Tool Framework for Preliminary Structural Design Using Beam Models in Collaborative Lightweight Vehicle Development

Abstract

Early-stage vehicle structural development requires fast iteration while still providing meaningful insight into load paths, stiffness trends, and mass sensitivity. Reduced-order beam models enable rapid system-level assessment, but their coarse representation limits the evaluation of localized structural behavior and the systematic refinement of subsystems. At the same time, fully detailed global models typically conflict with concept-phase requirements due to modeling effort and computational cost. This thesis presents the methodology and proof-of-concept (PoC) implementation of a modular, web-based framework for conceptual beam-based structural modeling and solver-driven evaluation. The platform follows an artifact-first integration approach with explicit interface contracts to support tool replaceability, traceability, and automation readiness. Individual tool modules are encapsulated in isolated execution environments and connected through RESTful APIs, allowing parallel development and technology-neutral integration of heterogeneous services. A beam-based global master model is introduced as a shared reference representation, together with an automated pipeline for model generation, solver execution, and standardized response extraction. In addition to a direct global optimization workflow, a sequential global–local strategy is formalized based on subsystem extraction, displacement-based boundary transfer, and reinjection principles. A standardized numerical demonstrator, including load and boundary-condition setup, design-variable parameterization, and a constrained mass-minimization problem statement, is defined to enable a consistent comparison of both approaches. The study highlights practical considerations and trade-offs of global versus global–local optimization in a concept-phase setting, including modularity, feasibility robustness, and the interpretability of subsystem-level refinements. Limitations of the current PoC are discussed, including simplified joint modeling and linear analysis assumptions, and directions for future work are outlined, such as asynchronous job execution, extended response measures, and tighter coupling strategies for global–local iteration.