### Abstract:

Independent fission product yields are primary input nuclear
data in the evaluation of nuclear reactor after-heat when using a
nuclide-decay summation method. The error in yield values is one
of the major causes of uncertainty of this calculation. Since only
a small fraction of the fission product yields has been measured
directly, the adoption of model-predicted unknown yields is inevitable.
The uncertainty involved is usually quite large.
This dissertation presents a systematic approach to correlating
the relevant experimental and calculated ²³⁵U thermal fission
data with other nuclear data, so that with application constraints
from exact physical laws, a consistent set of primary fission fragment
yields can be generated. These are for use in improving the
precision of after-heat calculations by constraining yield errors
and thereby reducing the uncertainties caused by them.
Fission fragment kinetic energy data and nuclear mass table
data are used to calculate the total excitation energy of a fragment
pair. This energy is partitioned between light and heavy fragments
assuming equal temperature of the fragments at the time of scission.
The prompt de-excitation of the primary fission fragment via neutron
and gamma ray emissions is simulated by nuclear evaporation processes.
A FORTRAN IV Monte Carlo Code called EVAPOR was developed for this
purpose. Neutron yields and emission probabilities for various numbers
of neutrons from over 1,000 possible fission fragments were
generated by EVAPOR.
The major uncertainties of the emission probabilities calculated
by EVAPOR can be simulated by lumping them together as though they
were the result of uncertainties in the mean excitation energy of a
fission fragment. A sensitivity study of the effect of this uncertainty
thus permits an a priori estimate of emission probability
error. The resulting emission probabilities and their errors can
then be used as an input for inference of primary fragment yields.
This inference, a form of probabilistic data unfolding, has
been explored for a sample case. Results to date have shown that a
consistent set of primary (before neutron emission) and secondary
(after neutron emission) yields can be generated using combined data
for light and heavy fragments of complementary charge number. However,
there are significant discrepancies between the secondary
yields obtained by this process and those used as input to it, which
stem from the ENDF/B-IV evaluation. The sources of these discrepancies
must still be explored.