A Graph-Based Approach to Selection of Feasible Compositions for Compositionally Graded Alloy Design

Abstract

Compositionally graded alloys (CGAs) offer a promising solution for engineering components that must perform under spatially varying conditions and requirements. However, selecting the appropriate compositions along a gradient is challenging because of potential chemical incompatibility, mechanical mismatch, or thermal expansion differences, which can compromise manufacturability or part performance. This study expands on recent advances in graph--based computational approaches for CGA design by addressing the question of how to work with a design space containing multiple isolated regions of feasible compositions. Using the quinary Nb--Cr--V--W--Zr system, the thermal and mechanical properties were computed using a combination of CALPHAD simulations, rule--of--mixtures models, and empirical estimates. The resulting data were embedded into a labeled property graph (LPG) structure and filtered using constraints that reduced the potential for solidification--related defects, promoted phase stability across the gradient, and ensured processability via additive manufacturing. A set of constraints produced four isolated subgraphs containing groups of connected compositions that were further evaluated using key property metrics to highlight their respective strengths and weaknesses in the context of CGA design and manufacturing. Overall, this work establishes methods for subgraph analysis and selection to provide guidance on narrowing design spaces down to one traversable group of compositions that aligns with performance goals and manufacturing constraints for any CGA design scenario.