Research Results Digest 74

February 2006

Failures of train doors are the biggestsingle cause of delay and disruption of railtransit service; therefore, in-depth studyand analysis of train door problems canlead to significant increases in the reliabil-ity of transit service. This digest describes(1) the development of a relational data-base for data on the causes and conditionsof train door failures and (2) some prelim-inary analyses of the data.

SUMMARY

The American Public TransportationAssociation (APTA) Rolling Stock Equip-ment Technical Forum (RSETF) devel-oped the Train Door Project, whose goal is to help transit managers resolve doorequipment problems that adversely affectrail transit reliability and service. The pri-mary output from this work has been thecreation of an on-line questionnaire anddatabase detailing door failure causes andconditions.

The Train Door team consisted of vol-unteer participants from transit agencies,railcar manufacturers, door equipment man-ufacturers, and consulting companies. Theproject received strong support and partici-pation from chief executives and staff at fiveleading U.S. transit agencies: San Francisco

Bay Area Rapid Transit District (BART),the Chicago Transit Authority (CTA),New York City Transit (NYCT), The PortAuthority Trans-Hudson (PATH), and theWashington Metropolitan Area TransitAuthority (WMATA). Others have ex-pressed interest in participating.

The Train Door project plan consistedof six steps: (1) identify door population,(2) define segment of study, (3) define fieldinvestigation strategy, (4) manage collectedinformation, (5) formulate problem-solvingactions, and (6) communicate findings totransit community.

For steps 1, 2, and 3, the Train Doorteam created a four-part questionnairecovering railcar fleets, door equipment,operations, and maintenance for the fiveparticipating transit agencies. Part I of thequestionnaire covers railcar fleets in ser-vice, including car classes in service, num-ber of cars per class, years of service foreach class, average speed, annual miles ofoperation, minimum and maximum consistlength, and operating and maintenance re-sponses to door incidents. Part II coversdoor equipment and component technicalspecifications for each car class, includ-ing covered locations, types, manufactur-ers, models, and original/retrofit conditionfor major electrical, electronic, mechanical,

TRAIN DOOR SYSTEMS ANALYSISThis digest is the final report for TCRP Project J-6, Task 62, “Field DoorSurvey Project.” Appendixes to this digest are available as TCRP Web-OnlyDocument 28. This digest is based on research conducted by TransportationSystems Design, Inc.

Subject Areas: VI Public Transit, VII Rail Responsible Senior Program Officer: Dianne S. Schwager

TRANSIT COOPERATIVE RESEARCH PROGRAMSponsored by the Federal Transit Administration

and switch components. Part III includes train opera-tions, rules governing delays and reliability calcula-tions, and operating procedures. Part IV covers doormaintenance issues and practices for major door components.

Following collection, review, and public presen-tation of data from the first questionnaire, the TrainDoor team identified seven common door failuresand conditions that affect all the surveyed transitagencies: (1) door fails to open or close when com-manded from the operator location; (2) door statusinterlock failures; (3) incorrect door opening; (4) in-correct door operation (operation/wayside error:wrong side opening or open when not berthed): (5) ob-struction detection failures/drags; (6) freewheelingdoor panel; and (7) door fails to completely close andlock and/or to indicate closed and locked.

The Train Door team then created a second ques-tionnaire about component causes of the seven com-mon door failures. For each failure, the questionnairelists generic door system components that could causethe identified door failure. In addition to requestinginformation on component causes, the questionnaireasks for solutions to these common problems, includ-ing maintenance procedure changes, operationalchanges, and/or equipment design changes.

The resulting data from both questionnaires includes answers to 34 groups of questions, with 191 data items covering 21 railcar classes. The data,collected in 2004, were unique, in depth, and ofgreat value. To permit processing and broad use ofthe data, the Train Door team built a web-accessible,

MySQL database that provides a permanent locationfor the data, provides tools to enable access, ensuresresponses are in the expected format, is easily scaledto accept input from more transit agencies, and doesnot require clerical processing for new transit agencydata to be entered. The website, to be installed atwww.traindoors.com, is implemented in proven com-mercial and open-source software, including MySQL,Linux, Apache, PHP, Smarty, and XHMTL. (Thesponsors have delayed public release of the train-doors.com website and database, while they resolveissues concerning access to the Train Door data.)

The MySQL database consists of 29 tables, struc-tured to parallel the questionnaire data. The websitelets users download selected tables into Excel orother file management applications.

