Technology Transfer: An Integrated Culture-Friendly Approach I.J. Bate, A. Burns, T.O. Jackson, T.P. Kelly, W. Lam, P. Tongue, J.A. McDermid, A.L. Powell, J.E. Smith, A.J. Vickers, A.J. Wellings, B.R. Whittle 1 Rolls-Royce Systems & Software Engineering University Technology Centre Department of Computer Science University of York York Y01 5DD England Contact Author: Iain Bate e-mail: ijb@minster.york.ac.uk tel. number: 44-1904-432773 ABSTRACT FOR WORKSHOP PAPER This paper describes the ongoing interaction between the York University Technology Centre (YUTC) and Rolls-Royce plc. The mission of the YUTC is to transfer technology, in an integrated manner, existing and novel methods into Rolls-Royce plc in order to achieve improvement in the process of developing aeroengine control systems. The YUTC are principally concerned with the following areas: requirements engineering, reuse, safety case development, timing analysis, process and product metrics. The project has been very successful to date with the YUTC s costs being recouped by savings within the company, and further potential for payback has been identified. This is in contrast to many other technology transfer initiatives which have failed because academics have not truly understood the industrial domain, have not gained the confidence of the industrial engineers and have not been prepared to perform the technology transfer in a suitable incremental and consultative manner. 1 Dr. B.R. Whittle now works at the British Telecom Laboratories, Martlesham Heath.
1. Introduction Technology transfer is a term used to describe activities that have the aim of establishing measurable process improvement through the adoption of new practices. In recent years, Rolls-Royce plc have created a number of University Technology Centres in strategic areas such as Control Engineering, Material Science, and Systems and Software Engineering. The University Technology Centres are responsible for identifying technology innovations in particular areas and determining their usefulness to Rolls-Royce. If the University Technology Centres can demonstrate sufficient benefits then the technology is transferred to the company. In 1993, the York University Technology Centre (YUTC) was established to address the area of Systems and Software Engineering for safety critical systems. In its first two years the YUTC has concentrated on technology transfer in the development of electronic controllers for gas turbine engines. Control on modern civil engines is provided by Full Authority Digital Engine Controllers (FADECs). The FADEC consists of engine sensors and actuators controlled centrally from the Electronic Engine Controller (EEC), a twin channel computer system. The FADEC is a computer controlled management system for the engine, taking inputs from the mechanical and electronic components of the engine, other engines on the airframe (plane), the airframe itself, and the cockpit. The FADEC system is embedded and safety-critical. With the increasing complexity of engine functionality coupled with the need for increased efficiency, the number and complexity of tasks that FADECs must perform are likely to increase in the future. Therefore, part of the YUTC mission is to prepare Rolls-Royce with the techniques necessary for the development and assessment of future systems. Section 2 of this paper will present an overview of the areas in which we have carried out our process improvement initiatives. Section 3 will describe our strategy for technology transfer. Section 4 will present what the YUTC consider to be the advantages of our approach. Section 5 is the conclusions of the paper. 2
2. Overview of the Process Improvement Areas This section will give a brief overview of the six areas that have been identified by Rolls-Royce and the YUTC as having potentially the most reward in terms of cost savings and product quality improvement. The process improvements areas are requirements, reuse, timing, evolvable safety cases and metrics which should be supported with appropriate tools. 2.1 Metrics Software metrics has been a necessary supporting activity throughout the process improvement initiatives in four ways: To evaluate potential pay-back a central measurable task plan provided a useful contract between academic and industrial cooperators and helped us to focus the problems into a coherent and effective strategy for change. To control the change process measures for each facet of the improvement task were made explicit and as a common objective used to monitor progress against the change plan. To manage the new process metrics were introduced as central to the new processes. To evaluate actual payback the common plan allowed us to measure efficiency of our process improvement work and identify how the relationship could work better in the future. The ability to measure, control and evaluate our process improvement tasks cannot be underestimated. Without the open, cooperative and introspective elements we could not guarantee a change for the better. Without a firm foundation for the management of new processes we could not make this a lasting change. 2.2 Requirements Our work has concentrated on the identification of fixed specification levels for the various requirements and design (which also contain requirements ) documents; identifying measures that will increase the uniformity of specifications across different functional areas; providing the means to assess consistency within and across different 3
