Group Title: Operating strategies and related performance evaluations for a moving merge control system by Christian S. Bauer
Title: Operating strategies and related performance evaluations for a moving merge control system
Full Citation
Permanent Link:
 Material Information
Title: Operating strategies and related performance evaluations for a moving merge control system
Physical Description: xviii, 474 leaves. : ill. ; 28 cm.
Language: English
Creator: Bauer, Christian Schmid, 1944-
Publication Date: 1975
Copyright Date: 1975
Subject: Express highways -- Florida -- Tampa   ( lcsh )
Traffic flow -- Florida -- Tampa   ( lcsh )
Traffic flow -- Data processing   ( lcsh )
Civil Engineering thesis Ph. D
Dissertations, Academic -- Civil Engineering -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis -- University of Florida.
Bibliography: Bibliography: leaves 462-472.
Additional Physical Form: Also available on World Wide Web
General Note: Typescript.
General Note: Vita.
 Record Information
Bibliographic ID: UF00097518
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000580873
oclc - 14092618
notis - ADA8978


This item has the following downloads:

PDF ( 15 MBs ) ( PDF )

Full Text


P Zight,
to the real,
.mmo Iality.

1 00

lace was respon-

ering work for the

oth people were a

system improvements

Traffic Engineer.

of his technical staff,

d David Osborne, have

bases of project work at

rotation was par-

ort provided by the

he many contri-

ett Owens,

ce, State Proj-

re her*






.......... 48

isplay..... 56

............ 57


.......... 58

............ 60

L ts....... 65

........ 67




4 1e




n the
















L Cas

.................... 436

S.................. 462

S.................. 472

.. ............... 473


Figure: Face

1 Hiierar-cal of Frec.'a, Pamtp 'C ntrcl-.
System Strategies.............................

2 Illustriaticn of sequence e of Ope-ra-
ti.crn i r a Gap-A.cceptance [lercjec
Control S. ;tei . ............................... 5

3 5chnmatic Diagram of the Ra'.'ti 'on
Corporation Pacer !lrge Co'ntrol
S'.s t m ..................... ......................

4 Schematic La of the. Ria. t, cn
Corporati.n GrCu c arcr d LnJ o'.'inl Mo.tr3-ge
Conti-rcol S tem. . .................. ............ 30j

5 green Barnd Systcem Cc'ntroil iSro ft:.are
Fl,.o. Chart .................. .................. 34

6 Green Band S stmiT- Control oft..ware Fi.!rc-
tioni Driagrlam. ................... ............... 35

7 Illustratiorn of Sch-ari U Lis. by Raytheo.n
for the Identificacion iof Double-Lop
Sensor Event Times............................ 38

8 Data iLoggJinir R.outine 7ap.e- Output Format
for lobturr Green i 'and Sc tem .................. 44

9 Grucn Band Display Element Parameter
D efinitions. .................................. 49

10 Possible Mlode Trarsitionr in t.le liuobiuriI
Green BanJ Sys' teii. .................... .... .. . 6

11 Woburn Green Banr Sy.stcm floJe
Decision Epacc: ................................ 6.

12 Five-Minute Volumes at Sensor o10
orn January 24, 11974 ........................... 74

Figure: Page

13 Fi'.'c-Mirnute VoluTimes Across sensors
Fl,., FPC, and [Fr on Januar' 24, 1974 ........ 75

14 TIampa .lMov'.in u.rye -, Instru-
m nt lrtatlion ubs '.' em. ........................... '7

15 Tamp.a Mi. M -.'inrg M.,-rL S; tm:
Di-play ibb s tem . . ................... ......... 22

16 The Elements of the Driver Display nr
the lamTi. 3 Green, Land S':tem ........................ 33

17 Conrtrol Eq.ipment Hard-,.arf for thli
Tamps Green Bardr S?/stemr ....................... .7

18 IldMce Charne Currcntl.' Impleme-nted for the
Taimpa Grcen and Sy'stem.r ....................... 97

19 IB tM System'360 CS'IP Proqram for the
Ainalysi s of Green Bard Trai-ct,;ries Undcr
VaLiaLie .c-ce'leration Frofiles in the
Stopp'e]-Gap C' p-eratirn? Mode.'. ................... 101

20 Computer Output from Storpped-Gap Earnd Tra-
jectory Anal:sis Prog ram ...................... 102

21 Linear Tampa Ramrip CGemctL .. ............. .. 106

22 Placement of Trafrfi. Co.,nters for Ilerge-
Point Location tui'y. .................. ....... 112

23 Distribution of Ramp vehiclese s Entering
Fre-. a; by Count Station, 7 A.M1. 11 A.M.,
Decemb.-r 13, 1'72 ............................... 114

24 Per Cent -of Ramp Vehicles Passing Count
Stations, 3 P.M. 7 P.M., Deccemb.f; 13,
1 72................................... ......... 115

25 Location of tlergre Point for the Tampa
S 'stem ......................................... 118

26 Sur.'eillance Area for Tra.'el Time
Stuj.'. ........................................ 120

27 Spacc-Time Diagram for ax:imum E::pected
Speeds........................................ 124

xi i

Figure: Pa 7e

28 Spa,-,---Tlime Diagram for tliri.L. i
E:xpc tei Spee s- ... .............................. ;

29 Space-Time Diaqramrr for t.ld-Pange speeds ....... 128

30 Space-Tim.c Diagrai,, for the Locaticrn
or Taipa P iFp EDet-ector F3 ..................... 120

31 Ev.'aluation Plan Flow Diagram.................. 141

32 Putlic- Opinion Ouestionnaire for 'lamp-a
Green Band S.stem E'.'aluation .................. 145

33 ScheriM UseJ for ssigqnrienrt .f Field S.-n-
aor Chainnels to Input Hard,.are Buffer
oar.-I in the i armpa Cr.-'en Band SE'stein......... 147

34 Data Lo.ggirig Routine Tape B.loc -i: FPcrriiat
for the Tampa Green Barid .:3t ..i. . ............... .. .150

35 Subroutine Structure for TOCAP Data
Reduct ion Soft ............................ 156

36 Output from Subroutine EVE!JT Sihewinq
Dec-oded 1- letMpe Message, Field Sens-or
Activity:, ard Driv'er Di-pl_-r Data ............. 160

37 Free'.:-. Vciicle i' Oujtput
From r TC'GC.'.P.. ................. ................. 168

38 Freeway Vehicle Tra-ie.tories by '.'-
hicle n number ................... ............... 170

39 Sensor Acti'it. Surrunar'.' frora TOGAPP............ 174

40 Freewa'' Vehicle A'ccel-era.tion 1OISE
Ciutput frori TCGAP. ............................. 177

41 Distribution of Freew.'a vehicle e
Vclocities at F1 .............................. 180

42 TOGAP Output for the Anal' sis of
Freeway Vehicle Lane Ch.anging Upstream
of the Merge Area............................. 181

Fi gure : Page

-43 TOG-AP Output Sho'.;ing Dlitribution
_,f '.'ehi- i T'ra''.'cl Time-s on the
A. hley S tr .eet RMi.p ..... ............ ......... 184

44 Suwimarl Output from TOG-AP Green
Ban.d Tamp .hicle Tracking .................... 136

4= Pictorial DiEpla. of TOGAF Green
Rand Ramp Vehicle Trackling .................... 190

46 Data for the Calibration of the Soni.-
Dl-ctector-. Uized in the Tampa Gre-en
Bard sterm........................... ........... 19

47 TOGAP Cutput from the Sonic Detector
A.ri iys of Lateral vehicle e Displac-ments
in the erilrgilug ........................ 195

48 Flo; Chart of F.outine MODEX:: .................. 198

49 TOGAP Output-. for Green Bard Control
S tsrtem iln.e and r lasting Actl'.'it ............. 202

50 Ccopute-r-Generated Plct of Ramp and
Frecwa, vehiclel e Tra iectori--s Produced
b:' TOGAP Software ............................. 205

51 TOGAP O:utputs for the Distribution of
Free:'..y Caps hIatched by F:Ranp .'ehicle.I
e-tected at Sensor R1 ......................... 208

52 Subroutine Calling Fates in the On-
Liine Control Program .......................... 230

53 Operating Erivironmernt for the Gre-n Eand
Control Pr.ogrami Simulator Illustrating
Inputs, Logic Eleimerts, and Outputs ........... 234

54 Subroutine Structure for Green Band
Control Program Simulator. ...................... 235

55 Scheme Used for TOGAP Data Storage
in Green PBanl] Simulation Routine .............. 237

56 Flow Chart of H'.'P Logic Used in
Control Program Simulator..................... 240

C igurL F .ag

57 FloV.' Chart of Suibroutine i I EP T ............... 24 2

58 Flo. Chart of Green Band Stntus
Rou tine ..................... .................. 45

59 Schemet for Filming 3imulIted
Gr-en Band OCutput- ................... ....... . 254

60 Initialitation Output from Cree:r
Band Simulator ................................ 257

61 Band Output from Control Program
S ii tor . . ..................................... 20

62 CI s t I Data . . . . . . . . . . . . . . . . . 76

63 Case IT Data .................. .. .............. 279

64 Case III D ta. ......................... ........ 281

65 Case 1'S. Data ................................... 82

66 Case V' Data ................... ................ -'84

67 Case iI Data.................................. 235

68 Case VII Data....... .......................... 2 7

69 Cas, 'VIII Data................................ 28

70 Case I: Data .................................. 2'0

71 Band Output '-ho',.in g GBS operation
Inhibited Ex::cept at Sensor Event Times ........ 291

72 Schematic Diagram of IMo'.ing Merge Con-
troller Using flicro-Comiput'L r echnoliog'...... 3.10

73 Projected Configuration of Packagcd Mlo'.'ing
'lerge Controller ................................ 3 9

74 Photograph Taken from the Screen of a
Surveillance Radar at Tamhpa, Intirnatilri on
Airport Showing Frve.,a" .'hicle Traffic on
1-75 Across Tampa Ba'y. ........................ 334

c.b tract of Dissertatic.rn F resentr d to the
Gra ilu.ja t- Co.unci of t e Univ .rsi t. of Flor-ida
in Partial Fulfl i m.rnt orf the P.eqiu rements
f.:'r the Degrue of Doctor of PhI-rlosophr

FOP F' l.:,...'I1G .lif.GE C:;TPOL H.S i' rEl


Cnristian S. Eauer, Jr.

March, 1975

Chairman: Fennetr, G. Co.uraqga
[lajor Department: Engiiieerinq

In 1970, the Pat;theon Corporation, under a contract

.'ithU thi Federal liigqrla~' Admirlistration, conducted prelimin-

art field tests n t:tc t'yp.s of moving merge control stemses

at a free.'a' ,ntrarnce ramp on Route 1i2 in 1-Joburn, flassachu-

setts. ile tests jdemonstrated the operational feasibilitl' cf

both s stems, and established a .jri'.er preference for the

green band system concept, which dispiaY.ed bands of light

moving along the ramp corresponding to acceptable qaps a.vail-

able in the freeway traffic stream for merging vehicles.

In 1972, the JUniversity- of Florida began a joint

research project with the Florida Department of Transporta-

tion in'.'.lvinq the installation and evaluation of a green

V11' 1 1

band =.Y3teirI on rthe Ashile. Street e-nt r rancc- r mp to 1-75 in

Tampa, Florid-. This ramni does nct meet current Jgeocmetric

design standards, anrI a m.3ajr objective of this pr oje-ctr 'as

to determine the feasibility l of improu.'in tiie- .op.rationr of

such sub-stan-lir ramps initt, : mio.-ing merge coritrol :.3tiem

as contrasted to tic traditional approach of recrnstrUicrti.irn.

modificationss and Imnpro',ement- to toe Wobuln ite

harrdare and soft.qare used to install the Tji3mpa .zt,'teI are

dc scribed, 5. "tei imr roL' ..:r emr-cnt' suggested by operational ex-

perience in the public opCeration of the Tampa ?.stem are pre-

sented, including the ,dev'eloprient of nc.j algorithms for on-

line sensor calibrationa, greeni barnd pr-opjaqtion, arnd system

mode control.

Techniques developed for the autorri.ate,- perf..rfirmnc.

evaluation of ramp operations under the direction of a merg-

inq control s'stc-m are described, arnd a trade-off any1:.'sl

of existing system functions 3nd requi rements leading to

recomrriended hardw:are.'softw3are configuration for future niergi.:

control systems is presented.

yr .:i,.4 E. '. IF1*1

cD U 2 t r c i n

lodj'ri c r r ffic Cnr gl rei ernl,1 prja:tic is concerned

t.: la re e..Ctel t with th,: saolttiono. to prolilems Jrising

fro..m free'.;a cnesti.on. ;-jin:. these problems are questions

rela.ting to the di'.ersiorn of .'hicles in a traffic s'y'stem

to .achie'.'2 bal sliced operat ion .if e.:istl ing facilities ran]

rc.aJways \within the system, ioln of traffic filoc

chLracteristics cn a gia ern r:a.d',.,a ta o achi'.'e specified

operating characteristics, and methods for pro'..'idir, ajFsist-

ance to drivers usin such facilities tc. reduce de-las and

provide all i cional 5afe-y .1.airi s.

The obiectl'e c f pro'.:d"i. inf dri'.er assistance has

been approached in fre.ewa.v systems primarily through the use

of ramp control systems. Figur 1 prerents a chart of the

tiicrarchy of frecuay ramp control system strategies. cl.issi-

fied with respect to the sophi;-tic3tion of the control

algorithm invol.e., and the degree of drir'er assistance pro-

vidc.J to the ramp driv'i in the merging manteu'er vb each

sy',tem. The earliest r mp control technique simply involved

the closina of certain entrance ramps during periods of peak

) H'''--'

4 '1 .
LW ;iU

LJ ^:

i- t- "
-: H ^ C






1~ .3

V t-




'" M '--


LJ I'2 "

l_l '





tr. ffic fl .:.. The primary c..-c'rr -ern '.'i.r th:r re-gulation of

total free.,I- t-raff i.c fl'..' c hi-v- Jerairel c'perati rn:

coriditi..ot-, a.j the .'Gree of ass- i t :anrce to ramp ri i.ers in

such sy--st ms i.' rj il inii al. .It coul '.l be a Urq-u J ho6.e-.' rhat rriSmp ir-.'-rs i:.'-rr- from'c, Clos.e raml'pi .'*le a tL, to

'airn acc..e s t, >S.t ti f re .. a by an t her C, r ut-, the.' .er.e t'hemIr, -

-i.eC .1di ty :he overall sy cem of ramp closur--, as the-

rree;..5a traffic flo ic ..a3 mir ntaineid at le'.'e-ls consistent '.itlh

t ficie nt op-r i c. r .

