ML20032D845
| ML20032D845 | |
| Person / Time | |
|---|---|
| Site: | Crane, Cooper  |
| Issue date: | 10/19/1981 |
| From: | Novak T Office of Nuclear Reactor Regulation |
| To: | Atomic Safety and Licensing Board Panel |
| References | |
| TASK-AS, TASK-BN-81-31 BN--81-31, BN-81-31, NUDOCS 8111180114 | |
| Download: ML20032D845 (2) | |
Text
-
8 4
UNITED SCATlS
- f'
, NUCLEAR REGULATORY COMMISSION h-i M,
.E WASHINGTON, D. C. 20555
.h.
1
/'
OCTOBE DISTRIBUTION
+
./
HSilver EHughes b
h'D5cket File)
KL PUR
'TNovak JHard
?
A 80 U
L PDR DEisenhut NJollensten-5 s
% p08 Docket No. 50-289 TERA ECase GVissing 2
/ 8/A NSIC RJacobs
%4f ORBf4 Rdg EHylton
- A
% d%.s DSnyder Q
MEMORANDUM FOR: Atomic Safety and Licensing Board for TMI-l
, g \\\\
FROM:
Thomas M. Novak, Assistant Director for Operating Reactors, Division of Licensing
SUBJECT:
BOARD NOTIFICATION - TMI-l RESTART HEARING (BN-81 31) r
~
This notification concerns the integrity of reactor pressure vessels when subjected to thermal shock and subsequent repressurization during an overcooling transient.
Enclosed is a copy of the October 9,1981 memorandum from William J. Dircks (ED0) to the Commissioners concerning the " Status Report on Pressurized Thermal Shock." This memorandum enclosed an interim report by the Oak Ridge National Laboratory (ORNL) on a NRC research program on pressurized thermal shock which presents a preliminary assessment of the threat of pressure vessel failure.
The ORNL results of preliminary analyses predicted failure of the Oconee 1 pressure vessel for the Rancho Seco type transient of 1978, turbine trip with stuck-open bypass valve, and main steam line break.
,The calculated threshold times for failure were 20, 3, and 4 EFPYs (respectively.
The owner of the plant, Duke Power Company, was asked to review the analysis for accuracy and a joint NRR/RES team is reviewing the report. The team i plan's to complete its assessment with a draft report regarding the validity of the ORNL report within the next two weeks. We will notify the Board of the results of this review.
~
"0RIGInn grgpjn Thomas M. Novak, Assistant Director for Operating Reactors Division of Licensing
Enclosure:
As Stated cc w/ enclosure:
See next page 8111100114 811019 DR ADOCK 05000289
^
p PDR
/
omcr >... 0RBf.;.0L.......CQ. ggfd.:DL.r..... AD. :
- D.....
. -.. -. - ~ ~.
suRnee>..... G......n,g,,c f.,,J,S,,tc
,,0.E..
10/
./. 81 1
- 1....
......./...b.../ 81 10/
1 s
i
-... ~ ~. -.. -
~ ~ ~. ~. -... I omy NRC FORM 318 (10-80) NRCM 024a OFF1CIAL RECORD COPY uso m ini--m no
g.
UNITED STATE.S OF AMERICA NUCLEAR REGULA, TORY C01tilSSIO,N BEFORE THE ATOMIC SAFETY AliD LICENSING SOARD In th's Matter of METROPOLITAH EDISON COMPANY,
)'
(Three' Mile Island, Unit,1)
CERTIFICATE OF SERVICE Mr. Henry D. Hukill, Vice President and Director - TMI-1 i
Metropo.litan Edison Company i
P. O. Box 480 Middle, town, Pennsy,1vania 17057
. Ms...arjorie M. Aamodt
- Ivan V. Smith, Esq., Administrative u
R.D. i5 Duite A::.i: Safe:y & Licensing E:ard Panel cea:esville, PA 19320
~
U.S. Naciez,r Regulat:ry Cer.ission Vashin:::n, D.C.
20555 Mr. Thor.as Gerusky Eureau of Radiation Fr:tection
~
Dr. Waiter H. J:rdan, Administrative C-sp of Envir:nrental Resources P.O. 5:x 205I Jud:e Harris:urg, Pennsylvania 17120 E51 W Guter Drive Oak Rid:e, Tennessee 37530 Mr. !!arvin I. Lewis
~
Dr. Lir.da W. Little, Administrative 5504 Eracford Terrace Philadelphia, Par.nsylvania 19149 Judge 5000 He mitage Drive Metropolita'n Edison company Raleich, North Carolina 27512 ATTN:
.J.G. Herbein, Vice President
~
Gecree F. Trowbridge, Esq.
P.O. Box 542 Shaw Pittman, Potts & Trowbridge Reading, Pennsylvania 19503 i-
~
1500 M Street, N.W.
Washincton, D.C.
20006 Ms. Jane Lee R.D. 3; Box 3521
~
~
~
Karin W. Carter, Esq.
Etters, Pennsylvania 17319 505 Executive House Walter W. Cohen, Consumer Advocate-
~
P. 0. Box 2357-Harrisburg, Pennsylvania' 17120 Department of Justice Strawberry Square,14th Ficor Harrisburg,JFennsylvania 17127 i
Hen:rable Mark Cohen 512 D-3 Main Capital Eu'iding Harrist;:rg, Pennsylvania 17120 I
, -. _ = - ~ - -
_m-
Tho.as J. Gemico
,4 John Levin, Esq.
t-De uty Attorne,y General Pennsylvania Public Utilities Co::rn.
Di ision of Law - Room 316
-2 mox 3265 1100 Ray =ond Boulevard Harrisburg, Ppnnsylvania 17120 Newark, New. Jersey 07102 Allen R. Carter, Chairman Jordan D. Cunningham, Esq.,
Joint Legislative Committee on Energy Qx,FarrandCunningham ca20 North' 2nd Street rest Offs.ce oox 142-Harr9sburg, Pennsylvan$a )7)10
'~*
Suite 513 Senate Gressette Building
. Louise. liradford Columbia, South Carolina 29202 1011 Green Street
~
Harrisburg, Pennsylvania 17102
,.c,ert Q. Pollard 609 P.:ntpelier. Street Ealtimore, Paryland 21218 Ms. Ellyn R. Weiss Chauncey Kepford Harmon & Weiss 1735 I Striet, N.W.
vudith H. Johnsrud Suite-505 Environmental Coalition on Nuclear Power Washington, D.C.
20005 453 Orlando Avenue State College, Pennsylvania 16801 Mr. Steven C. Shelly Union of Concerned Scientists Ms. Frieda Serryhill, Chairman 1725 I Street, N.W.
Cc'alition for Nuclear Power Plant Suite 601.
. Postponement Washington, D.C.
20006 2510 Grenden Drive Wi16ington, Del. pare 19508 ACR5 liechers Gail P. 5radford
.03?J Mr. Myer Eender 545 W. Philadelphia St.
Dr. Max W. Carten Erk, Per.r.sylvania 17401 Mr. Jesse C. Ebersele
.Mr. Har:1d Etherington
- .t:mic Safe y and Licens'ing Appeal 5 card Tr. William Kerr U.S. N;: lear Reguia cry Comissien Dr. Harold W. Lewis Unshin;.cn, D.C.
20555 Dr. J. Carsen l' ark P.r. Uilliam M. Mathis
'A :m.: Safety and Licensi.; 5:ard Panel Dr. Cade W. Moeller U.S. N;:' ear Regulat:ry Cea.ission Dr. David Cirent Dr. Milt:n 5. Plesset Washin: ten, D.C.
20555 Mr. Jeremiah J. Ray Dr. P.aul G. Shecon
' Secretary U.S. Nuclear Reculatory Comission Dr. Chester P. Siess ATTH:
Chief, DEcketing.& Service Br.
Mr. Davis A. Ward Washington, D.C..
20555
~
William 5. Jefdan, III, Esq.
~ Harmon & Weiss 1725 I Street, N.W.
g
/Ng_ Oh3((
. ng on, D'.C.
20005
/
1.ucir.da Lcw Swartz
~~
Counsal for NRC Staff ee 6
UNITC9 STATES h? ateeg
- (
p
,[
NUCLEAR REEULATCRY COMMisslON g
wassmaTow. o. c. mosas g
(
"%,,g #
October 9,1981 MEMORANDUM FOR: Chaiman Palladino Commissioner Gilinsky Comissioner Bradford Comissioner Ahearne Comissioner Roberts FROM:
William J. Dircks Executive Director for Operations
SUBJECT:
STATUS REPORT ON PRESSURIZED THERMAL SHOCK In infomation paper SECY-81-286 on pressurized themal shock of pressure vessels and in the subsequent briefing of the Comission on June 11,1981, the Comission was infomed that Oak Ridge National Laboratory was pre-paring a status report on pressurized themal shock. An interim report focusing on Oconee-1 has now been completed; a preprint copy is enclosed.
As noted in the enclos'ed report, the purposes of this ORNL work are to identify what is presently known about this problem including major areas of uncertainty and sensitivity, to identify further infomation needs, and to propose and evaluate possible mitigative measures. This l
interim report organizes what is presently known and presents a preliminary l
assessment, based on present analysis, of the threat of pressure vessel l
failure. Although the rominal calculations indicate a proximate threat, the report points out a number of flaws in the current thermal-hydraulic analysis which reflect a lack of realism. The owner of the plant, Duke Power Co., was asked to review the analyses for accuracy; the Duke representatives also challenged the validity of currently available analyses.
~
In order to evaluate the conclusions of the report and to assess their significance, a. joint NRR/RES team has been set up to review the report as soon as it arrives in NRC.
In particular, the staff will evaluate the probability of occurrence of the severe overcooling transients assumed and 8
i l D. nl n],
s g.
v 7 - )
v
The Comissioners the conservatisms used in other portions of the Oak Ridge analyses. The team will be led by the NRR task manager, Roy Woods, and'it plans to com,
plate its assessment with a draft report in 2 weeks.
We will keep the Comission informed of the results of this review.
William J. Dircks Executive Director for Operations I
Enclosure:
NUREG/CR-2083. Evaluatiion
~
~ of the Threat.to PWR Vessel Integrity Posed by Pressurized Thennal Shock Events cc w/ enc 1: PDR cc w/o enc 1: SECY OPE OGC.
e N
l l
l 1
l t
l l
\\
CCERUf REPORT s
NUREG/CR-2083 UE ORNL/TM-8072 Disc. Category RG Contract No.
W-7405-eng-26 Instrumentation and Controls Division EVALUATION OF THE THREAT TO PWR VESSEL EIT?CRIIT POSED BY PRESSURIZED THERMAL SHOCK EVE! prs R. C. Kryterl T. J. Burns 2 R. D. Cheverton3 R. A. Hedrick' F. B. K. Kam5 C. W. Mayo 4
~
Oak Ridge National Laboratory Oak Ridge, TN 37830 Manuscripted Completed - October 7,1981 Date Published -
CAUTION The document has not been even final posent clearance and the dieseannesson of is informodon is only for ofHcial use. No reisene to the put46e shed be made without the approval of the Law Department of Union Car 06de Corporeboa Nuclear D6-vtsion.
' Instrumentation and Controls Division, Oak Ridge National Laboratory.
2Engineering Physics Division, Oak Ridge National Laboratory.
3Engineering Technology Division, Oak Ridge National Laboratory.
AScience Applications, Inc., Oak Ridge 3 ranch Office.
50perations Research and Development Division, Oak Ridge National Laboratory.
Prepared for the U.S. Nuclear Regulatory Commission Office of Nuclear Regulatory Research Under Interagency Agreements DOE 40-551-75 and 40-552-75 NRC FET No. E0468 Prepared by the OAK RIDGE NATIONAL LGORATORY Oak Ridge, Tennesne 37830 operated by UNICN CARBILE CORPORATION for the DRAFT DEPARTMENI 0F ENERGY DRAFT ECERDi REPORT
[qe 3f ai,,
'~
y e
'i i
INTERDi REPORT Evaluation of the Threat to PWR Vessel Integrity Posed by Pressurized Ther=al Shock Events Task Coordinator:
R. C. Kryte r'-
2 3
Contributing Authors:
T. J. 3 urns, R. D. Cheverton,
5 R. A. Hedrick*, F. 3. K. Kam, and C. W. Mayo"
'.0 I"TRODi!CTION Pressurited water reactors (PWRs) ars susceptible to certain types of hypothetical accidents that, under some circumstances, including operation of the reactor beyond a critical time in its life, could result in failure of the pressure vessel as a result of extensive propagation of crack-like defects in the vessel wall. Accidents of particular concern are chose that result in rapid cooling (thermal shock) of the inner surface of the reactor vessel (RV) wall, par".icularly if they also involve substantial primary-system pressure.
(Such accidents have been referred to as " overcooling accidents" (excssive cooling for a particular pressure) and/or " pressurized thermal shock. "]
For a particular accident and operator and system response, the tendency for preexistent vessel flaws to propagate during the: mal-shock loading conditions is a funcrton of the relative magnit* ces of the stress field or stress intensity factor (K ) and tha material fracture-r and arrest-toughness values (KIC and Kr ).
These toughnesses decrease with decreasing temperature and increas,ing fast-neutron fluence, and K increases' with increasing stress and is greater for a surf ace flaw 7
enan for a buried flaw. Thus, flaws on the inner surface of the RV wall are of greatest concern for thermal-shock loading.
I l
The positive gradient in temperature that exists within the wall during a ther=al transient and the negative gradient in fluence together in K, that provides a mechanis= for result in a positive gradient r
a rrest of a fast-running crack. However, if the primary-system pressure is high enough, the gradient in K7 nay be such that l
i
!sstrumentation and Controls Division, Oak Ridge National Labo rato ry
-2 Engineering Physics Division, Oak Ridge National Laboratory l
3 Engineering Technology Division, Oak Ridge National Laboratory
Scietce Applications, Inc., Oak Ridge 3 ranch Of fice 50perat' ions Research and Developnent Division, Oak Ridge National Labo ratory i
I m
1-2 arrest will not take place, and the flaw will then extend through the vessel wall. Depending on the temperature and pressure of the primary system and the length and orientation of the flaw at the time of its wall penetration, the opening produced could either be negligible in size or sufficient to preclude adequate cooling of the reactor core.
For instance, previous oeercooling-accident calculational indicate that in the event of a double-ended pipe-break loss-ef-coolant accideut (LOCA), which produces perhaps the most severe of all thermal shocks but also a very low vessai pressure, flaws presumably will cot bu driven through the wall.
In another case 2 the internal presr. :e remains rather high, the coolant temperature remains well tbove saturation for atmospheric pressure, and RV failure with a sizable opening is predicted.
As already mentioned, the tendency for crack propagation increases with increasing reduction in material toughness and thus with increasing fluence. An additional factor that influences the extent of the radiation-induer.a reduction in toughness of present-day reaccor pressure vessel materials is the presence of -impurities such as copper and, to a lesser extent, phospho rous. Within '. sits, the higher the concentratiou of these two elemanes the greater the radiation-induced reduction in toughness for a given fluence.
In the context of calculated flaw behavior under pressurized thermal-shock loading conditions, a broad range of copper concentrations exists among PWR pressure vessels currently in service. Some of the vessels in the high-copper category appear, on the bases of selected hypothetical accidents, assumed initial flaws, and presumably conservattve acalyses, l
to be susceptible to failure at early dates, while vessels with low j
copper category are not susceptible to failure for an extensive period.
