Input reduction for nonlinear thermal surface loads

Abstract A multiplicity of simulations is required to optimize systems with thermal transient processes in the presence uncertain parameters. That why model order reduction applied minimize numerical effort. The consideration heat radiation and convection parameter-dependent transfer coefficients results a nonlinear system many inputs as these loads are distributed over whole surface limiting attainable reduced dimension. Therefore, new input method presented approximating matrix based on load vector snapshots using singular value decomposition. Afterward, standard methods like Krylov subspace or balanced truncation can be applied. Compared proper orthogonal decomposition, number training decreases significantly reduced-order provides high accuracy within broad parameter range. In second step, discrete empirical interpolation used limit evaluation nonlinearity few degrees freedom decomposition allows fast adaptation emissivity. As result, becomes independent original dimensions computation time drastically. This approach enables an optimal combination depending performed model.