision Solar hey are differen. The Leoek and Precision Solar raings of he wo groups were individually compared wih 3.00 (incandescen raing) and i was found ha all he groups found i o be saisically darker han incandescen. 5.3. Circular Green 5.3.1. Brighness Thiry one drivers gave hree raings for brighness of all he manufacurer s LED modules and he sample saisics are presened in Table 5.7. T-ess were performed o find ou if he average raings were saisically differen from 3.00. The resuls (shown in Table 5.7) srongly indicae ha all he LED modules are brigher han incandescen lens. Manufacurer N Sd. Deviaion T (1-ailed) Inference Conclusion LEOTEK 31 3.53 0.56 5.27 5.379E-06 Rejec 31 4.01 0.58 9.70 4.679E-11 Rejec 31 3.38 0.54 3.91 2.434E-04 Rejec 31 4.29 0.61 11.72 5.039E-13 Rejec Table 5.7.Resuls of -ess for brighness of Circular Green. Brigher han Inc Brigher han Inc Brigher han Inc Brigher han Inc 23

The resuls of he paired -ess are shown in Table B.19 in Appendix B. The resuls srongly indicae ha he decreasing order of brighness is Precision Solar, Gelcore, Leoek and Dialigh. The comparison (resuls shown in Table B.20 in Appendix B) of he raings of he drivers wih glasses and wihou glasses indicaes ha he wo raings are saisically same for all he four brands. Excep for Gelcore his conclusion holds even if he confidence level were lowered o 50%. When he raings of he wo age groups were compared saisically (resuls shown in Table B.21 in Appendix B) he resuls indicae ha he wo raings are saisically same for all he four brands. However, i should be noed ha for Dialigh he raings would be differen if he confidence level were lowered o 86%. Also for Leoek he variances were found o be saisically differen and herefore Welch s es was used. Alhough saisically no significan, we can observe ha, younger drivers found all he LED modules o be brigher han older drivers. 5.3.2. Doiness Thiry one drivers gave hree raings for doiness of all he manufacurer s LED modules and he sample saisics are presened in Table 5.8. T-ess were performed o find ou if he average raings were saisically differen from 3.00. The resuls (shown in Table 5.8) indicae ha all he LED modules are doier han incandescen lens. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 31 4.07 0.41 14.49 2.276E-15 Rejec 31 4.12 0.60 10.39 9.348E-12 Rejec 31 3.95 0.87 6.03 6.480E-07 Rejec 31 3.17 0.45 2.10 2.211E-02 Rejec Table 5.8.Resuls of -ess for doiness of Circular Green. More Doier More Doier More Doier More Doier 24

The resuls of he six paired -ess are shown in Table B.22 in Appendix B. The resuls indicae ha Gelcore, Dialigh and Leoek are saisically same and are doier han Precision Solar. I should be noed ha he confidence level of he ess comparing Precision Solar wih oher LED modules is very close o 100% which srongly indicaes ha Precision Solar is he leas doier of he four brands. The comparison (resuls shown in Table B.23 in Appendix B) of he raings of he drivers wih glasses and wihou glasses indicaes ha he wo raings are saisically same for all he four brands. The raings of he wo age groups were compared (resuls shown in Table B.24 in Appendix B) and he resuls indicae ha he wo raings are saisically same for all he four brands. However, i should be noed ha for Dialigh, he variances of he wo groups were found o be saisically differen and herefore Welch s es was used. 5.3.3. Color Tweny nine drivers gave hree raings for color of all he manufacurer s LED modules and he sample saisics are presened in Table 5.9. T-ess were performed o find ou if he average raings were saisically differen from 3.00. The resuls (shown in Table 5.9) srongly indicae ha all LED modules are darker han incandescen lens. Of he four brands, saisically Precision Solar was found o be leas darker han incandescen lens. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 29 3.55 0.65 4.57 4.501E-05 Rejec 29 3.44 0.43 5.56 2.983E-06 Rejec 29 3.52 0.52 5.39 4.819E-06 Rejec 29 3.37 0.55 3.59 6.201E-04 Rejec Table 5.9.Resuls of -ess for color of Circular Green. Darker han Inc Darker han Inc Darker han Inc Darker han Inc 25

