Large-scale fluidized beds for commercial processes commonly require heat transfer surfaces. Design then demands that heat transfer coefficients be specified. Empirical correlations are unable to cover the wide range of variables and conditions encountered. Mechanistic models are more reliable, but must be chosen carefully. For bubbling beds, the packet model approach gives reasonable predictions for the convective component of transfer, but further work is required to provide reliable estimates of two required time constants, dependent on the hydrodynamics. For industrial-scale circulating beds, a mechanistic model that incorporates the key factors influencing heat transfer, assumes fully developed transfer, and utilizes results from large-scale units is recommended. (c) 2004 Elsevier B.V. All rights reserved.