requirements and design documents; and introducing review guidelines to support these process improvements. 2.3 Reuse Despite their complexity, FADECs share core functionality. Whilst previous research by the YUTC concentrates on reusable code, more recent research has focused on reuse at the requirements and design level, and the reuse of test procedures. Much of this work has taken ideas from existing work on domain analysis [GOMMA95]. 2.4 Timing Rolls-Royce (like most other producers of safety critical systems) currently use a Cyclic Executive scheduler and the Ada programming language for implementing the system timing requirements. The YUTC have investigated fixed priority scheduling as an alternative. Fixed priority scheduling has a number of benefits over cyclic executive scheduling including: more efficient use of resources; determinism is achieved through analysis rather than test; and the scheduler is simpler to maintain [LOCKE92]. The YUTC has performed a case study for the application of a fixed priority scheduler to an actual system, considered scheduler architecture issues and their associated certification requirements, tested a scheduler on the actual Rolls- Royce rigs, defined an appropriate process model, and performed a safety analysis of the fixed priority scheduler. 2.5 Safety The Evolvable Safety Cases workpackage is researching into ways to better support the maintenance and potential reuse of elements of a system safety case [FENE95]. The safety case required for certification of a safety critical system typically embodies the results of a wide range of activities, including design, dependability analyses, testing and operating experience. Through means of a structured argument these results are shown to demonstrate satisfaction of high-level certification or contractual safety requirements (e.g. for the civil industries, DO178B [RTCA92]). 4
2.6 Tool Support As part of the process improvement in requirements, reuse, timing and safety, we have identified various tools that support many aspects of the process. Where parts of the process that would rely on these tools are commercially unavailable, we have developed (and continue to develop) prototype tools that support and prove key features of the process. The development of tool support is not a principal strand of our work. However, we have found that the rigors of tool development have helped us to refine and improve our suggestions for process improvement, and have helped Rolls-Royce understand and evaluate process improvements by case study. 3. A Strategy for Technology Transfer Recognising the relationship between technology transfer and process improvement is fundamental to our approach. Technology cannot be changed without some impact on the process that utilises the technology, be it an increase in productivity, a reduction in cost, or a fundamental change in method. In our experience, a number of key issues must be addressed if technology is to be successfully transferred. These can be divided into technology issues, process issues, and management. Part of our strategy to technology transfer has been to address the cultural reasons for the failure of technology transfer from academia to industry. In on our experience, most academic approaches to technology transfer is to take a specific problem and write a report on the solution. The YUTC has not followed this approach since engineers working on safety critical systems need confidence before they adopt a new approach. This section will consider two technology transfer approaches (pilot study and case based approaches) for examining techniques within a realistic industrial framework. 3.1 Technology Transfer by Pilot Project - Off-line Approach The traditional approach to transferring technology into industry is to make use of pilot projects, i.e. off-line from the actual project. In a pilot project, the new technology is applied to a small non-critical project in isolation from the line. If the solution produced using the new technology can be shown to be more effective than 5
the previous means, then a larger number of line staff are trained in the techniques. The new technology is then used on the next project that the company undertakes. Pilot projects should be relatively small and low risk, but sufficiently similar to other projects in which the company normally participates in order to provide a realistic proving ground for the new approach. There are a number of problems with this approach: Pilot projects should be small and well bounded so that the exact impact of the technology change can be measured. Most small projects tend to be demonstrator projects which are subject to a great many changes. In such an environment it is difficult to gauge the impact of new technology. The small size of a pilot project can in fact act against it because managers are skeptical about the scalability of the technique. Pilot projects have different problems and issues than the line projects and the engineers will have greater difficulty relating to the project. A pilot project rarely comes along, since companies cannot afford to fund them in the current economic climate. The expenditure on staff training used on the pilot project can be seen as a loss of skills and a waste of training if the pilot project staff are not transferred on to a new project. This is particularly likely in larger organisations where permanent R&D or pilot/demonstrator departments exist. The technical staff whom we would view as vital to the success of a pilot project are also key to the company s survival. 3.2 Technology Transfer by Case Study - In-line Approach The previous section described some of the disadvantages of transferring technology to an organisation through pilot projects. In this section we outline the basic principles of what we refer to as the case study based approach to technology transfer, i.e. as part of an actual project. This approach describes the transfer of technology directly into the production line, rather than through the use of independent pilot projects. The case study based approach is centered on the idea of small demonstrator project components running in parallel with the main project. This enables the effectiveness 6
