ML19211A734

Technical Evaluation of Draft NUREG-0630

ML19211A734
Person / Time
Site:	Yankee Rowe
Issue date:	12/10/1979
From:	YANKEE ATOMIC ELECTRIC CO.
To:
Shared Package
ML19211A732	List: ML19211A731 ML19211A734
References
RTR-NUREG-0630, RTR-NUREG-630 NUDOCS 7912200556
Download: ML19211A734 (16)

Text

Attachment 1 Appendix A TECllNICAL EVALUATION OF DRAFT NUREG 0630*

Prepared by Yankee Atomic Electric Company December 10, 1979

• D. A. Powers and R. O. Meyer; " Cladding Swelling and Ruptur:: Models for LOCA Analysis";

November 8, 1979.

1630 288 7912200 $ N"

Table of Contents Section Page 1

Introduction 1

2 Comments on the Data Base 2

3 Comments on the New Correlations 4

3.1 Rupture Temperature 5

3.2 Burst Strain 9

3.3 Assembly Flow Blockage 10 4

Summary and Recommendations 11 References 14 i630 289

1.

Introduction On November 8, 1979, a draft report prepared by tbe staff of the Office of Nuclear Reactor Regtlation, " Cladding Sw. lling and Rupture Models for LOCA Analysis" (DRAFT NUREG 0630), was issued by the USNRC.

It is presently being circulated for review by the technical community for the purpose of obtaining a technical critique. This report presents such a technical evaluation by Yankee Atomic Electric Company.

The importance of increasing the industry's existent understanding with respect to fuel rod behavior during LOCA and of improving the applicable data bases and phenomena correlations and models is recognized. These goals define the basis of this critique.

1630 290

_1_

2.

Comments on the Data Base 2.1 Applicability of the Data The selected data base, which has been restricted to experiments in aqueous environments that utilize either internally heated fuel pellet simulators or fuel pellets in-reactor, is generally appropriate. The extent of Zircaloy cladding swelling at rupture has been demonstrated to be influenced strongly by the degree of cladding oxidation which exists prior to and occurs during cladding heatup.(I' According to Kassner et al., oxidation and temperature nonuniformities in cladding as would occur in-reactor will decrease the ballooning of the Zircaloy cladding, in contrast to data obtained in inert gas atmospheres.

Data obtained in steam with direct heating of cladding may still be appropriate for determining a cladding rupture criterion (see Section 3.1); however, such methods are generally designed to produce uniform cladding temperatures and hence result in unrealistic cladding strains which should not be used in devising or qualifying a LOCA predictive model.(4) 2.2 Data Base Limitations This restriction of the data base results in a sparse data field, especially in the regime of high temperature / low preastra burst. Whereas the report attempts to quantify the effects of heatup ramp rate on rupture, swelling, and blockage, the data base is incomplete. No data are included which are within the high temperature regime (>1000 C) for slow cladding temperature ramps.

2.3 Data Qualification The final report should provide some detail with respect to experimental measurement techniques and data interpretation 1630 291

met hod s. For instance, the ORNL Multi Rod Burst Test Program reports burst temperature measured by external and internal thermocouples as the highest observed thermocouple reading regardless of burst location.(5) The FR2 nuclear tests report burst temperature calculated from thermocouples above and below the burst and corrected for the extent of burst deformation.(6)

Estimates on the underprediction or accuracy of the reported data with respect to actual temperatures at external and internal cladding surfaces should be provided.

Similar evaluations of other measured parameters (e.g. pressure and strain) should be included.

Since the research programs and data evaluations examining these phenomena are not complete, this information should be considered both necessary and valuable.

2.4 Data Evaluation Data have been reported which show large discrepancies in important measured parameters.

Such data values should be explicitly identified in the analysis, correlation development, and associated figures.

For example, Rods #B1.5 and #Bl.2 of the FR-2 (6) showed almost identical ramp rate, burst pressure tests (engineering burst stress), and burst temperature measurements; however, burst strains are reported as 60 and 25 percent, respectively. Rods #A1.1 and #B3.2 exhibited similar engineering rupture strains at equal burst pressure; the burst temperature measurements which differ by 105 C are not explained by the slight difference in temperature ramp rate.

