AN ANALYSIS OF COMPUTER APPLICATION TO TRAFFRC CONTROL SYSTEMS. thesis for the 939m at! M. S. MCHIGAN STATE UNWERSITY iames Rifixardg‘ Yaylor $9,633 LIBRARY Michigan State University AN ANALYSIS OF COMPUTEh APPLICATION TO TRAFFIC CONTROL SYSTEMS By James RiChard Taylor AN ABSTRACT OF A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE School of Police Administration and Public Safety 1963 APPROVED L/ . “7W" oy E. Holladéijhairman) %mond T. Calvin ziember) ,A ,4- .. (i) ’7 . "' ““w: “\ j, I w /" / / Ca, 242: 4:7)»; (.- 2-«2 3 Earle B. RobertSVTMembér) ABSTRACT AN ANALYSIS OF COMPUTER APPLICATION TO TRAFFIC CONTROL SYSTEMS by James Richard Taylor Traffic control systems are employed to provide a rapid and efficient flow of traffic and to reduce traffic delays and congestion. Traffic signal control systems cur— rently employed in our urban areas lack the flexibility and responsiveness to perform these functions adequately. Com— puter controlled systems have the potential of providing the flexibility, responsiveness, accuracy and speed of re- action required to control and regulate urban traffic. This study covered the description and analysis of three computer controlled systems and a comparison of the following selected factors: (1) the detection devices used to provide the systems input data; (2) the methods of proc- essing the input data; (3) the control measures activated by the systems; (4) the time required to react to changing traffic conditions; and (5) the degree of human control required to Operate the system. The following are the methods used in the study: (1) a review of the literature relating to the subject of computer application to traffic control systems, and (2) personal interviews and correspondence with manufacturers flnO~ a '--A.U fi‘F‘ ‘1‘-“ A... n. \‘U‘uaé . wm h. ‘« \.v wk. ..v r. .4 r“ .r.. .. .1 "7“- .. .— N\~ ~N~ he James Richard Taylor of computer and detection equipment, traffic engineers of cities employing or planning to employ computer controlled systems and research centers and organizations involved in similar or related studies. The literature was surveyed to determine what has been accomplished in this field and what trends were devel- Oping. Interviews and personal correspondence with inter- ested agencies and individuals were to determine the types of systems currently in operation, the results obtained from limited pilot operations, and the modifications and improvements projected for future systems. This descriptive study of computer controlled systems included an analysis of their purpose, operational principles, vehicle detection units employed, processing of input/output data, and the control measures activated to solve or reduce traffic problems. The efficiency of the systems in reliev- ing traffic congestion by reducing vehicle delays and in- creasing road production were compared. It was concluded that the employment of computer controlled traffic control systems would provide a more efficient and responsive control system. Substantial de- creases in vehicle delay, increased road production and a resultant decrease in traffic congestion were the major benefits to be realized. AN ANALYSIS OF COMPUTER APPLICATION TO TRAFFIC CONTROL SYSTEMS By James Richard Taylor A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE School of Police Administration and Public Safety 1963 ACKNOWLEDGMENTS My sincere appreciation to the Provost Marshal Gen- eral, the Military Police Corps and the United States Army for making possible this Opportunity to complete my gradu- ate studies. I am indebted to Mr. Roy E. Hollady, 1r. Raymond T. Galvin and Mr. Earle B. Roberts who, through their time, effort and guidance, made possible the completion of this thesis. My sincere appreciation to the many individuals, active in the field of traffic engineering, who willingly gave of their time and knowledge to assist in conducting this study. My special thanks to my wife Mary who encouraged and assisted me throughout the writing of this thesis. I O INTRODUCTIC'IIO O I O O O O O O O O O O O The Problem . . . . . . . . . . . . . Statement of the problem. . . . . . Limitations of the study. . . . . . -‘ Importance of the study . . . . . . A heview of the Literature. . . . . . . Statement of Methods Used . . . . . . Definitions of Terms Used . . . . . . Crganization of the Remainder of the II. ANALYSIS 0? TRAFFIC CONTROL SYSTEMS . . Evolution of Traffic Control Systems. Traffic Systems Currently Employed. . Traffic officer control . . . . . . Airborne surveillance . . . . . . Television surveillance . . . . . Existing traffic signal systems . . Traffic control signal Operations System operation. . . . . . . . . Simultaneous system . . . . . . . Alternate system. . . . . . . . . Fixed time operation. . . . . . . Simple progressive system . . . . \J'IH U1 «J «a \3 LI) o :1: p “U H m m w H (I) >u F" (T F4 (D 'C "5 O 0'; *1 (D (I) .7) H <: (I) (D L A ’0 Cf (D ,3 III. COMPUTER CO"TPCLL3D ThAFFIC SYSTEMS . . . Computer real-time control of traffic Simulation. . . . . . . . . . . . . System operation. . . . . . . . . . Detection system. . . . . . . . . . Reaction time . . . . . . . . . . . Fail-safe Operation . . . . . . . . Evaluation of the system. . . . . . Automatic flow control of traffic . . Shock wave generation . . . . . . . System operation. . . . . . . . . . Detection system. . . . . . . . . . Reaction time . . . . . . . . . . . Fail-safe operation . . . . . . . . Evaluation of the system. . . . . . Centralized traffic signal control by general purpose computer. . . . . . System Operation. . . . . . . . . . Special programs. . . . . . . . . . Detection system. . . . . . . . . . System results. . . . . . . . . . . Evaluation of the system. . . . . . IV. COMPARISON OF SELECTEJ FACTORS. . . . . . "0 iv v: :1» O L.) \N \hl O\ -b I; -b UJ an xx \0 O\ m \0 (1) U1 |,_J 54 56 56 51 63 65 66 as {U V. Detection Devices . . . . . . . . . Pneumatic detector. . . . . . . . Electrical contact (permanent). . Electrical contact (temporary). . Photo-electric detector . . . . . hadar detector. . . . . . . . . . Magnetometer. . . . . . . . . . . Ultrasonic vehicle classification Ultrasonic deppler detector . . . Ultrasonic presence detector. . . Induction loop detector . . . . . Infrared detector . . . . . . . . Lane occupancy. . . . . . . . . . Methods of Processing Input Data. . Control Measures Activated. . . . . Reaction Time . . . . . . . . . . . Degree of Human Control Required. . Comparison of Selected Factors. . . SUMMARY AND CONCLUSIONS . . . . . . . Current Traffic Control Systems . . Traffic officer control . . . . . Traffic control signal Operations Simultaneous system . . . . . . Alternate system. . . . . . . . Fixed time Operation. . . . . . detector '2. / 84 84 85 85 85 86 86 90 93 97 101 103 103 103 104 104 104 105 Simple progressive system . . . . . . Flexible rogressive system . . . . . '(J Traffic actuated system . . . . . . . Computer Controlled Traffic Systems . . Computer real—time ccntrol of traffic Automatic flow control of traffic . . Centralized traffic signal control by general purpose computer. . . . . . Comparison of Selected Factors. . . . . DeteCtion d'EViCBS o o o o o o o o o 0 Methods of processing input/output data Control measures actuated . . . . . . Reaction time . . . . . . . . . . . . Degree of human control required. . . Conclusions . . . . . . . . . . . . . . Recommendations . . . . . . . . . . . . BIBLIOGRAPHY. . . . . . . . . . . . . . . . . . . F—J F“ H} 1'“ Q) ;L\ \J l \fi C) F“ (A) (D 107 107 108 H l—J H H [‘0 C) F4 +4 I\) 113 114 115 115 117 118 L1; F11 Schematic Diag am of Electronic Computer Func- tions . . . . . . . . . . . . . . . . . . . . . 37 Traffic Control Simulator . . . . . . . . . . . . 43 Three logical Functions of the Traffic Control Simulator . . . . . . . . . . . . . . . . . . . 4 Components of the Automatic Signal Control System 7 Comparison of Vehicle Detectors . . . . . . . . 5 Comparison of Selected Factors. . . . . . . . . 102 CHAPTER I INTRODUCTION Major cities throughout the world are faced with the ever increasing problem of devising traffic control systems which will provide a rapid and efficient flow of traffic in and through metropolitan areas. Initially, one thinks of a smooth, well coordinated and efficient traffic control system in terms of personal convenience and pleasure. Every automobile driver has at one time or another been caught in the throes of a traffic jam in which individual tempers and engine temperatures rise at an alarming rate. Irritated drivers who are late getting to work, or to the beach, or to other destinations imprecate other drivers, the police, the city, and anyone else who may be at fault for the traffic system or lack of a system. Thus irritated, a disregard for safe driving practices and attitudes develops and all too often the end result is an accident or series of accidents, which reflect their cost in property damage, crippling injuries and death. In addition to these well known and well publicized costs, there is the high economic toll exacted by traffic delays and congestion. These costs are not readily apparent to the average motorist and taxpayer. The estimated annual cost of traffic congestion in New York City is Sl,082,200,000. p .r; 3 ‘04. aka .aA Fl 1'“ 5‘ ‘4‘. L. v!" I“ “V: III ‘5 UV 4 Vin use.“ V‘fi’n «a». i A M... .nL a: a e a?“ t u S 5. Q .‘i‘ 1“ This cost includes excess taxi meter fares, excessive oil and gasoline costs, automotive repairs, more rapid depre- ciation of automobiles, increased insurance premiums due to more accidents, diversion of additional police personnel to traffic duties, and other costs created by the delay in services.1 This cost, when estimated on a national SOOpe, presents a vast waste of our economic resources. There is also a destructive effect On downtown businesses, prop- erty values, municipal tax structures, and on the economic health of communities in general. It has long been recog— nized that clogged traffic is not only a source of public inconvenience, but that it adds heavily to the cost of trans- portation and the price of goods and services to all con— sumers. Traffic congestion problems have been building up for years. Urban streets represent a tenth of the nation's road system, yet must accommodate 50% of the total traffic. The result is that almost all major arterials in metropol- itan areas have become chronic bottlenecks.2 The question that naturally comes to mind is what is being done about this situation? Traffic congestion has 1Report of the Citizens Traffic Safety Board, Inc., New York City, New York, 1953, p. 3. 2John W. Gibbons and Albert Proctor, "Economic Costs of Traffic Congestion," Urban Traffic Congestion, Highway Research Board Bulletin 86 (Washington, D. C. National Academy of Sciences-National Research CounciI, 1954), p. l. 1 '1 o‘vuv..‘ rss‘” ‘1 9H; d... 223“,“ UH. V1 \ reached intolerable proportions on most of our arterial routes. Construction of urban freeways will provide sub- stantial relief to cities, but arterial routes must remain the urban transportation system's bulwark for providing traffic distribution services. At present, urban arterials are over-taxed in their ability to provide this traffic service.3 To add to the problem, the number of automobile registrations grows at a much faster rate than our road improvement programs. The following illustrates the rapid and continuing growth of the number of registered motor vehicles in the United States.4 YEAR TOTAL VEHICLES REGISTERED 1895 4 1900 8,000 1910 468,500 1920 9,239,161 1930 26,531,999 1940 32,035,424 1950 49,195,212 1960 73,795,182 1962 78,660,000 These totals, if they were to remain at their pres- ent level, would be problem enough, but it is estimated that by 1975 there will be 109,500,000 registered vehicles and by 1980 the figure will have increased to 120,000,000 3Jack Berman and Arthur A. Carter, Jr., "Increasing Traffic Carrying Capability of Urban Arterial Streets, " Increasing Traffic Ca acit Of Arterial Streets, Highway Research Board Bulletin 2 l TWashington, D. 0.: National Academy of Sciences— National Research Council, 1960), p. 1. 4Automobile Facts and Figures (Washington, D.C.: Automobile Manufacturers Association, 1963). p. 18. A 17"!“ 'y-AL‘ "(I a: . a r I quit 7 LA .ld « v CC :6 Ca — u «.1» . 5 "VI AA; ,0 i .i .1 1'“ VF. nuv Ad .u. I: 4. tr“ NEH “I I‘H a» are A‘L vehicles.5 It is obvious that the problem is not going to solve itself. Providing new and better highways is only a partial solution. Part of the solution must depend upon maximum utilization of existing and projected facilities. Major arterial streets must be used to their maximum capac- ity and efficiency. Utilization of these arterial streets depends to a degree upon the control measures exercised. Traffic control measures currently in use can be divided into two primary or basic elements: human and me- chanical types of control. In theory, a traffic officer could be employed at every potential trouble spot and pos- sibly eliminate much of the congestion and delay. In prac- tice, the number and quality Of police personnel who can be assigned to traffic duties are limited. Mechanical con— trol, although of great assistance, is limited in its con— trol function. The primary limitation is that it is not responsive to immediate and changing conditions. Even the combination of human and mechanical controls, operating in a coordinated system, does not solve the problem. The basic limitations of each remain and, despite every effort, traffic congestion and traffic jams occur regularly. The field of automation proffers a possible solution to this problem. It has the potential to employ automated systems controlled by electronic computers which will pro- vide a control system less dependent on human control and 5Ibid., p. 46. more responsive to changing conditions. Speed, accuracy, responsiveness and flexibility would be the characteristics of such systems. I. THE PROBLEM Statement 22 Egg problem. The purpose of this thesis is to provide an analytical study of the use of computer controlled traffic control systems in relieving traffic congestion in urban traffic systems. This will include a descriptive study of proposed systems and a comparison of the following selected factors: (I) the detection de— vices employed to provide the system in—put data; (2) the methods of processing the in-put data; (3) the control meas- ures activated by the systems; (4) the time required to react to changing conditions; and (5) the degree of human control required to operate the systems. Limitations 9; the study. This descriptive study of computer controlled traffic systems and the comparison of selected, similar factors of each system will be directed at the capability of the systems to perform the proposed task of relieving traffic congestion by reducing vehicle delays. Certain limitations must be placed on the compari— son of these systems since they are either in the experi— mental Or pilot operation stages and as such there has been little sustained field observation to test their actual capabilities. The frame of reference for the comparison must be based on a logical evaluation of the selected fac- tors without regard to fixed standards of effectiveness. This is necessary since no operational standards have been devised or formulated against which the systems can be ef- fectively compared. Each system operates on different prin— ciples of traffic flow and control and as such it is doubt- ful whether any attempt will be made to formulate a general standard for the automated systems. At this stage in the systems deveIOpment there is insufficient test data to pre- scribe standards that would apply to a specific system. It is the consensus of traffic engineers involved in sys— tems development that operating standards will not be for- mulated until additional test data are available and then the standards will vary according to the location of the system, weather and road conditions, and the type traffic involved.6 With this limitation the comparison of selected factors employed in the systems will not provide sufficient data to draw conclusions as to which system is most effec- tive, but only to make a logical deduction as to which sys- tem or combination of systems will probably be most effec- tive. 6Information obtained from personal correspondence with J. H. Auer, Jr., Principal Research Engineer, General Railway Signal Company, Rochester, N.Y., May 7, 1963; L. Casciato, Chief Engineer, Traffic Research Corporation, Limited, Toronto, Canada, March 3, 1963; Edward Gervais, Project Manager, Michigan State Highway Department, Lansing, Michigan, March 28, 1963; and Adolf D. May, Jr., Director, Expressway Surveillance Project, State of Illinois, Depart- megt of Public Works and Buildings, Chicago, Illinois, May 7, 19 30 Importance 2f the study. Providing an efficient and rapid traffic flow system in urban areas is a major police problem. The proposed use of computer controlled systems in selected critical areas or for city-wide use and in conjunction and coordination with other highway sys- tems opens new horizons in the field of traffic control. An automated system would receive a continuous flow of in- formation relative to traffic density, lane occupancy, and vehicle speed and time in the system. It would analyze the raw data and react by making almost instantaneous de- cisions to increase or decrease speed limits, actuate traf— fic control devices and reroute traffic as required. A system of this type would provide the police with a tool which would permit better utilization of police personnel and provide a more efficient and responsive control system. As an added benefit, it would provide a source of accurate and current traffic data to assist in future planning for civic improvements. This study provides a source of information as to what is being done to solve the problem of relieving traf— fic congestion, what systems are available, what systems are proposed and what results can be anticipated from the use of these systems. II. REVIEW OF THE LITERATURE Reference material in the field of computer controlled traffic control systems is difficult to obtain. Since the field is relatively new, there are no published volumes or universally accepted systems which can be used as a re- search guide. Instead, as different cities and research organizations develop new systems or elements of systems, they publish articles in professional journals or present papers before interested organizations. The literature review for this study is concerned primarily with these articles and presentations. It is here that trends and advancements in the field of traffic control are found. The Highway Research Board, an affiliate of the National Academy of Sciences-National Research Council, Washington, D.C., publishes in their periodic bulletins and board proceedings the results of the studies and exper- iments of traffic engineers and research organizations. In an extensive search of the literature dealing with cur- rent traffic control systems it is apparent that in most cases theories concerning vehicular traffic are inadequate or restrictive. The problem is approached from the view— point of past experiences and future predictions with little emphasis on what is occurring at the instant time. This approach results in systems characterized by a lack of re- sponsiveness to immediate changes and fluctuations in the traffic picture. From this situation is created our traf- fic congestion, traffic jams and resulting slowdown or KO breakdown of the control system.7 According to Cass and Casciato, the trend has been toward the periodic appearance of specialized types of traf- fic signal equipment on the market which respond in one way or another to traffic movements. The growing tendency has been toward fairly heavy investments in the form of traffic signal modernization. While the equipment has added somewhat in improving traffic flow, its value is limited. It cannot detect congestion and often will systematically aggravate rather than improve critical traffic conditions. The solution lies in the direction of signal systems which respond quickly to variations in traffic flow and which do a great deal to reduce congestion and decrease vehicle delay.8 Gibbons noted that more recognition is gradually being given to the theory that intersections are the lim- iting factor in street capacity and the frequency of inter- 9 sections is the chief cause of delay in urban traffic. Lewis, in his work on traffic platooning, found that 7Philip A. Perchonok and Sheldon L. Levy, "Applica- tion of Digital Simulation Techniques to Freeway On-Ramp Traffic Operations," Highway Research Board Proceedings, Vol. 39 (Washington, D.C.: National Academy of Sciences- National Research Council Publications, 1960), p. 506. 