Predictive Reduced Order Modeling of Chaotic Multi-scale Problems Using Adaptively Sampled Projections

An adaptive projection-based reduced-order model (ROM) formulation is presented for model-order reduction of problems featuring chaotic and convection-dominant physics. efficient method formulated to adapt the basis at every time-step on-line execution account unresolved dynamics. The ROM in a Least-Squares setting using variable transformation promote stability robustness. strategy developed incorporate non-local information adaptation, significantly enhancing predictive capabilities resulting ROMs. A detailed analysis computational complexity presented, validated. shown require negligible offline training naturally enables both future-state parametric predictions. evaluated on representative reacting flow benchmark problems, demonstrating that ROMs are capable providing accurate predictions including those involving significant changes dynamics due variations, transient phenomena. key contribution this work development demonstration comprehensive targets capability chaotic, multi-scale, transport-dominated problems.