Quantitative software-reliability analysis of computer codes relevant to nuclear safety
- ed.
This report presents the results of the first year of an ongoing research program to determine the probability of failure characteristics of computer codes relevant to nuclear safety. An introduction to both qualitative and quantitative aspects of nuclear software is given. A mathematical framework is presented which will enable the a priori prediction of the probability of failure characteristics of a code given the proper specification of its properties. The framework consists of four parts: (1) a classification system for software errors and code failures; (2) probabilistic modeling for selected reliability characteristics; (3) multivariate regression analyses to establish predictive relationships among reliability characteristics and generic code property and development parameters; and (4) the associated information base. Preliminary data of the type needed to support the modeling and the predictions of this program are described. Illustrations of the use of the modeling are given but the results so obtained, as well as all results of code failure probabilities presented herein, are based on data which at this point are preliminary, incomplete, and possibly non-representative of codes relevant to nuclear safety.
- Research Organization:
- Argonne National Lab., IL (USA)
- DOE Contract Number:
- W-31-109-ENG-38
- OSTI ID:
- 5194692
- Report Number(s):
- NUREG/CR-2186; ANL-81-84; ON: DE82019342; TRN: 82-018895
- Resource Relation:
- Other Information: Portions of document are illegible
- Country of Publication:
- United States
- Language:
- English
Similar Records
A Statistical Testing Approach for Quantifying Software Reliability; Application to an Example System
Quantitative software reliability in nuclear safety