Testing activities to gain specific design knowledge play an essential role in engineering design, when a structure needs to be developed, for which knowledge from analytical models or documentation is missing. As research into these testing activities is difficult, few insights into successful approaches exist.

In this contribution, we investigate testing activities to gain specific design knowledge through a laboratory task, where 10 engineering students optimize a system using a web-based process chain including rapid prototyping and testing. Design and testing data are acquired from 110 prototypes in 3 hours. A differentiation of performance is conducted and approaches of high- and low-performers are investigated to identify patterns.

Based on these patterns, hypotheses, and metrics indicating successful and non-successful approaches are derived as basis for development of metrics for testing to gain specific design knowledge. A successful approach was overstep the limit, where participants accept destruction of their system to identify boundaries. An unsuccessful approach was the change of many parameters in later tests. These hypotheses and their metrics can then be used in development and validation of support for testing.