Tnlrer n,-::t cate-gor.:.: *f ramp control systems is one

inr.i .'in mi teringq ,:f .-chlicles on'Dt thl- freei.a'sy from an en-

trance camp. ECrly s ti-mS u -eJ p:. lic- officers stationed

.t the entrance to the i free-..'ay c- mps ..ho '.'er r- rte-ponrsible

for allo%.ina i cc-r'tain numbeLr *f -.'hicls per minute tro

enter the freew :.. The a.-_91 s of such s'sto.nia .-*ere to rc-duce

quueuein problems in the rmeriq- afr-tas associatel .:ith su-ch

ranmips i9 n to pce'.'enrt cOnetiiort an t thi.- f reway entrance

point as.sociateJ '.ith thie cairp,.. '.'.anced sy2 stcims- of thi

type t.'pically use con'enrional 1fiy-ed-1ti~ce traffic control

-quii i pent to establish ramp -meteriing rates in cornjunction

..ith conr.'e~ :ioni traffic siqnals to alia-'iate the need for

Jhuiman personnel 3 tatrioneLl on the ramp, and can include fea-

rur,-- 'ihch imanilk thuse ;ystemrs traffic ire p.unsi..', that is

having the capaL lit:' to adjust metering rates t.o account

Ecr. different obser.'e1 flow. pattern- on either- the- ramp or

frecw :e,. or both. iJiot that the .sphi st._i3ca c. or cc, ,ontrol

a.lgorith iii in ramp ieiterfin s.;stem.I is of nece~-; it ;i ie.hat

hiher tihai that inv'.-ol'ved ''i thi the r ,amp clo' re ;systems,

*..hich simply inv.oleJ, in most cases, imeci-h ical barriers

t, the raminrs ocr ch n.:leiabie ni ~ar ;1.-ins ir dicati'rl' that

the ramps '.;erc closed.

Highe-r on th-e Sophistication s-ale are tti; s:-

called stopped ymp-.iaccptalnce v--.tems wih.i e .rp. ratio.. is

depicted in Ft,:ure 2. Ini sich systems, sen s .r-S inE-StallJ onr:

the fre.'.way upFtrecmi o f thl.- entrarn-c ramp me.isur.t_ japs iF t.-Ie

freewa-' traffic, and acti .-iate a con'.rti.or, l tra ffii c i. r, al

urn the ramp at a time projr cted to. all.:'.i the ra.p '. tiicl.- to

enter tihe freeway qap ..,hen it r-ac.l'ic the merge area. Suchj..I

systems represent an att.lempt tc. fit the Jt-'ir r iriI' avail-

able frees;i-y gaps dir;ctl,', and thuj rtpre-cnt -A cc-,niderlabl~ -

improvemeint in tlhie potcnti i for pr.. '.idig a Csaistance tfhL'

offer to ramp drivers. '. Fhcrtc:,mimgi of ep-ac-ceptanc._e s.I-

terms, however is that the. ex::rt no control .i *.r the driver '

actions once he- lhas left the control signal area. Thus,

assumptions are ne.ceasear' i.,n a'.'..r.cae ri'. er performalnIe in

the acceleration maneuver bet':cen the time hi' leaves a Si= -

nal and the time hc is e:-.pectedo to rah te fr-ee;ay gap and

problems can arise if '.-hicle traiectorie- Jo not meet the

expected characteristics.


...U I -ILE CA

Fr ,l..t O F T .ET L.E E t..-

..- . ---

l' LE i C

Figure 2. I1lustratlnri of Sequence- of Cperations in a Gap-
.cceptance :1er,]c Control System

io achieve .addi tic ar.a i r,~a c c tt-. c.rtro f rCampr. v.'e-

hicles in the mr ng m-rarleu'.'r er aii-thh r class of system

known as mv.'ing merge -orntro.Il ysctum has been develop ed,

representing g additironal s1c.phisticati on in the cO7ntl: ri a- qo-

rlthm sa well as an addliti crina p:oteltA i a 1 f or p r'.'idir.g

drri'ver asaistarnce in t hj merqginq nmneu .vr. In cthI. sa

teFmLs, gaps in tlhi free.;. ,- trAffr] a :re detc.. .-r j t he are

in gap-ac.1cept. nc-ir s'stemn, but arc then u-sed to ;per.5te

displays al ong side the entran;r, ramp to guiJe t he rjiamip V -

hicles into t(h: rfrc cay' gaps. Thnee S st.erns [ r'.'i. j adjdi-

tiona1 information to tIhe ranp over arier Ir sterns,

as thicy are typic 11 designe.-1 t, isplay a pictorlai repre-

zsntatior, of hot, well the rasp .'vhci le d*ri.ver is su :-:eed in

in his attempted merging rraneuvi er. 7.c CatL-qories .: f [i,;

merqe control sy;ste i tes-ted tr.-. dre are depicted in FiiguLq

1, respec.ti'.'ely the Psacr-t'-pe system, rwh!ich used a ciosed-

loop a lgori thmTi, 5nd the green hand type, whichh -a.-s an ad-

''isory system using an open-loop c: tri'I strategy. Eazh

s'.stem concept ill be re Lie'ed in Chapter 2, but the

tions of tl'hc systems in Figure I g9'.es an idea 'f the re-

sults of research findings relating to their use. The green

band system and Pacer system both represented aditicrlal as-

sistance to. ramp drivers ov.her that associated .ith earlier

gap-acceptance systems: the rclati'.'cely higher degree of so-

phistication displayed in the Pacer system control alqorichlt

,.'.'.:r- that used 1.n t1- grh -. i b and sy' t.-lm dlid not miat,-ria 1ly

increase it- utility to ramp drv.r:rs c'.rir that provided by

tI- green band system.

7. possible future diire-.tin iof ra-mp-merging control

3'y--;m rse r-rch is that pictureJ at the top right of Figure

i, ...hichi represents autromat-d '.- hicl e ma rg.: control systems

juingq on-board com:iutrs in t .et icle. With this approach,

informiatco n on free,.. ay gap.: s bryoc.-ced by a central computer

and trans.mittlJ rt: ai controller in each .ehilrle Then, as

a .'rhicle proceeds J.'.-.n the ramp., t is under the sunp vision

c.f trhii e. uipm: int and th,- rarip dr.'er can eithe- f, llo.- con-

trcl in. tri.: tions display.-e in the car or, in ad'.-anced. ss-

.e s2, let th- .system :cntrc.l the' po-sition of the vehiclel e

throughout ther entire mer,]irig u arineu.-e I'hile Zuch 'systems

ha.c- nort ct been demo3ntrated, the continuing rapid de'.-elop-

ment of computer te.:hn,:log, arnd -olid-state electroniC, s's-

tenm makes their operation technically feasible in the near

f utur_.

Prc *,:jlIs Pesearch

-onsiderable bod-., of litEritur, e:-:ists on the sub-

ject of free.*:a; anJ r amp control systems and related areas

:f .,-oncern. Wttle.:c.rth [1) discussed some of the basic

theoretical considerations in the operation of frec-:ay, systems

and discussed several control schemes for the alleviation of

tree''a. congqesti:n, include irg a Iret3cd of u:ing linear pro-

graiming to determine the optimum miete irinr rate of '.eh-ihle

entering a conr gei ted freei.ay. Dre'. [21 studied fre.-.,.

traffic c.:ngestion usin a .' of mathIema tical model- ',f

freec.jay traff17 f low and applied the result-s :c thi a worf. to

data obtained from the Gulf Free,:' in H.ouston, 'Te:xas. Earl,'

field i.'ork in free'.'ay ramp control -.. a di scu,,s -ed by flay [i3

.'hn treated work done on the Cocnrres: E:x:pre-sst:ay irn Chica. :o

ii-.'ol'irig the implementation and ev-a luac ton of ramp closure

and Lamp metering based ton occupancy iieas-uremenrt taken up-

stream ..of the entrance ramp, and by G'r.jai [4] who di.,cuj.sed

\..ork done on the John C. Lodg~e Fre::ay; in Detroit invol .'inq

ramnp closure e::perimnents dating oaci: to i960. 0 oti of thee

approaches iri.'nl''ed the use of humans in the control ic.-p -.o

mace the appropriate pmetering decisions and did rnot attempt

to automate the control procs.

In 1963 a research]ect begun b the Te:,:as-

Transportation IritL.tLute sponsored by tn Tc::as High,..a' Do--

partment and the II. S. Fureau of Public Foad- Inow.' the rcd-

eral High.a .' Administraticori i ith the objective of "de.'elop-

inq a criteria for the design and operation crf autcmati.c Eur-

veillances and control systems which unu.ld permit the at-

tainment of acceptable levels f serce or hea-.-ily traveled

urban freeways during peak period of demand" [5, p. 2. Th-i wa S

the beginning of a period of intensive research ucti'.'lit in

rimp control thec'r an.- application by the Te:-.:a Transporta-

tion Institute throughout -Lhc decide of the 196I' '3- and re-

sulted in many researcrn studies which ha'.'e contributed mat.e-

riiiy to thc q;gnrral understanding ofr t e behavior ':fr free-

way tr-ffic s stems [6-13 Famp miteriny studies cn the

Gulf Frcway' were conducted in 1964 using police officers

stat icned or thec ramip as thc- control clement. System develop-

me-rt corntinrued. with the installation, of traffic siq nals and

.associated advisory sign on eiqht ramps over a 6.5-mile

actionn' of the G.ulf Freeewa".

In early 1i66, a prototype ;ner.~in control system

t'a installed on the Telephonr Foad Interchanic entrance ramp

cn the Gulf Freeway using a hardware controller which was

traffic responsive and corriplietely automatic. This special-

purpose analog computer ;was one of the first examples of

hardware developed specifically to perf-orm the control func-

tions of a merge control system. The equipment had the capa-

bllit' for using either of cwo .basic ramp control strategies

to a. ist in the mc-rging process. The first of these the

gap acceptance mode of control which searched for freeway

gaps upstream of the merge area and activated a traffic

light on the ramp at a time calculated to allow the ramp

vehiclel e to arri'.' at the mere arca at a time corresponding

to the expected time of arrival of the freeway vehicle at the

same location. The second control strategy implemented was

a denand-c:pa cit'_' mode of ntion .hi.:h ad justel the

me' tering rate c f a r ,-rp si3nal in ac ,or iance 7r, the meo.-

sured freewa-' flo'.' rate upst r.ami of the entrance r.hi p ic .a-

icrn tc .eep t-ie total fr *a" 'cluire do.u'- re m ..ft tI. c.riJi

at or below free.'' capacity at that point, thus reduciila

the probability o.f f.Lee' 1rn dE-e lo[ in? at that

locate ion.

Hardw:.are ideclo[.nment co:nt inued. in the -Gulf Frr.;a,

Project wiith instal lation of firt -t- .3neration ro du.-rt:nri

controllers based on thie earlier Fr.prot.otype hard.;are i C.:-

tobcr, 19667, and irrnpro'.e-l scrond-geericrati% n C,.rtrc.l erI s in

lc.r.,ember, 19. 3. The later e g2rneation projectred ops in the

free-:.ay traffic based on the mea-L red sp.:.e; rof the ieaj "e-

hicl- of the freeway gaps rather than ctn tih speed of

the.? a'.eraie fr.:.-\ ay 'p hii cl e speed. as .:a-_ tie case :itl' their

first-acneration counterpart.-. Th=;- units autoi-aticj 1ly

selected one of four sets of detectorts. c in thle riee-ay to us.:

for gap projection bas.-d on tice prevailing free..a/ speed at

the time. .*apabilities were also pro..'ided for multi-"eliiclt

mrerrges and for monitoring the tra.rl time of a *.'ehicle re-

lea-ed from the ramp signal to acI-ountr for trucks an-d other

-loaw mo'.'in~ vehiclele : which did not clear the mere 3area in

the e.:pected time. Given that such a condition .4as detected,

the system could ther hcld the ramp signal in red Lintil the

pr-c iouus '.ehici hadl cleared the merge area, thus pro'.'idli n

an 3addi- iional safety factor in the me-rq in opera tiC-n s.

In July of i96?, an IE.;1 13jlj pr..,cess control co.m-

puter installed in the Gulf Free..way '.O*ntrol Ce.rnter to

proi.'iJe a capatbilict for research iito methilds- of "sy stem"

control of the freeway .as opl.Led to t he ramp

corLtrol that had beer, possible with the special purpose

sing l--ramp crnctrol lers. E:-:perilTientatior was begun 'with

c-omputear croncrol of trh ramps in March of 1969. Tw.o factors

associated with the use of the digital computer as a ramp

control system led the researchers t to he hypothesis that

"e'.'en isolated entrance ramps cari be better controlled with

digital equipment" [17, p. 3] These factors were 1'i the

ability to change the control philosophy of a digital process

controller by simply' changlirng the stored program, and ij,

che fact that the process iniputs and outputs associated with

a ramp control system are themselves digital signals, e.g.,

the system Inputs are in the form of relay contact closures

from the detectors in the field aind the contr ol signals to

be seit to the field are in the form of on-off comrimands or

discrete state change com an is to the '.'arous ramp signals

and displays.

