Early robust design (RD) can lead to significant cost savings in the later stages of product development. In order to design systems that are insensitive to various sources of deviation in the early stages, specific design knowledge (SDK) plays a crucial role. Different design situations result in higher or lower levels of derivable SDK, which leads to different activities to achieve the development goal. Due to the variety of design situations, it is difficult for design engineers to choose a more robust concept to avoid the costly iterations that occur in the later development stages. Existing RD methods often do not adequately support these differences in design situations. To address the problem, this paper outlines an adaptive modeling method using the Embodiment Function Relation and Tolerance model. The method is developed in two contrasting design situations, each with a high and low level of derivable SDK, and evaluated in another two corresponding case studies. It has a consistent structure with five stages and gates. At each stage, the derivable SDK is taken into account and the individual modeling steps are adapted. This method provides design engineers with concrete support for early robustness evaluation of a product concept in different development scenarios.