This case study was prepared in November 1994 for the Advisory Group on computer graphics (AGOCG) Technical Report 26
animation, cinegrams, composite nodes, diagarams, digital video, hypermedia interfaces, multimedia, technical documentation, time-based media
This case study describes the design and production of a hypermedia technical document. The term 'cinegram' has been coined to refer to this new time-based and interactive document type. A prototype is currently being developed under HyperCard 2.2. The application domain of the prototype cinegram is the oil system of Rolls Royce's new Trent 700 turbo fan engine. Rolls Royce plc in Derbv is a collaborator in this research project.
Unlike printed technical documentation, cinegrams can show processes in motion, e.g. oil flows, valve operation, temperature chanqes, etc.
In printed documentation the standard approach is to present a text supported by subordinate diagrams. The proposal here for hypermedia docurnentation is to reverse the relationship. In cinegrams, the user navigates through the system via interactive, animated diagrams which can show components in different states and point at a network of associated pictures, animations and words (as print or speech). The idea is to illustrate the functioning of components in several ways: through navigation (browsing and searching), sequential presentations, and embedded interactive simulations.
Where computer based systemLs for creating, storing and retrieving technical documentation have been introduced commercially to date, these have largely used the structure of the traditional book as the model for organisation.
The cinegram approach utilises the interactive and time-based capabilities of the computer which can offer far rnore than mere electronic page turning. This throws up many interesting questions about possible ways of using this new technique and the responses to them by users.
'Technical documentation' is a generic term used to refer to a range of visual materials that describe how things work and how to use and maintain them. The things can be mechanical equipment, computer software, or industrial processes of all sorts. The documentation can be in the form of training materials, operating instructions, maintenance guides, illustrated parts catalogues, etc. Whilst the paper-less office is still someway off, more and more companies are following the trend of putting their technical documentation into digital form. In view of the growing complexity of products and accelerated production cycles, electronic (possibly on-line) documentation techniques offer a number of advantages. They are expected to reduce information access time, facilitate the creation and updating of documents, enable automation of repetitive and book-keeping tasks, and facilitate document exchange, conversion and the production of different versions.
The existing applications in this area are mainly text-based. For the sake of consistency, they often retain the structure of printed documentation since both electronic and printed documentation are still expected to be used interchangeably.
Typically printed documentation comprises two principal modes of communication: text - or 'verbal graphic language', as Twyman (1982) would have it - and images - or 'pictorial and schematic graphic language' (ibid).
It should be noted that, despite the maxim 'a picture is worth a thousand words' diagrammatic representations are usually used in < subordinate role to the text, which provides the principal access structure to the document, through its organisation into chapters, heading, sub-headings, summaries, indices, etc. The diagrammatic material is usually arranged to support and amplify the written word, and the reader is directed to it through the text (as we are doing now! ) See figure 1. Biderman (1980 p232) has noted the 'segregated and subordinate status of graphics' [diagrams] and compared them to the cultural roles of minorities.
The dominant role of text is generally maintained when the digital form is used. If pictures and diagrams are included at all then they are not usually interactive. The potential of animation and digital video have only recently begun to be exploited in interactive technical documentation.
With the availability of anirnation and digital video for interactive computer systems, the design, implementation and expected usability of mixed-mode or active hypermedia documents that can contain time-based media have become topics of research interest (e.g., Feiner and McKeown 1993; Fischer and Richards 1993a, 1993b, 1994; Hardn.an et al 1994; Thomas et al 1994) .
Much current research is focused on the technical integration of time-based media into space-based informational structures (Horn & Stefani 1993) . An important example is the development of the SGML-extension 'HyTime' (Newcomb et al 1991) and the development of the IETM (Interactive Engineering Technical Manual) standard (cf. Mil-M-8726 GCSFUI 1992) .
This report describes the developrnent of the concept of the cinegram and the current cinegram procotype developed in the Visual and Information Design (VIDE) research centre at Coventry University. Cinegrams are a document type which use interactive diagrams to point at a hypermedia network of composite nodes. Each of these nodes can call up a number of data files such as pictures, animations and text. The separation of data from the actual application allows for selective updating and potentially, distributed storage of the data. In other words, a cinegram could act as as transparent interface to data kept and maintained on remote servers.
The cinegram development started with a week-long attachment to the department of Visual Communication at Rolls Royce, Derby, during which many informal meetings and interviews took place. The overall impression concerning document design and use at Rolls Royce was that because of the size, complexity and segregation of the organisation a co-ordination of information even on the trivial level of document exchange formats poses huge problems.