The resuls of he six paired -ess are shown in Table B.25 in Appendix B. The resuls indicae ha, Leoek is saisically same as Gelcore and Dialigh bu darker han Precision Solar and Dialigh is darker han Precision Solar and Gelcore which are saisically same. However i should be noed ha he confidence levels associaed wih all he ess, excep for Leoek and Dialigh are very close o 90% and herefore hese indicaions should be acceped wih cauion. The comparison (resuls shown in Table B.26 in Appendix B) of he raings of he drivers wih glasses and wihou glasses indicaes ha he wo raings are saisically same for all he four brands and hese conclusions hold even if he confidence level of he es were lowered o 30%. When he raings of he wo age groups were saisically compared (resuls shown in Table B.27 in Appendix B) he resuls indicae ha excep for Precision Solar he wo raings are saisically same for all he brands. The Precision Solar raings of he wo groups were individually compared wih 3.00 (incandescen raing) and i was found ha boh he groups found i o be saisically darker han incandescen. Alhough saisically no significan, for he oher hree brands arihmeically, he older group found he LED modules o be darker han he younger group. 5.4. Yellow Lef Arrow 5.4.1. Brighness Tweny eigh drivers gave raings for brighness of all he manufacurer s LED modules and he sample saisics are presened in Table 5.10. T-ess were performed o find ou if he average raings were saisically differen from 3.00. The resuls (shown in Table 5.10) indicae ha Leoek and Dialigh are saisically same as incandescen and Gelcore and Precision Solar are brigher han incandescen lens. However i should be noed ha he Gelcore would also be same as incandescen if he confidence level were raised o 96%. The resuls of he paired -ess are shown in Table B.28 in Appendix B. The resuls indicae ha Gelcore, Leoek and Dialigh are saisically he same and all are 26

dimmer han Precision Solar. I should be noed ha he confidence level for he comparison beween Leoek and Gelcore is very close o he 90%. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 28 3.04 1.00 0.19 4.257E-01 Rejec Sa same as Inc 28 3.29 0.85 1.77 4.407E-02 Rejec Brigher han Inc 28 3.11 1.31 0.43 3.349E-01 Rejec Sa same as Inc 28 3.61 0.83 3.86 3.179E-04 Rejec Table 5.10. Resuls of -ess for brighness of Yellow Lef Arrow. Brigher han Inc The raings of he drivers wih glasses were compared wih he raings of he drivers wihou glasses (resuls shown in Table B.29 in Appendix B) and he resuls indicae ha for Dialigh he wo raings are saisically differen while for he ohers hey are saisically he same. I should also be noed ha drivers wih glasses gave a raing below 3.00 indicaing ha i is dimmer han incandescen while he drivers wihou glasses gave a raing greaer han 3.00 indicaing ha i is brigher han incandescen. The raings of he wo groups were individually compared wih 3.00 (incandescen raing) and i was found ha he group wihou glasses found i be saisically brigher han incandescen while he raing of he group wih glasses is saisically same as he incandescen lens. When he raings of he wo age groups were saisically compared (resuls shown in Table B.30 in Appendix B) he resuls indicae ha he wo raings are saisically same for all he brands. However for Precision Solar he raings would be differen if he confidence level were lowered o 88%. 5.4.2. Doiness Thiry wo drivers gave raings for doiness of all he manufacurer s LED modules and he sample saisics are presened in Table 5.11. T-ess were performed o find ou if he 27

