Effectiveness of Alternative Detector Configurations for Option Zone Protection on High-speed Approaches to Traffic Signals

Transcription

1 University of Tennessee, Knoxville Trce: Tennessee Reserch nd Cretive Exchnge Msters Theses Grdute School Effectiveness of Alterntive Detector Configurtions for Option Zone Protection on High-speed Approches to Trffic Signls Jinwen Si University of Tennessee - Knoxville Recommended Cittion Si, Jinwen, "Effectiveness of Alterntive Detector Configurtions for Option Zone Protection on High-speed Approches to Trffic Signls. " Mster's Thesis, University of Tennessee, This Thesis is brought to you for free nd open ccess by the Grdute School t Trce: Tennessee Reserch nd Cretive Exchnge. It hs been ccepted for inclusion in Msters Theses by n uthorized dministrtor of Trce: Tennessee Reserch nd Cretive Exchnge. For more informtion, plese contct trce@utk.edu.

2 To the Grdute Council: I m submitting herewith thesis written by Jinwen Si entitled "Effectiveness of Alterntive Detector Configurtions for Option Zone Protection on High-speed Approches to Trffic Signls." I hve exmined the finl electronic copy of this thesis for form nd content nd recommend tht it be ccepted in prtil fulfillment of the requirements for the degree of Mster of Science, with mjor in Civil Engineering. We hve red this thesis nd recommend its cceptnce: Lee D. Hn, Arun Chtterjee (Originl signtures re on file with officil student records.) Thoms Urbnik II, Mjor Professor Accepted for the Council: Crolyn R. Hodges Vice Provost nd Den of the Grdute School

3 To the Grdute Council: I m submitting herewith thesis written by Jinwen Si entitled Effectiveness of Alterntive Detector Configurtions for Option Zone Protection on Highspeed Approches to Trffic Signls. I hve exmined the finl electronic copy of the thesis for form nd content nd recommend tht it be ccepted in prtil fulfillment of the requirement for the Degree of Mster of Science, with mjor in Civil Engineering. Thoms Urbnik II Mjor Professor We hve red this thesis nd recommend its cceptnce: Lee D. Hn Arun Chtterjee Acceptnce for the Council: Lind Pinter Interim Den of Grdute Studies (Originl signtures re on file with officil student records.)

4 Effectiveness of Alterntive Detector Configurtions for Option Zone Protection on High-speed Approches to Trffic Signls A Thesis Presented for the Mster of Science Degree The University of Tennessee, Knoxville Jinwen Si December 2006

5 ACKNOWLEDGEMENTS First nd foremost, I would like to thnk my cdemic dvisor, Dr. Thoms Urbnik II, for his guidnce nd support during my grdute studies t the University of Tennessee. I would lso like to thnk Dr. Lee Hn nd Dr. Arun Chtterjee for serving on my committee. They provided gret help to my reserch nd study. I m lso grteful to my collegues nd friends, Fng Yun nd Shun Quyle, for their ssistnce in my studies. Finlly I would like to extend my grtitude to my fther, Si Xinyong, for his endless love. I could never complete this work without his understnding nd support. ii

6 ABSTRACT Sfety nd efficiency re both prime issues of concern t high-speed signlized intersections. However, lthough sfety problems t high-speed signlized intersections cn be mitigted by the ppliction of detector configurtions with option zone protection, such configurtions cn themselves crete problems. For exmple extending the green indiction to the mximum time cn crete mx-out situtions which termintes the phse without regrd for vehicle loction. The detector configurtions with option zone protection my lso produce higher dely times for vehicles on other pproches t the intersection. In this study, four detector configurtions with option zone protection fetures- - the Single Detector configurtion, the SDITE configurtion, the Beirele configurtion, nd the Bonneson configurtion, re compred through computer simultion on their effects reltive to the number of vehicles in the option zone, mx-out occurrences, nd verge vehicle dely. The comprison suggests tht lthough ech configurtion hs its own dvntges nd disdvntges, the Bonneson detector configurtion, in most circumstnces, yields lower number of vehicles in the option zone per cycle nd less verge dely. iii

7 TABLE OF CONTENTS Chpter Pge CHAPTER I... 1 Bckground... 1 CHAPTER II... 2 Effectiveness of Alterntive Detector Configurtions for Option Zone Protection on High-speed Approches to Trffic Signls... 2 Introduction... 2 Option Zone (or Decision Zone)... 3 Option Zone Boundry... 4 Difference between Option Zone nd Dilemm Zone... 5 Option Zone Protection Design Considertions... 5 Detector Designs... 7 Computer Simultion Anlysis of Simultion Results Number of Vehicles in Option Zone Mx-out Occurrences Dely Times CHAPTER III Additionl Study for High Volume Condition Number of Vehicles in Option Zone Mx-out Occurrences Dely Time iv

