Enhancing Engineering Pedagogy through Sustainability-Driven Design Projects from Theory to Practice

Abstract

Engineering capstones increasingly require students to deliver solutions that balance economic and environmental objectives under data scarcity. This paper proposes a guarantee-aware pedagogy that frames each project as a bi-objective program with explicit uncertainty sets and a lightweight correctness backbone. The pipeline comprises p-group reduction and factor screening; admissible surrogate modeling with a nearest-PSD repair for quadratic fits; e-constraint generation of representative Pareto sets (augmented to include selected non-supported points); and a transparent MCDA/LCA decision audit. Uncertainty is handled via three teachable Dials-Scenario (sample-average with concentration bounds), Budgeted-Robust (price-of-robustness parameter G), and Fuzzy a-cuts (a-dominance bands)-with an optional Wasserstein DRO extension. We define a single Credibility Index C = coverage – overfit that combines calibration and parsimony, and pair it with learning gains measured by Hedges’ g and optional 2PL IRT. A compact demonstrator (three e-levels) shows how the method yields interpretable Pareto fronts and auditable choices, while uncertainty dials trade protection for cost in predictable ways. Complexity tags and resource scheduling rules (assignment TU, queueing r < r\*) keep workloads feasible for classroom scale. Results indicate that the approach raises methodological transparency, improves reproducibility, and supports defensible suctainability-driven design decisions within a single semester.