8Sam Case and L. Casciato, "Toronto Successfully Pioneers Automated Control of Traffic Signals by a General Purpose Electronic Computer," Traffic Engineering and Con— trol, Vol. 4, No. 2 (London, England: Printerhall, Limited, June, l962), p. 86. 9Gibbons and Proctor, pp. cit., p. 7. 10 the timing of signals to produce a coordinated progressive system on an urban street has been the subject of consider~ able research but that there appears to be a dearth of lit- erature on the subject of maintenance of vehicle platooning on urban streets and none on the extent to which vehicles remain together on an open road after leaving a Signalized intersection. The study of platooning shows that a defi- nite pattern of vehicle performance prevails and a traffic signal placed some distance down the highway from the issu- ing intersection would cause less delay to the traffic stream if coordinated in some way with the first signal than if allowed to operate in a random manner. The use of electronic computers to coordinate and actuate traffic control signals apparently has the capability to perform this important traffic function.lO Mathewson found that there is a trend to utilize electronic computers as simulators to provide the solution to such traffic problems as investigating the effects of traffic control devices in advance of installation and pre- dicting the effects of proposed changes on the capacity of a facility. The concept of vehicle flow rate or, alter- nately, distribution of gaps finds general utility in ap- proaches of both analysis and simulation. The physical 10Brion J. Lewis, "Platoon Movement of Traffic from An Isolated Signalized Intersection," Vehicle Performance as Affected p1 Pavement Edge Lines and Traffic Signals, Highway Research Board Bulletin 178,1958, p. l. 11 model (simulator) encompasses both the structure (fixed facility) and the dynamics of the movement of intersecting streams of vehicles in terms of flow paths, queuing, pro— ceeding ahead and turning subject to delays caused by cross- traffic and pedestrians.11 Gerlough expressed the purpose of simulation on a computer as an experimental determina- tion of phenomena which are too complex to study analytic— ally and which may not be conveniently studied empirically in the real life situation.12 The body of this study will contain information ob- tained from, as yet, unpublished articles by D. L. Gerlough, F. A. Wagner, Jr., and the International Business Machines Corporation.13 Each of these studies indicates that the implementation of computer controlled traffic control sys- tems will relieve traffic congestion by reducing vehicle delays. 11J. H. Mathewson, D. L. Trautman and D. L. Gerlough, "Study of Traffic Flow by Simulation," Hi hwa Research Board Proceedings, Vol. 34, 1955. p. 525. 12D. L. Gerlough, "Simulation of Freeway Traffic by an Electronic Computer," Highwgy Research Board Proceed- in s, Vol. 35, 1956, p. 543. 13D. L. Gerlough, Notes 2n the Egg gf Eng Di ital Computer for Traffic Signal ControI_TCanoga Park, California: Thompson—Ramo Wooldridge, Inc., n.d.); D. L. Gerlough and F. A. Wagner, Jr., Simulation 3f Traffic En g Large Network 9; Signalized Intersections IPaper to Be presented at the 'Second International Symposium on Theory of Road Traffic Flow, Sponsored by the British Road Research Laboratory, London, England, June 25-27, 1963); International Business Machines Corporation, Computer Real Ting Control 2: Traffic (A Presentation to the Commissioner ofTraffic, CleveIand, Ohio, June 4, 1962). III. STATEMENT OF METHODS USED This is a descriptive study of computer controlled traffic control systems and a comparison of selected, sim- ilar factors of each system. The descriptive study of the systems includes a detailed analysis of their individual operation and functions. It includes the purpose, Operat- ing principles, detection devices, analysis of in-put data and the controls actuated to solve or reduce the problems presented. The frame of reference for the comparison of the selected factors will be a logical evaluation of their effectiveness as compared with each other. The following methods are used in this study: (1) a review of the literature relating to the subject of com- puter application to traffic control systems, and (2) per— sonal interviews and correspondence with the manufacturers of computer equipment, traffic engineers of cities employ- ing or planning to employ computer controlled traffic con- trol systems and research centers involved in similar or related studies. The purpose of surveying the literature was to learn what has been done in the field of automated traffic con- trol and to determine what trends, if any, were developing in this field. The amount of information obtained from the review of the literature of itself was not sufficient in quantity or depth to reach any conclusions as to definitive actions 13 being taken in the application of electronic computers to traffic control systems. For the most part, the published articles rarely described or discussed a complete or oper- ational system, but rather dealt extensively with improve- ments or innovations to parts of systems. Similarly, manu- facturers of computer equipment did not release or publish the results of their finished systems studies because of the competitive nature of the electronic computer industry and their reluctance to release this information until the system had been found acceptable and saleable. Therefore, great reliance was placed on the results of personal inter~ views and correspondence with recognized experts in this particular field. Mr. J. H. Auer, Jr., Principal Research Engineer for General Railway Signal Company, Rochester, New York, was contacted since his firm deals extensively with the development of traffic detection equipment. They installed the equipment used on the John Lodge Expressway, Detroit, Michigan, and the Congress Street Expressway, Chicago, Illinois. Each of these installation projects were in fact field testing laboratories for evaluating surveillance and detection equipment. Valuable information was made avail- able On detection equipment and also on a new parameter to the traffic control problem, namely lane occupancy. Mr. W. K. Campbell, International Business Machines Corporation, was interviewed relative to a new system, 14 Computer Real Time Control of Traffic, developed by his company and heretofore not available in published form. Mr. L. Casciato, Chief Engineer, Traffic Research Corporation, Limited, Toronto, Canada, was contacted because of the test program which he is conducting in Toronto in- volving centralized traffic signal control by a general purpose computer. This is one of the few locations where a pilot Operation has been in use for a period of time and where preliminary test data were available for evaluation. Mr. Robert S. Foote, Manager, Tunnels and Bridges Research Division, The Port of New York Authority, New York, New York, was contacted because of the continuing research performed by his staff in the field of computer control of traffic control systems. Mr. D. L. Gerlough, Head, Traffic Systems Section, Thompson-Ramo-Wooldridge, Inc., Canoga Park, California, was contacted because of the extensive work that he has performed in the field of Intersection Traffic Control, Computer Use for Traffic Control and in the use of com- puters for simulation of traffic in Signalized intersections. Mr. Edward Gervais, Traffic Research Engineer, Mich- igan State Highway Department, Lansing, Michigan, was inter- viewed because of his experience in this field, particularly in view of his work in connection with the John Lodge Ex— pressway, Detroit, Michigan, and his experimentation with a computer controlled, single intersection in Lansing, Mich- igan. Mr. Adolf D. May, Jr., Director, Expressway Surveil— lance Project, State of Illinois, Department of Public Works and Buildings, Chicago, Illinois, was interviewed because of his long eXperience in the field and his recent work with the Congress Street 3x~ressway, Chicago, Illinois. 'd IV. DEFINITIONS OF TERMS USED Analog computer. The analog computer is a type cal— culating machine that Operates with numbers represented by directly measurable quantities, such as voltage or resist— ance. There is a direct correspondence or analogy between the quantities undergoing calculation and certain electrical quantities existing at various points within the computer. Digital computer. The digital computer Operates with numbers expressed directly in a decimal, binary or other mathematical form. The items of data are represented by coded combinations of signals in which each signal may exist in one of two discrete conditions. Arterial. An arterial route is one designed to ac— commodate through traffic. 91213. A traffic signal cycle is the number of sec- onds required for one complete sequence of signal indica- tions. Offset. The offset in a traffic control signal sys— tem is the number of seconds that the green signal indica- tion appears at a given signal indicator at a specified reference time. :,.¢ o..l .A.. -v‘ (r ; s“. §.‘ 1‘. 41;, "1' (l' ‘4 I? .53 l6 £h§§g. A traffic signal phase is the total time interval allowed for any one direction of movement or for movement in any two Opposite directions. This includes the green interval during which movement is permitted and the clearance interval during which traffic is warned to stop. §plit, A traffic signal split phase is the split- ting of one or more Of the normal phases to accommodate heavy left turning movements. Normally similar indications for Opposite approaches are arranged to be contrary for short intervals to permit traffic left turns. It is accom- plished by either an advanced or a delayed green indication. Shock waves. Shock waves are one of the factors which interfere with efficient traffic movement. They are periodic reductions in the flow and speed Of traffic along the roadway. Simulator. A simulator is a means Of representing real phenomena with other characteristics. It is based on a system Of mathematical and logical relationships. Traffic control system. A traffic control system consists of two or more signal installations on a common street which are interconnected and controlled tO provide an efficient flow Of traffic. IV. ORGANIZATION OF THE REMAINDER OF THE THESIS In Chapter II, an analysis of traffic control systems l7 is developed. The discussion of the systems currently em- ployed include the following: (1) historical development, (2) traffic Officer control, and (3) traffic control signal operations, which include systems Operation, simultaneous Operation, alternate systems, fixed time Operation, simple progressive system, flexible progressive system and traffic actuated control. In Chapter III, a descriptive study and analysis Of computer controlled traffic systems is made. This includes a discussion Of the following: (1) computer real-time con- trol of traffic, (2) automatic flow control of traffic, and (3) centralized traffic signal control by a general purpose computer. In Chapter IV, a comparison of selected factors Of the computer controlled traffic systems is made. This in— cludes a discussion, analysis and comparison Of the follow- ing factors: (1) detection devices, (2) methods of process- ing input/output data, (3) control measures activated, (4) reaction time, and (5) degree Of human control required. Chapter V contains the conclusions and recommenda— tions drawn from the descriptive study and the comparison of the selected factors. Recommendations as to additional areas requiring further research and analysis are proposed. CHAPTER II ANALYSIS OF TRAFFIC CONTROL SYSTEMS I. EVOLUTION OF TRAFFIC CONTROL SYSTEMS The earliest traffic control system can probably be ascribed to the ancient Romans. They established one- way traffic flow on selected streets, special Off-street parking and specified hours for unlimited use Of the city streets. The first traffic regulations were posted in Baby- lon about 2000 B.C. Moving up through the years we find pavement lane markings making their appearance in Mexico City around 1600 A.D. The earliest efforts at police con- trol of traffic were in New York City in 1850. In 1904, William Phelps Eno initiated traffic survey methods in deal- ing with traffic control problems. In 1916, E. P. Goodrich employed Speed and Delay studies in working with the prob— lem Of traffic congestion.1 Modern electric traffic control signal devices, as Opposed to signs, markers and other devices, are of rather recent origin. They may be considered descendants Of the manually Operated semaphores which were first used in 1910. lHenry K. Evans, Traffic Engineering Handbook (sec- ond edition; New Haven: Institute of Traffic Engineers, 1950), pp. xi-xii. 19 Detroit set the pace in the use Of semaphores, eventually fitting them with colored lanterns for night time use. The first recorded electric traffic signal control was employed for a brief period in Cleveland, Ohio, in 1914. This sys- tem utilized four red signal lights on the near corners and four green lights on the far corners of the intersec- tion. The red and green lights operated alternately by means Of a crude timing device. This traffic signal con- trol system, which enjoyed only a brief operational period before being replaced by the system of police Officer con- trol, was invented by James Hoge in 1913. This crude sys- tem was the progenitor Of the three color light signal which first appeared in New York in 1918, and then throughout the rest of the country in the 1920's. The United States led the rest of the world in the development and use Of traffic control signal devices. They did not come into general use in Germany until 1926 and in England until 1928. Preceding the use Of traffic control signals was the short era Of traffic towers. These were first used in De- troit in 1916 and later in New York City. An experiment which was doomed to failure was attempted in Philadelphia in 1924, when large searchlights were mounted atop the city hall building to control street traffic over a wide area. 2 History and Function (Toronto, Canada: Department of Traf- fic Engineering, 1957), par. 12.00. J. T. Hewton, Traffic Signals: Ag Outline 3; their 2 20 The overwhelming confusion and congestion that resulted from this experiment ended the use of a single signal for large areas. It did, however, indicate the awareness of a need for the coordination of signals over large areas. Between 1922 and 1927, the traffic towers and traffic sig- nal devices fought for supremacy in the field Of traffic control. By 1930 the traffic tower had all but disappeared.5 The first interconnected traffic signal system was put into use in Salt Lake City, Utah, in 1917. In 1922, E. P. Goodrich publicly proposed a plan of timing signals for progressive traffic movement. It is reasonable to as- sume that the first Time-Space diagram for progressive traf- fic control systems was drawn up between that date and 1927. The first traffic actuated signals made their debut in New Haven and East Norwalk, Connecticut, and Baltimore, Mary- land, in 1928.4 Although various means of controlling and regulating traffic have been in effect for many centuries, the use of signal control devices is of recent origin. Unfortunately, the develOpment and sophistication of these devices has not kept pace with the rapid growth in the number Of auto- mobiles using the highway and street systems. II. TRAFFIC SYSTEMS CURRENTLY EMPLOYED Traffic Officer control. The methods of employing 3Ibid. 4Evans, pp. cit., p. xii. police Officers for traffic control purposes varies accord- ing to the type and location of the city, the surrounding area, the layout of the streets and the type vehicular traf- fic encountered. Other factors which affect the type of control exercised are the policies Of the department, the number of personnel available for traffic control duties and the quality of the personnel assigned. Although many variations exist, there are certain basic requirements that are common to all departments. These are to keep traffic flowing smoothly and as rapidly as is consistent with safety. Another similarity in basic approach, although not in the specific method Of employment, is the method of re- lieving traffic congestion or traffic jams. The method, simply stated, is to locate the source of the trouble and take corrective action. At this point we find a great vari— ation in the means employed to relieve congestion and re— duce traffic jams. One man is rarely able to handle a traffic congestion area. When several Officers are assigned to a small geographic area, as is common in our larger met- ropolitan areas, or when spotter vehicles and aircraft are used in conjunction with traffic control personnel, then these different agencies attempt to coordinate their activ- ities to keep traffic flowing through a larger area. This system Of control, regardless of the number of men and the amount of equipment employed, has certain limitations. Foremost among these is the limited distance that each 22 officer can see. In many instances his View of approach- ing traffic is limited to one city block and often less. His knowledge Of what is transpiring elsewhere in the sys- tem is severely restricted. Even if several O fficers are in the immediate visual range of each other, their ability to recognize an abnormal traffic situation is usually after the situation has developed. The time required to contact the other Officers, coordinate their actions and initiate corrective action limits their effectiveness to reducing existing situations rather than preventing situations from occurring. Again these several Officers working together take corrective action without knowing what effect their actions will have on other portions Of the traffic system. Airborne surveillance. The use Of airborne spotters in conjunction with traffic control personnel has relieved this problem to a limited degree. The Observer is able to see the traffic pattern in all directions, rec— Ognize emerging conditions, and in turn notify traffic per- sonnel of impending or develOping trouble areas. Here again there is a time delay in recognizing the situation, report— ing the situation and the subsequent reaction to the situ— ation. The time factor in reacting to the situation is a critical limitation of the efficiency Of the traffic Officer. 5Robert S. Foote, Development of an Automatic Traffic Flow Monitor and Control System (New York: The Port of New York Authority, 1961), pp. 4-5. 61bid., p. 4. 6 23 In resolving a traffic problem the traffic Officer will normally perform the following pattern of actions: 1. Detection of a situation by sight or hearing. 