Pedestrian delay at Pelican crossings in areas operating under SCOOT

TRANSPORT AND ROAD RESEARCH LABORATORY Department f Transprt RRL Cntractr Reprt 249 Pedestrian delay at Pelican crssings in areas perating under SCOOT by J G Hunt, J D Griffiths and K M Hughes (University f Wales Cllege f Cardiff) The w~ reprted herein was carded ut under a cntract placed n the University f Wales Cllege f Cardiff by the Transprt and Rad Research Labratry. The research custmer fr this wrk is Traffic Plicy Divisin, DTp. This reprt, like thers in the series, is reprduced with the authrs' wn text and illustratins. N attempt has been made t prepare a standardised frmat r style f presentatin. Cpyright Cntrller f HMSO 1991. The views expressed in this Reprt are nt necessarily thse f the Department f Transprt. Extracts frm the text may be reprduced, except fr cmmercial purpses, prvided the surce is acknwledged. Traffic Operatins Divisin Traffic Grup Transprt and Rad Research Labratry Old Wkingham Rad Crwthrne, Berkshire RG11 6AU 1991 SSN 266-745

SUMMARY This reprt cnsiders the peratin f Pelican crssings as part f a SCOOT crdinated system. Surveys f current practice by Lcal Highway Authrities indicated cncern abut increased pedestrian delay at Pelican crssings perated under SCOOT. Site studies f 15 Pelicans perating under SCOOT cntrl in 5 areas cnfirmed that pedestrian mean delay is substantially higher when cmpared with delay at Fixed Time Pelicans perating independently. A mdel based cmparisn f the csts f pedestrian delay at a typical Pelican crssing perating independently in an uncrdinated area, under SCOOT cntrl, and under TRANSYT cntrl shwed that SCOOT has lwer pedestrian delay csts than TRANSYT and that the financial disbenefit f UTC cntrl t pedestrians is small in cmparisn with the verall benefits t vehicle ccupants. Alternative strategies fr reducing the delay t pedestrians at Pelican crssing within SCOOT systems have been assessed. t is cncluded that SCOOT culd prvide benefits fr bth pedestrians and vehicle users prvided dual vehicle precedence bards are fitted t all Pelican cntrllers and Pelicans are always dubled cycled unless the degree f saturatin exceeds 9%.

Ownership f the Transprt Research Labratry was transferred frm the Department f Transprt t a subsidiary f the Transprt Research Fundatin n 1 st April 1996. This reprt has been reprduced by permissin f the Cntrller f HMSO. Extracts frm the text may be reprduced, except fr cmmercial purpses, prvided the surce is acknwledged.

PEDESTRAN DELAY AT PELCAN CROSSNGS N AREAS OPERATNG UNDER SCOOT CONTENTS NTRODUCTON Page N: 2. BACKGROUND 2.1 Pelican crssings 2.2 SCOOT 2, CURRENT PRACTCE BY HGHWAY AUTHORTES OPERATNG SCOOT 3 4. STE STUDES OF PELCAN CROSSNGS OPERATNG UNDER SCOOT CONTROL 4 4.1 Objectives 4 4.2 Site selectin 5 4.3 Data cllectin and abstractin 5 4.4 Data analysis and presentatin 5.. MODELLNG PELCAN CROSSNG OPERATON N AREAS UNDER SCOOT CONTROL 7 A COMPARSON OF THE COST OF PEDESTRAN DELAY AT A TYPCAL PELCAN CROSSNG OPERATNG NDEPENDENTLY N AN UNCOORDNATED AREA, UNDER SCOOT CONTROL AND UNDER TRANSYT CONTROL 1 6..-ntrductin 6.2 Mdelling prcedure 1 1 6.3 Cmparisn f csts 12. AN EVALUATON OF ALTERNATVE STRATEGES FOR OPERATNG A PELCAN CROSSNG N AN AREA UNDER SCOOT CONTROL 13 7.1 Alternative strategies 13 7.2 Mdelling prcedure 15 7.3 Assessment f strategies 16, CONCLUSONS 1 7 9. 1. 11. RECOMMENDATONS 1 8 ACKNOWLEDGEMENTS 1 9 REFERENCES 1 9 FGURES TABLES APPENDX - QUESTONNARE SENT TO LOCAL HGHWAY AUTHORTES OPERATNG SCOOT

. NTRODUCTON SCOOT[] was develped by the Transprt and Rad Research Labratry in assciatin with the traffic signal cmpanies t prvide an nline Urban Traffic Cntrl System. Fllwing extensive trials in Glasgw and Cventry, SCOOT has becme accepted, bth in the United Kingdm and verseas, as a system which prduces significant savings in vehicle delay when cmpared with fixed cycle systems such as TRANSYT[2]. Selby and Pwell[3] have reprted that SCOOT is estimated t reduce average queues by abut 2% cmpared with a fixed time system based n plans which can nly represent average cnditins ver a cnsiderable time. Widespread acceptance f the benefits f SCOOT, in terms f reduced vehicle delay and cngestin, is demnstrated by the grwing number f twns and cities with peratinal systems; in early 199 it was estimated that mre than 4 systems had been installed. While the traffic cntrl and delay minimisatin benefits f UTC systems such as SCOOT and TRANSYT have been clearly demnstrated the rad safety implicatins and effects n ther rad users are less well established with Lcal Highway Authrity Engineers smetimes quting specific examples f perceived changes in accident frequency accmpanying r fllwing the installatin f a UTC system. A recent reprt by Hunt, Griffiths and Mses[4] has, hwever, established that SCOOT has nt adversely affected the accident frequency in six United Kingdm cities fr the three years fllwing installatin; in ne area, Suthamptn, there was a significant reductin in accident frequency. Bth TRANSYT and SCOOT seek t minimise a 'Perfrmance ndex' which is based entirely n 'csts' assciated with vehicles. The treatment f pedestrians is dependent n the skill and mtivatin f Highway Authrity Engineers perating within the cnstraints impsed by the UTC system and the assciated signal cntrllers. At signal cntrlled junctins the greater efficiency f the UTC system, in cmparisn with islated signal cntrl, may prvide additinal pprtunities fr intrducing pedestrian stages. Hwever there is general agreement that, because pedestrian precedence perids are available with less frequency than at islated Pelican crssings, pedestrians wishing t crss the rad at midblck are less well served in areas under UTC cntrl. A previus study[5,6] has cnsidered, in detail, pedestrian delay at Pelicancrssings perating in areas under TRANSYT cntrl. This Reprt attempts t evaluate pedestrian delay csts in areas under SCOOT cntrl and explres alternative strategies which culd be used t reduce pedestrian delay.. BACKGROUND 2.1 Pelican crssinss Pelican crssings are installed in the United Kingdm t prvide pedestrians with a safe methd f crssing the rad with acceptable delay between junctins withut impsing excessive delay n vehicular traffic. Criteria fr the prvisin and design f Pelican crssings, and the peratinal cycle, are described in the Department f Transprt Advice Nte[7] and Departmental Standard[8]. The peratinal cycle, in which perids f pedestrian precedence