8 CHAPTER IV Conclusions nd Future Study Recommendtion LIST OF REFERENCES Vit v

9 LIST OF TABLES Tble Pge Tble 1 Vehicles in option zone with rteril vehicle desired speed t 30 mph Tble 2 Vehicles in option zone with rteril vehicle desired speed t 40 mph Tble 3 Vehicles in option zone with rteril vehicle desired speed t 50 mph Tble 4 Mx-out occurrence percentges (%) on rteril rod Tble 5 MAH s of lterntive configurtions for 50 mph Tble 6 Vehicle verge totl dely (s) with rteril vehicle desired speed t 30 mph Tble 7 Vehicle verge totl dely (s) with rteril vehicle desired speed t 40 mph Tble 8 Vehicle verge totl dely (s) with rteril vehicle desired speed t 50 mph Tble 9 Vehicles in option zone under high volume condition Tble 10 Mx-out occurrence percentges (%) on rteril rod under high volume condition Tble 11 Vehicle verge totl dely (s) under high volume condition vi

10 LIST OF FIGURES Figure Pge Figure 1 Beirele recommended detector configurtions for 50-mph design speed... 9 Figure 2 SDITE recommended detector configurtion for 50-mph design speed Figure 3 Bonneson recommended detector configurtions for rurl intersections with 50-mph design speed vii

11 CHAPTER I BACKGROUND Historiclly mny crshes hve been reported t high-speed signlized intersections. The re on the pproch to the intersection, clled option zone or decision zone, poses high ccident potentil for drivers becuse of the difficulty to mke decision on stopping or proceeding, especilly t high speeds. This sitution of uncertin decision cn potentilly led to rer-end or right ngle crshes. There re mny detection configurtions for option zone protection tht could reduce the chnce vehicle flls into the option zone, thus improving the sfety t intersections. This study investigted four detection configurtions extensively used in the United Sttes. In order to compre their opertionl nd sfety effectiveness, the detector plcements nd ssocited timing were modeled in simultion softwre under different speed/volume conditions. The simultion results hve been nlyzed from both sfety nd opertionl spects. Finlly, conclusion nd recommendtions were mde following the nlysis. 1

12 CHAPTER II EFFECTIVENESS OF ALTERNATIVE DETECTOR CONFIGURATIONS FOR OPTION ZONE PROTECTION ON HIGH-SPEED APPROACHES TO TRAFFIC SIGNALS This chpter is slightly revised version of pper by the sme nme to be presented in the nnul conference of Trnsporttion Reserch Bord in 2007 by Jinwen Si, Thoms Urbnik II, nd Lee D. Hn. My primry contributions to this pper include: 1) performing computer simultion nd gthering dt results fter simultion; 2) nlyzing the dt results from sfety nd opertionl spects; 3) writing nd modifying the pper. Introduction There re mny detection lyouts used t isolted signlized intersections throughout the United Sttes, even when only concerned with single type of opertion like high-speed through lnes. The issues re further clouded by the complexity of issues ssocited with high-speed control including the timing of the yellow nd red clernces. This study compres four detection lyouts. For simplicity in this evlution of detector lyout nd timing, some of the complexity will be removed through series of ssumptions. The focus is on how well the lyouts of the detection zones nd the ssocited signl 2

13 times minimize the number of vehicles exposed to yellow indiction while in the option zone (or decision zone). In this pper, we reserve the use of the term dilemm zone, to the sitution in which the driver is neither ble to cler the intersection before conflicting green or sfely stop. Dilemm zones cn be eliminted s result of the pproprite timing of the yellow nd red clernce intervls. Option Zone (or Decision Zone) The option zone (lso known s decision zone ) is defined s length of rodwy in dvnce of the intersection where n individul driver my experience indecisiveness upon seeing the indiction of yellow (1). When the signl indiction to trffic pproching n intersection chnges from green to yellow, driver hs to mke decision whether to stop or ttempt to cler the intersection. This decision is bsed on severl fctors: vehicle s distnce from intersection, vehicle s speed, the driver s cceptble decelertion rte, etc. If the vehicle is fr enough wy from the intersection t the onset of the yellow, the driver cn stop with reltively low decelertion nd with consequently low risk or rer-end crsh. In this cse, most drivers will decide to stop. Also, when the vehicle is very ner the intersection t the onset of yellow, clering the intersection cn esily be ccomplished with little risk of side-impct, so most drivers in this sitution will choose to cler the intersection. Problems occur when the vehicle is within the option zone. Different drivers exhibit different rections to the indiction of yellow s result of their perception- 3

