AutomataScales for digital triplets: Integrating humans, models, and data in early design

Early phases of engineering design require fast, explainable models that do not rely on reduced-order or black-box surrogates, which often omit essential physics needed by engineers. This paper presents advances in AutomataScales, a discrete, multi-fidelity cellular automata framework for coupled microchannel flow and basic heat transfer. To improve predictive fidelity, new supplemental transition and boundary condition rules are introduced and evaluated through cross-fidelity validation against a 3D ANSYS Fluent model using a 2D AutomataScales implementation. Accuracy is assessed using two design performance metrics: pressure drop and cold plate temperature. Across multiple test scenarios, AutomataScales is demonstrated to predict pressure drop with differences of 0.69–5.45% relative to the reference results. The results demonstrating reliable trend and magnitude capture for early-stage design decisions when operating within the framework’s standard practical regime, defined by an inlet-to-maximum velocity ratio between 0.5–1.99. In addition to the simulation advances, the study outlines a traceable workflow in which Python manages simulation execution and SysML-based representation used as a common language for structured documentation across multidisciplinary design activities. Within the scope of this work, this workflow is presented as a preliminary step toward a digital-triplet-based research framework that links physical validation, predictive simulation, and traceable project information for early-stage design exploration. The contributions of this work include: (i) an accuracy-enhancing rule extension strategy for discrete multiphysics simulation, (ii) quantitative benchmarking of a computationally lightweight 2D model against 3D CFD for key thermal fluid metrics; and (iii) a traceable workflow concept for more transparent early-stage design exploration.