Experiences of an implementation

a database & SGML-driven authoring and viewing system at Rolls-Royce & Bentley Motor Cars Ltd

Alan Sockett CEng, MIMechE

Rolls-Royce and Bentley Motor Cars

Crewe

United Kingdom

alan.sockett@rrmc.co.uk

Leszek Wawrzyniak MSc

ASC - Advanced Software Corporation

Moosburg

Germany

leszek.wawrzyniak@t-online.de

Find

Abstract

In September 1995 Rolls-Royce and Bentley Motor Cars Ltd decided to develop a completely integrated system for the creation, management, publication and display of Service and Parts technical information. This was named IETIS (Integrated Electronic Technical Information System).

IETIS used integrated SGML-Authoring, SGML-Management and Database technologies to replace desktop publishing and post-processing for electronic display. The post processing resource and CD viewer were supplied to Rolls-Royce and Bentley Motor Cars by a third party supplier. This process was first used in June 1995.

The main reasons for moving to electronic display was to reduce the cost of publication (print costs) and to provide a better process for updating information (replacing the CD instead of sending individual amendment sheets).

The primary reason for the introduction of IETIS was to reduce the cost of producing technical information.

This objective was achieved by:

1. Creating information which could be automatically compiled and displayed without additional processing costs.
2. Creating user-friendly authoring tools, this enables a single author to efficiently create the complete rage of technical information for a feature.
3. Enabling automatic input of the language translations as variants of the source language.
4. Making the reuse of information an easier process for the authors e.g. import information instead of re-keying.
5. Providing management information relating to the work status

The IETIS System was designed to cover management of all types of technical publications (including non-SGML) such as: Workshop Manuals, Parts-Catalogues, Wiring-Diagrams, Technical Bulletins, Man-Hour-Schedules (Service/Labour-Times) etc.. in a single, consistent, object-oriented environment based on Object-Management-Engine (OME) supporting among other features, hierarchical tree-structures, hyperlinks, versioning, multilingual-variants as well as Rolls-Royce Motor Cars specific PIN classification (Part Index Number) and VIN (Vehicle Identification Number).

IETIS has been also designed to meet the requirements of SAE J2008 (Automotive-exchange standard).

The IETIS CD-Viewer is capable of running on Windows 3.11, Windows-95/98, Windows-NT 4.x. Also it has been integrated into the Rolls-Royce & Bentley Diagnostics system to enable a technician to view both Service and Diagnostics information on the same machine.

Management Information - describing: the above introduction, plus the measurable financial and quality benefits, the overall Architecture of the IETIS-System, some examples of user-friendliness of IETIS-Tools, experiences of 2-years of practical, productive exploitation and maintenance

Technical Information - describing: some basics of implementation including the OME Architecture, Versioning, support of Language-Variants and Language-Translations, reuse/sharing of Information, some details of the integration of SGML-Editor, OCR-Engine,...

Some background will be given why certain tools and technical implementation-decisions have been taken (ex ArborText as SGML-Editor, ORACLE as Database...).

Contents

Background
1. Objectives of IETIS Integrated Electronic Technical Information System
2. IETIS concepts
Implementation approach
1. Intermediate authoring environment
2. IETIS "production" authoring environment
IETIS architecture (see )

Background

Objectives of IETIS

In September 1995 Rolls-Royce and Bentley Motor Cars contacted various companies, selected on advice sought from the UK SGML User Group and members of the SAE J2008 working party. The original requirements were;

To provide an authoring environment that would be affordable considering the size of our business (funding would only be available from cost savings achieved by the proposed system that was planned to support a new product range. These savings would be from reduced time to create information, reduced costs for preparing data for CD-ROM and processing language translations)
To ensure that the changes to working methods could be implemented rapidly e.g. they would be acceptable and within the capabilities of existing staff (with minimal computer experience, mixed skills including technical authors who used DTP software (Frame), Parts planners who drafted parts lists).
To provide a single authoring environment for the various types of information that could be used with minimum training requirements e.g. Remove and refit instructions, Service schedules, Illustrated parts catalogue, Electrical wiring information, Service bulletins. This was essential for the introduction of our engineering author proposals.
To promote the use of authors who were basically skilled vehicle technicians not skilled DTP operators.
To minimise repetition authoring of information, reduction of costs.
To be capable of exporting data which could be processed by a third party for display on the existing ASSIST Viewer
To export information that met the requirements of SAE J2008.
To facilitate data management e.g.: Versions Variants Workflow and status Check in and check out User privileges
To import and enable the reuse of Company and Suppliers data

The following items were added during the project:

To provide a replacement for the original ASSIST 'Viewer'.
To provide ability to create and deliver Service bulletins 'online' to our dealers world-wide.