14 rection time, decelertion tolernce, nd inclintions for stopping. Crshes my occur becuse of the vritions in drivers rections. Option Zone Boundry The option zone is not precise region; indeed it vries t different pproch speeds. It is often defined by the stopping probbility for which there is only limited dt. Reserchers hve observed nd tbulted the probbility of vehicle s stopping if given yellow indiction X feet in dvnce of the stop line t V pproch speed. Prsonson nd Zegeer hve defined the limits of the option zone s greter thn 10% chnce of stopping to less thn 90% chnce of stopping (2, 3). Generlly, the option zone boundries re defined reltive to the stop line (1). Avilble reserch hs shown tht 90% of motorists will decide to stop, nd 10% will decide to cler the intersection, if they re bout 4.5 to 5 seconds of trvel time from the intersection t the time when the yellow signl indiction comes on. Only 10% of motorists will decide to stop, nd 90% will decide to go through the intersection, if they re 2 to 2.5 seconds of trvel time from the intersection t the onset of yellow. Therefore the option zone hs length of 2 to 3 seconds of trvel time. For the purpose of this reserch, the option zone is defined s 2 to 5 seconds in front of the stop br. 4

15 Difference between Option Zone nd Dilemm Zone The term dilemm zone ws often used to indicte the sitution cused by option zone. The dilemm zone in this pper is limited to the sitution where driver is confronted with the dilemm on the pproch to n intersection creted t the moment when the signl indiction chnges from green to yellow tht the vehicle t certin speed is neither ble to be stopped before the intersection nor ble to cler the intersection before the conflicting green. The dilemm zone issue cn be solved by modifying the clernce intervl of the intersection. The boundries of the dilemm zone nd option zone sometimes overlp in certin prts, but re fundmentlly different issues. The option zone cnnot be eliminted by djustments in the yellow nd red clernce time. Only the dilemm zone cn be eliminted by pproprite yellow nd red clernce times. Option Zone Protection Design Considertions To provide option zone protection t high-speed intersections, one or more detectors re usully plced upstrem of the stop line. Advnce loop detector design is generlly bsed on the loction of the option zone for the rnge of speeds commonly found on the pproch. The pssge time (nd perhps the detector unit s cll-extension) setting is set such tht vehicle within the design speed rnge will cler some design point. Often, two or more loops re used to reduce the pssge time setting. The phse is then extended 5

16 when the trvel time is less thn the design ssumption of ny pir of loops. The detection system will gp out when hedwy occurs tht is greter thn the Mximum Allowble Hedwy (MAH) (4) of the system for ll pproching lnes. For the purposes of this pper, we ssume lne by lne detection (5) to eliminte issues with the interction with gps on djcent lnes. Bonneson nd McCoy (4) hve defined MAH s follows: MAH = MAH MAH MAH s = PT + CE = PT + CE + MAH n s s D1 Dn + Ld + V L + ds + L V v + L v (1) where, MAH = mximum llowble hedwy of the detection design for the pproch (s), MAH = mximum llowble hedwy of the dilemm zone protection system (s), MAHs = mximum llowble hedwy of the stop-line detectors (s), PT = pssge time setting in the controller (s), CEn = cll-extension setting for detectors in the dilemm zone protection system (s), D1 = distnce from the stop line to the leding edge of the dilemm zone detector frthest from the stop line (ft), Dn = distnce from the stop line to the leding edge of the dilemm zone detector nerest the stop line (ft), 6

17 Ld = length of the dilemm zone detectors (ft), Lv = detected length of vehicle (ft), V = verge running speed on the intersection pproch (ft/s), CEs = cll-extension setting for the stop line detectors (s), nd Lds = length of stop line detectors (ft). Eqution 1 ssumes tht detection systems crry vehicle through its option zone. Another eqution to clculte the MAH for crrying vehicles to the stop line hs lso been provided by Bonneson nd McCoy (4). If ll hedwys re smller thn MAH until the phse reches its mximum vlue, the detection system will terminte the green by mxing out without regrd to vehicles in the option zone. Detector Designs The option zone problem hs been ddressed by severl different configurtions for high-speed intersections (2). Four lterntive configurtions hve been selected for evlution: the Single Detector configurtion, the Beirele configurtion, the Southern District Institute of Trffic Engineers (SDITE) configurtion, nd the Bonneson configurtion. Single Detector Configurtion This is the simplest option zone protection configurtion, employing only one 6x 6-ft (1.8x 1.8-m) detector on ech lne, plced t the strt of the option 7

