The design of products for sustainability involves holistic consideration of a complex diversity of objectives and requirements over a product’s life cycle related to the environment, economics, and the stakeholders in society. These objectives may only be considered effectively when they are represented transparently to design participants early in a design process. Life Cycle Assessment (LCA) provides a credible prescription to account for environmental impacts. However, LCA methods are time consuming to use and are intended to assess the impacts of a completely defined design. Thus, more capable methods are needed to efficiently identify more sustainable design concepts. To this end, this work introduces a fundamental approach to formulate models for normative decision analysis to accurately account for these multiple objectives. Salient features of this novel approach include the direct accounting of the LCA formulations via mathematical relationships and their integration with derived expressions for compatible life cycle cost models, as well as a methodical approach to account for significant sources of uncertainty. Here, a semantic ontological framework integrates the information associated with decision criteria with that of the standards and regulations applicable to a design situation. Since this framework shares the context and meaning of this information and design rationale across domains of knowledge transparently among design participants, this approach can influence a design toward sustainability considerations while the design complies with regulations and standards. Hypothetical equivalents and inequivalents method is represented and deployed to consistently model a designer’s preferences among the criteria. Material selection is a very significant factor for the optimal concept selection of a product’s components. A new method is detailed to estimate the impacts of material alternatives across an entire design space. Here, a new surrogate model construction technique, which is much more efficient than the construction of complete LCA models, can prune the design space with adequate robustness for near optimal concept selection. This new technique introduces a feasible approximation of a Latin Hypercube design at the first of two sampling stages to overcome the issues with sampling from discrete data sets of material property variables.