Research work on other freo?-'ays proceeding in

parallel with the initial development work on the Gulf Free--

wa.' in Houston. The Michigan Department of State Highways

and the U, S. Bureau of Public ,Ea.IJs had [jc.rncred earlier

research on the JLhn C. Lodge Freet'a ri etroit. ihi pro-j-

ect evol.'ed i.ntc a multi-sEate coioperati.e effort called rrie

national G-roCund fo.r Fr.- -'S~ i Sur ei ll arce Conrtr -:.l

aiid Electronic :.ids [ll]. For-al re.e rch wor : .f the ia-

tional Prov'nlg r round ended in 1' .6 =in the t.merLicn .-..jci-

ation of Stat,- Highi'-a.' fficia 1s assumied trher spor' ori s ri p :'f

the proie.-t .rn the LW.1o e Free,'a" placing it irn the Iiati.:r,al

Cooperati-.'e Highwa Research Pr:Fgrai INCHFPI he program

was designated 1CHPP Project '10-3 and began on December 1i,

1966, *,ith th, Txas rrarnpcori tat i.r Instritute ser.iri asc trhe

research agency. Fre-,ious wi;r-l: done at the sit? had included

the instsallaticlno of tcllc'.'i onr camera: inj ar, =ttr.tmataic de-

tectiorn system~ for imonir t.ring cf the free..~iy. Cr.ntr.-. Data

hlojel 8090 computer \as installed with ;uffi.iiert icpacity

for control in eight ramps on the d riLnt;i..n tietr'oit area uzsi iL

both the ]Fap acceptance and demand cap.ait: control mi:des.

The control .=tratecg' for the s'ys~em -an baseJd O:n thWi linear

programming technique de'.-eloped b/y Wattlew.arh [ll aind modi-

fleJ for use inr an online control strat-, gy. The linear pro-

grarinrtngr model dtlrmined for each fvv.e-minute ccr.trcl p.-riod.J

the optimum metering rates on each of the entrance ramps in

the controlled area. With the control system in operation,

a significant impro'.'ement irn the- operational efficiency' of

the freeway system i;a r obser'ed, including a reduction in the

a.'era c travel timre cf a '.chicle through the 3.5-mile free-

'-.'a; s section dur ing i-eak '.e r iod from 2'? r i nuj c,- t 1o 17 minutes

111, p. 23j.

Research continu'.d iii ad'valincd crrr.drz control ys-

tcris at there Te:-:as iransp'ortatj ion In -tit ute for the Iie'::t rew:

ears incli.dling the develcop.iment inl testing of the firct

"''miT, ing merge" ramp control sr'stem [i6 Traditional ramp

control z. s-te-m have as a primary ob]ecti'.e the reduction of

f -'ee.;a' c.r, jestion in per iods of peak n demands. .. secc-n..lary

obj,-cti'.'e .of suciL s'ystemiir is to a-sist the indlivid' al ramp

v.ehicle dri'vrs in carr'yini out their mTrin. maneu'.'ers. Gap

aCCe-ptance Sy-teris c ai .accnplis1 .h both objectiv 1 s by detect-

ing a.-ceptabl gaps orn ti freeway and attempting to fit the

ramrp vehicles into tncs.: gaps.. Durilngr perioJ'z of peak demand

the ramp vehicle arrival often e:-:ceeds the availability of

acceptable free'.iay 3aps and. it bccomii-s nce*ssart' to delay

ramTip '.'lieicls until suitablee aeaps are available.. Ho':ever, in

c-ff-peak conditions their. are typically ,iore free.waiy gaps

available than thcre are ramp vehicles to fill them; there-

fore, stopping ever/ ramp vehicle e as a part of the ramp con-

trol alg0.-ritl hr can introduce unnecessary delays to the ramp

.'eliicle traffic. Therefore, a basic design goal for moving

merge control sys'temr is to achieve the matchiirng of ramp ve-

hicleo and freeway ops c lthout the need for stopping all

ramp .'ehicles.

The Texa-:a Trans.porta t ion Institute prototyp, F io''inr,

mcriie control s'ysteni jeeigned to test the tecr hnic1a fea-

.ibility' and driver acceptance of the g imervre ,.conepFt.

Tests .'ere r condrict.d o. n a i u t 1 e.- erntrance r ap mrr1"e'i out

on a concrete rurn'.ay o f th- ie:-:ae .-,. UI Un 'ver- it. Pe. r ejrc An-

De.. P. outr tandar.r three- face tr.iaff sZ iiral '.: er2e as5

thc control element tchl. ramp. It; '.:re r pis e.- a-t

150-f.:ot inter.'-als ,ith the first sir al ,i600 feet froT t.he

nurse of the ramp. The 1iq na 1 .er- cr tr.i i e by a ta, rJ.a~n

fi:..ed tinrm sigqr n i controller .';ith the cjms ariranl ci to prl-' id

the ] -sireJ tiring q se uience for rte light -.r .ai-.i

speed sign pl aced on the ramp: 515 fe-ct upstream ..rf tn--

nose between the first t:o =igqnals. The irdijicatiorn .-n this

sian was changed manually. The ra.liO comurrimjni tion syste'

' :as used to transmit free'.-.'a' speed informn-aticr, t.: the ramp

vehiclel e dri'.er. Tests ..'ere conjuct eI arn e .'luat jt o:n the

basis of i.'herher the ramp vneh ie rhad be&n placed in an a-c-

ccptabl7 position for thr- mergiin maneuver or not. A con-

stant gap for the simulated freeewa- .'.ehicle- of r sec .n3

\as used for all test r..rins. Drivers of the l adirqg and

trailing cars of the freeway gap ..ere ins rtructed to

a given velocity and spacing vehicle traji-ctory data ..ere jb-

taincd by loop detector's placed on the ramp acce leration lane

and out-sitjc freelv.ay lane. Loop actuation lar a ere recorded

and used to prepare time-space -dia-arams for each merge. The

feasibility of the moving merge concept w::i demonstrated by

the e.x:perin-rita, runs conduct, Ini one ,f te-ts

th.-_ ramp driver :s .'ere placed in a p.:.-itin to mer -je .ucc scs-

fully ?i01. c f the time and in anc.thL-r series usirg an instriu-

m nte.,1 .',cinicle, th-e -juc:ess rat.e wasJ bett-r thar, 50 Fec-

,imn er. nations mere maudi e for irsntalla i ion anJ testing ocf a

mO:'"inr merqg control sys'te-m :on an actual free;i.ay entrance

ramp tor test the cricept under reaui-world c':.rnd it ions. This

rcr,-liclu erd.atl,:,n lei e n'.r lntual ly t: the ss 'it ems Jes signed and

impliierer:r d bl. the Raythe,:n Cr:i poratirn which h arc discussed

later iri this pa.per.

A number :of rese-archers repr'-ctet i %.ork irn ..'hat might

be -:alled the .'ysterms area of ranlp c.oitrol from 196C.C to the

pres.rntr [19-?3]. Weiner '21] modeled the velocity profiles

cf free,-.'a:.' shoulder lane and ramp merging v.'ehicloes as stocha.-

tic processes, arid dc'.el,'ped a method of comiputintg the opti-

mum iristant of reiea.-e -,f a ramp vehiclee in a gap acceptance

control s'stc-n toj insure the mna.-:imur probability if its ar-

rival at the mrer.-e area in synchronization i.'ith the freeway

-gap. 'aagoda and Piqnrtaro f[25] developed relationships be-

tween merglinq capacity and system parameters for several gap-

acceptance control s:stermis operating philosophies. lece 1331

developed an algicri hmi f"or synchronizing ramp vehicles into

free..:ay -gaps based onr a ccD'iideratict', of the interactiGns bc-

twe,-n the fr--eeway vehicles defining a gap as described by

car fEllwc'inq theory.

C;naiderable wrrl: in the general area o'f traffic

flo.: theory e::1sts ?If i.hich the case of rajip -' hicii e 'f r-ea-

."chicle interactions in tile nrjrqli'r procD-' s is a

lHuch of the early v.orT- in thi are= a [3-4-,3 ) c:r ce.rns tr, dj;-

namics of '.-chicle mro.-erm.nt *irnd: r fre flo;,. c.nd iti.n J.:,ne-

ani F-3tts 13:] le.jmevelop d what has b-corn.- a '..'idj ly ui;,;j

eter for the descriptio-n rof the quality of Eiclh-ile fiL:'. ai:.r -f

a road-.'ay. Tle;' sugqge-;ted the i ie of a quarnr ity called,. "ac-

celra tion noise,"' defined as the r_. t-i ean .qulare v-.*lue -of

the average acceleratlorn e::pcri-nced b' a "- hic le c.'.r a

period of tieu. Conceptuali. a vehicle with a 1:' accelera-

tion nois- w r I'a be con idere.: to be operatingn under c.or it ions

of smooth traffic flow., an.] con'.trrely, a vehicle a.ith a hiIh

acceleration noise is typically in a situation r..j: cr, th.

operations of other 'ehicics hav.i-- a great =ff ct on it? .

[riotionr. Other researchers [3?--. ] ra'.'~ reatcij oth-er areas

of traffic flo-'. theory. .1 i the case ;.ith other scientific

JiLsciplines, m'ich research .,ork has been devoted to s'-ecific

problremn in areas and n.o general unified theor-' of traffic

floa has yet treern .eveloped which i:ill acc-unt for all of

the conditions of vehicular traffic flow;, density, and speeds

experienced in the modern traffic cinvironmert.

The-- question of th, c.xlstence, stauLlity, and dis-

tribution of *gaps, definedJ as the distance in space or tine

between t'..'o SLICcessIe.e .'ehicles in a freeC-ay traffic stream,

of Cruci al irmripocrtan,.c i- th-. d iT l'h .3 .i 0[_..r1 tL.on o: f c r -imp

inerq rInq control s Ltemri.S. 0i-.Lr 156 i discui.Ased. basic methods

of l..c;.ir:ng at t'h- stati: stical ti. r -'-'roce- that generat

the obs-er. d g.* ZS i r a tr. a ffi.- tream. Onc approJch i as to

focus ~'n -' Q a ttC[Litir' on t he i-i j ir str eam -:f triffi_ inr order

to then study a o.arieti of fi .. anid st-.ra.'i- si situation that

are cr.aterd bC -.ari ous rctssinL g cr merfcinZ traff ic .strleams.

Th.e -.c a ~ptrL.r. h .- -a to, fOc.-Trilat the L st t-1 ti ca pr cess

in the major triajm of traffic inr tcrms of observed con.ji-

tions i n the m nrc.r 3 Litre' An ec.uameile is th, unt c.f a ran-

a.o- i vehic-le a1 cri'.al at .anr interseccrion for a gap of a cc-r-

tain lsze r in the maj -.r str-eam. Solberg and [57]

treated the juiSr t j..-oi cf lag aiid jgap acce-pta nct.- s at toc '-

controllcd a t-grjde Interiecti i.n in a paper of limited use-

fulness in the analysis of free.wam.'- merge control E stems,

but mi._ntirt i en here for i.-ts intcret-ir ] .]i scussE ion of the

compaLr so1r of rcsulIts obtiai ted from three different methods

of'iulatin.q the informia tion eon rejected and accepted

gap d.istributiorns. Drc.. (53] discuZ-.Lse the dat3 collection

and reduction rnecessa r- to. dete-rrine the gap acceptance and

rmergiing del a: character itics fc r the si>x entrance ramps

to be controlled in the Gulf Freei.a.''.' Surv'.illance aind Control

Project mentioned earlier. I-lerging v.'hicles were classified

into t.:o groups, 111 those in whichh tie .dri.'er rejected gaps

before finally accepting. the gap, and l2i those in which the

drl'.ier of a ramp '.hicle ,cce.- tej.c the fir..t gqap ob.-r.-,.-i

at the merge ar-a. The former 'as ref-rred a topped..-,

hicler arid thie latter Fs no'.'iir v.ehlici The cr- itical ijp

for 'tc.pped vehicles wa.' f-urind tc b.. so e :i0 hil-ihe-r than

that for mc'.'in ve'hicies. Ir aidi it ior, le onc l.u.

the "critic l g.Ap," de-fin-ed as a ,p nh e probabit il: L ,f

accc-ptance is 0.5, for the mer eI irinr mnarne '.er fr.:mn a cntrac- is Lndecpendent of the frc: -.'a., olum but i-i9 apparent

affected by ramp -.ecometric3 and ramp conrtrL meia =.ijres.

Qucueij-iL thec.ry used to Jd lri e e:.:Fpr -.=. ionr, for a.. era1 eJ

'.:alttlrn t ina for a ramp '.'hicl- hef-re. merqinrg -n te e::-

pected queLi length on the ramip .;.aitirn to imere. .,pplica-

tions of the dcri'.v'c Jata c-.ntroller settrinr. f:r ramp cr:n-

troi sy'.'-temT-s :were .1 i sc .l5?.ed Mil1 r (59 p;ei-ented a coln-

parative study of ecio ofcti nine different miethods f:.r

analyzing .as acceptac._- Jdata. Pahl 1601 us-ed data frrom

aerial photographs to determine the average space and timri,

sizes of accepted -aps and .ap manriej.'ers for :-:iting free..;a

vehicles close to their intended off-ramp and for thri:ugq

vehicle as a function of the distance fromii the -u ff-ramp.

Again, ..hile this paper is of limited usefulness in the anai-

y'si of on-ramp control system.m it ir inter-est:ino .'ith re-

spect to the insilghts it prov.'ides into the behavior of free-

way vehicles.

rlhe important q iesZticn of me "'tab, lit''." r.f a free-

:.a'y 7iap, whichh carn Le defined as th Fprobability trnit a gap

m asuidrc.d it certain si;'e Jilll rremaina that r1:- at a point

jlr .rn:tream f the initial ,bser'.'3ati n :oiint tre.ated in

.omrr. Jtai ll b; Bu Lhr [ i], and fTliun al, lieu, and Lat rc nce 621.

