ML20205B000
| ML20205B000 | |
| Person / Time | |
|---|---|
| Site: | Crystal River  |
| Issue date: | 05/24/1985 |
| From: | Dircks W NRC OFFICE OF THE EXECUTIVE DIRECTOR FOR OPERATIONS (EDO) |
| To: | Asselstine NRC COMMISSION (OCM) |
| Shared Package | |
| ML19284E734 | List: |
| References | |
| REF-GTECI-A-49, REF-GTECI-RV, TASK-A-49, TASK-OR PTS06, PTS6, NUDOCS 8506050224 | |
| Download: ML20205B000 (7) | |
Text
C ci
.TD y'
MAY 2 41985
\\ 7-
/
~
~
Y MEMORANDUN'FOR.
Commissioner Asselstihe_
FROM:
William J. Dircks Executive Director for Operations
SUBJECT:
RESPONSE TO YOUR MAY 1, 1985, QUESTIONS REGARDING SECY 85-60, FINAL PTS RULE The staff has prepared the enclosed responses to your questions regarding the final rule on pressurized thermal shock.
I5'gfHND William L flickq William J. Dircks Executive Director for Operations _
Enclosures:
1.
Responses to Questions Regarding SECY 85-60,
~ ~ ~
Final PTS Rule.
2.
Letter Report to E. Throm, NRC from G. Willicutt, Jr.
(LANL) TRAC-PD2 Calculation of the Crystal River 3 m-Transient of Feb. 26, 1980-Using Revised Assumptions (4/5/83).
3.
Paper by T. G. Theofanous and K. Iyer; Mixing Phenomena of Interest to SBLOCAs.
Invited paper for specialists meeting on S8LOCA analysis for LWRs, to be held in Pisa, Italy, June 23-27, 1985.
cc:
Chairman Palladino Distribution Commissioner Roberts Central Files FSchroeder Commissioner Bernthal GIB Rdg KKniel Commissioner Zech DST Rdg NAnderson OPE
.WDircks
.RWoods OGC JRoe A-49 File dECY TRehm EDO Rdg VStello HSmith - 000594 contact:
RMinogue CBarnes R. Woods, NRR GCunningham BMoore 492-7414 JHeltemes CTwigg S
W TSpeis hod o
N
-~
'$ 7 0 6 6 5~ 0 2 2 0
\\h
% [4) Y yA Sb -Jot-S Yshrt
+See previous sheet for concu.1rrences DST:GIB
- DST:GIB
- DST:ADGP
- D: DST
..j..g.
..g.U '
FSiihroedir TSpeIs
~ DEisenhui HDenton WJDircks
~
5)li/85' 5/i5[8E 5)I5/8E 5/ii)85 5/,/85 5/, /85 5/ /85 DATE) e.a ro u ve i,c c, v.cu em OFFICIAL RECORD COPY
- w. u + - n. ~
1 e
Responses to May 1, 1985 Question F om '
Commissioner Asselstine Regarding the PTS Rule QUESTION 1:
Is the analysis in the Nuclear Safety article accurate?
Response
No. The article stated that areas of the pressure vessel could have been exposed to temperatures below 250'F during the event, which is not correct.
However, the authors did note that detailed analyses needed to be done to determine the lowest temperatures, which is correct.
the staff sponsored such detailed analyses of the Crystal River 2/26/80 event.
The:results are reported in a letter report to E. Throrr (NRC) from G. Willicutt, Jr. (LANL) dated 4/5/83(LA-UR-83-1078)(attached). A better thermal mixing model is included in the analyses, resulting in predicted temperatures in the downcomer, where the critical welds are located, not lower than 476*F.
When the work was performed for the Nuclear Safety article in 1981-82, a major uncertainty was the possibility of cold water stratifying in the cold leg and impinging plumes of cold water along the downcomer pressure vessel wall.
Experimental and theoretical research funded by EPRI and NRC at Creare and by
~
NRC at Purdue University has since led to a good understanding of the mixing phenomena involved. Mixing is much greater than was previously assumed before this research. An excellent overview of this research is given in the attached paper by T. G. Theofanous and K. Iyer, Mixing Phenomena of Interest _to SBLOCAs.
Present results show that, even under conditions involving complete stagnation of all loop flow, extensive natural-convection-driven flow paths are established.
For example, these paths result in circulation of warm water from the lower plenum up through the downcomer into the cold leg pipe all the way to the cold injection point.
This results in mixing of the cold injection flow with warmer water before the mixture reaches the critical welds.