Because of the apparent severity of overcooling accidents and the obvious complexities associated with defining accidents and their likely f requency of occurrence, performing realtstic systems analyses to determine appropriate input temperature and pressure transients for the vessel integrity studies, and accurately evaluating the mechanical integrity of the pressure vessel, thorough plant-specific studies are in order.
In May 1981, the U.S. Nuclear Regulatory Commission (NRC) requested assistance 3 from the Oak Ridge National Laboratory (ORNL) in attaining such an understanding of the severity of the threat posed by pressurized thermal shock occurrences, subject to the constraint that an interim report which would consolidate, evaluate, and summarize all the pertinent data and analyses identified and collected must be produced in four months time. This short time f rame precluded undertaking new studies and calculations of significant magnitude, so the evaluated results cited in this report are necessarily drawn f rom known previous work and literature search.
t
..,.,,,,,.,_n-_
1-3 The maj:r goals of this ORNL integrative effort were to (1) identify what is presently known about the pressurized charmal shock problem, including the major areas of uncertainty and the sensitivity of the estimated severity of threat to these uncertainties; (2) identify what is not known about the problem, including suggested :neans for correcting any such deficiencies; and (3) propose and evaluate possible sitigative measures. Le work required to meet these goals was divided into six principal tasks:
1.
Define the problem elements that dominate in establishing the overall likelihood of RV failure and develop a scheme for assessing the relative safety significance and likelihood of occurrence for the spectrum of possible inittacing and subsequent events.
2.
Review presently existing thermal-hydraulic analyses of various postulated overcooling scenarios and critically assess their
- realism and usefulness in defining a generic spectrum of overcooling events.
Identify critical assumptions and input uncertainties and estimate their probable effects on the predicted temperatures and pressures.
3.
Review the function of plant-specific control and safety systems, along with procedure-directed operator actions. Consider system modifications which would help to lessen the severity and f requency of overcooling transients.
i-4.
Estimate the overall severity of threat to RV integrity imposed by pressurized thermal shock occurrences.
5.
Propose potential corrective actions which might be effective in reducing the severity of threat. Discuss probable effectiveness and relative ease of implementation.
6.
Provide recommendations for extending the study in an effective i
manner in FY 1982 to obtain a broader, more balanced understanding i
of the problem as it relates to the spectrum of current plant designs.
The selection of the first representative plant to be studied was somewhat arbitrary but in consideration of an extensive history of thermal-hydraulic upsets in Babcock and Vilcox (B&W) plants and the low thermal inertia provided by the 3&W once-through steam generscor l
design, a reactor built by this manufacturer seemed a reasonable choice.
Since Oconee-1 has a RV with longitudinal welds having a relatively high copper content, is the lead 3&W plant (coemerical operation began in July 1973), and has a larger cumulative power history (~4.9 EF7Y to date) than its sister units, this plant was selected (with NRC concurrence) to provide a basis, so far as practical, for our initial study. On the other hand, because thermal-hydraulic behavior needs to be further evaluated as recommended later in this report and because their are special control systems provisions in Oconee-1 limiting transients, more analysis needs to be done before their results are applied to Oconee-1 or generalized to other plants.
(
1 -
--a
,e-
1-4 REFERENCES - CRAPTER 1 1.
R. D. Cheverton, S. K. Iskander and S. E. Bolt, Applicability of LEFM to the Analysis of PWR Vessels Under LOCA-ECC Thermal Shock Conditions, ORNL/NUREG-40 (October 1978).
2.
R. D. Cheverton and S. K. Iskander, " Thermal-Shock Inrestigations Heavv-Section Steel Technology Program Quarterly Progress Report",
for Januarv - March 1981, ORNL/TM-7822, pp. 76-83.
3.
Letter, R. M. 3ernero (NRC) to A. L. Lotts (ORNL), " Report on Pressurized Thermal Shock," dated May 11, 1981.
O
/
e
- _ ~
___y--,_
2.0 OVERCCOLING TRANSIENTS IDENTIFIED AS SAFETY CONCERNS There are three basic mechanisms for rapidly tooling the primary coolant system: depressurization of the primary or secondary system, injection of cold fluid, and rapid removal of energy through the steam generator. Four general classes of transients can be identified as encompassing one or more of these cooling mechanisms:
o Large-Break Loss-of-Coolant Accident (D LOCA) o Small-Break Loss-of-Coolant Accident (S3LOCA) o Main Steam Line 3ceak (MSL3) o Runaway Feedwater Transient (RFT)
The severi:y and probability of occurrence of each of these transients is dependent on plant-specific characteristics.
The G LOCA produces primary fluid temperature temporal derivatives on the order of 36,000*F/hr, arresting at a base temperature of ~350*F.
To this system is injected 40-85'F high pressure injection system (HPIS) fluid and 90*F core flood tank (CFT) fluid, which results in rapid chilling of the fluid next to the RV and causes an effectively conduction-limited temperature transient in the vessel wall.
i The SBLCCA, in contrast, produces order of magnitude lower primary fluid temperature temporal derivatives than the GLCCA, generally less than 2200*F/hr, d2pending on the size of the break. Also, depending on break size, the CFT system may actuate in addition to the HPIS. A critical difference between SBLOCA and L3LOCA is that the HPIS can repressurize the system for many break sizes.
The MSU usually produces primary fluid temperature camporal derivatives that lie between those of the GLCCA and SBLOCA. These decreasing temperatures result from the rapid primary system energy removal produced by flashing of the fluid on the secondary side of the steam generator. The lowest primary fluid temperature achievable in j
this transient is determined by the performance of the steam generator feed train, HPIS, and CFT.
i The RFT is essentially a variant of the MSG, but without the initial I
rapid steam generator secondary-side blowdown and the resultant rapid removal of energy from the primary system. The primary system temperature temporal derivatives for the RET are usually the lowest among the four classes of transients. The progress of the RFT is totally controlled by the performance of the steam generator feed train.
i T
m
,sy-vy,,
w-wy-
=
-r-yr9 mme
..m-
3.0 SEVERITY OF THE THREAT 3.1 Probability of Occurrence of Initiating Events Ultimately, the probability of a thermal-stress-induced challenge to the reactor pressure vessel is dependent on the f requency of requisite iritiating events. However, the concern to this study is not the probability of individual initiating events themselves, but rather the total probability that the thermal-hydraulic transients resulting f rom the initiating events produce pressure and camperature conditions which approach the structural limitations of the RV.
This total probability can be viewed as the multiplicative combination of three probabilities:
(1) the probability of an initiating event, (2) the probability that the control and safety systems fail to respond to the transient in an appropriate manner to protect the vessel, and (3) che probability that the reactor operator fails to diagnose the exact nature of the transient and therefore fails to take appropriate action or possibly takes action which actually exacerbates the transient.
As noted previously, four transients were identified as possible thermal shock initiators: a small-b reak LOCA, a large-break LOCA, a main steam line break, and a runs.ay feedwater transient. Due to limitad time and resources, a detailed characterization of the various factors (i.e., initiating events and system / operator responses) for each transient has not yet been performed. An estimate of the prob bility of each' initiating event which could be a precursor to conditions having the potential for thermal shock to the RV was made.
Probability Per Reactor Year Estimate gg SBLOCA*
3 x 10' 3 x 10-5 e3 3 x 10-3 I3 LCCAa i x tow 1 x 10-5 to 1 x 10-3 MSL3a 5 x 10-3 1 x 104 to 1 x 10-"
ByT 1.0 0*I-V??
To 1.0 The RFT event is the most complex with a very high probability for the i
initial event but requiring failure in order to produce ther=al shock.
i This is very plant specific, and for Oconee-1 it appears that sultiple independent failures are required (see Section 4.5.4).
aEstimated f rom Raactor Saf ety Study, WASH-1400.
a
,.,, +
,-n
3-2 Table 16.
Summary of Pressurized Thermal Shock Evaluation Mechanistic Results INITIATING EVENTS La rge-3 reak Small-B reak Main Runaway SUBSEQUENT LOCA LOCA Steamline Feedwaeer EVENTS (GLOCA)
(SBLOCA) 3reak Translenc (MSI3)
(RFT)
Thermal-TRAC simulation (a) Rancho Seco (a) TRAC simula-(a) Rancho Seco Hydraulic (Westinghouse (actual plant tion (b)IRT simula-Infor ation Plant) t ransient)
(b)IRT simulation tion Sources (b) TRAC si=ula-tion Operator None (a) ' Rancho Seco (a) TRAC:
(a) See S3LOCA Actions operator over-initiates auxil- (b) None Taken or rode automatic iary feedwater r
Assumed trip of main at 30 s.
Main feedwater pumps feedwater (b) TRAC: ini-ramped at 40 s.
tiates aux. feed (b) IRT: none water at 30 s.
Main feedwater ramped at 40 s.
Thermal-No repres-(a) Rancho Seco:
(a) TRAC:
(a)See SBLOCA Eydraulic surization of
- Repressuriza-
- No rep res-(b)
Indications primary coolant tion surization
- Repressuriza-or system
- Tsin = 280*F Tmin - 350*F tion Predictions (b) TRAC:
(b) IRT:
- Tsin < 150*F analysis
- Rapressurf za-terminated tion prematurely
- ! sin < 150*F Vessel Crack initiation (a) Vessel fails (a) TRAC:
(a) see 53LOCA F racture and arrest (no at 20 EFPY Analysis not yet (b) IRT: Vessel Mechanics vessel failure)
(b) Insufficient available fails at 3 EFPY l
Predictions at 20 EFPY information for (b) IRT: Vessel i
analysis fails at 4 EFPY i
Limitations No (a) Ranch Seco:
(a) TRAC:
(a) See SBLCCA and rep ressuriza-
- Pressure &
- Mild case (b)
Concerns tion, so of tempe rature
- Feedtrain
- Assumes secondary data not as tables f eedwate r 1
concern entirely ade-(b) IRT:
control quate
- Assumes feed-failure vater control
- Feedtrain (b) TRAC:
failure treated with
- Press / temp.
- Feedtrain tables data incomplete as cables l
?
3-3 3.2 Pressure Vessel Integrity For the purpose of these studies the integrity of the pressure vessel was considered to be jeopardized if the f racture mechanics analysis indicated that an inner surface flaw would propagate through the ussel wall as a result of an overcooling accident. Four specific overcooling accidents for which f racture-mechanics analyses were performed are the large-break LOCA, Rancho Seco (1978), turbine trip with stuck-open bypass valves, and main steam line break. Results of the preliminary analyses indicate that the vessel will not fail as a result of the L3LOCA, but failure was predicted for the other three accidents. The calculated threshold times for failure were 20, 3 and 4 EFPYs, respectively, based on a fluence race of 0.046 x 1019 neutrons /ca'/EFPY, which is similar to that for B&W plants.
The size of the break that results f rom propagation of the flaw through the wall is of utmost importance since it is a factor in determining whether the vessel will be able to retain sufficient water to cool the co re.
Because cooling temperatures at some, if not all, locations in the primary system are expected to be well above 212*F at the time of predicted failure (excludes L3LOCA), there is a large amount of stored energy that will be released, and thus a potential exists for a rather large opening in the vessel wall. A more quantitative assessment of the problem awaits completion of detailed studies.
n J
(
l 4.0 PLANT CONTR01 AND OPERATIONS -
4.1 Introduction Plant control systems and operator actions were reviewed to define system secpoints and capacities relevant to pressurized thermal shock transients. Potential feedwater control failures and operator actions were investigated and the control system and operstor actions which took place in the Rancho Seco overcooling transient were reviewed.
These data were used to evaluate control system response assunptions employed in the thermal-hydraulic analyses available to us and to develop conclusions and recommendations concerning control system modeling for pressurized thermal shock thermal-hydraulic analysis, operator actions, and potential problem areas.
4.2 Reactor Protection System The Reactor Protection System (RPS) is a safety-grade system designed to trip the reactor according to the values of a variety of input parameters, and thereby to protect both the core from fuel rod cladding damage and the reactor coolant system from overpressurization.1 Reactor trip directly influences main feedwater control through the Integrated Control System? (ICS) unit load control. The neutron power signal obtained from the RPS can also modify main feedwater demand if its mismatch with the ICS reactor demand level exceeds a set tolerance.
Following a reactor trip, the heat generated by the reactor is' determined by the shutdown rate.
In order for the remainder of the unit to " follow" the reactor, the unit load demand (and hence the total' feedwater demand) will track the actual megawatts generated at a i
maximum rate of 20% per minute. For transients involving initial c~ pressurization, the RPS will trip the reactor at a low pressure set point of 1925 psi. For transients initiated by a turbine trip, the reactor will be tripped at the start of the transient.
4.3 Eigh Pressure Injection 4.3.1
System Description
1 The Eigh Pressure Injection System (HPIS) injects water into the four reactor vessel inlet pipes upon actuation of appropriate trips in the engineered safety features system.
The EPIS comprises three high-i pressure pumps; the flow can be controlled manually and the pumps can be aligned manually in several different ways.3 Normal HPIS actuation will inject full flow from two of the three pumps, with suction taken from the borated water storage tank C3WST).
l
4-2 l
4.3.2 Capacities -
The HPIS pump characteristic performance curve is shown in Fig. 4-1.
The 3WST has a volume of 388x103 gallons. The RPI valves will be fully open within 14 s from an actuation signal, and the pumps will be up to speed within 6 s.
4.3.3.- Set Pointsl The EPIS will actuate when the reactor coolant system pressure drops to 1500 psi.
4.4 low Pressure Injection 4.4.1
System Description
The Low Pressure Injection System (LPIS) injects water into the reactor vessel downcomer through two pipes located on opposite sides of the core and at ninety degrees from the reactor vessel outlet nozzles. Low pressure injection is provided by three low pressure pumps (operated in parallel) and two accumulators.* Pumps are normally aligned to draw from the SWST, but can be manually transferred to take suction from the reactor building sump. The pump flowrate can be controlled manually.
4.4.2 Cacacitiesl n e LPIS pump characteristic performance curve is shown in Fig. 4-2.
The LPI valves will be fully open within 15 s af ter actuation, and the pumps will be up to speed within G s.
3 As stated previously, the 3WST has a capacity of 388x10 gallons. When considering the total inventory of borated water available to the LPIS, it must be noted that the containment spray system, if actuated, will also draw from the BWST. The accumulators have a combined capacity of 21x103~ gallons.
4.4.3 Set Pointsl Ihe accumulators will discharge water into the reactor vessel when the pressure falls below 600 psi, and the LPIS punps will be actuated whec the primary pressure falls below 200 psi.
4.5 Main Feedwater Control 4.5.1 System Descriotion Main feedvater control is one function of the ICS. A feedwater denand signal is developed, based on unit load denaud but also eaticed and
40 I
l I
/
I NPSH
/
.=
e
/
30 C
i l
5 7000 pTDH 25 i
e C 6000 20 0 5000 15 4000 w
10 C
l l
g a 3000
[
i i
l 1
l 1
o 2000 l
l l
1 1000 O
100 200 300 400 500 600 CAPACITY, gpm i
i
.ig. 4-1.
High Pressure Injec:1cn Pt:mp Charac: eristics (F ce
(
Occuee FSAR).
i l
4-4 500 400 N
c'<
d I
c W
y 300 I
o TDH
\\
c N
w k
c o
s 200 o
NPSH 20 l
s
'l 15 X
100 to 0
1000 2000
'3000 4000 5000 CAPACITY, spm Fig. 4-2.
Lov Pressure Injec:ica Pump Charac:aristics (Fre::
Occuee FSAR).
