ML20213D147
| ML20213D147 | |
| Person / Time | |
|---|---|
| Issue date: | 05/29/1979 |
| From: | Soon Kim Office of Nuclear Reactor Regulation |
| To: | Meyer R Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML111220504 | List: |
| References | |
| CON-WNP-0333, CON-WNP-333, RTR-NUREG-75-087, RTR-NUREG-75-87 NUDOCS 8003240281 | |
| Download: ML20213D147 (5) | |
Text
-
Attachment A of Value-Impact Analysis
,( pa aseg,70 UNITED STATES y$g'i erg NUCLEAR REGULATORY COMMISSION
,.g wasmastow o.c.2 eses
%,' - QA,5 IIAY 2 3 3 3 MEMORANDUM FOR:
R. O. Meyer, Section Leader, Reactor Fuels Section, CPB, DSS FROM:
S. S. Kim, Reactor Fuels Section, CPB, DSS
SUBJECT:
COMMENTS ON THE PROPOSED APPENDIX TO THE STANCARD REVIEW PLAN " FUEL ASSEMBLY DESIGN ACCEPTANCE CRITERIA FOR DYNAMIC LOADS" Enclosed is my connents on the subject acceptance criteria.
c'4'",[ '
, M ang r,. Kim Reactor Fuels Section Corp Perfonnance Branch Division of Systems Safety Distribution:
K. Kniel 4
e 60 0h 9
k 4 1
.o F
ENCLOSURE
_ In y opinion. it _is too early to eliminate existing safety factor in the current practice and propose a new acceptance criteria in SRP Appendix. We only reviewed the Westinghouse method so far and we do not know the details of other vendor methods (CE. Exxon Nuclear andB&W). We may be in a position to write a formal acceptance cri-2 teria only after ve complete review of all the vendors and, hopefully, some full scale test become available. - We need this learning curve because the method contains highly complex non-linear analysis which is new to us.
I reconnend to continue to use an interim safety factor of 1.35 (instead
.of 1.75) on grid strength as an engineering safeguard until km reach a point of better understanding.
Following are the reasons for the safety.'4ctor.
I also made several technical coments on the proposed new criteria as a backup data as to why such criteria is premature and technically unsound.
A.
Coments on the Proposed Acceptance Criteria
- 1) EG&G's proposal assumes that one can devise a calculational method for a grid impact with no error bound (perfect model).
To state that non-linear analysis predict best-estimate value without benefit of experimental verification seems to be an unsupportable statement and may turn out to be on an undefend-able position. (At one time, EG&G recommended a full scale experiment including reactor and internels because grid im-pact may be difficult to evaluate by calculation alone.
t 2-
- 2) Two of the major items proposed by the EG&G have technical deficiencies.
a) Steam flashing: safety factor 1.3 This is a new load previously not considered by the ven-dors. Assigning a safety factor at this time is premature since the Analysis Branch with the primary responsibility in the area has not conmented on it. This new loading is a typical example of a moving' target.
b) Sensitivity: safety factor P
EG&G proposes that " plant specific sensitivity should be performed and additional safety factor should be imposed on the grid force if the sensitivity calculation shows that such safety is needed." Reason for the above proposal was Based on the EG&G sensitivity calculation where accelera-tion of core plate was used as a variable parameter rather than displacement function. tiowever, when a displacement time history was used, as done by all the vendors, no such sensttivity was apparent.' Therefore, we need a better jus-tification before requiring vendors any plant-specific sensi-t.ivity ca'1cu14tions.
- 3) Mean value vs. 95x95 LTL It wac noted in the new proposal that the grid strength may be determined by selecting a mean value of the test data rather thar. some upper bound. We msy be breaking a new ground in that allowable stress of safety component is selected by a l
P I:
n- -,.
.,n-,
-.--~.n7
m----,,--
l
.~ s l
\\
3 mean value (half of the component is allowed to fail?).
'I believe that not even a secondary system component allowable was determined by a mean value. It was argued that the mean value is allowed because consequence of failure is small.
However this argument was used several times explicitly or implicitly in developing the new criteria. If so, we might as well declare that the spacer grid fs not a safety component, and drop from the SRP item,
- 4) It is stated in section 3.a that we provide standard grids to all the vendors in an audit for the grid strength detennination.
This approach is premature and may create a problem. We do not know, at present, what is right procedure to determine a dynamic strength capability. Therefore, when we have a scat-ter in grid strength from different vendors, as we suspect, there is no way of deciding which value is correct. It is best not to do it now and accommodate such uncertainty by a safety factor. We will have better ideas on grid strength determina-tion once we review all the vendor methods.
B.
Recommendation I reconsnend to use safety factor 1.35 as an interim measure for the i
spacar grid. The numerical value of the safety factor was obtained from the following:
- 1) Comoonent (spacer grid) safety factor: 1.1 ASME(aswellasanyotherengineeringpracticesuchasASCE) re:ommends minium of 1.1 in safety factor for a service level D (faulted). With a possible exception of CE, all the vendors
--m
, c,
' I follows ASME recommendation. It is simply a bad engineering practice not to assign any safety factor for the components especially for a nuclear plant application.
- 2) Allowance for uncertainty in analysis: 0.25 The magnitude of the uncertainty factor in grid load calculation is somewhat arbitrary, However it provides the following advan-tages; i) encourage vendors to provide rigorous verification program.
- 11) keep review simple (no plant sensitivity analysis nor de-i tailed mechanistic review) iii) flexible to adapt new situation such as accomodating moving target.
I also recomend to retain 95x95 lower tolerance limit for the reason stated previously.
4 5
I I
O i
I l
i
-