This paper describes the challenges faced by aerospace avionic systems developers in the 1990s, as they use ‘in anger’ both best-practice processes and state-of-the-art tools to try and achieve that most elusive goal in a large system development: Re-useable designs, meeting all customer needs, on-time, and within budget. The Boeing 777 electrical load management system (ELMS) developed by Smiths Industries Civil Systems, Cheltenham, has involved up to 250 engineers since 1991. It is an example of military technology being applied in the civil sector, and integrates over 50 aircraft utility systems as well as managing the supply of electrical power. The author has been Engineering Manager for this system since its inception, and takes this project as an example of large system development. He shares the advantages of those process approaches and types of tool which worked, and the disadvantages of those approaches which were less successful. The objectives of this paper are: • ⊗ To share practical experience of current methods and types of tool for requirements capture, structured analysis, object-oriented design (which is very different to methodologies used only a decade ago), verification/validation and certification. • ⊗ To discuss where these modern methods and tools assist reuseability, and where they impede it. • ⊗ To describe the fresh management challenges they present, such as lack of visibility (-it's all in the system, if you know how to access it!). • ⊗ To examine the design review strategies which and most value with these methods. (More peer-level reviews; less large formal meetings.) • ⊗ To review how well these modern methods cope when faced with customer specification changes — both small changes and large-scale changes. (Is there much value in huge volumes of traceability information if it is all to be scrapped by a large change?) The paper also emphasizes the human elements in a successful development. Examples are the value of co-locating good engineers with customers (and good customer engineers within the developer's teams) and how this compares with technology such as datalinks, conference calls and videoconferencing; the values of concurrent engineering within the project; and good communication at all levels (technical and managerial). The author believes that even with best-class processes and tools, you still need good engineers to have a successful outcome!
Read full abstract