Inclusion of these experimentally demonstrated mixing phenomena is a principal reason why our more recent results show a diminished risk from PTS events compared with results of earlier analyses, such as those that Yesulted in the 250*F temperature prediction for the Crystal River event.
y
y p-
. / ;~.
2 s'
Question 2:
If so, how was the Crystal River event ~ factored,.into the staff analyses leading to the proposed screening criteria in s
$50.61(b)(2)?
Response
During development of the proposed screening criteria, the letter report analysis results (476*F minimum temperature) referenced in the Response to Question #1 were not yet available, so the Crystal River event was assumed to have resulted in a minimum downcomer temperature of 250*F (SECY-82-465, Enclosure A, Nov. 23, 1982).
Even after the letter report analysis results were available, when the proposed PTS rule was being prepared for publication for public comment (SECY-83-288, July 15, 1983), the staff continued to use 250 F to represent the Crystal River event.
This was because the staff:
(1) did not at that time have available the full results of the EPRI, Creare, and Purdue thermal mixing research referenced in the response to Question
- 1; and (2) wanted to assure conservative treatment of events that had been experienced.
It is perhaps worth noting that the Nuclear Safety article you referenced and the data base used by the staff in development of the PTS screening criteria (SECY-82-465) come from the same origin.
In the early stages of the staff's work on PTS (1981), the staff asked the ORNL precursor research project to combine the preliminary event trees developed by ORNL to describe possible PTS sequences with ORNL's key-word LER index.
The PTS precursors so identified were published as NUREG/CR-2789, a draft of which the staff used in identifying the PTS precursors reported in SECY-82-465.
The authors (Phung and Cottrell) then later also published their work in the Nuclear Safety article you referenced.
Question 3:
Why should not the screening criteria in 6 50.61(b)(2) be below the temperature that apparently occurred at Crystal River?
Response
The screening criteria of 270*F for axial welds and plates and 300*F for circumferential welds are below the lowest temperature reached by the Crystal River event, 476*F, based on the detailed thermal-hydraulic analysis referenced in the response to Question 1.
j,,
~
3
~
However, at the time the screeni g critefia wer.e developed, as reported in the response to Question #2, the value being used.,
s for the lowest temperature in the Crystal River event was 250 F.
That temperature is below the criterion.
However, the occurrence of a temperature less than the screening criteria during a transient is not sufficient to imply high probability of vessel failure for the following reasons.
RT is a single defined point in the approximately 100 F tehraturerangewhereamaterial'stoughnesswillchange most rapidly with temperature variations.
It is not a step i
change, i.e., one does not suddenly " fall off of a cliff" if the temperature goes below the RT Another way of stating thisisthatconsiderabletoughne$kT.still exists at l
temperatures below the RTNDT*
Therefore, in order to determine PTS risk from vessel failure due to any given transient in a vessel with a given RT must perform a detailed fracture mechanics analysis. IkT's"*
i not sufficient or meaningful to merely compare the lowest temperature reached to the RT The other important factors l
tobeconsideredare:,howfa$kT.the cooling occurs, how long the temperature remains at low values, the reactor vessel internal pressure, and whether or not a flaw (crack) exists in the critical area.
For example, at we reported in Enclosure C to SECY-83-288 (July 15,1983) a conservative deterministic fracture mechanics analysis of the Crystal River event based on the pressure and temperature time histories, but assuming i
a minimum temperature of 250*F (rather than the 476 F minimum l
discussed above) yielded a " critical RT
" value of 277*F.
NDT The " critical RT is the value of RT below which crack initiationisnobredictedbydeterminMicfracture mechanics analysis, given the temperature ar.d pressure history in the event.
4 On the basis of many such detailed analyses, of many actual and postulated events, the staff reached its conclusions regarding the screening RT r
later in the proposed PTS M e.eported in SECY-82-465 and Based on the work reported in SECY-82-465 and extensive, more recent confirmatory work, the staff continues to believe that PTS risk is acceptable for any PWR with RT below the proposed screening limits.
NOT The staff continues to believe that setting a screening limit 4
below the lowest temperature that occurred (or was believed to have occurred) during any experienced event, without consideration of the temperature and pressure time-histories during the event, would result in a screening limit that is unnecessarily over-conservative.
--.n.
. / [ '.
~
4 Question 4:
Why did staff not recommend a re uiremdnt that iicensees (not numbered analyze and correct, as appropriate, their plan't designs and
'in your procedures to ensure that there are no credible single or memorandum) common cause failures (including human error) that would cause a pressure vessel thermal shock event?
Response
To accomplish their basic purpose, nuclear plants must contain massive, complex systems designed solely to remove large amounts of heat from, and create pressure within, the primary system.
