THE PLANNING AND DEVELOPMENT PROCESS

Chapter 6

THE PLANNING AND

DEVELOPMENT PROCESS

133

INTRODUCTION

The main purpose of a rail rapid transit system isto transport passengers with speed, safety, and de-pendability. The train control system provides theprotection (ATP), operational control (ATO),supervision (ATS), and communications necessaryto accomplish this purpose.

The older rapid transit systems, such as CTA,MBTA, and NYCTA, were designed to performmany train control functions manually. Until re-cently, the major uses of automation have been fortrain protection functions (ATP) and certain super-visory functions, such as dispatching. The develop-ment of new technology within the last decade or sohas made it possible to automate other train controlfunctions, and so the older rapid transit systems arenow in the process of converting to higher levels ofautomation, especially in the areas of train opera-tion and supervision.

Rail rapid transit systems built in recent years(PATCO and BART) and those now under con-struction are tending to make use of more extensiveautomation and more sophisticated train controlthan the older existing systems. Various forms ofadvanced ATC technology seem to figure in theplans of system designers from the very outset.Thus, it appears that the general trend in both exist-ing and future rail rapid transit is toward increasedautomation. In light of this, the process by whichtrain control systems are conceived, planned, pro-cured, and tested assumes great significance; and itis important to investigate the way in which theATC design evolves within the context of overallrapid transit system development.

The evolutionary cycle of ATC, like the totaltransit system of which it is part, has three majorphases: planning, development, and testing. Thesephases are generally sequential but there arenumerous interactions and iterative steps. Forsimplicity of discussion, however, the features andissues of each phase will be treated separately. Atthe end of this chapter is an examination of the sub-ject of research activities that support the overallplanning, development, and testing process.

The evolution of an ATC system can be lengthy,often as long as the evolution of the transit systemitself. Table 31 identifies the significant dates for 16systems—the five existing and three developmentalsystems considered in detail in this report and eightother systems for general reference. The CTA,

MBTA, CTS (Cleveland), and NYCTA programs in-volve addition of new ATC equipment or extensionof an existing line. For the others, the programspans the conception and development of the entiresystem. The times listed include the evolution ofgeneral train control system concepts and thedetailed engineering development.

The major issues associated with planning anddevelopment are examined in the order in whichthey generally occur in the system evolution proc-ess.

Planning (Concept Formulation and PreliminaryDesign)

The concept of the ATC system is usually for-mulated early in the overall transit system planningprocess, The major issues are concerned with theorigin of the ATC concept, the influences whichshape it, the selection of a desired level of automa-tion, and the criteria and techniques used to evalu-ate the concept and translate it into a preliminaryengineering design.

Development (Final Design and Procurement)

The final engineering design and procurementprocess may cover several years, during which theoriginal concept may undergo substantial change,The most significant issues relate to how theengineering design specifications are written, howcontractors are selected, how the developmentprocess is supervised and managed, and howemerging differences between concept and imple-mental ion are dealt with in the developmentprocess.

Testing

Testing is a continual process that begins as soonas specific items of ATC equipment are engineeredand, manufactured and ends when the entire systemis ready for revenue service. The issues in this areahave to do with the types of tests conducted, thetiming of the tests in relation to the developmentcycle, and the methods by which the ATC system isevaluated for serviceability and conformance tospecifications.

Research and Deve lopment (R&D)

R&D is a supportive activity that runs concur-rently with planning, development, and testing. Theissues to be examined include the types of R&Dbeing conducted, its application to the design of

135

TABLE 31.—Significant Dates in the Engineering Planning, Procurement, and Testing of ATC for Various Transit Systems

CTA CTS(Lake (Airport

St Exten.

Line) sion)

N Y C T A

(System

M B T A Im -

(Red prove.Dade Co. D/FW MARTA Line) NFTA ments)

Twin

Cities

SEA. Area

P A A C PATCO RTD T A C MTC W M A T A

————

1959 1958 1969 1967 1967 1960

Balti -more BART

1962 Before1965

1964 1965 1962 Before 1967 (f-l

1966

Planning b 1962

1967

1951

1953ATC

PreliminaryD e s i g n

S t a r t e d

ATC

PreliminaryDesignCompleted

ATC Pro-curement

Specifica-

tions

Issued

ATC Pro-

curementContractAwarded

ATC

BebuggingStarted

ATC First

UsedInPassengerService

Initial Plan-ing

(Years)

ATC Evolu-

tion( Years)

Total TimeSpan

( Y e a r s )

1962 1969 1969 1966 1966 1973 Before 1963 1970 1967 1969 1962

1963

1973 1966 1965 (1975) 1970 1974 1970 1973 1972 1967

(d)(1977) 1966 1966 (1978) 1971 1974 1966 1969 1971

1967 1966 1971 (1976) (1975) 1966 1969 1971

1973 1971 (1978) 1968 1972

(1981) 1972 1967 1966 (1984) 1974 (1977) 1971 (1981) (1978) 1969 1973 (1981) 1976

5 4 4 1 6 (f) 4 5

5 3+ (15) 5 (11) 5 (8) (f) (14+) 6

5

(14)

(19)

2

19

21

1 1 2 2

5+ (12) 14

6 (14) 165 3+ (20) 9 (15) 5+ (14)(f)

(18) 11

a Unless otherwise noterf, the dates Ilstwi are for the start of the actlvltv Dates enclosed In parenthtws ( ) are planned (iates All act]vltws are fnr new svstems except as notecfb ATC plannlng IS generally Influenced by a numlwr of system IIeclslons w this date IS the start of the nverall syswm planning“ Prellm\nary d~slgn IS cons] cierwi to In[.lu[ie snme conceptual work as well as ciemonstratlnn protects where applicable{i

Act]v)t y IS cllrrently In progresse Most transit systems are [.onstructwi In phdses The program duration listed IS for a single phasr or the first phdse of the programs Becauw early planrrlng of mult~ple prngrams IS usIIally comprehenmve

the t~me requlrwi for such planning WI]] generally Iw longer than required for d ~maller single-phase effOrtf At NYCTA the process of equipment replacement and updat!ng 1$ vmtually continuous

new systems, the use of test tracks, andneeds in the area of ATC technology.

major R&D

ISSUE D–1: DESIGN CONCEPTS

How do ATC design concepts originate, andby what criteria is the level of automationselected?

For new systems, ATC design conceptsemerge from policy and planning decisions aboutthe general transit system concept. Initial selec-tion of the level of automation tends to be in-fluenced more by social, economic, and politicalconsiderations than by engineering concerns. Inalready operating systems, where ATC is in-stalled to upgrade or extend service, engineeringconcerns–especially evolutionary compatibilitywith existing equipment-are predominant. Forboth new and old systems, the experience ofothers (particularly their mistakes) has an impor-tant influence.

Some preliminary notion of the type of train con-trol system desired is usually included in the state-ment of the basic transit system concept preparedby the policymaking body responsible for planningthe system. For all of the transit agencies investi-gated in this study, the policy and planningauthority is a commission or board of directors cre-ated by legislative act. The size and compositionvary. Some are elected; others are appointed. Themembers are usually not engineers and seldomhave technological backgrounds in the area of tran-sit operation and train control, but there is alwayseither a technical staff or an engineering consultantfirm to assist the board in planning activities. Some,particularly transit systems already in operation,have staffs of considerable technical competence.For example, the CTA and NYCTA staffs do all theengineering planning for new developments andoversee procurement and testing. In general,however, the local policy and planning agency aug-ments the technical capability of its staff by hiringconsultants who conduct studies to support plan-ning decisions and flesh out the basic design con-cept. In some cases, the consultant firm may also beresponsible for the subsequent engineeringdevelopment of the system.

The activities of the planning agency are in-fluenced by many factors: State and Federal legisla-tion, regulatory agency rules and decisions, UMTA

policy, economics, public opinion, local social con-cerns, labor relations, and political interests, toname a few. Technical, considerations often playonly a small part and may be overridden by theseother concerns. Specific examples from among thesystems investigated will help to illustrate thenature and diversity of the ways in which ATCdesign first takes shape.

The PATCO Lindenwold Line was planned andconstructed over an n-year period. It is not clearwhen the basic ATC design concept was formu-lated; but an engineering consul tant reportpublished in 1963, about midway between the timeof the initial decision to build the system and thetime the line was opened for service, recommendedthe use of ATP and ATO. The tone of the reportmakes it plain that the nature of the train controlsystem was still an open question 5 years after theplanning process started. The primary justificationadvanced by the consultant for ATP was safety, andfor ATO efficiency of operation.