alternate with perids f vehicle precedence, is shwn in Figure. The 'Green Man', t enable a pedestrian t establish his presence n the carriageway, and 'Flashing Green Man', t enable any pedestrians already n the carriageway t cntinue t the farside, have duratins which depend primarily n rad width. Pelican crssings may be either Fixed Time (FT) r Vehicle Actuated (VA) and fr thse pedestrians wh nly start t crss the rad during the 'Green Man' the maximum delay will depend n the vehicle precedence perid (f) which is set n the Pelican cntrller as described belw. (i) (ii) At FT Pelicans : the length f the minimum vehicle precedence perid (f) is usually set t ensure vehicle capacity is adequate at all times. Currently f is subject t a maximum f 6 secnds with lwer values recmmended fr mst situatins. At mst FT Pelicans perated independently f is set at 28 r 32 secnds. At VA Pelicans : the length f the minimum vehicle precedence perid (fmin) is set t ensure that all halted vehicles between the stpline and the detectr clear the stpline; currently fmin = 7 secs is recmmended. Once fmin has expired the signals will respnd t a pedestrian demand if a suitable gap in vehicle flw is available. The maximum pedestrian waiting time depends n the duratin f fmax which fr mst VA cntrllers des nt begin until the pedestrian demand is registered. (The mst recent peratin cycle[8] prvides fr fmax t begin at the start f vehicle green but this mdificatin has been implemented by very few Highway Authrities). fmax may be set in the range 1 t 6 secs. Fr bth FT and VA Pelicans the signal timings must be set up n site at each Pelican cntrller and thus cannt easily be altered. At islated Pelican crssings the maximum delay between a pedestrian registering a demand and btaining a 'Green Man' signal wuld nly rarely exceed 3 secnds with average delays fr thse pedestrians waiting fr the 'Green Man' signal usually less than 15 secnds. n practice, because many pedestrians crss during the 'Red Man' signal if there is a suitable gap in vehicle arrivals, verall pedestrian mean delay is much reduced particularly where the vehicle flw is lw. 2.2 SCOOT Within SCOOT traffic flws appraching each nde (signal cntrlled junctin r Pelican crssing) are mnitred by detectrs, lcated as far upstream f the stplines as practical. The SCOOT system uses data frm detectrs t determine the ptimum cycle time fr..the area and the ptimum ffsets and green splits fr each nde. SCOOT represents a Pelican crssing as a junctin signal with tw stages in each cycle; the pedestrian stage is f fixed duratin based n an estimated pedestrian precedence perid. Current perating practice fr Pelican crssings which are linked within a SCOOT netwrk requires that (i) the Pelican perates n a Fixed Time basis, and

(ii) the pedestrian precedence perid ccurs at a time, and with a frequency, that satisfies the demands f crdinatin f vehicular traffic. These cnstraints ften have the effect f allwing a pedestrian precedence perid t ccur nly nce per UTC cycle. Upper (typically 88-12 secnds) and lwer (typically 32-44 secnds) limits n cycle time can be preset by the traffic engineer. Within these limits the cycle time is altered by SCOOT t ensure that the mst heavily laded nde perates at a maximum degree f saturatin f abut 9 per cent. Sme ndes, including Pelicans, may have a degree f saturatin which is sufficiently lw t allw duble cycling n a cycle time which is half that fr the area. Duble cycling f Pelicans is hwever subject als t the cnstraints impsed by the Pelican cntrller, which fr Fixed Time cntrllers may preclude duble cycling at area cycle times belw 8-9 secnds dependent n rad width and the default f set by the traffic engineer. SCOOT may autmatically alter a junctin frm single t duble cycling r vice versa, and can increase r decrease the cycle time f a sub area if calculatins shw that alteratins between single and duble cycling will yield a net benefit. The traffic engineer can, if he wishes, restrict specified ndes t either single r duble cycled peratin. This is particularly imprtant fr Pelican crssings since SCOOT des nt explicitly cnsider pedestrian requirements in its ptimisatin prcedures and a reductin in cycle time belw 8 secnds may, because f PeliCan cntrller limitatins, prevent duble cycling f a Pelican. The verall effect, when the cnstraints impsed by SCOOT are superimpsed n the perating limitatins f mst Pelican cntrllers, is t substantially increase pedestrian delay in cmparisn with that experienced at islated Pelican crssings. The pedestrian maximum signal impsed delays may be as high as 1 secnds with an average f abut 5 secnds at lcatins with a high cycle time where duble cycling is nt specified. Overall pedestrian mean delays, taking accunt f thse pedestrians wh crss during the 'Red Man' stage are much lwer but still give cause fr cncern when cmpared with delays at islated Pelicans as shwn in Figure 2, which is based n simulatin.. CURRENT PRACTCE BY HGHWAY AUTHORTES OPERATNG SCOOT n rder t mnitr current practice in the peratin f Pelican crssings under SCOOT cntrl a brief questinnaire was sent t each f the 27 Highway Authrities with SCOOT installatins; replies were subsequently received frm 26 f the Authrities. The questinnaire is shwn in the Appendix and the results are summarised in Table. The main cnclusins are as shwn belw: (i) (ii) Pelican crssings represent 28% f the SCOOT ndes cvered by the survey (283 Pelicans ut f a ttal f 111 SCOOT ndes). Pelicans under SCOOT cntrl represent a very small prprtin f the ttal number f Pelicans currently in peratin thrughut the United Kingdm. (Estimated as 85). 73% f Pelicans were perated Fixed Time in the perid befre the installatin f SCOOT (194 ut f 266 with 17 nt knwn). This restricted the selectin f f t a minimum f 2 secnds

at mst sites. Mst Authrities used f values f 3 secnds r mre fr sme Pelicans under SCOOT cntrl. The use f higher values fr f restricted the range f cycle times at which the Pelicans culd be duble cycled. (iii) Fr mst Authrities the minimum cycle time was in the range f 32 t 56 secnds and the maximum was in the range f 14 t 12 secnds. Mst Pelicans culd duble cycle at the higher cycle times. The lwer cycle times wuld prvide pedestrian delays at linked Pelicans which are similar t thse at islated Pelicans. At intermediate cycle times duble cycling wuld nt nrmally be pssible and pedestrian delays wuld be higher than is desirable. A number f Authrities frce the cycle time higher during the ff peak perid t facilitate duble cycling f Pelicans. At lwer cycle times f 32-4 secnds a Pelican with a vehicle precedence perid f f=2 wuld be unable t prvide pedestrian precedence in cnsecutive cycles. (iv) (v) (vi) Almst all Authrities indicated that the peratin f Pelican crssings influenced their chice f cycle times. Abut 3% f Authrities indicated that they perated ne r mre Pelicans ff link in a SCOOT area. n mst cases the Pelicans perated ff link carried a very lw pedestrian flw. The Lndn Traffic Cntrl Systems Unit cntinue t cnvert their Pelican crssings t perate dual vehicle precedence perids with 1 secnds emplyed when under SCOOT cntrl and 2-32 secnds set fr lcal cntrl. This facilitates duble cycling at lw UTC cycle times while retaining adequate vehicle capacity in the event f system breakdwn necessitating a temprary return t lcal cntrl. (vii) A number f Authrities indicated, either implicitly r explicitly, in their replies that they were cncerned abut increased pedestrian delay under SCOOT r ther UTC systems. 4. STE STUDES OF PELCAN CROSSNGS OPERATNG UNDER SCOOT CONTROL 4.1 Objectives The main bjectives f the study described in this Reprt are t quantify any pedestrian disbenefits at Pelican crssings perated under SCOOT, and evaluate any ptential benefits f alternative strategies fr the peratin f Pelican crssings within areas under SCOOT cntrl. SCOOT, and ther UTC systems impse additinal cnstraints n Pelican peratin and als have the effect f frming vehicle platns which mve thrugh the netwrk. Pedestria~ delay and behaviur is likely t be affected by the changes in perating prcedure and the sequence and distributin f the vehicle inter arrival times. The bjectives f surveys at sites where Pelican crssings were perated under SCOOT, were t (i) (ii) identify and quantify any differences in pedestrian delay r behaviur in cmparisn with islated Pelican crssings, and prvide data fr peratin at typical Pelican crssings within a SCOOT system fr applicatin in mdelling studies.

4.2 Site Selectin The criteria fr site selectin were initially similar t thse specified fr the surveys f islated Fixed Time, and Vehicle Actuated Pelican Crssings[9,1]. Hwever it was als necessary t mnitr changes in SCOOT cycle time during the perids that data were cllected at the Pelicans and nt all areas perating SCOOT were able t ffer that facility. The chice f sites was als limited by the relatively lw number f Pelicans currently perating under SCOOT s that 'typical' sites were difficult t identify; it was als necessary t avid sites which had a lw level f pedestrian flw. Sites were selected in each f the fllwing areas: Cventry, Leicester, Suthamptn, Walsall, Wlverhamptn, and Wrcester. A full list f survey sites is shwn in Table 2. 4.3 Data cllectin and abstractin Data were captured by vide recrding activities at the selected sites fr perids f up t six hurs between the mrning and evening peak perids. The ff peak perid was selected as mst apprpriate fr cmpatibility with surveys at islated Pelicans, because f cngestin/slw mving traffic at sme sites during the peak, and principally because the scpe fr imprving pedestrian prvisin is much greater during the ff peak perids when vehicle demands and cycle times are lwer. At mst survey sites the M14 messages utput by the SCOOT system were recrded n an Amstrad PPC by cnnectin t a terminal line r an OTU at the Traffic Cntrl Centre. This prvided a recrd f cycle time and data defining the cyclic flw prfile at the Pelican lcatin during the'perid fr which activities were vide recrded. Data were subsequently abstracted frm the vide recrdings using event recrders linked t a micrcmputer. 4.4 Data analysis and presentatin An verview f the characteristics and peratin f each f the Pelican crssings fr which data were cllected is shwn in Table 2. The range f vehicle flw during the bservatin perids was similar t that bserved at islated Fixed Time and Vehicle Actuated Pelicans. There were hwever few Pelican crssings where the pedestrian flw was high included in the survey. This may be due t chance as the selectin f sites was very limited, r be a reflectin f lw utilisatin by pedestrians because f perceived high delays. n all areas except Suthamptn the Pelican crssings previusly perated n an islated Fixed Time basis; under SCOOT at the majrity f survey sites the vehicle precedence perid f was set at r clse t the minimum pssible time f 2 secnds. At mst sites duble cycling f the Pelican was nly pssible if the UTC cycle time exceeded 8 secnds. Hwever, where a vehicle precedence perid f