I 9,,_
+ - *,
" ~ " "
=
e 1
4-5 limited by a number of feedback functions. The loop demand signal is compared to the measured feedvater flowrate so as to regulate the main feedwater control valve. The total correctad feedwater demand signal is modified to control the feedwater pump speed, and so to meet the feedwater demand and to maintain constant pressure drop across the main control valves. Steam generator low-level and high-level limits intercept the feedwater demand to pravent underfill and overfill conditions.
The main feedwater pumps are supplied water from the' condenser hotwell, the surge tank, and the condensate storage tank through three condensate booster pumps and three hoewell pumps.
A variety of feedback and trip cunctions affect the main feedwater cor~rol. Manoal control of all feedwater pumps and valves is also possible.
l 4.5.2 Capacities 3 lbs/hr from the hotwell per The full main feedwater flow is 6539x10 steam genarator. The maximum water inventory in the feedwater system is 295x103 gallons.
4.5.3 Set Points o The total feedwater demand will run back at a maximum rate of 20% per minute to track generated megawatts following a reactor trip.2 Operators are presently required to trip the reactor coolant l
o pumps follouing actuation of the engineered safety features system. Tripping the reactor coolant pumps will, in turn, cause the feedwater demand to run back to a maximum value of 20% at a rate of 50% per minute.Z o The Oconee-l unit has a steam generator high-level limit that I
will trip the main feedwater pumps if the steam generator is l
filled to this level.a (This trip may not be present on other 3&W units).
o The Oconee-1 unit will also trip both main feedwater pumps if there is a loss of ICS power.3 (This trip may not be present on other 3&W units.)
All hotwell pumps will be trippeda when the hotwell level o
l reaches " emergency low."
o All ecadensate pumps will be tripped when the condensate booster pump suction header pressure decreases below 43 psig.3 4.5.4 ICS Failure Modes and Effects The ICS is a complex, nonsafety grade control system. A failure modes and effects analysis (FMEA) review was therefore performed to
t l
C.O.5. l... l
.O.
. 04.._A..G J_
.W.
.W 4
.W.
T T
- ,,3
-l Q-Q-
{
e-..
4..
,,.3 i
I
.=:.
1-@
@-I
.=:.
I I
@-1 "
n-I ne.
r I "=rn" - Q r........
@-I"=rn-I
@-^i y
[^_r.
^-@
t
^
2 i--i
-i-o-
i i ~'
i l
I
_l.. W....... l-@
g-l
..W...... L
_L
_L
_3
_I
_L i
................................................l.........1....................I.......T.
co..
to.
ease sn..
euw se..
u.we.
=am
- u...
u..
WA..o W.W.o coa. W.o.
coa.W.o W.
A WA A
WA 3
Fig. 4-3.
ICS Ceneral Area lireakdown.
e
4-7 investigate potential failures in the ICS that might lead to excessive feedwater. This review divided the ICS into three general areas, as shown in Fig. 4-3.
The first area constitutes higher levels in the ICS, where significant feedback and feedwater flow limiting actions '
will be effected from a variety of other process signals. The second area is characterized by limited ICS feedback, but manual control of the feedwater system is still possible. The third area encompasses failures below all control points.
The results of this analysis, which are summarizad in Table 4-1, show that single control failures belos he m,nual control points always leave one or more alternate means by which the feedwater may be manually controlled. Manual control is required, in general, to assure proper feedwater control.
ICS failures in region 2 will limit i
feedwater flow to the steam generato; high-level limit.
In the case of Oconee-1, if the ICS high-level limit does not act, a separate high-level signal will trip the main feedwater pumps. Feedwater flow will be limited for ICS f ailures in region 1 by both the high-level limits and by feedback from other parameters.
It should be noted that without the steam generator high-level trip for the feedwacer pumps, failure of the startup level signal to a " low" condition can result in unlimited overfeed to the steam generator.
ICS f ailure or power failure will generally lead to a loss of feedwater, due to a zero demand speed signal being presented to the main feedwater pumps. Oconee-1 also has a trip to stop the main l
feedwater pumps on loss of ICS power. The presence of feedwater control for a loss of ICS power condition is known to be highly plant specific.
4.6 Rancho Seco Transient A significant overcooling transient occurred on March 20, 1978 at the Rancho Seco reactor following a failure of power supplies that fed both control room indicators and the ICS. The initial plant response was a loss of feedwater transient combined with incorrect indications presented to the operators. Auxiliary feedwater was also supplied to one steam generator through an ICS control path (no longer present in 3&W units). As a result, the operator was presented with an indication of 0% feedwater demand for one loop and 100% feedwater demand for the other.
Eis response was to manually remove the main feedwater pump trips and so restore main feedwater. Upon restoration of the instrumentation power, it was discovered that the reactor had been overcooled and corrective actions were taken.3 The instrumentation and control system power supplies have since been i
upgraded.
The Rancho Seco incident clearly demonstrates that significant overcooling trantients can be induced by operator action.
In this l
case, the operator's actions were the result of unsuspected and
e p1 r
y e
l d
l d
t
)
l e
l o e
a s-r a l
or s
w.
t ep eu l
rt o
d y i
f ve s
pn o
t n l
el m
ol s l
oa r
no c
el
.i a
l r m t
oc f a d yl t vs i
p n
c e
u nl n
a.
F) d o
d b
pn al l efh p
(
ne c
d n ua ae d o i
t il na n
t m euv n
1 r
i al e
u a
r vne el - t
.m t o h
od y c
ad l al poe yi nr r el t e am-er ep l l it n
ael s
S s v
h dt nm l
an a
pa e
o d g nnou al mo c
esu v
l l ei iocp ue c
g n
l
.c ol h cO nv o
d apa a
e rl n
r ae t e e
k mm v
l e t oo el e
ml b
e au bb nrt pa
.t l
l e
p eP d k
a ot ouea dh o
o vo s
l e
c l o cnr nl w eg r
r l s l
o is oo l abd l i t
t al p
r
.l l
al
- ct iMi e l h n
n va m
o) o b
va n
a a
se o
u o
o fd r a
eer f
sf ro i
C C
Wn p
et r
n pb e
.o t t F a t s n e
sa o
n sd pl n
d m
c d
poo t
ec ne en e
or n
r e
n el c a
v sag vad v co a
e ss a
cc w
l e
ec l mee t
T oe x
e d
av v
m aeel ea s
l v e
enb e
vl l ea vd p-h r e
g cl v
e e
a ave sh e
r n
a l
th n f.
k v vl t rr g
nn u
o l V ay cwa y
c as e e pi ae l
l l
vl e( c pl ol l v t t mh cg i
i l
f ul l o o
s aau a
w.
k a f ee t a b r rkl wwpe em F
w cu evv r u t
t cl dd t
va S o on l l an on noi eed a l
e r
Cl l a
)
aa t a wo ol F eenr at k
e I f B m P vv S m Tc Cb(
Ff aa V s taw dee F
e o
t v
d i
e n
s e
o s
p n
n n
n i
e e
s e
e e
e t
c d
p p
p p
c x
o l
O O
O O
e E
M l
r u
s s
s s
r e
f l
l l
l i
o r
d e l
i i
l w
v f
u o
a a
a a
w o.
nl l
t F
F F
F o.
1 i a w
i v
I e
a s
e e
e e
s s
i F
l v
v v
v l
l l n v
l l
l l
l l
l l
e e
a a
a a
a a
a ap H
F V
V V
V F
F F o AE k
H c
F a
b d
e 1
n e
o F
4 i
t r
e a
k e
l t
c t
b S
o e
a
)
l r
r m
T l
s B
e e
a a
(
t t
r i
p r
r a
a a
w r
n i
m ee e
we w
P o
a u
t v t
d v d
l l
e e
t r M
P al a
el ee r
F v
ao wa w
ea ev e
e.
wr S
r dV d
FV Fl h
r d r I
C e
e e
a t
e uE I
t el e
pl pV O
t p
e a
F o F
uo u
a u
Fd w
w r
e t r t k w
w t
n o
d nt nv rt r c o
d r
pa l
e io il an ao l
e a
om e
e ao aa t o t l e
e t
- o. De B
F HC MV Sc Sif H
F S
l
. s pd yt) me l i.
ue l ml pf aie ul v rs n
e ea.
al l th r me a
e vh w! w d eg d - o eep el i l
h ee l d F nS ont oC raa cI t
.O l
n
,l e
f o
oso z
o r
ct r e.
t it et s n
emn ris o
bi o f mo C
l c il n
sl m
alid e
n r
cTn vd o e
B a l e T
p aep m
, e vpi g
uss sr n
pt a l - t i e ow o
d mr rop l
nic tl m al n n
u i
oop ss ct est r
voi wse l r m oet aCi l oa V( l F gw SC o
I t
o n
t o
i r
e td e
d cn w
o ea o
H rm P
i e e
dd C
r A
u ne l
i s f
i a
o a
se F
l r s
i c s
an o.
Fi I
kca b
dee
)
F de r
u e
n t
i e
t m
n a
d o
r n
C a
a
(
P l m ak r
i r
t e
e o I w
h Tr o
1 t
e P
4 O
rtoa C
e e
w r
A l
v pd e
b o
oe h
S a
b
- o. Fe t
C T
A O
I I
I i
I 9
L
4 4-10 widespread information failure.
It should be notad that as a result of this and other power-supply-failure-induced control system transients, all 3&W units have since been required to review their system power supplies and make modifications and develop procedures as necessary to reduce the probability of such events and to provide operator guidance for controlling the unit.
4.7 Operator Actions Our review of the ICS failure modes and effects analysis (FMEA) indicated that a few single or double control failures can lead to uncontrolled feedwater supply, but that the majority of potential failures have feedback paths that will act to reduce feedwater automatically. It is concluded that there are a larger number of ways that excessive feedwater can result from operator errors of commission than from errors of omission.
Oconee-1 event sequences for small steam line break and excessive feedwater were reviewed for indicated operator actions. These event sequences assumed multiple system failures and failure to control different systems properly along the event paths. Potential severe consequences were identified where key problems remained uncorrected.
These event sequences show the necessity for correct operator actions along a number of paths.5 4.8 Review of Thermal-Hydraulic Calculations Available thermal-hydraulic analyses were reviewed for correctness of assumed control system response. These analyses were found to include operation of the safety systems as designed but to employ rather arbitrary assumptions about feedwater system operation.
In particular, the TRAC calculations for MS13 and S3LOCA assumed that the main feedwater continues to supply 100% flowrate and is then reduced to zero by operator action over a two-minute period starting 40~
s into the transient. Emergency feedwater is assumed to be initiated 30 s to the remaining intact steam generator for the MS13 and to at both steam generators for SBLOCA. In the absence of such operator actions, automatic control system actions would perform the same functions, but the effect of the poscible difference in timing on the thermal-hydraulic transient is not known at this time.
binq l
l l'
l
'I:
I -
,__,is,
f 4-11 The IRT calculations for MSI3 and RFT assume that the main feedwater continues to supply 1001 flowrate to one steam generator for the duration of the feedwater supply.
Owing to the wide variety of feedwater control feedback and trip functions present, as described in Sect.
4.5.,
achievement of 100% flow rata during this tranisent is not possibl+s;.che actual achievable flow rate and its effect on vessel integrity are not known, but the lower flow rate would result in a lesa severe transient.
Therefore, the IRT and TRAC feedwater flow assumptions in both the IRT and TRAC can be considered to be approximately bounding cases for excessive feedwater supply. Another perspective would be to view the cases as representing two events of different probability. The TRAC case assumes correct operation of the feedwater control system, whether by ICS or the reactor operator. The IRT casa corresponds to an overt operator error (manually supplying full feedwater flow) or a multiple control system failure with lack of corrective operator action. On this basis we consider the IRT transient to be less probable by a factor of 10-3 to 10-6, 4
4.9 Summary The plant control and operations review identifies a variety of potential failures that could possibly result in excessive feedwater supply. Most of these failures have feedback or trips that can be expected to reduce the feedwater in a fairly short period of time. The operator can also take action to terminate main feedwater for all single and double ICS failures identified.
The thermal-hydraulic calculations reviewed employed somewhat arbitrary but approximately bounding feedwater response assumptions. The assumptions used in the Err analyses, in particular, are considered to be conservative, with regard to the severity of the transient.
l More feedwater supply calculations would require modeling the feedwater demand runbacks, limits, and feedwater system trip _ points that were l
described in Sect. 4.5.
Using a model of tbis type, it would be necessary to investigate several of the potentia 1' control system failures to identify the worst credible case. The probability of excessive feedwater is likely to be dominated by the probability of overr operator error, since the control system f ailure requires two independent failures plus lack of operator corrective action.
i l
t
REFERENCES - CHAPTER 4 1.
Oconee-1 Final Safety Analysis Report.
2.
Oconee-1 Instruction Book for Integrated Control and Non-Nuclear Instrumentation Systems.
3.
Sacramento Municipal Utility District Followup Report to Reportable Occurrence 78-1, Re Docket No. 50-312, Operating License DPR-54, dated March 31, 1978.
4.
Oconee Unit 3 Piping Drawings:
PO-101-A, 3 High-Pressure Injection; PO-102-A Low-Pressure Injection; PO-121-A-3, 3-3A, 3-3B, D-3 Condensate, Feedvater, Emergency Feedwater Systens (from Duke Power Company).
5.
Oconee Small Steamline 3reak and Excessive Main Feedwater Event Trees and S4fety Sequence Diagrams (from Duke Power Company).
6.
Transient Response of Babcock & Wilcox - Designed Reactors, NUREG-0667 (May 1980).
7.
Integrated Control System Reliability Analysis, SAW-1564 8.
Duke Power Company, personal communication.
f 9.
" Loss of Non-Class-1-E Instruma.ntation and Control Power Systems Bus During Operation," USNRC IE Sulletin 79-27.
w.e
+
=
---,r
e r
5.0 "HERMAL-HYDRAULICS 5.1 Literature Search for Accumulated Experience Data 3ases. Thirteen dat., bases o 3RRA o
CIM o COMPENDEI o
CONF o EDS o
EIA o FEDEX o
ISMEC o NSA o
RSI t
I o WELDASEARCH were searched for thermal-hydraulic system data relevant to thermal shock tranisents. Capsule descriptions of the data bases are given in
[
Appendix A.
Approximately 600 thermal-shock-related entries were found; however, the majority of the entries were L3LOCA emergency core l
l cooling injection studies and generally contained no system thermal-hydraulic information. There were also a significant number of licensee event reports (LERs); however, they focus on the incident p recurso rs, not the transient data, and so are of limited usefullness j
to this study.
l Oconee Licensee Event Reports. Five LERs of interest, covering four events, were found for the Oconee Nuclear Power Station (see Appendix l
3).
The four events were distributed as one each for Units 1 and 2, and two for Unit 3.
Three events were in the class of steam generator overfeeds. (RFT), and the other was an open power-operated relief valve (53LOCA). None had major consequences, since the operators took timely action.
Specific Documents. Twenty specific documents, as listed in Appendix C, were also reviewed. The first seven of these references were particularly interesting since they contain system thermal-hydraulic data.
5.2 Thermal Shock Plant Transient Data The major source of actual plant data for transient overcooling is the March 20, 1978 Rancho Seco event (Appendix C., Refs.
I and 2).
The pressure and temperature data employed by ORNL as input for f racture mechanics calculations (Ref. 2) are shown in Figs. 5-1 and 5-2, respectively. The pressure surges contained in the original data (Ref.
- 1) were removed for simplicity (they may or may not be 'real").
Owing to the locations of the inlet temperature measuring instruments,
which are placed in wells at the suction side of the reactor coolant pumps and are therefore upstream of the EPIS injection, the use of l
these actual plant data as the RV forcing functions fot a 3&W plant I
introduces some uncertainty.
