Skeletal model reduction with forced optimally time dependent modes

Skeletal model reduction based on local sensitivity analysis of time dependent systems is presented in which sensitivities are modeled by forced optimally (f-OTD) modes. The f-OTD factorizes the coefficient matrix into a compressed format as product two skinny matrices, i.e. modes and coefficients. create low-dimensional, dependent, orthonormal basis capture directions phase space associated with most dominant sensitivities. These highlight instantaneous active species, reaction paths. Evolution equations for coefficients derived, implementation skeletal described. For demonstration, conducted constant pressure ethylene-air burning zero-dimensional reactor, new reduced models generated. laminar flame speed, ignition delay, extinction curve predicted compared against some existing literature same detailed model. results demonstrate capability to eliminate unimportant reactions species systematic, efficient accurate manner.