Originally the plan was based around changes to the text authoring process, however during the concept it was realised that our aim of integrated information could be fully achieved by including the whole range of after-sales technical information.

This paper describes the changes introduced by Rolls-Royce & Bentley Motor Cars Limited to significantly reduce lead-times and costs whilst improving the performance of the after-sales information supplied to the dealer. It will detail how with our Information System partner we have re-engineered our business process by transforming from a "Technical publications" department to a Customer Information Group.

It describes how we have utilised the latest information technology to create and manage the information that we are responsible for delivering. This enabled us to develop our authoring skills in identifying, gathering and where necessary creating the essential data and tools required by our dealers to support our motor cars. This has been achieved in a way that did not involve extensive re-training. Existing personnel who have product and customer experience were able to become more efficient in the process of supplying After-sales information. The project provided intuitive tools that required a minimum of training for both the authors and the dealer staff.

The project was called IETIS.

IETIS concepts

A prime driver was the possible requirement to implement SAE J 2008 with the requirement for structured information and the use of SGML for text information.

However, discussions with the selected partner encouraged a much more fundamental appraisal of the business process of producing technical information. In particular the introduction the CALS concepts of "create once reuse many times" and "life cycle support", provided a much wider area of change than originally planned. Figure 1 shows the information flow model.

Also there was a requirement for our department to support project-based working which demands a much more flexible approach to the production of after-sales information.

The outcome was to specify a custom built authoring system that would be able to:

1. Receive and manage all reusable data e.g. part numbers
2. Create and manage all text-based information including the graphics contained in this data.

This gave the following benefits:

Simplified authoring tool
Information produced in shorter time
Changes can be implemented rapidly
Dealer gets replacement not updates
Facilitates customisation of information for specific models
Enables reuse of information (CALS)

The technology is based on Object Orientation and uses an OME that enables the management of complex, hierarchically structured, objects with versions and multilingual variants, using a classical, relational database management system. The IETIS system is based on client-server architecture using Windows NT, ODBC and an Oracle database. Thanks to ODBC other RBD systems could be deployed. The applications used by Authors include an integrated Arbortext SGML editor, Illustrated Parts Catalogue Editor, Electrical Circuits Editor and a number of other custom-made data management tools, etc.

Figure 1 . Information flow diagram

Table of Contents

Implementation approach

Intermediate authoring environment

The IETIS Project started in Nov-1995 with a short (one-week) feasibility Study, which lead to the recommendation and finally a decision in January 1996: to implement an "Intermediate-Authoring Environment" , which should provide very basic SGML-Authoring , SGML-Management and Graphic-Management capabilities within 3-4 months. This enabled the final decision on the full project to be taken later, based on the departmental experience of the new technologies before a final commitment was made thus reducing the risks involved. However the information produced with this system had to be re-useable in the final system. The product plan demanded this if we were to meet the proposed launch date.

The benefits of this approach over traditional approaches that include long (many months) and therefore expensive Design phases (consisting of Analysis and Specification) included:

The technical authors and illustrators were able to familiarise themselves with the new, so far unknown SGML technology rapidly, without an excessive training (1-day of SGML-Basics only). Almost from the start they were able to produce information for the new product, (design features guaranteed that this would be convertible in future DTD's automatically and with minimal effort). The number of DTD's was minimalised (2), and their structure and amount of tags was optimised, so learning curve was minimal)
The use of the system in a true production environment enabled us to produce a high quality specification for the requirements and features of the final version. Communications play a vital part in this process, it was vital that the system developers were able to translate the users requirements into achievable specifications and agree these in terms understood by the end users.
Risks were reduced, due to the fact that the limited time (3-4 Months) and cost investment (8-Man-Months) was involved. If the risk analysis following the preliminary testing identified excessive risk the project could have been discontinued without jeopardising the product launch date.
The MIS (IT) organisation had the opportunity to gain experience of technologies new to the business and identify how they could support the project, particularly in approving the sign-off process for company approval.
The direct involvement of users and management in the design helped to ensure a positive attitude towards the development of the system. Also a key benefit was the effect on the end users, by involving them directly it enabled a change of approach to their job. The removal of page formatting etc. focused them on information creation alone, using technology to mange this information for them, thus enabling them to increase output.

After a very short (1-week) specification-time in March-1996, the very interactive and iterative implementation of the Intermediate Authoring Environment started. It was successfully finished in May 1996 and the overall acceptance of the Authoring team and Management was very high.

