A Multidisciplinary Design and Optimization (MDO) Agent Driven by Large Language Models

TL;DR

This paper tackles the inefficiency and innovation bottlenecks in traditional mechanical design by introducing a Large Language Model–driven Multidisciplinary Design and Optimization (MDO) Agent that semiautomates the design workflow from natural-language intent to verified, optimized CAD models. The approach combines NL-driven parametric modeling, Retrieval-Augmented Generation for knowledge grounding, and ReAct-style orchestration of engineering tools to perform end-to-end design, verification, and optimization. Validation across a gas-turbine blade, a machine-tool column, and a fractal heat sink demonstrates substantial reductions in manual scripting and setup effort, with automation levels achieving roughly 70–80% in modeling and FEA tasks and 100% script generation in optimization steps, while enabling broader exploration of design concepts. The work highlights a practical path toward human–AI collaborative mechanical engineering and outlines future directions for domain-specific LLMs, improved multimodal reasoning, and manufacturability integration.

Abstract

To accelerate mechanical design and enhance design quality and innovation, we present a Multidisciplinary Design and Optimization (MDO) Agent driven by Large Language Models (LLMs). The agent semi-automates the end-to-end workflow by orchestrating three core capabilities: (i) natural-language-driven parametric modeling, (ii) retrieval-augmented generation (RAG) for knowledge-grounded conceptualization, and (iii) intelligent orchestration of engineering software for performance verification and optimization. Working in tandem, these capabilities interpret high-level, unstructured intent, translate it into structured design representations, automatically construct parametric 3D CAD models, generate reliable concept variants using external knowledge bases, and conduct evaluation with iterative optimization via tool calls such as finite-element analysis (FEA). Validation on three representative cases - a gas-turbine blade, a machine-tool column, and a fractal heat sink - shows that the agent completes the pipeline from natural-language intent to verified and optimized designs with reduced manual scripting and setup effort, while promoting innovative design exploration. This work points to a practical path toward human-AI collaborative mechanical engineering and lays a foundation for more dependable, vertically customized MDO systems.

Figures (15)

• Figure 1: Schematic of the MDO Agent. The framework couples a Designer, Modeler, Verifier, and Optimizer in a closed loop, grounded by RAG for concept generation, text‑to‑CAD for parametric modeling, and ReAct‑style tool orchestration for FEA and optimization, with a human‑in‑the‑loop for guidance.
• Figure 2: Text‑to‑geometry mapping. Natural‑language intent is translated into structured parametric code and geometric features to construct editable CAD models.
• Figure 3: Interactive editing mechanism. The agent refines geometry through natural‑language edits that are compiled into code modifications and reapplied to the model.
• Figure 4: Design fusion mechanism. The agent merges elements from multiple parametric designs under natural‑language guidance, reconciling geometry and constraints to form a unified model.
• Figure 5: Chain‑of‑thought–based task decomposition. High‑level goals are broken into ordered sub‑tasks, each mapped to code generation or tool calls, enabling robust execution on complex geometry.