
 DOENASTD30162006
Qualitative identification of controls and ensuring their adequacy is the centerpiece of the safety
evaluation process. In a qualitative hazard analysis, the hazard analysts are concerned with how each
control may fail, how to prevent such failures, and whether redundant components, verifications, or
diverse systems need to be considered to ensure adequacy of controls for each hazard/accident scenario.
This analysis aids DOE/NNSA in making an informed decision on whether to authorize the operation.
The information provided must also include the evaluation of how each control meets the safety function
derived from the analysis. Controls should be prescribed at a level of appropriate detail to ensure their
effectiveness.
7. INTERFACE WITH SARs
The HAR must evaluate all hazards that could impact the NEO and must serve as the final safety basis
integration document. Another DSA (e.g., a SAR) may provide analysis and resulting controls for
hazards that are relevant to the NEO. However, the HAR must verify the analysis and controls are
adequate for the hazard.
DOESTD3009, or superseding directives, provides adequate criteria for evaluating NPH in regard to
systems, structures, and components associated with NEO.
8. CHARACTERIZATION OF PROBABILITIES AND UNCERTAINTIES
Probability estimates for weapon responses, safety function failures, and intermediate events as part of an
accident sequence should:
a) Provide reasonably approximate, orderofmagnitude pointestimates commensurate with the
secondary role that estimation of accident scenario probabilities play in the safety basis
documentation,
b) Characterize the degree of uncertainties from the range of variability in supporting information
that was used to develop the pointestimate probability,
c) Be reasonably conservative, and
d) Be associated with properly and thoroughly defined events.
The application of this model for characterizing probabilities and uncertainties can be accomplished as
described below.
1) Establish pointestimate probability for any event from the associated underlying probability
distribution (e.g., 95/5, 80/20, mean and standard deviation, or median and uncertainty factor) to
those values. When using percentiles, they represent the confidence levels associated with the
upper and lowerbound probability values. The PPC will use the appropriate two parameters of
the supplied information to develop a lognormal distribution, and derive the mean of the
distribution as the representative and reasonably conservative pointestimate for the probability of
the associated event.
2) For those probability estimates based on detailed modeling or data resulting in a reasonably
conservative estimate, it may not be necessary to assume an underlying distribution associated
with the variability of the pointestimate probability.
Probabilities previously established using the binning methodology do not have to be changed for
compliance with this standard.
9. NEW INFORMATION
As the weapon stockpile ages and technology develops, the DAs continue to learn more about the
behavior and aging of weapon systems and components. This understanding is accomplished through a
6

Privacy Statement  Press Release  Copyright Information.  Contact Us 