In contrast, an ATC design concept for BARTwas established very early in the planning processand took over 20 years to evolve. Original planningstudies conducted by engineering consultants toBART in 1953 to 1956 advanced the general conceptof completely automatic operation at high speedand short headways. An onboard “attendant” wasenvisaged, not as an operator but as an aide topassengers, much like an airline stewardess. Theidea of building a glamorous “space age” systememploying the most advanced technology seems tohave been a dominant concern in BART from thevery beginning, Th i s app roach was c l ea r lymanifested in the ATC concept. The justificationmost often given was that advanced train controltechnology was necessary for the, high-speed, short-headway operation needed to attract patrons,

CTA, in planning the conversion to cab signal-ing, appears to have been most strongly influencedby operational and engineering factors, Cab signalswere seen by CTA as an improved method of assur-ing train separation and preventing overspeed, i.e.,as a way of enhancing safety. Compatibility withexisting signal equipment and other elements of thesystem was also a factor (as it is in MBTA wherecab signal conversion is now being implementedand in NYCTA where it is in the planning stage).Engineering and equipment concerns are also adominant concern in the planned expansion ofPATCO, where the existing ATC system dictatesthat the new lines have the same operational

137

— . .-

characteristics andbe integrated with

level of automation in order tothe present line.

Operational transit systems for airports (such asSea-Tac and AIRTRANS) feature automatic, crew-less train operation. These systems were plannedand built in a rather short time span (6 years forSea-Tac, 9 for AIRTRANS). The concept of un-manned vehicles was inherent in the nature ofthese systems from the beginning. It was felt by theplanning agencies and their consultants that fullyautomatic operation offered significant savings inlabor costs and was the only way to make thesystem economically viable.

There are sometimes general engineering deci-sions made during the planning process that maylimit the technology that can be employed for ATCequipment. For example, a number of transit plan-ning agencies have decided to employ only equip-ment already proven in use by other operating tran-sit systems. For WMATA, the schedule set by thepolicy makers did not permit extensive R&D andengineering studies before selecting a train controlconcept. Therefore, WMATA engineers decided tospecify an ATC system that could be realized withproven, existing hardware.

The formulation of the ATC system concept isalso strongly influenced by events in other transitsystems. The community of rail rapid transit agen-cies, consultants, and suppliers is a small fraternity.There is a continual exchange of informationamong the members and a high degree of mutualawareness of plans, problems, and operation ex-perience. Because the supply of qualified transitconsultants and engineers is limited, there alsotends to be a steady interchange of personnelamong transit properties, consultant firms, andequipment manufacturers. These forms of interac-tion assure that the experience of others will bereviewed during concept selection and preliminarydesign,

However, the review of others’ experience isoften rather narrowly focused. There is a tendencyto be swayed more by specific problems and inci-dents than by overall statistics and the general pat-tern of operations. “Avoiding others’ mistakes”seems to be a more dominant concern than emulat-ing their success. For instance, the problems en-countered by BART were in part responsible for themore conservative approach adopted by WMATAand Baltimore MTA. Atlanta’s planners also havechosen a train control system less sophisticated

than that originally proposed by their consultants(PBTB, who were responsible for BART), partly asa reaction to the experience in San Francisco. Cau-tion is a prudent course, but the rapid transit indus-try could also benefit if there were a more com-prehensive body of comparative performance datato help make decisions on an analytical, rather thana reactive, basis.

The salient points that emerge from an examina-tion of the initial planning process are that ATCdesign concepts originate (sometimes early, some-times late) in policy-level decisions about thegeneral nature of the system. The methodologyemployed to arrive at concept definition is often in-formal and influenced strongly by engineering con-sultant firms engaged to assist in planning thesystem. Except in the case of modernizing an exist-ing system, technical considerations of train controlsystem design seldom predominate. Route struc-ture, service characteristics, vehicle design, right-of-way acquisition, cost, and local sociopoliticalconcerns tend to be given greater importance at theearly stage of planning. The engineering aspects oftrain control are most often deferred to a latter stageof planning, when design specifications are to bewritten, As a result, the embryonic ATC design isusually not defined in detail until other parts of thesystem have taken shape, The preliminary ATCconcept thus tends to develop a life and perma-nence without being subjected to engineeringscrutiny and cost-benefit analysis to determine itsappropriateness for, and compatibility with, the restof the system,

There seems to be a crucial difference betweenexisting and new systems. The former give greaterweight to engineering concerns and specific opera-tional needs in defining an ATC concept. Newsystems tend to take a broader, more informal, andless technical approach, The engineering-orientedapproach offers the advantage of assuring a worka-ble ATC system tailored, although perhaps notoptimally, to specific local needs. But there is a dis-advantage. The scope of the ATC concept inupgrading an existing system tends to be limitedand constrained by what already exists. The bolder,“clean sheet of paper” approach employed by manynew systems results in a more technologically ad-vanced concept and greater coherence betweenATC and the system as a whole, but the practicalproblems of development and engineering may notalways be given sufficient attention,

138

I S S U E D – 2 : S Y S T E M DEVELOPMENT

How is the ATC system concept translatedinto preliminary and final functional design ?

Most system development work is done byengineering consul tants , except for largeestablished rapid transit systems where it is doneby the in-house staff. The methodology varies,but there is a trend toward a more systematic andsophisticated approach using simulation, systemanalysis, system assurance studies, and testtracks.

The first step in the development process forATC systems is preparation of a preliminary func-tional design, expressing the basic concept and itsunderlying policy decisions in engineering terms.The preliminary design defines performance re-quirements and organizes the ATC system withrespect to functional relationships among systemcomponents. At this stage, the ATC system is sepa-rated into its major subsystems (ATP, ATO, ATS,and communications), and the functions required ofeach are specified. Further analysis may separatethe system into carborne and wayside elements.The preliminary design also defines the interfacesbetween ATC and other parts of the transit system.

For most of the transit agencies investigated, thetechnical staff plays some role as engineering plan-ner in preliminary design. However, the extent ofstaff involvement varies widely. In establishedoperating agencies, such as CTA, CTS, andNYCTA, the engineering staff does almost all of thepreliminary design work. In new systems, wherethe technical staff may be quite small, especially inthe early planning phases, engineering consultantsare generally and extensively used. Heavy par-ticipation by consultants is also characteristic inestablished systems undergoing a major program ofnew construction or modernization. While the pro-portion of staff to consultant participation varies,there appears to be wide agreement among transitsystem managers that staff involvement should notfall below a certain minimum level, roughly 15 to 20percent of the design work. In this way, theauthority can maintain technical involvement inthe preliminary design process and exercise propercontrol over system evolution.

Several kinds of methodology may be employedin preliminary design. The specific methods differwidely from authority to authority, and it is difficult

to discern any common thread, beyond the generalbelief that technical studies are needed to gatherand analyze information about the performance ex-pected of the system. In the new systems now underdevelopment, there seems to be an increasingreliance on the so-called “systems approach”79 andthe use of techniques such as simulation, ridershipanalysis, function/task analysis, and cost/benefitstudies. Several agencies (BART, CTA, NYCTA,Sea-Tac, and PAAC) have also conducted studies attest tracks on their properties to gather informationneeded for preliminary design.

The application of system analysis techniquesdoes not appear, however, to extend very deeplyinto the design of the ATC system itself. There is atendency, for instance in cost/benefit studies, totreat ATC as a whole, without examining thechoices that may exist within the train controlsystem as to degree of automation or alternativemethods of achieving a given level of automation.One reason is the general lack of empirical data onthe performance of ATC systems, which precludesa precise formulation of potential benefits, A sec-ond reason is the overriding nature of the safetyfactor which strongly influences designers to auto-mate the train protection function, without regardfor the cost/benefit relationship of ATP to otherfunctional elements of ATO or ATS. Also, since theentire ATC package typically amounts to only 5percent or less of the total capital cost of the transitsystem, there is a belief that cost/benefit analysisshould be concentrated in areas where the payoffwill be greater.

Thus, the process of developing a preliminaryfunctional design of the ATC system still tends tobe more art than science, but there is a trend towarduse of more objective, quantitative, and systematictechniques. This is particularly evident at the points

7~he “sy.sterns approach, ” which derives mainly from aero-space technology, is a collective designation for techniques usedto solve complex problems in a methodical, objective, and oftenquantitative way. The systems approach involves a logical andreiterative analysis of the system into its constituent parts, eachrepetition leading to a greater degree of specificity, Othercharacteristics of the system approach include measurability ofparameters, constant recognition of subsystem interdependence,and parallel analysis of elements. The heart of the systems ap-proach is the *’System Engineering Cycle” which involves foursteps: (1) convert system requirements to functional require-ments, (2) convert functional requirements to specific detail re-quirements, (3) conduct analysis to optimize parameters, and (4)convert specific detail requirements into end products, (Grose,1970)

139

.-

of interface between ATC and other subsystems,where mutual influence and interdependence canbe reduced to quantitative expression and theparameters of performance can be manipulated.Even here, however, ATC system characteristicstend to be treated as dependent variables, i.e., thedriving concerns are other system characteristics, towhich the ATC system design must be accommo-dated.

System design is a continual and reiterativeprocess, preliminary functional design merging intofinal engineering design without any clear line ofdemarcation, The process culminates in the state-ment of specific equipment and performance re-quirements, suitable for incorporation in procure-ment specifications. Often, final design coincideswith the preparation of procurement specifications,and it is difficult to separate the two activities.However, for the purpose of this discussion, finaldesign is considered to include all activities neededto define the detailed technical requirements of theATC system, up to but not including the actualwriting of procurement specifications.

