|M.Sc Thesis||Department of Architecture and Town Planning|
|Supervisor:||Prof. Emeritus Goldschmidt Gabriela|
|Full Thesis text - in Hebrew|
This work is a study of the usage of various visual representations in an engineering context - and more specifically, problem solving by teams in the semiconductors industry. We investigate the contribution of visual representations to teams’ mental models and to a shared understanding of the problem and a vision of its solution.
The visual representations used in problem solving processes include generic and domain-specific diagrams (e.g., microscope pictures of the device at numerous magnifications, and GDS - a schematic layout of the device); we compared the use of various diagrams by integrated problem solving teams.
16 teams were asked to solve two typical manufacturing problems, similar to the ones they encounter in their routine work in a foundry. The data the teams received differed in terms of the visual material provided. The solutions the teams arrived at were assessed by three naïve judges who are experienced senior engineers in the foundry.
The results showed that whereas the visual data had a demonstrated impact on teams’ performance and helped create their shared mental models of the problems, different kinds of diagrammatic representation contributed to a different extent to the process of problem solving. Surprisingly, GDS diagrams contributed next to nothing and many participants reported that they were not able to understand them and use them.
We proceeded to interview another group of process engineers from the same foundry about their experience with GDS layouts. Here too, we found that even experienced engineers reported difficulties using these diagrams. We checked their responses against academic background, gender and amount of experience and found no correlation with the ability to understand and use GDS layouts.
We conclude that a separation between design and manufacturing which exists in the semiconductors industry has caused the two to adopt different representational conventions, where domain-specific diagrams like the GDS layout are in use only at the design end of the industry. At the manufacturing end the technical personnel is not proficient in the use of diagrammatic and other domain-specific standard visual representations of industry. This affects teams’ problem solving efficacy in terms of solution quality and time to solution, with significant financial consequences. This state of affairs raises questions regarding the use of domain-specific diagrams and the degree to which it is possible to integrate them into dispersed and multi-disciplinary user contexts, where shared mental models based on them are a key to effective work.