It is very important to mention that the expectations of the authoring team were always carefully managed during the project. The gradual introduction of functionality, as well as the compromises between implementation-time and functionality were communicated so that the team's expectations were managed in a positive way. It was important not to raise expectations that the final specification would be delivered in the early releases of software.

IETIS "production" authoring environment

The success of the "Intermediate" solution was essential to assist in the removal of the various barriers (human, financial, political etc.) to the successful introduction of this new technology into the business. In July 1996 the decision to complete the project IETIS was confirmed.

This phase started with a 4-week period during which the project-plan and detailed specifications were defined. The work was carried out in the same way as for the 'intermediate' system i.e. a highly interactive and iterative implementation.

At this time the IETIS "Production" Authoring Environment was to provide an integrated environment and tools to produce and manage structured, hyper linked SGML and Graphics Information including Illustrated Parts-Catalogue and Electrical Circuits. This information related only to the new product and was planned to be exported to third party who would process the information to enable it to be distributed by CD-ROM and displayed on a PC using a proprietary viewer.

The implementation plan was subdivided into deliverables of well-defined, controllable and measurable functional-blocks which were gradually introduced to testing and usage by the final-users at defined Milestones. As with the 'intermediate' introduction, the users were given an early introduction to the new functionality, and it was possible to generate change-requests resulting from production usage of the software.

It is a well known but often ignored fact, that regardless of time spent on producing a detailed specification there will be always some details which are forgotten. Even with the best team of users and developers issues will arise because of misinterpretation, misunderstanding etc. or the specifications were not detailed or precise enough. Perhaps an even more important consideration is that the requirements can be directly affected by changes that are continually happening in the technology and the business/organisation over a period of time.

The result in both cases is delivery of a product which does not deliver to the expectations of the business at the time of it's introduction.

All participants of the project i.e. the users, project management as well as the implementation team, must be aware of the need to be flexible in coming to terms with these unforeseen issues, and be prepared to cope with these not always easy, but not uncommon situations.

It is also crucial that the System-Architecture and implementation-techniques provide enough flexibility at reasonable cost to cope with changing requirements.

The implementation of IETIS "Production" Authoring Environment, planned originally for Spring 1997, would have been finished on-time, however a change in circumstances caused a review of the project and additional requirements. At this time a risk was identified using the third party to process the information and it was agreed that we could reduce this risk by adding features to the IETIS project.

Therefore 6 months before the launch of new series of vehicles, it was decided to add:

automated CD_Composition engine
new CD_Viewer

However this decision demanded that we would also have to convert the current information supplied for our existing product range to enable us to have a single viewing application at our dealerships. Also it required that IETIS had to have the capability of handling this legacy information including the functionality to modify existing or add new information. Although the data was available in an electronic form, it was only psuedo-tagged (proprietary Tagging, without parsing) and error prone. The data contained ASCII-Text, Text-Tables, TIFF-Graphics, Hotspot-Tables for Parts-Catalogues etc.

This resulted in the decision to develop automated conversion tools to import the legacy data into the IETIS-Authoring environment. Although in terms of the overall architecture of IETIS System, these were just new, additional functional blocks with minimal extensions to the database, it put a lot of pressure on the implementation team and time-schedules, as well as requiring some additional resources on both sides.

The development of the CD-Builder and CD-Viewer has been successfully finished in November-1997

The legacy-data import proven to be quite complex, mainly due to the fact that the manually tagged legacy data contained many errors and exceptions from defined rules. This required additional, significant development and testing.

The final additional requirement was to integrate the delivery of Technical bulletins into the IETIS viewer with automated, remote download and update over FTP-Network. The Implementation of this new functionality across Authoring, CD_Builder and the Viewer with some other minor improvements required additional 3 months, so the IETIS Project was completed on April-1st 1998.

The following diagram shows a simplified Timeline of the IETIS-Project

Figure 2 . IETIS time-line

Table of Contents

IETIS architecture (see Figure 3)

Overview

IETIS-Authoring system manages objects, and using Object Oriented Techniques for storing and managing the complex IETIS Information elements.

The IETIS tools used Visual C++ and MFC (Microsoft Foundation Classes) programming languages

The basic IETIS- Objects are:

Structured Texts, contained in SGML-Text-Elements called SIE's (Service Information Elements-J2008) - with links to other objects
Tables with links to other objects
Graphics -with links to other objects
Part numbers

Out of these basic building nodes, complex hierarchical object representations are built and managed.

