The experimental validation of the available codes and nuclear data to calc
ulate correctly the effects of streaming paths in the shield of a fusion re
actor was provided by a streaming experiment carried out at the 14 MeV Fras
cati Neutron Generator (FNG). The experiment consisted in the irradiation o
f a mock-up of the shielding blanket (stainless steel and water equivalent
material, 1 m thick) and of the toroidal field coil. A void channel with hi
gh aspect ratio and a cavity at the end of the channel were realised in the
bulk shield. Some nuclear quantities were measured in the cavity at the en
d of the channel and behind it up to the simulated superconducting magnet.
The neutron flux was measured using the activation foil technique while the
nuclear heating was measured using thermoluminescent detectors. The experi
mental data were compared with the results of calculations performed by the
Monte Carlo code MCNP-4B, using the Fusion Evaluated Nuclear Data Library
(FENDL, version 1.0, 2.0) and the European Fusion File (EFF, Versions 3.0,
3.1). The comparison shows that all the used libraries reproduce the neutro
n flux generally within +/- 20% total uncertainty both in the void penetrat
ion and in the bulk shield behind it. The trend to underestimate the high e
nergy flux (E > 8 MeV) with increasing penetration depth is found again as
in previous bulk shield experiment, showing that the relevant transport cro
ss sections at around 14 MeV need further investigation. The comparison bet
ween experimental and calculated nuclear heating shows an agreement within
+/- 30% for all tested nuclear data libraries and in all experimental posit
ions. The comparison with the case of bulk shield without penetration is al
so discussed. (C) 2000 Elsevier Science B.V. All rights reserved.