It is accustomed to call robust design a design that is resilient to noise. A product could be designed to be robust by methods such as Taguchi's. The idea is to manipulate the design parameters that could be controlled by designers to minimize the effect of the noise on the planned behavior in the designated environment. We are concerned with a broader perspective of robustness, one that arises from many environmental uncertainties including those related to technical knowledge, customers, and market conditions. In such interpretations, the product behavior includes physical behavior as well as customer satisfaction, cost, as well as any parameter that is related to the technical and market success of the product. In this context, we define a product as robust if a large variety of potential environmental uncertainties have little impact on its behavior.
While the broad perspective of robustness applies to all design stages, we are in particular interested in the conceptual design stage that is considered to be the most critical step in product development. In this stage, an abstract description of the product is created that serves as the basis for subsequent design stages and decisions. To a large extent, the quality of the product concept determines the fate of the product. In (Ziv Av and Reich, 2005) we presented a method – SOS (subjective objective system) – for generating optimal concepts in diverse disciplines. In this work, we extend SOS to generate robust product concepts.
In the context of SOS, robustness is defined as the stability of the optimal concept or configuration generated by SOS with respect to (1) variations in designers’ subjective judgment, (2) variations in available technology, (3) variations in organization context, and (4) variation in customers’ preferences. All these could have an impact on the results obtained by SOS. In order to assess the robustness, we run different tests with simulated changes and analyzed the results. For example, robustness with respect to designers’ judgment was tested by varying such judgment and checking the stability of the solution to such variations. Robustness with respect to customer preferences was calculated by sampling different preferences and finding their related optimal concepts. This data was subsequently analyzed to find robust concepts as well as risky concepts. For each preference we also analyzed the local robustness of the solution; that is, how much can we change the customer preferences from the available estimation and maintain the same solution. If these variations are large, our confidence in the solution increases. The robustness with respect to other variations is analyzed similarly.
In the context of SOS, we define two types of robust concepts. In general, robust concept is a product concept that remains stable as different evaluations in SOS varies due to different circumstances. The first type of robustness – global robust concept – is defined as the concept that is most prevalent if we let SOS input values vary randomly in their allowable range. This is an operational definition because it specifies the method to find that concept. The second type of robustness – local robust concept – is defined as a concept designed for a particular set of inputs and that is remain intact even if these input values change significantly from their present values. This is also an operational definition.
Since SOS automatically generates the concept from its inputs, we can run simulations with different input values and obtain the results that allow assessing the global and local robustness of a concept.
A last perspective on concept robustness arises from the concept of a product family. In doing so, we depart from the traditional work on product family and platform that mainly deals with complete or detailed designs. In contrast to others, we deal with product family and platform at the concept generation stage. In the context of SOS, instead of finding a design concept that is prevalent across the space of SOS input values as we did in the global robustness analysis, we define robust platform concept to be a product concept family that addresses several markets and whose common platform is almost completely specified. Consequently, implementing the platform concept in multiple markets involves minimal customization.