As in preliminary design, the final design is ex-ecuted either by the technical staff of the transitagency or by engineering consultants. Here, too, theolder and established agencies tend to rely more ontheir own personnel, and new agencies more onconsultants. Usually, a single consultant is hired forfinal design of the complete ATC system –carborne,wayside, and central control elements. Thisconsultant is often, but not always, the same firmthat carried out the preliminary functional design ofthe ATC system. Once reason for selecting a singleconsultant for the entire process is to assure con-tinuity and coherence of the ATC design as itdevelops, It is also considered advisable to have asingle consultant for all parts of the ATC system toensure integration of the design of vehicle andwayside equipment and their all-important inter-face.

Many of the factors that shape the preliminarydesign of the ATC system continue to have signifi-cant influence during the final design process. Non-technical factors still play a strong, but perhapsdiminishing, role as the system moves from plan-ning to engineering. The continuing influence ofnontechnical factors is not surprising since they areusually built into the design criteria and guidelinesthat emerge from preliminary design and are ap-plied to the final design. Still, as the system ap-

proaches the hardware stage, it is to be expectedthat purely engineering considerations should cometo the fore. Generally speaking, however, theprocess of generating detailed engineering require-ments from preliminary design criteria is basicallyan interpretive effort, with the experience and judg-ment of the designer playing the dominant part.However, there are two more formal designmethods that are being used increasingly in newt r a n s i t s y s t e m s . T h e y a r e s y s t e m s a f e t ymethodology and quantitative reliability, main-tainability, and availability analysis.

Most of the systems now being planned are in-cluding a formal system safety study, involvingdefinition of safety criteria, analysis of potentialsafety problems, and identification of ways to elimi-nate or minimize hazards, Some designers considerthis approach to safety superior to the traditionalmethods of “fail-safe” design. Others disagreesharply.80 It appears, however, that much of thecontroversy over the “fail-safe” and “systemsafety” methods is semantic; and it is premature todetermine whether the results of the two ap-proaches will differ, The important point is thatdesigners are turning, at least in the area of safety,to more systematic and quantitative methods ofanalysis,

Until recently, it has not been the practice in thetransit industry to specify safety requirements inquantitative form, i.e., as a numerical statement ofrisk or probability of occurrence. Many believe thatthe levels of safety which must be achieved are sohigh that it is difficult, if not impossible, to statemeaningful quantitative standards and to devise anacceptable and practical method of verifying thatthey have been met. This view is not universallyheld, and the topic is highly controversial .However, it does appear that future ATC specifica-tions will place strong emphasis on formal pro-cedures by which potential safety hazards can beidentified, evaluated, and reduced to “acceptable”levels. An effort is being made to put hazardanalysis on a quantitative basis, but much of thework is likely to remain qualitative and judgmental.(Again, this view is not shared by all in the transitindustry, ) Along with the emphasis on quantitativemethods, there is also a trend to define safety in asense that is broader than just train protection andto deal with the safety aspects of the total system.

1111%111 chapt(~r !I. p~igr tlh for (i (1 is(:llssion of this topi(; .

140

The second formal design method that is cominginto wider use in the transit industry is quantitativereliability, maintainability, and availability (RMA)analysis. A discussion of this design technique ispostponed to Issue D–4, where it is considered aspart of the general question of how these aspects ofsystem performance are written into procurementspecifications.

ISSUE D—3: PROCUREMENTSPECIFICATIONS

How are ATC design requirements specified,and is there a “best” way to write such specifica-tions ?

There are two basic approaches to writingthe design (equipment-specific)specifications—

approach and the functional (performance) ap-proach. Each has advantages and disadvantages.The only generalization to be made about the“best” way is that, whichever approach is used, itis of crucial importance to specify equipmentperformance standards and to define explicitlythe means of testing.

The final design of the ATC system is docu-mented in procurement specifications in terms ofrequired performance for ATC functions and/orequipment components. However, the procurementspecifications have a much broader scope than justa listing of required ATC system performance. Re-quirements for documentation, scheduling, installa-tion, management visibility and control, andvarious types of testing may be specified togetherwith numerous contractual and legal provisions.The procurement specifications include all of thedetailed information required for a prospective sup-plier to prepare a bid.

As a general rule, the organization that does thefinal design of the ATC system also prepares thetechnical portions of the procurement specificationfor that system. At times, another consultant writesthe procurement specifications in cooperation withthe final designers. In this way some additional ex-pert knowledge is incorporated into the specifica-tions.

The most common method of preparing procure-ment specifications is by drawing on availablespecifications for similar equipment, from prelimi-nary proposals submitted by equipment suppliers,or from experience gained through testing or use of

similar equipment. Often, a general incorporationof test and use experience is achieved by requiringthe use of “proven technology,” which means thatthe same or similar equipment must have been usedor tested successfully on an operating transitproperty in the United States.

There are two basic approaches to writing pro-curement specifications. Requirements can bestated in functional terms (performance specifica-tions) or in equipment-specific terms (designspecifications). The two are not mutually exclusive,and in practice something of each approach is used.Thus, implicit in even the most design-orientedspecification is the expectation that the equipmentshould perform in a certain way,

The design type of specification indicates, to agreater or lesser degree, the equipment or systemcomponents needed to perform individual func-tions. In the extreme case, design specifications callfor particular items, for which only a narrow rangeof substitutes, or none at all, are acceptable. Suchspecifications are often issued by transit agenciesthat have similar, satisfactory systems in operationand wish to assure compatibility of the new equip-ment with that already in place. Recent procure-ments of cab signaling equipment by CTA typifythis approach. Somewhat less restrictive is thedesign specification that calls for a type of equip-ment with stated characteristics but leaves the sup-plier some room for choice. The WMATA traincontrol system specification is an example of themodified design-oriented approach, which hassome of the features of a functional specification.

Functional (or performance) specificationsdefine what functions are to be accomplished butnot the way in which they are to be accomplished.For ATC systems, the BART specification comesclosest to the purely functional approach, TheDiablo test track was operated for the purpose ofdetermining the feasibility of new ATC concepts(not to select a system). At the end of the testingperiod a functional specification was written to ac-commodate any of the concepts successfullydemonstrated (and many others). For example, thebasic train separation system could have used radar,track circuits, or any other device that met thestated functional requirements,

Table 32 below is a rough classification of thetype of specification used by seven transit systemsin recent procurements. The development of the sixnewest systems (Baltimore, Dade County, MARTA,

141

————-——

NFTA, RTD, and Twin Cities) has not yet advancedto the point where ATC system specifications havebeen written.

TABLE 32.—Type of Specification Used in Recent ATC Pro-curements

TYPE OF SPECIFICATIONSYSTEM Design Functional Combined

AIRTRANS x

BART x

CTA x

NYCTA x

PATCO x

SEA-TAC x

WMATA x

The use of a design specification permits thebuyer to exercise a high degree of control over theequipment purchased. At the same time, however,it requires considerable experience and technicalcompetence on the part of the buyer to be sure thatwhat he specifies will perform as intended. There isalways the risk that individually procured sub-systems will not prove compatible, with the buyerhaving no recourse but to go through a process ofredesign or retrofit. If a testing procedure has beenestablished in the specification, product evaluationand acceptance is usually easier for the buyer whohas followed the design approach. To the extentthat design specifications are equipment-specific,they lock the buyer into a given technology and donot allow taking advantage of innovation, economy,or other improvements that the seller might other-wise be able to effect.

One of the major advantages of a functionalspecification is its independence from particularmeans of implementation. It gives the supplier greatlatitude when innovation is desired or when a widerange of hardware is acceptable. This approach ismost compatible with a new system being builtfrom the ground up or with an independent part ofan existing system. In effect, the functionalspecifications transfer some of the responsibility forsystem design from the procuring agency to theequipment supplier.

Functional specifications, because they are lessdetailed, may be somewhat easier to prepare thandesign specifications. On the other hand, it is some-

what harder to define the desired end product withprecision. The functional specification allows thesupplier to be creative, but it can also provide theopportunity for cutting corners. Litigation, as in thecase of the BART train control system contract, isalways a possibility if differing interpretations aretaken or if the method of testing system perform-ance is not well defineed. From the buyer’s stand-point, one difficulty with functional specificationsis that it may not be possible to determine if the pro-duct will meet performance requirements until thecomplete system is assembled,

The re i s no un ive r sa l ag r eemen t on t hesuperiority of either type of specification, Eithercan be employed successfully so long as the buyerrecognizes the shortcomings of the selected ap-proach and so long as the standards for an accepta-ble . product are clearly and fully defined. Theresults of the WMATA specifications, which com-bine a functional and a design approach, will beawaited with great interest to see if they offer acompromise solution to the problem of specifyingequipment requirements and characteristics.

It is of crucial importance that both the criteriaand methods of testing the equipment be made ex-plicit in the procurement specification, From a prac-tical standpoint, the design type of specificationmay offer some advantages over the functionalspecification in terms of the ability to define andmeasure reliability and maintainability-a problemthat lies at the heart of the difficulties encounteredby most new systems. Because of its importance,the topic of how RMA requirements are specified istreated as a separate issue immediately following.

ISSUE D--4: SPECIFICATION OFRELIABILITY, MAINTAINABILITY, AND

AVAILABILITYAre the methods of specifying reliability,

maintainability, and availability (RMA) adequ-ate to assure that ATC systems will give goodservice ?