The website allows the user to display or exportresults from the Train Door database; enter new ormodified data for a transit agency, in response to thetwo questionnaires; participate in a discussion forum;access resources including technical papers about traindoor technology and database documentation; and getcontact information for the website and database.

The first results obtained from analysis of the in-formation in the Train Door database suggest im-provement areas and directions for further researchand analysis. Figure 1 shows the distribution of typesof door failures by car class. Further analysis is neededto understand how to reduce the number of door in-cidents, and why a failure type is quite significant atone transit agency or in one car class but less signif-icant in another.

2

Opening en Route Phantom Operation

Loss of Operator No Motion Problem

Sticking

Other

0%

20%

40%

60%

80%

100%

A1 A2 A3 A4 B1 C1 C2 D1 D2

Car Class

Co

ntr

ibu

tio

n (

%)

D3 D4 D5 D6 E1 E2 E3 E4

Rounding caused some totals to add to slightly less than or more than 100%.

Figure 1 Door failures by type.

Other initial results showed the following:

• Improved door switch reliability could have abig positive impact on door and transit service.

• Certain car classes account for a dispropor-tionate share of door failures, and a focusedeffort to address those car classes could havea big positive impact on service.

• Winter weather correlates with increased doorproblems, and reliability improvement effortsrelated to weather should focus there.

• Improved maintenance tools, techniques, train-ing, and design provisions in support of main-tenance could have a big positive impact onservice.

Users should be aware that the data in the TrainDoor database come from only five transit agencies.The possibility exists that the sample of data, whilecovering a great breadth of topics and detailed tech-nical information, may not be deep enough to pro-vide conclusive evidence of a trend or possibility.Accordingly, the Train Door database at this stageshould be considered one tool, but not the last word,in analyzing train door conditions and failures.

Next steps for the Train Door database and train-doors.com are to determine how to handle sensitivedata such as manufacturer details; to use the on-linequestionnaire to guide collection of data in uniformformat; to expand the database to cover maintenanceand other performance aspects; to solicit and encour-age participation from other transit agencies; and tocontinue to expand the data analysis to understandthe trends shown in the data and investigate prob-lems and possible solutions within and among carclasses and transit agencies.

Next steps for the transit industry are to evaluatevalue and utility of traindoors.com and determinewhether the concept should be extended to collectsimilar or expanded data on other equipment relia-bility and performance characteristics.

1 INTRODUCTION

Starting in 2002, the APTA RSETF assessedrolling stock problems that rail transit professionalsconsidered critical to their operations with the intentto develop a project to help transit managers resolveone set of those problems.

Transit industry managers from heavy rail, com-muter, and light rail transit agencies identified criticalproblems involving railcar body structures, doors,propulsion and dynamic braking, air braking, trucks

and suspension, wheels and axles, heating ventilatingand air conditioning, couplers and draft gears, com-munications, lighting, and train control. Door systemfailures stood out because transit managers identifiedthem as having the greatest negative impact on railcarreliability and transit service, particularly on heavyrail rapid transit systems. As a result, the RSETFselected train door system failures as the recom-mended project focus.

This digest reports on the results of the RSETFTrain Door Project and its accomplishments to date.

2 DEVELOPMENT PROCESS

2.1 Five Participating Transit Agencies

The RSETF proposed to focus on train doors in ameeting with the CEOs of leading U.S. rail transit sys-tems at the 2003 APTA Rail Transit Conference. Thechief executives endorsed and supported the proposal,and they committed key technical and project staff tosupport and help implement the Train Door Project.As a result of this high-level support, the RSETF TrainDoor Project team had enthusiastic participation andcooperation from the transit agencies and their techni-cal and operating staff and from door manufacturers,railcar builders, and transit industry consultants.

To control the project scope, the team selectedfive heavy rail transit agencies as the first group foranalysis and action: BART, CTA, NYCT, PATH,and WMATA. These agencies were chosen becauseof their readiness to participate in the project and be-cause they collectively operate a wide range of rail-cars, door equipment, and service levels in diverseregions and climates. The team understood that thepilot assessment would entail a learning curve, andthe limited volunteer resources of the Train Doorteam were matched to the limited initial data sampleprovided by the five transit agencies.

2.2 Project Plan

The Train Door Project plan consisted of the fol-lowing steps:

1. Identify door population: equipment suppliers,users, and car fleets

2. Define segment of study: door system compo-nents, technology, and life cycle phases to beinvestigated

3. Define field investigation strategy: create aquestionnaire, perform interviews, and docu-ment findings

3

4. Manage collected information: create a data-base and analyze data

5. Formulate problem-solving actions: investi-gate problems, plan corrective actions, anddocument lessons learned

6. Communicate findings to transit community

An overview of the process is provided in Appen-dix D.

