The KIM Project  is a £5.5 million research programme involving eleven UK universities and funded primarily by the EPSRC  and ESRC . The Project's tagline is 'Knowledge and Information Management Through Life', and it is primarily focussed on long-lived engineering artifacts and the companies that produce and support them. The driver for the research is a 'product-service paradigm' that is emerging in several industrial sectors, whereby a supplier is contracted not only to deliver a product such as an aircraft or building, but to maintain and adapt it throughout its lifecycle.
The second KIM Project Conference was held on 2–3 April 2008 at the University of Reading. As with the previous conference , the purpose of the event was to allow the Project's researchers to present their findings so far to the rest of the Project team, and to garner feedback from industrial representatives.
The first session of the Conference was the showcase for papers of particular relevance and interest to industry, and ranged across the various different aspects of the Project.
Alison McKay of the University of Leeds started the proceedings with a paper on integrating product, process and rationale models. The move to product-service offers a supplier new opportunities for improving design work in the light of experience gained from the later lifecycle stages of existing products. It also presents challenges, as it is impossible to predict at the time of design how the product (and its attendant manufacturing and in-service processes) will need to be adapted over the course of its life. Instead of trying to future-proof designs, what is needed is a set of process and rationale models that can be linked to and superimposed on the product model, and updated as the understanding and usage of the product changes over time.
Stewart Johnstone of Loughborough University presented some early findings of research into the impact that the move to product-service is having on suppliers' Human Resource (HR) management. Some of the more pressing challenges are the stark divide often found between the product and service sides of the business, with the service side often receiving a low priority, and the need to align the personal goals of the staff with organisational goals. Some of the strategies adopted include changing the organisational structure to increase integration, developing programmes of service training, and recruiting globally to find adequately skilled staff. There are corporate benefits to allowing HR departments greater influence within the organisation and including them in decisions on strategy.
Michael Lewis of the University of Bath School of Management challenged a common belief that contracts and relationships are mutually exclusive methods that customers use to govern the business they do with suppliers. While contracts and relationships have different dynamics and evolve differently over time, they should be seen as complementary; good governance depends on getting (and maintaining) the right balance between the two.
Yuyang Zhao of the University of Bath Innovative Design and Manufacturing Research Centre (IdMRC) presented the results of a case study looking at how information was evaluated at three different levels of management in seven different organisations. Evaluation practices tended to vary more with functional context than managerial level, however. In an archival context, document properties rather than full text content were used as the basis for judgement, whereas for information published to an intranet, context was key. The research pointed to the utility of applying a profile-dependent 'value tag' to documents, enabling more effective searching and better information system co-ordination.
Yee Mey Goh, also of the University of Bath IdMRC, looked at three different ways in which in-service information could be captured and fed back into the system. Many aerospace companies are equipping their products to produce real-time performance logs, but as this data is held by the customer, the suppliers cannot take advantage of the feedback it provides. Service reports are usually semi-structured, but are hard to analyse due to the variety of ways in which free text portions are completed, and the number of locations over which they are distributed. NASA's Lessons Learned Information System, in contrast, is a highly managed system, where staff are trained to enter records in an objective and consistent manner, and each record is checked for compliance. This system still has cultural barriers to overcome — unwillingness to admit mistakes, for example — but its value is being proven by the reuse of the records.
The final presentation of the session was given by James Wasiak, again of the University of Bath IdMRC, who has been conducting a case study into the use of email by one of the Project's industrial collaborators. Some of the key statistics were that 35% of the engineers studied spent more than two hours a day using email, and 40% felt they didn't use it effectively. Where systems alert them to new email, engineers read email within six seconds of it arriving; thus email was highly disruptive to their work. Study of email content revealed that during product development, 70% of emails related to organisational aspects rather than the project or product, although the product came to dominate email correspondence later on. Notably, 70% of emails were purely informative, and while approximately a third concerned problem solving, hardly any decisions were recorded within emails.