2. DeciSion as to whether corrective action is required. 5. Confirm the existence Of the problem. 4. .ggll for assistance or alert others (police or drivers) of the problem, if necessary. 5. Divert, stop or speed up traffic to elim- inate the problem. 6. Assistance to the elements causing the prob— lem. 7. Restore traffic to its normal pattern. These seven steps can be broken down into three main components-~the time needed to detect the problem, the time needed to respond and the time needed to restore normal Operations. This is a relatively slow process and normally takes place after the problem has developed. Thus we are reducing an actual situation rather than preventing a po— tential situation.7 Another limitation not usually considered is the reduction Of the effectiveness of police Officer control by the inability Of the vehicle operators to see the police Officer clearly and to understand the instructions or changes 7Ibid., pp. 6-9. ' 1 u it .~ 5 c. .u. a; . u N « N. ‘ A ‘V .TL WFV H ..A.~..N 1.. o . 1. .A 24 Of instructions given. Television surveillance. Of fairly recent ori- gin is the use Of television surveillance equipment in traf- fic control work. This has Often been erroneously described as an automated traffic system. Actually, the use Of tele- vision can be grouped with devices such as vehicles and aircraft. It is primarily a means Of extending the traffic Officers' vision. It is particularly helpful in areas where, during peak traffic periods, interference to the traffic flow would be reflected over large areas. Continuous sur— veillance would permit the police to take corrective action in a shorter period Of time. This has been aptly demonstrated on the John Lodge Expressway in Detroit, Michigan, where a series of TV cameras transmit a continuous series Of pic— tures Of traffic conditions to a master control room. The TV monitors are scanned by traffic personnel, who make de— cisions, based on the traffic flow, to raise or lower speed limits or to restrict traffic in certain lanes. They have the necessary centralized controls to effect changes in the control signals along the route. This system would best be described as a semi-automated system since the 8Joseph B. Bidwell, Control Problems 3: Automatic Hichwa s, Conference on Electronic Controls and Traffic Safety (New York: Safety Education Project, Teachers Col- lege, Columbia University, 1958), p. 4. 25 9 decisions are made by humans. Existing traffic signal systems. The flow Of traf— fic on city streets is generally limited by the capacity of intersections. Where traffic control requires the in- stallation Of traffic signals, the intersection capacity is inherently reduced by the ratio of the red signal time to the total traffic signal cycle. Further reductions in capacity result from the time required to get a traffic stream moving once it has been stopped by turning movements, interference and cross traffic. By virtue of these condi- tions there is thus a boundary to the effectiveness of a traffic signal system. Existing traffic signal systems, however, are in many cases a long way from this boundary, and thus can be subject to substantial improvement.10 Although crude techniques are available for handling traffic along a thoroughfare or within a network, still a large proportion of traffic signals throughout the United States are essentially controlling traffic on an individual- intersection basis. In the city of Los Angeles less than 20% of the traffic signals are connected in such a manner 9Information Obtained from personal correspondence and interviews with Mr. Arthur C. Gibson, Traffic Engineer, City Of Detroit, Department of Streets and Traffic, Detroit, Michigan. 10D. L. Gerlough, "Some Problems in Intersection Traffic Control," Theory g: Traffic Flow (Proceedings of Symposium on Theory of Traffic Flow held at General Motors Research Laboratories, December, 1959), p. 10. 26 that they may be maintained in an appropriate time relation- ship tO other nearby traffic signals.11 Grunow noted that the movement Of traffic on urban streets is dependent upon the degree of congestion or delay. Traffic signal delay is the major cause of congestion.12 Traffic control signal operations. Traffic control signal Operation may be classified according to the number of alternating traffic movements permitted, the methods Of time apportionment, and the time relationships between the start of the green intervals at adjacent inter- sections on a common street. In general, the simplest mode of operation which will handle the required traffic move- ments in safety will be the most efficient and should al- ways be used.13 System Operation. A system Operation results when two or more signal installatiOns on a common street are set up and timed in such a way that the appearance of the common street green indication at one intersection will be followed at a definite repetitive time interval by the Ibido , pp. lO-llo 12R. N. Grunow, "Vehicle Delay at Signalized Inter- sections as a Factor in Determining Urban Priorities," High- wa Needs Studies, Highway Research Board Bulletin NO. 194 (Wéshington, D.C.: National Academy of Sciences-National Research Council Publication, 1958), p. 42. 13J. T. Hewton, Traffic Signals (Toronto, Canada: Department of Traffic Engineers, 1957), par. 14.00. 27 appearance of the same indication at the next Signalized intersection. In general, all signals within 1200 feet of each other on the same street should be formed into a sys- tem, since otherwise delay and inconvenience will probably result for traffic using the street.14 The time lag between the appearance of the common street indications at adjacent intersections is known as the offset and the relationship between the Offset and the overall traffic cycle length determines the nature of the resulting system. System Op- eration may be established and maintained in two ways, namely: non-interconnected and interconnected systems. Simultaneous system. In this system the Off— set is zero and all signals show the same indication to the common street at the same time and therefore exactly the same phase times must be used at every installation forming a part of the system. In its simplest form one controller may be used to Operate a series Of signals.15 This is not recommended since any defect or breakdown will render the entire system inoperative. The disadvantages of this system are: l. Simultaneous stopping of all traffic along a common street prevents continuous movement and results in high speed between stops, but low overall speed in the system. l4;pgg., par. 14.40. 15Ibid., par. 14.41. 28 2. Cycle lengths and interval proportioning are usually determined by the requirements of one or two major intersections in the system. This creates ineffici— l6 ency at the other intersections. Alternate gystem. In this system the Offset is one-half the cycle length and therefore adjacent signals show Opposite indications to the common street at the same time. This system can provide for continuous through move— ment if all conditions are favorable.17 This system has the following disadvantages: 1. Equal phase intervals are required for both the main and side street traffic which is likely tO be in- efficient at most intersections. 2. It is not adaptable to streets having blocks Of unequal length. 3. It reduces the road capacity during periods of heavy traffic since the rear part of a vehicle platoon clearing one Signalized intersection will be stOpped by the next signal in the series when the signal changes.l8 Fixed time Operation. In this system the signal indications are switched on and off in a predetermined, repetitive sequence. The length of each interval and the 16 17 18 Evans, gp. oi ., p. 229. Hewton, 2p. ci ., par. 14.42. Evans, pp. ci ., pp. 229-230. \L- F0 time required for a complete cycle are always the same, except that in modern expansible equipment these lengths can be varied at pre-set times Of the day. This mode Of Operation has the advantage that the control unit is sturdy and relatively simple while the sequence of signal indica- tions can be quickly and simply changed to allow for addi- tional phases. The units may also be simply and positively interconnected for systems Operation without the need for any additional or special equipment.19 The principal disadvantage is the fact that short term volume variations cannot be readily allowed for and that undue delay result for main street traffic during low volume periods. Stopped time delay for any vehicle which is required to stop is much greater with fixed time control than with any other system.20 The motorist views this system only as it produces delays to his travel. The control cycle is changed with— out regard to the delay being produced or the amount Of traffic actually present. That is, the control system is not responsive to the needs of the traffic actually present.21 19Hewton, pp. cit., par. 14.21. 2OWayne N. Volk, "Effect of Types Of Control on In- tersection Delay," Highway Research Board Proceedings, Vol. 35 (Washington, D.C.: National Academy Of Sciences-National Research Council Publication, 1956), p. 523. 21D. L. Gerlough, "Some Problemsin Intersection Traf- fic Control," Theory pi Traffic Flow (Proceedings of Sym— posium on Theory Of Traffic Flow, held-at General Motors Research Laboratories, December, 1959), p. 11. 30 Simple progressive system. In this system the same cycle length must be used at all signalized intersec- tions, but the Offset bears no fixed relationship to the cycle length as it is adjusted so that a vehicle entering the system at one end will be able to pass through all in- tersections without being stopped. The Offset between ad- jacent intersections therefore depends on their distance apart and the permissible vehicle speed. The time allotted to the various phases at different intersections can vary, though naturally the shortest time given to the common street will determine the maximum number of vehicles which can make a non-stop through movement.22 A simple progressive system is far more efficient than those previously described but is normally only suit- able for use on one-way streets having reasonably constant traffic volume and vehicle speed.23 Flexible progressive system. In this system the same cycle length must be used throughout but the phase time can be varied in accordance with traffic requirements and the Offsets can be varied to allow for unequal signal spacing. The cycle, phase time and Offset can be independ- ently or collectively varied to suit different traffic con- ditions at different times of the day. Thus for a normal 22 Evans, cit., p. 230. Q. 23Hewton, pp. cit., par. 14.43. traffic artery, the cycle lengths, phase times and offsets can be set up and automatically selected for: 1. Morning peak hours with Offsets favoring inbound movements. 2. Normal day conditions with Offsets arranged to give the best possible two-way movement. 3. Evening peak hours with offsets favoring outbound movements.24 This type system must be of the interconnected type with a central master controller supervising all time re- lationships and coordinating all cycle function changes. modern expansible equipment can provide a maximum of three different cycles, with each having three different Offset positions, thus providing nine combinations Of cycle and Offset timing which may be automatically selected for use at various times of the day.25 This system has the following advantages: 1. Continuous movement of groups or platoons of vehicles is possible with reduced delay and at an aver— age speed as planned for the system. 2. Platoon movement tends to form gaps which may be utilized for safe vehicular or pedestrian crossing at non-Signalized intersections. 3. High and low speed travel is discouraged 241bid., par. 14.44. 51bid. 32 since both would be penalized by delay at the next signal- ized intersection.26 Each of the described systems has certain advantages and disadvantages. One disadvantage which is common to these systems, but which has not been discussed individually since the present day equipment is not designed to cope with the problem, is that the systems are not responsive to traffic conditions at any given moment. The cycle lengths are set for a particular period based on what occurred in tne past, not what is occurring at the instant time. If the traffic conditions affect the pre-planned patterns to the extent that a change in the system is required, then several courses of action are available to effect the change. The police Officer on duty may do one Of the following: 1. Switch the signals off completely and manu- ally direct traffic until the situation is back to normal and then restart the signals. 2. Switch the lights to flashing Operation and accept the additional delay and congestion until the traffic volume decreases enough to resume normal operations. 3. Manually Operate the signals and trust his own judgment to recognize immediate traffic requirements and arrange the cycles accordingly.27 Evans, pp. cit., pp. 230-232. 27 22 Hewton, cit., par. 16.00. As can be seen, these courses may reduce the prob- lem but the time required to do so may be extensive. The end result is traffic control signals which are designed to reduce the degree Of human control required for this function eventually are dependent on that human control when conditions differ from those which the system was pre— set to accommodate. Traffic actuated system. This system acts in accordance with traffic demand. The length of each interval and of the overall cycle are almost infinitely variable. The principal advantage of this system is that delay, es- pecially at minor intersections, is held to a minimum. The principal disadvantage is that each unit acts independ- ently without regard or knowledge as to wh‘t is transpiring elsewhere in the system. This individual reaction to traf- fic demand creates as many delays elsewhere in the system as it eliminates at its own location.28 Although this sys- tem is Operated on a traffic demand basis, fixed time sig- nals which are properly coordinated can result in less overall delay within the entire system than the individual 29 traffic actuated signals. This is not due to the system itself, but rather to the manner in which the agency Oper- ates it. 8Evans, pp. cit., pp. 257-268. 29 . p . LeWis, pp. Cit., p. 3. \JJ .Cx Operation of intersections without controls of any type would result, in theory, in the least vehicle delay. From this idea of minimum control the traffic actuated sys- tem could be more efficient then fixed time controls. In either system, the delay to vehicles on the side streets . . . 30 would be apprOXimateiy the same.“ Thus we have briefly examined the systems currently employed in controlling traffic. Each s‘stem has its own advantages and disadvantages. Each was designed with a particular function in mind. Each was and is installed at considerable expense. The remarkable and awesome growth of vehicular traffic in and through our metropolitan areas could not have been foreseen or predicted when these sys— tems were designed and installed. This is also true of the planning and construction Of arterial networks on which traffic patterns must be based and upon which control sys- tems must be superimposed. That the systems are not ade- quate to perform their intended function is apparent when one observes the congestion that metropolitan drivers have been forced to accept as an inevitable happenstance of prog— ress. Present systems are characterized by the absence Of feedback loops; that is to say, information is not fed back into the equipment relative to traffic density, lane 30 - . Volk, pp. C1 ., p. 533. \JJ \.“1 occupancy or delays and congestion. The equipment is pre— set to a certain cycle and remains there until re-set. It does not have the capability of analyzing current traffic demands and re—setting the signal cycles accordingly. The lack of responsiveness to immediate and changing conditions is a primary disadvantage of existing traffic signal systems and is directly caused by this open loop system. A closed loop or feedback capability would provide a traffic system with the capability of receiving information relative to traffic delay or congestion and vehicle density. With this information, a fully actuated system would provide maximum efficiency in maintaining optimum traffic flow. CHAPTER III COMPUTER CONTROLLED TRAFFIC SYSTEMS In discussing the use of computer controlled traffic systems it is necessary to understand the functions and limitations of the electronic computer. A computer is a highly complex electronic machine which is capable of performing many complex computations in a very short period of time. Basically it consists of four logical units: memory, input/output, arithmetic logic, and control. The memory unit is the location where infor- mation is stored. Vast amounts of external information can be accepted and stored for indefinite periods of time. This stored information can be continuously up-dated as changes occur. The input/output unit is the medium through which information is transmitted to and from the computer. The input portion performs functions very similar to the sensory organs of man. Input devices may take various forms, such as: punched cards, paper tape, magnetic tape, type- writers and other instruments.1 In traffic control systems detectors are the input devices. Output devices take the same general form as input devices with the exception that 1Franz L. Alt, Electronic Digital Computers (New York: Academic Press, Inc., 1958), p. 26. 37 in traffic control systems the output would be reflected in the control devices actuated. The arithmetic logic unit performs the function of the basic arithmetic operations and lOgical manipulations. The control unit coordinates and controls the other units and performs the integrated functions of the computer. CONTROL J \. \ i I 1‘. Y \\ i x' I . ____). MEMORY ——-———€>- OUTPUT 1 ARITHMETIC 1 INPUT ! L FIGURE 1 SCHEMATIC DIAGRAM OF ELECTRONIC COMPUTER FUNCTIONS The electronic computer can perform many operations in rapid time, but it has limitations. It will do only what it is programmed to do. Man must put the basic infor- mation into the computer either through his own direct ef— forts or through the use of sensor equipment. This infor- mation serves as the base from which the computations, 38 comparisons and decisions are made. In effect, the computer can do only what man can do, except that it is able to do these things faster and with less chance of error. Computer real-time control 2; traffic.2 This sys- tem envisions a traffic control system that is completely responsive to traffic demand rather than being based on a pre-set, fixed time operation or independent traffic actu- ated systems. To illustrate the need for this type of con- trol, picture an intersection where a long stream of vehic- ular traffic is suddenly stopped by a red traffic signal. There are no vehicles on the cross street taking advantage of the green signal. The light has turned red for no ap- parent reason, because there was no demand from the cross street for traffic to go across the intersection. The reason for this senseless delay is fairly obvious with present day methods of control. There is a set time control on the traffic signal. In other words, there are a certain number of seconds for the red, amber and green signals. Real-time traffic control is designed to eliminate this needless delay. If traffic is heavy on the artery, real-time control would not change the traffic signal un- less there was a demand from the cross street. There is a 2International Business Machines Corporation, Com— puter Real- Time Control of Traffic (A presentation made to the Commissioner of Traffic, Cleveland, Ohio, June 4, 1962 39 maximum cycle length established for the signal lights, since it would not be desirable to have a signal light show- ing one phase for extended or indefinite periods. Nor would it be desirable to have the phase changing every few sec- onds just for the sake of change. In effect, what the sys- tem will do is change the timing of any signal or series of signals in response to demands made on the system. De— pending on the number of vehicles at each intersection and depending upon the other lights in the system, the computer would calculate the time the light should be red or green at different times of the day. This may vary at every in- stant of the day.3 Simulation. The system under discussion is not an operational system in the sense that it is actually controlling traffic. It is a system devised through simu- lation of actual conditions by means of an electronic com- puter. The purpose of simulation on a computer is the ex- perimental determination of phenomena which are too complex to study analytically and which may not be conveniently 4 studied empirically in a real life situation. In the past, efforts to solve a traffic problem have 3Ibid., p. 2. 