This has been one of the most troublesomeareas of ATC system design and development.Transit agencies are becoming increasingly con-cerned with RMA problems, and an effort isbeing made to write specifications in more pre-cise and quantitative terms. In their present state,however, RMA specifications still fall short ofwhat the transit industry (both buyers andmanufacturers) consider satisfactory.

142

RMA specifications can be divided into twoclasses—those that state quantitative requirementsand those that do not. Before issuance of the BARTspecifications, most transit agencies followed anonquantitative approach to RMA specifications,and some still do. The BART specifications were apioneering effort to introduce in the transit industrythe quantitative methods used in the aerospace in-dustry for specifying RMA. This was a major in-novation at the time and, like nearly everything elseassociated with BART, controversial. However, allthe agencies planning new systems are now incor-porating some form of quantitative RMA require-ments in their specifications.

Historically, reliability and maintainability havebeen treated only in general terms in procurementspecifications by transit agencies. Some form ofwarranty was called for, but specific requirementsas to reliability (mean time between failure, orMTBF) or ease of repair (mean time to restore, orMTTR) were not stated. Certain transit agenciescontinue to follow this practice for a number ofreasons. In some cases, the procurement consists ofadditional equipment similar or identical to pastpurchases. Thus, the expected performance of theequipment is understood by buyer and seller to belike that already in use. Another reason has to dowith the size and nature of the transit industry.There are only a few buyers and even fewer sellers,all of whom have been in business for many years.Hence, the needs of the former and the capabilityand reputation of the latter are well known. In suchcircumstances, it is considered unnecessary to drawup elaborate and detailed statements of RMA re-quirements. The seller is familiar with the kind ofequipment now in use by a transit system, and thetransit agency knows that the seller must standbehind the product in order to remain in considera-tion as a source of supply. A third reason for takingthe nonquantitative approach, especially in smalltransit systems, is that the managing authority maynot feel it is cost-effective (or they may not be ableto get the funds) to prepare specifications that in-volve extensive engineering analysis, and perhapstesting.

The quantitative method of specifying RMA hasfound increasing favor in the transit industry fortwo basic reasons. First, the type of equipment nowbeing purchased, especially for ATC systems, ismuch more complex and technologically sophisti-cated, creating a need for the document thatgoverns the purchase of the equipment to become

increasingly detailed and precise. Second, the num-ber of suppliers has increased and now includesfirms without a long and established record in thearea of train control equipment manufacture andinstallation. Starting with BART and continuingwith WMATA, MBTA, and a number of newsystems being planned, transit agencies are turningto a quantitative approach.81 Still, a decade after theBART initiative, the specification of RMA require-ments remains a developing art,

There are significant differences in how quan-titative RMA requirements are written, dependingupon whether the procurement document is adesign or a functional specification. In a functionalspecification, the buyer defines generic types offailures, their consequences, and required systemperformance. The seller is (in theory) free to con-figure the system in any way seen fit so long as thefunctional requirements are met and the systemperforms as expected. In a design specification, thebuyer develops a specific equipment configuration,evaluates the consequences of failure of each com-ponent (equipment items not functions), anddefines the component performance requirements.The seller must then meet the performance require-ments on an item-by-item basis. Thus, the sellermay well have no responsibility for the perform-ance of the total system, but only for the parts as setforth in the procurement specification. In effect, thefunctional specification transfers much of theresponsibility for detailed system design to theequipment supplier, whe rea s w i th a de s ignspecification this responsibility is retained by thepurchaser.

With regard to RMA, the difference betweendesign and functional specifications centers aroundthe definition of failure. In design specifications thedefinition is reasonably clear-cut and precise.Failure means that a given component does not re-spond to a given input or fails to make a particularoutput within stated tolerances. In a functionalspecification, failure is defined not in terms ofspec i f i c equ ipmen t pe r fo rmance , bu t moregenerally as the inability of the system (or sub-system) to perform certain functions. Some func-tional specifications (such as those prepared forBART and Sea-Tac) also identify the consequencesof failure that are of concern.

MMARTA, Dade County, Denver RTD, NFTA, PAAC, andTwin Cities are all contemplating the use of quantitative RMAspecifications.

143

. .—

A problem of interpretation can thus arise inevaluating equipment procured under a functionalspecification. Some failures and their consequencesare defined; but others are not, even though thesame piece of equipment may be involved. Whatthen is a failure? And what particular equipmentmalfunctions are to be counted in determining thereliability of the purchased equipment? There is adisagreement, and litigation in progress, betweenB A R T a n d t h e A T C e q u i p m e n t s u p p l i e r(Westinghouse Electric Corporation) as to the in-tent of the specification on these very points.

The WMATA train control system procurementspecification, written with the BART experience inmind, attempts to deal more clearly with the defini-tion of failure. In the WMATA specification, failureis defined as “any malfunction or fault within anequipment which prevents that equipment fromperforming its function in accordance with thespecification. ” Thus, it appears that WMATA RMArequirements pertain to all equipment failureswithout regard to the effect on train operation.However, the specification does not clearly indicatewhat modes of operation are to be counted and howequipment operating time is to be reckoned incalculating MTBF, In some systems, ATC units arelocated at each end of the train and actually controlonly half the time. If a failure occurs in a unit notinvolved in train operation at the time of malfunc-tion, is this to be counted as failure? And if so, howmany hours has it been operating? All the time thatthe car has been in revenue service, or only that partof the time that the ATC unit has been used to con-trol the train?

Without belaboring the example, it is clear thatthe transit industry still has not reached a full anduniversally accepted understanding of how tospecify and test equipment reliability. A recentstatement by a representative of an equipmentmanufacturer (King, 1975) highlights the continuingproblem.

Success and failure of transit equipment andsystems must be defined in relation to their mis-sion. Indeed, the term “mission” itself probablyrequires redefinition. Many industry specifica-tions in recent years have not agreed on suchpoints as whether a transit vehicle completes itsmission at the end of one trip or the end of a fullday, or when that day ends, or whether the vehi-cle must be available during all peak serviceperiods. If the function of transit equipment is

carrying passengers, has a mission failed if anequipment outage occurs during nonrevenueservice? These are some of the fundamentalquestions which must be answered to define tra-ditiona1 reliabi1ity in a manner acceptable totransit industry application.

One of the significant problems affecting theability of the transit industry to draw up meaningfulRMA specifications is the lack of a data basedescribing the performance now being achieved inthe industry. Individual manufacturers have someinformation, as do individual transit systems, butthere is no uniform method of reporting and noavailable industry-wide data base.

This need has been recognized by transit agen-cies and equipment manufacturers; and, throughtheir industry organization (the American PublicTransit Association), an effort is underway to dealwith the problem. APTA task group, known asRAM (for Reliability, Availability, and Main-tainability), has been assigned the responsibility ofdeveloping recommendations for a standardizeddata collection and reporting procedure. Theproblem of making these data generally available,free from local transit system bias and manufac-turers’ proprietary concern, is still unsolved,

ISSUE D–5: ‘ EQUIPMENT SUPPLIERS

What firms supply ATC equipment? Is theretransfer of ATC technology between automatedsmall vehicles and rail rapid transit systems?

Historical ly, two U.S. f i rms—GRS andUS&S--have supplied most of the ATC equip-ment to the rapid transit industry. In recentyears, several new firms, supplying either specialproduct lines or control equipment for smallvehicle sytems, have entered the market. Themajor transfer of ATC technology is from railrapid transit to small vehicle systems, but not thereverse.

The suppliers of ATC equipment to the rail rapidtransit industry fall into two distinct groups: thosethat provide a broad line of services and equipmentand those that have limited lines or specialtyproducts. There are many firms in the latter catego-ry, but the former includes four companies, GeneralRailway Signal Company (GRS) and Union Switchand Signal Division of the Westinghouse Air Brake

144

Company (US&S)82 are old, established firms thathave a long history in the signals and communica-tion business and have dominated the market. Re-cently, two new suppliers have entered the com-peti t ion. Westinghouse Electr ic Corporat ion(WELCO) supplied the ATC system for BART,where they were low bidder against GRS and US&S.Transcontrol is furnishing the ATC system for theSan Francisco MUNI light rail system and for theToronto Transit Commission in Canada,

There are many more suppliers of ATC equip-ment for small, automated-vehicle, fixed-guidewaysystems. In addition to GRS, US&S, WELCO, andTransControl, the list includes Philco-Ford, TTI(now TTD) and Varo Monocab.

The number of firms supplying small-vehicleATC systems, and the organizational relationshipsamong them, change from year to year. Some dropout of the market, new ones enter, and others formjoint ventures or acquire each other. It is a marketwhere there are many more companies offeringsystems than have actually received contracts forinstallations, Further, the resulting contracts areusually rather small. The complete “Satellite Tran-sit System” installation (guideway, vehicles, andcontrols) at the Seattle-Tacoma airport was about$7 million, while the AIRTRANS system at theDallas-Fort Worth airport was about $3 I million.The ATC portions of these systems were about 7 to12 percent of the total contract prices.83

To date, transfer of technology between conven-tional rapid transit systems and the new small vehi-cle systems has been in one direction—from theconventional to the new systems. Reverse transfer,and entry of small vehicle system developers intothe conventional rail rapid transit market, has notoccurred, perhaps due to the much larger size of thecontracts and capital commitments required tocompete in the conventional rail rapid transitmarket, or perhaps due to the failure of AGT sup-pliers to develop workable systems for rail rapidtransit application.