18 zone for the design speed. No considertion is given to slower-speed vehicles. Three seconds of pssge time will be used to llow design-speed vehicles to trvel out of their option zone t the design speed. This configurtion must operte with locking memory since there is no stop-line detector. Beirele Configurtion The Beirele configurtion uses one-second pssge time throughout the detection zone (2). The controller opertes in fully ctuted locking mode. It utilizes 6x 6-ft (1.8x 1.8-m) presence mode loop detectors. The detector lyout is bsed on sfe stopping distnce for vehicles with different speeds. The outermost detector is plced where vehicle with design speed cn stop sfely. The second detector is locted t the sfe stopping distnce for speed 10 mph less thn the design speed. Other detectors closer to the intersection follow the sme procedure, with 10 mph less ech time, until the lst one is within 75 ft of the stop line. The Texs Stte Deprtment of Highwys nd Public Trnsporttion (now TxDOT) modified this configurtion with the AASHTO stopping distnce criteri. The detector plcement in the Beirele configurtion for 50-mph design speed is shown in Figure 1. 8

19 Figure 1 Beirele recommended detector configurtions for 50-mph design speed 9

20 SDITE Configurtion This configurtion ws developed by the Southern District Institute of Trnsporttion Engineers (6, 7). It uses bsic ctuted controller nd multiple 6x 6-ft loops, but it opertes in non-locking mode. The pssge time is 2 seconds. This configurtion utilizes primrily engineering judgment to determine the loction of detectors. The outermost detector is positioned t pproximtely 5 seconds of trvel distnce t design speed to give option zone protection. The second detector should be locted to llow the 50-mph vehicle to hold the green. The other detectors re plced to ccommodte vehicles with reduced speed, nd the stop-line detector prevents premture gp-out during queue dischrge. Figure 2 is the SDITE recommended detector configurtion for 50-mph speed. Bonneson Configurtion This configurtion ws developed by Bonneson nd McCoy (4) in 1994, nd it hs two different recommendtions for rurl nd for urbn intersections. The recommended detection for rurl intersections will be discussed here. The Bonneson configurtion hs fetures tht include locking controller memory, pulse-mode detection, no stop-line detector, nd two-second pssge time. 10

21 Figure 2 SDITE recommended detector configurtion for 50-mph design speed 11

22 Advnce detectors re locted t the beginning of the option zones for their design speeds. The outermost detector hs the sme design speed s the rod, nd every subsequent detector hs design speed 10 mph less thn the one before it. There re two possible design gols in this configurtion: either to crry the lst vehicle through its option zone before the onset of yellow or to crry the lst vehicle to the stop line before the onset of yellow. The configurtion with the first gol will be used in our discussion becuse it provides greter efficiency by terminting the green indiction erlier. The detectors in our simultion re operted in presence mode insted of pulse mode due to the current limittions of the simultion model. This results in modest inccurcy equl to the time it tkes vehicles to trverse the 6 x 6 loop (8/100 of second t 50 mph). Figure 3 is the Bonneson recommended configurtion for rurl intersection with 50-mph design speed. Computer Simultion Computer simultion ws used to compre the effectiveness of these detection configurtions described bove. A micro-simultion softwre, VISSIM (8), ws selected s the trffic-simultion model for this investigtion. An intersection between two-lne rteril rod nd one-lne minor rod ws used with the following conditions: through movements only, two-phse, fully ctuted signl control, nd 5-second chnge intervl. There is only 12

23 Figure 3 Bonneson recommended detector configurtions for rurl intersections with 50-mph design speed 13

24 one direction of trffic on the rteril rod in order to eliminte the influence of trffic clls from the opposite direction, nd two directions of trffic on the minor rod. The trffic is composed of 98% of pssenger crs nd 2% of trucks. For our study, we used the defult vehicle settings in VISSIM. Ech configurtion ws implemented bsed on its specific chrcteristics of detector loction nd trffic-controller timings utilizing NEMA controller in VISSIM. The detector lyouts for ll lterntives were designed ccording to n rteril design speed of 50 mph. The minimum nd mximum green times were set t 15 nd 60 seconds for the rteril rod, nd 10 nd 20 seconds for the minor rod. Three seprte cses of desired speed-- 30mph, 40mph, nd 50 mph-- were used for the vehicles on the rteril rod in order to evlute the effect of design-speed, s well s slower-thn-design-speed trffic on the rteril. Three levels of trffic volume for ech cse of desired speed, 400 vphpl, 500 vphpl, nd 600 vphpl, were lso used on the rteril to investigte the detection configurtions performnces under different volume conditions. A reltively low desired speed (30mph) ws used on the cross street, suggesting configurtion bsed on two 60-ft-long stop line presence detectors on both minor rod directions. The trffic volume ws set t constnt 300 vphpl for the two minor rods. Ech lterntive ws run 30 times for 900 seconds under ech rteril vehicle speed nd volume condition. After the simultion process, the outputs from totl of 1080 independent 900-seconds runs were nlyzed nd compred. 14