4Aaron Glickstein, Leon D. Findley and S. L. Levy, "Application of Computer Simulation Techniques to Interchange Design Problems," Freeway Design and Operation, Highway Research Bulletin No. 291 (Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1961), p. 146. 40 been by analysis and trial, analysis being a mathematical expression used to represent a traffic process and then manipulating it to determine the values to be used in chang- ing to better conditions. Trial involved a change in a real life traffic situation. Simulation is a combination of both methods. It permits an attack on the most compli- cated of processes which the analysis method did not. It does not affect traffic until a solution has been reached, unlike the trial method which may disrupt the traffic sys- tem completely before a final solution is reached.5 Employ- ing simulation techniques avoids the expense of installing new equipment or modifying existing equipment to test a proposed system. As can be seen, if the proposed system is not feasible, then, through simulation, a new system can be developed without additional changes in the exist— ing system. Simply stated, the simulation model duplicates in a digital computer a real life situation. The concept of movements and delays is the central feature of this sim- ulation.6 5Harry H. Goode, Carl H. Pollmar and Jesse B. Wright, "The Use of a Digital Computer to Model a Signalized Inter- section," Highway Research Board Proceedings, Vol. 35 (Wash- ington, D.C.: National Academy ofISciences-National Research Council Publication, 1956), p. 549. 6D. L. Gerlough and F. A. Wagner, Jr., Simulation 9; Traffic i§,g Large Network 9f Signalized Intersections, Paper to be presented at the Second International Symposium on Theory of Road Traffic Flow. (Sponsored by the British Road Research Laboratory, London, England, June 25-27, 1963), p. 2. 41 The simulator can be used to investigate different controls of traffic signals. In other words, depending upon demand which is the number of vehicles at each approach to a signal light and depending upon the other lights in the system, the computer will calculate the time the light should be red or green at different times of the day. There will be no fixed time for red, amber or green on a particu- lar light, rather it may vary at every instant of the day. Instead of going out to a location and hooking up detectors and vehicle counters and tying them into the computer, ve- hicle counts are taken along the route to be controlled. This information is used in the computer simulation so that the computer calculations correspond with the actual traf- fic situation in the selected area.7 In the typical urban area the settings of cycle length, split and offset are determined not from the num- ber of vehicles on the street at any given moment, but rather from the number of vehicles that were on the street on the days the traffic was counted. With the exception of traffic actuated signals, present day systems do not respond to the second by second demands of traffic at each light. In fact, they are indifferent to the number of ve- hicles waiting at each intersection at any given moment. Real time control is designed to increase utilization of 7International Business Machines Corporation, pp. Cito, pp. 2-40 42 the intersection and thus reduce vehicle delay.8 System operation. In employing the traffic control simulator typical distribution of vehicle arrivals and different traffic signal control methods are applied for a geographic area and fed into the simulator. The con- sequences of signal timings and various vehicle arrivals are given in terms of a series of signal control decisions and resulting traffic flow (see Figure 2).9 The computer real-time traffic control system is basically divided into three logical functions. The first function is the real traffic generator. This generator may be based on random vehicle arrivals or from pulsed flows. The second function is the simulator. It computes the intersection outputs and queues in each arm of all in- tersections. The traffic patterns at all intersections at the end of each generation or other fixed period of time are determined. The results of this simulation are analo— gous to a motion picture taken from above, looking down at the entire traffic area system. Each generation re- sembling each frame of the film is a photograph of the continuous traffic flow. Therefore, the simulator treats the traffic flow as a series of discrete photographs. The third function is the control function which may contain any two types of control mechanisms. The first control Ibid., p. 4. 91bid., p. 7. 43 Typical Distribution of Vehicle Arrivals Y Different TRAFFIC Traffic Signal Traffic Signal Control Decisions CONTROL fi. Control Methods and Resulting SINULATCR Traffic Flow FIGURE 2 TRAFFIC CONTROL SIKULATION 44 mechanism being a real time adaptive feedback method in which the decisions are based on real-time queues which are weighted in relation to the cumulative green light sig- nal time. The second control mechanism is the existing fixed time signal control system. The choice between the two methods of control is made by the "on" of "off" status of an external switch on the computer console. The computer makes the decision as to which control method will be em- ployed. The results of the control function are merely a time sequence of reds and greens for each signal which, in the light of the traffic demand patterns, will determine traffic behavior (see Figure 3).10 The following specifications must be determined and inserted into the computer as the source input: 1. Road status a. number of lanes b. road width 2. Per cent commercial type vehicles 3. Per cent left turns 4. Intersection lane capacity 5. Green arrows 6. Initial number of vehicles in intersection lanes. 7. Arrival distributions Ibid., pp. 8-9. Real Traffic Generator 45 Random Vehicle >: SIMULATOR Arrivals Control Choice Fixed Time Control l FIGURE 3 Time Sequence of Red and Green for Each Signal Real Time Control l THREE LOGICAL FUNCTIONS OF THE TRAFFIC CONTROL SIMULATOR 46 8. Signal status a. green, amber, red b. elapsed time since last change 9. Traffic flow rate The number of vehicles in each intersection lane, the signal change pattern and the total delay for the en- tire area system are provided as the simulation output.ll Detection system. The real-time computer sys- tem utilizes information from detectors that pick up vehicles leaving the intersections. The detection of vehicles is sensed by the computer through either a wire or radio com- munication link. Based on the vehicle count and speed in— formation, the digital computer makes decisions as to what signals should be changed. The instruction to change is (transmitted over the communication channel to the local intersection controller. Thus there is a closed loop con- trol of intersections with the following sequence: 1. Detection 2. Communication 3. Computer 4. Local controller The first step of the closed loop sequence is detec— tion. The sooner that vehicles are detected before the next intersection,the more time there is available to Ibid., pp. 9-11. 47 accomplish a change at the intersection. Vehicles should be detected as they leave the intersection as well as when they enter it. For intersections with special left turn lanes, the detector should be located much nearer the inter- section since the vehicles that are crossing the intersec- tion are going to be in motion while the vehicles in the left turn lane may be moving very slowly or stopped. In the case of vehicles going through the intersection, a non- presence type detector, e.g., radar and pressure road switch type detectors, would serve the purpose. For the turn lanes, a presence type detector, e.g., induction, sonic or infra- red type detector, would be required. The condition of traffic at the position in the street where the detector is to be employed will determine the type detector required.12 The detection devices that can be employed in this system are classified into two basic types: 1. Presence detectors which will detect ve- hicles at all speeds, including stOpped vehicles. 2. Non-presence detectors which will not de- tect vehicles travelling at speeds less than 2 mph. The detectors are further classified by their prin- ciple of operation. 1. The induction type detects only metal ob- jects which, on passing over a loop of wire embedded in Ibido ’ pp. 13-150 48 the surface of the pavement, cause the inductance of the loop to be changed and thence recorded. 2. The infra-red type detects any object which passes through the intersection line of the plane of infra- red transmission and the acceptable plane of the infra-red receiver. 3. The sonic type detects the delay time be- tween the time a speaker transmits a tone and the time a receiver hears the reflected tone. The shorter the elapsed time, the taller the detected vehicle. This is the only detector which can classify vehicles by size. 4. The radar type detects the doppler phase shift in the energy wave that is beamed at the oncoming traffic. A vehicle has to be travelling at a speed greater than 2 mph to cause a phase shift great enough to be de- tected. 5. The pressure road switch detects the pres- sure of an automobile's tire, causing a high rate spring to deflect and make an electrical contact. The road switch is embedded in the pavement in a long square-ended prism. The unit is encased in a rubber-like potting compound. 6. The earth magnetic type detects the change in the earth's magnetic field with a tightly wound coil installed below the pavement. A car passing over the coil distorts the magnetic field. 7. The photoelectric type detects the absence 49 of light reflected from a mirror positioned on the other side of the road. 8. The permanent treadle type detects the pres— sure of a tire on an 8"-wide contact plate. By deflecting a high rate Spring beneath the plate, an electrical contact is made. The permanent treadle type uses the same principle as the road switch type; however, the treadle type is much larger- Studies indicate that the radar type detector would perform the required tasks at the lowest cost, although it has the disadvantage of not detecting very slow moving or stopped vehicles.13 Reaction time. By definition, when a vehicle is under a detector and the detector causes a relay control to close, the closed contact condition is called the "on" state. When the vehicle has left the detection zone, the relay contact opens. This condition is called the "off" state. The computer determines the number of vehicles that have passed under a detector by rapidly scanning the detect- ors and noting the number of detector "on-off" changes that have occurred, with the contact sense capability of the computer. If the sampling rate is rapid enough, so that one complete scan of all the detectors will occur in less time than either an "on or off" state, then there is an 13Ibid., p. 15. 5O assurance that the computer will avoid counting two vehicles as one or one vehicle as two.14 In order to determine the time required to sense all of the detectors in a system, the "on and off" times that are typical for the type detectors employed must be determined. For induction type detectors, the "time on" length equals the length of the vehicle plus the effective detector length. For sonic, radar and infra-red types, the "time on" length equals the length of the vehicle plus the small effective detection pattern width. For the road switch type, the "time on" length equals the length of the tire contact patch plus the road switch effective width. The "time on" length is a function of the time on and the vehicle speed. The "time off" is a function of the time on length, the vehicle spacing and the vehicle speed.15 The minimum time interval for a change of state for the detector has been determined to be five milliseconds (5 ms.). The computer employed in this system has the capability of performing 25,000 detector point scans per second. This would average out to a system whereby one computer could effectively control 30 Signalized intersections, each hav- ing 4 detectors. This would include a range of accurate l6 detections up to Speeds of 68 mph. If speed limits were 51 higher than this figure, it would necessitate reducing the number of controlled intersections. Conversely, lower speeds would increase the number of intersections that could ac- curately be controlled by a single computer, since the lower speed would permit a greater time gap for the computer to scan the detectors. In addition to scanning the detectors, the computer also has the capability of processing a con- trol program at the same time. Thus the computer is scan- ning the detectors and recording the number, direction and speed of the vehicles within the system and at the same time it is processing, based on programmed data, the se- quencing of traffic signals to provide the Optimum traffic flow. Fail-safe operation. The computer real—time system is designed to overlay an existing traffic signal control system. This provides a minimum installation ex- pense. Present traffic signal devices operate on a three or four wire connection depending on whether the signals are controlled from a central master controller or by local intersection control. If the hook-up is a local intersec- tion control, then the four wire lead is interconnected to the computer or if it is a central master control, then the three wire lead is interconnected. With this inter- connection, the intersection controllers are Operated from the computer. If the computer fails, then the existing timing dials at each controller or at the central master 52 controller would still Operate the traffic signals, since the original system is not altered. This also permits the option of discontinuing computer control during low vehicle volume hours, so that the existing control system can oper- ate without requiring adjustments or secondary installations.17 The results of the simulation of the traffic control model for the Willow Freeway, Cleveland, Ohio, indicated that the existing traffic control system resulted in a uti- lization of from 40-60% of road capacity. The real time computer control system would provide a utilization of 90% 18 This increase of road capacity utilization of capacity. would result in a 50% reduction in vehicle delays. This would seem to justify its installation even if no other benefits were accrued. In addition to the reduction of vehicle delay, real—time control of traffic provides the following benefits: I. It delays the time when street and roadway expansion becomes necessary. This is realized from the increased utilization of existing road capacity. 2. It identifies specific areas which need improvement and then simulates different types of improve- ments to determine which is best. By simulating prOposed improvements, major changes, such as installation of one- way streets, additional control signals, installation of 18 17Ibid., pp. 24-26. Ibid., p. 23. 53 special traffic lanes or major construction can be pro- grammed to determine if they will meet the requirements. Future improvements and long range planning requirements can be predicted by the computer. 3. It assists in street layout planning by providing the means to simulate various traffic plans to determine both their traffic handling capability and also the effect it would have on adjoining areas. 4. City evacuation plans can be expedited by computer control. It is questionable if personnel assigned to direct evacuation traffic can or will be able to remain at their posts. The computer can be programmed to provide the best traffic flow pattern in accordance with a pre-de- termined evacuation plan. 5. It records changes of traffic patterns. The computer has the capability of providing a permanent "write out" record of information relative to traffic pat- terns, vehicle density and speed and traffic trends. This information can be utilized in future improvement planning. 6. It assists in routing emergency type ve- hicles. By sectioning off the city and preparing computer sub-routine programs, the computer has the capability of selecting the shortest and quickest route to the emergency route and controlling the traffic signals along the route to permit emergency vehicles to travel without interruption or delay. 54 7. It reroutes traffic due to accidents or other situations. By installing additional motorist aids, such as: directional arrows and electrical detour signs, at key intersections, the computer has the capability Of recognizing abnormal conditions within the traffic system and then actuating the directional or detour devices, to reroute traffic around the abnormal area. It would main- tain this detour until the situation was resolved and then 19 revert back to the normal traffic pattern. Evaluation gf the system. Since this system is a simulation model, it is difficult to assess its actual 20 The principle of the system is logical capabilities. and,within limits, feasible. As Gerlough noted, while field data were not available with which to make quantitative comparisons with Operational systems, the results attained were qualitatively consistent.21 That is to say that sim- ulation models, based on different traffic situations, vol- umes, speeds and locations, are consistent in the results that can potentially be obtained. It is anticipated that real-time control of traffic will appreciably reduce vehicle lgIbid., pp. 19-22. 20 See pages 38-39 for detailed discussion. 21D. L. Gerlough, "Simulation of Freeway Traffic by an Electronic Computer," Highway Research Board Proceed— ings, Vol. 35 (Washington, D.C.: National Academ of Sci- ences-National Research Council Publication, 1956 , p. 547. 55 delay and thus relieve traffic congestion. The estimated efficiency of a 50% reduction in traffic delay is question- able, since the results Of simulation have yet to be com- pared to the actual traffic process.22 Although the com- puter has the capability of recognizing major trouble areas and changes in traffic flow, it does not have the SOphis- tication, as yet, of recognizing and reacting to the minor causes of delay. Specifically, the instances where a ve- hicle slows down to observe happenings along the sidewalk or in store windows, or stOps to discharge a passenger in the middle of the street, or to slow down when undecided as to which direction to follow, are as yet beyond the cap- ability of the computer to anticipate, recognize or react to. For these reasons, it appears that a more probable reduction of vehicle delays would be in the range of 10% during peak periods and 20% during mid-day off-peak periods. This is not to depreciate the potential capabilities of the system but rather to state a practicable range of cap- abilities in view of the present day develOpment of computer systems. 