Summary of Ph.D. Thesis

Mehmet H. Göker

My Ph.D. thesis ('Einbinden von Erfahrung in das Konstruktionsmethodische Vorgehen') deals with the development, the influence and utilization of experience in the engineering design process as well as the development of methods to capture, store and reuse design experience. The basic idea behind it is that engineers, like all other human beings, reason and thus design by means of their experience. Our experience enables us to understand our surroundings, classify our observations and generate solutions to our problems.

To be able to capture and utilize the immense amount of knowledge that distinguishes an expert designer from a novice and to better understand the experience based problem solving approach human beings employ, I conducted experiments in cognitive-psychology and analyzed how the experience of a person develops, is used and influences the way problems are solved. The effects of experience were analyzed both on a behavioral level through observation and "loud thinking", and on a physiological level by means of EEG measurements and coherence calculations. The results showed that the test-subjects preferred to re-use known solution elements instead of generating new solutions and that the regions that are being activated in the human brain depend on the experience a person has with a similar problem.

Nevertheless, this experience-based problem solving approach is neglected, or even prohibited ("solution-neutral thinking"), in current engineering design education. Engineering Design Methodologies (EDMs) are used as a prescriptive framework and serve as a kind of Design Grammar. Similar to a grammar of a language EDMs provide structures and methods by means of which existing designs can be analyzed and new designs generated. However, trying to design by using EDM only is like trying to speak a foreign language with a grammar and a dictionary. It works, the results are good too, but that is not the way people like to speak.

Based on the results of the behavioral experiments, Case-Based Reasoning was selected as a suitable and "cognitively adequate" method to simulate and support Experience-Based Problem solving on a computer. The results of the behavioral experiments also gave clues as to how design experience should be indexed, structured, retrieved, represented, and evaluated in a case-based reasoning system.

By using the methods and structure provided by engineering design methodology as a design grammar and the case-based reasoning approach to provide the framework for experience based problem solving, I developed a system architecture for a conceptual design assistant and implemented the major elements of this architecture as prototypes. The goal of the system is to support the design engineer both in terms of methodology and with experience during conceptual design. During my research I considered it most important to find a way to extract, transform and to use knowledge in a way that comes 'natural' to the human being.