The temperature of the fluid at the RV inlet nozzle might therefore be expected to be lower than neasured by the instrumentation, unless two phase thermal equilibrium flow exists.
l l
L
e 3-2
-3
-S R
E E
x a
-3oa 2
_g a
c c
u 2
U E
d E
E 5 w
2o E
k f.l.
Ou Q
c.1 0
-3 l
3=
e*
i
-o
=0
- n.
~~ R a
s s
i s
k k
(VISd)1HASS3Md W31SAS
3-3 3
8
_.z=c
-2 x
Un=
=n 8
n wn=
c
~
z=
-w
-3 I9 w-30 E i-eu 3z e
-3 '
-3===
m em
-s s
l
-a O
6 i
e i
i 8
8 8
I 8
8 o
(d SE3WO3GT*dW311NV7CO3
.. ~.
5-4 On the other hand, for transients in which the vent valves open (not likely for the Rancho Seco event, since the reactor coolant pumps are running, but quite likely if the pumps are shut off), mixing in the downcomer could be significant and the fluid' temperature at the RV wall could be higher than the measurement (again, barring two-phase thermal equilibrium flow). Due to the unknown nature of these competing effects, the RV wall temperature forcing function is not easily derived from standard instrumentation in 34W plants during overcooling events.
5.3 overcooling simulations 5.3.1 Results of Analyses To our knowledge, two computer codes, IRT and TRAC, have been used to predict the thermal-hydraulic characteristics of overcooling transients for Oconee-type plants. The pressure and temperature predictions for five scenarios analyzed to date are shown in Figs. 5-3 and 5-4 Note that primary system repressurization is predicted for all cases except case 4, which ::orresponds to MSL3 as simulated with TRAC, and case 5, for which the transient predictions are incomplete.
Repressurization does not occur in case 4 because of an assumption of thermal equilibrium made in modeling the pressurizer. The initial depressurization is similar in all cases, except that IRT does not predict so sudden a primary system contraction as TRAC. The lower worth assumed for the control rods in case 2 is responsibla for the quick return of system pressure, as compared to case L.
Figure 3-4 shows the degree of primary system overcooling to be similar for cases 1 and 3.
The lower rod worth of case 2 is again evident in the race of cooldown. Case 4, TRAC MSL3, is the only one which does not show great overcooling; this difference is attributable to assumed operator termination of sain feedwater flow at 40 s into the transient and the use of emergency feedvater to the intact steam generator.
5.3.2 Limitatio ns All the current simulations possess limitations which give concern for the realism of the thermal-hydraulic predictions. These limitations are, in part, inherent in the codes and also result from modeling deficiencias and questionable input assumptions, as discussed below.
Feed Train. Owing to the multiplicity of heaters and pumps and the various. input conditions and feedbacks present, it is no t. easy to calculate the steam generator secondary-side inlet conditions.
Accordingly, they are supplied by look-up tables in both 1RT and TRAC, and the values entered are the result of simplifying assumptions.
Moreover, since this tabular input is fixed for the duration of the transient, the course of the RFT and the latter stages of the MSL3 cases are almost completely determined by the values entered initially in the look-up tables.
ub,
i 1
00 i
2 1
00 8
5 i
1 4 4 1
6 3 4
0 0
e ms s
a a r r D c c 00 N S S, i 0 1
E pp A
i is G
sT T C
E SO e e L
n L L n8 S i
ih B
bs s L uuS M, S.
s 0
n T,T,M,C C s
o i0 A A i
T T T RRaR R i
3 t
l l l T T 1 2 3 4 6 a
lum e
i ee S
S mo 2
r f
s 0
i0 e
7 l
i for
)C P
E 0
S e
(
e06 E
r I
s M
u T
ser P
0 i0 e
6 t
isop mo 0
C e0 4
3 5
0 i0 g
3 i
F 0
i02 s
0 1
i0 1
1 0
0 0
0 0
0 o
0 0
0 0
0 4
0 6
2 a
2 2
1 4
1
- bs4$Esb a
,t ub s
s lo
~
a t
l 6 e4 o3 n
- o. e.
o i
i t
nt n saa r e cc D
- s. S, N
pp l i A
0 E
e s TT C
i00 s
G ee BO 1
n E
nn t L o
L e kS S S l
mwSb L M. S.
i t
t T,TMAA CC a
TTT i
RRR RH u
- 8
- l. TT o
I lo m
1 2346 s
i S
mor 4
f o
ios se l
i for 0
P 0
7 e
ru t
)C a
2 E
r S
e J
(
o p
io E m
s M e
IT T
e t
l 0
u
~
g0 o
6 p
mo c
o 0
4 4
5
.g iF
~
h s
2 o
f i
6 o
0 0
0 0
0 0
0 0
0 0
0 0
4 6
4 3
2 1
C gESkgg.4f 8 i!!
. 4
3-7 Fluid Mixing.
It is obvious that the degree of sixing between the EPIS and primary-system fluids will have a marked effect on the RV wall temperature in the downcomer region. Further, as shown in Appendix C, Ref. 3, for caaes where the vent valves open, the degree of fluid mixing assumed to occur in the upper downcomer completely determines the severity of these transients.
Neither TRAC nor IRT contains models which are suitable for analyzing these special cases.
IRT has only equilibrium capability for the HPIS and vent valve mixing, whereas TRAC has non-equilibrium capability.
It is our understanding that the cases modeled thus far do not p redict vent valve opening, so some basic code differences may not be manifested in the simulation results.
Control Systems. The performance of the control system was assumed and specified before the cases were run; the assumptions were rather arbitrary and, in most cases, quite conservative. Direct feedback control system modeling for these transients is not currently possible in either TRAC or IRT.
Repressurization. Repressurization is a key phenomenon, both as to its eventual occurrence and the specific time at which it occurs within the t ransient, since this relates to likely operator actions. Only bounding cases have been run thus far, i.e., full equilibrium or noninterchange pressurizer models have been used. Actually, more than the pressurizer is involved here; the upper head and the entire hot leg act as *pressurizers" during these transients and their performance during primary system refill will also affect the race of repressurization.
l t
i Flow Distribution. An ability to calculate flow in the primary system over conditions ranging f rom full forced flow to natural circulation is required to creat these transients properly. How well the codes perform such calculations has not been deter =ined, although to date the IRT calculations have apparently been " driven" by input and have not utilized a momentum equation solution.
Existence of Two Phases. Owing t; void formation at the top of the
" candy cane," loss of natural circulation will oct 2r in B&W plants during many overcooling transients.
IRT cannot treat this phenomenon correctly, which obviously limits the transients to which this code is applicable.
Primarv Metal. The effect of heat transfer to the primary fluid f rom the primary metal has not been fully evaluated. Such heat transfer has been included in some cases (IRT-RFT, TRAC-MSI3 and S3LCCA) and not in others (IRT-MSI.3). The effect could be significant in some transients,
and so should be included in realistic evaluations.
IRT Cases 1, 2, and 3.
These are two cases of turbine trip with open bypass valves (Cases 1 and 2) and a main steam line break (Case 3).
In all cases full main feedwater flow is assu=ed.
This is a very conservative assumption. With one steam generator flooded and the a
L 5-8 intact steam generator isolated, the operator would have to go. to extremes to keep the main stean-drivsa feed pumps running, having lost his primary steam source (the steam generators). The hot well and condensate booster pumps do not have enough head to maintain fun flow (at least in Cases 1 and 2).
In cddition, a multiplicity of ICS failures would have to be assumed to prevent automatic runback and trip of the main feedwater train.
The feedwater temperature was assumed to be a simple ramp down to the hot well temperature, 91*F, over one minute. This is again conse rvative. The several pumps and heatars and the length.of piping win an have considerable capacity to hold the temperature up.
In addition, the high pressure heaters take their steam f rom the main steam lines and not the turbine, so they will not be isolated by the t rid.
The rate at which the cooled primary system fluid is transferred to the pressure vessel is not properly calculated. With the primary pumps tripped and the system depressurized, voiding in the " candy cane" will inhibit natural circulacion until the system is refined by the HPIS.
The repressurization by the HPIS is also overpredicted. The non-i interchange models u ed for the upper head and pressurizer result in a eteam compression calculation producing the pressure. Since the steam and water are assumed not to interact, the steam " bubble" is compressed at a volume reduction rate equal to the HPIS charging capacity.
TRAC Case 4 This is a main steam line break and is more realistic in its assumptions of feed train performance. The operator is assumed to start auxiliary feedwater at 30 s and terminate main feedwater at 40 s.
The ICS would have performed the same functions in the same time f rame had the operator net acted. The auxiliary feedwa',ar comes straight f rom the hot wen, so its temperature is more easily determined.
The adequacy of TRAC to tr2at " candy cane" voiding and consequent loss of natural circulation and, therefore, the transport of cooled fluid to the pressure vessel before repressurization by the HPIS is not known.
However, the repressurization race is known to be too slow. Full equilibrium is assumed in the node above the water level in the p ressuriza r.
This results in concensation of the steam. The ref o re, repressurization win not occur until au steam has been condensed and the pressurizer is fun; this is unrealistic.
37 comparison, case 4 is more realistic and also much silder than-the IRT-MSL3, Case 3, as can be seen in Figures 5-3 and 5-4 Under the limitations noted, these two. cases could be considered bounding analyses.
5.4 "3est Judgment
- Pressure anu femperature Forcing Functions s
Owing to the above deficiencies, the available simulated forcing functions must be regarded as approximations to the true functions; the magnitude and sign of the error is not presently known. Whateve r their deficiencies, the Rar.cho Seco data represent a recorded event, not a simulation, and so provide the closest realistic vessel forcing 4
functions currently available.
.~ _. _ _ -. _ _ _,..., _
5-9 References, 3y Legend Number, for Firs. 5-3 and 5-4 1.
" Runaway Feedvater Af ter Turbine Trip Report," letter, M. Levine to N. Zuber dated July 2, 1980.
2.
Ibid.
3.
" Analysis of a Steam Line 3 reak with Primary System Overcooling for a Typical 3&W Reactor," letter, R. Carbone to R. Kryter dated August 14, 1981.
4
" Completion of Scheduled Analysis on Pressurized Thermal Shock Scenarios," letter, S. Fabic to C. Serpan dated June 22,198..
5.
Ibid.
6.
" Parametric Analysis of Rancho Seco Overcoo11ag Accident," letter, R. Cheverton to M. Vagins dated March 3,1981.
i O
=
6.0 ESTIMATION OF NEUTRON FLUENCE AT THE REACTOR VESSEL WALL 6.1 Intr'oduction A realistic evaluation of a postulated " pressurized thermal shock" accident requires a knowledge of the neutron fluence (E > 1.0 MeV) and its uncertainty throughout the reactor vessel wall. The fluence values must be known at the location of those materials (welds or plates) which are most likely to be affected by the conditions attained during the transient.
Currently the Code of Federal Regulations (10CFR50, Appendices G and H) and Regulatory Guide 1.99 (Ref.1) require licensees to provide estimaces of the neutron fluence in the reactor vessel. beltline region as a part of their surveillance programs. The methodologies adopted by different vendors and service laboratories to obtain the fluences can be separated into three parts:
a.
Neutron transport calculations b.
Dosimetry measurements c.
Analysis of uncertainties The techniques for accomplishing parts "a" and "b" do not vary significantly from vendor te vendor.
However, the uncertainty analysis involves combining differential and integral dosimetry data, both measured and calculated, in a consistent fashion so as to obtain absolute fluence va?.ues (and their uncertainties) in the RV wall as a function of energy. These fluences must be extracted from an analysis of measurements performed at the location of a surveillance capsule, which may be distant from the RV vall. Although considerable work has 1-5 been expended in developing methodologies to achieve part "c", the application to power reactors has just begun.
A preliminary list of uncertainties affecting the calculation and measurement of neutron flux and fluences in LWRs was compiled by the ASTM E10.05.01 Task Group cn Uncertainty Analysis. This list, with a f ew additions, is given in Table 6.1.
6.2 Babcock and Wilcox Methodology 6.2.1 Neutron Transport Calculations The calculated energy group fluxes are determined using a discrete ordinates solution of the Bolt =mann transport equation. The codes used are ANISN6 and DOT.7 Table 6.2 below gives the steps *in the B&W transport calculational procedure.
6-2 Table 6.1.
Uncertainties for Calculation and Dosimetry Measurement Procedurec in LWRs Source of Uncertainty Estimated Uncer-tainty (%)
I.
Calculational Procedure,b a
A.
Source Term
- 1. Fuel management - uncertainty in the fuel cycle 10
- 2. Fuel position
<5
- 3. Burnable poison 10
- 4. Core power distribution (cycle and cycle-to-cycle variation) 30
- 5. Power / time history (cycle and cycle-to-cycle variation) 10
- 6. Iocal power at core periphery vs. total power a.
axial 20 b.
radial and azimuthal 20
- 7. Control rod position 5
- 3. Transport c
1.
Flux synthesis (reduction of 3-D to 1-D and 2-D calculations) c 2.
Energy group structure c
3.
Quadrature (S and anisotropic scattering P )
c 3
g 4.
Cross sections c
5.
Spatial ses'i e
6.
Geomet ry c
7.
Time-averaging vs. changing core condition e
8.
Extrapolation (lead factor) e h
II. Dosimetry Measurement Procedures e
A.
Nuclear data (reaction rate cross sections, branching rations, etc.)
c 3.
Coopecing reactions e
C.
Photofission corrections c
D.
Flux / time history c
E, Counting calibration c
87alues listed compiled by ASti E/0.05.01 Task Group od Uncertainty Analysis.
b3AW-1485 discusses several of the sources of uncertainty, but does not specify the effect on fluence escisates at the RV welds, t
cCurrently unavailable.
6-3 Table 6.2.
S&W Method for Cbtaining Neutron Fluences a.
Obtain a pin-to-pin, time-averaged power distribution b.
Obtain P3 and Pt 22 group CASK cross section sets for 1-D and 2-D calculations, respectively c.
Perform a 1-D, P, S 3
6 discrete ordinates transport calculation d.
Perform a 1-D, P, S6 discrete ordinates transport calculation g
e.
Obtain a ? Pg correction factor from the 1-D calculations to 3f apply to a 2-D calculation f.
Perform a 2-D, x-y calculation with the surveillance capsule g.
Perform a 2-D, x-y calculation without the surveillance capsula h.
Obtain a capsule perturbation factor from the 2-D calculations to correct the measured activity or calculated fluxes 1.
Perform a 2-D, P, r-6 calculation g
j.
Perform an axial 2-D, P, r-z calculation g
k.
Obtain synthesized 3-D group fluxes in the reactor vessel 1.
Correct estimates of the group fluxes, based on the P /Pg and 3
capsule perturbation factors 6.2.2 B&W Surveillance Dosimetrv Measurements The surveillance program for Oconee-1 comprises eight surveillance capsules designed to monitor the effects of neutron and thermal environment on the materials of the reactor pressure vessel core region. The capsules, which were inserted before initial plant startup, are positioned between the thermal shield and the RV wall.
Six of the capsules, placed two in each holder tube, are positioned near the expected peak axial and azimuchal neutron flux. The remaining two capsules (designed to monitor thermal aging) are placed in an area of essentially zero neutron flux.
Capsule OCl-F was removed during the first refueling shutdown of Unit 1, and capsule OCl-E was removed during the second refueling shutdown.