22Perchonok and Levy, pp, cit., p. 518. 23D. L. Gerlough and F. A. Wagner, Jr., Simulation of Traffic iglg Large Network 2: Signalized Intersections 'TPaper to be presented at the Second International Sympos- ium on Theory of Road Traffic Flow, Sponsored by the Brit- ish Road Research Laboratory, London, England, June 25-27, 1963): p0 13- 23 56 Automatic flow contrOl gf’traffic.24 This system is based on the principle that, in considering the flow of traffic along an arterial, Optimum behavior results when the traffic moves in bunches or platoons from one signal to another, arriving at the second signal before it turns green. This type of Operation is desirable for two reasons: in the first place, it is well established that an inter- section will handle more traffic if the traffic arrives and passes through the intersection without stopping, thereby eliminating the dead time required to get a stream of traffic moving; and second, the drivers find such a system more desirable because the necessity of stopping is eliminated.25 The automatic flow control of traffic system Oper- ates on the principle of introducing gaps in the flow of traffic so as to maintain traffic speeds above the level where shock waves are generated. Shock wave generation. EXperiments have shown that the accordian-like action which is characteristic of congested traffic flow is caused by periodic shock waves.26 24Robert S. Foote, Development of an Automatic Traf- fic Flow Monitor and Control System (New York: The Port of New York Authority, 19617. 25D. L. Gerlough, Notes on the Use of the Digital Computer for Traffic Signal Control, n. d., pp. 1- 2. 26Robert S. Foote, Kenneth W. Crowley and Alan T. Gonseth, Development of Traffic Surveillance Systems at the Port of New York Authority (New York: The Port of New York_ Authority, 1962), p. 2. 57 The shock waves are periodic reductions in the flow and speed of traffic. They are caused by the inability of that section in the roadway which has the least capacity to handle all of the traffic flow. Shock waves start at the point of lower capacity and move back through the traffic stream. They cause the vehicles to slow down or stop as they ap- proach the low capacity points. Research into the problem of shock waves led to two main findings: that shock waves occur spontaneously under certain conditions; and that shock waves are a cause of congestion as well as an effect.27 By maintaining surveillance on the traffic stream to deter- mine when and where shock waves are to be generated, and taking action to prevent or absorb them, it is possible to minimize the loss in capacity due to the shock wave. The purpose of inserting gaps in the traffic stream and speeds higher than those usually found in congested traffic is to increase traffic production. The momentary interruption of traffic entering a roadway does not delay approaching motorists. Since production on the roadway ahead is at a higher level, all motorists' trips are ex- pedited by these gaps. This effect was demonstrated whereby congestion was reduced by one—third when gaps were inserted.28 ide, p0 30 Robert S. Foote, Kenneth W. Crowley and Alan T. Gonseth, Instrumentation for Improved Traffic Flow (New York: The Port Of'New York Authority, 1960), p. 15. A problem that arises in initiating a system of gap insertion in traffic flow is that of public acceptance. Since the Operation requires a periodic interruption of traffic entering the roadway for no apparent reason, it is necessary that the motorist be assured of a smooth and rapid trip so that one of the benefits of the system will be apparent to him. This is not the main benefit of the system. The main purpose is to gain greater traffic pro- ductivity through the critical roadway and thereby reduce the overall extent of congestion and delay.29 By maintain- ing higher production along the roadway an overall reduc- tion in delay to motorists on the roadway is experienced. In line with the need of public acceptance, the size and timing of the gaps must be carefully determined. With ex- cessive gaps, speeds will be maintained but overall produc— tion will decrease. V The insertion of gaps tends to create a vehicle pla- toon system of traffic flow. Studies have indicated that, if Signalized intersections are not spaced too far apart, platoons of vehicles tend to maintain a constant time-dis— tance relationship. This, however, is subject to change. When a free-flowing platoon encounters a red signal and enters a stationary state, vehicles are compressed, filling 29Robert S. Foote, Kenneth W. Crowley and Alan T. Gonseth, Development 3f Traffic Surveillance Systems g5 Egg Port pf New York Authority (New York: The Port of New York Authority, 1962}, pp. 5:61 the intersection approach. When the signal changes to green and the free-flowing state is re-entered, the controlled movement results in an increase in distance from the front to the back of the platoon, thus decreasing the number of vehicles which can pass through the intersection.30 TO understand how the behavior of platoons affect traffic flow, it is necessary to understand two processes which occur in vehicle traffic flow. One process is the decrease in speeds of successive vehicles in platoons. This decrease is caused by the fact that, since the vehicles are in platoons and therefore driving relatively close to each other, the fluctuations of speeds from one vehicle to the next will usually be such as to result in lower speeds. Fluctuations which would speed up successive ve- hicles are not likely nor desirable since the probability of having an increase in accidents is greater when vehicles follow too closely. The second process is that the traffic flow is highest when speeds are in a particular optimum range. When these two processes are combined, it is evident that, as a vehicle platoon becomes longer, due to increas- ing traffic demand, and the speeds of successive vehicles become slower, the traffic flow will tend to fall below 30D. L. Gerlough, Notes on the Use of the Digital Computer for Traffic Signal Control, n. d., pp. 1- 2. 60 its capacity. This can be avoided by introducing gaps to prevent these long platoons from developing.31 To provide a base from which to Operate the gap in- sertion system and to avoid excessive gap delays, the auto- matic flow control system is based on a one minute re—cy— cling basis, during which the system monitors both speed and volume and establishes automatically an input rate for traffic flow for one minute based on traffic conditions during the preceding minute. In Operation, the computer classifies speed conditions at the point of congestion and infers density indirectly by measuring speeds upstream from the point of congestion to determine when shock waves begin to propagate. The computer controls the density by count- ing the number of vehicles passing through the point of congestion in one minute, and setting the number of vehicles to enter the roadway in the following minute higher or lower than the point of congestion flow depending on whether more or less vehicle delay is required. If speeds at the point of congestion are slow, fewer vehicles are allowed to enter the roadway than passed through the congested area in the minute before, and density thereby is lowered. This con- tinues until the lower density results in raising speeds 32 at the point of congestion. The system is dependent upon 31Foote, Crowley and Gonseth, gp. cit., pp. 15-17. 32Ibid., pp. 5-9. 61 a continuous measure of the actual number of vehicles on a section of the road. When the vehicle density approaches a critical level, then the system must have the capability of exactly measuring the speed of each vehicle at points where shock waves might generate. As long as speeds are maintained, densities can be allowed to increase and a high flow rate will be realized. As soon as a trend of decreas— ing speeds becomes evident, gaps should be inserted until speeds are restored. As can be realized, this system re- quires numerous speed measuring devices as well as devices to measure actual density. System operation. In operation, the first step is the surveillance of the roadway. This can be accomplished by means of closed circuit television, police traffic con- trol personnel and vehicle detectors. The effectiveness of the surveillance system is determined by how rapidly it detects abnormal traffic situations. Although the use of closed circuit television and police traffic control personnel have a definite value in the surveillance system, it was determined that the least accurate part of the system was the human observer. Attempt- ing to exercise continuous control of traffic flow for long periods of time is extremely demanding on the individual. The observer must regularly evaluate information being re- ceived from several sources as well as consider the prob- able effects of his decisions on the traffic flow. In view 62 of the frequency and rapidity with which the traffic flow changes, it is not possible for a man to provide the fre— quent traffic alterations required to maintain an efficient traffic flow. Added to the inability of man to maintain the high state of efficiency and accuracy required, is the relatively high cost of providing sufficient personnel on 33 a daily basis. The least expensive and most efficient method of surveillance is by use of vehicle detectors which have the capability of measuring individual vehicle speeds at frequent intervals. This is based on the theory that a drop in speed will provide the earliest indication of actual or potential congestion and delay on the roadway. The speed information is then furnished to the traffic flow control computer which automatically determines when gaps should be inserted in the traffic stream. The surveillance system will provide an automatic alarm to indicate when and where speeds have dropped below a critical pre-set level. The flow computer scans the detector units and con- siders the number of vehicles which have passed the point of congestion in the preceding minute and then, by next considering the speeds of traffic approaching and at the point of congestion, establishes a maximum number of vehicles which may enter the roadway in the next minute. It should 33Robert S. Foote, Development of an Automatic Traf- fic Flow Monitor and Control System (New York: The Port of New York Authority, 19515, p. 23. be noted that although the system is programmed on a one minute re-cycling basis, any time suitable for local con- ditions may be substituted. After deciding each minute the number of vehicles which should enter the roadway in the next minute, the computer automatically adjusts a traf- fic Spacer. By means of ultrasonic or induction loop ve- hicle detectors placed near the intersection or potential congestion point, the traffic Spacer will count the number of vehicles entering the intersection or congestion point. When the amount which has entered in less than a minute is equal to the amount predetermined by the flow computer, the traffic Spacer will automatically change the signal light to red for the remainder of that minute. The mini- mum stop time is seven seconds, which insures that result- ing gaps in the traffic flow would be at least that long in duration. If the gap time was less than this, then there would be a tendency for vehicles to Speed up and close the gap resulting in an increased size of the platoon and de— creased overall road production.34 Detection system. Sensing traffic behavior is the critical point of any traffic control system. The automatic flow control system employs radar, ultrasonic and induction 100p detectors. The radar speed sensor de- tector transmits tone pulses to the monitoring equipment 34ibid., pp. 23—25. which measures the traffic flow, speed and density. The induction loop detector provides a similar level of detec— tion performance at an equal cost. The ultrasonic detector was determined to be the most accurate and the least expen- sive.35 Most traffic signal systems have included a minimum amount of sensing capabilitv. The sensing that has been used is generally volume counting. More adequate sensing of traffic behavior will improve traffic control systems. As an example, consider the situation when a vehicle platoon is moving in a progressive system in which the Signal is timed to turn green as the platoon reaches the intersection. From time to time, there may be a queue of vehicles at the intersection. When such a queue exists, the platoon is required to slow down or stop. The number of vehicles which can clear an intersection from a standing start is much less than the number which can move through without being required to stop. By using a combination of presence de- tectors to sense the presence of a queue and Speed or vol- ume detectors located well back from the intersection to sense the arrival of the approaching platoon, it will be possible to give an early green signal to clear the inter- section before the arrival of the platoon.36 65 The detection system has the capability of providing data to the computer so that a prediction of the platoon arrival time is made. The ultrasonic detectors are placed well back from the intersection. The computer examines passage inputs until a substantial gap occurs; then it looks for a group of vehicles following the gap. When such a gap occurs, the time at which the first car of the group passed the detector and the speed of traffic are used to predict the time at which the first car of the next platoon will arrive at the intersection.37 When Side street traffic is allowed to cross or enter the main street, minimum disruption to the main street traf- fic will occur if the arrival gaps in the main street traf— fic can be predicted and the signal timing changed accord- ingly. Gaps can be sensed by the detectors which are set back from the intersection. When the Speed of traffic as well as the time at which the gap front passes the detector are known, the arrival of the gap at the intersection can be predicted and the signals changed to permit Side street traffic to move in the intervening period. Reaction Elgg. The automatic flow control of traffic system is a relatively slow reacting system. Al- though the computer scans all detectors and processes a program in less than a second, it cycles itself on a one 371bid., p. s. 66 minute basis. Thus it normally uses a fifty—three second period to evaluate traffic conditions and trends in the cycle period. Once the computer recognizes a potential congestion point, it reacts instantaneously to the situa— tion. In this system the Speed of reaction is not as im— portant as the ability to determine accurately the number of vehicles in the system and the individual Speed of each vehicle.58 It has the capacity of scanning all detectors in the system in less time than the "on or off" state of any Single detector which is in the range of five milli- seconds. Fail-safe operation. The stoppage computer is the heart of the automatic detection system. In opera- tion, the computer does two jobs simultaneously. First, by remembering the number of pulses it has received in the past few minutes, the computer establishes a flow rate. Secondly, the computer measures the time that has elapsed since the preceding vehicle has passed the detector. When flow is heavy past the detector a large number of pulses will be received in a few minutes. Then the computer will recognize that the passage of a relatively small amount of time without a vehicle passing the detector might be cause for alarm. When traffic flow is light and few pulses are received the computer would not generate an alarm until 38Foote, Crowley and Gonseth, gp. cit., p. 14. 67 a much longer period passes with no vehicles passing the detector. Thus the system is fail-safe, because unless the system continues to operate and vehicles pass the de— tector an alarm is generated.39 The alarm can be to the central control station for evaluation and action or to the local intersection controller where it can automatic- ally revert tO the existing signal system. Limited experimentation with this system realized a 5% increase in traffic production plus several collateral improvements in traffic Operations when Shock waves were 40 This modest increase had a sharp effect on prevented. the degree and length of congestion and delay on the road- way, reducing it by 30%. The discussion of the system indicated that the de- tection, evaluation and control of abnormal situations was best handled through automation, since man lacked the abil- ity to perform the multiple functions required on a contin- uous basis without tiring and making errors in control and judgment. Further experimentation with this system indi- cated that at certain points in the system human control was necessary. These locations are the sections where traf— fic enters the system roadway. The traffic Signals which were activated by the traffic Spacer and flow computer were 39Foote, pp. cit., pp. 9-10. 4OFoote, Crowley and Gonseth, pp. cit., p. 4. 68 not effective in spacing the traffic flow. The length of time that traffic was required to stop to form the gaps usually ranged about ten seconds and since there was no apparent danger motorists were usually quite slow in stop— ping. The presence Of a police officer at these critical points was effective in producing the required stops and . - - a - a 41 prov1d1ng effective Spacing. Evaluation 2; the pystem. This system provides a means of increasing road traffic production and appreci- ably reducing traffic delay and congestion. But the loca- tions where this system would be beneficial are limited. The nature of the system which involves eliminating Shock waves, increasing production and reducing delay by means Of inserting gaps in the traffic flow limits its applica- tion to roadway approaches to tunnels, bridges and sections requiring a change Of phase to a lesser number Of lanes. For these traffic situations, the system is good, but to attempt to employ it in conjunction with a traffic area pattern would necessitate a very large outlay in terms of the number of detector units, stoppage computers, traffic Spacers and flow computers. TO install a master computer large enough and with sufficient SOphistication to monitor and direct the local flow computers would involve a cost 41Robert S. Foote, Kenneth W. Crowley and Alan T. Gonseth, Instrumentation for Improved Traffic Flow (New York: The Port of New York Authority, lg60), p. 23. 69 far out of proportion to the job to be performed. Centralized traffic Signal control py_g general ppr- 42 This system is an experimental pilot study pggg computer. conducted in the city of Toronto, Canada. In metropolitan Toronto, as in most large and growing urban areas, the ex- isting traffic Signal control system is inadequate. Local traffic engineers concluded that a modernized traffic sig- nal using the latest in available equipment would still not be able to cope with existing traffic situations. Avail- able equipment was not sufficiently flexible to meet chang- ing conditions.43 In view of this, serious consideration was given to the possibility Of evolving a new concept of traffic control which would have unlimited flexibility in the man- ner in which the traffic signals could be made to respond to traffic situations. From this study developed the pilot program Of controlling traffic with a general purpose com- puter. System pperation. The computer is the heart of the automatic system. A standard business type digital computer which uses a magnetic drum and magnetic tape units for storage of numbers and instructions is employed. Input/ 42Sam Cass and L. Casciato, Centralized Traffic Sig- nal Control py g General Puppose Computer—(Washington, D.C.: 1960 Proceedings, Institute of Traffic Engineers). 