During this explorative stage of the design process, one document type caught our attention since it appeared to be both popular and used in very diverse contexts. The document type is that of systems diagrarns, also called Dunwells (after the inventor) or ETGs (for Engineering Technical Graphics). Systems diagrams show an entire engine sub-system, (such as the oil system) in a lot of technical detail, mostly using component cross-sections linked together through the colour-coded flow of oil (see figure 2a) . However, systems diagrams are not part of the standard documentation handed out to the airlines. The standard documentation is developed according to a hierarchical catalogue sequencing standard used throughout the airline industry and certified by the ATA (Air Transport Association) . The main reason why systems diagrams are so useful is that they provide a level of integration non-existent in ATA-certified documentation. Systems diagrams emphasise functional unity and are thus very useful whenever a general understanding of system behaviour is required - whether they are used in system familiarisation training courses or as troubleshooting aids where they can help solving a particular problem be visualising the larger system context.
The multi-purpose character, functional complexity and visual richness of systems diagrams seemed to render them a good starting point for the developrnent of hyperrnedia documentation. The prototype design was inforrned by regular visits to the departments of Visual Communication, Technical Publications, Customer Services, and the Customer Training School. We often presented and discussed the developing prototype, and many people made useful suggestions . We realised that what people expected from the system varied greatly according to their working background. The notion of a 'core cinegram' based on the structure of the physical engine emerged. We thought of the core cinegram as a framework which could be arnended for different purposes and contexts of use. For example, the training school might utilise the possibility to create tailored sequences out of the cinegram material which could be used as a presentation mediurn in instructor-led training courses, while Customer Services might use an animated cross- section of a bearing charnber as a point-and-click access tool to data on oil pressure, ternperature and f low speed under various operating conditions .
We also thought cinegrams should allow customisation by individual users who would be able to add to the cinegram pointers to documents they were currently using, and 'get rid of' those bits that were not needed.
In designing the cinegram prototype, we have used a variety of sources, like systems diagrams, the engine manual, schematics, photographs and technical reports, which all in some way refer to the reference system, i.e., the physical engine. In selecting relevant aspects of the reference system and reconstructing these as a cinegram, design decisions are undeniably biased towards a particular kind of anticipated use. Another constraint that has a strong impact on design decisions is the layout of the source material. For reasons of economy (and also in order to secure consistency with the printed oil system diagram), the cinegram designer would be ill-advised to attempt a complete redesign of the laboriously produced technical illustrations serving as the source material for the cinegram animations. An example may illustrate this constraint and also show how the diagrammatic purpose can inform particular spatial re-constructions.
The technical illustration underlying the pressure pump and filter component animation (figure 3) shows both pump and filter in one diagrammatic cross section. In fact, these two functionally independent components are again part of an even bigger ensemble also housing all scavenge pumps. For the design of the overview map (figure 2b), this poses an interesting question: should the overview emphasise spatial or temporal relationships? While pressure pump and scavenge pumps are far apart in terms of temporal flow cycle, they are placed in the same case and even driven bv the same gear shaft.
The technical illustration itself (on which the animation in figure 3 is based) is actually an ingenuous montage: two different cross sections are combined into one image in order to emphasise the functional relation within the overall assembly of the pump casing. The view focuses on those technical details that contribute to an understanding of the function. While it is conceptually precise, it modifies the actual physical arrangement.
The term cinegram has been coined to refer to interactive hypermedia diagrams that include time-based media such as digital video or animation. The cinegram approach is distinctly different from the traditional text-to-diagram, relationship used in books and their digital equivalents. In contrast to a text-oriented approach, an animated diagrammatic representation of the physical system (in our case the oil systern of the Trent 700 engine) is used as the principal rneans by which readers access information (See figure 2b, 3 and 4) .
In a cinegram (as in Hypemedia in general) the designer can first of all translate his or her conceptual model of the reference domain into a network of interconnected nodes. The resulting structure can be dynamic and account for more than one decomposition. The possibility of tucking away large amounts of information reduces the information overload and visual noise found in many diagrams which compress a detailed view of rather complex system into a single field.
Cinegrams are then based on a simple hierarchical structure. The top level is the respective system itself, e.g., the Oil-, Fuel-, or Heat Management system. Each system then contains a number of components, and each of these components can be shown in a number of different states and presenced in a variety of document types, like animated cross-sections, phovographs, or line drawings. In terms of prototype implementation in HyperCard, systems correspond to stacks, components to backgrounds and states to cards.