Four activation detectors with reaction thresholds in the energy range of interest were placed in each surveillance capsule. The properties of interest for the detectors are given in Table 6.3, and the results of the averaged measured reaction rates are given in column 3 of Table 6.4 l
Table 6.3.
Surveillance Capsule Detector Data Detector Threshold Energy Product Half-life (MeV) 59Co(n,Y)60Co Thermal 5.26 y 237Np(n,f)l37Cs 0.5 30 y 23ag(3,f)137Cs 1.5 30 y 5"Fe(n,p)SS S4Mn 2.0 313 d Sani (n,p) Co 2.5 71.3 d
6-4 Table 6.4.
Comparison of Calculated to Experimental Reaction Rates for Oconee-1 Capsule Reaction Experiment, E Calculated, C Ratio,d E/C (uci/g)C (uci/g) a CCl-F 547,(a,p)s4Mn 17.6 2 0.95 21.0 0.84 Sani (n.p)5aco 495.0 2 2 422.0 1.17 25ag(a,g)137Cs 1.36 2 0.21 0.58 2.34 237Np(n,f)l37Cs 7.86 2 0.10 2.90 2.70 b
OCl-E 5"Fe(n.p)S"Mn 536 2 62 729.3 0.74 sagg(3,p)5aco 975 2 163 1,266.0 0.77 238g(a,g)137Cs 1.94 2 0.18 1.719 1.13 237Np(n,f)l37Cs 9.32 2 1.18 8.799 1.06 aSAW-1421 l
C54Mn and 58Co values are given in units of per gram of target for CCl-E and per gram of dosi eter for CCI-F.
i
(
dNormalization constant
~
I e
e o
6-5 6.2.3 Determination of the Neutron Fluence (E)1 MeV) at the Reactor _
Vessel The energy-dependent neutron flux is not directly available frem activation i
detectors becar.se they provide only the integrated flux on the target 3
sacerial as a function of both irradiation time and neutron energy. To obtain an accurate estimate of the average neutron flux incident upon the detector, the following parameters must be known:
(1) the operating history of the reactor, (2) the energy response of the given detector, and (3) the neutron energy-group fluxes at the detector location. Two means are available to obtain the desired spectrum:
iterative unfolding of experimtseal foil data and neutron transport methods. Due to a, lack of sufficient threshold foil detectors satisfying both the threshold energy and half-life requirements necessary for a surveillance program, the neutron energy spectrum was obtained by the transport mechsd (Sec 6.2.1), instead of by spectrum unfolding.
The calculated spectrum is used in the following equations to obtain the calculated activities used for comparison with the experimental values. The basic equation for the activity, D (in uCi/g) is:
-A (T-tj)
M
-A t
- i t3 g
I h
ft[ag(E)$(E)[Fj (1-e
)
(6.1) le D
=
i i
3,7 x to 3.t s
where normalizing constant C
=
Avogadro's number N
=
atomic weight of target material i A
=
g th F
.=
either weight fraction of target isotope in the i g
material or fission yield of desired isotope group-averaged cross sections for material i e (E)
=
g group-averaged fluxes calculated by DOT analysis
$(I)
=
E fraction of full power during j time interval, e F
=
3 d
th decay constant of the i material A
=
interval of power history t
=
j sum of total irradiation time, i.e., residual time in reactor T
=
and wait time between reactor shutdowh and counting cumulative time from reactor startup to end of j"h th t
=
j period, i.e, T
=
e j
k k=1
6-6 The normalizing constant, C, can be 'obtained by equating the right hand side of Eq. (6.1) to the measured activity. With C specified, the neutron fluence > 1 MeV can be calculated from 15 MeV
$(E > 1.0 MeV) = C
$(E)
Ft (6.2)
E1 j1 3d, where M is the number of irradiation time intervals.
The last column of Table 6.4 (Ratio E/C) shows the spread of the normalizing constant as a function of the threshold reaction used in the measurements.
3AW-1436 states that the 238U and 237Np reactions from the OCl-F capsule have been deleted in all current evaluations on a basis of inconsistency.
6.3 Results and Uncertainty Analysis The estimated fluences (E > 1 MeV) at the axial welds (Table 6.5) were determined from Tables 6.6 and 8.1 of B AW-1436.
The procedures used in obtaining these estimates are given in BAW-1485 (proprietary). The estimated uncertainties in the surveillance capsula analysis are also provided in Sect.
4 (Tables 4-1 and 4-2) and Appendix F of B AW-1485 (proprietary). Fluences at-the vessel wall locations may be as high as 250%.s The procedure outlined in 3AW-1485 identified many of the sources of uncertainty stated in Table 6.1 but did not specify all their values.
6.4 Conclusions and Recommendations The methodology used by B&W is similar to that used by other vendors and service lab' oratories. One weakness in the methodology is the analysis of uncertainties. This analysis should provide not only estimates for the sources of uncertainty identified in Table 6.1, but a statistically sound technique for combining all the individual estimates to arrive at an overall uncertainty for the fluences at the RV wall. This task requires considerable effort and funding on the part of the vendors, and only recently has work been done in this area.1-5 Another weakness relates to surveillance dosimetry measurements and the subsequent analysis to obtain fluxes > 1 MeV or other suitable neutron exposure parameters. To address this problem,
~ ~ ~.. - -
Table 6.5.
Predicted Fast Fluence in Oconee-! RV at the Axial Wolds for 8 EFPYa nettline I.ocation Ma t e rial Su rvei llance Inside of RV t/4 3T/4 Outside of locations Wall RV Wall tipper long. weld SA-1073 4.95Etl8 2.93Etl8 1.63Etl8 3.70Etl7 1.'s9Etl7 til alille long. weld SA-1493 4.83Ettb 2.86Etl8 1.59Etl8 3.65Etl7 1.36Etl7 1.ower long. weld SA-1430 6.42Etl8 3.80E+18 2.llEtl8 4.79E+17 1.8tEtl7 Peak flux location 7.30Etl8 4.32Etl8 2.40Et!8 5.45Etl7 2.07Etl7 i
v an AW-1436, Tables 6-6 and 8-1
6-8 a pressure vendel benchmark facility for power reactor surv1111ance dosimetry validation and certification is.needed to help identify and reduce uncertainties.
It is thought that, with care, an overall uncertainty of *10-30% for the fluences at the vessel vall should be achievable.
REFERENCES - CHAPTER 6 1.
W. N. McElroy, Ed., LWR Pressure Vessel Surveillance Dosimetry I:provement Program: PCA Exoeriments and Blind Test, NUREG/CR-1361 (1981).
2.
W. N. McElroy, F. 3. K. Kam, E. D. McCarry, A. Fabry, LWR Pressure Vessel Surveillance Desimetrv Improvement Program, NUREG/CR-1747 (1980).
3.
R. E. Maerker, J. J. Wagschal, 3. L. 3roadhead, Development and Demonstration of an Advanced Methodology for LWR Dosimetry Applications, (EPRI report to be published in 1981).
4 J. J. Wagschal, R. E. Maerker, Y. Yeivin, " Extrapolation of Surveillance Dosimetry Information to Predict Pressure. Vessel Fluences," pp. 631-632 in Trans. Amer. Nucl. Soc., 1980 Annual Meeting, Vol. 34.
5.
W. N. McElroy, et al., Surveillance Dosimetry of coerating Power Plants, Hanford Engineering Development laboratory, HEDL-SA-25e6 (1981).
6.
W. W. Engle, Jr., A User's Manual for ANISN, A One-Dimensional Discrete ordinates Transport Code with Anisotronic Scattering, Oak Ridge Gaseous Diffusion Plant, USAEC Report K-1693 (1967).
7.
W. A. Rhoades, D. B. Simpson, R. L. Childs, and W. W. Engle, ~he DOT-IV Two-Dimensional Discrete Ordinates Transoort Code with Scace Deoendent Mesh and Quadrature, Oak Ridge National Laboratory, ORNL/TM-6529 (1979).
8.
W. N. McElroE et al., Surveillance Dosimetr*r of Ooerating Power Plants, REDL-SA-2546 (October 1981).
O e
s=m-e M
-*y,g
-g n
&--,--..-,-w- - - - - - -,. - - -
we-e-m y-w
7.0 PRESSURE VESSEL MATERIAL PROPERTIES 7.1 Material Properties Required for Vessel Integrity Studies The material properties required for' studying vessel integrity can be grouped in accordance with the three types of analyses that must be performed (Table 7.1).
l Table 7.1 Reactor Vessel Material Properties Needed for Vessel Integrity Studies Thermal Analvsis Thermal conductivity (k)
Specific heat (c )
p Density (p)
Stress Analysis I
Linear thermal coefficient of expansion (a)
Modulus of elasticity (E)
Poisson's ratio (v)
Yield and ultimate strengths (a, a )
y u
{
Fracture-Mechanics Analysis Static crack-initiation toughness (7 )
Static crack-arrest toughness (Kr,)
Reference nil ductility temperature (RTNDT)
I j
- 7. 2 Dependence of Mat 6 rial Properties on Chemical Composition, Temperature, and Fast-Neutron Fluence f
Generally speaking, all of the sacerf al properties in Ttble 7.1 are functions of material chemical compcsition, temperature, and fast-neutron fluence and must be treated accordingly in carrying out the vessel integrity stwiies.
7.2.1 Material Chemical Comoosition and Heat Treatment With the exception of a few of the earliest reactor pressure vessels, all belt-line regions of U.S. commercial reactor vessels presently in service were fabricated from the three sacerials described in Table 7.2.
TVo additional materials that zust be considered are the weld j
filler material (used to join sections of rolled place and forging rings) and the vessel cladding (applied by depcsiting weld setal). The weld filler material is similar to the base sacerial, whereas the cladding is an austenitic stainless steel (18 Cr-8 N1).
7-2 A chemical element of special interest in both the vessel base material and associated welds is copper, since it enhances radiation damage, which resu~.cs in reduced fracture toughness. High concentrations of copper exist in some welds because the weld wire was placed with copper; this element is also present as an impurity in the base material.
The chemical composition of the vessel materials influences all the parameters in Table 7.1, while the various vessel heat treatments (tempering and stress relieving) affect primarily e, e, and y
u RTNDT.
7.2.2 Temperature Dependence of Properties The temperature dependences of the parameters in Table 7.1 are reasonably well known, and each parameter (with the exception of RTNDT and v, which has a negligible temperature dependence) or an appropriate combination thereof, is included in the ASME Pressure Vessel Code as a function of tempersture. Values for k, k/pe, a, E, c, and e p
y u
i Table 7.2 Materials Used in the Fabrication of U.S.
Commercial Reactor Vessels Wgt. Percent Composition for Designated Materials Place SA 302 Place SA 533 Forging SA 508 Element GR 3 GR 3, C1 1 C1 1 C
0.25 max 0.25 max
- 0.27 max Cr 0.25-0.45 Ni 0.40-0.70 0.50-1.00 Mo 0.45-0.60 0.45-0,60 0.55-0.70 Mn 1.15-1.50 1.15-1.50 0.50-0.90 Si 0.15-0.30 0.15-0.30
,0.15-0.35
?
0.035 sax 0.035 sax 0.025 =ax 5
0.040 max 0.040 sax 0.025 sax Fe balance balance balance
7-3 t
as functions of temperature are inc1'uded in ASME Section III, while Kre and Kr, as functions of' T - RTNDT. are included in ASME Section XI for temperatures (T) less than that associated with t~
upper shelf.2 The uncertainty in the parameters included in ASML t
Section III is not large and is expected to have a negligible effect on
- the evaluation of vessel integrity. However, the uncertainties in and K, vs T - RTNDT is substantial, and there is also a Kre g
significant uncertainty (approximately 220*F) in the determination of RTNDT.
The curves for Kre and Kr, vs T - RTNDT in ASME Section II represent the lower bound of a limited amount of data that were obtained some years ago for A533 and A50S material. The use of this lower bound would appear to be conservative; however, recent 3
experiments with large test cylinders
- 4 indicate that long cracks in large structures will usually behave in accordance with the lower bound of data obtained f rom a large number of small (IT to 3T-CT) specimens.
The uncertainty in the ASME Code lower bound is under investigation at i
this time.
The uncertainty associated with the use of RTNDT as a normalization factor is also under investigation. An alternative to relying on RTNDT is to determine, through laboratory testing, Kre and Kr, vs T for each vessel. However, this approach appears to be impractical because of the large number of specimens required, since RTNDT 'is a function of fluence.
l During a reactor vessel thermal transient of medium duration, the outer portion of the vessel wall remains at temperatures corresponding to ductile behaviour (i.e., the up;,er-shelf portion of the Kre vs T i
curve). It is not likely that cleavage (brittle) f racture can proceed through this zone; however, the crack may tear at relatively low crack-i tip velocities to a depth at whic.h plastic instability is achieved.
Tearing-res'istence material property data are required for an accurate l
analysis, and such data are, at present, very limited. An alternative approach currently used is to assume what appears to be a conservative upper-shelf toughness that is essentially independent of temperature and fluence; this upper shelf value is then compared to the calculated stress intensity factor. For very severe accidents such an approach is probably adequate, but for less severe cases the tearing resistance may terminate crack propagation. The degree of conservatism in the p resent model for the less severe cases is ' not. known.
7.2.3 Decondence of Material Properties on Fast-Neutron Fluence Of the material properties listed in Table 7.1, those that have a significant dependence on fast-neutron fluence are c[o, a K
us Ic' Kr,, and RTNDT. Radiation damage increases a and, a lesser 7
o, while Kre and Kr, are decreased and RTNDT is
- extent, u
l increased. The average of c and a is used at the conclusion of 7
n l
the f racture-mechanics analysis-to see if the uncracked ligament has l
become plastic under pressure loading. Usually, strength values for
7-4 the unieradiated material are used, and this approach introduces some conse rvatism. Scae data on elevated strength are available, and their use in the analysis would result in a higher permissible pressure during a thermal transient.
In accordance with ASME code procedure, the decreases in Kre and I, due to radiation daange are estimated by shifting the Kre t
and Kr, vs T curves along the temperatur. axis by an amount ARTNDT = f(F, Cu, P), where F = fast-neutron fluence (E > 1 MeV) and Cu and ? are the copper and phosphorous concentrations, respectively.
Values for ARTNDT = f(F, Cu, P) are included in Reg. Guide 1.99, Rev.
1,5 which was thought to be conservative at the time of its issuance.
A more recent evaluations of the available data indicates that the AR~NDT for materials containing a high concentration of nickel, which appears to enhance the effect of copper on radiation damage, agrees rather ven with Reg. Guide 1.99, while lower concentrations of nickel result in conservative values for the reference temperature shift.
There are indications that many of the velds in the older ?%A reactor vessels have high concentrations of nickel, and thus estimates of -
ARTNDT from Reg. Guide 1.99 presumably are not excessively cons ervative. However, the data base is ::nch smauer than would be desired and win take some time to increase substantiany, even though su:ve111tsce spacimens from power reactors are becoming available and irradiation programs at materials testing reactors are under way.
To date, radiation damage to the cladding is of little concern to the analysis of overcooling accidents because it has been assumed that the initial flaw win extend through the vessel cladding into the base material, and also that the flaw win be very long on the surface so that cisdding resistance to crack extension is not important. However, an assumption of long initial flaws may be unnecessarily conse:vative, since the presence of cladding may prevent short flaws f rom extending, particularly if the cladding retains its high tearing resistance at high fluenc'es. ~he:e is a limited amount of data 7 for veld cladding that indicates a substantial reduction in Charpy upper-shelf energy
(~100 to 30 ft-lb) at a fluence of ~8 x 1013 n/cd and an irradiation temperature of 550*7.a Thus, it is not clear that the cladding will p revent short flaws from growing long.