While some foreign-made ATC equipment isutilized in the United States, the market is not really

8ZUnlon Switch and Signal is also referred to by the acronymof its parent firm, WABCO.

@aIn relative terms, this proportion is somewhat greater thanthe 3 to 5 percent of total contract price that is typical for railrapid transit systems. The absolute dollar amounts, however, arequite small.

receptive to foreign incursions. There are severalreasons. Some procurement specifications excludeforeign suppliers by requiring prior transit servicein the United States or by including restrictions onforeign-made components. Also, U.S. transit agen-cies tend to doubt that foreign suppliers would beable to provide continuous long-term service.Finally, there are some major differences betweenU.S. and foreign ATC technologytechniques.

ISSUE D–6: CONTRACTOR

and engineering

SELECTION

How ore contractors for ATC design andengineering selected?

The lowest technically qualified bidder isusually selected. Competitive bidding and awardto the low bidder is required by law in manyStates.

Usually, two or more suppliers will compete forthe opportunity to design, build, and install ATCsystem hardware and software in response to thetechnical specifications describing required systemcharacteristics. Ultimately, responsibility for selec-tion of the supplier rests with the directors of thetransit authority. Most frequently, the directors relyon their technical staff for evaluation of the pro-posals and for monitoring the work of the selectedcontractor. This procedure was followed at CTA,CTS, MBTA, NYCTA, and PATCO. However, atBART, the general engineering consultant (Parsons,Br incke rho f f -Tudor -Bech t e l ) was de l ega t edauthority for some of the contractor selection andmanagement. Interviews with personnel at newsystems in the planning or early constructionphases (MARTA, RTD, WMATA, Balt imore,NFTA, and the Twin Cities) indicate that theseagencies will also utilize consultants to assist incontractor selection and management,

The increasing involvement of consultants incontractor selection and management for new railrapid and small vehicle systems reflects the increas-ing complexity of new rail rapid and small vehiclesystems. The design and development of suchsystems is often beyond the capability of the limitedstaff maintained by most transit agencies. It shouldbe noted, however, that consultants may havesomewhat different motivation and may use some-what different evaluation criteria than the transitauthority,

145

Contractor selec tion is relatively simple when The opportun“off-the-shelf” equipment is to be used and thecompetition reduces to a matter of price amongprospective suppliers, all of proven capability.Often, however, the available equipment does notsatisfy all of the specifications and requirements.Contractor selection then involves identifyingqualified suppliers, publishing an invitation for bid,evaluating the bids received from the prospectivesuppliers, and awarding a contract. Table 33 sum-marizes the contractor selection approaches used bytransit authorities in several recent procurements.

Several of the transit authorities require that aprospective ATC system supplier be a manufac-turer of equipment proven in use on operating tran-sit systems in the United States. If the ATC systemat BART is considered to be proven, this restricts thelist of qualified ATC equipment suppliers to justthree companies: GRS, US&S, and WELCO.However, technical personnel at some authoritiesdo not accept the BART ATC system as proven.Thus, only GRS and US&S are presently consideredqualified by these authorities. The list could beenlarged by including Transcontrol if Canadian in-stallations were accepted.

ity for a new company to becomequalified as a supplier of ATC equipment is offeredby several authorities, who will permit the com-pany to install and demonstrate ATC equipment ata test track location on the authority’s property. Iftesting proves that the equipment has desirable per-formance features together with acceptable safety,quali ty, rel iabi l i ty, and maintainabil i ty, theauthority’s technical staff may approve this com-pany’s qualifications to bid for the next ATC equip-ment procurement. The prospective supplier mustbear the expense of the demonstration equipment,installation, maintenance, and testing in this pre-qualification program.

Prior to 1969, the Dallas/Fort Worth AirportBoard conducted an investigation of possible sup-pliers of an automated system. As a result of this in-vestigation a Varo/LTV/GRS team and Dashaveyorwere selected as the two (and only) qualified candi-dates. These two submitted preliminary engineer-ing reports in October 1969. In 1970, Varo/LTV/GRSand Dashaveyor received technical study grants fordemonstration of their systems at the plant. Initialbidding for AIRTRANS took place in March 1971,with Varo/LTV/GRS and Dashaveyor being the

TABLE 33.—Contractor Selection Approaches

Transit Bidder Evaluation ContractSystem Qualification Process Award

BART (d) (b, c) WELCO

CTA (b) GRS, US&S

CTS (d) (b) GRS

D/FW (e) (b, c) GRS

MARTA (d) (b, c)

MBTA (d) (b) GRS, US&S

NYCTS (a, d) (b) GRS, US&S[f)

PATCO (d) (b, c) us&s

SEA-TAC (b, c) WELCO

WMATA (g) (b) GRS

Demonstration at test track.Low bid.Proposed performance.Manufacturer of proven equipment.Demonstration at plant was an original requirement.

At a second bidding there was no such prequalification.R-44 and R-46 procurements.Preliminary proposals.

146

only bidders allowed. One bid was rejected as toohigh, and the other was rejected as not responsive tothe specifications. In May 1971, a second biddingtook place with four bidders: Bendix/Dashaveyor,VSD-LTV, WABCO Monorai l Division, andWELCO. VSD-LTV was selected as the supplier.The subcontract for the train control system wasawarded to GRS by VSD-LTV.

The “invitation to bid” requests a cost quotationfor supplying the ATC system and services definedin the procurement specifications. The solicitationmay also require submission of a technical proposalthat describes how the bidder intends to satisfy therequirements of the procurement specifications. Inaddition to the technical requirements, provisionsfor documentation, program planning, managementvisibility and control, quality control, acceptanceand systems assurance testing, and the many otherfactors specified as important to the procurementmust be taken into account by the prospective sup-plier in preparing his bid. Experience shows that itis very difficult to add or increase a requirementonce an “invitation for bid” has been published andthe prospective suppliers’ responses have beenreceived.

As a general rule , competitive bidding isemployed by the transit authorities; and, in mostcases, competitive bidding is required by State lawor local ordinance. Usua l l y , howeve r , t heauthorities reserve the right to reject all bids andhave a new solicitation. This study has disclosed noinstances where a sole-source solicitation had beenemployed.

The established transit agencies select an ATCequipment contractor from previously qualifiedsuppliers on the basis of the lowest price. Otheragencies employ a single-step process where tech-nical capability and cost are weighed together.WMATA was unique in that they used a two-stepprocess in which the responsiveness of prospectivecontractors’ proposals to the procurement specifica-tions in a prebid solicitation was used to make aselection of qualified bidders. Subsequent selectionof the winning contractor from the two qualifiedbidders was based solely on cost.

To date, cost estimates and award to the low bid-der have been based solely upon the capital costs ofsystem development and construction. Life-cyclecosting, which would require cost competitionbased upon both the capital and operating costs, isan alternative costing method that has not been

used but may find increasing favor as energy andeconomic conditions cause a shift in values.

Once a contract has been awarded, data onprogram status and control over program directionavailable to the transit authority management arelimited to that specified by the contract. Therefore,it is important that the contract provide the meansfor monitoring the contractor’s progress and for ex-erting some directive control over contractor ac-tivities.

Management control is achieved in many waysranging from a resident engineer at the contractor’splant to formal design status reviews, RMA predic-tions, progress reports, and other such techniques.Traditionally, management control of an ATCsystem contract has been achieved by assigning sig-nal engineers from the authority’s staff the task ofmonitoring the work of the ATC contractor. Theseengineers are expected to know the status of thecontractor’s program at all times throughout thecontract, and, in particular, to be aware of anyproblems and the work being done to solve them.They also direct contractor progress by exercisingapproval of designs proposed by the contractor,

Maintaining management control has become in-creasingly difficult as ATC systems have grownmore complex. BART, PAAC, and WMATA ATCsystem procurement specifications included provi-sions for system assurance programs, periodicdesign reviews, and other modern managementtechniques. Several transit authorities expect to hireseparate consultants to plan, specify, and monitorthe system assurance programs for their ATC pro-

14772-683 0- 76 - 11

— .

curements. These consultants will report directly tothe transit authority technical staff.

One important method for achieving manage-ment control is independent review of the ATCmanufacturer’s design. This review may be con-ducted by the transit authority engineering staff orby engineering consultants. The manufacturer is re-quired to correct all the deficiencies identified.Besides providing an independent evaluation of themanufacturer’s design, this procedure also educatesthe reviewer on the details of the design. This par-ticularly is important in new systems where thestaff may not have lengthy transit experience. Avariation of this approach is being used at MARTA.Periodic reviews of the MARTA train controlsystem design are being held under the auspices ofUMTA, with the DOT Transportation SystemsCenter serving as a technical consultant.