The session concluded with a feedback period, where the industrial delegates could discuss the papers presented. It was encouraging to hear them reiterate and amplify many of the points raised by the speakers. It was also a little uncanny that one of them should call for some sort of integrating framework that linked the research up in an easily digestible way, as this lead neatly into the next paper.
Work package structures are good for dividing research work into logical threads, but not so good for explaining how the threads tie together. One of the deliverables of the Project's administrative work package was to produce a framework that did link up all the research into a cohesive whole, and to this end the EGIPT Model was born. Simon Austin of Loughborough University explained how the research could be thought of as involving the engineering context (Environment), people (Groups and Individuals) and things (Practices and Tools). (As an aside, the Practices element of the model was briefly known as Techniques, but this would have yielded too amusing an acronym.)
Each of the three main work packages contribute to all five divisions of the model, although Work Package 1 — Advanced product information representation and management — is more concentrated in the Practices and Tools areas, whereas Work Package 3 — Managing the knowledge system lifecycle — is more concentrated in the Environment, Group and Individuals areas. Work package 2 — Learning throughout the product-service cycle — is fairly evenly spread.
The remainder of the Conference was explicitly organised around the EGIPT Model.
Jurong Zheng of the University of Bath School of Management kicked off the second session with a comparative case study of two PFI hospitals, looking particularly at the asymmetries in how the public and private partners learned from each other. In both instances, the private partner seemed better equipped for learning, with more mature processes in place — meeting structures, knowledge databases — while the public partner's approach was more ad hoc, though the asymmetry was different in the two cases.
Wisdom Kwawu of the University of Reading considered the place of incentives in contracts and business relationships, and noted the need for vigilance over the course of a contract to ensure that the incentives are working. Incentives need to match up with well-defined performance measures, but handling these measures can be tricky as business requirements evolve over time. Wisdom presented a case study of a 45-week project to build a £4.5m office building complex. The project was dogged by misaligned incentive systems through the supply chain (the primary contractor did not want to pass the incentives down the chain), the performance measures were subjective, and poor communication made it hard for managers to prioritise the work.
Florian Täube of Imperial College, London looked at organisational learning in the context of a UK-based firm with subsidiaries in India and China. The case study found few mechanisms for disseminating knowledge, and found that technological, geographical and cultural distances were still significant barriers. Indeed, the case study highlighted the advantages of a co-located design team over a distributed team, and found that a scheme of temporarily embedding individual members of the subsidiary team into the headquarters proved to have a highly positive impact.
Kathryn Fahy of the University of Lancaster examined the role of tacit knowledge in engineering expertise. The tacit–explicit axis is just one dimension along which knowledge can be placed; others include the individual–social axis, the axis between intellectual knowledge and acquired skill, and the axis between prescribed procedure and actual practice. The case study found that a whole range of knowledge types were key to solving problems: written documents, knowing one's way around the archive system, having a 'feel' for problems, understanding the past, and so on. The picture painted was of bricoleurs bringing together diverse resources in order to come to an understanding of the situation.
Finally in this session, Jawwad Raja of the University of Reading presented a further case study on the theme of the impact of the move to product-service on HR management. The study concerned two service businesses, both of which were undergoing continuous restructuring and were changing their HR structures to suit. In moving to a 'business partner' model , HR was given a more strategic role within the business, with more personnel functions devolved to line managers, and a consequent reduction in the number of HR staff. However, this change in role was only implemented patchily, due to strong resistance to the changing practices in some units.
The final session of the first day began with Khadidja Grebici of the University of Cambridge presenting a technique for reducing risk in the design embodiment process by explicitly recording the uncertainty and maturity of the information used. The technique involves analysing the uncertainties associated with a piece of information (errors, imprecision, instability) and assessing the likelihood that the uncertainty is high, medium or low, according to a given set of metrics; these factors are used to generate the risk likelihood. Similarly, the likelihood that maturity indicators (data quality, validity, sensitivity) are characteristic of high, medium, or low maturity is assessed, and the results used to generate the risk impact. The risk likelihood and impact together determine the priority to assign to the risk, and the probability assessments indicate where greater certainty is needed in order to reduce the risk.