2.3 First Questionnaire: Fleets, Equipment,Operations, and Maintenance

The Train Door team created a questionnaire thataddressed four general information categories re-lated to doors: railcar fleets, door equipment, opera-tions, and maintenance.

2.3.1 Part I: Fleet Survey

Part I of the questionnaire covers the particularsof railcar fleets in service at each of the five partici-pating transit agencies:

• The car classes in service• Number of cars per class• Years of service for each class• Average speed• Annual miles of operation• Minimum and maximum consist length

Questions also cover the operating and mainte-nance responses to door incidents. To provide a visualperspective of door opening and door operator loca-tions, the questionnaire requests a door schematicfor each car class, showing plan and elevation viewsof the car, side door openings (two, three, or four),and door operator locations (under-seat, wall pocket,or overhead).

The five participating transit agencies operate atotal of 32 car classes. Responses were provided for21 of the car classes.

2.3.2 Part II: Door Equipment Survey

Part II covers door equipment and componenttechnical specifications for each car class. The surveycovers locations, types, manufacturers, models, andoriginal/retrofit condition for each of the following:

• Door operators and master door controllers,relays, cams, and micro switches

• Wiring• Mechanical linkages

• Door panels, sensitive edges, hangers, thresh-old plates, and bottom door-panel guides

• Microprocessor/electronics equipment (at doorlevel and at car level)

• Inter-car communications, train line wiring,electric couplers, and electric portions

2.3.3 Part III: Operations

Part III covers train operations, applicable rules,and standard operating procedures, including thefollowing:

• The definition and calculation of train delays• Railcar performance reliability• Basis for calculating mean distance between

failures (MDBF) or mean time between fail-ures (MTBF)

This part studies the distribution of door failuresin each car class. The questions on operational fail-ures experienced, factors affecting satisfactory op-erations and reliability, and incidents leading to pas-senger injuries are rated on a percentile basis foreach, totaling 100%.

2.3.4 Part IV: Maintenance

Part IV covers door maintenance issues and prac-tices for major door components: master door con-trollers, door operators, mechanical linkages, doorpanels, door-panel sensitive edges, door hangers,micro switches, relays, microprocessors, electronicsequipment, wiring, threshold plates and door guides,and coupler electric portions and pins. Questions coverthe following:

• Repair reporting method• Preventive maintenance (PM) intervals• Average time spent on door equipment during

each PM• Percentage that in-car system components

contribute to door incidents• Most common types of failure associated with

each door component• Percentage that trainline components contrib-

ute to door incidents• Details and elements of any car body/door

component interfacing problems that contrib-ute to incidents

2.4 Getting Industry Data

In early 2004, the Train Door team began col-lecting data. With the full support of the transit

4

agency CEOs, a volunteer group visited each of thetransit agencies and conducted interviews to com-plete the questionnaires, working closely with thekey transit agency staff in operations, door mainte-nance, and engineering with door responsibilities.

The tool used for data collection was a wordprocessor–generated questionnaire, and the formatfor data collection was entry into the word proces-sor file of the questionnaire, or hand-entry onto aprinted copy of the questionnaire.

With the data in hand, the Train Door team beganto review and assess the data, and develop first results.The data showed the current state of door systemequipment at the selected transit agencies and pointedto problem areas related to specific equipment.

During the analysis, the Train Door team realizedthat the paper format was limited as a medium to hold,distribute, manage, and analyze the data and con-cluded a better tool was needed to enable thoroughanalysis of the data. Such a tool would help the teamunderstand specific door problem causes, pinpointcritical areas where operational mishaps and equip-ment failures adversely affect door system perfor-mance, and learn how some problems were corrected.

2.5 Communicating with the Industry

At the June 2004 APTA Rail Transit Conference,the Train Door team presented “Hold That Door,” atechnical session on its work to date and its plans tomove forward to the transit industry. The sessioncovered the project purpose, goals, the five partici-pating transit agencies, and the project plan. The pan-elists represented a transit operator, car manufacturer,door system manufacturer, and industry consultant.

The session accomplished three things:

• Garnered interest within the industry. Therewas a lively question and answer period afterpresentations were completed. Executivesfrom several heavy rail, light rail, and com-muter rail transit agencies expressed an inter-est in becoming involved in the Train DoorProject.

