Talon-P: Decision-Tree-Driven Multi-Fidelity Testing for Engineering Cycle Time Reduction

This paper is a contribution to a special session focused on the analysis and optimization of a vehicle inspired by the Stratolaunch Talon-A vehicle, the baseline case of which we will refer to as Talon-P. The work presents a decision-tree-driven framework for optimizing multi-fidelity testing strategies in hypersonic vehicle development, demonstrated through application to the Talon-P vehicle. The framework addresses critical cycle time reduction needs in the design-test-build-fly process while managing costs and risks. The methodology systematically integrates three fidelity levels: (1) Configuration-Based Aerodynamics for low-mid fidelity rapid screening, (2) Kestrel Computational Fluid Dynamics for detailed high-fidelity analysis, and (3) wind tunnel experiments using the Boeing/AFOSR Mach-6 Quiet Tunnel at Purdue University. The decision framework employs entropy and information gain metrics weighted by time and cost factors across the preparation, testing, and analysis phases. A nose geometry design alternatives use case demonstrates the framework’s application, with comprehensive time tracking revealing substantial resource disparities: Configuration-Based Aerodynamics (5.4 hours), Kestrel Computational Fluid Dynamics (662 hours), and wind tunnel testing (2,557 hours). Results demonstrate a 3.1% time saving when using a sequential testing approach compared to a simultaneous approach. Additionally, parameter sweep analysis reveals that optimizing agreement thresholds can achieve up to 5.3% time savings, with maximum efficiency occurring at 120 cases using a 0.78 agreement threshold. The framework provides a systematic approach to fidelity selection that reduces reliance on expert judgment while optimizing resource allocation in time-critical hypersonic vehicle development programs.