3 secnds was used, as fr example at Lndn Rad a SCOOT cycle time f at least 1 secnds wuld be required in rder t duble cycle the Pelican. The cnstraint impsed n duble cycling during ff peak perids by the limitatin f the minimum vehicle precedence perid is clearly evident fr a number f sites where pedestrian requirements are less well served in ff peak than in perids f peak vehicle flw. n Suthamptn, where mst Pelicans perated n an islated Vehicle Actuated basis befre the intrductin f SCOOT, the use f a lwer vehicle precedence perid f 12 secnds prvided fr mre flexibility in allwing Pelicans t duble cycle under SCOOT cntrl. The data which were lgged by event recrders were brken dwn int perids f apprximately 3 minutes fr each site and prcessed t determine fr each recrding perid:- (i) (ii) the number f Pelican cycles called, the range f SCOOT cycle times, and fr each side f the rad: (iii) the vehicle flw (veh/h), (iv) (v) (vi) the pedestrian flw (ped/h), the mean delay t delayed pedestrians (secs), the percentage f pedestrians delayed, (vii) the mean delay t pedestrians (secs), and (viii)the percentage f pedestrians arriving/departing during each pedestrian signal aspect. Figures 3 t 5 shw percentage f delayed pedestrians, percentage f pedestrians crssing during 'Red Man' aspect, and pedestrian mean delay respectively pltted n a base f cmbined vehicle flw fr the survey sites. Figure 3 indicates that, as expected, few pedestrians are able t crss at Pelican crssings withut incurring sme delay. Figure 4 shws that a high percentage f pedestrians crssed during the 'Red Man' aspect at sme f the survey sites but there is n clear increase with increase in vehicle flw, as was evident in cmparable data fr islated Pelican crssings. This may be related t bth the varying effect f vehicle platning n crssing pprtunities at a particular vehicle flw and t the length f time each pedestrian wuld have t wait fr the 'Green Man' aspect. The latter effect is mainly a cnsequence f the way in which the Pelican is perated within SCOOT.

Figure 5 shws a wide variatin in pedestrian mean delay, bth between survey sites, and between data fr each 3 minute perid at sme f the sites. The latter effect which is mre prnunced than at islated Pelican crssings may be attributed t varying perating cnditins under SCOOT cntrl. At mre than half the survey sites the 95 percentile pedestrian delay crrespnding t the mean delays shwn in Figure 5 will be abve the 3 secnd delay level at which it is cnsidered that pedestrians becming increasingly impatient and may begin t take unacceptable risks. SMPEL, a simulatin prgram which was carefully calibrated fr islated Fixed Time and Vehicle Actuated Pelican crssings[11], has been applied t calculate, fr each 3 minute recrding perid at each survey site, data fr an islated Pelican crrespnding t thse recrded n site fr Pelicans under SCOOT and shwn in Figures 3 t 5. The input data supplied t the simulatin mdel was identical t rj that applying at the SCOOT Pelicans but in simulatin each Pelican was allwed t perate as islated Fixed Time. The cmparisn between data fr islated Fixed Time Pelicans, based n simulatin, and data fr the Pelicans under SCOOT cntrl as recrded n site, is shwn in Figures 6 t 9. These Figures clearly demnstrate that - (i) (ii) pedestrian mean delay is substantially higher at mst f the Pelicans perating under SCOOT cntrl, there is a substantially higher prprtin f pedestrians crssing during the 'Red Man' aspect at the Pelicans perating under SCOOT cntrl, (iii) the percentage f pedestrians delayed is similar at islated Fixed Time Pelicans and the Pelicans under SCOOT cntrl, and (iv) the frequency f pedestrian precedence perids (number f cycles called) is substantially lwer fr Pelicans under SCOOT cntrl.. MODELLNG PELCAN CROSSNG OPERATON N AREAS UNDER SCOOT CONTROL The site surveys cnfirmed the bservatins f Lcal Highway Authrities that pedestrians delays are ften higher at Pelican crssings under SCOOT cntrl than fr islated Pelican crssings in nn UTC areas. n rder t cst any disbenefit t pedestrians, against a backgrund f significant savings in vehicle delay, under SCOOT, and t examine pssible alternative perating strategies it was necessary t apply a simulatin mdel. Pelican crssing peratin can be mdelled using the simulatin prgram SMPEL[5,6] which was develped t deal with mst types f Pelican crssing perating independently and extended t include peratin within a UTC area. The mdel f peratin within a UTC area requires the synchrnisatin f the UTC and Pelican cycle times and the applicatin f apprpriate cnstraints n the time when the pedestrian precedence perid can be initiated. The generatin f vehicle inter arrival times als requires careful attentin as the size and distributin f time gaps between vehicle arrivals will

influence the number f pedestrians crssing during the 'Red Man' perid. Within a UTC area vehicle arrivals will usually vary n a cyclic basis with perids f high vehicle flw ('platns f vehicles'), fllwed by perids f lw vehicle flw. This cntrasts with areas where there is n crdinated cntrl where vehicle inter arrival times can usually be adequately represented by a displaced negative expnential distributin. (Mst areas which are nt crdinated, have islated VA signal cntrl at junctins, frequent uncntrlled junctins, demand dependent facilities, and ften allw a substantial amunt f time limited parking n street). The traffic mdels fr bth TRANSYT and SCOOT calculate vehicle delays by the manipulatin f cyclic flw prfiles representing vehicle arrivals and departures at the signal stp line. The TRANSYT mdel has an utput ptin which will prvide details f the cyclic flw prfiles at each signal stp line. These prfiles can be used t determine pstulated prfiles fr vehicle arrivals at a dwnstream Pelican crssing lcatin by applying a frmula[2] which displaces the prfile in time and applies dispersin t allw fr traffic interactin dwnstream f the junctin. Fr Pelican crssings perating within a UTC area under a TRANSYT develped plan it is thus a relatively simple matter t btain the prfile representing the pstulated vehicle arrival prfile at the Pelican lcatin. The simulatin mdel must then generate vehicle inter arrival times fr each side f the rad which satisfy the fllwing cnditins. (i) the inter arrival times, when aggregated int intervals ver the netwrk cycle time, must prduce the required cyclic flw prfile, (ii) the verall vehicle inter arrival distributin shuld be similar in simulatin and in practice, (iii) the simulated sequence pattern f vehicle inter arrival times shuld be similar t that bserved in practice, and (iv) the frequency distributin f vehicle arrivals in each cycle shuld be similar fr simulatin and in practice. A methd t generate vehicle inter arrival times which satisfied the abve criteria was develped and has been described by Hunt and Griffiths[12]. The technique was hwever cmputatinally inefficient. Fr fixed time UTC systems, TRANSYT and SMPEL used tgether prvide a valid and representative apprach t determining the csts f any pedestrian disbenefits and evaluating alternative perating strategies. Fr a UTC netwrk under SCOOT cntrl the applicatin f the abve methd is much mre cmplex than in the case f peratin under a TRANSYT develped plan. Under SCOOT cntrl the netwrk cycle time may be cntinuusly varying s that the definitin f an average prfile is inapprpriate and the determinatin f a typical prfile is subjective. The mst satisfactry apprach fr the current study wuld have been t use a SCOOT Simulatin System, such as that emplyed fr demnstratin purpses at the SCOOT wrkshp. An apprpriate netwrk culd then have been set up and the SCOOT simulatin applied t predict vehicle arrival times at the Pelican and the times f the pedestrian stages. These data culd then have 8