• Developed the future focus for the project. Thequestion and answer session provided the TrainDoor team with ideas about door problems andareas where future focus was required.

• Initiated the involvement of TCRP. Followingthe session, the RSETF and the Train Doorteam developed the work plan for TCRP Proj-ect J-6, Task 62, “Field Door Survey Project,”

which enabled the project to develop a data-base for the collected train door information.

2.6 Second Questionnaire: Common DoorFailures and Component Causes

Following review of data from the first ques-tionnaire and the “Hold That Door” session, theTrain Door team identified a set of seven commondoor failures and conditions that affect all the sur-veyed transit agencies and contribute to delays andother door problems. The team realized that furtherinformation was needed to expand the understand-ing of the causes of these door failures. As a result,the team created a second questionnaire. The secondquestionnaire asks about the component causes ofthe following seven common door failures:

• Door fails to open or close when com-manded from the operator location. Thequestionnaire provides 14 choices for itemscausing the failure, of which 11 are doorcomponents.

• Door status interlock failures. This seriousin-service failure is particularly troubling be-cause door status interlocks are designed toprotect train passengers against other hazard-ous door open failures or conditions. Becauseinterlock functions are sensitive to operatingpractices and driver actions, operator error wasincluded as a possible cause.

• Incorrect door opening. This door systemfailure is safety related and potentially dan-gerous. In this case, a single door opens with-out command because of a failure in the dooroperator mechanism. This is a safety problem,since one of the most important safety func-tions of train doors is to stay closed and lockedexcept when it is safe to open. As with inter-lock failures, preventive designs to address in-correct door openings have also gradually in-creased the complexity of door systems andpossibly reduced reliability.

• Incorrect door operation (operation/waysideerror: wrong side opening or open when notberthed). In this case, all doors on one side ofthe train open on the wrong side or when thetrain is not safely stopped at a platform. Thiscondition generally relates to operator, train-level, or wayside error. For passenger safety,doors must open on the correct side at eachstation, which poses real design and operational

5

challenges. Human factors, wayside controlvariations, and track and platform configura-tions constantly test both the operator’s abil-ity and the equipment’s provisions to open thecorrect doors.

• Obstruction detection failures/drags. Thesecond most important safety function of traindoors is to prevent the train from moving withan object stuck in the doors. The mechanismsand procedures to achieve this sometimes con-flict with consistently keeping doors closedand locked when the train is moving. Multipleschemes and processes have been deployed toreduce the possibility of moving a train with aperson caught in the doors. The performanceand reliability of these systems is the mainfocus of this question.

• Freewheeling door panel. A door panel thatfails to respond to commands must be manu-ally placed in the closed and locked positionand mechanically locked out. Normally asso-ciated with mechanical linkage failures or lossof power, this type of failure has begun to re-emerge. The team’s collective experienceshave identified recent occurrences of this fail-ure not caused by the traditional failure mode.The introduction of microprocessor- and soft-ware-controlled door systems has increasedthe occurrence of this failure. Latent softwaredefects and control logic failures in the ad-vanced control systems have been increas-ingly reported as the root cause of this type offailure.

• Door fails to completely close and lock and/or to indicate closed and locked. This failureis by far the most numerous and most frus-trating of the seven common door failures.Measures that prevent unsafe train operationssignificantly add to the quantity and complex-ity of door system equipment. Increased partcounts and complexity provide more opportu-nities for causing this failure. Along with in-creased system complexity comes increasedtime to identify root causes and carry out cor-rective repairs. This failure impacts both theoperation of doors and the reliable operationof the rail system as a whole.

For each of the seven door failures, the question-naire lists generic door system components that cancause or contribute to the identified door failure.Many components can contribute to more than one

of the seven common door failures. In addition tothe causes, the questionnaire asks for solutions tothese seven common failures, including maintenanceprocedure changes, operational changes, and/or equip-ment design changes.

Associating the information from the second ques-tionnaire on seven common door failures with thefirst questionnaire information on railcar fleets, doorequipment, operations, and maintenance was ex-pected to yield new insights into relationships amongfailures and the components and conditions that com-bine to create them.

3 THE DATABASE TOOL: TRAINDOORS.COM

The five participating transit agencies gave strongsupport and extensive responses to the two question-naires. The resulting data include answers to 34 groupsof questions in the categories of railcar fleets, doorequipment, train and door operations, maintenance,and common door failures. There are 191 data itemscovering 21 car classes. Paper copies of the resultsfill a 3-inch notebook. Clearly, there is a lot to learnfrom the experience of these five leading transitagencies.

