Global Sensitivity Analysis with Small Sample Sizes: Ordinary Least Squares Approach.

Area of Science:

• Chemical Engineering
• Combustion Science
• Computational Chemistry

Background:

• Complex chemical-kinetic mechanisms are essential for modeling combustion processes.
• Accurate identification of sensitive reactions is crucial for mechanism reduction and understanding.
• Existing global sensitivity analysis methods often require large sample sizes, limiting their application in complex systems.

Purpose of the Study:

• To develop a novel global sensitivity analysis (GSA) method for efficient identification of key chemical reactions.
• To enable accurate ordering of sensitivity coefficients for the most influential reactions in complex chemical-kinetic mechanisms.
• To facilitate the study of chemistry in realistic combustion devices with reduced computational cost.

Main Methods:

• Integration of GSA with statistical, machine learning, and optimization tools.
• Development of a calibration strategy for small sample sizes under realistic combustion device conditions.
• Application of multiobjective optimization principles to manage the trade-off between false positives and false negatives in reaction ordering.

Main Results:

• Achieved accurate ordering of sensitivity coefficients for the top 10-30 most sensitive chemical reactions.
• Demonstrated a reduction in required sample sizes by approximately a factor of 10 compared to previous algorithms.
• Successfully calibrated the method using operating conditions from a compression ignition engine.

Conclusions:

• The new GSA method offers a significant advancement in analyzing complex chemical kinetics with reduced computational resources.
• The developed calibration strategy effectively handles challenges associated with low-temperature combustion chemistry.
• This approach is highly valuable for studying and optimizing chemistry in realistic combustion devices.