of the new technique to be assessed using the same data, and tackling the same problems, as the real project. In some instances a number of case studies of increasing size can be used to reduce initial scepticism and iterate towards a solution. At each stage the case study is kept at a size where it can be staffed, part time, by the engineers from the project, with the assistance of case study specific staff/researchers. Viewed at a high level the case study based approach has the following steps: 1. Identify and bound a specific problem or situation for improvement. 2. Find a possible solution from existing research and by consultation with existing line staff 3. Prototype the new solution in parallel with the main project and demonstrate its efficacy. 4. Write up the results of the case study, jointly authored with the line engineers, and make a presentation to the manager regarding the costs and benefits of the solution. 5. Obtain a decision regarding the solution, e.g. the need for a larger scale case study, a change in the technique, evaluation on an actual rig facility with the real system etc. An important part of the YUTC strategic approach is to provide Rolls-Royce with help with their short term problems. Even though this distracts the YUTC staff from their research and process improvement it does have two principal benefits. Firstly, the YUTC staff gain a better understanding of the systems domain and the problems to be solved. Secondly, the Rolls-Royce staff gain confidence in the fact that the YUTC are not typical academics, but that they understand their process and problems, and that they have the ability to solve the pertinent problems. 4. Advantages of Our Approach We have found that this case study approach gives us the following advantages: The techniques are realistically proven on production line problems. Large scale technology transfer can be undertaken and assessed in manageable steps. The engineers used are those from the line, reducing costs of training and time to take up the technology should it be adopted. 7
Engineers are less skeptical as they buy into a technology that they have helped to develop. An appropriately sized case study and timescale can be generated that will be sufficient to prove the technology to those whom you are targeting, probably the production line managers. Any successful outcome from the case study is automatically used in the product. Case studies are a continuous approach to process improvement, they do not require large levels of management commitment, and they can survive small setbacks. Case studies are owned by the line, rather than the management and therefore stand a greater chance of being used in the future. This should be contrasted with pilot projects where the line does not own the solution. By working with the engineers in their environment, they have gained confidence in our understanding of the domain and of our ability to solve their problem. Part of our work has been helping to solve their short term line problems to help the industrial engineers and increase our understanding of the domain. As such we regard well managed case studies as a more culture friendly approach to technology transfer than pilot projects. 5. Conclusions One of the key benefits resulting from the technology transfer approach advocated in this paper is that it provides a means for a company to progress into the learning curve for new process innovations in a low-risk way. We have suggested that the use of case studies has been the primary vehicle for achieving confidence in novel or radical techniques. Furthermore this approach has provided opportunities for line engineers to take an active role in the development and refinement of new techniques. The case studies carried out within Rolls-Royce by the YUTC have proven the feasibility of a number of new approaches in key process areas. However, as we argued earlier, the demonstration of the value of a technology does not necessarily imply the take-up of that technology unless both the production management and the engineers are convinced of the benefits of the change. Again, we have found that the case study approach has allowed both parties to assess the impact of new techniques 8
and provided the necessary confidence to initiate take-up. Perhaps the most important break through has been demonstrating to the company that an outside academic party can gain detailed understanding of the company s process and methods to allow stateof-the-art techniques to be introduced in a manner that supports the many diverse business constraints. The end result of these case studies is that the company has accepted the responsibility to champion the new techniques by spawning internal initiatives (creation of working groups) that will see through the adoption of the methods. This is an important development, fruits of the labour in some senses, and proof that the approach has facilitated sufficient technology transfer to convince the company that they have adequate grasp of the techniques to manage the implementation. 6. References [RTCA92] Software Considerations in Airborne Systems and Equipment Certifications, DO-178B, December 1992. [FENE95] Safety Cases for Software Application Reuse, P. Fenelon, T. P. Kelly, J. A. McDermid, in Proceedings SAFECOMP 95 September 1995 [GOMMA95] Reusable Software Requirements and Architectures for Families of Systems, H. Gomma, Journal Systems Software, 28:189-202, 1995. [LOCKE92] Software Architecture for Hard Real-Time Applications: Cyclic Executive vs. Fixed Priority Executives, C.D. Locke, The Journal of Real-Time Systems, Volume 4, pp. 37-53, 1992. 9