The collected information was unique, in depth,and of great value, but also presented a challenge tothe Train Door team regarding how to use the in-formation and make it useful to the broader transitcommunity. The team decided a database would giveimmediate and continuing results.

TCRP Project J-6, Task 62, “Field Door SurveyProject,” enabled the team to develop and imple-ment a database for the Train Door Project infor-mation. The resulting database

• Provides a permanent location to record thecollected data;

• Provides tools to enable access to and analy-sis of the recorded data;

• Is easily accessible to the world community oftransit professionals;

• Ensures that responses to questions are in the expected format, so that answers to a sin-gle question from several transit agencies canbe directly compared;

• Is easily scaled to accept added input fromnew transit agencies as well as updates fromthe original transit agencies; and

• Does not require clerical processing of data tobe added for new transit agencies.

6

The selected structure for the Train Door databaseis MySQL, an industry-standard structured query lan-guage (SQL) relational database, hosted on a webserver and accessed via website clients. The hostwebsite, to be installed at www.traindoors.com, isimplemented in broadly accepted, proven, commer-cial and open-source software:

• MySQL, the industry-standard SQL relationaldatabase system

• Linux, the reliable open-source operating sys-tem similar to Unix

• Apache, the mature open-source web server• PHP, the broadly used open-source program-

ming language, to interconnect on-line, webbased forms to a web-based server and a data-base manager

• Smarty, an open-source PHP templating engine• XHMTL/CSS, the standard markup language

for web pages and browser-viewable infor-mation

The MySQL database currently consists of 32 tables. Of these, 28 tables hold 191 data items foreach of the 21 car classes. Each car class adds an ad-ditional 112 pieces of data to the database for thattransit agency. Three other tables contain invariantdata (constants) and one holds computed data. Thereare about 2,750 pieces of data in the database col-lected for the five transit agencies.

A user can retrieve data for a selected question ortopic. The data are presented in the same format asthe survey form, for the selected transit agency or carclass, or for all transit agencies and all car classes.

To aid users who want to explore data relation-ships within the collected data, the website lets usersdownload selected tables as a Microsoft® Excel file,which can be ported to Microsoft® Access, File-Maker®, or other commonly used desktop file man-agement systems. To make user processing easier, thedatabase tables are not normalized or structured tominimize redundant data entry; the contents are for-matted for display as numbers, dates, etc.; and foreignkeys are not used as is typical with normalized datatables. Appendix A describes the database designapproach and provides technical information on datatables, contents, and formats. Appendix B providesdesign information on the traindoors.com website.

This open-source, web-based approach chosenfor traindoors.com allows continuing extension, en-hancement, and maintenance of the database and thewebsite without expensive and proprietary software

development tools. Making database access via awebsite enabled several other important benefits:on-line entry of new data using an interactive ques-tionnaire, a technical forum bulletin board service,and a technical library are all easy to include anduseful for the transit community.

The team considered other approaches such asan Access database on a standalone PC. However,these approaches did not offer the same advantagesfor enabling broad access to data across the transitindustry and entailed levels of technical complica-tion for end users that were at cross-purposes to theTrain Door team’s intentions.

4 TRAINDOORS.COM AND THE TRAINDOOR PROJECT DATABASE

4.1 Using traindoors.com

The Train Door Project database will be hostedat www.traindoors.com. As of the publication ofthis digest, the sponsors have delayed public re-lease of the traindoors.com website and database,while they work to resolve issues concerning accessto the Train Door data. For the same reasons, datain the following figures—such as transit agencynames, car classes, door equipment manufacturernames, and part numbers—have been substitutedwith alphanumeric identifiers. This substitutionenables understanding of the scope and potentialvalue of the database, but does not publicly dis-tribute comparison data and the whole database.

The traindoors.com home page is shown in Fig-ure 2. The navigation tabs provide the followingchoices:

• Home: The page shown in Figure 2.• See Results: These pages allow the display or

export of results from the Train Door database.• Take Survey: These pages enable new or mod-

ified data for a transit agency to be entered in re-sponse to the two questionnaires.

• Discussion Forum: These pages are a set oflinked topic discussions covering train doorsand the website.

• Resources: This page links to technical pa-pers about train door technology and providestechnical documentation for the Train Doordatabase.

• About: This page gives contact informationfor the website and database.

7

4.2 Getting Train Door Data

To get data from the Train Door database, a userchooses the See Results tab on the home page,shown in Figure 3.

