Reduction of thermal models

Reduction techniques for nodal models of the linear thermal systems are reviewed, and applications to problems involving finite wall structures are illustrated. Several modal techniques involving the introduction of matrices into the equation of state are examined, and the method of Eitelberg (1980), in which the weighted equation error between the complete and reduced models is optimized by solving a Liapunov equation, is described in detail. The latter approach is used to obtain reduced versions of a 10-node model of a building wall subjected to steady, single-pulse, or periodic heating. In one case the model is successfully reduced to a single node corresponding to its internal surface.