Carys Siemieniuch of the Loughborough University demonstrated a decision making system for product-service projects. The system consists of agents (those making the decisions), activities, infrastructure and technology, and knowledge and information; these affect or are affected by internal variables, external variables, organisational culture and the level of decision-making. This system is currently being used to analyse a set of incidents (nicknamed the 'disaster database') to discover the reasons behind the decisions that lead to the incidents. The next stage of the research will be to derive recommendations from these findings.
Weisheng Lu of the University of Reading gave the last presentation of the day, and livened up proceedings by positioning his paper as a sequel to the Matrix trilogy of films. His subject this time was a decision support system that uses a weighted SWOT analysis, the conclusions of which are derived using matrix mathematics. The theoretical basis for the system is a mixture of both the rational-analytic and naturalistic schools of decision making; decision making is seen as a process of analysing the situation, generating a set of options and choosing the best option from this set. The proposed system would constitute an artificial intelligence for generating the set of options, but would need a well-curated expert knowledge base from which to work.
Day Two of the Conference began with three papers looking at the design record and tools for creating and organising it.
Lian Ding of the University of Bath IdMRC explained how lightweight representations of 3D computer-aided design (CAD) models, coupled with 'stand-off' (separate) XML annotation files, can be used to reduce costs and simplify the process of curating designs and communicating them across the extended enterprise. The annotations allow non-geometric information to be layered on top of the geometry, but being 'stand-off' they can be applied equally well to the original model or its lightweight derivative. The concept is being proven using a full-weight CAD system (NX) and two lightweight formats: X3D and 3D PDF.
Alastair Conway of the University of Strathclyde demonstrated complementary methods for recording synchronous and asynchronous design work. On the synchronous side, the Media Enhanced Minuting System (MEMS) allows one to upload in real time textual and photographic records of what happens in a meeting, along with any digital materials used in the discussion. It is also possible to link these in with video and audio captures of the meeting. On the asynchronous side, each activity can be represented in XML according to its inputs, outputs, controls and mechanisms. These XML records can be linked up to provide a complete chain, and transformed via XSLT into reports. The two modes can be reconciled into a single process chain by considering reports of asynchronous work as inputs to synchronous work, and synchronous work as setting the agenda for asynchronous work.
Yee Mey Goh of the University of Bath IdMRC returned to the podium to give a presentation on the faceted classification of in-service issue records. The technique chosen was to transform a corpus of in-service issue records into structured XML documents, where each element represents a facet by which the record can be classified. Fields with controlled vocabulary (such as operator, product variant) were simple to turn into facets. For the free-text part of the record, the corpus had to be examined carefully to see which types of information came up most regularly; in this case, subjects such as function, failure mode (e.g. corrosion, crack, wear), and type of usage proved most suitable. It is hoped that text mining techniques can be used to generate these facets. The advantages of using faceted classification in this context is that it makes it easier to spot systemic issues with a product, and allows one to spot recurring queries.
The final two sessions of the Conference featured some procedure-related tools, alongside theoretical and philosphical papers on Project themes.
Raymond Sung of Heriot-Watt University showed how motion study techniques, such as chronocyclegraphs and therbligs, can be applied to logs of interaction with virtual reality (VR) design software in order to highlight inefficiencies and training needs. In one case study involving cable harness design, it was found that the designers were spending 47% of their time navigating through system menus, but only 27% of their time designing; this revealed that the menu navigation needed to be simplified and improved. Other applications of this system of analysis include finding the best way to go about a maintenance activity, and deducing the rationale behind a given design decision.