been lgged t a database which SMPEL culd access thus circumventing the requirement t generate vehicle inter arrival times. Unfrtunately because f the cmmercial sensitivity f SCOOT it did nt prve pssible fr the authrs t have access t the SCOOT simulatin system and an alternative less satisfactry apprach had t be used. The alternative apprach invlved assuming that SCOOT is a dynamic versin f TRANSYT and can therefre be represented by a series f runs f TRANSYT in each f which an apprpriate cycle time is specified. The results frm these TRANSYT runs can then be aggregated based n apprpriate time segments t take accunt f the varying cycle time under SCOOT peratin. As a first stage the apprach using SMPEL and TRANSYT was validated by cmparing fr sme f the survey sites: (i) percentage f pedestrians delayed, (ii) percentage f pedestrians crssing during 'Red Man' stage, (iii) pedestrian mean delay, and (iv) number f Pelican cycles called as recrded n site and as predicted by Simulatin. The site data were btained directly frm recrded infrmatin fr each lcatin. Fr simulatin the basic signal timings, pedestrian flw, rad width etc., were as recrded fr each lcatin; the cycle time, ffset t the start f the pedestrian precedence perid, and cyclic flw prfiles fr each side f the rad had t be derived frm the M14 messages utput by the SCOOT system, r by direct measurement n site. Effectively each 3 minute recrding perid was regarded as a time segment during which the cycle time, ffset, and cyclic flw prfile remained cnstant. During an ff peak perid this appeared reasnably valid as illustrated by Figures 1-14 which shw the variatin in UTC cycle time and the Pelican cycle time (this indicates whether the Pelican was single r duble cycled) ver the perids fr which M14 messages were available at the survey sites. The determinatin f typical cycle times, ffsets and cyclic flw prfiles fr each side f the rad was bth subjective and time cnsuming. The M14 messages fr each site were prcessed t prvide, fr each 4 secnd time interval, (i) (ii) the interval number, vehicle flw n side A, (iii) vehicle flw n side B, and (iv) the stages f the Pelican signals as shwn in Table 3. The cyclic flw prfiles fr each side f the rad, cycle and ffset infrmatin were displayed in histgram frm n a VDU and scanned manually t identify a typical cycle.

The crrespnding cyclic flw prfiles fr each side f the rad were then adapted fr input t SMPEL. A typical set f prfiles fr ne f the survey sites is shwn in Figure 15. Because f the time invlved in the rather labrius manual prcess fr determining typical cyclic flw prfiles the validatin was restricted t ne 3 minute perid at each f 12 lcatins. Wrcester, where cycle times, ffsets and cyclic flw prfiles had t be derived frm data abstracted frm vide recrdings in the absence f M14 messages, was included in rder t represent all survey areas in the validatin. The results f the limited validatin are shwn in Figures 16 t 19. The Figures shw - (i) (ii) reasnable agreement between site recrded data and simulatin predicted values fr number f cycles called, percentage f pedestrians delayed and pedestrian mean delay, and generally pr agreement, withut any indicatin f undue bias, between site recrded data and simulatin values fr the percentage f pedestrians crssing during the 'Red Man' stage. The differences, fr individual sites between bserved and predicted percentage f pedestrians crssing during the 'Red Man' perid illustrates the difficulty in predicting this parameter based n typical ffsets and cyclic flw prfiles. Relatively small time differences in the arrival f the main platn in each cycle can have a substantial effect n the availability f crssing pprtunities fr pedestrians. Overall the validatin indicated n real evidence f bias and n this evidence it is reasnable t apply the simulatin mdelt evaluate pedestrian crssing delays and cmpare alternative strategies fr Pelicans perating under SCOOT.. 6.1 A COMPARSON OF THE COST OF PEDESTRAN DELAY AT A TYPCAL PELCAN CROSSNG OPERATNG NDEPENDENTLY N AN UNCOORDNATED AREA, UNDER SCOOT CONTROL, AND UNDER TRANSYT CONTROL ntrductin Discussins with Lcal Highway Authrity Engineers, data cllectin and simulatin studies have all indicated that, when cmpared with the delays experienced at Pelican crssings perating independently in an uncrdinated area, pedestrians suffer increased delays at Pelican crssings perated as part f a UTC system. While the percentage increase in pedestrian mean delay appears t be large, n attempt has been made t link these delays with assciated waiting time csts. These pedestrian delay csts must be cnsidered against a backgrund f the substantial savings in vehicle delay and cngestin which accmpany the intrductin f a UTC system. Any prpsal fr mdificatins t SCOOT r its perating hardware and prcedures, particularly if there is a disbenefit t vehicles and their ccupants, wuld need t be justified in terms f substantial reduced csts t pedestrians. This sectin f the Reprt cnsiders the ptential fr pedestrian delay cst savings by cmparing the situatin at a typical Pelican perating (i) independently in an uncrdinated area with vehicle actuated signals at junctins, 1

(ii) as part f a TRANSYT crdinated area, and (iii) as part f a SCOOT crdinated area. 6.2 Mdelling prcedure The cmparisn was based n a typical Pelican which was assumed t be lcated n a tw way rad f width 1.m. The Pelican cntrller was assumed t have the timings set as belw: a = 3 secs b = sec c = 5 secs d = 7 secs e = sec f = 28 secs The delay t pedestrians and assciated behaviur was determined using the simulatin prgram SMPEL which was referred t earlier in this Reprt. Each f the Pelican perating envirnments listed in sectin 6.1 was mdelled as described belw. (i) Pelican perating independently in an uncrdinated area t was assumed that vehicle actuated signal cntrl was used at junctins within the area s that the cycle times and cyclic flw prfiles wuld vary cntinuusly with time. n previus studies[t1,12] f Pelicans perating independently in areas where traffic signals were nt crdinated, the displaced negative expnential distributin was fund t prvide the best descriptin f vehicle inter arrival times. The accuracy f the representatin, by a displaced negative expnential distributin, will vary depending n link lengths, traffic flw variatins and vehicle green duratins at the upstream junctins. Hwever fr mst situatins any errrs in using the displaced negative expnential t represent vehicle inter arrival times, at a Pelican crssing lcated in an area with VA signal cntrl at junctins, wuld be small and accrdingly the displaced negative expnential distributin was used in this study. (ii) Pelican perating as part f a TRANSYT crdinated area A simple UTC netwrk cnsisting f fur cnsecutive junctins with a Pelican situated half way between the middle tw junctins, as shwn in Figure 2, was used t represent a typical UTC netwrk. An alternative wuld have been t cnsider a netwrk f just tw junctins with the Pelican situated half way between them. Hwever this wuld have meant that the vehicle arrival prfiles at the stplines f all the links feeding the Pelican links wuld have been unifmn, and thus nt ttally representative f a UTC netwrk. The fllwing assumptins were made abut the netwrk: (a) (b) The Pelican links were 15m lng and all ther links were 2m lng. The flws alng each link were balanced. 11

(c) Fr all links entering a junctin 8% f the traffic travelled straight n, 1% turned left and the remaining 1% turned right. The netwrk was set up fr TRANSYT with the pedestrian stage cnsidered as an intergreen perid. TRANSYT was used t calculate the ptimum signal settings fr each specified cycle time and vehicle flw. TRANSYT was then run t determine the ffsets and vehicle arrival cyclic flw prfile fr each f the links feeding the Pelican crssing. Pedestrian delay was calculated frm SMPEL run with the apprpriate pedestrian flw. UTC cycle time, ffsets and vehicle arrival prfiles were cnsistent with the TRANSYT utput. (iii) Pelican peratin 8 as part f a SCOOT crdinated area As described previusly the apprach used was t assume that SCOOT is a dynamic versin f TRANSYT s that pedestrian delay and behaviur culd be based n an aggregatin f data fr a series f TRANSYT and SMPEL runs crrespnding t a specified cycle time and ffset variatin. As it was difficult t prepare a specificatin fr a typical SCOOT envirnment the cmparisn was based n the Pelicans fr which data were cllected during the site survey. SMPEL was applied t calculate pedestrian mean delay etc., fr each f the fllwing sets f ttal pedestrian and vehicle flw - Pedestrian flw (ped/h) 2, 4, 1 Vehicle flw (veh/h) 5, 1, 15, 2 TRANSYT and SMPEL were run in cmbinatin fr each f the abve flws with the fllwing UTC cycle times - UTC cycle time (secs) 48, 6, 72, 84, 96, 18, 12 The pedestrian mean delay, percentage f pedestrians delayed and percentage f pedestrians crssing during the 'Red Man' fr each f the abve data cmbinatins are shwn in bar chart frm in Figures 21-23. 6.3 Cmparisn f csts The unit f cst per pedestrian hur f delay may be taken as 4.24. This is based n Highway Ecnmics Nte 2 which gives pedestrian wrking time a value f 8.24 per hur and nn wrking time 4.15 per hur; 98% f pedestrian time is assumed t be nn wrking t~me. The delays calculated frm simulatin were used t prduce the database shwn in Table 4 where the pedestrian delay cst fr each cmbinatin f vehicle and pedestrian flw is shwn fr a Pelican perating within an area with each TRANSYT cycle time, and fr a Pelican perating independently in a uncrdinated VA signal area. Fr each f the survey sites the average pedestrian delay csts crrespnding t each Pelican perating mde can be derived frm Table 4 by interplatin based n the average vehicle and pedestrian flw recrded n site and, in the case f TRANSYT and SCOOT, 12