From here, the user can select from three tabs:

• One At A Time. This tab displays selected orcomplete results for a single transit agency.Here a user can examine complete survey re-sults by transit agency and use the data to lookat the details of the selected transit agency’sreplies to questions on door failures, failurecauses for each type of failure, narrative de-scriptions and definitions such as door MTBF,failure rate and impact, the presentation ofoperational data, details of door componentsand equipment, and fleet descriptions within atransit agency. Figure 3 shows the page forchoosing these outputs.

• Compare Data. This tab displays selected re-sults for equivalent items across all transitagencies. For example, asking for a display ofcausal environmental factors as indicated byeach transit agency will display the responsesfor the selected variable by all transit agencies

in table form. For a question about door equip-ment, the table will give the responses for eachcar class at each transit agency.

• Export Data to Excel. This tab provides datatables for further off-line analysis and process-ing by the user. The user can import the datainto a relational database and make combina-tional queries, bring the data into a spreadsheetand plot distribution histograms, or undertakeany analysis which uses the base data.

Figure 4 shows typical data output from the Com-pare Data tab, a comparison of train door sensitive-edge equipment. The output shows, for each carclass, the type of door sensitive-edge equipment, itsbasic configuration, and its retrofit status. AppendixC shows examples of other types of output.

The capability to export data to Excel for off-line analysis is an important one. The volume anddetail of data invite explorations of relationshipsamong equipment, operations, failures, climate, re-gion, and other differences. The Train Door team didnot consider limiting or presetting the combinationsthat could be analyzed to be practical or desirable.Accordingly, the Export Data to Excel tab allows aninterested user direct access to the primary data.

8

Figure 2 traindoors.com home page.

For the user who wants to probe deeper into theTrain Door database or to view it in a different form,the traindoors.com site permits the user to downloadselected tables from a drop-down list into an Excelfile. This file or files can be used for manipulation asa spreadsheet or flat file for sorting, graphing etc., orin turn be exported to Access or other databaseapplications for the conditional extraction of data.Appendix A provides the complete structure of theMySQL database tables, as well as introductory ma-terial describing the database structure.

4.3 Entering New Data

The Train Door team plans to expand the TrainDoor database by including more transit agencies.Therefore, the traindoors.com website allows easyentry of a new railcar class or an entire new transitagency and all its railcar classes.

The Take Survey tab on the home page brings theuser to a log-in screen, shown in Appendix C. Once anew user has an account established by the train-doors.com administrator, the user can access the sur-vey forms on line. Figure 5 shows a typical entry form.

9

Figure 3 traindoors.com ‘see results’ page.

The on-line data entry form has several keyadvantages:

• The questionnaire is readily available at alltimes.

• There is no added work for a clerk to translatethe answers from one form to another.

• Each entry is pre-formatted, so each new dataset will be in a data format compatible with allthe other data.

5 FIRST RESULTS

Preliminary analysis of the data collected withthe two questionnaires suggests failure areas requir-ing improvement and directions for future researchand analysis.

5.1 Component Causes of Common Door Failures

Figure 6 compiles the frequency of componentcauses of the seven common door failures studied inthe second questionnaire. The chart shows the distribu-

tion of tallies of component causes for all of the sevencommon door failures, weighted by the frequency withwhich the response was given, for all transit agencies,for all car classes, and for all types of door failures.For each of the seven door failures, the questionnaireasks the transit agency to rank the relative frequencywith which each of fourteen possible causal compo-nent items contributed to the door failure. Causalcomponent items reported as having frequency ‘1,’least often, were not tallied in the totals for Figure 6.

The seven causal components—switch/sensor, in-terlock, local door controller, threshold/bottom guide,door push button, door operator motor, and unlockmechanism—account for 55% of the total door fail-ure incidents reported by the five transit agencies. Thecomponents causing the other 45% of the reporteddoor failure incidents are trainline, car network, elec-tric coupler, door panel, short or open circuit, designproblem requiring modification, and other.

Figure 6 shows that switches/sensors were thecausal component in 12% of the common door fail-ures. Door push buttons were the causal component in6% of the common door failures. Because switches/sensors and door push buttons are switch-type de-

10

Figure 4 Sensitive-edge comparison results.

vices, Figure 6 shows that switches of some sort wereidentified as the causal component in 18% of all re-ported door system failures. The figure suggests thata focused effort to improve door switch reliabilitycould have a big positive impact on door and transitservice.

