Abstract
Reproduction of temperature distribution in furnaces of different size is shown, by dimensional analysis, to require direct proportionality between power input and linear dimensions of the furnace, if transfer of heat by conduction is a controlling factor. Experiments with scale models indicate the importance of correct temperature gradients as a criterion of similarity. A 1:5 linear scale model of an industrial furnace for calcining anthracite has been successfully run according to this criterion. The principle has been extended to electric smelting furnaces with good results.