Basically, cinegrams can be used in three different modes:
1) The user navigates through the hyper-document in a more or less constrained manner. Hypermedia is based on the notion of interactive access to documents by means of navigating links, i.e. through browsing or search mechanisms. The basic pattern is in many ways similar to looking something up in a traditional encyclopedia. The temporal unfolding of information in the process of interaction with the document (and/or other humans, sources, etc.) is in the hands of the user. In a future implementation, history mechanisms and dynamic maps may provide navigational feedback on request, while user configurations (like link bases) may generate particular cinegram instances. Results of user interaction may be saved in some way, e.g. for versioning control or for performance assessment in interactive courseware.
2) The user interacts with dynamic simulations of particular system aspects. Some cinegram elements are responsive in a way different from simple links. For example, looking at a diagram of the heat management system, the user might change the oil temperature through a simple point-and-drag operation over a thermometer icon. The system will then respond by animating a chain of events leading to the further opening or closure of the air modulatinq valve of the air-oil heat exchanger.
3) The user selects 'canned' sequences showing important processes within the system (e.g. the effects of a fault, component function, or maintenance procedures). The most general sequence would include all system components and arrange them along a linear path in the order in which they are traversed by the oil flow: frorn the oil tank to the bearings and finally back to the tank
Time-based media do not sit comfortably in an interactive frame- work because they carry their own temporal definition: they conform to a presentation paradigm rather than the interactive paradigm. Once triggered, they tend to suspend user action; the user stops interacting and just watches the sequence. Most current systems constrain interaction with time-based media to the few actions which have crystallised in the video panel scheme. In cinegrams, we want to explore the affordances given in timed embedded controls. How can the availability and functionality of embedded links be signalled to the user? But more fundamentally, how can a process be broken up into discrete segments, and how can invisible aspects (like oil pressure or flow speed) be visualised?
One sequence of pressure pump and filter component view (figure 3) demonstrates the application of cinegrams for the instruction on maintenance procedures. The animation shows the functioning of an invention that makes life easier for the flight mechanic. When the filter element is removed, an anti-leak valve is automatically closed to prevent oil spill (figure 3b).
The schematic drawing of the component shows both states of filter on each side of a vertical dividing line. Some technical imagination is required to complete the other half for each state and mentally perform an 'in-between' between both states. A static technical illustration might show both states separately. An animation has the advantage of showing the transition between both states over time. This way, the functioning of the automatic shut- off mechanism becomes immediately obvious.
Another sequence shows the development of a fault condition (see figure 4). Fault conditions can cause cascade effects that rapidly propagate beyond the system boundaries to affect the whole engine. Regular maintenance ensures that those conditions rarely become reality. But the very fact that such conditions are rare and consequently rarely observed indicates that cinegrammatic presentations of those conditions and their effects may help the people who fly and maintain those engines to get a better conceptual understanding of the system.
The fault sequence shown in figure 4 explains the blockage of the scavenge filter through debris in the oil. The clogging of the filter element results in a pressure drop past the filter and pressure rise before the filter. When a differential pressure of 13 psi has been reached, a differential pressure meter sends a signal to provide flight deck indication of impending by-pass (figure 4b). At 20 psi, the spring-loaded pressure relief valve is blown (figure 4c) to let the oil bypass the blocked filter element (figure 4d).
The entire fault sequence falls into two different processes on both sides of a peak in the sequence: the slow building up of pressure until the signal is sent, and the blowing of the valve shortly after. The first stage may take two weeks of operation during which the filter gets increasingly clogged; the second may take only a fraction of a second until the filter bypass state is reached.
The animation takes considerable license in its treatment of time. In order to accentuate the relatedness of both stages, it radically accelerates the first stage and to some lesser degree, slows down the second state.
The primary aim of our planned evaluation is to assess the usefulness of cinegrams as documentation tools and learning aids. Beyond that, we will concentrate on the way users interactively unfold system relations and processes. We want to explore how the affordances of the cinegram interface relate to the user's understanding of cornplex systems. The evaluation will happen in three stages:
1) Formative interface evaluation with novice users (students of technical communication) confronted with the system for the first time; the user is asked to carry out a simple task, e.g., describe the main functionality of the oil system while witness protocols are taken and analysed. The results will inform the prototype design so that some usability problems will be sorted out before the next stage of evaluation.