REFERENCES - CHAPTER 7 1.
ASME Soiler and Pressure 7essel Code,Section II.
2.
T. U. Marston, Ed., ~71aw Evaluation Procedures,* ASME Section XI, Electric Power Research Institute, EPRI NP-719-SR (August 1978).
3.
R. D. Cheverton and S. K. Iskander, RSST Program Quarteriv Progress Recort for October-December 1980, NUREG/CR-194L, pp. 37-50 (March 1981).
m
-r g-wes-y p--.
w+----
7-5 References (Continued) 4 R. D. Cheverton et al., ESST Program Ouarteriv Progress Recort for Januarv - March 1980, NUREG/CR-1477, pp. 16-28 (July 1980).
5.
U.S. Nuclear Regulatory Commission, " Effects of Residual Elements on Predicted Damage to Reactor Vessel Materials," Regulatory Guide 1.99, Rev. 1 (Sept. 1976).
6.
P. N. Randall, U.S. Nuclear Regulatory Conanission, personal communication (Sept. 1981).
7.
F. J. Loss, Naval Research Laboratory, personal communication (July 1981).
8.
It is of interest to note that because of a recent change in core loading, the anticipated fluence by the end of 32 EFPY for Oconee-1 is ~1.1 x 1019 neutrons /cm2 (personal communication, Duke Power Company, October 1981).
l I
o O
L.
1 INTEGRITY OF REACTOR VESSELS DURING OVERC00 LING AC'!IDENTS i
8.1 Description of Basic Problems l
During an overcooling transient in a PWR the reactor pressure Vessel is subjected to a thermal shock in the sense that thermal stresses are created in the vessel wall as a result of rapid removal of heat f rom its inner surface. The charmal stresses are superimposed on the pressure stresses, with a result that tr.e not stresses are positive (tensile) at and near the inner surface of the wall and are substantially lower and perhaps negative elsewhere, depending on the magnitude of the pressure stress. The concern over the high tensile stresses near the inner surface is that they result in high stress intensity factors (K ) for any inner-surface flaws which may be g
p resent. To compound the.atter, the reduced temperature aoJ the relatively high fast-neutron fluence near the inner surface result in relatively low fracture toughness values (Krc and T-h) for the vessel material in the same area. Thus, there is a possibility of crack p ropagation. The positive gradient in temperature, combined with the negative gradients in stress and fluence through the wall, tends to provide a mechanism for crack arrest deeper in the wall. However, if the crack is very long on the surface and propagates deep enough, the remaining vessel ligament will become plastic and the vessel internal pressure will ultimately result in rupture of the vessel. Thus, fo r each thermal transient there will be a maximum permissible pressure that is a function of time.
Crack propagation may also be limited by a phenomenon referred to as warm prestressing (WPS), which has been demonstrated in the laboratory l
i with small specimens and also in a rather large, chick-walled cylinder during a thermal shock experiment.2 In such cases, WPS simply refers to the inability of a crack to initiate while Ky is decreasing with
- time, i.e., while the crack is closing. While this special situation j
is encountered during some specific overcooling accidents, caution must be exercised in taking credit for WPS because changes in the pressure that affect little else can delay or eliminate the requisite conditions l
'or WPS.
l l
The integrity of a reactor vessel during a postulated overcooling event is evaluated in terms of the continued ability of the vessel to contain the coolant. in such a way that melting of the reactor fuel will not occur. Generally speaking, this means that the water level must be maintained above the core, and to do this there must not be a significant breach in the vessel wall below the level established by the top of the core. Therefore, it is necessary to determine if a preexistent flaw will propagate through the wall, and'if it will, to estimate the probable site of the breach and its resistance to leakage.
f
4 8-2 The investigative effort thus far has been directed at understanding the bebsvior of flaws during thermal, pressure, and thermal plus-pressure loading conditions. It has been assue d on the basis of limited available data that if the temperature of a major portion of the coolant in the primary system is well above 212*7 at the time a long flaw penetrates the wall, the final opening may be excessive in the sense that flooding of the core could not be maintained. Methods for estimating the size of the opening more accurately will be evaluated in the near future.
In the following paragraphs a calculational model for predicting crack behavior during overcooling accidents is described, and a summary of results for specific accidants is presented and discussed. 2e reactor plant analyzed for these studies is Oconee-1, and the postulated accidents include a main steam line break, a turbine trip followed by stuck-open bypass valves, a small-break LOCA, a Rancho Seco-type t ransient, and a large-break LOCA.
8.2 Calculational Model The calculational model consists of three basic parts: a thermal analysis of the vessel wall, a stress analysis, and a f racture-mechanics analysis. The 'thennal analysis is performed for cylindrical geometry, is one-dimensional (radial direction), includes an insulated outer surface, and accepts a transient coolant temperature at the wall's inner surface. A time-independent inner-surface thermal resistance is used that is the sum of the fluid-film resistance and the cladding resistance, in which case the heat capacity of the cladding is igno red.
Since an inportant input to the thermal analysis is the temperature of the coolant in the downcomer region, some of the assumptions used in obtaining this temperature need to be mentioned. Depending on the nature of the overcooling accident, the temperatures of the coolant entering the downcomer may be diffarent for the diff arent inlet coolant pipes. Thus, there can be azimuchal and axial variations in the downcomer coolant tempe rature. An accurata determination of the temperature distribution as a function of time would be very difficult, and the subsequent use of two-or three-dimensional stress and f racture-mechanics computations would be impractical for a parametric-type analysis. An additional complication in this regard for 3&W reactors is that relatively warm water f rom the core outlet may enter the upper portion of the downcomer region through the, vent valves and may thus raise the temperature of the downcomer coolant. For the purpose of the present analysis the coolant temperature (temperature transient) used as input to the vessel thermal analysis corresponds, with one exception, to the lowest of the cold legs (inlet lines). The degree of conservatism associated with this assumption is unknown.
8-3 1
The stress analysis is also one-dimensional and is performed for a cylinder; the cladding is excluded, as is the flaw, consistent with the method used for criculating the stress intensity factor. Loads on the cylinder consist of a radial temperature distribution and internal pressure, both of which are treated as functions of time.
Linear elastic f racture mechanics (LEFM) is used for predicting flaw behavio r.
The flaw assumed for this particular study is an inner-surface, long, axial 17 uriented, sharp crack of uniform depth along its length. Thus, an accurate, two-dimensional (radial and azimuthal) model can be used. Consideration of a long axial surface flaw is realistic and necessary in the sense that sho rt flaws will tend to become long flaws under thermal-shock loading, and the stress intensity factor is greater for long axial surface flaws chan for any others.
The two-dimensional model does, however, introduce some conservatism since there exists an axial gradient in fluence (and hence in toughness) that is ignored but which will provide some additional resistance to crack propagation. Furthe rmo re, the cladding, which tends to be a much tougher material than the base material, may suppress the surface extension of short flaws that has been predicted and observed in the absence of cladding.
As already mentioned, although the cladding is included in the thermal analysis it is not included in the f racture-mechanics analysis. The presence of cladding reduces slightly the tensile stress in the base material during a thermal transient. Howsver, if the flaw extends through the cladding, the K value is significantly greater than if the cladding were ignored, partfeularly for shallow flaws. Thus, the minimun critical crack depth for crack initiation would be less and the threshold fluence for crack initiation and vessel failure would also be less. A detailed quantitative assessment of this cladding effect is not yet available.
F racture-co'ughness curves (I and K vs T - RTNDT) for this study XIoff$eASMEdode,3 and an uppe r-shelf were taken f rom Sect.
toughness of 200 ksiVTn? was added for both I and K,.
t Input to the f racture-mechanics analysis inclu!es (1)'$he temperature and fast-neutron fluence distributions thruugh the wall, (2) che thermal and pressure stresses without the presence of the flaw, and (3) the copper (Cu) and phosphorous (?) concentrations.
The temperature and fluence distributions, coupled with the Cu and P concentrations, are used to calculate K and K, radial distributions at various times re r
it the transient, and the stresses are used to calculate K values for r
a number of crack depths, ranging f rom ~3 to 90% of the wall thickness.
S-4 The thermal, stress, and f racture-mechanics analyses were performed with the OCA-I computer code," which uses a superposition technique to accurately calculate Kg values for long flaws using stresses for the unflawed cylinder. The purpose in using the superposition tee'anicue was to achieve the accuracy of a finite-element analysis at a f raction of the usual cost, thus making parametric studies feasible. A block diagram describing the code is shown in Fig. 8.1.
Required input for the OCA-I analysis is also indicated in Fig. 8.1, and specific values used for the Oconee-L studies included herein are given in Table 8.1.
For these preliminary studies only five transients were analyzed, but for each transient several inner-surface-fluence values were included so that the threshold fluence (and thus the number of years of operation) for incipient crack initiation could be estimated.
8.3 Transients Considered for oconee-1 8.3.1 Main Steam Line 3 reak Ti:ae-dependent pressure and temperature curves for this case were submitted to ORNL by 3rookhaven National Laboratory (3NL) on August 14, 19815 and art given in Fig. 8.2 and Table 8.2 (corresponds to case 3 in Section 5).
For the thermal analysis of the vessel the fluid-film heat-transfer coefficient was assumed to be 1000 Stu/hr f t2..y, which corresponds to full-flow conditions (primary system) and a total surface conductance of 330 Stu/hr*f t2..y, i
3.3.2 Rancho Seco Transient
in January 1981 and is shown graphically in Fig. 8.3 and in tabular form in Table 8.3.
As described in Sect. 5, the coolant reaperatura is sensured upstream of the injection point for the HPIS and is thus probably somewhat higher than actually exists at the entrance to the downcome r.
The fluid-film heat-cransfer coefficient at the vessel wall was assumed to be 1000 Stu/hr*f t2.*y, 8.3.3 Turbine Trip Followed by Stuck-Ocen 3ypass Valves This transient, which corresponds to case. in Section 5, was submitted 7
to NRC by 3NL in July 1980 and to ORNL by 3NL in anust 1980a and is shown graphically in Fig. 8.4 and in more detail in Lble 8.4 The portion of the transient beyond 1240 s was added by ORNL, assuming that the HPIS would purap against the relief valve setting (2500 psi) and that the temperature of the coolant in the downcomer would remain at 140*F.
The fluid-film heat-transfer coefficient was again assumed to be 1000 Stu/hr f t2.*F.
m g-ww
-.A-.--w.--s.
- -, _, + -, -,,.
.e,.
y w
,.~--
~
ORNL-DWG SI-1663R EID k
CYL HTNDTo F = f(a/wl h
p p
e,, K
- ID l
ARTNDT o
" II I T,
c>
OD Cu,P
- f(F, Cu, P) s I
o o
o e
e COOLANT TEMP TilEllMAL ANAL.YSIS g**g' 8
8 WALL TEMP VS y
3 VS TIME TIME TIME AND a/w K /K K /K,
a,/w g
g q,
VS
=
VS d.
PilESSURE STitESS ANALYSIS K,
TIME AND a/w TIME VS O
TilEGMAL AND VS TIME PitESSullE STflESSES TIME AND a/w IlEQUlftED INPUT J L 1 6 J L TYPICAL DATA INCLUDED, OPTIONAL INPUT a
CYL E
3D K* = f(a** al TYPICAL DATA INCLUDED, AND NO OPTION i
] CALCULATION AND OUTPUT Fig. 8-1.
tilock Diagram of the OCA-I Computer Code, Ind icating flauf c Input.
Calculations, and Ots t ptit.
- 9
6 8-6 Table 8.1.
Input to OCA-1 for Oconee-1 Analysis Vessel Dimensions, in.
Outside Dia.
189 Inside Dia.
172 Coolant Temp vs Time Specific Accident Pressure vs Time Specific Accident 2
Heat Transfer Coeff. (h), Stu/hr*f t..y Large-break LOCA 200 Others 330 Initial Wall Temp, 'F 550 RTNDg, r 40 Copper Concentration (Cu), ;
0.31 Phosphorous Concentration (P), ;
0.012 K;e and K, Upper Shelf, 7
'ui 'i in.~
200 Fluence at Inner Surface (F )
Range of Values o
3RTNDT,= zero-fluence RTNDT (initial value) i e
i e
.en n
n g-
,-ma q
p-,w
-ev->
e 3-7 OMNL-OWG S1-16490 ETO I
i i
i i
soo.
2sco 4co 2000
- L w
t.=
3 TEMPERATURE zoo.
1500 =
g s.2
=.
a a
\\
PR ESSUR E 2co toco too -
500 I
I I
I l
o a
o 200 4c0 soo sco 1c00 1200 TIME (il
~
~
Main Steam line break te=peracI5e and pressure Fig. 8-2.
~I
- _cransients, (EPIS reminn active),.
8-8 Table 8.2.
Main Steam Line Break Temperature and Pressure Transients (HPIS Remains Activated)
PZR Pri:na ry Time Level Pressure (s)
(ft)
(psia) 0 18.23 2157.8 1
17.44 2139.1 5
11.01 1964.35 10 2.10 1677.02 15 0.0 1440.25 20 0.D 1323.79 25 0.0 1216.75 30 0.0 1106.18 35 0.0 904.86 40 0.0 760.18 45 0.0 743.5 50 0.0 725.19 60 0.0 689.04 70 0.0 650.19 80 0.0 610.61 90 0.0 578.52 100 0.0 559.57 120 0.0 530.47 140 0.0 513.61 160 0.0 523.63 180 0.0 517.91 200 0.0 507.52 220 0.0 497.72 240 0.0 487.96 260 0.0 478.35 l
1
'O 0.0 468.92
.0 0.0 459.61 J50 0.0 438.92-400 0.22 429.69 450 1.40 443.36 500 2.69 459.39 550 A.12 478.53 600 5.67 501.16 650 7.30 527.56 700 9.04 558.72 750 10.84 595.07 800 12.69 637.46 850 14.56 686.83 900 16.43 747.1 950 18.30 818.56 1000 20.14 903.93 1050 21.96 1006.48 1103.58 23.84 1139.87 1123.18 24.51 1195.84
e e
8-9 CANL-CwG 41 4080 ETO
- so e
4 i
i L
=
m 8
16.000 E
=
y 12.oco,
r, e
m h
8" 3
o smo g too i
50 4 coo o
o 20 40 60 80 100 TIME (ment l
Fig. 8.3.
Temperature and pressure transients for Rancho Seco.
l 8
1t-
8-10 Table 8.3.
2ancho Seco Temperature and Pressure Transients Time Temperature Pressure (min)
('F)
(psi) 0 590 1500 10 490 1710 20 412 1880 30 356 2020 40 318 2110 50 296 2130 60 282 2100 70 280 2050 80 284 2000 90 299 1950 100 320 1900 i
I i
i
(
l
e e
3-11 CANLC.M 81-1M91 ETD l
l l
l l
l l
Sc0 25c0
- 400 2000 o_
TEM 9ERATURE l
~x
~
n W
72 1500$
5 w
8-
~
200 1000 100 500 PR ESSUR E a
1 I
I I
I I
I g
tst 0
200 400 000 200 10C0 1200 1400 1800 I
i l
l 1
l l
(mins 0 5
10 15 20 25 i
UME I
l l
Tig. 8-4 Turbine ::1p vi:h stuck-open bypass values (Scra= 'Jorth = 0.061 Ak/k): Tenperature and l
Pressure Transients.
l l
i i
I l
I l
r
8-12 Table 8.4 Turbine Trip vith Stuck-Open 3ypass
~
Valves (Scram Worth = 0.061 Ak/k):
Temperature and Pressure Transients Ti:na Core Inlet Primary System (s)
Temperature (*F)
Pressure (psia) 0 556.00 2,192.00 5
563.64 2,289.31 f5 528.73 1,570.92 125 444.09 523.95 215 328.42 320.38 340 271.34 275.21 630 197.45 365.93 800 167.70 463.54 900 157.08 561.81 1000 149.97 723.80 1050 147.34 847.90 1100 145.20 1,020.84 1150 143.45 1,273.30 1200 142.06 1,663.70 1240 141.17 2,149.54 e
O 9
w n,
D s
8-13 8.3.4 Small-Break LOCA The SBLOCA case was defined and the thermal-hydraulic analysis was performed by los Alamos National laboratory; it was reviewed by NRC in June 19818 (case 5 in Section 5). The break was assumed to be in the cold leg downstream of the main circulating pump and ahead of the HPIS injection nozzle and was sized at 10% of the pipe area. The charmal-hydraulic analysis was performed for the first 250 s only, and the resulting pressure and temperature transients are shown in Fig. 8.5 and Table 8.5.