Established transit properties such as CTA andNYCTA have traditionally required the manufac-turer to continue to correct equipment deficienciesuntil the equipment performance is acceptable tothe chief engineer. Management control by theseauthorities succeeds, in part, because of the limitedmarket for ATC equipment. If an ATC equipmentmanufacturer wishes to remain in business, he mustnecessarily satisfy his customers, and these two arethe largest in the country. The major change inmethods of management control for the new ATCsystem procurements is the introduction of require-ments for detailed program planning by the contrac-tor. The increased management involvement per-mits control action to be taken immediately when adeviation from the program plan is noted. Thismakes it possible for management to avoid potentialproblems rather than waiting until they occur andrequire drastic action to correct.

Upon completion of the manufacturing process,the ATC equipment is delivered to the transitauthority, installed, and tested. Test procedures aredescribed in the next issue.

ISSUE D-8: TESTING

How are ATC systems tested? What kinds oftests are conducted, for what purposes, andwhen in the development cycle?

There are three categories of ATC systemtesting,system

148

each beginning ata different stage in thelife cycle and satisfying different needs.

Engineering testing occurs early in the develop-ment cycle and provides data for detailed systemdesign and modification Assurance testing is per-formed to evaluate how well the equipmentmeets procurement specifications. Acceptancetestiing is performed when the whole ATCsystem has been installed and debugged and maybe performed on significant subsystems beforetheir ‘integration into the total system. Accept-ance testing is the final demonstration that thesystem meets specification. There is room for im-provement in several areas-test planning, docu-mentation, and dissemination of results.

Testing serves a number of important functionsin the development process. It provides the datanecessary to support ATC design. It serves to iden-tify actual or potential problems during manufac-ture and installation. It is the means to verify thatthe resulting system meets specified requirements.

There are three basic types of testing: (1)engineering testing, (z) assurance testing, and (3)acceptance testing. Each is initiated at differenttimes in the system life cycle, and each satisfiesdifferent needs, but they are not mutually ex-clusive. They frequently overlap in time, and dataobtained in one type of testing may be useful for thepurposes of another. Although all three types oftests are initiated prior to opening of the system,they may extend well into the period of revenueservice.

The results of testing are of primary interest tothe transit agency installing or modifying an ATCsystem and to its system contractors. The testresults may also be of value to other authoritieswho are planning a similar system. Careful plan-ning of tests, description of test procedures, anddocumentation of results is essential to maximizethe value of testing.

Of particular interest for this report is the ade-quacy of the testing process in terms of planning,procedures, and documentation of results. Also ofinterest is responsibility for testing and evaluationof test results. Finally, the degree to which testresults for one system are utilized at others plan-ning similar systems deserves exploration.

Engineering Testing

Engineering testing begins early in systemdevelopment and includes tests of components and

subsystems to verify that they perform as expected.There are also tests undertaken to diagnose thecause of a problem and assist in its solution. Thissecond category of tests is called “debugging.”

Engineering tests are generally performed by theATC system contractor to support equipmentdesign and manufacture. Results are not alwaysdocumented and are generally not submitted for-mally to the transit authority, A representative ofthe authority may be in residence at the supplier’splant and may monitor engineering test results.NYCTA, for example, follows this procedure.

Because engineering testing occurs early insystem development and there is higher order test-ing later on, it is probably not necessary to havemore formal documentation and wider dissemina-tion of engineering test results than is presently thecustom. Furthermore, manufacturers frequentlyconsider the results of these tests to be proprietary.

Assurance Testing

Assurance test ing includes inspect ion andquality control during production and tests to en-sure that the equipment meets procurementspecifications.

In general, the procurement specifications in-clude provisions for the quality control program.Unfortunately, quality control programs are not al-ways adequate, For example, the BART ATCsystem procurement specifications provided forsuch a program, but strong and effective qualitycontrol was really not achieved. An effectivequality control test program must include not only agood inspection and test program but managementprocedures to follow up and correct deficiencies.

Besides quality control, tests are conducted todemonstrate that equipment meets specificationsfor performance, safety, reliability, maintainability,and availability. Such tests are performed on in-dividual components at the factory or as they are in-stalled, then on subsystems, and eventually on thewhole ATC system. Failure of the equipment toperform according to specifications leads to diag-nostic testing to isolate faults and correct them—another type of debugging. Ideally, these testswould be completed and all deficiencies correctedbefore revenue operation. However, the length oftime required for some kinds of assurance tests(notably rel iabi l i ty) and pressures to beginpassenger service often dictate that operations start

before the tests are completed. Some transitauthorities recognize this necessity by indicating inthe procurement specifications those assurancetests that must be completed before revenue serviceand those that will be accomplished during revenueoperation.

It is important to note that statistically significanttests to demonstrate ATP safety probably cannot beconducted, The required levels of safety are so highthat a valid quantitative test for safety would takeyears or even decades to complete, even if acceler-ated testing methods were employed. As a result,assurance of ATC safety is accomplished by a com-bination of analysis and testing. The analyticalwork is done to identify possible design or engineer-ing defects that could produce an unsafe condition.Testing then concentrates on these areas. While itmay not be able to produce statistically significantresults, test data of this sort can lend credibility toengineering judgments made about safety.

Acceptance Testing

Acceptance testing is the final set of tests on thecompleted ATC system to demonstrate that thesystem meets all procurement specifications, Ac-ceptance testing is specified in detail as part of theATC system contract and usually consists of an’ in-tegrated series of tests which take place overmonths or years. Acceptance testing tends to con-centrate first on safety features, then on perform-ance, and finally reliability and maintainability,Formal tests of the personnel subsystem and man-machine integration are seldom, if ever, conducted.Problems in this area are detected and corrected asthey arise in the course of other testing or opera-tions. The ATC system is accepted by the procuringagency when i t has been demonstrated thatspecification requirements and contractual accept-ance provisions in the contract have been met.

The planning, conduct, and communication oftest results are basic to all three categories of test-ing, The adequacy of documentation of plans, testprocedures, and results was reviewed during thisstudy in order to evaluate the testing process. Thegeneral conclusion is that documentation of testplans has been less than adequate,

From interviews with representatives of transitsystems now being planned, and from examinationof procurement specifications, it is apparent thatthere will be increased emphasis on formal docu-mentation of test plans in the future, For example,

149

the PAAC ATC system procurement specificationsrequire the contractor to prepare and submitvarious test plans appropriate to the differentcategories of testing. As another example, MTARapid Transit Development Division (Baltimore)expects to hire a reliability, maintainability, andsystem safety consultant who will be required toplan a comprehensive and integrated program forthe entire transit system, including the varioussystem assurance tests pertinent to RMA andsafety. This consultant will work with both thegeneral engineering consultant and the ATC systemdesign subcontractor.

Confidence in test results is determined to a largedegree by the detail to which testing procedures aredocumented. Careful attention to details such as ac-curacy, precision of measurement, and control ofthe test environment is important. In some cases, itis difficult to assess the quality of testing that hasbeen conducted in existing transit systems becausedocumentation is lacking or inadequate.

For testing to be of maximum value, the resultsmust be communicated to interested parties. Withina single organization, this may be accomplished in-formally by oral report or internal memoranda.However, in an integrated test program, more for-mal reporting procedures are necessary to assurethat the test results are properly disseminated. As intest planning and performance, there is room forimprovement in the dissemination of results, par-ticularly outside of the transit agency.

R&D may be defined as discovery of newknowledge and its development for use in practicalapplication. R&D must be distinguished from ap-plications engineering which refers to the solutionof specific technical problems. With this distinction

in mind, the following summarizes the organiza-tions which might be expected to perform R&D inATC and their involvement in such activity.

R&D Programs

Operating transit agencies perform very littleATC R&D. Fiscal realities of the operating environ-ment do not support such activity. Operating agen-cies do conduct ATC applications engineering.

Agencies planning new rail rapid systems andtheir subcontractors perform R&D in the course ofsystem development--chiefly design and develop-ment of new hardware, test track demonstrations ofnew concepts, and basic analytical work. Fundsmay be provided for such purposes by the FederalGovernment as part of technical study programsand capital grants. Transit agencies sometimes usetheir own funds to support such work.

The American Public Transi t Associat ion(APTA) is the principal rail rapid transit industryassociation. Some of its committees are active inareas related to ATC, principally safety andreliability. Such work is paper-and-pencil studiesand is supported by member organizations. TheTransit Development Corporation is an industry-organized R & D corporat ion. No programsspecifically related to ATC have been undertaken,

Some R&D in ATC reliability and small vehiclesystems is done by manufacturers. This work issupported primarily by private investment, Therehas been some private investment in test trackdemonstration programs. (See Issue D-10, p. 151.)Most industry work in ATC for rail rapid transit isapplications engineering.

Educational research organizations, such as theUniversity of Minnesota, Northwestern University,Aerospace Corporation, and Applied PhysicsLaboratory, have funded contributions to thel i terature for small-vehicle, f ixed-guidewaysystems. They have not made substantial privatecontributions to rail rapid transit R&D for ATC.

The Federal Government is the principal sourceof R&D funds. Major Federal support to assist test-ing and demonstration of ATC equipment for con-ventional rail rapid systems was given in themid-1960’s in conjunction with the BART andTransit Expressway test tracks. (See Issue D-10, p.151.)