5.2 Door Failures by Railcar Class

In the first questionnaire, question 4 in Part II,Operations, deals with train delays caused by doors,reported by car class. Other parts of the questionnairedivide delays into three levels: Level 1, where thetrain operator clears the door fault (e.g., by re-cyclingdoor open/door close command); Level 2, where thetrain operator reports to central control for assistance(i.e., a service delay is reported); and Level 3, wherethe train is removed from service and passengers dis-

11

Figure 5 Train Door questionnaire data entry form.

12%

45%

6%

9%

6%

9%

6%7%

door push buttonsdoor operator motorunlock mechanismothers

switch/sensorinterlock failurelocal door controllerthreshold/bottom guide

Figure 6 Distribution of reported causes.

embarked (i.e., a significant delay). However, thisquestion 4 includes any of the delay levels.

The collected data, shown in Figure 7, showgreat variation in the impact of doors, depending onthe railcar and transit agency. The figure suggeststhat a focused effort to address failures of doors oncertain railcar classes at transit agencies B, C, D, andE could have a big positive impact on service.

5.3 Door Problems by Season

In the first questionnaire, question 6 in Part II,Operations, deals with train delay variation by sea-son. The collected data, shown in Figure 8, showvariations in the impact of door problems by seasonamong transit agencies. Some variations seem cor-related to weather and some do not.

For example, transit agency ‘A’ in a moderateclimate has the least variation in weather, and theleast reported weather effects. The other transit agen-cies are in four-season climates and show some signof seasonal effects. The figure suggests that wintereffects are the most consistent and, therefore, impliesthat a design effort concerning weather effects anddoor reliability should start with those possible win-ter effects.

5.4 Types of Door Failures

In the first questionnaire, question 7 in Part II,Operations, deals with train delay by type of doorfailure. The collected data, shown in Figure 9, showa substantial variation by agency and car class.

Door opening en route is a substantial hazard, asit exposes riders to the risk of falling out of a moving

train. Phantom operation occurs when doors open orclose under safe conditions but without driver com-mand. Sticking doors do not fully open or fully closeand require driver intervention. Loss of operatormeans that the door cannot operate at all and must betaken out of service.

The ‘A,’ ‘B,’ and ‘C’ transit agencies havemany failures because of no motion indications,while the other transit agencies show few inci-dents of this type. All transit agencies and carclasses report substantial problems with sticking,except the car class ‘B1’ at transit agency ‘B.’Phantom operation is a problem in some cars butnot in others.

The variations in these data suggest that furtheranalysis will be useful to understand why a failuretype is significant at one transit agency or in one carclass but not significant in another.

5.5 Factors Affecting SatisfactoryOperations and Reliability

In the first questionnaire, question 8 in Part II,Operations, deals with factors affecting satisfactoryoperations and reliability, including maintenance, pas-sengers, design, environment, and employee actions.The collected data, shown in Figure 10, show a sub-stantial variation in effects.

Maintenance affects door operations for everyreporting car class, and at transit agency ‘E’ it is theprincipal effect. Passenger use is a consistent causeof door incidents, as is to be expected. Most of the“other” cases indicate no trouble was found duringmaintenance, which often indicates an intermittent

12

0%

10%

20%

30%

40%

50%

60%

70%

A1 A2 A3 A4 B1 B2 B3 B4 C1 C2 D1 D2 D3 D4 D5 D6 E1 E2 E3 E4 E5

Railcar Class

Per

cen

tag

e o

f A

ll D

elay

s

Figure 7 Train delays due to doors.

design condition. Environment and design are smallto medium factors.

The data suggest that improved maintenance tools,techniques, training, and design provisions in sup-port of maintenance could have a big positive impacton service.

6 THE PATH FORWARD

To the best of the team’s knowledge, the collec-tion of data presented on traindoors.com is unique inthe transit world. Nowhere else is it possible to accessdetailed, usable, timely data on a rail transit vehiclesubsystem from leading transit agencies.

Such data are important to a specialized commu-nity. While the transit equipment engineering com-munity is barely known in the broad world, its workhas a big impact on rail transit reliability and safetyand on the daily commuting experience of millions ofpeople, every day.

6.1 Recommendations and Next Steps

6.1.1 Next Steps for traindoors.com

• Determine how to handle sensitive data suchas manufacturer details.

• Use the on-line questionnaire to guide collec-tion of data in uniform format.

13

C D E

Transit Agency and Season

0

1

2

3

A B

Transit Ag

Rel

ativ

e Im

pac

t(1

= lo

, 2 =

med

, 3 =

hi)

Summer Fall WinterSpring

Figure 8 Door problems by season.