These are:

Workshop information (Remove and refit instructions
Frames of Illustrated Parts Catalogue
Electrical Circuits

The core of IETIS is the OME (Object Management Engine designed by CSC Ploenzke). This is an object management layer build on top of a traditional, relational Oracle database (see Figure 4) which manages objects with attributes, versions, subversions and variants, Also the complex hierarchies of objects and additional links between objects in different hierarchies are managed by OME. Hierarchies can themselves (as any other object) exists in multiple versions, subversions and variants.

Figure 3 . System architecture of IETIS

General object management concept (see Figure 4)

Since one of the Key objectives is to provide maximum reusability and data sharing, IETIS has to manage the configuration of objects, Figure 4 shows an example of the object hierarchy used

The system must be able to:

Create an object
Create object Versions, Subversions and Variants
Change object protection
Change object status
Lock and check out a specific object
Deliver it to the editor
Check object back in (without creation of a new version, if not yet released, or with creation of a new version if already released)

Figure 4 . Object hierarchy

Versions and variants (see Figure 5)

Objects can have multiple versions and subversions that are created when a released object has been checked in after a modification, e.g. an update or error correction.

However check out and check in of an unreleased object does not create a new version but only replaces the content.

Variants of an object will occur when the content has not changed but the 'appearance' of the data has. An example is the various translated languages.

Figure 5 . Version and variant relationship diagram

Workflow and status

The following table shows the status levels used to report on workflow and provide the control for conditional export of data for CD build.

Status	Control parameters
Work in progress	Can be checked out, modified and returned
To technical check	Checked out and locked
Return from technical check	Check in, no change from this point without change to version
To legal checkTo legal check	Checked out and locked
Return from legal check	Check in
Send for translation	Legally checked version translated
Return from translation	Check in as variants of legally checked version
CD build	Latest version used in conditional export for CD build

Table 1

Object attributes

Each object has the following attributes to facilitate data management:

Class/Name
Unique object identification number
Applicability - VIN, model, model/year
PIN (Part index number)
Version
Sub version
Variant
Status
Creation date

IETIS elements

Authors functions (see Figures 6 to 13)

The tools produced for authoring information include:

Applicability (VIN, model etc.) attribute builder - see Figure 6
Parts group (PIN) attribute builder - see Figure 7
Structured Text-Editor based on SGML technology - see Figure 8 & Figure 9
Tables-editor
IPC-Illustrated Parts Catalogue -Editor (with integrated 'hot spot' editor) - see Figure 10 & Figure 11
OCR engine for automatic recognition of Item-numbers on illustrations
Print engine for 'paper' output e.g. bulletins
ECE-Electrical-Circuits Editor - see Figures 12 & 13
Import and export tools for suppliers-data and language-translations etc.
Maintenance tools
Status and Workflow management
Graphics database
Version management
Library functions
Integrated import & export tools
Import of parts-information from Engineering & Manufacturing
Export of parts-information to the Parts department
Import-export of textual-information for language-translations
Import tools for suppliers-information
Interactive IETIS CD-Viewer
Import filters for legacy-data. i.e. data created at RR prior to IETIS.

Figure 6 . Model, model-year and VIN applicability builder

Figure 7 . PIN attribute builder

Figure 8 . Authoring SIE selection and management

Figure 9 . Typical SGML entry

Figure 10 . Illustrated parts catalogue editor

Figure 11 . Typical parts frame composition

Figure 12 . Electrical circuit editor; data entry tool

Figure 13 . Electrical circuit editor; connector detail entry tool

CD builder - see Figure 14

The CD builder prepares the data for burning the master CD; it takes the exported files including tables of content, links, and repositories.

Figure 14 . CD builder controls

CD - Viewer (see Figures 15 to 21)

IETIS Viewer uses Visual C++ and MFC programming and is highly customised to integrate with the author output. It use a hybrid HTML and special table viewer to take advantage of Author output together with a highly customised graphic viewer.

The viewer has search engines for VIN, PIN, part number and words, multilingual functionality, graphic and text print, display options (graphics/text split screen displays), database or CD selection and on-line help.

The benefits of o-o also carry through to the viewer program since by carrying the unique identification number information can be related. This enables a 'note' to be added to the displayed data this is linked to the object i.d. number and can be retrieved irrespective of the CD issue being used.

Also o-o is used in the dynamic indexing which is used to display a single index for the PSIS information held on the hard drive (from the electronic download) and the CD for previously published bulletins.

The following figures show viewer screen shots to illustrate the functionality;