TL;DR
This paper explores the application of Goal-Oriented Requirements Engineering (GORE) to bridge the gap between aircraft design and manufacturing. Through three iterative attempts with industry experts, the authors demonstrate how GORE can help model and understand the complex dependencies between these two domains, leading to more integrated and potentially optimized development processes.
What Problem Does This Paper Address?
Traditionally, aircraft development follows a sequential process: design first, then manufacturing. This often leads to the manufacturing system being constrained by early design choices, potentially resulting in suboptimal performance. To address this, the aerospace industry is moving towards collaborative and concurrent engineering approaches, where design and manufacturing teams work together earlier in the process. However, effectively managing the diverse expectations and intricate dependencies between these teams is a significant challenge. This paper tackles this problem by investigating how Goal-Oriented Requirements Engineering (GORE) can be used to model these interdependencies and foster a shared understanding between aircraft designers and manufacturing engineers.
Methods
The authors employed a three-stage iterative approach to apply GORE with industry experts from Airbus.
- First Application (Free Notation): This initial phase focused on building a common vocabulary and a high-level understanding of goals and dependencies between aircraft design and its industrial system. Experts used a language-free, online collaborative tool, relying on colors and notes to represent concepts. This resulted in a large, informal model.
- Second Application (iStar Language): Building on the first attempt, the researchers introduced the iStar GORE language. This provided a more formal structure for modeling goals, tasks, resources, and dependencies. The goal was to refine the model and improve its precision.
- Third Application (Narrower System of Interest): To manage complexity, the focus was narrowed to the aircraft nose section and its assembly line. This allowed for a more detailed and manageable model, leveraging the experience gained from the previous stages. The iStar 2.0 language was used, with a focus on understanding key elements of assembly line performance for a given nose design.
Datasets
Not discussed in the paper. The study relies on expert knowledge and iterative modeling sessions rather than pre-existing datasets.
Key Insights and Ideas
- Bridging Silos: GORE proved effective in helping experts, who traditionally work in silos, gain a global perspective of the co-design activity and understand inter-domain goals and dependencies.
- Eliciting Motivations: The GORE approach helped experts articulate the underlying motivations behind their activities, which are often missing in formal company processes.
- The Challenge of Abstraction: Experts often struggled to think at the "goal level," tending to jump directly to solutions or tasks. This required significant guidance and iterative refinement.
- Language Matters: While a free-notation approach can be useful for initial elicitation, a formal GORE language like iStar significantly improves the structure, clarity, and expressiveness of the model.
- Complexity Management: The complexity of modeling the entire aircraft and its industrial system was a major hurdle. Narrowing the scope of the system of interest was crucial for creating a more usable and insightful model.
- Trade-offs are Key: The concept of "trade" relationships, representing the need for negotiation between conflicting goals, emerged as important in the free-notation phase and was addressed in the iStar translation.
Key Results
- First Application: A large, informal goal model with 123 goals and 120 relationships was created, providing a broad overview but proving difficult to exploit due to its size and lack of formal structure.
- Second Application: An iStar model with 69 elements (30 goals/soft goals, 24 tasks, 15 resources) and 96 relationships was developed. This model was more structured and focused but still faced complexity challenges.
- Third Application: A more refined iStar 2.0 model was produced for the aircraft nose assembly line. This model, with seven actors and detailed dependencies, was considered more usable and facilitated discussions about potential automation of certain tasks.
Limitations and Future Work
- Expert Familiarity: The initial difficulty experts had with GORE concepts highlights the need for better training and integration of these approaches into industrial practice.
- Model Complexity: Even with a narrower focus, managing the complexity of dependencies remained a challenge. The paper suggests that iStar's limitations, such as the inability to represent dependencies between tasks of the same actor, required workarounds like duplicating soft goals or splitting actors.
- Future Work: The authors propose adapting GORE to better support decision-making by exploring alternative vocabularies (e.g., "criteria" instead of "soft goals") and more flexible dependency rules. They also suggest that further research could explore how to automate parts of the design process based on these goal models.
Summary updated