Opening en Route Phantom Operation

Loss of Operator No Motion Problem

Sticking

Other

0%

20%

40%

60%

80%

100%

A1 A2 A3 A4 B1 C1 C2 D1 D2

Car Class

Co

ntr

ibu

tio

n (

%)

D3 D4 D5 D6 E1 E2 E3 E4

Rounding caused some totals to add to slightly less than or more than 100%.

Figure 9 Door failures by type.

• Expand the database to cover maintenance andother performance aspects.

• Solicit and encourage participation from othertransit agencies.

• Continue and expand the data analysis. Addresskey questions such as– “What should I change in the specifications

for my next door equipment procurement?”– “I’m having door switch and sensor prob-

lems. Anyone else have this problem?”

6.1.2 Next Steps for the Transit Industry

• Evaluate value and utility of traindoors.com.• Determine whether the concept should be ex-

tended to collect similar or expanded data onother equipment reliability and performancecharacteristics.

6.2 Cautionary Note

Users should be aware that, although there is aplethora of information in the Train Door database,it comes from only five transit agencies. The possi-bility exists that the sample of data, while coveringa great number of topics and including a substantialamount of detailed technical information, may not bedeep enough to provide conclusive evidence of a trendor possibility. Accordingly, the Train Door databaseat this stage should be considered one tool, but notthe last word, in analyzing train door conditions andfailures.

6.3 Conclusions

The general conclusions from the team’s col-lection effort to date have confirmed to the team thefollowing:

• The steps reported here to collect, manage, andreport on train door data are effective in col-lecting transit industry information that canprovide important and useful results.

• The common door failures questionnaire pro-vides a valuable source of data to focus on spe-cific door system components.

• The data reported by the transit agencies pointto problems caused by switches, sensors, andpush buttons.

7 ACKNOWLEDGMENTS

The generous and energetic contributions ofscores of supporters and participants are recognizedand appreciated.

Thanks to the APTA RSETF, whose ranks pro-duced many volunteers and contributors to this proj-ect. The team membership grew to include 30 par-ticipants, including rail car manufacturers AlstomTransport and Kawasaki Rail Car; door equipmentmanufacturer Faiveley Rail Corporation; engineer-ing consulting firms Booz•Allen & Hamilton, LTKEngineering, Interfleet Technologies, TransportationSystems Design, and Turner Engineering Company;and transit agencies BART, CTA, Delaware Area PortAuthority (DRPA), NYCT, PATH, and WMATA.

14

Figure 10 Factors affecting satisfactory operations and reliability.

0%

20%

40%

60%

80%

100%

A1 A2 A3 A4 B1 C1 C2

Co

ntr

ibu

tio

n (

%)

D1 D2 D3 D4 D5 D6 E1 E2 E3 E4

Car Class

Environmental

Employee Operations

Maintenance

Passenger Use

Design

Other

Rounding caused some totals to add to slightly more than 100%.

Thanks to TCRP and to the National Academies,whose encouragement has been vital to performing theresearch, maintaining the momentum, and enablingthe development and building of traindoors.com.

Grateful acknowledgement to the pivotal contri-butions and hard work of the staff and CEOs of thefive transit agency participants: BART, CTA,NYCT, PATH, and WMATA. Their data are valu-able to the community, and it, collectively, can yieldvaluable benefits for the transit agencies and the rid-ing public.

Special thanks to the working group for TCRPProject J-6, Task 62, for their dedication to this proj-ect, in particular Paul Messina, MTA New YorkCity Transit; Christopher Pacher, LTK Engineering

Services; and Harry Burt, LB TransportationConsultants. Also for their dedication to this project,special thanks to Connie Hwang, Aleph Associates,and David Turner, Turner Engineering Company.

Finally, gratitude and fond remembrances toTom Sullivan, whose skills, enthusiasm, technolog-ical grasp, courage, and gentle leadership were soimportant to starting, guiding, and executing thework reported here.

APPENDIXES

Appendixes are posted on the TRB website asTCRP Web-Only Document 28 (www4.trb.org/trb/onlinepubs.nsf/).

15

Transportation Research Board500 Fifth Street, NWWashington, DC 20001

These digests are issued in order to increase awareness of research results emanating from projects in the Cooperative Research Programs (CRP). Personswanting to pursue the project subject matter in greater depth should contact the CRP Staff, Transportation Research Board of the National Academies, 500Fifth Street, NW, Washington, DC 20001.