apprpriate cycle times. Fr TRANSYT the csts were based n a cycle time twards the upper end f the range f SCOOT cycle times recrded n site as shwn in Figures 1-14. Fr SCOOT the csts were determined by aggregatin f csts based n TRANSYT cycle times and apprpriate time segments t apprximate the variatins recrded n site. The results are summarised in Table 5. t shuld be nted that the cmparisn is intended t determine the pssible ptential fr savings in pedestrian delay csts under SCOOT s that the tabulated csts fr TRANSYT and SCOOT represent the upper end f the expected range. n particular it has been assumed that all Pelicans are single cycled; this des nt invalidate the cmparisn between TRANSYT and SCOOT since it is assumed that SCOOT culd if required be cnstrained t perate as a TRANSYT system, e.g. the cycle time culd be held at a high level t allw duble cycling f Pelicans. t shuld als be nted that the csts are based n net pedestrian delay after allwing fr pedestrians wh crss during the 'Red Man' stage. Since crssing during the 'Red Man' ccurs mre frequently at Pelicans linked within UTC areas, this has the effect f reducing the cst differentials. At mst f the Pelicans pedestrian delay csts are higher under SCOOT r TRANSYT cntrl than fr independent peratin in an uncrdinated area. SCOOT, because f its ability t reduce cycle times during perids f lwer vehicle flw, has lwer pedestrian delay csts than TR~NSYT. Because f the relatively lw pedestrian flw at the survey sites and the mderating effect f crssing during the 'Red Man' at Pelicans under:utc cntrl, the ptential financial disbenefit t pedestrians f perating Pelicans under SCOOT cntrl is small in cmparisn with the financial benefits t vehicles and their ccupants as may be illustrated by cnsidering the SCOOT system in Wrcester. t has been estimated[13] that the replacement f islated signal cntrl by SCOOT in Wrcester has prduced a net benefit in terms f reduced vehicle delay f millin per annum at current csts. This is based n savings fr week day traffic during the hurs 745-1745. The annual disbenefit t pedestrian in terms f delay csts at the 8 Pelicans perated under SCOOT in Wrcester is 6.5 thusand n the basis f the average pedestrian delay csts given fr the three Wrcester sites in Table 5. The pedestrian delay csts were calculated fr the ff peak perid 1-16; there is little scpe fr reducing pedestrian delay at peak times. Clearly, the mnetary disbenefit t pedestrians is marginal in cmparisn with the savings in vehicle delay csts which fllwed the intrductin f SCOOT in Wrcester.. 7.1 AN EVALUATON OF ALTERNATVE STRATEGES FOR OPERATNG A PELCAN CROSSNG N AN AREA UNDER SCOOT CONTROL Alternative strategies Based n Wrcester as a typical example, the disbenefit in financial terms f perating Pelicans n a single cycle basis under SCOOT cntrl is numerically insignificant when cmpared t the benefits which accrue t vehicles. t fllws that any reductin in 13

pedestrian delay at Pelicans in areas perating under SCOOT can nly be justified in financial terms, if it is nt accmpanied by a substantial increase in vehicle delay csts. n rder t reduce pedestrian delay at Pelicans linked within a SCOOT cntrlled area Lcal Highway Authrities have adpted a number f strategies mst f which are expected t have a marginal effect n vehicle delay and cngestin. These may be summarised as belw. (a) Duble cycling This prcedure, where the Pelican is perated n a cycle time f ne half f the regin cycle time allws fr tw pssible pedestrian precedence perids in each UTC cycle and s reduces pedestrian delay. SCOOT will seek t duble cycle ndes, prvided the degree f saturatin des nt exceed 9%, n the basis that there will be a reductin in vehicle waiting times. The Lcal Highway Engineer may als cnstrain SCOOT s that a particular nde is always duble cycled. The principle f duble cycling is accepted fr SCOOT althugh there is little published infrmatin n the lss f crdinatin, and increased vehicle delay, where a nde is perating a duble cycle between adjacent ndes which are perating a single cycle. This culd be particularly relevant if Pelicans were t be duble cycled fr regin cycle times belw 8 secnds. Prvided the SCOOT netwrk cycle time des nt exceed 12 secnds the duble cycling f Pelicans fr netwrk cycle times f 6 secnds and abve wuld prvide pedestrian maximum waiting perids bradly cmparable with thse expected at independent Pelicans n rads carrying a high traffic vlume. Currently, except in Lndn where dual vehicle precedence bards are fitted, limitatins f the Pelican cntrller prevent duble cycling fr cycle times belw 9 secnds. (b) Duble windwing Where duble cycling is nt pssible, because f the Pelican cntrller cnstraints, duble windwing is smetimes emplyed as an alternative. Under duble windwing tw windws, during which a pedestrian precedence perid may be initiated, are available in each cycle. Hwever nly ne f these perids is available in any ne UTC cycle since the minimum time between perids which is determined by the length f the vehicle precedence perid f must expire. With duble windwing it is pssible fr the peratinal pedestrian precedence perid t becme lcked nt the less favurable f the tw available windws. Duble windwing culd be expected t be f mst benefit fr Pelicans with a relatively lw pedestrian flw.,. A number f ther strategies which may benefit pedestrians are described belw: (c) Wider windws The standard windw fr initiating a pedestrian precedence perid at a Pelican is f tw secnds duratin. The use f this small windw has the disadvantage that few pedestrians will experience a rapid respnse t a demand via the push buttn in cmparisn, fr example, with a FT Pelican perating independently during the perid after the vehicle minimum precedence time has expired. During perids f lw 14

vehicle flw it may be pssible t use a wider windw, f, fr example, 12 secnds duratin. (d) (c) Extended 'Green Man' Under SCOOT cntrl there may be perids f lw vehicle flw between the arrival f the main platns. This culd prvide an pprtunity t ffer mre pedestrians the chance t crss withut delay by extending the duratin f the 'Green Man' perid. ndependent peratin Where Pelicans are situated n lng links r in areas where crdinatin is pr because f traffic management prblems, there may be ptential fr allwing the Pelican t perate independently withut impsing a significant disbenefit n vehicular traffic. (f) Restricted Vehicle Actuated peratin With this strategy the Pelican is allwed t perate n a linked vehicle actuated basis with the start f the pedestrian perid prhibited during ne r tw specified perids in the UTC cycle. This strategy attempts t avid maintaining the 'Red Man' t pedestrians during perids f lw r zer vehicle flw. Under light traffic flws this apprach wuld be likely t prvide many f the benefits t pedestrians assciated with independent peratin whilst maintaining the prgressin f the main vehicle platn. 7.2 Mdellin 8 prcedure The bjective f the evaluatin was t determine within a SCOOT system the net benefits/csts f the abve strategies in cmparisn with a Pelican perating n a single cycle basis. The mdelling prcedure invlved applying TLA~SYT and SMPEL fr the netwrk shwn in Figure 2 with the Pelican timings given in sectin 6.2. Pedestrian and vehicle delays were cmputed fr each f the cmbinatins f vehicle and pedestrian flw, and cycle times given in sectin 6.2. The sequence f prcessing was as described belw. (i) (ii) TRANSYT input files were prepared fr the five nde netwrk fr each cycle time, vehicle flw and duratin f the pedestrian stage. TRANSYT was run fr the set f input files and the results prcessed. n all assessments the sum f the vehicle delay was extracted directly frm the utput fr (a) (b) the whle netwrk, and the tw links assciated with the Pelican The arrival cyclic flw prfiles at the Pelican were recrded fr later use with SMPEL. n rder t btain the vehicle delay fr independent peratin f the Pelican, the flw int the links immediately upstream f the Pelican was assumed t be exclusively frm a unifrm surce. The prcedure allwed TRANSYT t calculate delays fr a Pelican which effectively single cycles in a unifrm flw situatin and may underestimate the vehicle delays which wuld accmpany independent peratin f the Pelican in a situatin where the vehicle flw varies n 15

