Errors in Dose Assessment Computer Codes and Reporting Requirements Under 10 CFR Part 21
See IE Information Notice No. 85-52 entitled as above and dated July 10, 1985. This notice alerts licensees to: (1) errors in a dose assessment computer code supplied by a vendor, and (2) in general, computer codes can be considered basic components under the requirements of Part 21, and non-conservative errors leading to substantial underestimation of radiation exposures would be considered reportable under 10 CFR 21. The health physics position was written in terms of 10 CFR 20.403, but it also applies to "new" 10 CFR 20.2202.
IEIN-85-52 was issued following an evaluation by NRC staff of an event where errors were found in the prediction of offsite doses using computer software supplied by Nuclear Data, Inc. In the incident, a large discrepancy between the result of the offsite dose calculations made by the licensee and the regional office during an emergency preparedness exercise was noted. The licensee and Region V office used the same input parameters (radiological source term and meteorological conditions); however, the offsite calculated dose determined by the Region V office was an order of magnitude less than the licensee's estimation.
The licensee found errors in the dose assessment computer programs that were used to estimate environmental doses for both routine and emergency operation supplied by Nuclear Data, Inc. In coordination with Nuclear Data, the licensee corrected the errors and notified other licensees via INPO's electronic "notepad" of the inherent program error that led to predicting less atmospheric dispersion than the code should have calculated.
If errors result in substantially underestimating or overestimating offsite doses, it could result in inappropriate protective measures. An error that substantially underpredicts offsite doses (non-conservative) would be reportable under 10 CFR 21. The underestimation could cause a delay or deferral of protective action leading to unnecessary exposure to a person in an unprotected area, thereby creating a "substantial safety hazard."
An error that substantially over predicts (conservative) is not strictly reportable under 10 CFR 21, since it is unlikely that such an overestimation could result in personnel radiation exposures exceeding the referenced guidelines. However, because of potential non-radiological negative impact from unnecessary protective actions resulting from overly conservative dose estimates, licensees should continue to cooperate with vendors and share information concerning common problems with generic computer codes.
The following NRC staff guidance on the amount of radiation exposure that can be considered to represent a "substantial safety hazard" is taken from NUREG-0302 (Rev.1):
A substantial safety hazard means the loss of a safety function to the extent that there is a major reduction in the degree of protection provided to public health and safety. Note that the term "public heath and safety" includes both members of the public and licensee workers / employees.
From a radiological perspective, a criterion for determining whether substantial safety hazard exists includes "moderate exposure to, or release of, licensed material." a. Guidelines for determining what "moderate exposure to ..." means: greater than 25 rem to the whole body (or its equivalent to other body parts) to occupationally exposed workers; or exposure of 0.5 rem to the whole body (or its equivalent to other body parts) to an individual in an unrestricted area. b. Guidelines for determining what "... release of, licensed material" means: release of materials in amounts reportable under the provisions of 10 CFR 20.403 (b) (2) [or 10 CFR 20.2202 (b) (2)].
Regulatory references: 10 CFR 21, NUREG-0302
Subject codes: 2.2, 7.3, 12.12