Statistical properties of the subgrid-scale stress tensor, the local e
nergy flux and filtered velocity gradients are analysed in numerical s
imulations of forced three-dimensional homogeneous turbulence. High Re
ynolds numbers are achieved by using hyperviscous dissipation. It is f
ound that in the inertial range the subgrid-scale stress tensor and th
e local energy flux allow simple parametrization based on a tensor edd
y viscosity. This parametrization underlines the role that negative sk
ewness of filtered velocity gradients plays in the local energy transf
er. It is found that the local energy flux only weakly correlates with
the locally averaged energy dissipation rate. This fact reflects basi
c difficulties of large-eddy simulations of turbulence, namely the pos
sibility of predicting the locally averaged energy dissipation rate th
rough inertial-range quantities such as the local energy flux is limit
ed. Statistical properties of subgrid-scale velocity gradients are sys
tematically studied in an attempt to reveal the mechanism of local ene
rgy transfer.