43Ibid., p. 203. output is normally accomplished by means of punched cards. However, during periods Of traffic control, information is passed directly to and from the control panel Of the electronic computer by means of a modified control board which receives the pulses from the detectors and transmits impulses to actuate the signal ccntrOls (see Figure 4).44 In taking over and maintaining control of the sig- nals, the computer is guided by a master control program. The master control program contains several sub-routines which enable the computer to control any or all of the sig- nals according to a specific Signal control plan. These may be fixed-time, semi-actuated, full-actuated, volume density, or progressive type control systems. For each intersection there is stored within the computer a table of constants or parameters which essen— tially describe the individual intersection. This table Specifies the number of lanes on each approach, the distance Of the detector from the cross walk, the minimum time which should be allowed for the green signal and the maximum time for a green Signal which can be tolerated in any direction. When the computer actuates an intersection according to a certain control plan, it specializes the control to suit the data recorded in the parameter table. It is the con— trol plan which enables the computer to keep track of the 44Ibid., p. 206. 71 Ememwm Aombzoo A mo zomHm4 :4 Is: >4 :& :Owumowaflmmmao afloagm> oaeom-mspas pcpoeopocmmz pmumm cappooamaouonm >4><><>< AhnmpanOev pompcoo HmOprOOHm Apcocmspmmv pompcoo Hwowupowam M OwomESocm COHmefiMwmmmHo kocmazooo nooam ocQH concom cowpfincoo oocmmcum ommmmmm moeomexm mmwe 88 center Of a traffic lane, the lane occupancy at any instant would be defined as the percentage of the line covered by vehicles. Vehicles packed bumper to bumper would produce a lane occupancy Of 100%. Vehicles spaced at three vehicle lengths apart would produce a lane Occupancy Of 25%. Although lane occupancy is quite similar in behavior to vehicle density, there are significant differences. Traf- fic consisting of a mixture Of passenger cars and commer- cial type vehicles can create a wide variance in the rela- tive degree Of lane occupancy and density. If one mile of roadway contained 30 automobiles, the density would be 30 vehicles per mile and would have a lane occupancy Of approximately 10%. If the same lane were occupied by 30 trucks, it would have a density Of 30 vehicles per mile, but the lane occupancy would jump to approximately 20 to 30%, depending on the length of the trucks. It is Obvious that, although the density rate is the same, the degree of lane occupancy or road congestion can differ. This dif— ferentiation is an important parameter for computer use. Since the computer would actuate traffic signals to increase road production based on the vehicle presence data trans- mitted by the vehicle detectors, the system would tend to over or under estimate the actual amount of road space avail- able to receive traffic. By experimentation it has been found that the degree of congestion in a traffic lane can be reliably ascertained 89 by Observing the degree of lane occupancy. The ultrasonic vehicle classification detector has the capability Of dis- tinguishing between vehicle height and length and thus pro- viding the computer with the necessary data to compute ac- curately the actual road production.l7 None of the computer controlled systems discussed in Chapter II employs the lane occupancy parameter as a function Of the system. The primary reason is that at the time Of the initial system development, the vehicle classi- fication detector had not been developed to an Operational stage.18 Inquiries directed to various cities throughout the country which are installing or planning to install computer controlled traffic systems indicated a wide variance in the type of detectors to be employed. Some reasons for the wide variance are cost Of installation and maintenance, degree Of accuracy and reliability required Of the system, location of the detectors and the type traffic which pre— dominates in the area. All agree that as the systems become Operational and sufficient data are Obtained to evaluate and modify the systems, that more sophistication and reli- ability in terms of detection ability, data processing and id., p. 3. D. L. Gerlough, Notes on the Use of a Digital Com- puter for Traffic Signal Control, n. d., p. 2. 90 control programs will be incorporated.19 II. METHODS OF PROCESSING INPUT DATA The vast amount of traffic data generated by the vehicle detectors create a problem of tremendous scope. The information must be obtained from the detectors, reduced to a form by which it can be handled by the computer, proc- essed and evaluated, compared with programmed data stored in the computer's memory unit, compared with data received in the previous time cycle, a decision made as to which traffic signal plan will provide the Optimum traffic flow, and the decision reduced to the appropriate pulse commands to instruct the local intersection controller. All of these actions must be performed in times ranging from fractional parts Of a second up to a few minutes. Obviously it is a task beyond the capability of man. A typical method of processing traffic data from the time it is received from the vehicle detectors until it is transmitted to local intersection traffic controllers in the formed signal change instructions would follow this procedure. The computer scans the vehicle detectors and accepts the information that has been recorded in the cycle 19Information Obtained from personal correspondence with traffic engineers of the following cities: Chicago, Illinois; Detroit, Michigan; Lansing, Michigan; Los Angeles, California; New York City, New York; Seattle, Washington; and Toronto, Canada. 91 time. The detector pulses are automatically encoded and converted to a pulse tone format which can be accepted, read and stored by the computer. The information received may be in the form of vehicle speed, density, presence, passage, lane occupancy, vehicle classification, time of day or combinations of these. The computer evaluates the input data in relation to programmed traffic parameters for each intersection and the traffic data received from previous cycles. Based on the programmed traffic control plans and existing conditions, the computer decides on the specific signal cycle, split and offset to be employed at each intersection. The com- puter then translates this decision into a series of pulsed tones to direct and actuate the individual traffic signals to follow the selected sequence and timing. The end result is the progression of traffic signal lights which direct and regulate the flow of traffic.20 During the normal sequence Of events, the computer will periodically analyze inputs from system detectors de- termining whether there should be any change from the sig— nal timing plan already in effect. When a change is to be made the computer will determine the new offsets and cycle updatings. The change may be made instantaneously or the computer may direct that a gradual transition to the new 2OAuer, pp. cit., pp. 4-7. 92 cycle take place. The choice rests with the computer and would be based on developing traffic trends and how rapidly the change must be implemented to achieve the desired re- sults.21 The computer real—time control Of traffic and cen- tralized traffic signal control by a general purpose com- puter systems operate on this complete centralized control method. All intersection traffic controllers are actuated by the computer based on a continuous flow of data from the detectors. The local controllers are completely depend— ent on the computer for actuation and any changes to the signal phase. The exception is when the computer fails or is disconnected. During these times the signals revert automatically to the existing signal system.22 The automatic flow control Of traffic system is not quite so centralized. In Operation the signal system is controlled by the stoppage computer and the traffic Spacer. These two devices actuate the signals under normal traffic flow conditions. The flow control computer monitors all input data and maintains a cumulative total Of the entire traffic picture.‘ In the event of abnormal conditions, the flow computer, which has monitored and evaluated the traf- fic situation, will direct the traffic spacer to actuate 21Gerlough, pp. cit., pp. 3-4. 22 cussion. See Chapter III, pages 52 and 73 for detailed dis— 93 the required traffic signals to relieve the abnormal con- dition.23 III. CONTROL MEASURES ACTIVATED The basic element of control activated by the com- puter system is the local intersection traffic control sig- nals. Additional signals can be installed and activated by the computer to provide a more flexible and responsive system. This would include traffic directional arrows to indicate turning lanes, speed information signs to advise motorists Of the Optimum speed to travel, detour signs and arrows tO reroute traffic automatically away from abnormal road conditions and route information signs to direct motor— ists to routes that will provide more rapid movement. The variations in this tvpe of signals are limited only by the imagination and ability of the traffic engineer. More important than the specific control device to be activated are the possible modes Of control that can be initiated. The following modes of control are advanced by comparison with current, conventional traffic signal control systems, but they are feasible for implementation by a digital computer under the present state of develop- ment.24 These modes are: 23See Chapter III, pages 64—65, for detailed discus- sion. 24Gerlough, pp. cit., p. 5. 94 Sophisticated cycle and offset selection. . Automatically adjusted progression. Platoon arrival prediction. Gap arrival prediction. Ul-b-WNH . Combination of the above modes. These control modes are programmed for the computer by means of a table look up procedure. Individual tables are provided for each combination of conditions. The en— tries in the tables are pre-determined settings of cycle, split and Offset for each parameter combination. Local signals may be operated by a regulated fixed-time controller or by means of a semi-actuated controller with a regulated background cycle.25 The advantages of employing these various pre-pro- grammed modes of control are that they are easily understood, provide traffic control programs for a variety of traffic situations, and the appropriate program is initiated auto— matically when fixed traffic input levels occur. The main disadvantage is that it requires a computer with a large memory storage unit to store the many pre-determined pro- grams associated with the ccntrol plans to be activated.26 It is possible for the computer to provide every timing command to all of the local intersection controllers. When this is done the computer will deliver a pulse to the 25Ib id., pp. 6-7. Ibid., pp. 3-4. 95 intersection controller at the end of each green, amber, walk, wait, or other interval. With this high degree of control, the information levels on the communication trans- mission lines will be very high. This will also require that a large percentage of the computer's work cycle be spent in issuing commands to intersections. In the oppo- site direction, it is possible to put a high degree of in- telligence in the local intersection controller and have the computer provide only supervisory information. This action increases the overall cost Of the system, increases the maintenance problem and in some cases decreases the system's flexibility.27 The computer real-time control of traffic and cen- tralized traffic signal control by a general purpose com- puter systems employ this highly centralized system of sig— nal control. However, the number of control plans avail- able are not as numerous or sophisticated as those mentioned. With the exception Of the periods when the computer is dis- connected or out of order, all signal changes are controlled directly by the computer. This provides for maximum system- wide coordination even though the type control plan utilized will vary from location to location and according to the 28 traffic demand generated. The automatic flow control id., p. 4. 28See Chapter III, pages 41 and 73, for detailed discussion. 96 of traffic system is less centralized and less flexible in the variety of control plans available. It operates on a selected re~cycling basis, during which gaps are inserted in the traffic stream by means of actuating the signal con— trollers. The length of the gap may vary depending on the instant traffic conditions, but the basic control plan re— mains the same.' Under normal conditions the stoppage com- puter and traffic spacer actuate the signal lights while the flow computer monitors and supervises the system.29 The computer real—time control of traffic and cen- tralized traffic signal control by a general purpose com- puter systems have a fail-safe Operation whereby if there is a failure in the communication link or a failure in the computer power supply the system will revert automatically 30 The automatic to a local intersection self-timing system. flow control of traffic system has a fail-safe operation that transmits a warning signal from the stoppage computer to the flow computer if variations in the established vehicle flow rate are detected. This causes the flow computer to scan the entire system, locate the trouble location and take appropriate action. If there is a power or computer failure the system reverts automatically to the local signal controllers.31 29See Chapter III, page 64, for detailed discussion. 30See Chapter III, pages 53 and 73, for detailed discussion. 31See Chapter III, page 67, for detailed discussion. IV. REACTICN TIME The reaction time for computer controlled systems should be instantaneous, although extreme care must be ex— ercised to prevent rapid fluctuations which have little effect on the total traffic flow picture from activating control plan changes. This is a major problem since it is Often difficult to determine valid changes in peak flow as against momentary surges, since both will be reflected in the detection devices and in the master computer.52 The computer real—time control of traffic has the capability of scanning all the detectors in the system in less time than either the "on or Off" state of any Of the detectors. The computer has a scan rate of 25,000 detector scans per second. Since the minimum time interval for a detector change of state is five milliseconds, the computer has the capability of making 125 detector scans per 5 ms., which is within the required time limit. During this scan interval, the computer is also processing a control program for the system. The output data or pulses, while transmitted instantaneously, do not actuate the signals in the same interval. The computer estimates the exact time the signals should change and instructs the controllers accordingly. 32Information Obtained from personal correspondence with Mr. K. E. Cottingham, District Traffic Engineer, Wash- ington State Highway Commission, Department of Highways, Seattle, Washington, May 28, 1963. 98 The delay in actuating the signals may vary from a few sec- onds to several minutes depending on the traffic demand.33 The automatic flow control of traffic system is a relatively slow reacting system. Although it has the capa- bility Of scanning all system detectors in less than a sec- ond, it Operates on a one minute re-cycling basis allowing a minimum time gap Of seven seconds and a fifty-three sec- ond period in which to evaluate the traffic situation and react accordingly. In this system, speed of reaction is not as important as is the requirement for accuracy in de- termining individual vehicle and the exact number of vehicles in the system.34 The centralized traffic signal control of traffic by a general purpose computer system has the capability of scanning 1,000 detector points a second. During this period, the computer reads all input data, updates exist- ing data and sequentially studies each intersection for possible revision to the control schedule in use. Although the computer has this capability, in Operation it has been limited to performing these steps every two seconds. The overall coordinating program has a slower reaction time, ranging from a few seconds to several minutes, depending 35 on immediate traffic conditions. 33See Chapter III, pages 51-53. for detailed discussion. 34See Chapter III, page 67, for detailed discussion. 35See Chapter III, pages 73-74, for detailed discussion. 99 V. DEGREE OF HUMAN CONTROL REQUIRED In operation the computer controlled systems are 36 to be completely a "hands—off" operation. This statement must be examined in terms Of personnel involved in the com- puter operation and those involved in traffic control func- tions. The purpose Of the computer controlled system is to increase road production and to reduce traffic delays and congestion. Whether this reduces the number of personnel used for traffic control functions is incidental. The system Operator will be primarily a supervisor to coordinate the activities of service and maintenance groups and the traffic control division of the police de— partment. The traffic engineer must develop traffic prO- grams for the computer. Although the computer analyzes and evaluates the traffic control plans in use, it is nec- essary for the traffic engineer to interpret the computer's analysis and prepare revised programs. Maintenance and repair of the equipment must be performed by skilled tech- nicians. With the reliability of transmissions from the computer to local intersection controllers, via owned or leased lines, there is little reason for human intervention except for the routine maintenance required and occasional 36Personal correspondence with Mr. L. Casciato, Chief Engineer, Traffic Research Corporation, Limited, Toronto, Canada, March 25, 1963. lOO verification of the computer programming results.37 Thus for normal computer operations and maintenance there is a need for human control. In develOping a computer controlled system, provision should be made for human intervention and control at any stage in the systems Operation. A system which could not be over-controlled by humans would result in an inflexible system. This would be especially true in those situations which might occur and which are not programmed into the systems routine and special programs. There should be pro- visions made for the control operator to intervene at his own discretion.38 This may be nothing more than the relin- quishing control of one intersection for special events or it may involve relinquishing control of an entire sec— tion of the system on a need basis. A major point to be considered in installing a com- puter controlled system is whether it will permit a reduc- tion in the police manpower requirements for traffic con- trol. It is not anticipated that the implementation of computer controlled traffic systems will result in any sub— stantial reduction in the number of police personnel assigned to traffic control duties. However, it will result in better 37Personel correspondence with Mr. K. E. Cottingham, pp. cit. 38Personal correspondence with Mr. S. S. Taylor, City Traffic Engineer, Los Angeles, California, June 4, 1963. lOl utilization of those assigned. This means that many of the police Officers assigned to point control of traffic or manual Operation of signal lights during peak traffic hours or emergency conditions would be released from these duties and assigned to more useful traffic functions.39 In the computer real-time control of traffic and centralized traffic signal control by a general purpose computer systems, human control is not required except for emergency actions. In the automatic flow control of traf- fic system police officers were required to assist in gain- ing compliance with the signals employed to form gaps in the traffic stream. Public education will in all probabil- ity correct this deficiency. VI. COMPARISON OF SELECTED FACTORS Figure 6 illustrates the similarities and differences, the strengths and weaknesses, and the degree of efficiency Of the three computer controlled traffic systems. 