a cyclic basis. Since TRANSYT assumes that all pssible pedestrian precedence perids are implemented at the designated times in each cycle, vehicle delays assciated with the fllwing strategies culd nt be determined by the TRANSYT mdel - duble windwing (TRANSYT wuld verestimate vehicle delay), extended windw, (TRANSYT requires an effective fixed time fr start f pedestrian precedence perid), and restricted vehicle actuated (as fr extended windw) n rder t ensure cmpatibility between the prpsed strategy and the cnstraints impsed by TRANSYT and the Pelican cntrller it was necessary t adjust the vehicle precedence perid n the Pelican t facilitate duble cycling fr cycle times belw 84 secs. The alternative precedence perid used was 1 secnds t allw the ptential benefits f a dual vehicle precedence bard t be evaluated. Fr restricted VA, a D lp system with fmin f 1 and fmax f 3 was assumed. Duble windwing, which is used as an alternative where duble cycling is nt pssible, was nly evaluated fr cycle times f less than 96 secnds. (iii) SMPEL input files were prepared with cyclic flw prfiles and start f the pedestrian precedence perid based n the assciated TRANSYT utput. (iv) SMPEL was run fr the set f input flws and the assciated delay t pedestrians and vehicles was cmputed. The delays cmputed by TRANSYT and SMPEL were then scaled by the apprpriate cst ( 4.24 per pedestrian hur, 5.6 per vehicle hur). The results are shwn in Tables 6 t 12 and are generally intuitively sensible after the effect f changes in cycle times n vehicle arrival prfiles and the time f initiatin f the pedestrian precedence perid have been taken int accunt. 7.3 Assessment f strategies Mst f the strategies prvide a net benefit fr pedestrians which increases with pedestrian flw because f the scaling effect n csts. The use f a lnger windw f 12 secnds prduces nly a marginal net benefit t pedestrians which is nt viable in financial terms. At lw cycle times the highest benefit t pedestrians is assciated with the use f an extended 'Green Man' but this can have the effect f causing lng vehicle delays because f the preprtin f the cycle which is effectively red t vehicles. Overall, pedestrians benefit mst frm independent peratin except at lw cycle times where duble cycling (with the use f a vehicle minimum precedence perid f 1 secnds) yields substantial benefit t pedestrians. At the higher levels f the vehicle flw the benefit t pedestrians f duble cycling at lw cycle times is cunterbalanced by a substantial disbenefit t vehicles. The TRANSYT calculatin f vehicle delay at the Pelican may be cmpared directly with the SMPEL calculatin f the vehicle delay 16

. nly fr the cases f duble cycling and the 15 secnd 'Green Man' SMPEL is better able t represent cycle by cycle variatins and the effect f pedestrian flw and predicts delays which are f the same rder as TRANSYT as shwn by Figure 24. Hwever with a crdinated system the adptin f an alternative strategy fr perating a Pelican may create additinal delay at ther ndes by reducing crdinatin. Generally Figure 25, which cmpares the change in vehicle delay csts at the Pelican with the change in vehicle delay csts acrss the netwrk, indicates that the effects f the reductin in crdinatin are marginal with the inference that vehicle delays are transferred between ndes with nly a small verall increase. n view f the inability f TRANSYT t calculate the change in vehicle delay csts at the Pelican fr 4 f the 6 strategies, the net benefits shwn in Figures 26 t 29 are based n the SMPEL calculatin f bth pedestrian and vehicle delay. These Figures indicate the fllwing. (i) Duble cycling prvides a net benefit under mst circumstances prvided that the Pelican is perating within its vehicle capacity. Fr the cycle times f 48 and 72 secnds a dual vehicle precedence bard wuld be required. (ii) Duble windwing is generally neutral in effect prviding neither a substantial benefit fr pedestrians nr a substantial disbenefit fr vehicle ccupants. (iii) 12 secnd windw prvides an verall disbenefit except at lw vehicle flw. The disbenefit may be attributed t allwing a pedestrian precedence perid at a nn ptimum time in the cycle. (iv) 15 secnd 'Green Man' prduces an verall disbenefit because f the increased effective red time fr vehicles. (v) (vi) Restricted VA prvides a net benefit at lw vehicle flws fr higher cycle times. ndependent peratin prvides a net benefit at high pedestrian flws cmbined with high UTC cycle times. Overall it is clear that duble cycling prvides the mst satisfactry slutin. Every effrt shuld be made t ensure that all Pelicans in UTC areas are fitted with dual VP bards t facilitate duble cycling. ndependent peratin is wrth cnsidering where the pedestrian flw is high. CONCLUSONS (i) A survey f current practice by Lcal Highway Authrities perating SCOOT indicated cncern abut increased pedestrian delay at Pelican crssings. Almst all Authrities cnfirmed that the peratin f Pelican crssings influenced their chice 17

f cycle times with 3% f Authrities prepared t perate ne r mre Pelicans ff link if the pedestrian flw was lw. The Lndn Traffic Cntrl Systems Unit are cntinuing t cnvert their Pelican crssings t perate dual vehicle precedence perids thus facilitating duble cycling at lw UTC cycle times. (ii) Site studies f 15 Pelicans perating under SCOOT cntrl in 5 areas indicated that in cmparisn with Fixed Time Pelicans perating independently - (a) (b) pedestrian mean delay is substantially higher, and there is a substantially higher prprtin f pedestrians crssing during the 'Red Man' aspect. (iii) A cmparisn f the cst f pedestrian delay at a typical Pelican crssing perating independently in an uncrdinated area, under SCOOT cntrl, and under TRANSYT cntrl shwed the fllwing. (a) (b) Assuming similar cnstraints in respect f duble cycling, SCOOT because f its ability t reduce cycle times during perids f lwer vehicle flw, has lwer pedestrian delay csts than TRANSYT at ff peak perids. Because f the relatively lw pedestrian flws at Pelicans in SCOOT areas and the mderating effect f crssing during the 'Red Man', the financial disbenefit t pedestrians f perating Pelicans under SCOOT cntrl is small in cmparisn with the verall financial benefits t vehicles and their ccupants. (iv) An evaluatin f alternative perating strategies fr Pelicans perated under SCOOT cntrl shwed the fllwing. (a) (b) Duble cycling f Pelicans prvided substantial benefits t pedestrians with at wrst nly a small disbenefit t vehicles and their ccupants. Essentially duble cycling ensures that the Pelican perates n a cycle time which is similar t that fr Pelicans perating independently in uncrdinated areas. Where pedestrian flw is high and vehicle flw lw t mderate, the benefit t pedestrians f independent peratin f the Pelican is greater than the disbenefit t vehicles and their ccupants.. RECOMMENDATONS t is clear that the cmbined cnstraints impsed by UTC systems, including SCOOT, and Fixed Time Pelican cntrllers impse substantially higher delays n pedestrians and encurage a higher prprtin f pedestrians t crss during the 'Red Man'. This situatin culd easily be reversed by fllwing the example f the Lndn Traffic Cntrl Systems Unit and using existing SCOOT prcedures t duble cycle Pelicans fitted with dual vehicle 18