'Th first ,pa r .JI .'elop.Je r crefsio, Cl quatioi,-s o for gap size

dio.n.r str- am ur E gi'.'nr fr:ci .ay iocattion as a function of the

initial t ize :-r f thie gar3p hci fi -r t .obser--e.. anr. of the .speed

of the gap '-hen first obser'.ed. Trie latter paper

9aalytiz' l :1 *:-' --pes oris for -rr-'rs in gap production and

.F.pld e ti nation from :-.:per mental Jata obtjai, ed frcm aerial

p.-rtrqr.aph s. The sEnalyses discussel include the effects of

.ifrrerinrg sensor scan rates arn, detector location, cn the gap

size predictions. Conr;si.-erra tion i: also ~ i'.ven to the rela-

t i.'e accuracy o'..f sin lie-]oor, detect:t- rs .'ers'us .dout] r e-loop

jdetctors, for thi gap preJdict ion process. Thie usefulness of

the previous t..o papers related .directl" to the question of

the tripe and number of detectors to be used in a free;.ay

'control s; 5tcm to 'chie'.'e somi gcl.-un lr.l of gap preiJict ion

ac cuI ac' .

Results from the classical queueing theory of opera-

tioi'i research are cften useful in the design of a freeway

control s_'stem as the': can give some indication of relatr-.'e

e-:p-cted s'sterm performance characteristics as a function of

various control parameters as mentioned earlier in the Drew

pape er[S ], rther researchers '-:ho h'. e treated the h i anh'.

merg.in7 process as a queue in s-ytem include E'.ans, lirmarn,

and Weiss [63), Jewell [G4 and rlar:us 165),

Since- mereic- control sryst-ems use information : n

freeway vehicles to control the Tmo'.emernts o _f ramp -ehicle_,

it is LiTipcrt3nt to consid er hr.'. tn:h.P data iore -obta A

nimi-er .of researchers have st.iei--.i the queo.i ti :n of djetecto.r

performance ir the literature. Stern 166- :onsidc-red the

use of iiagnretic loop dJetectors o-ri:beJod in thc road s urf ac:

to collect data on vehiclee ch.ra1-teristics. Chri-stnsen [I7]

discussed the use of a cormiiluter and vehicle e length detectors

at signaliz.:d nte-rsoctions. While this ;.or -; does rnot re-

late directly to freeway v n.'rricles, it is orf se in the c.m-

putat ion of ramp vehicle queues and de.*1a i.hich ma ser'.-

as control inputs to advanced frree'.wa rar.ip control s--stem,.

Detector ac'curac' is discussed, and testing or-ce.jurs re-

vi.ew-.d. n a r.-port by the Sperr-' Rahad Corpora t ion [63].

Coura. e j691 discusses'c measurements of traffic

variables on a multi-lane free.'ay i using a combinationn of

sonic detectors an. optical detectors to measure traffi.-

stream parameters. Gazis and Fc.ote |701 present a calibra-

tion technique for magnetic ioop detectors using vehicle

data obtained from optical presence detectors as a b-asis.

Gordon, Dodge, and Scotc [71] discuss the surveillance as-

pects of a computer controlled traffic system. A it enough

the paper is conc-rn r&ed w..ith l.asln hin n, : Dr C., Urban ra ffic

Control S.stem, it is rie-.erthele ss interesting fr.:..m tlIh

standpLint of its discussion o: f the traniismission f V.erhii:!l

,-n- -i-r dat.a ,o'.'er leased commiunLic tio in E and its treat-

menrt of tcliniiiqucs for the correction of iristruienrtatLon

bias err ors.. Trle general theme of sensor rrror analysis is

-ontin e-d iI continued in papers by Pahl (7 ], Huckins 73],

and rlih:all: inr, F'Pane, and IrseKsn [-74]. The latter paper

treats q tion th i tithe estimtimion of speed information

from single-lop detectors, and concludes that the use of

estimation techniques based on the pro.essi ng of sensor n .r,-

timpe asscziaceLd iith the passage of individual '-enicles

rather than on period averages can reduce the errors associ-

ated :with computed speeds.

Ps2earch -jb]ecti"es

The Unir.'ersitc of Florida began a research contract

with the Florida Department of Transportation in 1972 in.volv-

ing the installation and operational evaluation of a green

bind t_'pe moving merge control system on the Ashley, Street

entrance ramp to 1-75 in Tampa, Florida. The scope of this

dissertation includes the following areas of research per-

formed in conjunction with this contract:

1. Systems analysis and design engineering

activities performed to adapt the proto-

type Woburn aie e h ard'are n-d noftw..are

to the specialized requirements of t~i-

Tampa site, and to correct or'rating

problem ex::perienced ' the earlier

,s. ; tern.

2. Th,- identification and .J l.-.'.prment of

computerized techniq..ues fc.r raEip sy SteCi

perfo-rTmnce .evaluation under existing

conditions and after thie irn tallatior, cf

the merJe control sy-stem, including both

the on-line recording cf control system

senior, mode, and field display activity,

and the subsequent off-line .anal.'.'sis of

such data.

3. Consideration of impro.:ed hardware and

softi:are design fa.ctors_ for use in future

implementations of moving merge control


Research Profile

Chapter 2 discusses the green band ;sytenr and Pjcer

system rmo'.'ing merge control 'istemes originally' de'.eloped by

the Raytheon Corporation and tested on a freeway enr.:nnce

ramp in ,Doburn, llassachuietts in 19'O, Desian elements of

tihee systems are rev.-ie'..ed rel]-ting to the need for changes

in the later Tampa System.

In Chapter 3, the Tampa Green Band Systc'-m, chances

mitade to thc oburn system- hardware and software used to im-

p[.]ment the green ban. system concept at the Tampa Site are

di cu3 ed. These changes :ere of three basic categories:

1. Changes required b:. the Tamrpa site

geo3metr, which i. considerably dif-

ferent from that which e:-.isted at the

Woburn ramp.

2. Corrections made tc known and antici-

pated problems in the Whoburn control


3. Operating improvements .hicih represented

extensions to the capabilities of the

Woburn system.

Chaptcr 4 discusses variousu s techniques devised for

the analysis of the Tampa system performance, with emphasis

on the computerized schemes used for recording data on-line

by the control system, and for the subsequent off-line analy-

sis of this information. The development of computer soft-

.ware to iriplerilenrt these functional i dsencribed, and pi-

of pro.-cessed s atem operating dJat. are discusa.ed,

Chapter 5, Designr Conarid;rations fcr Future ;lerqe

Control Sstans, .jis.-ess- thre-e t..asic areas inrcludingir. knour,

deficiencies of the currenL- t rampa system, theljrat.ical re-

quirements and impr-:.'emenits that could implemenrted in jId-

'anced merge control system:w, and the d%.,7i-pmenrt of stani-

alone controller l-ard-'are packages f o:r me-ge c,.rntrol systems.

Conclusions resulting from thle r-es..arch .ork c'm-

prising this discrtatior, and suggested areas for future re-

search irn.olving moving imrge control systems are di scussed

in Chapters 6 and 7.


Prototyp-e Herge Control Si.ystems
Tested in WiJobujr, MilssachusettS

In 1967, the Traffic S5stems Divi,.sion or the Bureau

of Public Roads bean a program with the Raytheon Company to

"_tudl. design, and faabricite a :dynamic merying control s:'s-

ter, ani to s:tu."' the problems of adaptive control traffic

tio' in their urbrn freec.'a corridor" [7f, p. 10l11. An urban

free;oway :as defined to bt t the freei.way itself Dlus as many of

the adjacent arterial streets, both parallel and perpendicu-

lar to the free..'ay, as could be considered to be supporting

traffic flow. in the same general direction as the freeway.

The following design requirements 'were establishedJ for the

hardware to be die'.eloped in the study [75, p. 109]:

1. Ramp .vehicles were not to be stopped unless

absolutely] necessary..

2. Dynamic control over ramp "-hicles was to be

use-d to regulate the time of arrival at the

ramp nose of the merging vehicle.

3. If circumstances caused the formation of a

queue on the ramp, it was not to be allowed

to e:.:tend back. to tihe cit-" tr'eets fe-ijinr,

the ramp.

4. The gaps rn the fred:.'av ,'.-ere t. e detecte-]

earl', enoiiql to initiate a c.ntr.oiled merg .

5. ri m.=rinr inlaneu'.':r '.*'a to b, ia nitorie 3and

tl-h sy-'st-m made respn-r.i-.e tc. cihang.c such ~

io.E of the freeway .:]p a id nort-ci: p lance :.f

thic r.amp vehiclel e 1s .ith the c ncr rol 1i.:irinal is-

played Lb, the s, tenm.

P. Pert urbaticir. ir fr;cuaiy fitw forced- mer e4 )

were tc be minimi zed.

7. The system wa' to : keep track -f vehicle on the

ramp ari. t t to-b respo-nse ie to the or-se t or c'En-


3. The systemm .war to be m, Tje f.! -safe.

9. The system was not to require a human ope-ratcr

for normal operation.

Consideration of tlh.: abo'.'e *bjecti:-e resuitc in

several differing hardware syFtem concepts to [:.r'..'iJ e con-

trol of a ramp-side- driver displayy [ 6-35]. T.: candidate

display concepts .wcre designed and tested. The first

based on a closed-loop control cacnfiquratic.n anj referrcJ to

as the Pacer s stem, and. the second iasm based on .an open-

lc.op control s'stel.m str te&', referred to aE. the green rand

system. F figure ? shor.- the elements r f the Pacer system.

Sevcnr: *.Joublzl-loop dietctor ,3tationns were placed in the rlnht-

hand lanre .tf State Route 12.' near the tow'.n of Hloburn, lMa sa-

chuse tt.t, upstrea.3i of the entrance ramp from Route 3'.`. Com-

iimunicat iorns *.'lre ..,ere run from these detector stations to a

Paytheon MEodel 703 computer located in a control center in

a trailer parkl-.' near th.- ramp. Freewiay Jetector in-

formnation was processed and used to project a single moving

green licht aljonlide the left side guard rail of the ramp.

this information 'was displ]jyed on 7' s ingle-lens green traf-

fic signals spaced 'i91qht feet apat. Eleven .ouble-loop

detector stations were used on the rad p itself to monitor

and track the progress of ramp vehicles as they,' moved through

the sysctm. The relat ior-hip of the time-space position of

each ramp vehiclee with respect to its pacer light .as mont-

tored and used in con Iuction with the reduced freeway data to

control the future motion of the pacer light. Thus, the

2'ztemrr control concept used in the Pacer system was closed-

loop in nature, as the measured information on the position

of the ramp vehicle-s wasz fed back to the system and used to

update thc prior display.

A second hardware configuration for a moving merge

control system was evaluated in loburi,. this was the green

Sf:'LL" Z Fi i; rI,

i L- L .
i>.,71' .r I O I u: :,-;-
I M. *r:-E6

I 1 _-_ "

I- -- -" It 1 ..
, :. .

', ". ,. ,"i Tlj .EL : iT

Figure 3. Schematic Diayra:n of the Raytheon Corporation Pacer
:lo.'ing 'i.erye Control 5;'stem


band '- -ster, ..'hici .ia:" to-. pro'.'id. no direct ad.Lartiv'.' control

of ramp .eniclecs, but rather designed to provide a '.'isual

indication of the estence the hest e te accc-ptable gaps in the

freeway traff ic strearr ..'hich could then be used as advisory

inf t.rmi,.stion fr vehicles s on the ramp. Figure 4 show's the

I.oburn 'reen Band F.: terr. The s.are sensor confti3ur tlion was

used on the frree;ay for tie giroen band systerni as that used in

the Pacer system. ThIe sensors, on the ramp wiere left ir posl-

ticn but *;ere not used as part of the control philosophy, and

nor. informiat, ion ;;.as prov'-ioed to the control alqorithuT, on the

position and motion. of ramip vehicless in the green band ,ys-

te m. ..s the green ban.d s;Lsten concept :w.s preferred by, more

orf thie drivers using both systerns than the Pacer [661, its

use .ija recommended in future merge control s-.terr implemen-

tations and rel:.-vant details of its operation .:'ill be de-

.cribei her-2 to, pro.'ide a frame.iorl- for the consideration of

the Tampa Green Band S.stem Pro]i-ct.

Green Land Sstem C'peratinq Mlodes

To m-et the requirements for responsi'.'eness to the

onset of congestion mentioned earlier as a design constraint

in a froc-'.:ay ramp system, the green band merge control algo-

rithm used in Woburn haJ three different operating modes.

The standard mode of operation *.was the so-called mov'inq mode

i L I'.t FII IN | .- '"
r- Hi~r ii' i Is ij i-l IT

1 - _( :_Iy; J _

,,, / .. / ..-
^b~~ 1rl /&' o g

:il l%

Figure 4. Schematic Dia.ran o.-f the Raythec.n Corpc.cation
GrCeen- B.-ind I-.''iqng lerge Control S'stem

Soarce: [(86

(:-1M;, which prjo]e.te' mo.'irng bands of green lignt along the

r amp display in -iccordance .:ith *.ap3 of acceptable length

,detected on th,.- fre--way. It is important to. note that this

was an open-loop control system, andi that the rmnotins of the

r i7p .*-hi,_ics were not moni tored ..'ith respect t o providiing

inforirn i..n r to the control system. 're bands. of light on the

idillay' 'were prrc.cteJ in acc'.rdl jnce witlhi a'.ailable g aps onlv.

with the onset ,of freei:aY contest i-.n letected by a

reduc,,t i-on in the a.'ecage detecrted '.eloc:tyj of the freeway '.e-

hicles at the last ..enser station JpstreaIm _f the inerge arcs,

the ireen t:.nd S.ystemi changed tao I stopped-gap acceptance

SG) mode of operation. In this i-mode, all ramp vehicless pass-

ing through the s;stcei were stopped at a standard traffic sig-

nal locate, approximately one-third don. the length of the

qreen hand .diplay. This system then searched for an accept-

able gap for the ramp v.'ehicle, and if one was found, released

the ramp v.'ehicle by turning the light green and projecting a

green band of constant length down the display in such a

fashion 3s to ggide the merging vehiclee into the proper posi-

tion for merging. Provisions were made to release the ramp

vehicle e from the light without an accompanying green band if

a specified maximum wait time in the red phase at the signal

had been e::ceeded. This mode of operation represented an im-

plementation of the earlier work conducted by the Te:
Fortation Institute and other research groups in the area of

Stopped- :.iap acceptance. The unri.-qu featr ie ofr the ij.tobtrn

system, ho..e.'er it u-Se of the .reen band di-l[iaY tO-

'ruile the ramp r vehicles s to tthe ruLetir-: ar-ea, as the eari .er

-.*,'stems haJ Ir. mearr :f centr, i 1 inil r imp traffic

than a stanr.ard traffic s-i. ral u:,-- to: h.ol *.* iIcl le ur.til

Suitable gap had been located. I..7].

