Automated systems for variant design can be used for design iterations in order to guide the designer towards solutions that are optimized with respect to weight, cost, lead time, or other vital properties. In this work such a system for computational design problems is presented together with examples of its application. The system performs design computations, computed aided design model configuration, production process planning, and cost estimation. The design rules and algorithms are captured in knowledge “chunks,” which are human readable as well as computer executable. The workflow governing the execution of these rules and algorithms is created using a dependency structure matrix (DSM) which is included in the system. Particular attention has been given to the need for transparency, modularity, and longevity of the system, which is a prerequisite for such a system to become a viable tool in industrial applications. Experiences from the proposed system indicate that the DSM workflow manager in combination with a human readable and modularized knowledge base provides clarity and transparency for both developer and user of the system.

Customer requirements provide objectives and constraints for all phases of the product development process. For complex system products with a high degree of customization, it is no mean task to ensure that the right persons at the right time have ready access to the selected requirement specifications that they should adhere to and strive to satisfy. In the present work a systematic sequence of development phases associated with computer implemented information structures for requirements, functions and systems have been studied at three companies with widely different products and business scenarios. Although individual adaptations are required, the overall processes for requirement decomposition and propagation appear surprisingly similar. To fully exploit the potential of such information systems, many companies would benefit from also including knowledge structures in their product models. Fundamental product and process knowledge often evolves slowly over time, can be gradually upgraded and be reused many times. It also constitutes one of the company’s most valuable assets and should be carefully maintained and enhanced. In the work presented here a few different approaches to integrating knowledge structures into the total product information structure have been developed and exemplified for the three companies studied.

Choosing methods and tools to automate engineering design tasks has farreaching implications. A system that is first established for a limited task may later require being scaled up and maintained for many years of operational life. As the system grows, the company’s reliance on the system increases and the need for a stable platform to start from becomes critical. The system must be tailored to the specific design process and knowledge structure at hand. The purpose of this work is to present a structured methodology to classify design task(s) and to give guidelines for how this definition should be linked to suitable solution strategies and computerised tools.