precedence perid bards perating with f=1 secnds under SCOOT cntrl and f=28 secnds under lcal cntrl. The specificatin fr new SCOOT systems shuld include the fllwing.. ALL PELCAN CROSSNGS SHALL BE DOUBLE CYCLED PROVDED THAT THE DEGREE OF SATURATON DOES NOT EXCEED 9%. ALL PELCAN CROSSNGS LNKED AS PART OF A SCOOT SYSTEM SHALL BE FTTED WTH A DUAL VEHCLE PRECEDENCE BOARD SO THAT THE VEHCLE PRECEDENCE PEROD CAN BE SET AT 1 SECONDS UNDER SCOOT CONTROL AND 28 SECONDS UNDER LOCAL CONTROL All Pelicans in existing SCOOT installatins shuld be cnverted t dual VP bards as sn as pssible. The implementatin f the abve recmmendatins will ensure that SCOOT ffers significant benefits t vehicle ccupants and pedestrians. 1. ACKNOWLEDGEMENTS The help and advice f Mr. D. Brethertn, Dr. K. Wd and Mr. J. Wall f the Traffic Grup at the Transprt and Rad Research Labratry, and the c-peratin and assistance f traffic engineers frm: Cventry, Wrcester, Leicester, Suthamptn, and Wlverhamptn is gratefully acknwledged. 11. REFERENCES i. Hunt, P.B., Rbertsn, D.., Brethertn, R.D. and Wintn, R.. SCOOT - A Traffic Respnsive Methd f C-rdinating Signals. Labratry Reprt 114, Transprt and Rad Research Labratry, Crwthrne, 1981.. Vincent, R.A., Mitchell, A.. and Rbertsn, D.. User guide t TRANSYT8, Labratry Reprt 888, Transprt and Rad Research Labratry, Crwthrne, 198.. Selby, D.L. and Pwell, R.J. Urban Traffic Cntrl System ncrprating SCOOT: Design and mplementatin. Prcsedings f the nstitutin f Civil Engineers, Part, Vi.82, Octber 1987, pp93-92.. Hunt, J.G, Griffiths, J.D and Mses, T. An evaluatin f the effect f SCOOT n rad safety. Reprt t the Transprt and Rad Research Labratry, April 199.. Hunt, J.G, Griffiths, J.D and Dyke, D. The peratin f pedestrian crssings within areas where traffic signals are c- 19

rdinated. Seminar D. Prceedings f the PTRC 15th Summer Annual Meeting, PTRC, Lndn, 1987, pp79-86... Hunt, J.G. Pelican crssing peratin in areas under UTC. Prceedings f the 3rd nternatinal Cnference n Rad Traffic Cntrl, nstitutin f Electrical Engineers, Lndn, 199, pp129-133. Department f Transprt. Design cnsideratins fr Pelican and Zebra crssings. Departmental Advice Nte TA52/87, Lndn, 1987.. Department f Transprt. Pedestrian crssings : Pelican and Zebra crssings. Departmental Standard TD28/87, Lndn, 1987.. Hunt, J.G., Griffiths, J.D., Williams, J.E. and Williams, S.L. The peratin f Zebra and Pelican crssings at sites in England and Wales. Prceedings f the nternatinal Cnference n Rad Traffic Signalling. nstitutin f Electrical Engineers, Lndn, 1982, pp25-28. 1. 11. Griffiths, J.D., Hunt, J.G. and Marlw, M. Delays at pedestrian crssings - : site bservatins and the interpretatin f data. Traffic Engineering and Cntrl 25, 1984, pp365-371. Griffiths, J.D., Hunt, J.G. and Marlw, M. Delays at pedestrian crssings - 3 : the develpment and validatin f a simulatin mdel f a Pelican crssing. Traffic Engineering and Cntrl 25, 1984, pp611-616. 12. 13. Hunt, J.G. and Griffiths, J.D. Pelican crssings - the develpment f a simulatin mdel t evaluate perating strategies. Prceedings f nternatinal Cnference n Transprt Planning and Cntrl. The nstitute f Mathematics and its Applicatins, Cardiff, 1989. Department f Transprt. UTC in Wrcester - Results f the Assessment Study. Leaflet issued by Department f Transprt, Transprt Planning Assciates, and Herefrd and Wrcester Cunty Cuncil. 1986. 2

FGURES Pelican perating cycle and timings. 2. A cmparisn, based n simulatin, f pedestrian mean delay at Pelicans with different perating cnditins...... The variatin f percentage pedestrians delayed with cmbined vehicle flw at the Pelican survey sites. The variatin f percentage pedestrians crssing during the 'Red Man' with cmbined vehicle flw at the Pelican survey sites. The variatin f pedestrian mean delay with cmbined vehicle flw at the Pelican survey sites. Cmparisn between percentage pedestrians delayed at islated Pelicans (by simulatin) and Pelicans perating under SCOOT (recrded n site). Cmparisn between percentage pedestrians crssing during the'red Man' at islated Pelicans (by simulatin) and Pelicans perating under SCOOT (recrded n site).. Cmparisn between pedestrian mean delay (secs) at islated Pelicans (by simulatin) and Pelicans perating under SCOOT (recrded n site).. Cmparisn between number f Pelican cycles called in 3 minute perids at islated Pelicans (by simulatin) and Pelicans perating under SCOOT (recrded n site). 1. 11 12 13 14 15 16 17. 18. Cycle time histry at the Cventry survey sites. Cycle time histry at the Leicester survey sites. Cycle time histry at the Suthamptn survey sites. Cycle time histry at the Wlverhamptn survey sites. Cycle time histry at the Wrcester survey sites. Typical set f cyclic flw prfiles fr ne f the survey sites. Cmparisn between bservatin and simulatin fr the percentage f pedestrians delayed at Pelicans perating under SCOOT cntrl. Cmparisn between bservatin and simulatin fr the percentage f pedestrians crssing during the 'Red Man' at Pelicans perating under SCOOT cntrl. Cmparisn between bservatin and simulatin fr the pedestrian mean delay (secs) at Pelicans perating under SCOOT cntrl.

19. Cmparisn between bservatin and simulatin fr the number f Pelican cycles called at Pelican perating under SCOOT cntrl. 2. Netwrk used fr TRANSYT simulatin f Pelican peratin. 21. Percentage f pedestrians delayed at Pelicans within different perating envirnments. 22. Pedestrian mean delay at Pelicans within different perating envirnments. 23. Pedestrians crssing during the 'Red Man' at Pelicans within different perating envirnments. 24. Cmparisn f vehicle delay at a Pelican crssing as predicted by SMPEL and TRANSYT. 25. Cmparisn f netwrk and lcal vehicle delay csts as predicted by TRANSYT. 26-29.Evaluatin f SCOOT Pelican strategies.

VEHCLE ASPECT GREEN AMBER RED FLASHNG AMBER GREEN RED MAN -,-- C--- GREEN MAN ~ d---- FLASHNG GREEN MAN eq ~fq RED MAN E F A B C PEDESTRAN ASPECT a = 3 secnds c = 4-7 secnds e = r 2 secnds b = r 3 secnds d = 6-12 secnds f = 2-6 secnds fr VA peratin fmin = 6-15 secnds fmax = 1-6 secnds Nte: The Green Man may be extended by 2 secnds, r an verlap f 2 secnds may be used between the Flashing Green Man and vehicle red if required by site circumstances. FiEure - Pelican peratin E cycle and timines

25 a FT Pel,can (~'=33 A VR Pel icn FT Pel cn in UTC 2 ix (J @s! Q t5 " C O ~E p, rn Z~ (- (/ ".." 1 r"l 5 1 2 Vehicle F1v - veh/h Figure 2 -- A cmparisn,based n simulatin, f pedestrian mean delay at Pelicans with different perating cnditins

Figures 3-9.16-19 Kev Fleshill Rd., Cventry Binley Rd.~ Cventry Walsgrave Rd., Cventry St. Nichlas Circle, Leicester Lndn Rd., Leicester Vaughan Way, Leicester + Shirley High St., Suthamptn Bitterne, Suthamptn Prtswd Rd., Suthamptn Blue Lane, Wlverhamptn Chapel Ash, Wlverhamptn Penn Rd., Wlverhamptn Fregate St., Wrcester # Barburne, Wrcester Deansway, Wrcester "j

;> x x x 4 4 X X 4,~ WK X X 4 ~ X 4 ~e ~:.4<~ ~z * ~ ~., *~* ~w~ ":~' ~*~t " + ~., ~ *,O + O O "F" 4.'1",, ~, -,~ ++,- F b.,,.. O.,.=~.. ~,, d,) s,,. i O C3 CO LO '~- C'~,=,,,,=, X - sul-;l~eped pe&tap.4. e6luej d