The temperature given is that calculated for the top of the dotcccomer; since the main circulating pumps were assumed to be tripped I
at 15 s into the transient and the HPIS was assumed to inject coolant i
for the duration of the transient, this coolant temperature is probably higher than would actually exist locally at the vessel wall, i.e.,
channeling of the HPIS coolant would occur.
8.3.5 Iarge-Break LOCA 2
The I2LOCA has been under detailed investigation for several years 10 and differs f rom the other transients considered in that no repressurizaton of the primary system takes place, and the downcomer temperature transient consists of an essentially step change in temperature f rom normal operating temperature (550*F) to ~70*F.
The fluid-film heat-transfer coefficient used in the thermal analysis of the vessel corresponds to f ree convection and was estimated to be 300 2.*F,2..y B eu/h r* f t which corresponds to a total surface conductance of
~200 Stu/hr* f t 8.4 3asults'of Fracture-Mechanics Analyses The f racture-mechanics analysis will indicate one of three possible results for each specific case:
(1) there will be no crack initiation i
for any reasonable assumed preexistent crack depth; (2) crack initiation will occur, but the crack will arrest permanently; or (3) crack initiation will occur and the crack will penetrate the wall.
The results of the analysis are quite sensitive to th. radiation-induced reduction in toughness and thus to the fast-neutron fluence, which is a function of the operating time of the reactor. The refo re,
the results are summarized la te:ms of the threshold fluence for incipient crack initiation and for failure of the vessel, and this is done for two cases:
(1) assuming WPS to be effective (if appropriate conditions exist), and (2) assuming WPS not to be effdctive (even if appropriate conditions do exist). A :,:nnmary of results for the five overcooling accidents analyzed is presented in Tabler 8.6 and 8.7.
Table 8.7 indicates the total number of EFPYs that a B&W-type reactor can operate before the overcooling transients considered would likely result in vessel failure.
The summary of results presented in Table 8.6 shows that for all cases analyzed the sinimum critical crack depths for initiation are in the
a 6-ia CRNL OWG $1 16492 ETQ l
l l
l 4
2500 500 TEMPERATURE 2000 CO 0
I
=
3 1500 q ll00 5
=g 9R ESSU R E y
2 20 1C00
- 00 500 100 0
0 0
50 100 150 000 250 11ME (s)
Fig. 3-5.
Small-Break LOCA ta=pera:ura and pressure ::ansien:s.
O
,,-y
8-15 Table 8.5.
Small-B reak I.0CA Temperature and Pressure Transients Time P
T (s)
(ksi)
(*F) 0 1.63 553 12 1.45 558 576
~
1 17.3 25 1.20 570 50 1.06 553 75 0.930 537 100 0.777 516 125 0.590 486 150 0.457 459 175 0.383 439 200 0.348 430 225 C 322 423 250 0.290 415 O
l I
8-16 range of 0.17-1.3 in.
This implies that at least some of the flaws, because of their size (upper and of the range), might have a high probability of being detected by condestructive means. Howeve r, fo r fluences somewhat greater than those associated with incipient initiation and failure the upper and of the range is =uch lover.
The calculated critical crack depth would be further reduced relative to the values in Table 8.6 by including the effect of cladding in the f racture mechanics analysis, assuming, as we are, that the crack extends through the cladding. As mentioned earlier, the inclusion of cladding in the analysis will also result in smaller threshold fluences. Thus, in this respect the results in Table 8.6 and 8.7 are somewhat optimistic.
In evaluating the data in Tables 8.6 and 8.7 there is a distinction that must be made between the large-break LCCA, and the other cases.
Since the GLCCA does not involve repressurization, a long, axially oriented flaw presumably would not extend completely through the wall, and even if it did the crack would remain tight and tnus leakage of coolant presumably would be negligible. For the other cases the primary system pressure is high enough, consistent with assumptions made, to force the crack all the way through the wall. Furthe rmo re, since the system is pressurized, the temperature of the coolant could be high enough to result in sufficient energy release.during blowdown to open the crack substantially. As mentioned earlier, a detailed analysis of crack opening under these circumstances has not yet been pe rfo rmed.
As indicated in the tables, LTS was predicted for each of the cases conside red.
It may be reasonable to take advantage of WPS for the GLOCA since, by definition, there is no repressurization; however, for the other cases, variations in the repressurization could preclude conditions for WPS without making significant differences in the results othe: vise. Thus, one cannot necessarily depend on WPS to reduce the consequences of the transient.
The fluences listed in Table 8.6 correspond to those at the inne r surface of the vessel vall at locations having the specified copper concent rations. Thus, to determine the number of EF?Ys in Table 8.7 it is necessary to know the fluence rate (fluence per EFPY) at the same locations. To establish the most limiting locstion one sust consider the combined effect of fluence, copper concentration and initial RUDT
( RUDT ). The location that would tend to have the highest RUDT o
(RUDT + ARUDT) would be the likely choice. Such a location was establ$shed for Oconee-1 in Ref.11, and the corresponding fluence race.
is 0.046 x 1019 neutrons /cm /EFPY. According to the information in 2
Sect. 6 the uncertainty in this value is 250%. The =ean value was used to obtain the threshold times (EFPYs) to vessel failure listed in Table 8.7.
,e 9
Table 8.6.
Suneaary of Flaw Ilehavior Characteristics for Several liypothetical Oconee-1 Overcooling Accidents (refer to the list of nomenclature for this table on the following page)
T 1
F a
t F
a t
(a/w)'
11 (a/w)*
if (a/w)it if 11 2
(n/cm x 10I9)
(min)
(n/cm2 x 10I9)
(min)
Large-Break 1.0CA WPS 0.4
% 0.G4
- 5. 5,
O.I5 0.9 0.02-0.22 1.5-8 Without WPS 0.15 0.I 30
- 0. 's 0.2 0.04-0.18 14-45 Rancho Seco WPS 1.5 0.I 40 1.0
- 1. 5 0.1 40 Without WPS 0.9 0.1 65 1.0 0.9 0.1 65 Turbine Trip /Open Bypass Valves WPS 0.2" 0.06 22 1.0 0.2" 0.06 2 2 e'.
o>
d d
Withont WPS 0.13 O.15 60 1.0 0.13 O.15 60 Main Steamline Break WPS 0.4 0.04 9
1.0 0.4 0.04 9
d d
Without WPS 0.2 0.06 18 1.0 0.2 0.06 18 Sawill-Break I.0CA No initiatione af f a range of crack sizes is not shown, shallower and deeper flaws will initiate at higher fluences.
bFor this case, failure refers to crack penetratton huyund a/w = 0.9.
Presumably the crack will not actually penetrate lhe outer surface.
CCrack depths greater than a/w - 0.2 ignored.
Uf the transient time were extended beyond 60 min., Fg wraald Le less.
l C )urat ton of thermal-hydraulic transient simulated too short to permit meaningful fracture mechanics analysis.
l
0 8
3-18 Nomenclature for Table 8.6 (a/w)gg f ractional crack depth for incipient initiation (a/w) g f ractional crack depth for final arrest following incipient initiation (a/w)g*
f ractional crack depth for first initiation that results in vessel failure (corresponds to threshold fluence for failure)
F, threshold fluence at inner surface of vessel wall for incipient crack g*
initiation F
threshold fluence at inner surface of vessel wall for incipient vessel p'
failure s
e time in transient for incipient crack initiation gg e
time in transient for incipient vessel failure gg e
O s
1 l
4
)
Table 8.7.
Estimated Threshold Times for Vessel Failure for a B&W-type Reactor for Various Overcooling Events Threshold Timea b
Commenta and Qualifications e
1.a rge-B reak IDCA 20 WPS assumed ef fective (4 EFPY w/o WPS). Crack not expected to penetrate RV wall; vessel lutegrity expected to be maintained, since rep ressu rization does not occur.
Rancho Seco 20 Wl'S not assumed effective (33 EFPY if it were).
Through-wall crack predicted.
Turbine Trip with Open 3
WPS not assumed effective (4 EFPY if it were).
Ilypaus Valves (RFT) C Through-wall crack predicted.
Main
- Steam hine B reak C 4
WPS not assumed effective (8 EFPY if it we re).
Through-wall crack predicted.
T U
Small-B reak 1ACA Thermal-hydraulic forcing functions calculated to only 250 s (at which time temperature and pressure are still decreasing), thereby preventing meaningful analysis of crack propagation in RV.
aEffective full power years at which failure of the vessel is predicted, given the pressure and thermal driving functions presently predicted for the transients, and the assiumptions used in this study.
halased on a fluence accumulation rate valuell of 0.046 x 1019 2
n/cm /EFPY. Depending on the specifica of surveillance programs and fuel management schemes, this value may have an associated uncertainty. of as much as 150%.
CFa11ure preedictions based on thermal-hyd raulic calculations containing conservative assumptions.
i s
u
8-20 REFERENCES - CHAPTER 8 1.
F. J. Loss, R. A. Gray, Jr., and J. R. Hawthorne, Significance of Warm Prestress to Crack Initiation During Thermal Shock, NRL/NUREG Repor 8165 (Sept. 29, 1977).
2.
R. D. Cheverton and S. K. Iskander, RSST Program Ouarterly Progress Report for October-December 1980, NUREG/CR-19t+1, pp. 37-50 (March 1981).
3.
ASME Boiler and Pressure-Vessel Code,Section II.
4.
S. K. Iskander, R. D. Cheverton and D. G. Ball, OCA-I, Code for Calculating the Behavior of Flaws on the Inner Surface of a Pressure Vessel Subjected to Temnerature and Pressure Transients, ORNL/NUREG-84 (August 1981).
5.
Letter to R. Kryter, ORNL, from R. J. Cerbone, BNL, Aug. 14, 1981.
6.
Letter to R. D. Cheverton, ORNL, from J. Strosnider, NRC, January 30, 1981.
7.
Letter to Novak Zuber, NRC, from M. M. Lavine, 3NL, July 1, 1980.
8.
Letter to R. D. Cheverten, ORNL, from M. M. Levine, 3NL, August 11,. 1980.
9.
Letter to C. Z. Serpan, NRC, from S. Fabic, NRC, June, 22, 1981.
10.
R. D. Cheverton, S. K. Iskander and S. E. 3olt, Aeolicabilit r of LEFM to the Analysis of PWR Vessels Under LOCA-ECC Ther*al-Shock Conditions, ORNL/NUREG-40 (October 1978).
- 11. J. Strosnider, NRC, personal coczmunication (September 1981).
O 9
9
.e-one w
- w-
.+
m
-g w
9 p
9
+
9.0 FOSSISLE MITICATIVE MEASURES 9.1 Operator Actions In the case of the Oconee-1 control system, reactor operators clearly have a manual capability to terminate excessive feedwater for a wide variety of failures. However, such actions require that the opcrator correctly recognize overcooling problems at a time early enough in the t ransient that his resulting response is effective. Diverse sources of information are available to the operators, from which a determination of overcooling can be made. It is therefore possible that operator response can be an effective deterrent to this problem.
It is conceivable that the operator might be able to control the outlet subcooling for certain accidents, as BW proposes. How practical this is, considering instrument locations and fluid transport times (particularly if the reactor coolant pumps are tripped), needs to be evaluated.
The questions of whether and/or when to trip the reactor coolant pumps in overcooling upsets need evaluation. Tripping the pumps will raise the temperature and delay the influx of cold water to the vessel for steam-generator-driven transients; however, maintaining a properly cooled core and promoting good downcomer mixing may necessitate leaving the pumps on.
It is evident that the operator needs a definitive indicator of adequate core cooling and vessel wall temperatures to achieve a proper balance between concerns for core cooling and vessel overcoolings.
9.2 System Changes 9.2.1 Oconee Changes Our review of the Oconee-1 control system revealed that several of the upgrades already performed will act to reduce the likelihood and extent of excessive feedwater transients in this plant. Among these changes are upgrades to the instrument and control system power supplies, identification of diverse information channels for operator use during power supply failures, automatic feedwater pump trip on loss of ICS i
l power, and steam generator high level limit trips for the main f eedwater pumps. This reduces the probability of excessive feedwater being supplied to that of two failures of the control system and failure of the operator to take corrective action or overt ope rato r error in manually actuating the feedwater.
t r-ar +'
e av-Tiw a
w**e+
-w-w-w-y-----t C'-
r--wvv't---+-
<s-me
,.y e9
--ve---D.,
-+---y+--we--
-v----w,,r yyyy
-e
-~--w
-v
^--e---+
9-2 Other potential changes can be identifed. An analysis of the condensate booster pump response following mein feedwater pump trip could be used to determine whether or not other feedwater system control actions should be initiated on loss of ICS power. Similarly, analysis could be performed to determine whether a main feedwater trip based on a specified pressure-temperature envelepe could be used to p revent inadvertent overecoling. A more detailed systems analysis could possibly identify other alternatives.
9.2.2 General Changes There are several other changes to the reactor system which deserve evaluation for their effectiveness, cost, and practicality of implementation. These are outlined below.
Borated Water Storage Tank (3WST) Temperature.
Increasing the temperature of the 3WST fluid would reduce the degree of overcooling caused by actuatloa of the safety injection systems. However, this measure obviously has limited effectiveness, since some tanks probably cannct be heated above ~80*F and, in any case, maintenance of temperatures above ~200*F would be impossible without tank p ressu ri:ation.
Feedwater Train Storage. Limiting the amount of water available to the f eeswater system would obviously reduce the severity of the steam-generator-driven transients. However, there is a trade-off here with p ractical requirements for oormal plant operation.
Containment Flooding.
This "fix" has been proposed and discussed in connection with other major accidents. It would obviously mitigate the consequences of a reactor vessel breach; however, inadvertent actuation of such a system might itself produce a vessel breach through rapid and extreme overcooling of the RV outer wall.
9.3 Restoring Pressure Vessel F racture Toughness 37 Annealing F rom a vessel design point of view, the most desirable solution to the ove rcooling-accident vessel-integcity problem is to restore the vessel material f racture toughness, which is gradually reduced during reactor operation as a result of exposure of a vessel wall to fast neut rons. Studies that have been underway for several years indicate that the toughness can be restored by annealing at temperatures in the range of 750-850*F for a period of approximately 200 hours0.00231 days <br />0.0556 hours <br />3.306878e-4 weeks <br />7.61e-5 months <br />
,2 The l
results of studies conducted by Westinghouse, under contract to EPRI, indicate that conducting the annealing treatment of the irradiated vessel is practical for some and perhaps nost of the vessels in service l
today.
t REFERENCES - CH. APTER 9 1 T. R. Mager (Westinghouse), personal communication to R. D. Cheverton (ORNL), September 29, 1981.