39Personal correspondence with Mr. L. Casciato, pp. cit. 102 SYSTEMS Computer Real Centralized Automatic Flow Time Control Traffic Signal Control of Of Traffic Control by a Traffic General Purpose Computer DETECTION Radar Radar Ultra-Sonic DEVICES J PROCESSING Centralized 1 Centralized Computer super- INPUT/OUTPUT control of control of Vises system all systems . all systems Operation. Signal control decentralized. CONTROL All signal A11 signal All signal MEASURES controllers. r controllers. controllers. ACTIVATED Variety of ; Variety of One basic control plansl control plans. control plan. FAIL-SAFE —system fails, System fails, System fails, OPERATION control re- control re- control re- verts to verts to verts to existing existing existing system. system. system. REACTION 25,000 scans/ 1,000 scans/ Scans detectors TIME second. second. every second. Continuous Performs Operates on a reaction. cycle every 1 minute re- 2 seconds. cycling basis. “HUMAN CONTROL None. None. Limited. REQUIRED EFFICIENCY Provides 90% Provides 11 to’ Provides 5% utilization Of road cap- acity. 50% reduction in delay. (Simulation) 25% reduction in delay. 20% increase in traffic volume. (Actual) (Actual) increase in road produc- tion. 30% reduction in delay. FIGURE 6 COMPARISON OF SELECTED FACTORS CHAPTER V SUMMARY AND CONCLUSIONS Traffic congestion is a relative thing which varies with the size of the city and the experience of the indi— vidual. It means more lost time in travelling from trip origin to destination, greater delays in movement of per- sons and goods, increased opportunity of being involved in a collision and increased consumer costs for services. I. CURRENT TRAFFIC CONTROL SYSTEM Traffic officer control. The employment Of police officers for traffic control functions varies from place to place. They ma’ be used for point control of traffic, or patrolling roadways, or monitoring traffic flow from aircraft or by means of television cameras. Although the means vary, the desired results of their activities are the same: to keep traffic flowing smoothly and as quickly as is consistent with safety, and to prevent traffic delays and congestion. Working alone or in groups, their efforts are characterized by an inability to see beyond their im— mediate location, difficulty in communicating with other traffic control elements, difficulty in coordinating their efforts with other areas within the system, limited ability 104 to predict when and where traffic congestion will occur, slow reaction time in recognizing and reducing abnormal traffic situations and a lack of knowledge of the effect that their efforts will have on the remainder of the traf- fic system. Their actions are essentially limited to cor- recting an existing abnormal situation rather than prevent— ing the occurrence of the situation. Traffic control signal operations. Traffic control signal Operations are classified according to the alternat- ing movements permitted, time apportionment and time rela— tionships of adjacent controlled intersections. The simplest method of moving traffic should always be used. Simultaneous pystem. This system employs a zero Offset with all signals on the common street showing the same signal indication at the same time. The simultane- ous stopping of all traffic on the common street prevents continuous movement and results in high speeds between stops, but low overall speed. Alternate pystem. This system employs an off- set which is one-half the cycle length. Adjacent signals show opposite indications to the common street at the same time. This system can provide continuous through movement under favorable conditions. Equal phase times are required for both main and side street traffic which is inefficient if traffic volume is higher on the main street. It is not adaptable to streets having blocks of unequal length and Q it reduces road capacity during peak hours of traffic. Fixed time gperation. In this system the sig- nal indications are governed by a pre-determined program. The length of each signal interval and the time required for a complete cycle are the same for a given program. It is possible to pre-set additional timing prOgrams for dif- ferent times of the day. This system is relatively simple and can be quickly changed to allow for additional phases. It can be interconnected with other traffic areas for a coordinated systems operation without adding special equip— ment. Fixed time control does not allow for short time variations in traffic volume and thus creates traffic de- lays for main street traffic during low volume periods. Simple progressive system. This system employs the same cycle length at all intersections, but the offset has no fixed relationship to the cycle length. A vehicle is able to enter the system and pass through every inter- section without being stopped. The offset is dependent on the distance between adjacent intersections and the per- missible vehicle speed. This system is normally suitable for use on one—way streets having a reasonably constant traffic volume and vehicle speeds. Flexible progressive system. This system em— ploys the same cycle length at all intersections, but the phase time can be varied in accordance with traffic require- ments. The offset can be varied to allow for unequal spacing. 136 The entire system can be varied to meet different traffic requirements. This system must be of the interconnected type with a central master controller supervising all time relationships and coordinating all cycle changes. The sys- tem provides for the continuous movement of vehicle platoons with reduced delay and at an average pre-planned speed. High and low speeds are discouraged since both would be penalized by delays at subsequent intersections. Traffic actuated system. This system is oper- ated by traffic demand. The length of each signal interval and the overall cycle length are infinitely variable. De- lays are held to a minimum at individual intersections, but since each signal unit acts independently of and without coordination with any other signal, it may cause delays and congestion at other locations in the system. None of these systems, with the exception of the traffic actuated system, is responsive to immediate and changing traffic needs. Their cycle lengths, splits, off— sets and timing are determined by what occurred in the past, not what is occurring at the instant time. They are inflex- ible in the sense that changes to the timing cycle require adjustment to the individual controllers or to the master controller. The time required to react to changes in the traffic situation reduces the system to the role of correc- tive rather than preventive action. 107 II. COKPUTER CONTROLLED TRAFFIC SYSTEJS Computer real-time control pf traffic. This system Operates on the principle of control based on traffic de— mand. It is completely responsive to present traffic con- ditions and not on pre—set, fixed time operations based on data collected in the past. The traffic artery carry- ing the high volume traffic will control the signal right of way until such time as traffic demand from a side street necessitates a change of signal. The change duration will be decided by the computer in relation to the traffic volume of the main and side streets. This system is not in operation. It is a traffic control system model developed through simulation of actual traffic conditions by means of an electronic computer. Simulation involves introducing data into a computer and testing various solutions to a problem. It involves the duplication of actual conditions in a computer. This sim— ulation model employs as its central feature the concept of traffic in relation to movement and delay. The control model employs a closed 100p control sys- tem in which no outside action or control is required. The system receives its input data from detection devices, trans- mits the data to the computer, computes the appropriate traffic program and then activates the signal controllers. The system is predictive in the sense that it predicts 108 traffic problems before they actually exist and takes the necessary control action to prevent their occurrence. The system employs a radar type detector to detect vehicle passage and speed. The computer has the capability of scanning all detectors in the system in less time than is required for a change of "on or off" state of any indi- vidual detector. With a capability of performing 25,000 detector scans per second, the computer can effectively control 30 Signalized intersections. At the same time it processes the necessary control programs to actuate and sequence the traffic signal controllers. The system has a fail-safe operation in that it reverts control automat— ically to the existing signal control system in the event of computer failure. Estimated results of the simulated systems effici- ency as projected from computer test results to an actual traffic process are a 90% utilization of road capacity as Opposed to 40-60% capacity with existing systems and a 50% reduction in vehicle delay. In addition, it provides a valuable tool for predicting and evaluating future traffic control plans. Automatic flgfl control pf traffic. This system op- erates on the principle of inserting time gaps in the traf- fic flow to form vehicle platoons which will result in main- taining vehicle speeds above the level at which shock waves are generated. The elimination of shock waves in traffic 109 flow results in increased road production and decreased delay to the motorist. Extensive reliance is placed on the accuracy of the detection system. Radar, induction 100p and ultrasonic type detectors were tested. The ultrasonic detection unit was determined to be the most accurate in Operation and the least expensive in overall cost. The vehicle detector transmits data pertaining to vehicle presence and speed to the traffic flow computer which determines when gaps should be inserted in the traffic flow and the duration Of the gap. The information is transmitted to the traffic spacer which adjusts the signal light sequence in accord- ance with the instructions. Once a flow rate is established, the detectors, stOppage computer and traffic spacer Operate independently, with the traffic flow computer serving to monitor the Operation. In the event the traffic flow rate changes, the flow computer automatically scans the system and issues new instructions to the traffic spacer until the flow returns to the normal flow rate. The system Operates on a one minute re-cycling basis in which the minimum gap time is seven seconds, thus pro- viding a fifty-three second period for the flow computer to analyze the traffic data and predict potential traffic problems. The traffic spacer is informed of what action to take and when to take it. The fail—safe Operation of the system is based on F4 p.» C the operating principle that unless vehicles are being re- corded by the vehicle detectors within reasonable limits of the established pulsed flow rate, the stoppage computer will alert the flow computer to evaluate the situation and take the necessary action to correct the situation. In the event of a computer failure, the control functions re- vert to the existing signal system. Limited test of this system realized a 5% increase in traffic production and a 50% reduction in traffic delay and congestion. A limited degree of human control was nec- essary to insure compliance with the short stop intervals which were used to create the time gaps in the traffic flow. This system is somewhat limited in its application to urban traffic control systems. It is best suited for roadway approaches to bridges and tunnels and areas where traffic lanes are suddenly reduced in number. The present state of the system development limits its use for normal traffic patterns and areas because of the large number of detection devices required and the requirement for a large computer with a high degree of SOphistication. Centralized traffic signal control py_g general pur- pose computer. This system employs a central master con- trol computer which has complete control over every signal within the system. The computer is programmed with a mas- ter control program which contains several sub—routines which may be applied to any or all of the signal controllers 111 within the system. The computer also contains a parameter table for each intersection which minutely describes the intersection. Based on traffic data transmitted by the vehicle detectors, the computer evaluates the information in respect to the parameters Of the particular intersection and selects a signal control plan to move the traffic through the intersection. This action is also coordinated with the other Signalized intersections. The system is responsive to all changes and demands of the instant traffic situation. The system employs pole mounted detector units which are adequate for the present system. The computer has the capability of performing 1,000 detector scans per second and simultaneously processing the control plan to handle the traffic requirements being generated. Special programs and traffic situations can be programmed into the computer for future use. The computer will automatically select the correct program in accord with traffic demand. In the event of a computer failure, the control func— tion automatically reverts to the existing signal system. The initial results Of the pilot Operation program indicated that traffic delays decreased by 25% during the morning peak hours and by 11% during the evening peak hours. This reduction of traffic delay resulted in an overall in- creased travel speed and a 20% increase in traffic produc- tion. 112 III. CCEPARISCN CF SELECTED FACTORS Detection devices. The analysis of the various types of vehicle detectors available and in Operation indicate that those with multi-sensing capabilities are more produc- tive and practical for use with computer controlled systems. Radar, magnetometer, ultrasonic vehicle classification, ultrasonic doppler, ultrasonic presence, induction 100p, and infrared type detectors possess these multi-sensing capabilities. Of these tests indicate that the radar, ul- trasonic and induction loop are the most practical and ac- curate for traffic control use. This selection may vary according to the particular requirements of the system. The ultrasonic vehicle classification detector, when coupled or coordinated with a vehicle Speed detector, offers the best accuracy and dependability for future systems. Methods 2: processing input/output data. The major difference in the input data processing cycle is the type and degree of control exercised by the master controller over the systems operation. The sequence of action is re— ceipt and transmission of traffic data to the computer, the analysis, comparison and computation of control programs by the computer, and the transmission of instructions to actuate the signal control devices. In the computer real-time control of traffic and centralized traffic signal control by a general purpose computer system the input data are transmitted directly 113 to the central computer where it is evaluated and compared with data stored from previous cycles. It is then compared against the parameter table for the particular intersection and with the traffic plan currently in effect. The computer determines if a change should be made and which traffic control plan should be implemented. .This is a highly cen- tralized processing system. In the automatic flow control of traffic system, the computer continuously monitors the systems input. It continuously analyzes the total situation. The routine analysis, comparisons and reactions are made by the local stoppage computer and traffic spacer. The master flow com- puter only directs the system when abnormal conditions arise. This system is less centralized and less flexible than the other systems. Control measures actuated. In all cases the systems actuate traffic signal control devices, which may be in the form of lights, arrows or sound alarms. The type device actuated is less important than the methods in which the devices operate. The computer real-time control of traffic and cen- tralized traffic signal control by a general purpose com- puter system employs a variety of traffic signal control plans which are placed in effect by the central computer. They are highly centralized systems with a maximum degree of responsiveness to traffic demand. 114 The automatic flow control system is less highly centralized, but it is equally responsive to changing con- ditions. It Operates from one basic plan with an infinite number of variations. Reaction time. The reaction time for the various systems differs according to the requirements of the system and the type computer employed. The computer real-time control of traffic system employs a computer which has the capability of performing 25,000 detector scans per second. The scan of the detectors must be performed in less time than the "on or off" state of the detector to avoid miscounting the number of vehicles. The minimu "on or off" time is five milliseconds, which gives the computer the capability of scanning I25 detectors in the minimum time. The computer reacts instantaneously to traffic demands, although it will phase its reactions to have the optimum effect on the traffic flow. The automatic flow control of traffic system oper- ates on a one minute re-cycling basis. It scans all de- tectors in the system in less than a second, but times its reactions to correspond to the cycle time limits. The cycle time can be varied to meet existing traffic requirements. The centralized traffic signal control by a general purpose computer system has the capability of performing 1,000 detector scans per second. The svstem is timed to scan all detectors every two seconds and prepare its traffic 115 programs accordingly. Although slower than the real-time system, it is completely responsive to traffic demands. Degree of human control required. The systems are designed to be a "hands Off" Operation. Each system re- quires human assistance and direction in preparing progr amxmed data, ins stallation and maintenance. The automatic flow control system requires police control to insure compliance with the momentary stop requirements for inserting the time gaps in the traffic flow. It is not antici pa ated that the use of computer sys- tems will reduce the number of personnel assigned to traf- fic control functions, but it will provide for better uti- lization of these personne Conclusions. The employment of computer controlled traffic control systems will provide a more efficient and responsive control system. Limited Operational use indi- cates that traffic delay can be reduced from 10 to 50?. The decrease in traffic delay and congestion will result in increased road production or number Of vehicles able to travel a road system in a given time. This increased road production will range up to 90% of maximum capacity. These improvements in t affic flow patterns of urban are as will provide a substantial decrease in the overall eco.cmic cost of traffic congestion. l. The computer real—time control of traffic system predicts the highest degree Of improvements for urban 116 traffic control. It cannot be stated that this will occur, until such time as the system is applied to actual traffic conditions. Simulation models attempt to test the system against all known variables, but the actions of inoividual drivers cannot be predicted with a sufficient degree of certainty to enable traffic engineers to program them ade- quately in a simulation model. 