25 Anlysis of Simultion Results The four lterntive detector configurtions were compred on the bsis of sfety nd efficiency performnce. The mesures of effectiveness for sfety performnce re the number of vehicles in the option zone t onset of yellow nd the number of instnces of the rteril green indiction being forced off due to the mximum timer. On the other hnd, verge vehicle dely is used s n efficiency mesure of effectiveness. Number of Vehicles in Option Zone The primry objective of the lterntive detection configurtions is to provide option zone protection. It is, of course, desirble tht there should be no vehicles in the option zone t the onset of the yellow indiction. However, vehicles could fll into the option zone for vriety of resons, including driving below the design speed or hving the green indiction terminte becuse of mx-out insted of gp-out. In relity, it is imprcticl to scertin the exct number of vehicles in option zone t signlized intersections in the field becuse the option zone loction is ssocited with vehicle speed; therefore there is specific option zone for every vehicle, mking it difficult to mesure the number of vehicles in the option zone t the moment the signl turns to yellow. 15

26 VISSIM cn record signl indiction, vehicle coordintes, vehicle speed, nd other informtion t the end of every simultion step, then sving it into output files. Therefore, t every simultion step, first the signl indiction is checked; then, if it is yellow, ll vehicles will be exmined if they re within their option zone bsed on their distnce from the stop line nd their speeds. Bsed on the VISSIM simultion results of 30 simulted runs, the cumultive numbers of vehicles cught in option zones t the onset of yellow (30 runs), s well s the verge numbers per cycle were tbulted in Tbles 1, 2 nd 3 for rteril speeds of 30, 40, nd 50 mph, respectively. Initil observtions of the numbers of vehicles trpped in option zones indicted the following: For 30 nd 40 mph conditions, the Single Detector configurtion (designed for 50 mph) significntly underperformed in comprison with the lterntives; nd For 40 nd 50 mph conditions, the Bonneson configurtion clerly outperformed ll other lterntives. The reltive performnce of the detector configurtions under other conditions ws less obvious. Therefore, the ANOVA test ws performed to compre the number of vehicles in option zone per cycle under those conditions. According to the test results, only under the 40 mph condition the dt 16

27 Tble 1 Vehicles in option zone with rteril vehicle desired speed t 30 mph b c d Single Detector SDITE Beirele Bonneson Arteril Rod Vehicle Volume (vphpl) Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle d 0.20 d 0.35 Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle d 0.25 Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle 0.13 b c b 0.20 Not compred with the others Significntly different from SDITE (α=.05) Significntly different from Beirele (α=.05) Significntly different from Bonneson (α=.05) c d c 0.35 b d 17

28 Tble 2 Vehicles in option zone with rteril vehicle desired speed t 40 mph Cumultive Vehicles in Single Option Zone (30 runs) Detector Vehicles in Option Zone per Cycle Cumultive Vehicles in Option Zone (30 runs) SDITE Vehicles in Option Zone per Cycle Cumultive Vehicles in Option Zone (30 runs) Beirele Vehicles in Option Zone per Cycle Cumultive Vehicles in Option Zone (30 runs) Bonneson Vehicles in Option Zone per Cycle Not compred with the others Arteril Rod Vehicle Volume (vphpl)

29 Tble 3 Vehicles in option zone with rteril vehicle desired speed t 50 mph b c d Single Detector SDITE Beirele Bonneson Arteril Rod Vehicle Volume (vphpl) Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle 0.04 b c 0.08 b c 0.16 b c Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle c 0.08 d c 0.16 d 0.32 d Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle 0.30 b d 0.33 b d 0.39 d Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle Not compred with the others 0.00 Significntly different from SDITE (α=.05) Significntly different from Beirele (α=.05) Significntly different from Single Detector (α=.05)