2) A group of inexperienced users (students of technical communication) will be given a basic introduction to the function of the oil system and a demonstration cinegram interface. The users will then be asked to find and write down in a given amount of time answers to a number of written questions all of which presuppose some conceptual understanding of the system. A control group will use written documentation (the 'system description and operation' section from the Trent engine manual) . A third group will have access to both types of documentation. Witnessing documentation use of the third group will reveal the turning points at which the information seeker may abandon one medium and turn to another, and to show when and why different media are used in conjunction. Users will be asked to write a short free-form statement immediately after the session, and additionally, to fill in a short questionnaire covering usability aspects of the cinegram which might have been left out in the free-form statements. Witness reports and user statements will be compared and analysed. The written answers to the problems and the results of the questionnaire will provide some 'hard' facts which we hope will provide an indication as to the usefulness of cinegrams as a documentation tool.
3) A nurrber of expert users in the technical support department at Rolls Royce will 'play with' and comment on the system. This will reveal what they would expect the system to do in their specific context. Possible tasks are accessing pressure, temperature, and flow speed values within one bearing chamber under various operating conditions. Users will be asked to write a short note on the problems they encountered; additionally, the time it took them to arrive at the answer could be measured and compared with the time for accessing this information using printed and CD-based engine manuals.
The aim of research was to explore potential applications of multimedia technology for technical documentation. The design process has indicated that the success of any design - its usefulness - will not primarily depend on the current state of technology but on the organisational background, i.e. the history of information requirements and documentation design and use in quite diverse working environments. We see the cinegram development as a pilot design project which has indicated some of the issues and opportunities of collaboration with industrial partners in speculative research of this kind.
The contact with Rolls Royce was frequent and generally fruitful. The nature of co-operation was relatively informal and since the concept of cinegram w as not a response to any clear-cut actual problem of any particular clienc within Rolls Royce, there was some hesitation in recognition of its potential usefulness within the documentacion scrategy of the company.
Nevertheless there was rnuch enthusiasrn by those who saw demonstrations of tne developing cinegram prototype. It was clear from the start that the emphasis was on pictures and diagrams as primary access structures to information; an approach many engineers (who tend to f lip through bulky documents searching for pictorial and diagrammatic cues, and only read text if they have to) felt very sympathetic about.
However, there is a major technological (and perhaps cultural) divide - while we are using Apple Macintoshes for the prototype design, Rolls Royce, like most industrial companies, use PC's and Unix workstations almost exclusively. One of the most frequent questions was 'Can you run this on a PC?' The answer, sadly, had to be 'No, not at the moment.' This divide continues to contribute to the difficulty of evaluating the prototype - people experience the prototype on a machine they are not familiar with and currently does not fit easily into their normal working context.
All this can present difficulties if an iterative design process requires the evaluation of early versions of the prototype under realistic conditions. The potential benefit of the design project needs to be clear to the industrial partner if the necessary commitment of resources is to be made for trials.
The collaboration entered a new stage when desk space for a researcher was arranged in the Department of Customer Services and therefore access to both documentation and communication with employees was facilitated. This has proved extremely beneficial to the development process.
This case study has described the development and function of a new hypermedia document, called a cinegram, which uses diagrann.latic techniques in order to show the functioning of complex technical systems. Cinegrams include aspects of navigation, sequential presentation, and interactive simulation.
The cinegram prototype described here is still under development. The concept is sufficiently general to allow application to a variety of other domains, but much further research is necessary in order to investigate the potential and scope of their application. In developing the concept of cinegram, we intend to reformulate the wealth of information design techniques found in the diagrammatic tradition (Richards 1984, Pedersen 1988, Tufte 1990) within an interactive and time-based framework. Thus we expect that the concept of cinegram will contribute to the development of future dynamic hypermedia environments.
We would like to thank John Buckland, manager of Visual Communications, for arranging attachments to Rolls Royce for one of the authors which provided many insights as to the way technical documentation is used within the company. Frank Thomas was particularly helpful in giving expert advice, providing many documents, and arranging numerous meetings with other people at Rolls Royce. Much is also owed to Jeremy Laffan at Customer Services, who spent much time explaining the technical intricacies of the Trent oil system and was a great help in specifying the prototype. Many thanks also Mark Johnson and Eddie Ashley, designers of the paper-based Trent oil systerns diagram for frequent discussions. Rolls Royce plc is gratefully acknowledged for perrnission to reproduce the simplified oil system overview shown in figure 1.