Recent Federal programs have generally beenassociated with support of major vehicle or system

150

concept development rather than ATC as such.These programs include the State-of-the-Art Car(SOAC), the current Advanced Concept Train(ACT I), the TRANSPO ’72 demonstrations, theStandard Light Rail Vehicle, PRT activities atMorgantown, West Virginia, and the now-canceledDual-Mode Program,

In small vehicle technology, a new projectdirected toward the development of a high perform-ance PRT (HPPRT) system has major ATC ele-ments. Also, the Applied Physics Laboratory (APL)of Johns Hopkins University has been providingmore or less continuous support to UMTA in PRTtechnology. Most of the APL work has focused onanalytical studies of operational and reliabilityproblems associated with PRT systems. APL hasalso provided general technical support to UMTA,notably as a technical monitor (with MITRE) of theTRANSPO ’72 PRT demonstrations.

Recent and current work in system assurance hasbeen closely allied with ATC technology and thequestion of manned versus unmanned vehicles. AnUMTA-funded ongoing program in these areas isbeing conducted by the Transportation SystemsCenter (TSC). One product of this work was areport entitled “Safety and Automatic Train Con-trol for Rail Rapid Transit Systems, ” published inJuly 1974. It is expected that the results of the TSCinvestigation of system assurance and the questionof manned/unmanned systems will be available in1976.

Except for the APL work, there has been littlesupport for the development of analytical toolsneeded to evaluate ATC (and other) problemsassociated with advanced technology systems. Thissituation now appears to be changing. A part of thenow-canceled Dual-Mode project was to have in-volved development of the analytical tools neces-sary to evaluate such general concerns as opera-tional strategies and reliability. Such a requirementis included in the later phases of the recently initi-ated HPPRT program.

There are indications that a more programmaticapproach to ATC technology for small vehicles willbe initiated. UMTA is currently developing anAutomated Guideway Technology (AGT) programwhich will deal with many system and subsystemproblems on a generic rather than project-specificbasis. If there are any significant contributions torail rapid transit system of these programs, they arelikely to fall in the area of the development of

methodology and analytical tools. Equipment re-quirements for AGT and rail rapid transit are sodifferent that contributions to rail rapid transithardware technology are unlikely. However, betteranalytical tools would be an important contribution.

Application of R&D

The application of the results of R&D varies ac-cording to the sponsoring organization, Privatelysupported R&D, such as is done by manufacturers,is generally proprietary and not fully available tothe industry. Unfortunately, this is where most ofthe expertise resides,

The results of federally supported research andthat conducted by educational institutions generallyfinds its way into the literature, Much of this workis more theoretical then practical in outlook.Further, such work is often concerned with auto-mated small-vehicle technology rather than moreconventional rapid transit. The increasing involve-ment of the Federal Government in rail rapid transitmay change the situation.

Transi t agencies planning new systems ormodifying old ones generally exchange informa-tion, on a personal basis, with their counterparts atother transit agencies, This helps to compensate forthe lack of research literature and the withholdingof proprietary data held by manufacturers.

ISSUE D-10: TEST TRACKS

What role do test tracks play in ATC R&D?Who operates and funds test tracks?

Test tracks are not built solely for ATC studiesbut to serve several objectives, and their valueshould be judged accordingly. For developmentof ATC, test tracks are used for R&D, demonstra-tion of conceptual feasibility, and hardware testand evaluation. By permitting scientific andengineering work in the absence of constraintsimposed by revenue service, test tracks are vitalto advances in transit technology. Some testtracks have short life spans. Others are more orless perrnanent facilities. They are operated andfunded by the transit agencies, manufacturers,and the Federal Government.

As used here, a test track is a facility built ex-pressly for the purpose of engineering and scientific

151

studies, and not revenue trackage that may be usedfor test purposes. Thus, the Morgantown project isnot a test track. The TRANSPO ’72 exhibition,while perhaps better classed as a demonstration, isincluded because of the post-TRANSPO testprogram. Test track programs discussed below arecategorized by the three types of organizationswhich operate them: transit agencies, manufac-turers. and the Federal Government.

Transit Agencies

BART Diablo Test Track.—The purpose of thistrack was to demonstrate the conceptual feasibilityof alternative subsystems for BART—not, as com-monly thought, to select hardware to be procured.The results of the program were used as a basis forwriting functional specifications for BART equip-ment.

The 41/2-mile test track was located betweenConcord and Walnut Creek, California. It was oper-ated in the mid- to late- 1960’s, at a total programcost of about $12 million. The Federal Governmentsupplied about two-thirds of the funds, and BARTthe remainder. Most suppliers participating in theprogram are believed to have invested substantialfunds of their own.

ATC was 1 of 11 different system elementsstudied at the track. Because the purpose was con-cept demonstration using prototype hardware,reliability and maintainability studies were not partof the ATC test program. Four ATC systems weredemonstrated, Suppliers were General Electric,General Railway Signal, Westinghouse Air Brake,and Westinghouse Electric.84 The results of the for-mal tests were that all four systems met the generalrequirements for BART ATC, with no one systemsignificantly better.

After final ATC specifications were prepared byBART, the winning contractor, WestinghouseElectric, was selected on the basis of low bid.Because the WELCO system was developed inresponse to new specifications and designed to beprice-competitive, it is not surprising that it differedfrom any demonstrated. This system was not subse-quently tested on the Diablo track before finalsystemwide installation. Whether such testingwould have avoided some of the la ter ATCproblems encountered in BART depends upon the

BqThe Philco Corporation also tested portions of an ATCsystem later, after the completion of the formal test program.

type of tests which might have been performed andthe criticality of the analysis of results, rather thanthe particular track used.

PAAC Transit Expressway Program Transit Ex-pressway.—This program, conducted by the PortAuthority of Allegheny County, ran from June 1963to November 1971 at South Park, 11 miles fromdowntown Pittsburgh. The objective was to designand develop a new technology—namely a fullyautomated system of medium-size, light weight,self-propelled vehicles which could be operatedsingly or in trains of 10 or more vehicles. The workwas done in two phases at a cost of $7.4 million.Two-thirds of the funds were provided by theFederal Government; and the remainder was pro-vided by Allegheny County, the State of Penn-sylvania, and Westinghouse Electric.

As the first fully automated transit system, sig-nificant development work was done on ATC. TheATC system underwent major changes between thefirst and second phases of the program. The finalsystem is comparable to BART, with the exceptionof the t ra in detect ion equipment which wasspecifically designed to detect the rubber-tiredvehicles planned for the system.

The importance and value of this program lies inthe many innovations demonstrated there and lateri n c o r p o r a t e d i n t o s y s t e m s n o w o p e r a t i o n a lelsewhere. The ATC technology has been used byWestinghouse Electric for the Seattle-Tacoma andTampa airport systems, for BART, and for the SaoPaulo METRO in Brazil. PAAC used the project todevelop procurement specifications for TERL, aprogram recently defeated by the voters.

Manufacturers’ Test Tracks

Manufacturers’ test tracks have been built pri-mari ly for work on automated small-vehiclesystems. These tracks are used either to developnew systems, to check equipment prior to delivery,or both. Federal funds may be used, as was the caseof the Dashaveyor and Varo test tracks which wereused for feasibility studies conducted by these com-panies for AIRTRANS at the Dallas-Fort Worth air-port. Some company test tracks that have been usedfor ATC development or checkout are:

● Dashaveyor, Pomona, Calif.

● Varo Monocab, Garland, Tex.

● WABCO Monorail Division, Cape May, N.J.

152

. TTD, Denver, Colo.

● Bendix, Ann Arbor, Mich.

● Alden, Milford, Mass.

Federal Government

TRANSPO ‘72.—Four automated small-vehiclesystems were demonstrated at TRANSPO ’72 andlater evaluated in a test program conducted be-tween August and November 1972. Federal fundsamounting to about $7 million were provided forthe demonstration and test program. There was alsosubstantial private investment. The exact amount isunknown, but it is thought to be of the same orderas the Federal contribution. The systems demon-strated and their manufacturers were:

. Dashaveyor System—Bendix Corporation

. ACT System —Ford Motor Company

● Monocab System—Rohr Industries

. TTI System-Otis

The systems were developed under tight timeconstraints with limited funds. This led to somecompromises in the ATC system design. The post-TRANSPO test program showed that some of theATC equipment had undesirable control charac-teristics, including long delay times and speedoscillation. It was concluded that the basic cause ofthese problems was the prototype nature of theequipment.

Apart from its value as a public demonstration ofn e w t e c h n o l o g y , t he ma jo r bene f i t o f t heT R A N S P O ’72 program was the increasedcapability in small-vehicle technology gained by thefour participating manufacturers, Because of basicdifferences in philosophy and operating charac-teristics between automated small-vehicle systemsand rail rapid transit and because of the lessstringent demands placed on a system in an exhibi-tion (in comparison to a revenue operation), theTRANSPO ’72 program had limited value in im-proving ATC systems for general transit industryapplication.