2 F. Loss et al. (NRL).
e e
O e
e
1 i
10.0 CONCLUDING REMARKS AND REC 0teiENDATIONS FOR FURTHER WRK 10.1 Concluding Remarks l
Despite a fair degree of recent effort in the study of pressurized thermal shock phenomena by a number of knowledgeable groups, the true severity of the threat is, at present, very difficult to ascertain with confidence. The principal problems contributing to uncertainty are:
The computer codes presently being used to simulate the hypothetical overcooling transients were not designed to I
treat some of the phenomena that take place and hence produce inaccurate (sometimes nonphysical) thermal-hydraulic forcing functions under certain circumstances. The results of the i.
f racture-mechanics analysis used to predict vessel failure are known to be sensitive to the temporal behavior of these forcing functions.
The temperature indications in the lone set of actual plant data available (Rancho Seco) provide only a nominal indication of true RV wall conditions (they could be too high or too low by an indeterminate amount), and the chart-recorded pressure traces are made suspect by the presence of large " spikes" of unknown and possibly nonphysical origin.
The thermal / stress /f racture-mechanics analyses presently used to predict ~ crack propagation resulting from the temperature and I
pressure forcing functions have limitations (e.g., 1-D thermal and stress analysis; lack of treatment of azimuthal and axial variations in downcomer coolant temperature; inability to i
account for the axial gradient in wall fluence; lack of treat-ment of vessel cladding in f racture-mechanics analysis) which introduce uncertainty of an unknown magnitude in the results.
The fluence at the vessel wall and at critical welds is probably known only to an accuracy of 130% (perhaps 150%), and this implies an uncertainty of like magnitude in the vessel " life remaining" figures.
The probabilities of occurrence for various overcooling accident initiating events have associated uncertainties of at least plus-or minus one order of magnitude, and the conditional probabilities for correct subsequent operator diagnosis of a transient, timeli-ness and correctness of operator response, appropriate, automated control and safety features responses, and the like are at present j
undetermined.
,,.-,-,,ry-+-,
--+ ew s ww,--,-,...-.>
.4---..-e=r,m.r.
.-,.e-,-w w-,,ve--,,---..e-,
ewer"
-w e---e-- -weew~~~---*--"--+*---*=---+-'-=--'~*"--=-------"-*e--"~~
. 2 Nonetheless, for all their shortcomings, the analyses at hand are the best presently available on a nonproprietary basis, and, owing to the apparent severity of the outcomes predicted f rirs the limited number of overcooling scenarios studied, it is oar opinion that pressurized thermal shock atst be regarded as a serious potential threat and merits s great deal more study using refined techniques.
10.2 Recommendations for Further k'ork In order to reduce the magnitudes of the uncertainties described above, we recommend that additional work be undertaken in the following areas:
Refinement of thermal-hydraulic simulation codes and associated models (in particular, creatment of the feed train, fluid mixing, the control system, primary coolant system repressurization and flow distribution, two phase phenomena, and the heat capacity of heavy primary metal).
Refinement of vessel thermal / stress /f racture-mechanics analysis techniques (in particular, a consideration of higher dimension-ality in several of the variables treated, and inclusion of the vessel cladding in the fracture mechanics).
Refinement'of the analytical methods and surveillance capsula data assessment procedures ~ required to estimate fast-neutron fluence in selected areas of the RV wall, in order that state-of-the-art accuracies (+10: ) may be realized.
A thorough study of the probability st:ucture of the various intertwined occurrences (among them, normal plant maneuvers, chance events, equipment and operator failures, plant recovery actions, etc.) that are necessary to produce the severe thermal shock conditions that constitute a serious threat to RV integrity.
Some facets of this recommended work are known to be in progress by the NRC and reactor owner's groups or will be initiated in FY 1982.
O dp
=aw*
- - w ee e o = = e w-o m. - e oms.
o APPENDIX A CAPSULE DESCRIPTION OF DATA 3ASES EXAMINED 1.
BERA - Fluid Engineering BHRA Fluid Engineering provides indexing and abstracting of world-wide information on all aspects of fluid engineering, including statics and dynamics, and laminar and turbulent flow. Theoretical research is covered, as well as the latest technology and applications. Data are taken from 3RRA's ten secondary abstract publications, which abstract over 530 technical journals as well as books, proceedings, standards, technical reports, and 3ritish patents. Major fields covered include civil engineering hydraulics, industrial aerodynamics, dredging, fluid flow, fluid power, fluid sealing, fluidics feedback, and tribology.
2.
CIM - Inventory of Models The Central Inventory of Models data base is maintained by ORNL for DOE, and includes energy-related bibliographic and numeric data bases, graphics packages, integrated hardware /sof tware systems, and models from DOE laboratories.
3.
COMPENDEX - Engineering COMPENDEX covers significant world-wide engineering ?.iterature (1970 to date) from ~2,000 serials and over 900 monographic
~
l publications (including books and conference proceedings). Fields of engineering and related subject areas include: aerospace engineering, agricultural engineering and food technology, automotive engineering, bioengineering, chemical engineering, civil engineering, computers and data processing, construction sacerials, control engineering, electrical engineering, electronics and communications engineering, engineering geology, engineering physics, fluid flow, and heat and thermodynamics. Also covered are industrial and management applications, instruments and measurements, light and optical technology, marine engineering, sacerial properties and testing, mechanical engineering, setallurgical and sining engineering, nuclear technology, ocean and underwater technology, petroleum engineering, railroad engineering, transportation, water and waterworks engineering, and pollution, sanitary engineering, and waste.
1 4
CONF - Conferance Pacers This includes scientific and technical papers (1973 to date) in the life sciences, physical sciences, and engineering areas that are presented at regional, national, and international =eetings, including small meetings having a cross-disciplinary focus.
l
[
2 4
5.
EDB - Energy DOE Energy is one of the world's largest sources of literature reference.s on all apsects of energy and related topics. It includes references to journal articles, report literature, conference papers, books, patents, dissertations, and translations.
.4 1 manner of energy topics are included:
nuclear, wind, fossil, geothermal, tidal, etc., as well as the related topics of environment, policy, and conservation.
6.
EIA - Energy Information EIA Citations are drawn from Energy Information Abstracts, and are compiled by the Environmental Information Center.
7.
FEDEX - Federal Government Activities Federal Index contains information (19'/6 to date) on federal government activities drawn from the Congressional Record, the Federal Register, The Weekly Compilation of Presidential Documents Commerce Business Daily, and the Washington Post. Additional sources from the F & S Index are also included, beginning in 1979.
The citations provide' access to the Code of Federal Regulations, the U.S. Code, Public I.aws, Congressional Bills, Resolutions and Reports. The information is indexed by acting government agency, affected industry, or institution and type of government action or function.
8.
ISMEC - Mechanical Engineering ISMEC covers mechanical engineering, production engineering, and engineering management. Subjects covered include production processes, tools and equipnent, energy and power, transport and handling, management and production, measurement and control, and mechanics, materials and devices. References (1973 to present) ar gathered from journal articles, technical reports, conference proceedings, and books.
9.
NSA - Nuclear Science The Nuclear Science Abstracts base presently contains more than 500,000 citations, covering the period 1967 to June 1976.
10.
NSC - Nuclear Safety The nuclear safety information data base is maintained by the Nuclear Safety Information Center, ORNL, under the joint sponsorship of DOE and NRC.
me
+.-
r,
.;&y<
-e-
-S
---1
--T
+
3 11.
NTIS - Government Soonsored Research NTIS covers U.S. government-sponsored research and development technical reports from over 200 Federal agencies and some reprints, federally-sponsored translations, and foreign-language reports in major areas of technical interest.
Its multi-disciplinary scope includes aeronautics, agriculture, astronomy and astrophysics, atmospheric s:1ences, behavioral and social sciences, biological and medical sciences, chemistry, earth sciences and oceanography, electronics and electrical engineering, energy conversion (non-propulsive), materials, and mathematical sciences. Also covered are mechanical, industrial, civil, and marine engineering, methods and equipment, military sciences, missile technology, navigation, communications, detection methods and counter-measures, nuclear science and technology, ordnance, physics, propulsion and fuels, and space Lachnology.
12.
RSI - Radiation Shielding Information The Radiation Shielding Infor=ation Center data base is maintained by ORNL and contains citations to literature describing computer codes that have been designed to perform radiation analysis and shielding calculations, neutron cross-section processing, and 1
experimental data analysis.
\\
l 13.
LILDASEARCH - Joining of Metals and Plastics The kTLDASEARCH data base provides primary coverage of the international literature on all aspects of the joining of netals and plastics and related areas such as metals spraying and thermal cutting., kTLDASEARCH includs material on welded design, welding metallurgy, and fatigue and fracture mechanics, as well as welding and joining equipment, corrosion, thermal cutting, and quality control. Approximately 5,000 new records are added to kTLDASEARCH each year from several thousand journals and research reports, l
books, standards, patents, theses, and special publications.
O
APPENDIX 3 OCONEE LICENSEE EVENT REPORTS (LERs)*
1.
Cooldown Rate Limit Exceeded Following Loss of ICS Power at Oconee-2 The RCS cooldown race limit was exceeded after power to the ICS was lost for about two and one-half minutes. No ES actuation setpoints were reached, and adequate RCS inventory was maintained, i
No damage was incurred. Loss of ICS power resulted from blown fuses in normal inverter (KI) and failure of transfer witch to transfer automatically to regulated AC power. When ICS power was restored, excessive feedwater flow caused a rapid RCS cooldown. A redundant transfer switch has since been installed, and personnel have been instructed on how to respond properly to loss of ICS power.
2.
Reactor Coolant System Cooldown Rate Excessive at Oconee-3 During a routine shutdown for mat,ntenance, a minor system transient occurred, which resulted in opening a power-actuated pressurizer relief valve with reactor power at 15%. The valve remained open and the RC system depressurizaton continued until the isolation valve was closed. The shutdown continued with a cooldown rate of 100*F/hr.
However, when the initial drop in temperature from depressurization was included, the rate exceeded the 100*F/hr tech spec limit by 1*F/hr. It was determined that boric' acid crystal buildup on the connecting pin of the lever arm of the pilot valve had caused the valve to remain open.
3.
Additional Information on Excessive Cooldown Rat'e at Oconee-3 Reactor, power was being reduced from 100% to 15% by the integratei control system for a maintenance shutdown. When 15% was reached, unit load demand was 65 MWe and power generation was 115 MWe.
This difference existed because the reactor was operating at its lower limit of 15% and could not folinw load demand. A transient occurred that tripped the reactor.
During the transient, a relief valve opened and failed to close. This transient was terminated by closing the isolation valve. Cooldown rate was 101*F/hr during the first hour. The relief valve failed because of heat.
expansion, boric acid crystal buildup on the valve lever, and bending of the solenoid spring bracket.
- Text has been modified slightly in scue instances to improve clarity and readability.
mv.
,y-
- .--wr,--+--t
,w,pr
,q,
,,-.m
.,.3
,-,.,m,
-.-99..,
g.
,w,.e y.,..
n w.
c
~
s.
2 4
Feedwater Transient Following Scram Actuates HPI at Oconee-1 4
On December 13, 1978, the I statalarm began to act arratically andaninvestigationwasinfEIated. During investigation (12/14/78) a power cord supplying T,y, recorder shorted, causing ar apparent (not real) drop in T,y, or 13*F and ICS attempted to correct T,y,.
Unit tripped on high pressure / temperature.
Feedwater transients during cooldown allowed OTSG "B" to go dry.
When it was refilled it caused RCS pressure to drop below 1500 psi, which actuated the HPIS. The cause of the T,,, cord short has not been identified. The feedwater transients were probably caused by improper valve operation. The power supply cora was replaced.
5.
Reactor Coolant System Cooldown Rate Exceeds Limits at Oconee-2 When a spurious cignal in the 230 kV switchyard circuit breaker failure relay circuitry resulted in the isolation of the switch-yard, the reactor scrammed from 75% power. The scram tripped the feedwater pumps. The emergency feedwater pumps started and filled the steam generators to the 95% level as designed. This high water level, plus normal required steam, resulted in a cooldown rate of 140*F/hr in one loop and 135.5'F/hr in the other, which exceeds the 100*F/hr limit. Reduction in water level set point is being studied.
l l
O l
l
t?
APPENDIX C SPECIFIC DOCUMENT REFERENCES 1.
" Reactor Coolant Pressure and Temperature Data for the March 20, 1978 Cooldown Event at the Rancho Seco Power Plant," letter, 4
S. Fabic to C. Serpan dated November 25, 1980.
2.
" Parametric Analysis of Rancho Seco Overcooling Accident," letter, R. Cheverton to M. Vagins dated March 3,1981.
3.
Effect of HPI on Vessel Integrity for Small B reak LOCA Event with Extended Loss of Feedwater, BAW-1648 (November 1980).
- 4. " Runaway Feedwater Af ter Turbine Trip Report," letter, M. Levine to N. Zuber dated July 2, 1980.
5.
" Transmittal of Preliminary Calculations of a Steam Line B reak Accident," letter, S. Fat,1c to C. Serpan dated May 14, 1981.
6.
" Analysis of a Steam Line B reak with Primary System Overcooling for a Typical 34W Reactor", letter, R. Carbone to R. Kryter dated August 14, 1981.
7.
" Completion of Scheduled Analyses on Pressurized Thermal Shock Scenarios," letter, S. Fabic to C. Serpan dated June 22, 1981.
8.
Analysis of Capsules CCl-F f rom Duke Power Comeany Oconee-1 Reactor Vessel Materials Surveillance Program, 3AW-1421 (August 1975).
l 9.
"Oconee Nuclear Station Docket Nos. 50-269, -270, -287," Lette r i
Report, W. O. Parker, Jr. to H. R. Denton dated July 23, 1980.
1 10.
Final Safety Analysis Report, Oconee Nuclear Station Units 1, 2, and 3, Rev. 19, Duke Power Company (May 5, 1972).
11.
IRT - A Pressurized Water Reactor System Transient Code, B rookhaven National Laboratory draf t repo rt dated December 1980.
12.
" Thermal Shock to Reactor Pressure 7essels," Letter Report, l
R. W. Jurgensen to D. G. Eisenhut dated May 14, 1981.
13.
" Reactor Vessel 3 rittle F racture," Letter Report, J. J. Mattimoe to H. R. Denton dated May 12, 1981.
14.
" Reactor Vessel Pressurized Thermal Shock," Latter Report, K. P. Baskin to D. G. Eisenhut (undated).
1 i
15.
EPRI Research on the Procerties of Irradiated Materials Pertinent to the Overcooling Transients, T. V. Marston, Ed. (April 1981).
4 p--r-
-- y e-4m-..w--_
c,6
.-m,_
_,,,__y,
-__,3-e p
.,,..,..,.. +
Q:
i 2
References (Cont'd) 16.
" Pressurized Thermal Shock," letter, D. F. Ross to R. Bernero dated May 19, 1981.
17.
" Rancho Seco Data," letter, J. Strosnider to R. Cheverton dated January 10, 1981.
18.
"?V Thermal Shock," letter, W. 3. Cottrell to G. D. Whitman dated October 21, 1980.
19.
"IRT Output," letter, M. M. Levine to R. D. Cheverton dated August 11, 1980.
20.
"IRT Results," letter, M. M. Levine to N. Zuber dated July-2, 1980.
J t
S O
e O
o r
e v-
~-
~~
^