Evaluation & Results

DOE-STD-1134-99

If parameter studies are performed and the keff results are plotted, the curves should be smooth

and continuous. Discontinuities, outside of statistics, are generally, but not always, a signal of a

mistake in the model. Such discontinuities should be understood and explained before taking

them as indicators of errors.

The reviewer should be aware that apparently minor changes in calculational models could have

significant effects on the reactivity of the system. It is common for an analyst to gloss over

minor changes or variations when reporting results and still report large differences in keff. The

reviewer should seek to understand the underlying model changes that caused the reactivity

changes. Typical model variations to watch for in Monte-Carlo calculations include:

Boundary conditions (specular, albedo, vacuum, etc.);

Reflector type (types of concrete, water, polyethylene, etc.);

Reflector thickness;

Presence or absence of moderators;

Moderation type;

Minor spacing variations;

Number of neutron histories;

Neutron start distribution;

Reversed geometry;

Density variations; and

Mistakes in material compositions and locations.

These kinds of potential variations are particularly important to watch for when the CSE

references previous work. The CSE should clearly document why calculational results from

previous CSEs apply to the system currently being analyzed and why the variations, if any, exist

in the reactivity results for very similar models. The previous CSE should be summarized so that

the reviewer knows what was determined in the previous study and how it applies. There is no

need to produce a complete copy of the prior analysis results.

The reviewer should perform spot checks of the "worst case abnormal" models to see if they

match those described in Section 5.0. A common practice in Monte Carlo is to slightly change

the models for the "worst case" situations (i.e., to reduce or eliminate conservative assumptions)

to "tweak" the reactivity result under some arbitrary cutoff. The reviewer should be aware of this

tactic and assure himself that it is inconsequential if it occurs.

Key Review Issues

Calculational results for normal and abnormal scenarios are present and summarized.

Variations in keff from calculation to calculation are explained adequately in terms of model

changes corresponding to process variations.

A "sanity check" using handbooks, previous analyses and/or hand calculations is consistent

with reported results.

Results applied from previous CSEs are adequately justified.

No unexplained discontinuities are present in parameter studies.

Look for geometry plot to check the model configuration.