2. The automatic flow control of traffic sys— tem performs its required function in relation to specific types of traffic situations. The limitations Of the system preclude its use as a general traffic control system. 3. The centralized traffic signal control by a general purpose computer system has been in operation on a pilot model basis and has demonstrated its ability to improve the existing traffic control system. It is a flex- ible and responsive system which can be employed in any urban traffic situation. 4. The ultrasonic vehicle classification de— tector is the most advanced, accurate and versatile detector unit in operation. It will provide the detailed and dis— criminatory data required by a computer controlled system. 5. The use of computer controlled systems will not appreciably reduce the number of police Officers required for traffic control functions. It will permit better uti- lization of personnel by relieving the police of the neces- sity of standing point control of traffic during peak traf- fic hours. r.) FJ «-1 Recommendeticns. Since the use of computer controlled ? traffic control systems is relatively new, it is recommended that additional studies be made of the following areas: 1. The applicability of the General Motors Corporation's Traffic Pacer system to urban and freeway traffic. 2. The feasibility of linking urban arterial and freeway traffic systems under a coordinated computer controlled system. 3. The development of a completely automatic predictive traffic control system for the Congress Street 7xpressway, Chicago, Illinois. 4. Follow up study on the results of the com- puter controlled system currently undergoing tests in Tor— onto, Canada. BIBLICGhnIHY A. BCOKS alt, Franz L. Electronic Digital Computers. New York: Acadenzic Tress, Inc., 19- 58 Andrews, W.E. Traffic Engineering. Detroit: Steidinger Press, lQLE. Automatic Telephone and Electric Co., Ltd. "Electro-Matic" Vehicle Actuated Road Signals. London: Strowger House, 1957. Evans, Henry K. (ed. ) Traffic Engineering Handbook. Second edition. New Haven: _Tnstitute of Traffic Engineers, 1950. Hewton, J.T. Traffic Signals: An Outline of their History and Function. Toronto, Canrda: Department of Traffic Engineering, 1957 Ingraham, Joseph C. Modern Traffic Cpntrol. New York: Funk and Wagnalls, 195A. Johnson, Clarence L. Analgg Computer Techniques. New York: McGraw-Hill Book Company, Inc., 1956. Popkess, Athelstan. Traffic Control and Road Accident Prevention. London: Chapman and Hall, 1951. Seburn, Thomas J. and Bernard L. Marsh. Urban Transportation Administration. New Haven: Bureau of Highway Traffic, Yile University, 1959. Traffic Control Methods. Purdue University, Lafayette, Indiana: Public Safety Institute, 1952. Weston, Paul G. The Police Traffic Control Function. Springfield, miwnois?‘ Thomas Publishers, 1960. B. BCOKS: PARTS OF SERIES Gerlough, D.L. "Simulation of Freeway Traffic by an Elec- tronic Computer," Highway Research Board Proceedin s, ' Vol. 35, Washington, D. C. : National Academy of Sciences- National Research Council Publication, 1956, pp. 543-547. 119 Gerlough, D.L. "Traffic In-Puts for Simulation on a Digital Computer,” Highway Research Board Proceedings, Vol. 38, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1959, pp. L80-h92. Gerlough, D.L. "Traffic Study Techniques and Instrumentation -- 1980 Vintage," Proceedings 2: Institute 9§_Traffic Engineers, Vol. 32,1962, pp. 6h-66. Goode, Harry H., Carl H. Pollmar and Jesse B. Wright. "The Use of a Digital Computer to Medel a Signalized Inter- section," Highway Research Board Proceedings, Vol. 35, Washington, .C.: National Academy of Sciences-National Research Council Publication, 1956, pp. 548-557. Hall, Edward M. and Stephen George Jr. "Travel Time - Ag Effective Measure of Congestion and LeveI'of Service," Highway Research Board Proceedings, Vol. 38, Washington, .C.: National Academy of Sciences-National Research Council Publication, 1959, pp. 511-529. Mathewson, J.H., D.L. Trautman and D.L. Gerlough. "Study of Traffic Flow by Simulation," Highway Rgssarch Board Proceedings, Vol. 34, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1955, pp. 522-530. Meskowitz, Karl. "Waiting for a Gap in a Traffic Stream,". Highway Research Board Proceedings, Vol. 33, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 195h, pp. 385-395. Perchonok, Philip A. and Sheldon L. Levy. "Application of Digital Simulation Techniques to Freeway On-Ramp Traffic Operations," Highway Research Board Proceedings, Vol. 30, Washington, D.C.: National Academy or Sciences- National Research Council Publication, 1960, pp. 506-523. Volk, Wayne N. "Effects of Type of Control on Intersection Delay," Highway Research Board Proceedings, Vol. 35, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1956, pp. 523-533. C. PUBLICATIONS OF THE GOVERNMENT, LEARNED SOCIETIES AND OTHER ORGANIZATIONS Analytical Procedures for Metrgpolitas Area Traffic Studies. Michigan State Highway"Department, Planning and Traffic Division, 19h9. Automobile Facts and Figures, 1963 edition. Washington, D.C.: Automobile Manufacturers Association, 1963. Bartle, Richard M., Val Skoro and D.L. Gerlough. "Starting Delay and Time Spacing of Vehicles Entering Signalized Intersections," Effects sf Traffic Control 23 Street Capacity, Highway Research Board Bulletin No. 112, ashington, D.C.: National Academy of Sciences-National Research Council Publication, 1956, pp. 33-hl. Bauer, John and Peter Costello. Transit Modernization and Street Traffic Control. Chicago, Illinois: Public Administration Service, 1950. Bellis, W.R. "Capacity of Traffic Signals and Traffic Signal Timing," Increasing Traffic Capacisy 9f Arterial Streets, Highway ResearCh Board Bulletin No. 271, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1960, pp. A5-67. Belsar, C.H. Eno Foundation for Highwaerraffic Control. Saugatuck, Conn., l9hB: Berman, Jack and Arthur A. Carter Jr. "Increasing Traffic Carrying Capability of Urban Arterial Streets," Increasisg Traffic Capacity 9§_Arterial Streets, Highway Research Board Bulletin No. 271, Washington, DiC.: ‘ National Academy of Sciences-National Research Council Publication, 1960, pp. 1-13. Bidwell, Joseph B. "Electronic Highways," Driving Simulators and Application g£_Electronics 2g Highways, Highway ResearcH'Board Bulletin No. 261, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1960, pp. 50-60. Bidwell, Joseph B. Control Problems 9§_Automatic Highways. Conference on Electronic Controls and Traffic Safety. New York: Safety Education Projects, Teachers College, Columbia University, 1958. Campbell, E. Wilson, Louis E. Keefer and Ross W. Adams. "A Method for Predicting Speeds Through Signalized Street Sections," Trip Generation and Urban Freeway Planning, Highway Research Board Bulletin No. 230, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1959, pp. 112-125. Edie, Leslie C. and Robert S. Foote. Effect 2£_Shock Waves on Tunnel Traffic Flow. New York: The Port of New York fithority, 1960. Edie, Leslie C. and Robert S. Foote. Experiments ig_Sin 1e Lane Flow is Tunnels. New York: The Port of New York Authority, 1959. Eno, William Phelps. Simplification 9: Highway Traffig. Washington, D.C.: The Eno Foundation for Highway Traffic Regulation, Inc., 1929. Eno, William Phelps. The Story 9: Highway Traffic Control. Saugatuck, Conn.: The no Foundation for Highway Traffic Control, 1939. Eno, William Phelps. Uniformity is Highway Traffic Control. Saugatuck, Conn.: The Eno Foundation for Highway—Traffic Control, 19h1. Foote, Robert 8. Development gfggg_Automatic Traffic Flow Monitor and Control System. New Ybrk: The Port of New IBrkAuthority, 1961. Foote, Robert 8., Kenneth W. Crowley and Alan T. Gonseth. Dgyelspment g£_Traffic Surveillancs Systems at_ghg Port 9; NewYErk Authority. New York: The Port of New York Authority, 1962. Foote, Robert 3., Kenneth W. Crowley and Alan T. Gonseth. Instrumentation for Improved Traffic Flow. New York: The Port of New York Authority, 1960. Freewsy Operations. Institute of Traffic Engineers, Washington, D.C., n.d. General Electric Company. Traffic Analysis Using the GE 225 Electronic Compyter. Bethesda, Maryland: General Electric Computer Department, Washington Information Processing Center, 1963. Gerlough, D.L. "Some Problems in Intersection Traffic Control," Theory 9£_Traffic Flow. Proceedings of Sym- posium on Theory of Tralec Flow held at General Motors Research Laboratories, December 1959. Amsterdam, Holland: Elsevier Publishing Company, 1961, pp. 10-27. Gerlough, D.L. Use pf Poisson Distribution in Highway Traffic. Saugatuck, Conn.:—The Eno Foundation for Highway Traffic Control, 1955. Gibbons, John W. and Albert Proctor. "Economic Costs of Traffic Congestion," Urban Traffic Congestion, Highway Research Board Bulletin No. B6, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 19Sh, pp. 1-25. Glickstein, Aaron, Leon D. Findley and S L. Levy. "Appli- cation of Computer Simulation Techniques to Interchange Design Problems," Freewsy Desigp and Operation, Highway Research Board Bulletin No. 291, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1961, pp. 139-162. Grunow, R.N. "Vehicle Delay at Signalized Intersections as a Factor in Determining Urban Priorities," Highway Needs Studies, Highway Research Board Bulletin No. 19A. Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1958, pp. h2—A8 Hague Congress. Traffic Congestion is the City Center. The Hague, 1957. Herman, Robert (ed. ). Symposium on the Theory of Traffic Flow, Warren, Michigan,l 959. Amsterdam, Holland: Elsiever Publishing Company, 1961. Joint Committee 9p Uniform Traffic Control Devices for Canada. Ottawa, Canada: Canadian Good Roads Aesociation 1960. Joint Safety Research Group. Pennsylvania Turnpike. Pittsburgh, Pennsylvania, 195A. Kock, Winston E. Auto- Radar Systems and Traffic Saf e_y. Conference on Electronic Controls and Traffic Safety. New York. Safety Education Project, Teachers College Columbia University, 195 . Lewis, Brion J. "Platoon Movement of Traffic From an Isolated Signalized Intersection," Vehicle Performance as Affected hy Pavement Edgs Lines and Traffic Sighals Highway Research Board Bulletin No. 178, Washington, D. C. : National Academy of Sciences-National Research Council Publication, 1958, pp. 1 11 Maier, Eugene. Application of Electronic Controls to Urban Transit. Conference on Electronic Controls and Traffic Safety. New York: Safety Education Project, Teachers College, Columbia University, 1958. Manual of Traffic Engineering Studies. New York: The Accident'Prevention Department of the Association of Casualty and Surety Companies, 1953. Manual of Uniform Traffic Control Devices for Streets and Highways. Springfield, Illinois, 1958. Pearson, Robert H. and Michael G. Ferreri. "Operational Study - Schuykill Expressway," Freeway Design and Operation, Highway Research Board Bulletin No. 291, Washington, D.C.: National Academy of Sciences-National Research Council Publication, 1961, pp. 104-12h. Platt, Fletcher N. Operations Analysis of Traffic Safety. Dearborn, Michigan: The b‘ord Notor Company, 1959. Platt, Fletcher N. Traffic Safety Research. A Unique Piethod 2f Measuring Road, Traffic, Vehicle and Driver Charac- teristics. Dearborn, Piichigan: The Ford Motor Company, 1962. Report of the Citizens Traffic Safepy Board, Inc. New York City, New York, 1953. The Traffic Institute of Northwestern University. Application 2; Electronic Compgters 29 Traffic Engineering. Publi- cation No. 2&02, 1959. The Traffic Institute of Northwestern University. Estimating Future Traffic. Publication No. 2566, 1959. The Traffic Institute of Northwestern University. Selecting and Testing Alternate Route Locations. Publication No. 307 9 19590 The Traffic Institute of Northwestern University. Street and Highway Capacity Studies. Publication No. 200A, 1959. United States Department of Commerce, Bureau of Public Roads. Highway Capacity Manual. Washington, D.C.: United States Government Printing Office, 1950. United States Federal Bureau of Investigation. Traffic Control and Accident Investigation. Chapel Hill, North Carolina: Institute of Government, University of North Carolina, 1947. VE- DET, RCA Vehicle Detectors. Industrial Computer Systems, Radio Corporation of America, Camden, .ew Jersey, 1963. Wiley, T. T Urban Traffic Safety Related to Electronics. Conference on Electronic Controls and Traffic—Safety. New York: Safety Education Project, Teachers College, Columbia University, 1958. Zworkin, V. K. and L.E. Flory. An Electronic Sy§tem to Control Motor Vehicles on the Highway. Conference on Electronic Control and l'raffic Safety. New York: Safety Education Project, Teachers College, Columbia University, 1958. ".5 D. PERIODICALS Cass, Sam and L. Casciato. "Toronto Successfully Pioneers Automated Control of Traffic Signals by General Purpose Electronic Computer," Traffic Engineering and Control, Vol. a, No. 2, June 1902. Freer, J.A. "Los Angeles Installs America's First Computer System for Traffic Control," Traffic Engineering and Control, Vol. t, No. 3, July 1962, pp. 1h3-152. Gerlough, D.L. "Automatic Com uters for Traffic Control," Municipal Signal Engineer, Vol. 17, No. A, August 1952, pp 0 LPO'AZ o Hillier, J.A. "Instrumentation for Traffic Studies," Traffic Engineering and Control, Vol. 2, No. 1, May 1901, pp. 49—51. Hopkins, B.C. "Vehicle Detection for Traffic Analysis and Control," Traffic Engineering, Vol. 31, No. 10, July 1901, pp. 1h-10. Kreml, Franklin M. "Traffic and Transportation Looks Ahead," Traffic Digest and Review, Vol. 8, March 1900. Mertz, William L. "The Use of Electronic Computers," Traffic Engineering, Vol. 30, No. 8, May 1900. Mertz, William L. "Traffic Assignment to Street and Freeway Systems," Traffic Engineering, Vol. 30, No. 10, July 1900. Newell, G.F. "The Flow of Highway Traffic Through a Sequence of Synchronized Traffic Signals," Qperations Research, Vol. 8, No. 3, June 1900, pp. 390-h05. New Control Concept, Traffic Engineering, Vol. 31, No. 10, July 1901, p. 38. ‘ Osofsky, Sam. "The Multiple Regression Mehtod of Forecasting Traffic Volumes," Traffic Quarterly, Vol. 13, July, 1959. Parker, E. "Trafficometry," Traffic Engineering and Control, Vol. 4, No. 0, October 1902, pp. 3181323. Ricker, Edmund R. "Monitoring Traffic Speed and Volume," Traffic Quarterly, Vol. 13, July 1959. 125 "Spot Speed Survey Devices," Traffic Engineering, Vol. 32, N0. 8, May 1902, pp. h5-S3. Stouffer, Lloyd. "They Are Taking The Jam Out of Traffic," The Reader’s Digest, Vol. 81, No. A88, December 1902, pp. 125-128. Taylor, 8.3. "Freeways Alone Are Not Enough," Traffic Quarteriy, Vol. 13, July 1959. Volume Survey Devices," Traffic Engineering, Vol. 31, No. 0, March 1901, pp. hh—Sl. Webster, F.V. "Future Development in Traffic Signals," Traffic Engineering and Control, Vol. 3, No. 2, June I961) pp. ' ° Wohl, Morton. "Simulation - Its Application to Traffic Engineering," Traffic Engineering, Vol. 30, No. 11, August 1900. E. UNPUBLISHED ARTICLES Aitken, W.S. Don't Measure, Compute. Paper presented at the 17th Annual Instrument-Automation Conference, Instrument Society of America, October 15-18, 1902. Auer, J. H., Jr. A Systemb for the Collection and Processing of Traffic Flow Data _y Machine Bethods. A paper pre- sented to the Freeway Operations Committee, Department of Traffic and Operations, Highway Research Board, Washington, D.C., January 8, 1902. Bruening, S.M. Traffic Forecasting fly Linear Graph Model. Engineering and Computer Laboratory Seminar, Michigan State University, East Lansing, Michigan, January 29, 1963 O Cass, Sam and L. Casciato. Centralized Traffic Signal Control §y_ A General Purpose Computer. 1900 Proceedings of the Institute of Traffic Engineers, Washington, D. C. Gerlough, D.L. Notes On the Use of the Digital Computer for Traffic Signal Control, n. d. Gerlough, D. L. and F. A. Vagner, Jr. Simulation 3; Traffic in a Large Network of Signalized Intersections. Paper to be presented at the Second International Symposium on Theory of Road Traffic Flow, Sponsored by the British Road Research Laboratory, London, England, June 25-27, 1903. f_J I) (A International Business Machines Corporation. Computer Real Time Control 9; Traffic. A presentation to the Commis- sioner of Traffic, Cleveland, Ohio, June 1902. Irwin, Neal A. Computer Control 9f Traffic Sigpals. Transportation Engineering Seminar, Michigan State University, East Lansing, Michigan, February 19, 1903. Rhee, S. Young. The Urban Traffic Control Simulator. Paper presented at the 1902 Systems Engineering Symposium. International Business Machines Corporation, October 1902. F. PERSONAL CORRESPONDENCE Allgaier, Earl. Manager, Driver Education Division, Traffic Engineering and Safety Division, American Automobile Association, Washington, D.C., March 1, 1903. Auer, J.H., Jr. Principal Research Engineer, General Railway Signal Company, Rochester, New York, May 7, 1903. Campbell, W.K. State Government Account Representative, International Business Machines Corporation, Lansing, Michigan, February 8, 1903. Casciato, L. Chief Engineer, Traffic Research Corporation, Limited, Toronto, Ontario, Canada, March 25, 1903. Cottingham, K.E. District Traffic Engineer, Washington State Highway Commission, Department of Highways, Seattle, Washington, May 28, 1903. Davies, Robert. Technical Director, Mathematical Sciences, Research Laboratories, General Motors Corporation, Warren, Michigan, April 1, 1903. Foote, Robert 8. Manager, Tunnels and Bridges, Research Division, The Port of New York Authority, New York, March 1, 1903. Gerlough, D.L. Head, Traffic Systems Section, Thompson-Ramo- Wooldridge Inc., Canoga Park, California, June 12, 1903. Gervais, Edward. Project Manager, Michigan State Highway Department, Lansing, Michigan, March 28, 1903. Gibson, Arthur C. Traffic Research Engineer, City of Detroit, Department of Streets and Traffic, Detroit, Michigan, March 14, 1903. an it? Gray, G.w. Radio Corporation of America, RCA Laboratories, David Sarnoff Research Center, Princeton, New Jersey, “arCh 7, 1903. Harris, E.W. Assistant to the Director, Highway Research Board, National Academy of Sciences-National Research Council, Division of Engineering and Industrial Research, Washington, D.C., May 15, 1903. Irwin, Neal A. Vice-President, Traffic Research Corporation, New York, March 18, 1903. Manikos, J.G. Traffic Safety and Highway Improvements Department, Ford Notor Company, Dearborn, Michigan, March 15, 1903. May, Adolf D., Jr. Director, Expressway Surveillance Project, State of Illinois, Department of Public Works and Buildings, Chicago, Illinois, May 7, 1903. Nussbaum, Ernest. Civil Engineer Group, Washington Information Processing Center, General Electric Company, Bethesda, Maryland, March 19, 1903. Sweeney, W.T. Educational Administrator, UNIVAC, Division of Sperry Rand Corporation, Detroit, Michigan, March 13, 1963. Taylor, 8.8. City Traffic Engineer, City of Los Angeles, California, June 4, 1903. illlllllfllll‘jlflfllfllllulmvwllllllilllilfllljllllil 7 0