ABSTRACT
Thermal and fluid mechanics problems are governed by partial differential equations. The numerical resolution of such systems usually requires a fine spatial discretization of the corresponding domain. This yields to a large number N of degrees of freedom and consequently to a high CPU time consumption. When the objective of the modeling is the inversion of experimental data or the real-time control of a process, this drawback may become an impossible task to accomplish. In order to circumvent this issue, a helpful approach is to find another model that reproduces accurately enough the behavior of the system, but with a much lower number of degrees of freedom n(n N). Model reduction methods provide interesting solutions to obtain such low-order models.