A third ode f o: peratirn <.as prt..."i.-ed the -top

meteri1ri. (lc.i. mo de of o.,peri atI c -.. In T chir m.:d rar-np '---

tlicl cs were stop;pe: at the t-rafcic lir ht c:.n t-c r.iampl arid,

released at a fi.:ed mel t-rin rate ietermnirned b con -tjnt

'.'al ue i untne control prog rar.. Thi miicd. of o3peLjration ..'aS

entered ..hen frcc'.;iy 'elo ities indicate. tre onset .f lea.

ccnqe&tion ori th,.2 free..'a area, a d .as irri i.atr t'o earcl-;r

i-.rk in ri:.:ed time mete Lrini used at a cf o:the r i n-

stallation- rn'.ol'irnc ramp con ntrol p .jiects [i .

The s.',stem ,a.j a fc.urth mrac.e, repre ent ing a i nl-

tiali-ation phase which entered :'nen the -,'stEim ..i

initial]-l activated each d.a or s.toppel *.,riclcs *..*ere de-

tected in the ramp merge area. This con istc.j of a 15-

second period diurin.j which h free a'y *.'ehicle. were tracked

t-. establish Jata tables f.,r s'l,jbiequnt use b. the green

band program, and d rini.] :..hich the ramp displa" is rni ted

onl, y of a fl -.hin :i amber in. i.lcat ion .a L the stop liiht to

indicate a oroceed-'.lth-caution *-ondition to the ramp '. e-

hicle drivers.

To accomplish their objective. of automatic operation,

the i'oburn Green Band System wa- give-r the capability .f

automatically. switching between th-e v'ariou. operating movies

in respons-. to the changing freewa" traffic cornd itiions.

Sensors downs treatm .f the mere area in the rignt-hanr free-

wa lane .ere used to measure total free'.a' '.-olume across

th roadJway. Problems were encountered with reliability

factors inr.' these ~ sensors in wloburn, arn the program

logic .JesigncJ to switch operating moeles based on volume

.;as disabled, as ..ill be discuassed in the ne:x:t chapter ori

the 'Tampa Green Band System. Control logic to command mode

trjnit ions based .'nly on. a'.'erace freewaya spec-es measured

up trceam of the entrance ramp .-3as used tic achieve the jgoal

of s-stem operation without human intervention during normal

condo itions.

Woburn Green Band System Operation

A flow chart of the computer subroutines in the on-

line control program for the ioburn Green Band System appears

in Figure 5. All control software for the Ihoburn Green Band

System wjritt-en in the basic assembly language code for

the Raytheon 703 7:4 Series Computers. In Figure 6, the

operational task of each program is outlined in the sequence

that the programs are called. The functioning of each pro-

gram is discussed below.

Figure 5. Green Band System Control Software Floo; Chart

Source: Green Band Design, Vol. '.'II, flerging
Control System Final Report, Report l[J. E.70-
420U, Equipment Di-'sicn, Raytheon Compan', .,ujd-
bury, Massachusetts, October, l'1970, p. 4-2.


--I h T 1 L 1 i
T i.E F rI. I[. '

S r L r. ll L i
I fb- L I N
,' I 1 1 :,II" r\'

i*i 'T ii F '.

LT I'I' -(I Li t
S LI 'T LL" I l

''' L F F W.

I 1 I i Pi 1 I] T % 1 I I,
I rt' Kr Ii


[ ,A 1% I ., l .. :C I
I n .T11 1 i I .l i T [ l ,'F ,
,_,F p+T:, ',IT ,' : tilt

\ I ,I I T ' I h ] 'fI .'L. i ,I,_T T

I A I L r 'i ri.i' T i I '
I .I*I T pi i 1T

i '- t i r. I Ft i t F t I E. it
O r F '11.1 1 1 1..;

,I --I L F i TOR i
; ] .r ,r. Ir, ;; 4, ;

ir ._ II,,T t FA Iii [
f .1 I A ,. ,,r 7

:L Ti .
\ '.)i I' ,T | -'.,r. ,Tr ir.- ii L ,, ,'i

C'r 7 If TF161L L r .I,
i. ,h. L'EF ;E rdPT

\t i

E TLraifil T1 .1
. :i C 41.1 / r 1, 1M .,,:
_ | I-i |i. 'L' I Ir F" i

cI .i T i -rj il

Figre 6, Green Band System Control Sft ware Function Dia-
q r a m

'- :acuti'.'e Subroutine

The e.-:e.cutive subroutine scheidu'les tn C call;, .t trh:

1ndid- visual subprograms m~jI'inr.j up the green stand -soft,.'are sys-

tem. It is entered at the time tih program is first turned

on or restart.-l, and transfers to initiaization entries at

each of the component subroutines to set up data tables in

preparation for the rne; :t run. Then, as th. system is ope-rrat-

ingr, the ee-_cut .ive program, is initerrupFt-dj bty an, e:.terna

hardware .lock .it a rate of 5030 times per second represent-

ing a cycle time ii; O 0.i 00L secndL). U p:nr each enCry toj the

executive from an interrupt, the first action is to call the

input service routine (ISRi ihich is co-located in core i-;itn

the executive program. ISP reads four input j rids from tle

field containing the loop detector data and prepare' tables

of time reports for vehicle crossings at these sensors. At

the end of input service r-utcine activity, the ex-:ecutl iv

program calls the data logging pro-ram to record any sensor

changes on magnetic tape for subsequent off-line analysis.

The e-ecuti've program then schedules calls to the v.arcous

system subroutines by comparing their ne.-:t scheduled starting

time with the current '.'lue c-f the system time iord, and

transferring control tc. tre various routines if appropriate.

Thus, in practice, the e:.-ecutive routine has no input or

output variables of its own e:-:cept timing information, and

its CoutpLutS are simply cali lto the *variou!u subrcoutines of

the green band system in the proper sequence.

Input Service- Rouitne

The input service routine IISR) read- four 16-bit

Jata wordss co.nttininrg thie field sensor input bits from an

input. output hardware interface. These four ,words are :orm-

pared i;ith th'ei r previous 'aliuez ie.g., the value. red at

the last interrfup.t) and any bit-level chances are processed

individual" b: the ISP. Tables of data are maintained for

each of the sev'.:ni freeway sensor stations. Uhen a new sen-

sor data list is completed, contaiinig times of entry and of a frecaiay vehicle at the detector loops in a station,

a flag ,eord is set by the itnpLt cr'.'ice routine to indicate

a requirement for subsequent processing by the highway v.e-

hicle- processor routine (H.PI .

Highway vehiclel e Processor Program

Consider the -scheme used by Raytheon to identify

the time of vehicle e arrivals and departures at each loop of

double-loop vo'locity sensor, as indicated in Figure 7. Time

T3 represents the time at which the vehicle .'as observed to

have crossed the leadrig edge of the first loop, T4 repre-

sents the time at which h the vehicle was reported to have left

the first loop, T1 represents the time the vehicle arrival


T4 T1 T2


Figure 7. Illustration of Scheme Used by Raythecn for th.
identification of Double-Loop Sensor E.'ent Times


iLas noted at tnhe second loop, and '2 represents the time at the vehiclee -_-a reported to have left the second loop.

The distances C and % represent respectively the- length in

fetL from the leading ed,7e cf the first loop to the leading

ed-e of the second loop, and the ::idth in feet of the second


The input ser-.'ice routine is responsible for pro-

cezsirng '.ehicle sensor event data into tables of time v.alues

for euch sensor i;hich then :.erve as the input to the ii'.'P

routine. Data for each vehicle passing through a double loop

s-=.,nsor jio nslt of a s.-nsor status i.ord nd i.-at ing the- cur-

rent state of the sensor data lists e.-. are they full,

eiriptc', or representative of a vehicle overdue at the ne::t

sensor doc-:nstream), and information on the Tl, T2 and T3 val-

ues for the current vehicle in the sensor. Also stored is

a time difference &, represe-nting the transit time of that

particular vehicle from the leading edge of the first sensor

to the leading edge of the second sensor. The processing of

completed sensor data proceeds as follows:

1. The time of vehicle e arrival at the leading edge

of the second loop, as represented by the Tl

data value passed from the input service routine,

is stored as the time of completion of the sen-

sor report for that particular vehicle.

2. The estimated timr .:-f .: a rr'.ai I(T.- AI T .f tlmh. lhiqn-

.a'. '.'chicle at the Jo:.:Istrrcaram mer 'g point i

rn'.: crompute d. Tht:e expression fo-r this "aluu:- iS

ETA = 'i + ( *5 '.I, .i e r is the *listarnce fr:mi

the sensor to the merc.e pir t. The result tf

this ca lcul nation ir i T.a ,. f.or the cur r-nt ".-

hic e. If this .'iia e frc ETAr. is 15 than

that for the ET'A c.miputation p-rf:.rmi.i' f.o'r the

last vehiclele, E'IA.. he assumptit n ir s riTade

that the cutrr,:-r t '.'h!.cle 1i. ric'.-In faster than

the pre-'.]ous '.hicle, arid an i ppliicat-ion o th;i

ia'. of conser:.'atCr.n o.f vI-hicies is TaJe, result-

ing in the assuLpti'iLo, that the iurrent vehiclee

..ill sic..'' do...n to a.'..-id .:-ertaking th last e-

hicLe i thi s Inrstanc.-i 3 "'ligh.. .a '.'ehi :i

minimum time sFaciJ. r' constant is aJjdde.d the

previ.ou.u .'ehicle 's ELT ,.lue to. establish the

ETA for the current '.'hicl- That is: ETA-.r, a

= ETAla.j + II, where H = 0. s.c. in the Wi'_burr

Grec-n Band System.

3. The velocity of the freera.'i' .eliclC at the cur-

rent sensor is computed by the e-:.pressi.n .' =

G-' which h results in a v.'alue representing tie

'-elocit',' n feet per second for the '.'ehilcl. If

the computed value for this v..elocit'- Is greater

tninr I L, fe.- t per seco,.Ind, or lessI than 5 feet

pel second, the assumpLicon is made that the cal-

-culat ion hs ns ae ri~meric.l error problem for

chis case and th-e .ehiLie i. ignc.riored until it

pa s es the n-r -:t sensor do..natream. This '.as the

factor '.-'hizh caused. the W';oburnr Greei BaErn Ss :-tem

to brekl'. dow.n in terms of its ability ; tco accu-

rately F.roject free&;a '.y vehicle mrrition during

times. of hea'v. free...ay congestion ard associated

lo,..-speed operation [h] .

4. The e.'hicle length iin time; is computed for the

curre-nt freew.'ay vehicle e by usirngj the ec.pressior

L = (T? TI (U.- V). The first term in this

expressi.:,n represents the occupancy time of the

vehiclee in the second loop of the freeway sensor

and the second term of the computation repre-

sents a correction factor to the occupancy time

which h acco'ints for the length c- there detector

lo.p. That is to say, the freeway vehicle is in

the seconJ sensor loop for a period: of time

equal to the time it takes to travel its own

length plus the i.;.th of the sensor loon divided

by the velocity of the vehicle as measured inr

the previous calculation.

The act'.'it'' or the hi9gh.;a vehiclee pfr'cescr pr.:o-

gram represr entrs a consideraLti- fraction of thle coiip;Jting '.',r l.

load encountered b te rn tne gren r band system hos; *ci'ioputer, .a

i ll b L .'id nt from the aisc.- ic.n cf Lte d.eelF.mer nt of

p imrui lati.rn cf this program, jhlici ;i ll bt- riI:; tie.; ii *Ch. -]-.ter

5. The output *of thi H''P ro.utin.i 1 a table cf 'alueij f_.r

each freewv'. vehicle ii, th-e area up:tr am of th. r,,er.:i, a[tte

monitored ,by the 3.s:n':.r st a Etions 3 Cci. ted .,itl ti, s'. tEim.

Four pieces of infrc.rnation for each *.ei-* cle are -tored 3z

outlined ab--e. The T1 value, the ETA at rerge pint -.alue,

tr e vehicle e lean th, and the '-locit"' of tli 'el-Iicle for each

up to nine vehicles s are touredd for each sc-ns:r stati.-n.

Since there are sc"en fre-.; .a s-ren 'or = stations, thi- r ;pr, -

serts a total storage in the table of. 7 :-: :-:; = 2'5 '..;ords

of freeway,.. hicle ii fcrmaticn.

Data LoJ.-iin Subpr.oramn

Thli? routine processed sensor data and .sst-tem imode

information into blc'cl -E .of data for rec-ording on a digital

magnetic tape recorder supplied w.ith th-e green barn control

system -hardware. The re:ultinq tape, were read and proces-ed

off-line for s:'stern pc-rfcrr7nance e.-aluatici s. Operation of

the data loggin'j program wi.a controlled b, s-.nse E'.itches

on the computer control panel.

The data lc:gging p-rogiram '.*as called every 2 milil-

3econdz after their input s-rvice subrout ine had finished its

duties. The data logging program read the four field sensor

statzi-i words fro t, lh' intLr face hardware and tested each

..',rd for chanqge If an.' one bit or group of bit r in a

u-:i'd had ch-nged state, the entice four-word r ensior block

..~. tore,- in one of tjo 14-" buffers in core. In addicicrn,

a time tag ';.c-r3 .'as read i from thri- timer .value word in core

and szrto- red wth the scE-nlso status ..'ord3. When the buffer

currently being written into '.as full, it w'.as *.rittcn onto

the tape b.' a cape recording subroutine. in the onbi-urn Green

R'and System, the data logging subprog Lr a \:.a modi fied from

its P:' counterpart to include drata on sy stem mode Ftatu.

and current green banrd Speed with the other information re-


A-s mentioned abo.:e, a tape recording subroutine in

the data logging subprogrami acconpli shFed the transfer of

stored buffer data to the tape dri'e unit. In addition, af-

ter every fourth sensor change the status of the drive-r dis-

plad' liititzs ''as stored in a 15-,..ord block in the tape out-

put buffer by' this section of the data logging program logic.

