Exhaustive test cases for the software reliability of safety-critical digital systems in nuclear power plants

摘要

Digital instrumentation and control (I&C) systems have several specific characteristics as compared with analog I&C systems. Although one critical feature of digital I&C systems is software, even after several years of research the quantification of software reliability remains an unresolved issue. This paper suggests a method to obtain an exhaustive set of test cases for safety-critical software. Without such a method, considering all combinations from the full range of software variables results in the number of test cases becoming so tremendous as to render testing physically impossible. In order to obtain a realistic number of test cases that still satisfies the "exhaustive" concept, the range of variables should be reasonably considered. In the developed method, all variables are first divided into plant variables and software internal variables. Input-profiles for the plant variables are obtained from the results of best-estimate thermal-hydraulics code, and for the internal variables, a gray-box analysis is conducted to identify the relationship between all internal variables. As an application to demonstrate the feasibility of this method, plant variables from the OPR-1000 were utilized to develop the input-profiles, and the internal variables of the bistable processors in the IDiPS-RPS were taken for a gray-box analysis. As a result, the novel method developed in this research yields an exhaustive yet realistic set of test cases for safety-critical digital software.