Nathan Eng of the University of Cambridge presented some early thoughts on the relative merits of graphical and narrative methods of presenting information. Human thought is narrative in nature, while endeavours such as complex engineering work and the systems that support it are increasingly non-linear and web-like in nature. There is an increasing trend to move away from textual reports and similar narrative structures to graphical representations such as concept maps, rationale maps and mind maps, but it is not clear what is gained and what is lost by this. One danger could be that without an over-arching narrative to interpret the information, the significance of the collection of graphs may be lost.
As if in reply, Rob Bracewell of the University of Cambridge then presented a comparison of traditional narrative design documentation with an integrated design documentation method using DRed, an editor for design rationale maps. The DRed maps linked in with the various informational inputs to the design process (Word documents, PDFs) as well as live spreadsheets where the calculations were done, while version control was accomplished using a Product Lifecycle Management (PLM) system. While the point was made that written records could be optimised for certain oft-repeated design tasks, the new integrated method did look easy to follow and could be used for novel design tasks.
Bob Young of Loughborough University set out a new way of capturing and re-using manufacturing best practice. The technique involves using the second edition of the Unified Modelling Language (UML2) to describe constant and conditional relationships between design part families (defined by their geometry, topology and functional requirements) and manufacturing part families (defined by how they are made). By expressing best practice in this way, it becomes possible to automate the process of identifying the most suitable manufacturing options for a given design part. A proof of concept is being implemented in NX, Teamcenter and E2KS.
Peter Heisig of the University of Cambridge discussed research into an integrated product-process-rationale (PPR) framework, to enable organisations to ensure they collect the right information and knowledge at the different stages of the lifecycle. The first branch of the research was literature-based and yielded seven different PPR models; among these models there was certainly no consensus over what terms should be used, nor even what the terms should mean, but some useful components were identified. In order to prioritise these components, a second branch of the research surveyed practising engineers about their knowledge and information management concerns. More analysis needs to be done before producing the first draft of the new model.
John Rooke of Salford University reported on meta-research concerning product-service developments in the construction industry. Rooke argued that more attention needs to be paid to the information carried by physical artifacts and environments. He also introduced a procurer–provider–user model of meta-relationships in product-service lifecycles, tentatively suggesting that the product-service concept should be expanded to include policy. Rooke's last point was a methodological plea in favour of the strong unique adequacy principle: only analysing situations in terms native to the context.
In the final paper of the Conference, Koray Pekericli of the University of Reading presented a vision of a construction industry that uses a shared data repository and radio frequency identification (RFID) tags to facilitate through-life knowledge management. RFID tags can be used not only to identify components, but also to report on environmental conditions and the performance of subsystems. Linking the RFID data to a data repository allows components to be tracked from design to disposal, and enables one to discover systemic failures or underperformance of given components. This workflow is equally applicable to other complex engineering contexts such as aerospace. A proof of concept implementation has been prepared.
Project Director Chris McMahon of the University of Bath IdMRC brought proceedings to a close with some words about the direction of the Project in its final year. Researchers who have primarily focussed on describing existing scenarios and behaviours were asked to think about recommendations they could make to industry, while those developing tools were asked to think about their technology readiness level . Above all, the key was to make the research accessible, relevant and useful to industry, so it can be used to improve knowledge and information management with immediate effect.
From the perspective of a researcher on the Project, it was encouraging to see how much progress has been made in the past year. The strands of research that seemed a bit abstract last year have now gained rather more substance, and the final stages of the Project look set to yield some exciting and relevant results. The Conference itself was smoothly run, and maintained its tradition for rigorous time keeping (this time speakers were threatened with crocodile-infested swamp). It was also a little emotional as it marked the retirement of the Project Manager, Garry Burt, who has done so much to ensure close collaboration between the Project and industry. All in all, it was a valuable and positive couple of days, and the Project leaders hope to conclude the series with a third conference in spring 2009.