The structure of th- 140-cl-ord data buffer as record-ed on

tape ~.'was a shoun in Figure 8.

S. SE!S'I' U'-l D

SEIISO5t'. 'i.'i .D

.- IlD in'rD 'li'^rD
*,-.lii I flh. i.5ill- U

AiD i[LL Uii U
WI. 1'F. i

,-L, TISi. -.JR.D: ._ G,, U F
1_ l i.O P:.DEf. i-Ii C"RFE(I C.I PPI'.LD
.1uPD TPI 1T
S-'..TEL I iI':D.E iOR.D LiFrLi'.
GF:ELN 5.ElD Sri' j J''."L'

iS 'API Ui'. l:D
Li..L. V 'C U C I' A' O'.'E
L',.I-A Gir-'O jIP '. .i ''. ', L
DATA L.'_i[ *;, ri. r

Figure 3. Data Loggin' RPo..utinr Tape Output Format for \lo-
burn Green Brnd 5 stem

Green Band Status F:outrint:
Green Barnd Generation
i .'b.rn Constant Pamip Spe .e-d Case)

Green bands -were genricated Lb thne ioburr S'stem

control program in the foil lowin manner. On each entLc. to

Lhe Green Eand Status routine fi'.5S), the rhigrJ.way'' vehicle

.senso;r lists prepared by the H:.'P ;ere checked to see if an

hiqgh.ay vehicle_ with mner e point ETA's greater than thi

most recent time of entran-:e to GES .were in the lists. If

no such vehicles were found, thi highi'.'a. segment upstream of

the ramrp eYt.-trding to tli: most distant sensor' location was

assumed to be free of traffic, ard the entire length of the

green band display ias Lurned On Lt indicate the availability

of a freewa- ''"gap" \;h osc length at least as lo:.g as that

of the display .

Ass.iuing,, that at least one vehicle Iwas

found ir the sensor lists, a band was generated by taking the

first '.elicle e.:petcd at the mec.-r e point Ithe one hosee ETA

at merge was the smallest future timer and using this vehicle

as the trailing highway '.vehicle for the first band on the

display the one nearest the donstream endi The leading

edge of the hand was set equal to the length of the display

161i feet for Wobur-n) and the trailing edge placement

found by the expression

X:TEFi j. = : 9 '..IF'. (ET( .T TIME THD'E),

where 392 is the total .distance from the start .:if the Wobjurr

Green [and dispi :' to the merge point in feet, '.'iFG is the

current green band speed in feet. second, ETA. i the l .-..n .tccc.

time of arri.'al of the frr-eew ;y .'eicl at t tre ier.-.:I.e pcint in

seconds, TIME' il the curre-nt timc in .seconds.r, and THTr' is a

trailing hiead.-:, a allowances representing a safet-t factor 1.2

seconds used fo r tris i.'alue 'in Ic W ijrn i. If the first

freeway ,ehlic.ei ETA wa3 s-.inall with res pFct to lithe 7a rent

time, the compute the ot trailing eJle ofI tlie a associate band niay

have been off the end of the display. If thi-s aS. iIn fact

the cUs-, tlhe band i.'s stored in a green and *output table

but was not displayed. The ban.J s-.ree-d f.r tlhe first band set equal to the current green ban.j : peed, V.'MF.'3. Agree n

Pand Status continued to search the highway ; list- for

pairs of leading and trailing vehicles. If .ny given v-

hicle ETA at nerge ;:as less than .,.25 seconds greater than

the previous .'ehicle, the assumption .'a miade that these data

represented a downstreamm vehicle w-hich had not yet been re-

mo'ved from the current sensor list, and the search continued

:ith the nexr vehicle in the lists. if a vehicle with an ar-

rival headway at least 2 seconds *reater than the pre'.'ous

vehicle ETA ..was found, the measured gap Let:'een, these vehi-

cles was computed by the e:.pression


where ETA aI nd ET,- reL peseent the estimtt-:d merge point ar-

r1'.,j1 times for the trailing and leadinr .vehicles, respec-

tivel-.', nrid LEUGTHL re.Fpreent the length in seconds of the

lead .*-hicle. i Le measured 'jp tested against a "re-

quire ga," who's- dJetermi nation ill tLe di scus.-ed in the

ne::t s c.t ion rand a decision tc. establish a band for th.e gap

in q.uesti.:n rwaS irande if the rrelasuredi gip ws greater thal- or

eq ual tc, trh required q.a time forr the two '.e.hicle If the

decisiorr .;as po.itive, the ith intemendi ate bn.d to be dis-

pla'ed w-as creat--. by setting

::L = :' '.1 RGILT. TIME + LHDY + LENGTH ),
i L L

:-:TE = .XL.E tIG.AP) iVL.'IRGI, and


wher.:- LDHY is a leading headwa' aillo.wan.ce safety; factor (0. 4

seconds in Woburn The process outlined in the di s.-cu sion

at.o:e w~s r._.peated until all pairs of fr'-e::ay vehicles in the

sensor lists had been processed.

When the la-t '.'ehiice in the sensor lists \:as en-, a band was created in the Green Band Table with

the following parameters

':LE = 892 VMRG(ETA TitIF + LHD' + LUJGTH i

TE.. = 0
L ;sbT

'EA-IID. = '.'[HPG.

If the last vehicle had a large LFA at merge point with r-e-

spect o t he time of generatioll, tre computed :'.LL 'alue *co:ul-i

ha.'-: been negati *e, repreSie-ntin-- a ban-d ..h-: -iose loAding e

%iculd rnot be on the phyl. ical display, anrd thiiu oire which

could nor be displ ayed, to the -eil:].

Figure 9 .shhwsD\. the green.C band plac.j-mrients on t i-

ficid display for thie g.eae ration ro tie first barid, incer-

me.diate band i, and tiLh last batr,, to be creat--e iri a given

call to the 'IES routine.

Green Banrd 3 tus Poutine
Determinat icn of an .'.ccert alle Fre *.'., Ca
Size for trne Green ajnj Launch '.ten

Since the major operational nmi.;-on of a mo"ing

merge control system is to match ramp .'e-hicles '.itlh a.'ailabie

gaps in the fre-wa'. traffic stream, some method of classify-

ing detected free.:ay gaps into the' two mutuall.. e-:clusi-.'e

categocies of those .-.'hich will be used for an attemiiptc'] merg-

inq maneuver and those which uill not be u.-t-d muit be prc-

vided. One Ltoehnlque for pLerformirIn this =sel--ciion. procedure

is to consider the statistical properties of the gaps. ac-

ceptLed and ree-cted by dr'.'crs at a giern ramp, thus c-eta.-

lishing a "critical gap," defined as the sie of a gap in


mp -'-


.-..- first




mp -



Pigure 9. Green Band Display Clement Par.ameter Definitions

time .hich 1ii be accepted b y 50 of the dr .'r:.r ui-ini the

ramp. Then, in the on-line corJntrl 'I : temr, all measured

free..uay .apz smaller than this r critical ga.p are rejected a-

unacceFtablie, and nLmeasured gqip' ? reat--r than r equal to this

value e are u.sed to program the ramp vh'.'i:le into rirgli ng

ai' icl on. T pica value- for, the critical qap r'aram.-ter 3t

fre.~.a': entrance ramp.s are in the range .f 1.0-4.0 se.-c n.i

[44], and nima var. ':it-h ,-hanling spe d and flo,; c,_niti-n s.

Note that this critical -ap 'value represents -eriapC the

most important dynai-ic ramp control .' teml o-peratinr3 par-m-

eter, and its '.alue .should be carefully selectLd, a- -.'lue-

'.hi.7h are toou Ic: ..;ill cause ramp ve.'E ic 1- s to l.e .r'.jected

into gcap whhich are to.a: mi all r adjcqulat vehiclee separa-

tionr, creating a potential safety hazard, and v.alues- for

the gap acceptance parameter ..hich are too high ..ill cause

unnecesrar7 dela;" tc be imparted tc rarr, v.ehicle- ..hile the

control system is sttemptiny to find ,aps cE the larger i.ze

un the fre '.:a.'.

.ncthr approach for cla'sif ing free."a.' intc

acceptable and unacceptable categori.L-L is oc consider thc

relative trajectories in timc and space .tf the vehicles s in-

volved in th- mer.gig proce-;s at the instant in time the g:ap

i detected, and to use this information to coi mputc a re-

quired gap to accomplish a safe me-rqinq operations for the

vehicles involved o.g., the ramp vehicle and the leading

a.inj rralin freeway v.'ehicle dsef rin.! the gaop). Then, if

the requiired p to thhe mere ai th approf r iat

safety margins is less than or equ~a LO the current measured

p on the free..ay i:he curr-ent gap bccomeS a car dida3te for

use L0 the merge control system. If the required gap is

g.reatet ti .a' rrh mSni ur-e:d ..!* p, th. current fre.~:-ay sap 1i

i-ct]ccted by thi merge controli 3systemr arnd. the search continues

it the ne::t .ap. This latter apprio,'h was used by

Faythen'r in both the Facer System and the Creen Band System

to evaluate freew..ay g.9ps for acceptability 193, pp. 4-1,8],

anid th, implementation scheme used to carry out this proce-

dure inr the on-line sy ;teim %.ill be re'vi;eed here to establish

the framew.'orl: of potential improvements Lo the Paytheon green

band processing logic to te discussed later in Chapter 5.

The required free;.a gap size for a ramp '.'chicle in

the green band system is c::pressed as the surr of two terms:

FGAP = HL + H,

'.lire PGAP = Lengch of requireJ ya.p (in seconds),

L = Required minimum time headway between ramp
1.ehicle and leading .'chicl of the freeway ,

alId HT = Pequired minimum time headway between ramp
vehicle e and trailing vehicle of the highway

The icoiipo'nents of the require-d *gup are .aiiputed ir,

the follo,.iing manner:

HL = 1- i + rl' + hi.

H_ = h + 1, + h + L'

whCIIL'L- and h rpresei r, t iad..' ai ,.;.anc,-. ba--,J

cn the relati-.e .- c ti -s o f the hi.'hria, 'e-

hicles defining the *gap- anri the r.amip "eli ,i.c- ,

,1 anrd h.,' reppr- ri hea..'ay alliances t5 c

account for the in pre.diicting fr--ee.'a_,

vehiclee arriv.-l triires at the merge area based

on point estimates n'f .-liicle .-0locitc talern

upstream or the mierqle area,

,, represents a mininmurim aceptable fre-.ja ga

threshold time t0..8 seconds ;.aa used rin iLcburr),

and L' is trie length 7f the ramp '.rehicle ir seco-1nds.

Recall that in the Pacrr E-ystemr, ramp vehicile.- ;.re

indi-.idual'l track-d by' the cn-line control prcoqLam under a

closed-loop control strategy, This supplic. tih control pro-

gram -..lth the values for length and ,.'clocit,' fcr each ramp

vehicle e identified as L' and ',' abo'.e. In the green band

system, howe,.'e.r, Indi.idual ramp -.'ehicles are not tracked b.

the control system, arnd the gap calculations are performed

using the currenLt green banri speed, 'i.F:G, as the ram p vehicle

.eioccit'y and a constant representing the value of the length

i, n time) of a *.'hicle 15 feet long is used. for the

of th? mjerlinr-g vehicle in the coJiputations.

Define the following '.ariables:

'. = velocityt y of 1-adin freeC .ay '.vehicle

V '= .elocity of trailing free.;a'y .vehicle

'."PG = Current Green band Velocity I assumed
'.'locity of ramp vehiclele )

= ..asiieid -.ehicle .i ecelerati-n = feet.'second2

= ..ssuriied 'velocity at which 3 vehicle has a
zer) accelcr.tion c-ap.pability = 14i..7 feet/

= A.-celeration capa3tilit; of a vehicle m'.'ing
at zero velocity = 5 feet,'secrond.

Then, the equations for hi and h, as a function of

free '.'vehicle velocities and green band velocity y are as

shown in 'Table 1.

As indicated earlier, h and h,' represent addi-

tional gap requirements based on uncertainties in the measure-

ment of frect.ay vehicle merge point arri.'al times due to up-

steam velocity measurements. Raytheon estimated this error






it h+ when the imeasurement was tarc i a t the fret.ew detector

2ta tion located l, 1,'.i feet ups treair if the assiimed mere [76, pp. -1. ],

Let ETL = Estimtdtist ime oif arrival .of leading.
free-.y vehicle at the merge point,

ETA. = Estimated time of arrival of trailing
rree.ii'i .enicle at the merge print,

TIMlL = Time of sensor report for leading. ve-
hiicle w.hen its E TA c-lculatiL-)n done,

and TIf-IE = Time c.f 3se.ri- or report for trailing .'e-
ricil-2 when its ETA, calculation was doar.

tihn, the RPayithon expressions for freeway vehicle e

ETA. measurem.ant error correction factors were as follo,.s:

hi' = i.i'i7' ) If (V' I ElIE TIiE )
1 3 L L

h = 10. -. / 1 ,,1 ) I ETA; TitlE )


f t'..') is defined a :

0.01, '.' 83 feet.'second

t 3(C' = 0.01 + ( 3 .'), 3.j5 14.7 feet/second
< .' < 88 fe-et.'second.

0. 10, V < 14. 7 feet. 'second.

This completes the mathematical specification of the

components of the required gap computation equations.

Green Banrd Ubpdate Proutir e:
Green Band Propagatir:n .-.L.-.r Fi :- 1 Display

Thc- appeF arance of rio)ti ,n of th: bands .:.f liqht .:.n

the- fi ld display re-Freientirg acc9.etabl.- freeway g aps- is

achie'.ed in ti.o i.a;s. First, each time a band is-. -cr-eated

by Gre .n R.arnd Status (GBS) data represernti n it; parara,,et-rs

lleadingJ edge, trailing ed-ge, current spee.l .re placed:: in a

table. of data w.'hicr car store jup Lto i balds. 'ra tnPer ro,-

tine, Gre en Bard Updatic- (GBiiJlD is then :al led to ijp.-Jate bLand

p.sittioris .and disFlay them to the f eid hardLare. UpdLiat inr

.is ever.' 1N mi1 l iiisecc-nd s, and c.nsista of applyiing r i-


II DX = V. it) ftr a constant sFpeed bL.,rid,

or 1 i = .' + ai tl and

[:X = '.' it + ]. ai t) for an ar, elera.tin.
hand ,

to the stored parameters. for all bands in the table before

displaying them to th. field.

