Global sensitivity analyses for test planning with black-box models for Mars Sample Return.

Area of Science:

• Planetary Science
• Computational Modeling
• Data Analysis

Background:

• The Mars Sample Return program requires robust experimental test planning under resource constraints.
• Complex black-box models are essential for simulating Martian sample return missions but present analysis challenges.

Purpose of the Study:

• To develop a systematic workflow for sensitivity analysis of complex black-box models.
• To provide quantitative insights into uncertainty drivers, impact direction, and interactions.
• To optimize experimental test planning for the Mars Sample Return mission with limited resources.

Main Methods:

• Application of optimal transport-based global sensitivity measures for multivariate outputs.
• Utilized sensitivity measures that accommodate dependent model inputs for univariate outputs.
• Employed multifidelity techniques to accelerate calculations using low-fidelity models while ensuring accuracy with high-fidelity samples.

Main Results:

• The sensitivity analysis successfully identified key drivers of uncertainty and their interactions.
• The workflow provided insights into model behavior, guiding focused experimental testing.
• Multifidelity approaches significantly sped up computations while maintaining accuracy.

Conclusions:

• The developed sensitivity analysis workflow is effective for complex models in resource-limited scenarios.
• This approach enhances understanding of model behavior and optimizes experimental design for mission success.
• The methods are applicable to similar complex modeling and experimental planning challenges in space exploration.