Pueblo Colorado Test Facility.—DOT’s HighSpeed Ground Transportation Center at Pueblo,Colo,, became operational in 1973, Managed by theFRA, the Center can test several types of groundtransportation systems. Both advanced systems andrail technology programs are conducted, Theformer programs include the Tracked Levitated

Research Vehicle (TLRV), the Tracked Air CushionResearch Vehicle (TACRV), and the Linear Induc-tion Motor Research Vehicle (LIMRV), For railtechnology programs, the Center includes 20 milesof conventional railroad trackage, used for studyingtrain dynamics under a variety of track and gradeconfigurations, a 9.1-mile oval rail transit trackwith a third rail for testing electrically powered roll-ing stock, and a Rail Dynamics Laboratory forsimulator testing of full-scale railroad and rail tran-sit vehicles. As a part of the now-canceled Dual-Mode Program, it was planned to build two guide-way loops at the site, each 2 miles in circumference.

Probably the most significant rail transit activityat Pueblo was the testing of the State-of-the-ArtCar (SOAC) in 1973. There was little ATC relatedwork associated with this R&D activity, and theATC provisions at Pueblo are all but nonexistent.There are several reasons for this. DOT has beenusing the facility for other purposes. Limitedfacilities are available. (For example, there are noprovisions for inserting signals into the rails,) Thesite is very remote from both operating propertiesand equipment suppliers. Most transit agencies feelit is essential to conduct final ATC developmentwork in the actual operating environment (at-mospheric, electrical, etc.) where the equipmentwill be run. Unless there are specific federallyfunded programs requiring that the work be con-ducted at Pueblo, it seems unlikely that significantamounts of ATC research for rail rapid transit willbe conducted there.

The MITRE Corporation (1971) conducted asurvey of rail rapid transit agencies and equipmentmanufacturers to identify problems that should beaddressed in a federally funded research program.Of the 11 top priority areas indicated by this survey,none had any direct relationship to ATC. Theresults must be accepted with some caution because

153

.: .’ - t

FIGURE 72 DOT Test Track, Pueblo, Colorado

154

none of the industrial firms surveyed were ATCequipment manufacturers and because the intent ofthe study was to identify problems for investigationat the DOT Pueblo test site. (As indicated earlier inIssue D-10, p. 151, the Pueblo test facility is notsuited for investigation of ATC problems.) Still, thesurvey does suggest that ATC is not viewed as amajor R&D problem by a significant part of the tran-sit industry.

During visits to transit agencies made by BattelleColumbus Laboratories as technical consultants inthis assessment, comments and suggestions weresolicited on R&D needs in rail rapid transit tech-nology, particularly those associated with ATC.Here again, the results indicate that ATC is clearlynot a major concern.

Operating transit agencies felt that the majorR&D needs were:

● Improvement of chopper control, multiplexingof train lines,85 and a.c. traction motors;

● Documentation of slip-slide tests for use inofficial and expert testimony in damage andinjury suits;

● Clarification of the trade-off values associatedwith such technical matters as analog vs.digital signals, control signal frequencies andmodulation rates, types of station stops, chop-pers vs. cam controllers, and the use of p-wire;

● Review of the availability and allocation ofradio frequencies for both voice and datatransmission by transit systems;

● D e v e l o p m e n t o f a d a t a b a s e a n dclearinghouse for rel iabi l i ty and main-tainability information for the benefit of tran-sit systems and manufacturers.

Transit systems in the planning and constructionstages had a differing set of priorities:

● Investigation of electromagnetic interferenceproblems;

● Improvement in the rel iabi l i ty of ATCsystems and related equipment;

. Study of techniques for, and the value of,regenerative braking;

. Establishment of a data bank on the safety,

●

●

reliability, and maintainability experience ofoperating transit systems;

Maintenance training programs to ensure thatnew and sophisticated transit equipment (in-cluding but not limited to ATC) can be pro-perly cared for;

Studies of collisions and crash resistance, par-ticularly for small-vehicle systems.

Since one of the main purposes of this tech-nology assessment was to weigh the need for R&Din the area of automatic train control, this topic wasgiven special attention. In addition to review of theliterature and collection of opinion within the tran-sit industry through the interviews cited above, thematter of research needs and priorities was madethe subject of a separate investigation by the OTATransportation Program staff and the OTA UrbanMass Transit Advisory Panel. This investigationdrew especially on the experience of individualpanel members and of various transit systemmanagers, equipment manufacturers, technicalconsultants, and DOT officials. The findings of thisinvestigation, as they apply to rail rapid transit, arepresented below.86

At the outset, it should be noted that there is noneed for a significant R&D effort to make major ad-vances or innovations in ATC technology for railrapid transit systems. The basic technology is suffi-ciently developed for present and near-term futurepurposes. What is needed now is research anddevelopment to refine the existing technology andto improve performance at reduced cost. The majorelements of such a program are discussed below.Figure 73 is a matrix, categorizing the importance ofthese R&D efforts against the estimated relativecost to carry them out.

Reliability and Maintainability

There are several aspects of reliability and main-tainability in which further work is needed.

Equipment Reliability and Maintainability

There is a major need to develop more reliableand maintainable equipment. This applies notjust to ATC but other types of rail rapid transitequipment.

aSThe underlined items are those directly or indirectly re-lated to ATC.

aeR&D needs for automated small-vehicle systems are ex-plored in a separate OTA report, Automated Guideway Transit:An Assessment of PRT and Other New Systems, June 1975.

155

RESEARCH AREAS

RMA Analytic Techniques

Equipment RMA

RMA Data Bank

RMA Standards

Safety Methodology

Technology Transfer

Handicapped Requirements

Standardization

●

FIGURE 73. ATC Research and Development Priorities and Relative Cost

Techniques for RMA Analysis

Improved and more quantitative methods areneeded to evaluate total system performance interms of rel iabil i ty, maintainabil i ty, andavailability. Component performance measuresexist. Total system performance measures donot. Total system measures would permit betterallocation of reliability requirements among sub-systems, better understandingtrade-offs, and better utilizationnance work force.

RMA Standards and Guidelines

An effort is needed to establishment standards and to clarify

of reliabilityof the mainte-

realistic equip-manufacturers’

responsibilities in the area of RMA. The stand-ards must be high enough to assure reasonableavailability of equipment but not so high as tomake the equipment unnecessarily costly.

156

Reliability and Maintainability Data

A pool of data from testing and operational ex-perience pertaining to equipment reliability andmaintainability y would be of great value to transitsystem planners, research groups, and manufac-turers. At present, there is no uniform way ofrecording and reporting such information, and noclearinghouse for collecting and disseminating itwithin the transit industry.

Safety

The safety levels of the rail rapid transit industryare high and exceed nearly all other forms of publicand private transportation. Still, there is a need forresearch in two aspects of safety.

Train Detection

The much publicized train detection problems ofBART (which are probably no more severe than

those experienced in other transit systems) haveunderscored the need for clarification of thestandard for train detection and the need for auniform method to test the performance of traindetection systems.

Safety Methodology

Controversy over system safety versus fail-safeprinciples abounds in the transit industry, Thereis also debate over how safety is to be measuredand how safe is safe enough. Research is neededto develop an objective and quantified methodfor evaluating the safety aspects of rail rapidtransit system performance.

Man-Machine Relationships

Function Allocation

There is great variability among transit systemsin the duties assigned to the human operator. Sig-nificant errors were made in the original designof the BART system because of the highlypassive role assigned to the train attendant. Theman-machine interface needs to be carefullystudied to determine the optimum role of thehuman operator in automated systems and to en-sure that provision is made for the operator to in-teract effectively with the system in abnormal oremergency situations. The role of personnelassigned in a supervisory capacity needs to besimilarly examined.

Cost-Benefit of Automation

Research is needed to determine the relative ad-vantages of manual and automated methods ofoperat ion with respect to energy savings,variability of trip time, equipment utilization,system capacity, and manpower costs. Such datawould be of value not only in the design of newsystems but also in the .modernization of oldones.

Application of Technology

Even though ATC is a rather mature and welldeveloped technology, there remain some problemsof practical application. Three areas are in need ofspecial attention.

Standardization

There are a number of technical and economicbenefits to be gained from reducing the diversityof ATC equipment now in use or planned for in-stallation in rail rapid systems. These advantagesmust be scrutinized and evaluated against thedisadvantages of inhibiting innovation and im-peding improvement that standardization mightbring.

Technology Transfer Within the Industry

There is a general shortage of persons with ex-perience in ATC system design, manufacture,and operation at all levels in the industry. Thisshortage is most keenly felt by agencies planningand building’ new systems. Research is needed todevise more effective methods for sharing infor-mation, exchange of experienced personnel, andtraining of new personnel.

Requirements for the Handicapped

Under the stimulus of the Federal Government,there is an increasing concern in the transit in-dustry with the transportation needs of the han-dicapped. As a part of the investigation of thegeneral social costs and benefits of providing railrapid transit service for the physically, visually,and auditorily impaired, there is a need to con-sider the specific influence of ATC. Among thematters of interest are acceleration and decelera-tion limits and their effects on system capacityand trip time, passenger assistance on trains or instations with a low level of manning, and thesafety of the handicapped and others in emergen-cy situations.

157