The second f3ct_-.r at '.:ork to achie.-e band rn.o-.'v-erenlt

is the band generation pr-.cess itself. At each entr: to GiE,

the qreen baid table band count variabi is cleared, effec-

tively removing all current band, from the field display ..ere

the table to be displayed at that instant in time. This pro-

:ess occurs e-,'ery 10' mill. 1 i conds and c .ntin ucs .:ith the

creation of a ne.q set of bands in the output tables. If new

sensor data nr.l:'- been received since the last ent r, to GBS,

new band p~ itionE m ili be generated for the gaps previously'

positiojnedr or the field display during the last :!al, ard

siriiarly, as the band placement fLormulas use the variable

TIME in the comiputatio.:,n of leading and trailin-g cdqg place-

rment, each call to BS will resLlit ir. new band pos itiorns

e.'cn if the zensor data base has remained the same.

Thus, the two distin:!t proc2eses of ill band genera-

tion, and 12I band updating and display, work togcthcr to pro-

'ide the apFparent forward rrnmtion of the *green bands along the

display. The creation o'f a new band table every 10 r miili-

sec.nids doe- allow the displayed bands to represent dynami-

:ali3. changing gap lengths on the freeway, but does give rise

to certain bajrd stability problems, which will be discussed

in Chapter 5.

Situation Display Subprogram

This routine generated the control signals used to

dri"' an operator display board in the controli center for

use with the ramp merge control systems. The board consisted

of a row of light bulbs representing the upstream freeway

outside lane traffic for both the greer. band and the Pacer

systems, and rows of lights representing both the ramp ve-

hiics and tne pac-er l;iht.l in th Pacer 'ys;tm- .aid zionly

the grrL-n band display element in tr!e iWit-urn GCren E.jn 5 ?-

t R-j. s ; t-he gree-n band cs prL'Qgrm id not uCe the r mpj

chicle processing of the a-'cer s2's.I.T em, the itu-

atiDion display h-oard contro-I, prr .igr m ,id, north nha'.'e inf.:.rrnij .,-n

available on the traj -.tcoriucs of ramp '-.niiic -h th_, .c.uld.

r-e used to drive i- .1pi I ht. reqpr-ps.r i' ,itng tihelr '.':lhicles.

Fault ! Sjut-prc.qr..

This routirne was Called to- perf:rm "*rics co.:mputer

ar ithmetic and sfensor p.rf:.rmance checl:s *:n the r imp,-,l

system hard...ars arnd to: inform the c.erator .i.3 the tele't'pe

supplied with the E'sytemi of potential err:.r _:.nditir-.n O ni

the 11 ramp sensors and 7 high;s.i-y s-ensor st.atilons ir ti e no-

burn installation wcre monitrced. Fault ol:rrarnce. (e.g.,

..'hat per-_entr age :f .ehicle?. out ,of .- f .-:ei riuribecr mi.a, be

missed by a sensor before a '.-arninr is printed) ,erer p.arjii-

eters in the control program l whichh could be adjusted to q.ive

the required degree of system se.niti'.' ity to potential err.:.r


Green Band '.'Jlcc it' and .'oluTmh Frc:,ra.i

This routine computed thc ..verage hiqgnhi,m' vehicle

e.'locity at the first senasor upstreamji of tIe mere area in

kc.burn tal:nr over a 3-minute a'.eraginq period, and c.1:mputed

the c.-orrmbin.e ramp anrd i'nilghuay traffic '.olume taken do;wnstream

of the miering are a acro s three freeway lan-:es. Th.:ee t..wo

av.e *rag- 3ier, updated every 10 mil lisecond upo-n entrc to

the green rand status program 'ia 3 call to th.e -.elocity arnd

'C olnie sub'proqgrami if require l jdue tO sensor acti'. at ion in

th: pre'. ious tim i nter'.a 1 This routin-e operated on "vehicle

'.'*loc t;l/ data store-. in core. by the -1 hig:; vehicle. process-

ing program arn also Jirectl'y .jce;sed the sensorss for the

do:. n-tream .u:ouihr cormutations b"' reading trh data word con-

Laininr Lh.; appropriated bit -aluLes from th.- irnput.'output

hj interface.

The logic used to perform the -e.'lo:.t a'.eraging

o";r a three-minute period was. iome.what unusual. A table of

'L60U 'ordjs of core storage contaiinqin the '.',locitieo and times

of s5en=or reports from free':way vehicle e data was maintained.

On .*ach entry to the '.elocity .'olume subprogram, the table

::as sear-ched for the .veicles includeJ in the previous three-

minute a'.eraging period and the totals were accumulate.J based

on a consideration of the results of thie search. This ap-

proach of storing past data in a relati'.ely large block of

core indicated that core storage spac was not a critical

problem ,with the green band system as implemented in Woburn.

Had this fact been a problem, the a.-eraging procedures could

have been accomplished trough some form of ex:ponenntial

smoothing to alleviate the need for storing past data values.

lroburn Iree.-n band Syr teni Mode Transition.,

Recall from tnr .-_:rlirr d iscus'3 ons that the IJ'Uurn

Green Band Systen had, four possible .-perating indes. liese

were init al3 at ior, t the 'ttop mete rintr id S the t.oppcJ-

,iap acceptance mode i an the moving ri.mode.- iflii i o,

as indicated earlier the s'..stemn had .arL .utE.matic capatlit'y

for chanciin bet:.en these imorides in re=Fsponse to e\:l-.'virn

higqh:.ay cor, nici ion tlgure 10 sho, S; the basic ch -ine be-

t\weer the three operating mo,.:le :f Sri1, .G, and I.1, and fl-'e

addit onal mo i transition' which .' ere possirb- betm.eecn the

initi al i'tion state and the .various o-per.ating imides. rigure

11 shows the green bar d system mode deci-siorn space w-hich

used to ectablisn the mrio-c in which the i ;stc-m would be op-r-

ating at a given p.:int in tieii,.

The basic control parameter for mcde transition is

the obsthe OL d thee-mlinte velocity average at the first up-

stream sensor. .'alues of '" and '.' in Fig rei 11 repre-ent

the boundaries of a transition zone that .:.:isted bet :ceen

stop metering and stopped-gap modes of control. The tran-

sition from stop metering to stop gap requires that the '.e-

locity be greater than V.. Similarly, the transition front

stop gap to stop metering reqLirrs that the velocityy be i-_-

low which ,as set at 25 milcs per hour, representing an

expected condition considerable re coneidrable reest o ion. The

Figure 10. Possible ! Transitions in the Woburn Green
Band System






~ju LL


hori :zntall area beti.een V. ard .'... represented a transition

zone which pro ,iied a hysteriis _tfrect co reduce the possi-

bilit. o.f fr.--quent mode transitions if tre free,..-ay traffic

ias 1 lating about som- f ixej speeJ point ..ithin this

ranqg. Similar transit tion zone allowances ..:,re used in the

rransiti c.ns to and from q o.'in mia e to Stoppe.-gap. Durinj

the first three minutes t f Uop ration, the 'v.'.alues for '.', V ,

'. and V art r.' specta..'. 1', 25 35, and 33i miles an hour.

.Aft-r trh- first three miii.nutie of system operation tlhese- bcunrd-

arie-s ar- change to 15, 30, 35 and 40 miils per hour, res-

p r- -ti'.'el'. Thi' s, th. basic corns i Jerat ion in the choice or

these values Lb-:comes the ability to predict the onset of free-

.':' and. or ramp congest ian, '. hicr :.ould reduce the I usefuIln',-

.,f the moving mod:e o*:f operation of the system, by the use of

real-cime freeway ".'elocit'. measurements.

lote the exnistenc-- of a ,.'olune lev''l in FigurI 11.

'.OLcF 'hicli is, also -or, ldered for mode conirtrol purposes.

This figure r.iprese-nted a .*olume whichch if reac-hed dow.n-

stream of the mere area, w.'ould cause the system to begin

the transition from mo.'ing mode to stop gap. Although the

capability for using this va.olume control as a mode switching

parameter existed in the W-oburn Green Band software, the

".alu of th, control parameter for volume switching iwas set

to such a high limit that it would have had no effect on nor-

mal system op[-raticn (a comrrurment in the control program list-

ing indicates that this- .-olume limit is su-ch that it .;ill

ne.v r be reached [83, pp. 5-5' .

ihe d-shed arrc:..'s in Figure 10 represent pos -, ble

mode transitions between the initialiration mode and the

.'ariou s operate ng ird.-s. These ar.- ind icate. to sh, .. that

ani of the three basic o*peratinq modes can b. ecnt.r.ef. from

the initialization L-outine \when the ytcr is starte.d. Addi-

tional mfrode changes were possible to the initi li iatlon

state in the Woburrn System due toc the fact thit t-he tico

modes of stopped-gap acceptinrce and moving mode both u;=e.d

the qreen band update routine to move bands that had beern

generated down the ramp displ.i;a, and when the Wcburn green

band update routine w'as being used, it monitored tenFor

occupancy in the merge area from thh presenc-e detectors in-

stalled in the acceleration lane, and had the capability y of

causing a system reinitialization if iny one of th: fiv.'

merge area detectors were occupied for more than 30 s:scords.

After a 15--second reinitialization period, the initialulatlon

routine would then cause the sy-.tnm to enter one or the-

three bLasic traffic processing modes for the continuation of

normal system activity. This latter action .was done even if

the presence area detectors were still occupied, there; in-

troducing the possibility that the system could c:.:hibit be-

hav.ior consisting of the display of bands or stopped-gap

-.,nicle iunching actv.ivity for a perlc d of 30 seconds, then

dropping back to initialization fo.r 15 seconds, and continu-

ing tnis cycic of 30 seconds of dis-play and 1i seconds of

init i i ion This could occurred in periods of high

merge 1area occupancy such as miight have been the case during

periods of heavy ram-p demands in the peak rush hour.

Another problem n ith the use of initlalization as

a system rc-spons- to detected congestion in the ierging area tnhe potential for hav.'in a sy'-tem come out of initiali-

:ation and begin generating green band which would then

direct .'-ichicles on theC ranrp into vehicles possibly backed

up from the merg- area into the ramp display area. This

would d have represented a con iderabie safety hazard, but

fortunately -:as not a problem in the ooboirn system, as ranp

congestion never became th.t se'.ere.

Results of the Massachusetts rests

The Woburn site tests of th- dynamic moving merge

control systems developed by Raytcheon had the dual joal of

11) assessing the technical feasibility of such s ,stcms under

actual field conditions, and 12) measuring the relative de-

gree of public acceptance of the control system concepts.

The successful public operation of both the Pacer and green

band systems in the latter months of 1970 mrt the first goal,

and results from a questionrn re eJd t:. f-ramp *.'*hicle

,ri'.'ers using the WiN burn E,.-site irJndicated anslE r.- to the

latter of the tw.-o oL.]ec.CL is iF6- ]. True and Roen ['-7 tli results of thl .- la .aclhusetrt test ',ilh the

following c*onclusions:

Of thojs- drv:.-trz u-inr3 tlh pFaczr
-ysteim, about ,):.. four. it lielpful w'lr:-
5'. thought tt- t it .-a, dif icult to f,..i-
lo'wv the pa:-r lights-. Witr the green ain'd
syst.TIm, S51 c.ns iJerad it helpftll a d- ..:
co[sii:-rad it di[ to .-ir e LerSiid
the green ba,:d. b-ecraci' thee r
inr.-oi-.' s '.-ral uestio. th. [-erce.-e ta
dJo not add up tc 100,. Sirr:e mani drV_.ere
useJ the- test rai',p'.' o0. er a time
perLio,] including both sy-'.st.em-. tests, it '
pos-sible to dtcrirmine their preference t-e-
t..'-Cnl thic tw'.o sy5 stems Of the 2li0 .iji ers
".ho used bo tl sy; ter-i inrd r turned i- quest in-
naires, there .-was a general pr.-feren.c_ for
thr gr.-n tsnd system.

Based or responses frum tihe qu-~s tion-
n r ires -inn the operate c.r ctf j P.ace.r ard
igreei band systems in the lMscichusett.
tests, the following fi.'e tatemi].E:it c-n be
adej about Ilhu.n-g 1merg.e control systems:

1. lo'.'inq nierge control s3 tems are
technically feasi lre.

2. Approximatel. 30i. of dri'.'ers feel
such syst mTis are Lei-cf icia1.

3. mergingg position! for ramp v.'eh1ies
are impro.ed.

4. Dri.'ers prefer the ocme.;hat simpler
green Land system .'hicl is also thne
iore ecunomlical system.

T.. The reeen lban s'stein deser.ves fur-
ther field test -:valuations.

Tr e ia3iri[:.J ; Greern Brand steri E::-perirnment

Eased on the results cited abo.'-, the next step, in

the e'colution of practice al mo.'rin merge control systems '.as

to. install .and e-a luat the pr-eferrcd green ban.- system on

a ramp ; .hlch -as ::Fpe ienc i ing serious operational miTrg.-ji

problems, as ocFos-ed t: the 'Ic'bu-jr r.amp, which ,:as of .'oc.

qge.mteric deslin characteristics an *-.ics not considered to

i-e been a z'.ere enLIjgh test of th merge concept.

Thus, a se- rch l.'as beaqun t.:- identil: a rarmp where an evaiu-

aticn could be performred tc ans.jer the basic *luestioi, of

"should a ireen bandr s,-tem rbe- considered a "iaile alterna-

tl"e for imripro.'inq the merging operations.- at a problem en-

trance r.amp?" [.17, p. 240]. To assist in the search for a

jrobiem ramp, the follo-;in.? list of test :ite- requirermenlit

'.j d-.i clope d:

1. The entrance ramp should be other than

the clo.'er leaf cype .ith a restricted

3lght distance.