Thanks are also due to all tnose people from Rolls Royce plc in Derby who in scme way contributed to our research. Amongst those who provided first-hand experience, helpful comments and critique are, in, alphavetical order, Tea Bardgett, Richard Barton, Simon Beevers, Mike Bowrnan, Graham Briggs, Macolm Grayburn, John Heathcote, Graharn Lewis, Brian Lomas, Simon Pank, Paul Rees, Alan Steele and Chris Timm.Finally, we also thank John Vince of Thomson Training & Simulation; Don Hinson, Sam Porter, Bob Newman and Niki Paul of Coventry University, Coventry School of Art & Design; John Lindsay of the School of Information Systems, Kingston University; Sam Deane and Tom Smith of the Ultralab, Anglia University and Karl Longman of Attica Cybernetics for helpful comments and discussions.
The prototype makes use of a number of different XCMDs amd XFCNs: Hilite XCMD, LocalToLocal XFCN, UpdateWindow XFCN Rob Bevan 1994; Say XCMD, Pollocks XCMD Sarn Deane 1994; QTmovie XCMD Ken Doyle, Apple Computer 1994; WindowScript 1.5 XFCN Heizer Software 1993; Textoid XCMD, Pictoid XCMD, Listoid XCMD Frederic Rinaldi 1994.
Biderman, Albert D 1980 'The graph as a victim of adverse discrimination and segregation'. Information Desiqn Journal, vol l, no 4, pp232-241.Cypher, Allen and Stelzner, Marilyn 1991 'Graphical knowledge-based model editors'. In: Sullivan, Joseph W and Tyler, Sherrnan W ecis.: Intelligent user interfaces, pp403-420. New York: ACM Press, Addison Weslev. Feiner, Steven K. arla McKeown, Kathleen R. 1993: 'Automating tne generation of coordinated multimedia explanations'. In: Maybury, Mark T. ed.: Intelligent multimedia interfaces, ppll7-138. Menlo Park CA / Cambridge MA: AAAI Press/ Ths MTT Press Fischer, Detlev and Richards Clive J 1993a 'The presentation of time in interactive animated systems diagrams'. Proceedings of the British Computer Society International Conference Multimedia Systems and Applications, 7-8 December 1993, Weetwood Hall Conference Centre, Leeds, UK. Fischer, Detlev and Richards Clive J 1993b 'Trade-offs in the display design for on-line technical documentation'. Proceedings of the ELVIRA (Electronic Library and Visual Information Research) First International Conference, 3-5 May 1993, De Montfort University, Milton Keynes, UK. Fischer, Detlev and Richards Clive J 1994 'Interactive systems diagrams (cinegrams) and their potential for exploratory learning'. Proceedings of the Conference on Computer Aided Learning in Engineering, 5-6 September 1994, University of Sheffield, Department of Mechanical and Process Engineering, Sheffield, UK. Hardman, Lydia; Bulterman, Dick C A and van Rossum, Guido 1994 'The Amsterdarn hypermedia model: adding time and context to the Dexter model' . Communications of the ACM, vol 37, no 2. Horn, F. and Stefani, J B 1993 'On programming and supporting multimedia object synchonization'. The Computer Journal, vol 36, no 1. Mil-M-8726 (GCSFUI) Interactive Electronic Technical Manuals. General content, style, format and user-interaction requirements. 20 Nov. 1992. Newcomb, Steven R; Xipp, Neill A and Newcomb, Victoria T 1991 'The "HyTime" hypermedia/time-based document structuring language' . Communications of the ACM, vol 34, no 11, pp67-83. Pedersen, Martin B. 1988 Graphics diagram 7: the graphic visualisation of quantitive information, procedures, and data. Zurich: Graphics Press Corp. Richards, Clive J 1984 Diavrammacics. PhD dissertation. London: Royal College of Art. Thomas, Frank; Hinson, Don and Fischer, Detlev 1994 'Keeping the show on the road. The progress of knowledge disseminacion through time'. Communicating 94. The ISTC Annual Conference, 7- 9 October 1995, Loughborough University, Burleigh Court, Loughoorougr!, UK. Tyman, Michael 1982 'The graphic presentation of language'. Information Design Journal, vol 3, no 1, pp2-22 Zellweger, Polle T. 1592 'Toward a model for active multimedia documents'O In: Blattner, Meera M. and Dannenberg, Roger B. eds: Multimedia interface design, pp39-52. New York: ACM Press.