We first review a classical scale-similarity model used to simulate the mot
ion of large eddies in a turbulent flow. The kinetic energy balance of this
model is very unclear in theory. Experiments with it often have reported t
hat an additional Smagorinski type subgridscale term is needed. This term i
s not benign; it can alter significantly the predicted long term dynamics o
f the large eddies. However, we also show that the principal of scale-simil
arity (introduced in 1980 by Bardina, Ferziger and Reynolds) can also give
rise to other scale similarity models which have the correct kinetic energy
balance. (C) 2000 Elsevier Science Ltd. All rights reserved.