The effects of modeling requirements in early phases of buyer-supplier relations

Författare: Göran Backlund; Linköpings Universitet; []

Nyckelord: ;

Sammanfattning: This work is an attempt to attack waste of resources during development of embedded systems. Modern systems, in aircraft or automobiles, constantly grow in complexity and also become more integrated. The task of specifying system requirements really shows how crippled the human mind is when we try to master the complexity by the use of natural language. In order to clearly resolve the requirements, other methods need to be employed.Focus on system requirements means focus on the early phases of systems development. It is particularly crucial to clearly understand the requirements in collaborative development, since the system requirement specification usually is a part of the contractual agreement. The research questions are: How shall models be used in collaborative systems development? What is the effect of using models?The framework of reference for this work is the product development process, collaborative development and requirements analysis. The standard systems development processes are studied and their importance, apart from providing an orderly way of conducting development and establishing a common terminology, is the fact that they are needed as a tool for measuring organizational efficiency. Largely due to complexity, outsourcing of systems development is common in the aircraft industry - the buyer cannot have all required competence in-house. Consequently, parts of the development work is outsourced to suppliers who share risk, and this collaborative development leads to close relations between buyer and supplier. The airworthiness certification process is a contributing factor; it is not easy to change suppliers. The collaborative development relies on commercial contracts where the system requirements are a central part. This motivates focus on the requirements analysis phase; well understood and communicated requirements are a prerequisite for a successful collaborative development project.It is also a part of the lean philosophy - always attack waste of resources - to bring in the supplier early, and to focus on the early phases. Good requirements must have specific characteristics, hut if we try to specify complex systems in natural language, the sheer number of requirements represent complexity that limits the overview, and characteristics like completeness, consistency and unambiguity are often violated. Therefore, formal methods are needed for specification since they force the engineers to state the required functionality exactly. There are formalisms available, like Statecharts or ROOMcharts, that can capture the functional behavior of a complex system, and offer testing of the specification since the resulting model is executable.Management is responsible for process improvements, hut often lack tools for decision support; justification for investment in a process change. There have, until recently, not existed any tools for simulation of process improvement, for highly iterative processes. Design Structure Matrix is a method used in this work that actually can simulate such processes.The use of requirement models in collaborative systems development, and the resulting effect, is studied in two cases. These cases are subprojects in the Gripen fighter program at Saab AB, where two other organizations are suppliers to the projects. One case represents a situation before introduction of modeling requirements, and the other case represents the after situation, since modeling is already a natural way of working in that case. Modeling was tested in one of the cases, within the framework of this study. The results are positive, from both cases, and are summarized as recommendations to future projects. The results prove the current text-based requirement specifications to be inadequate and to cause problems later in systems development.Model integration efforts, prior to this work as well as efforts executed in this work, are summed up and result in recommendations to future projects.A method for simulating the effects of process improvement was tested on one of the cases. The Design Structure Matrix (DSM) method was applied to a collaborative systems development process, which was derived from interviews in the case studies. The process change studied was the introduction of requirements modeling, and the result was a substantial reduction of total project time. This result was corroborated by study­ ing problem reports in that project, since many of the detected problems were of the kind that they probably had been detected early through modeling of functional requirements. The other case, which early adopted a model-based approach, actually found some design errors, and due to their early detection the result was a reduction of project time. However, it is difficult to account for the problems "that never occurred", since many of them were avoided very early in the project, prior to any problem reporting having been established.Finally, a "Digital Mock-Up" room, which is a concept for model-based design currently applied to airframe development, is recommended for carrying over to the systems and software engineering areas. In such rooms the systems development teams can meet around requirements models, performance models, etc. and actually make the models the foundation and design base of their decisions and engineering work.

  Denna avhandling är EVENTUELLT nedladdningsbar som PDF. Kolla denna länk för att se om den går att ladda ner.