30 compred were not significntly different (α=.05). Becuse ech configurtion ws run 30 times (10-20 cycles ech time) for ech cse, the number of vehicles in option zone per cycle (the men of ech run) ws ssumed to follow the norml distribution. Hence for the other cses tht the number of vehicles in option zone per cycle ws significntly different (α=.05) between configurtions, the Tukey test, which requires the dt tested follow the norml distribution, with α=.05, ws performed to compre these results in order to find out which two results re significntly different (α=.05). The results of the sttisticl test were lso incorported into Tbles 1, 2 nd 3. The Bonneson configurtion ws found to result in fewer vehicles trpped in options zone per cycle thn ll other detector configurtions with confirmed sttisticl significnce. The only exception ws for the 30-mph condition with demnd of 600 vph; in this cse, Bonneson, while still outperforming the Single Detector nd SDITE, did not sttisticlly outperformed Beirele, (i.e., the α=.05 significnce level ws not met). Results from the SDITE nd Beirele configurtions re not sttisticlly different (α=.05) under 6 out of 9 conditions. The Single Detector configurtion genertes fr greter numbers of vehicles in option zones per cycle under 30 nd 40 mph, s observed nd reported erlier, wheres in the under-50-mph the results were significntly lower (α=.05) thn the SDITE nd Beirele. Overll, the Bonneson configurtion provides the best option 20

31 zone protection under lmost ll circumstnces, while the Single Detector configurtion only provides effective option zone protection when rteril vehicle desired speed is t the design speed (50 mph for the simulted cses). The performnces of the SDITE nd Beirele configurtions re similrly mediocre. Mx-out Occurrences The term mx-out refers to the immedite termintion of green when it hs lredy been extended to its mximum llowble time. If the green is terminted by mx-out insted of gp-out, the signl controller will not provide ny option zone protection. Generlly speking, higher occurrence of mxout indictes less effective option zone protection. The results of mx-out occurrences on rteril rod for ech lterntive re shown in Tble 4. As cn be seen, the Single Detector configurtion hs no mx-outs, wheres the SDITE configurtion genertes the lrgest mx-out occurrence. Tble 5 shows the MAH s of ll four lterntive configurtions when vehicles re pproching t rteril rod design speed. The 6-ft detector length nd 20-ft vehicle length were used in the clcultion of the MAH s of the four lterntive configurtions. It cn be concluded from Tble 4 nd 5 tht mx-out occurrence is strongly relted with the detector configurtion s MAH, s most vehicles pproching intersection t design 21

32 Tble 4 Mx-out occurrence percentges (%) on rteril rod Single Detector SDITE Beirele Bonneson Arteril Vehicle Speed (mph) Arteril Vehicle Volume (vphpl)

33 Tble 5 MAH s of lterntive configurtions for 50 mph MAH (s) Single Detector 3.4 SDITE 7.0 Beirele 4.3 Bonneson

34 speed or lower. The SDITE configurtion hs the lrgest MAH s nd the lrgest percentge of mx-out occurrence, while the Single Detector configurtion with the smllest MAH s produces no mx-out t ll. Since mxout is the sitution in which signl control system loses option zone protection for vehicles, it is desirble to limit MAH to resonble vlue. Dely Times The verge totl dely time of every vehicle trveling through both rteril nd minor rods is used to compre the opertionl efficiencies of the different lterntives. The verge totl dely is the difference between rel trvel time nd theoreticl trvel time. The theoreticl trvel time could be reched if there were no ny other vehicles or signl controls for the given rod section (8). By reducing dely, the level of service t intersections could be improved; therefore, configurtions with less verge dely will be fvored in terms of their opertionl spect. The verge totl dely of every vehicle is clculted by dividing the totl dely time by the number of ll vehicles in the simultion network. Similrly to our nlysis method in the number of vehicles in option zone per cycle, the ANOVA test ws first undertken to compre results from the lterntives. Under every condition there ws t lest one group of results significntly different from the others (α=.05). Then Tukey test with α=.05 were 24

35 performed. Simultion nd sttistic test results re shown in Tbles 6, 7, nd 8. In the 30-mph rteril vehicle desired speed cse, the Single Detector configurtion hs significntly lower (α=.05) dely time, nd the SDITE configurtion hs significntly higher (α=.05) dely time thn the other three configurtions. In 40- nd 50-mph cses, the Single Detector configurtion nd the Bonneson configurtion produce significntly lower (α=.05) dely times; wheres the SDITE configurtion produces significntly higher (α=.05) dely time thn the other configurtions. The results with the Beirele configurtion rnk in the middle in ll three cses. Generlly, the Bonneson configurtion nd the Single Detector configurtion re the most opertionlly efficient configurtions in producing less dely; the SDITE configurtion hs the highest verge dely time. 25

36 Tble 6 Vehicle verge totl dely (s) with rteril vehicle desired speed t 30 mph Arteril Rod Vehicle Volume (vphpl) b c Single Detector c c b SDITE Beirele Bonneson b b b b b b Different letters indicte mens re significntly different under the sme volume condition (α=.05) 26