.,; s x. Xx X X 4 X X X 4 <].= X q.~. O +13 ~ ',,,, + ~ + ~ g m QJ.O. ) -pill O' 7. -,udw p~, 8u;~np 8u~sJ ~ut.~soped.~ eslueje d [.-, L.. r,:.,

> X <J X <~ X X Xx X x X X X X ("J.-4 ~z 4# x ~# 4z. + x vi. = <3 f :3 ~ ;> & ~ ~ O 4= Q W-~ ~, -~,+...,,.::. CD - = /D j:,~ > c_ ~.pq,d~ L ~ f t~ i c. ~ LO se~; - ~lep ueuj ut~l, sepe d

e- ]. 9@ x ; 8 -- 7 ~ ~z~" ~ ~...a" T.i ~- 5 b- 4 1.1. = 3 2 2O - 1 to 2 3 4 5 6 "7 8 9 to SCOOT survey dafa Figure 6 - Cmparisn between percentage pedestrians delayed at islated Pelicans (by simulatin) and Pelicans perating under SCOOT (recrded n site) e- 9O 8 ~ 7. GO : 5O "O 4 J. ~, 3 z 2O - lo " ~ ~,,.,,.,, n~nl to 2 3 4 5 6? 8 9 SCOOT ~urvey data Figure 7 - Cmparisn between percentage pedestrians crssing during the 'Red Man' at islated Pelicans (by simulatin) and Pelicans perating under SCOOT (recrded n site)

6 e- ~- 5 E "~ 4 e,l ~- 3 p- x,, 2 e~ to X & i i i i 1 2 3 4 5 6 SCOOT ~urve~/ data Figure 8 -- Cmparisn between pedestrian mean delay (secs) at islated Pelicans (by simulatin) and Pelicans perating under SCOOT (recrded n site),-j 5 4 :3 +.~.%'+ ~ + + / 3 -- " 2 L~. 6,) 1!! to 2 3 4 5 SCOOT 3urvey data Figure 9 - Cmparisn between number f Pelican cycles called in 3 minute perids at islated Pelicans (by simulatin) and Pelicans perating under SCOOT (recrded n site)

Figures - i~ Nte Bth regin and Pelican cycle time histries are shwn in these Figures. The Pelican is represented by the brken line which appears as a higher density- trace where verprinting ccurs. m~

12 -~ lo - to - 9- i 8-7- 5 6 1 4O 3O BOO 9 1 11 12' 13 14 15 Time- hrs (a) Fleshlll Rd. 16 17 tso 12-11 - U 1-9- i 8-7- "6 6-5- 4-3 8O 11!11 lllul 11 ~ ill lilllll 'L~j~ i 9 i 11' 12' 13' 14 15 16 17 18 Time - - hrs (b) elnley Rd. 12-11 - ~, - ~' 9- J 8- - 7- -~ 6- m,, c., 5-4-,L~.,J~LR!_ lli i eli i lii i nl 3 8 9' ' i ' 12' 13' 14 15 16' Tlme- hr= (c) Walsgrave Rd. 17 18 Figure to - Cycle time histry at the Cventry survey sites

O 12 11 1 9 8 m 7O 6 5 4 3 -- 8 U ~~ ~ l m J llil lllr lllll 7 'i 9' 1 11 12 13' 14: 15' 16 17 18 Time- hr+ (a) St. Nichlas Circle 1 3 @ v~ @ E "6 Q~ 12 11 1 9O 8O 7 6 5O J 4O 3 8 9' 1' 1i' 12 13 14 15 16 17' ta Time- hrs (b) Lndn Rd. 12 11 ~, 1 @ 9 8 _E 7 6 5 4 3 8 fll lllmlmll U. ".. 9' 1' 11 12' 13 14' 15 18' 17 18 (c) Time- hr= Vaugl",an Way Figure t - Cycle time histry at the Leicester survey sites

J. 2 - i 11 -~ 1 - i 8-7- U 6! 5 4 3. 8OO 9 1' 11 12' 13 14 15' 16 17 18 Time -- hrs (a) Shirley High St. 12-11 -. ], q ~. - ~' 9-8- E 7-6- )k. ~-' 5-4- 3O 8 9 1 11 12' 13' 14 15 16 17 [8 Time -- hrs (b) altterne 12-11 - - 8 - i 8-7- a 6- ~' 5-4- 1 3 8 9 1' 11 12' 13' 14' 15' 16 17' 18 Time- hrs (c) Prtswd Rd. Figure 12 - Cycle time histry at the Suthamptn survey sites

12 i~,i 9 ' 8 E 7O _J t 6 5 4 3O 8O lll i f. '""" '""'" "' V~ ~'""'"" ~' """ ""' ~ 9 1' 11 12 13 14 15 16 17 tso Time- hrs (a) 81ue Lane West 12 ii ~1 9O 8 -~ 7O -~ 6 5 4 3 "1 8, i,.,,,, i ~'~"~.! ~, ~~ _,,,,, 9 ' 11 12 13 14' 15 16 i7 18 Time- (b) Chpe hrs Ash 12-9 ' 8 _E 7 6 5 4 3 m a m 8 r ~i~ i i_i_ ill lllj il fj llill '~11 lull U lllmiuilll ll ilium 9' 1' i ' 12' 13 14' 15' 16 17 18 Time- hrs (=) Penn Rd. Figure J.3 - Cycle time histry at the Wlverhamptn survey sites

12 -- -,,, 1-9- ' 8- _E 7- - i "-n n 6-5- 4-3O 8OO 9 [ L --tlr E _LLj; --L LF -- -f[ --Ll-LLr -- re -- _rt _JL... ' 11 12 13 14 15 16 t7 ~ Time- hrs () Fregte St. 12 -- 11 ~ 1 t 9O i 8 - E 7-6- U 5-4- 3O 8OO 9' ' ' 12' 13 14' 15 16 17 18 Time -- hrs (b) Barburne 12 11 1 9O i 8- _E 7- a 6-5- 4-3 8 9 F F ft _ [ -- --t _rt 1' 11' 12' 13' 14' 15' 18 17 18 Time- hrs (c) Oenswy Figure J.d~ - Cycle time histry at the Wrcester survey sites

Flw S~de A T ime - 4s in ter'va s Side 8 Flw Time - 4s r'l N i i i er'v a s Figure 15 - A typical set f cyclic flw prfiles frm ne f the survey sites

to 9 8.J b,j (i,. t/3 "7 GO 5 4 3 2 to [ i i i l O 2 3 4 5 6 7 8 9 to Obeervatln - SCOOT ~urvey data Figure 16 - Cmparisn between bservatin and simulatin fr the percentage f pedestrians delayed at Pelicans perating under SCOOT cntrl t 9 8..J ~J Q. p--1 L~ cn! C E? BO 5 4 3 2 to Q ~' / -! [ i i to 2 3 4 5 6 7 8 9 to Ob=ervatin - SCOOT ~urvev data Figure 17 - Cmparisn between bservatin and simulatin fr the percentage f pedestrians crssing during the 'Red Man' at Pelicans perating under SCOOT cntrl

GO 5 J UJ 4! = 3 2!! to 2 3 4 5 GO Observat ~n - SCOOT survey data Figure 18 - Cmparisn between bservatin and simulatin fr the pedestrian mean delay (secs) at Pelicans perating under SCOOT cntrl 5 4 _1 b. (1. s-! C 3 + el E 2 to O to 2 3 4 5 Observatin - SCOOT survey data Figure 19 - Cmparisn between bservatin and the number f Pelican cycles called at Pelicans SCOOT cntrl simulatin perating fr under

"',!/ tl m a. c,i i,~.1 f ',1 >- Z ee. O'J L_ O ) z u_ "/T'"

Figures 21 -- 23 Cmparisn f Pelican peratin within different envirnments Key 12 secnd cycle time 8 secnd cycle time 96 secnd cycle time 84 secnd cycle time 72 secnd cycle time 6 secnd cycle time 48 secnd cycle time ndependent uncrdinated in a VA signal area

q.) r- m e- (E +lql Q O m e- q; > t- "C3 n c~ t! (_3 "C.iml.Jl (_) t! e" q~ el (~ q,q+-., r- m 1 Q G c,4 f 'li...ij c- O m (E O 1. t 1! O O O O O O O O O O..1 % - pet~ol ep +u+ u+saped.~ eo+uauecl