It is not possible to ensure that no single or multiple failures can occur that would cause those systems to perform their design functions under conditions when that function is not wanted.
Ihat is, there will always be precursors to PTS events occurring at nuclear plants.
Every shutdown is, in some mild sense, a precursor in that it involves going to low temperatures.
It takes failures, however, to cause too rapid a cooling with concurrent pressure (conditions not present during normal shutdown).
Given that precursors will occur, the way to minimize PTS risk is twofold.
(1) Provide design features (high level trips, isolation, etc.) and procedures that will prevent cooling and i
pressurization under conditions when it is not wanted.
~
This lowers the frequency of precursors becoming severe PTS events, because these systems / procedures must fail in order for a severe PTS event to occur.
The more failures that must be assumed to postulate a severe PTS event, the lower the frequency of that event becomes.
(2) Ensure that operation will not occur with vessels embrittled to a level where frequently experienced or predicted events would cause vessel failure.
It is possible to postulate one failure followed by another failure followed by another, etc., until severe enough conditions are predicted in which any moderately irradiated vessel would fail.
It is thus necessary to ensure that the vessel is tough enough to withstand any failure or combination of failures that have a frequency high enough to be of concern.
.O 9
l 1
5 e
Thestaff'sPTSefforthasstudiedmanfsvents,realand postulated, looking at both the frequency and the embrittlement level (RT Wehavbo)ncludedthatallPWRsnowhave ai. which the event would cause vessel failure.
sufficient design features and procedures so that PTS-related risk is acceptable, so long as the embrittlement level remains below an RT equal to the proposed screening criteria.
NOT We did not look at the specific, detailed features of each PWR in reaching this conclusion.
We did, however, look at representative plants and we did include a level of conservatism we believe adequate to assure applicability of that conclusion to all plants.
1 It is thus quite possible that many plants have sufficient features and procedures (or could change the plant into such a configuration) that risk would be acceptable at higher embrittlement levels (above the screening level).
However, a
<dtailed analysis of the specific plant would be required aefore such a cenclusion could be supported.
That is why the proposed rule requires such an analysis before the screening level is reached.
Question 5:
I would like to know the upper bound frequency, the frequency 1
(not numbered which served as the basis for the staff position, and the in your rough statistical confidence staff has in the latter
~~
memorandum) frequency.
Response
At the time the staff selected the recommended screening criteria (SECY-82-465, November 1982), we estimated that the (somewhat conservative) best estimate vessel through-wall crack propagation frgquency for a vessel at the screening criteria was 5 x 10 per reactor year. We did not then have a quantification of the uncertainties involved; we could not then quantify a statistical confidenca limit; and we could not then define an upper bound frequenc/.
However, since that time, in order to confirm and/or improve staff understanding of PTS, we have sponsored three plant-specific analyses at ORNL with thermal hydraulic analyses performed by INEL, LANL, BNL, and Purdue, with design and operating data contributed by three utilities.
One of the objectives of these analyses was to identify major uncertainties.
biases (accura..Two types of uncertainties are involved:
cy), and random uncertainties (precision).
f
=-. -.
s
..r '
1 6
t 4
The biases, some of which are conservative and.some j
i non-conservative, result from simplifying assumptions used to s
make tractable models.for analysis of thermal hydraulics, human actions, and fracture mechanics. We believe the biases in these NRC-sponsored analyses are, on balance, conservative, j
although the amount of bias is difficult to quantify.
The random uncertainties, on the other hand, result mainly from uncertainties in input data.
The major random uncertainty in our analyses.is the uncertainty in the number of pre-existing flaws in the vessel welds.
The ORNL analysis uses a value of 1 flaw / cubic meter of weld, with an uncertainty bound of 500-flaws / cubic meter.
This factor dominates the random uncertainty in the results.
i The methods for addressing uncertainty in the three NRC sponsored PTS analyses were improved as the study progressed, as were the analyses themselves.
The results for the last j
' (and most improved) of the three analyses (based on a-Westinghouse plant postulated to have reached the screening i
criterion) estimated the following range in likelihood of'a 3
i through-wall crack (due to random uncertainty, not corrected for bias):
upper bound, 95th percentile = 2 x 10 s per reactor year mean (expected) value = 8 x 10 S per reactor year j
median (50th percentile) = 2 x 10 s per reactor year i
Because the biases introduced in the analysis appear, on f
balance, to be conservative, we believe the values are t
conservative.
1 Thus, looking at a specific plant and using the best l
techniques developed since the time we first proposed the i
screening RT
, we note that the expected value is close to l
our original SItimate of 5 x 10 5/ reactor year.
N
+
l t
4 i.
l
~.,
.~.
- -.