ML17262A019
| ML17262A019 | |
| Person / Time | |
|---|---|
| Issue date: | 01/31/1978 |
| From: | Levine S Office of Nuclear Regulatory Research |
| To: | Case E, Minogue R, Casey Smith NRC, Office of Nuclear Material Safety and Safeguards, Office of Nuclear Reactor Regulation |
| References | |
| RIL-0019 | |
| Download: ML17262A019 (2) | |
Text
.!
.l'.'
'.f
---\\-, -*~ *
- l.
UNITED SH' ~ES NUCLEAR REGULATOl'Y COMMISSION WASHINGTON, 0. C. 20555 MEMORANDUM: Edson G. Case, Acting Director Office of Nuclear Reactor Regulation FR0.!\\1:
SUBJECT:
Robert B. Minogue, Director Office of Standards Development Clifford V. Smith, Jr., Director Office of Nuclear Material Safety and Safeguards Saul Levine, Director Office of Nuclear Regulatory Research RESEARCH INFORMATION LETTER #19 - GO METHODOLOGY ASSESSMENT This memorandum transmits the results of research on an assessment of GO, methodology for performing reliability evaluations of systems and subsystems, as requested informally by your office. Enclosed please find a copy of a report on the research giving a de;;c1*iption of the analysis performed, a description of the GO method and a detailed description of the application of GO to a typical reactor trip system. This research provides for an understanding of this method either to perfo1*m reliability prediction or to critique system analyses subrnitted by applicants. The results of the research were reviewed by the Offices of Nuclear Reactor Regulation, Standards Development, and lnspcction and Enforcement.
GO provides a method for system modelin~ and a computer code to calculate a prediction of system reliability. The study demonstrates that this method p1*ovides equivalent results to those obtaiuecl from fault ti*cc a11alysis, the analysis mc1.hod used in the Reactor Safety Study. A 1*ather unique feature of this method is the capability to calculate with one model the probability of the various states of system condition, e.g., early operati.011, normal, failed, etc. Other methods, such as fault tree analysis, require a different model for each system condition; however, this feature of GO requires that all system components be included in the model. The model resembles the system schematic or piping diagram which thus reduces the burden of modeling all system components.
Addressees e
{
Evaluation and Applic;i.tion
- An example of how GO can be used to model and calculate system unavailability is extensively described in the enclosed report. The description of the analysis shows how large complex systems can be modeled by using this method. The use of supertypes described in the report, can greatly reduce modeling time and effort when dealing with redundant systems. The method also allows for the inclusion of human errors into. the models.
GO has a potential sigri.ificant use as a means of determining system reliability or as a dive1*sc method for verifying system analysis performed using fault tree or similar modeling techHiques; howc,*cr, it does not provide the qualitative assessment capability that can be obtained from fault trees. The sensitivity results obtained using this method identify the components which have the greatest effect upon a systems unreliability; however, the combinatorial relationship of the sensitive components is probably_more easily discerned using fault trees. The choice of using either GO or fault tree analysis may depend upon the analyst's capability and preference.
The utilization and application of GO will allow for further evaluation of the practical uses and limitations of this technique.
We currently plan to place the GO program in the computer library at Brookhaven National Laborato1-y for access by NRC,users. If desired, training programs ca;1 be arranged by PAS to instruct personnel in the construction and programming of models used in this method.
~e,~
Office of Nuclea1* Regulatory Research
Enclosure:
"A Comparison of Results from the GO Methodology and Fault Tl*ec Analysis, 11 by D. E. Wood and N. J. Be car, August 31, 1977, Kaman Sciences Corp.