ML20246H364
| ML20246H364 | |
| Person / Time | |
|---|---|
| Issue date: | 01/31/1989 |
| From: | NRC |
| To: | |
| Shared Package | |
| ML20246H349 | List: |
| References | |
| GL-84-04, GL-84-4, NUDOCS 8903200127 | |
| Download: ML20246H364 (6) | |
Text
_.
s
'.j # **" %
UNITED STATES k
E NUCt. EAR REGULATORY COMMISSION g
WASHING TON, D. C. 20655 ik
- l y egy JANUARY 1989 STAFF EVALUATION OF EMBRITTLEMENT OF REACTOR VESSEL SUPP0 PTS The Advisory Comittee on Peactor Safeguards (ACPS) initially raised the question l
of greater than expected embrittlement of the reactor pressure vessel (RPV) sup-ports in a July 1987 memorandum, citing the embrittlement of the ORNL High Flux y
l Isotope Reactor (HFIR) and questioning the implications of the HFIR experience on RPV supports. The initial investigation, reported in September 1987, indicated ~
that the potential existed for significant embrittlement of RPV sup' ports. The 4
staff asked ORML to continue the investigation by perfoming plant specific in-vestigations of the limiting plants.
In the interim, the ACRS focused its l
questions, asking specifically if (1) there were any plants operating today where j
the operating temperature was below the support material's nil-ductility tempera-ture (NDT), and (2) if there would be any plants where the operating temperature-l
'would be below the NDT by the end of the design life.
In February 1988, the staff briefed the ACRS on the continuing ORNL study.
In-that briefine, the staff. asserted that looking only at the operating tempera-ture relative to the material's NDT would nnt adeouately evaluate the potential
[mpact of greater than expected irradiation damage.
Rather, a '" fracture mechanics" analysis of the support structures of the liniting plants was proposed. The ACPS subsequently stated tSat they were satisfied with the
{
staff's approach.
l The ORNL study began by grouping the support structures for operating plants i
~
into five categories--skirt supports, shield tanks, long columns, short columns, and suspension supports. Skirt supports, used in all but one BWR and all but one B&W PVR, were excluded from' further study because of the very Inw neutron flux in the vicinity of the skirts. Shield tanks (9 PWR's with Stone and Webster as the AE) and long columns (9 C-E plants and 2 Westinghouse plants) were not considered further in the OPNL study because of industry j
analyses that concluded there were no problems with.these structures. Only j
one plant, Big Rock Point, uses suspension supports so this type also was not l
considered further by ORNL. The short column designs (57 Westinghouse and C-E l
plants and the B&W designed Davis Besse) became the focus of the ORNL study because (1) they can have members with tensile stresses'in the relatively high i
8903200127 890313 PDR TDPRP ENVWEST B
PNU
.-___-___--a
2 I
flux regions of the cavity and (2) there were no other studies evaluating these support designs.
l I
ORNL identified two plants--Trojan and Turkey Point Unit 3--with the short 4
column design supports which also had members with tensile stresses exposed to the relatively high cavity flux. Although the initial survey of support-designs iden'tified 15 other plants with these features, these two plants were l
considered the limiting designs.
4 I
ORNL requested from the two utilities specific information on materials, cavity flux and neutron energy spectrum, operating temperature, loads, and construc-
{
tion details including preservice inspection reports.
Both utilities cooper-l l
ated fully although some of the information requested by ORNL did not exist.
l
~
As discussed below, the lack of specific materials information, loads, and operating temperatures has contributed to significant uncertainty in the j
results of the analyses and the need to carry this effort further.
-l I
I in addition to the OPNL analyses, Portland General Electric (PGE) commissioned j
a separate study of the Trojan plant's RPV supports. The PGE study paralleled
)
the ORNL study in many respects although there were some notable differences.
)
Perhaps the greatest difference comes in how PGF and OkM. treated the irradia-tion damage data from the HFIR. ORNL extrapolated the HFIR data to conditions l
pertinent to the vessel supports; PGE truncated the irradiation damage to zero l
17 if the neutron fluence fell below 10 neutrons /cm2 (E>1MeV). The staff l
advised PGE that this approach is not acceptable since it ignores data below 1
2 the 10 neutrons /cm level and since there is no physical basis for truncating 1
the damage at this level.
As a result of staff comments, PGE revised their analysis to include a value for neutron irradiation embrittlement below 1017 2
neutrons /cm.
The single greatest analytic uncertainty is the extrapolation of the HFIR data to conditions pertinent to RPV supports. The extrapolation schemes are based on expert opinion using the limited supporting data base.
The limited data base results in the analyses being sensitive to these extrapolations and is the main reason that ORNL and PGE have different conclusions.
i
~
, Both the ORNL and PGE analyses used linear elastic fracture mechanics to compute the critical size of flaws or cracks that could propagate unstably i
during faulted conditions. This " critical crack size" then can be compared to the size of crack that could exist in the structure tn determine a margin.
]
against failure. The ORNL study computed the critical crack sizes but did not provide an estimate of the crack size that could exist in the structure.
The PGE analysis included an estimate of the crack size that could exist in the structure.
I l
Plant specific information was not available for several of the analysis input parameters.