37 Tble 7 Vehicle verge totl dely (s) with rteril vehicle desired speed t 40 mph Arteril Rod Vehicle Volume (vphpl) b c Single Detector c c c SDITE Beirele Bonneson b c b c b b Different letters indicte mens re significntly different under the sme volume condition (α=.05) 27

38 Tble 8 Vehicle verge totl dely (s) with rteril vehicle desired speed t 50 mph Arteril Rod Vehicle Volume (vphpl) b c Single Detector b c c c SDITE Beirele Bonneson b c b c b b Different letters indicte mens re significntly different under the sme volume condition (α=.05) c 28

39 CHAPTER III ADDITIONAL STUDY FOR HIGH VOLUME CONDITION In lst chpter, the four lterntive configurtions hve been compred under norml volume conditions. In order to evlute the four lterntives performnce under high-volume condition, ech lterntive ws run for nother 30 times with vehicle desired speed t 30, 40 nd 50 mph under rteril vehicle volume t 1200 vphpl. The trffic volume nd desired speed on minor rod re the sme s used in norml volume nlysis. Number of Vehicles in Option Zone The cumultive number of vehicles in option zone nd the number of vehicles in option zone per cycle were tbulted in Tble 9. For 30 mph condition, the Single Detector configurtion significntly underperformed in comprison with the other lterntives. For ll the other conditions, similr to the method used for norml volume nlysis, first n ANOVA test, with α=.05, ws performed, which indicted the significnt difference (α=.05) between results under ech condition. Then Tukey test ws performed to test which two results were different. The test results were lso incorported into Tble 9. Results from the SDITE, Beirele, nd Bonneson configurtion were not sttisticlly differently (α=.05) under 30 mph condition. For 40 mph condition, 29

40 Tble 9 Vehicles in option zone under high volume condition Desired Speed of Vehicle on Arteril (mph) Cumultive Vehicles in Single Option Zone (30 runs) Detector Vehicles in Option Zone per Cycle 2.38 * b 1.78 ** 0.99 SDITE Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle ** 1.83 Beirele Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle 1.78 c 1.45 ** 1.41 Bonneson Cumultive Vehicles in Option Zone (30 runs) Vehicles in Option Zone per Cycle 1.82 bc 1.68 ** 1.29 * In 30 mph condition, Vehicles in Option Zone per Cycle of Single Detector is significntly different from the other three (α=.05) bc In 40 mph condition, different letters indicte results significntly different (α=.05) * * Every result is significntly different from the others (α=.05) 30

41 the Beirele configurtion outperformed the Single Detector nd SDITE configurtion, while genertes the results tht were not significntly different (α=.05) from the Bonneson configurtion. For 50 mph condition, the Single Detector configurtion hd much smller number of vehicles trpped in option zone per cycle thn the other three lterntives, nd in this condition the Bonneson configurtion produced the second best result. Although Bonneson configurtion did not outperform the other lterntives under every speed/volume condition, it performed resonbly ll the time. The Single Detector configurtion generted fewer vehicles trpped in option zone per cycle for design speed cse, but still did not work s well for other cses. Mx-out Occurrences The results of mx-out occurrences under high volume condition were consistent with norml volume condition. The SDITE configurtion with the lrgest MAH produced the highest possibility of mx-out occurrences, resulting in mx-out situtions in lmost every cycle. On the contrry, the Single Detector configurtion generted the lowest mx-out occurrences. These results were incorported in Tble

42 Tble 10 Mx-out occurrence percentges (%) on rteril rod under high volume condition Arteril Vehicle Speed (mph) Single Detector SDITE Beirele Bonneson

43 Dely Time The verge totl dely time of every vehicle under high volume condition is clculted following the sme procedure utilized in lst chpter, which is dividing totl dely time by the overll number of vehicles in network. The results were tbulted in Tble 11. The ANOVA test with α=.05 ws performed to compre the results of ech lterntive. Bsed on the sttisticl nlysis results it ws concluded tht there ws no significnt difference mong the dely time of the lterntives with the confirmed sttisticl significnce. From the comprison results of the verge totl dely time produced by ech lterntive, it could be concluded tht under high volume condition, the detection configurtion did not influence verge totl dely time much. When the trffic volume ws low to moderte, the more efficient configurtion yielded lower dely time by lwys providing enough green time to cler the queue nd servicing the conflicting movement s soon s cceptble gp occurs. However, s the demnd incresed on rteril rod, the controller hd to extend the green mny times to provide option zone protection, which cused more occurrences of mx-out or nerly mx-out. In this cse, significnt congestion occurred, nd therefore, no lterntive ws ble to function superior thn the others. 33