T. The length of the acceleration lane should

be 400 feet or les'.

3. The ramp to be used for a test subject

should have a hirih accident e:-:perience rate

with o'-er 10 accidents yecariy at the test


4. The rjamp shr~ ud 1i h 'e a re- lat1:.'ti '' hiqgi

off-pea : mergq ii. -.' luiime '.1 h t'-.'s wcil-

Jdfirnei. peal: pericodr of tr iff icl daily.

EDuring the--e FpeaI peri' ods, rap. *.volume

should re app 'ro::imiate ly 41'0 t-c 6' 1 --

hicles per ihour, with a rrgi.rgi iq problem

onto the freeway. 139 ,

5. Tne minlinum rjnip ieriqth s1-.lii ild re- 500


G. TIie rainp sLiould li.av.e a cuitahle area

av'.ailable or the left side cf tie en-

trjnce riap for instillati,,rio of tli green

band driver display elements.

7. There should be no iFutream frea:'l :rn-

ramp cl.os.r tnan 1,300 feet, cr dow.n-

stream off-ramp closer than 7,li00 feet to

the test ramp.

3. Test site should represent a ramp wiiere

reconstructcon of the rimp facilicie. to

eliminate the unsafe or c.apacity-reiduciinj

ccnditiorin would betm too costly' or iinpric-


r.-', nluT Lrer of state re ponrded ;. ith candidate, sites

for the ihr:talli at rion of the- ri.erge c.riontrol system. .r.ter a

thoir.,ugh ce'.se :f these ci.. bI'/ the- Federal Highway

Adminiriitratiic.n, thci. A hley Stre.t enrtrair c.. ramp tr.. I-75 In

TauTiF.-l Fl.crid. was se -cted ftc-r the irstalla tionr and e-.'alu-

ati.jn rof a green band m:.'.n.r merge control system under sub-

CsanidardJ meiirg n corinditiorns. Characteristics of the ?-.shle.'

Strec-t ramp .ir.h re.;spcct to the criteria mernt icried ab.ovc a. foll .-.:

1. The ramp and freca-.'ay .ere Lboth on elev.'ted

structures creating severe sicl ht restric-

t ions.

2. The accc.l[cration was 318 f.ct in length,

''hicn is conisidczlra'bly shorter than current

standard ftot entrance rarips to interstate


3. The Ashley Street ramp had a considerable

prc.blnem with accidents of the rear-end

varietyy 'with some 38 accidents of this

category cx.:perireced Iri 1971 alcnei. Con-

tributing to this accident potential was

the high angle of cori'.er-ence of the ramp

of approximately 8 degrees with the free-

w'ay traffic stream.

4. The .shley Street raiinp had an a'.'era.e

measured vol-ume during the 4:00 F.M. to

6: :i0 P.M.. weekday period of .approx: 1i-

matel-' 680 vehicles s per hour, with se.-ere

queuingi problems occurring ,n the rarif

during the;-s tires.

5. The ramp lernth Wva 650 fret, entry point

to nose.

6. There was an gourd rail on tne

ram3p structure avaiiabic for green band

dlipla'' mounting.

7. The nearest was 5,000 fec-t up-

stream, and the nearest off-ramp wri 7,500

feet do'Tistream frromi the A.shle' ramip.

3. Since both the Ashli--e Street ramp and th-

freewa;' it serves are -'Cn ele'.'ated struc-

tures in downtown Tampa, qeomietric modifi-

cation and recornstruction would h.'e b,;,n

e:treiiel'_ costly to implement.

In 1972, the University o.f Florida was awarded a

contract by the State of Florida Department of Transportation

to assist in the implementation of a mergini control system

on the Ashley Street ramp. A separate proposal was later

prepared and submitted to co.-er the evaluai.tion a.ctivitic`

for the perfjrmanr.-lc chliarjac teristics- of the ramp. The s 's-

tern proi.-.cosed for irntallatic, at the .- chiev .treet site was

thel :reer, band system developed ry. Paitleonr for the Federal

ii gh.:a.' '-dmiTiiristratric n anid ested at the Wioburri site. Con-

cr.actul1 re-S.pornsbilities f thel Un i.vers ity Df Flori-da

e'.,ol'.ed fr.:'ii c- nsul tinri or desinqr chiaracteristi1c.- o.f the

ssacei to. preparatl.r. of s .ft,..'are modifications to adapt the

s.steri to the Tamp.-a site, .,nd finr ll .' to the responsihil it

fcor impln ilrt inq, and testin thle sc!fc .are modifications to

g. t thie S--stiemi illtC. cperatlcin. TIhe natur-e :of thi' hardw.'are

ar d softw..i .rc chanc. s adopted to implement the Tampa system

are discussed in Chapter 3, and research itern cormponernts3 re-

lated to tie eval.uacion of this system are discussed in Chap-

ter 4.


I rntrojuct ion

The de.parture.-- from the earlier qreern bind ,

9s tcste.J in i lWolurn ulsed t.o install the Tamp.a sst'. em .e

three basic c*atcg'ories f90-95]:

1. Change. wihicih \.ere- required i o dec:.ribe

the Tairpa ;ite georioi-try tt LhC ;-rontrolt

program so.f t.a r inc:lujd n rL inforiam3ti.rn

ot sensor characteristics and 1.catiorns

and geometric 1 invitations on the peedi

of ramp vehicle caused ry a section ..-ith

a radius of cur-.'tur'e of 230 foot iii the

middle of the Ashle" Street r.amp.

2. Corrections w.lhich ;.ere mardc to kioI.rn an

anticipated problems in the ioburn .control

algorithms, including revision of ;hjat \;re

considered to be inipprorppriateo system roIi-

trol r.esponses to overtime o.ccupancies of

'.'hicles in the merge are acceleration


3, Operatin.3 improv.'emei-its rpiprcseinting

a.-i.itional .-aparilitlie o'.er those

pro-.'ided rin th_ l.aburn ireen Eand 3Ss-


Lach of the component elements or the Tampa Green

Banr System '.*ill be discussed inrdi'.'iually in this chapter,

including the ramp configuration, driver display design,

sensor crisid'ir.atiornI, zGmjputer requirements, and green hand

*.ojntrio algo-rithmn improv.',mentS.

The T.ampa Pamp

The sait selected for the installation of the Tampa

Green Band Control System was the Ahley Street entranc-'

ramp to 1-75 in Tampa, Floriia. This ramp provides a direct

entrance to the frec'.;ay from the downtown Tampa central busi-

ness district. The traffic volumes on the rtcility are mod-

erate throughout the diy, .,ith a substantial peak occurring

during the evening rush hour. The A'.'erace Daily Traffic (ADT)

values for the eastbound rO.adJ,.Jay of the free-.ay are 38,000

vehicles s per day ind 7,000 vehicles per dcay for the ramp.

Figures 12 and 13 show respectively data taken on January 24,

1974, representing f i.'e-minute '.'olumies at the entrance to

the Ashley Street ramp, and the sum of five-minute traffic

v'.clumes taken from sensors across the three free.'ay lanes


#..'" i-

'**_.--.--___' 1^.


C 0 0 C 4-, -,

4 ,., -

o i,--

'^ ?

U t- -,'

I s
- .J
j ,'-, ,








"..:- --t

*- ..r r.rl

", C.
^ ^

^-' L4

downstream of the entrance rampr acceleration nrgLte iar e. Note

the rc lat i el hijh ra-ip demand d'Jur irg the 4:00 F .M. to r. :00

P.M. peak period, w.hi.r. is typical of eek-da'y operation

of the ramp. Data for these fiqurer a wre pr-.duzei.J L.: the

computerii--eJ Jata collection system to be dliscuss.e. later.

Compari-son of obser.e rampi "cluri-s *durirrg peak F Jemianr peri-

od- with standardize.d expect atiors of rciramp s.steir capacity

[(:::] indicated ramp operation .t or abc..'e capacity, wnich

was confirmed by' the existence cf long vehicle queuies e:tnd-

ing upstream of the ranp entrance .JuLrin tiese times.

Sensor Installation at the Tamp.a :ite

Figure 1-1 shows the instrumentation s ytem i.sed to

monitor ramp and'a:' .-hii-le acti'.'it in the iTampa Green

Band System. The functions of the .'~rsou sensors used in

the on-line control program are as follo.s:

1. Double-loop sensors F7 through Fl .up-

ply information used to e tjbiish freeit.a

'.'chicle trajectories upstream of the i.merje

area. Sensor Fl is used to gerneratu 3-

minute free.,a' velocity an.'eraqles for use

in the .-.stem mode control logic.

2. Sinqle-loop sensor F.12 was originally in-

stalled to prno".'iJe a capability for ramp


r -

- J ^ ^i


D9 *


\ \-O\






L ..






queie detect ion in the stopped ,

of system op-'ratio.. This rfuction r .

later assicnr:-d t:o .sin. I--I cop s-enor

P3 t aii.-.' rapid :'steITm reiponr.-.

CapE.b L it t..-, the o*n ramp

t i:'n .

J. Sino-l. l.::ops deiiot:ed v.' F6E and RP .., -re -

pF.-'tivl.'y- se-rvo. as '.'ehicie chec-,;-in and

check-o.i.t- detectors at cte rainp traffic

signal locatio-n in the -toppedj model ,:f.

stem operation,

4. Sinqgl -lo, -Ic. ensr r F:? is used to c:.ntrl-.

tri, opera t:- n of a blank-out YIELD irgn

located at the end .f the ramp dispi. ..

This s gn iS illuJmrinated if a ramp '.e-

hicl:e crosses P3 without an accomrtpany in-,

creen band on the rain displa' oppeaite

this location.

5. Silgle-loop scnsor FO;., which h is the fir-t

loop cf the doublc-loop free:''ay velocity'

sensor PFOA 'E, iF uced to generate ccntinruouS

3-minute counts of free.'a" and ramp merging

-volumies downstream of the acceleration lane.

This informatici:n is used by the 5yStmi

miOde control i:.,ic in the Tampa soft-

wa r .

6. Acceleration lane presence _snsors ill

through r1i are used to inhibit green

band accti.'itc wrien occupiedd for a time

greater than a preset threshold -.alue.

The sernsors above are the onl-' dctcctors which are

actually used b; ctne on-line green band control program,

and thus are representative of a minimum set of sensors re-

quired for green band S,;tenm inplenientat ion at an:' ramp.

Additional s1 cnsrs were used at the Tampa ramp for

purpose., of data collection for system performance ev-alua-

ti.ns, as will be disEcusse i in Chapter 4. These additional

sensors are scanned by the control program at the rate used

for the on-line control sensors, and all field sensor acti.'i-

ty is recorded on magnetic tape and processed in a manner to

be discussed later. Additional sensors installed for evalu-

ation purposes included double-loop sensors at freeway sta-

tion F4 for rrmonitoring center and inside freeway lane veloci-

ties, single-loop sensors at Fl for center lane and inside

lane volume counts, double-loop sensors at station FO for

.:elocity profiles across the downstream freeway: lanes, and

single-loop sensors R11 through P1 e::tending throughout the

length of the ramp to monitor erg ing vehicle traiectcori.:.

The Woburn Green Band System ha.l double- locp .elolt it sensor

stations on the ramp left over from the Pacer syc' iriple--

mentation, but did not us.e these se-nsor=r in its cc.ntrol al g-

r i t lhin .

A.d it tional senscor-. '..ere installed or, two offr-ranmps

of I-75 in the study area and on th-e i'.'e' on cc tric :-outr

bound I-75 ramp e-tension on there Ashley- ramp, rLprCi.nented in

Figure 14 t b sensors El, EI', anJ E3, r.c:sp._ cti'.*el In addl-

tion, a series of eleven sonic detectors, modified electron-

icall,' to report pulse: widths proportional to vehicle e spac-

ing; in front of the sensors, were installed on the guard

rail of the acceleration lane and used to provide inf-,rr,-

tion on lateral :ehiile displacement- throughout the m.-rergin

area. Ulth th- c:.::ception of the sonic detectors, all senior-

were of the inductive loop type, positioned in grooves cut

in the roadw.ay. surfaces and covered with a hea.y-dui,' sealant

compound 195, p. A.151. Sensor dataw'-ere transmitted o'.er

hard-.:ired communication lines specifically installed for

this purpose.

Driver Display Configuration

The Uoburn Green Eand System driver displayy o.nsis-

ted of a series of 156 display clemicnts, each four feet in

length and containing t..o 75-w att flood lamps 'ired in paral-

lei, and co'.ered withr a -n :c-. of green acrl ic plastic.

Thiu:, te display in jperitiorn had the appearance of a series

or mo-i, ng points of green light representing the binds dis-

pli'ed by the syLstem. In addition, the other display ele-

ments as re'..iew':ed in Figure 4, Chapter 2, consisted of a

cr3.angeable message speed sign representiO-g one of three dif-

fEtrent speed levels for current band -speed, a standard tlire.-

face traffic signal positioned approximately one-third of

the -wa' aiong the display, a fri:ed message sign indicating

"Drive Beside Grcen Band," and sign at the end of the dis-

play element indicating a "Mlerge- ith Ciation" indication

..hich could 1:be illuminated .hen the s.'stem was unable to find

a green band for the- current ramp .veiicle.

Consideration was to the Wobirn display in

the design c.f the Tampa s'stcem elements and tne final con-

figuration implemented is shown in Figure.: 15 and 16. The

green band display consists of a series of 152 green fluo-

rescent light bulbs, each four feet long, housed in 76 alumi-

num channel sections placed end to end on the left-hand side

cf the guard rail, and covered with sheets of clear plastic.

The final configuration of this display element was deter-

mined by consideration of anticipated display visibility un-

der tryingng conditions of ambient illumination, and it was

felt that the appearance of continuous linear bands on the

display rather than the band composition scheme used in the

-, :
I I I f


\ \1 \

:-21 _a- '-

\ \\
L ',5

1, '2 ,_.

\ a

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs