FIG. 7 Cross sections of irradiated Zircaloy-2-clad UO, rods. 
At left is 0.72-in.-diameter rod irradiated for 60 sec in NRX hy- 
draulic rabbit; note central area that had been melted. Center 
tests (67), corrected for flux gradient 
in the case of the X-1 specimens, were 
used in calculating the integrals. 
Unambiguous melting has been ob- 
served in specimens irradiated in a 
water-cooled facility in the NRX reac- 
tor known as the “hydraulic rabbit,”’ 
where irradiations of short duration can 
be carried out. <A cross section of a 
sintered UO: specimen, DB, clad in 
Zircaloy-2 and irradiated for 60 sec 
(Fig. 7) indicates clearly the extent of 
melting. The method used to apply 
the observed radius of melting obtained 
from a photograph such as Fig. 7 to the 
graph shown in Fig. 6 is illustrated by 
Fig. 8, where [ro dé is plotted 
against the fuel radius, r. The curve 
shown applies specifically to specimens 
CR and DB, which were similar in di- 
ameter and enrichment. The oxide 
surface temperature of both was 
calculated to be 250° C. Hence, 
Tm 
aaa k(@) d@ is 66 w/cem for CR and 
62 w/cm for DB, at the melting tem- 
perature, 7. Specimen CR was irra- 
diated for only 30 sec. 
Later experience indicated that an 
irradiation time of 60 sec was required 
to ensure that virtual thermal equilib- 
rium had been reached. Thus, the in- 
tegral for melting in specimen CR is 
probably somewhat high. Specimen 
DO had a larger fuel-sheath diametral 
clearance than the other rabbit sam- 
ples. Hence, the integral might have 
been higher if the clearance had been 
comparable, as indicated by the verti- 
cal arrow on the graph. Since speci- 
mens ET, DI and DL did not melt, the 
ordinate of Fig. 6 could be specified 
but only an upper limit given to the 
abscissa. 
Small differences in surface temper- 
ature of the oxide can be corrected for, 
without introducing serious errors, by 
using an assumed value of conductivity 
in the difference term alone. To plot 
points on Fig. 6 for specimens CR and 
DB, a value of 14 w/em was assumed 
250° C 
for ip " R(8) 8. 
With the exception of sample DO, 
the fuel-sheath diametral clearances of 
all the specimens shown in Fig. 6 were 
small enough so that the oxide should 
have expanded to contact the sheath 
while the fuel was being irradiated. It 
is interesting to note that the ‘“effec- 
tive” thermal conductivity for such 
fuel appears approximately the same as 
the Hedge and Fieldhouse values (31) 
corrected for density. 
Effect of fuel-sheath clearance. A 
simple model was proposed by Robert- 
son et al. (58) to predict temperatures 
in fuel elements with large fuel-sheath 
diametral clearances. The UOz cyl- 
inder was pictured as remaining cen- 
trally located and intact under irradi- 
ation and expanding as if it were at a 
uniform temperature equal to its mean 
temperature. The interfacial temper- 
ature drop could then be calculated 
from the thermal conductivity of the 
gas in the surrounding annulus. 
It had been appreciated that the 
model was not physically correct, since 
it was unlikely that oxide pellets would 
remain central in the sheath and it was 
known that pellets would crack from 
the thermal stresses produced during 
irradiation. Evans had reported (37), 
for example, that cracking effectively 
relocated part of the original annular 
gap to the hotter interior of the fuel. 
It was only recently, however, that 
the inadequacy of the model was clearly 
demonstrated from tests in the EEC 
loop (61) and hydraulic rabbit (62) 
at Chalk River. The latter tests, on 
0.67-in.-diameter oxide pellets heated 
and right cross sections (0.41-in.-diameter rods) show effects of 
UO, composition after ~4,500-Mwd/tonne U irradiation; at cen- 
ter is UOo.o0 and at right is UO>2.15 
near the melting point in the center, 
indicated that the oxide surface tem- 
perature of specimens with a starting 
0.017-in. fuel-sheath diametral clear- 
ance was not more than 100° C higher 
than those with a 0.005-in. clearance. 
These results have led Robertson to 
suggest a more realistic model (63) in 
which cracked segments of oxide shift 
radially outward to contact the sheath, 
so that the interfacial temperature drop 
becomes mainly dependent on the inter- 
facial pressure and the properties of the 
contacting surfaces. The few avail- 
able irradiation data support the newer 
model. If the approach is valid, large 
differences in assembled diametral 
clearance could be tolerated in fuel ele- 
ments where the sheath-collapse prob- 
lem mentioned earlier was of no con- 
cern, since such differences would have 
a relatively small effect on the fuel tem- 
perature during irradiation. 
Effect of O/U ratio. Several tests 
(58, 64) have demonstrated that non- 
stoichiometric UO: exhibits much more 
grain growth and liberates more fission 
gas than near-stotchiometric pellets ir- 
radiated under comparable conditions. 
The most unambiguous comparison 
was obtained from the X-2-n loop test 
where specimens, identical except for 
their O/U ratio, were irradiated in ad- 
jacent positions for the same period. 
After irradiation, two samples of steam- 
sintered and hydrogen-cooled UOs.oo 
were radially cracked with no apparent 
grain growth, whereas two samples of 
steam-sintered UO..;; exhibited crack- 
ing and extensive grain growth, as illus- 
trated in Fig. 7. The amount of fission 
gas released was 100-200 times higher 
from the nonstoichiometric oxide, 
The X-2-n experiment clearly indi- 
cated that the value of fk(@) d@ pro- 
ducing grain growth is markedly low- 
ered if extra oxygen is added to the 
65 