44 Tble 11 Vehicle verge totl dely (s) under high volume condition Desired Speed of Vehicle on Arteril (mph) Single Detector SDITE Beirele Bonneson No result is significntly different from ech other (α=.05) 34

45 CHAPTER IV CONCLUSIONS AND FUTURE STUDY RECOMMENDATION It cnnot be concluded tht ny one detector configurtion is better thn nother bsed on the mesures of effectiveness used in this pper. However, mong ll the detection configurtions studied under most desired speed nd norml volume conditions, the Bonneson configurtion produces lower number of vehicles in the option zone nd lower verge totl dely time s well s moderte mx-out occurrence, while it still provides the second best option zone protection under ll the desired speed condition for the high volume condition. By contrst, the Single Detector configurtion hs much more vehicles in the option zone if the vehicles pproch speed is lower thn the design speed of the configurtion regrdless of the trffic volume on rteril rod; but under norml trffic volume condition, this configurtion cuses no mx-out in simultion becuse of reltively smll MAH nd produces, s well, smll verge totl dely time. As under high trffic volume condition, this configurtion still produces much less mx-out occurrence thn the other lterntives. The performnce of the Beirele configurtion is in the middle mong the lterntives. No remrkble dvntge or disdvntge hs been found with this configurtion. 35

46 The lst configurtion, SDITE, hs one min disdvntge: the long MAH. Consequently, this configurtion hs more vehicles in the option zone, greter possibility of mx-out, nd n verge totl dely time. Also, this configurtion uses six detectors for 50-mph design speed, which is not desirble from n economicl stndpoint. This study ws intended s n explortory study to develop better understnding of detector configurtion performnce thn ws previously possible with strictly nlyticl nlysis. Additionl work would be desirble in order to look t more cses using speed distributions bsed on ctul field conditions. In ddition, besides the detection configurtions discussed, there re some ltely developed signl control systems, such s Detection-Control System (9) developed by Texs Trnsporttion Institute. It would lso be desirble to test the performnce of these systems using simultion method. 36

47 LIST OF REFERENCES 37

48 LIST OF REFERENCES 1. Prsonson, Peter S., Signl Timing Improvement Prctices, NCHRP Synthesis of Prctice 172, Wshington, D.C., Februry Sckmn, H., Monhn, B. Prsonson, P. S. nd Trevino, A. F. Vehicle Detector Plcement for High- Speed, Isolted Trffic-Actuted Intersection Control Vol. 2, Mnul of Theory nd Prctice, FHWA, My Zegeer, C. V. Effectiveness of Green-Extension Systems t High-Speed Intersections. Reserch Report 472, Lexington, Ky.: Bureu of Highwys, Kentucky Deprtment of Trnsporttion, My Bonneson, J. A. nd P. T. McCoy. Mnul of Trffic Detector Design, Second Edition, Wshington, D. C.: Institute of Trnsporttion Engineers, My Smglik, E., D.M. Bullock, nd T. Urbnik, An Evlution of Lne by Lne Vehicle Detection for Actuted Controllers Serving Multi-Lne Approches, In Trnsporttion Reserch Record: Journl of the Trnsporttion Reserch Bord, No. 1925, TRB, Ntionl Reserch Council, Wshington, DC, pp ,

49 6. Lrge Are Detection t Intersection Approches, A Section Technicl Report, Technicl Committee 17, Southern Section ITE, Trffic Engineering, June, Smll Are Detection t Intersection Approches, A Technicl Committee Report No. 18, Southern Section ITE, Trffic Engineering, Februry VISSIM 4.10 User Mnul, PTV Plnung Trnsport Verkehr AG, Krlsruhe, Germny, pp. 210, Bonneson, J. A., Middleton, D., Zimmermn, K., Chrr, H., nd Abbs, M., Intelligent Detection- Control System for Rurl Signlized Intersections, Report NO. FHWA/TX-03/4022-2, Texs Deprtment of Trnsporttion, Austin, Texs, August Lin, F.B. nd F. Mzdeysn., Dely Models of Trffic-Actuted Signl Controls, Trnsporttion Reserch Record 905, Trnsporttion Reserch Bord, Wshington, DC,

50 VITA Jinwen Si ws born in Jiozhou, Chin, in He received Bchelor of Engineering Degree of Trffic Informtion Engineering from Nnjing University of Aeronutics nd Astronutics t Nnjing, Chin, in After his undergrdute studies, he went to the University of Tennessee to pursue Mster of Science Degree in Civil Engineering, with concentrtion in Trnsporttion Engineering. During his grdute studies, he worked s Grdute Reserch Assistnt. 40