Design space exploration

Eliciting and analyzing conceptual designs

In a project with collaborators from the Systems Engineering Design (SED) group in Sweden and the University of Auckland, we set out to explore conceptual design spaces in a systematic and quantitative manner to narrow them down to a handful of feasible and valuable concepts for further development and detailed design. We pose the question:

How can we analyze conceptual designs and quantify their ability to absorb a change in design requirements?

To answer this question, we borrowed concepts from design margins because of their ability to act as a buffer against change in design requirements . We used an Enhanced Function-Means (EF-M) tree to enumerate possible concepts and analyze their design margins .

We used the example of an aeroengine component to demonstrate our proposed concept synthesis and analysis method. We enumerated four concepts by combining the available vane cross-sections with the lean configuration of the struts shown below. Design margins are present in each concept due to the discrete nature of each concept’s available means (e.g., choices for vanes and lean angles). The more margin is present, the more change can be absorbed but the worse is the performance of the strut (e.g, the weight and manufacturing costs of the strut). These effects are described by Change Absorption Capability (CAC) and Impact on Performance (IoP), respectively.

We use a library which implements the margin value method to quantify these effects and plot the results on a Parallel Coordinates Plot (PCP) and scatter plot for comparison.

lean_0
lean angle θ=0o
lean_0
lean angle θ=30o
thin vane
height = 15mm
vanethick
height = 17mm
Try hovering over the rows above to visualize each conceptual design.

The most valuable design is the one that corresponds to the red line in the PCP and red dot on the scatter plot. It is trivial to analyze and choose the most suitable design(s) when our selection is limited to just four concepts. However, our method is most useful when the number of combinations grows large. We demonstrate this on an example with 6,552 concepts shown below. Using the PCP, or the scatter plot, the designer can pinch and narrow their selection down to a few candidates.

PCP of more than 3000 different conceptual alternative. Try pinching each vertical axis by clicking and dragging. You can also drag a selection box across the scatter plot to filter designs. You can also reorder the vertical axes by dragging the axis title.

publication   library