average raings were saisically differen from 3.00. The resuls (shown in Table 5.11) indicae ha all he LED modules are doier han incandescen lens. I should be noed ha hese conclusions hold even if he confidence level were raised o 99.9%, which srongly indicaes ha all of hem are doier han incandescen. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 32 3.72 0.73 5.58 2.098E-06 Rejec 32 3.66 0.87 4.29 8.113E-05 Rejec 32 3.84 0.68 7.05 1.236E-08 Rejec 32 3.56 0.80 3.97 1.962E-04 Rejec Table 5.11.Resuls of -ess for doiness of Yellow Lef Arrow. More Doier More Doier More Doier More Doier The resuls of he six paired -ess are shown in Table B.31 in Appendix B. The resuls indicae ha Dialigh is doier han Precision Solar and all he oher pairs are saisically same. These conclusions should be acceped wih cauion because, if he confidence level of he ess were lowered o 85% he conclusions would be significanly differen. The comparison (resuls shown in Table B.32 in Appendix B) of he raings of he drivers wih glasses and wihou glasses indicaes ha he wo raings are saisically same for all he four brands. When he raings of he wo age groups were saisically compared (resuls shown in Table B.33 in Appendix B) he resuls indicae ha he wo raings are saisically same for all he brands excep Leoek. Alhough saisically no significan, for he oher hree manufacurers also, he older group found he LED modules o be darker han he younger group. 5.4.3. Color Tweny five drivers gave raings for color of all he manufacurer s LED modules and he sample saisics are presened in Table 5.12. T-ess were performed o find ou if he 28

average raings were saisically differen from 3.00. The resuls (shown in Table 5.12) indicae ha all he four brands are darker han incandescen lens. These conclusions should be acceped wih cauion because, if he confidence level of he ess were raised o 95% he conclusions would be significanly differen. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 25 3.24 0.83 1.44 8.075E-02 Rejec 25 3.24 0.78 1.54 6.824E-02 Rejec 25 3.28 0.89 1.57 6.454E-02 Rejec 25 3.28 0.61 2.28 1.585E-02 Rejec Table 5.12.Resuls of -ess for color of Yellow Lef Arrow. Darker han Inc Darker han Inc Darker han Inc Darker han Inc The resuls of he six paired -ess are shown in Table B.34 in Appendix B. The resuls indicae ha all he pairs are saisically same and hese conclusions hold even if he confidence level of he ess were lowered o 60%. The comparison (resuls shown in Table B.35 in Appendix B) of he raings of he drivers wih glasses and wihou glasses indicaes ha he wo raings are saisically same for all he four brands and hese conclusions hold even if he confidence level of he ess were lowered o 60%. When he raings of he wo age groups were saisically compared (resuls shown in Table B.36 in Appendix B) he resuls indicae ha he wo raings are saisically same for all he brands. 5.5. Green Lef Arrow 5.5.1. Brighness Tweny seven drivers gave raings for brighness of all he manufacurer s LED modules and he sample saisics are presened in Table 5.13. T-ess were performed o find ou if 29

he average raings were saisically differen from 3.00. The resuls (shown in Table 5.13) indicae ha Leoek is saisically same as incandescen and Dialigh, Gelcore and Precision Solar are brigher han incandescen lens. I should be noed ha Leoek would be brigher han incandescen if he confidence level were lowered o 85%, while all he oher brands would remain brigher even if he confidence level were raised o 99.9%. Manufacurer N Sd. Deviaion (1-ailed) Inference Conclusion LEOTEK 27 3.19 0.88 1.10 1.418E-01 Rejec Sa same as Inc 27 4.15 0.72 8.31 4.324E-09 Rejec 27 3.89 0.80 5.77 2.239E-06 Rejec Brigher han Inc Brigher han Inc 27 3.63 0.69 4.76 3.188E-05 Rejec Table 5.13.Resuls of -ess for brighness of Green Lef Arrow. Brigher han Inc The resuls of he paired -ess are shown in Table B.37 in Appendix B. The resuls indicae ha he decreasing order of brighness is Gelcore, Dialigh, Precision Solar and Leoek. However i should be noed ha Dialigh would be same as Gelcore and Precision Solar if he confidence level were raised o 94%. Therefore hese indicaions should be acceped cauiously. The raings of he drivers wih glasses were compared wih he raings of he drivers wihou glasses (resuls shown in Table B.38 in Appendix B) and he resuls indicae ha for Dialigh he wo raings are saisically differen while for he ohers hey are saisically same. Drivers wih glasses gave a lesser raing han he drivers wihou glasses. The raings of he wo Dialigh groups were individually compared wih 3.00 (incandescen raing) and i was found ha boh he groups found i o be saisically brigher han incandescen. I should also be noed ha for he oher hree brands, he wo raings remain equal even if he confidence level were lowered o 40% which srongly indicaes ha he raings of he wo groups are equal for hese hree brands. 30