For example, in ORNL's Trojan analysis, fracture toughness 1
properties for the material actually used in the support structures were not available. Cnnseauently, generic values were estimated from data in the
~
I literature, but even this data base is limited. Although, PGE used the sane data base as NL, its evaluation resulted in higher fracture toughness properties fc,r the support material. Another difference in the evaluations is the effect of the concrete between the support pedestal and the cavity liner on the lead carrying capacity of the supports. - PGE assumed this concrete would be effective in carrying load. ORNL performed the analyses with and without the concrete but in the best estimate evaluation assumed it would not be effective because contact between the concrete and the supports could not be assured-durino faulted conditions.
Further, actual cavity dosimetry data are not available for, Trojan.
Thus ORNL and PGE relied on data from a "similar" plant, normalized to the Trojan operating history.
I Based on the ORNL study of the Trojan plant, the most likely fracture origin (the location with the smallest critical flaw size), is in a support structure detail--a flame cut grout hole in the tension flange of the horizontal support beam embedded in the concrete.
Based on their fracture mechanics analyses, ORNL concluded that the best estimate critical flaw size at the' grout hole is a concern at 32 EFPY.
The critical flaw sizes at other locations are considerably larger but not necessarily inconsequential. Both the original and revised PGE analyses. indicate there is no concern at 32 EFPY.
o.
. ORNL performed a fatigue crack growth analysis which shows that crack growth during service is unlikely. _ Consequently, any flaws that could lead to failure would have been introduced during fabrication. This leads to questions about the size of flaws that could be introduced during fabrication and either accepted or missed during the fabrication inspection.
It also leads to ques-j tions about local brittle zones and the potential for them to abruptly fracture during loading, producing a rapidly propagating crack that must be arrested by the surrounding material to preclude failure of the structure. Both PGE and ORNL did not specifically address local embrittlement zones.
Fracture resulting from local brittle zones may be conservatively evaluated using the crack arrest fracture toughness from the ASME Code K curve for pressure IR vessel materials.
The fracture toughness values from the ASME code K curve 7p are below the values estimated by ORNL in their analysis. Hence, a fracture mechanics analysis using the ASME Code K curve would indicate that the yp critical flaw size at the grout hole location would be less than the size
" estimated by ORNL.
The ORNL study of the Turkey Point plants--Turkey Point 3 and 4 have identical supports--is not as complete as the Trojan study. There are virtually no generic material property data for the materials used in the Turkey Point I
plants. As a result, the material property data used in the study came from the ASME code end reflect lower bound properties for this general class of steels. Also, the largest loads provided by the_ utility are not appropriate since they reflect DEGB of the primary loop piping.
At the time of the study, the utility provided loads for the double-ended break of the primary coolant loop pipes, contending that they were still part j
of the design basis since leak-before-break had not been requested. However, I
the Turkey Point plants were part of the Westingh use Owners' Group study that led to resolution of USI A-2.
Subseouent to the time that the loads were provided to ORNL, the utility submitted a letter requesting that the dynamic ~
effects of postulated primary loop pipe ruptures be eliminated from the design basis.
The staff has reviewed the licensee's sumbittal and concurs that the leak detection systems at Turkey Point 3 and 4 satisfy the requirements in
)
e o
i
, Generic Letter 84-04. The staff also finds that the primary loop piping in these plants complies with the revised General Design Criterion 4 (GDC-4) of Appendix A to 10 CFR 50.
Consequently, the dynamic effects of the postulated primary loop pipe rupture may be eliminated from the design basis.
While the large loads may t'e deleted, the utility has not yet determined the appropriate replacement loads resulting from a postulated DEGB in auxiliary piping.
These loads became the limiting loads used in the Trojan study and would have to be considered to complete the Turkey Point study. However, the Turkey Point study suggests that, at least qualitatively, the results will be similar to the Trojan study results.
Based on the results of these analyses, the staff reconnended that Generic Issue 15, " Radiation Effects on Reactor Vessel Supports," be reprioritized.
~
A preliminary reprioritization screening of the issue indicates the priority increasing from LOW to HIGH. Accordingly, ar ' action plan, potentially including experimentation, will be developed to resolve the issue.
It is expected that the plan will include the following specific actions.
Immediate Actions (1) Shortly af ter the Draf t ORNL report has been placed in the PDR, the staff will meet with the Westinghouse Owner's Group to discuss the results of the ORNL,and PGE analyses, the staff's concerns, and the Owner's Group conclusions.
(2) Organize specialists' meetings (NRC, Contractors, Vendors, Owners' Groups, and International Experts) to evaluate the overall scope of the problem, to assess uncertainties in the analyses, and to develop an acceptable analysis framework.
(3)
Initiate a study to assess the consequences of RPV support failure for a trial plant (Trojan), starting with reasonable assumptions of failures of critical components due to fracture propagation. The study is intended
'r e
. to evaluate the effects of redundant load paths and to determine if movement of the vessel would result in failure of components or systems needed for safe shutdown.
[ NOTE: The staff has performed a bounding study indicating that the RPV can be supported by the primary coolant loop piping without the benefit of the supports.)
(4)
If warranted, based on the resolution in (2) and (3), issue a generic letter indicating the staff concern and containing the recommended analysis framework.
Long Term Actions (1 - 5 years)
(1) Gather low-temperature, low-flux embrittlement data to further evaluate rate effects.
(2)
If warranted, and based on the results from the immediate actions, evaluate the limiting long column and shield tank supports to determine if they are subject to brittle fracture from neutron irradiation embritt-lement.
(3) Confirm that Big Rock Point's supports will not be subject to brittle fracture from neutron irradiation embrittlement.
(4) Perform a PRA analysis of vessel support failure.
_ _ -. _.. _.