For all data sets, the analysis should attempt to resolve these types of discrepancies and/or clearly identify that the discrepancies exist.

1630 292 3.

Comments on the New Correlations The correlations in the report are based upon a relatively restricted data set.

Reliance on three multirod burst test experiments at two cladding heat-rates is apparent. The objectives of the ORNL MRBT program have been to " delineate the deformation behavior of unirradiated Zircaloy cladding under conditions postulated for a loss-of-coolant accident (LOCA) and to provide a data base to facilitate assessment of the magnitude and distribution of geometrical changes in the fuel rod cladding in a multirod array and the extent of flow channel blockage that might result."(

To fulfill these objectives the program developed a test matrix designed to examine effects of fuel pressure and temperature at rupture in a multirod array.

Several features of the experimental setup have been chosen to characterize performance in relation to independent variables. These include steam flow rate, shroud heating and heatup rate, and uniform axial heat generation. To this extent specific, anticipated in-reactor LOCA conditions were not intended and have not been produced. Program goals of providing a test data matrix for predictive model validation are being met.

The correlations to " predict" cladding strain at rupture and assembly blockage tacitly assume that the data base includee results representative of LOCA conditions.

The analysis then envelopes this information with respect to the cladding temperature parameter. This process treats data within the envelope as dependent variable scatter with respect to the modeling parameter and ignores analysis of that data. The report should express this philosophy of correlation developaent, attempt to identify its limitations, and suggest potential model improvements which would predict, rather than interpret, the data. 1630 293

3.1 Rupture Temperature 3.1.1 Statistical Evaluation of the Correlation The rupture temperature correlation was developed within the ORNL MRBT program (0 as a function of rupture pressure and translated to engineering stress at rupture to apply to the extended data base of the DRAFT NUREG report.(9}

6

~ 20.4 S ~

8.51 x 10 g 1+H 100(1+H) + 2790 S where:

H=

Heat rate ( C/sec)/(28 C/sec)

T=

rupture temperature ( C)

S=

engineering stress at rupture (KPSI)

Since the data base for the correlation is mainly from the ORNL program, an evaluation has been performed to benchmark the correlation against the extended data base of the DRAFT NUREG report.

In this study, data points were eliminated from consideration if they failed either of the following tests:

1) ramp rate accurately known and 2) all information for a data point is known.

Applying these two conditions, a total of 112 data points were applicable to the analysis. The analysis consisted of evaluating the NRC correlation at each data point and comparing the calculated value of burst temperature to the measured valus.

General observations include:

(1) the RMS value of the difference tends to decrease with increasing ramp rate and (2) the RMS values and numbers of points which are underpre-dicted by the correlation are greater than those which are overpredicted. 1630 294

This difference indicates a bias of the correlation to predict lower burst temperatures. The value for the RMS difference (N25 C) indicates that temperature is a major factor which must be further quantified. Tabic 3.1 lists the RMS values for different subsets of the data. A bias of underprediction of the rupture temperature is observed for all data partitions.

3.1.2 Systematic Uncertainty in the Data As discussed in Section 2, systematic uncertainties are apparent in the ORNL MRBT data for burst temperature. Cha pman has indicated that such uncertainties can be evaluted and reduced by careful analysis of the burst versus thermocouple locations.

Burst temperature data values may be on the order of 15 C to 35 C higher than those presented in the 10) report.

Since approximately 80 percent of the data analyzed in Section 3.1.1 were developed in ORNL tests, this potential bias should be considered to be additive to that which has already been demonstrated.

3.1.3 Limits and Extrapolation of the Data Base As indicated in Section 2, data are severely limited for low temperature ramp rate conditions which have recently been identified as applicable to most anticipated reactor conditions during a LOCA event.(11) No data have been presented in the temperature regime above 1000 C for slow heating rates.

Therefore, the heating rate dependence of the correlation evaluated in Section 3.1.1 is derived from data taken primarily in the 700 C - 900 C temperature range.

Such extrapolation 1630 295,

d '

g W

ignores the additional physical processes (i.e., cladding oxidation) which are known to occur in this regime and to influence the rupture process.

At least some data in this temperature range which was employed in the correlation development was not obtained in an aqueous environment.(I Preliminary investigations indicate that heating rate dependence may not exist in the high tempercture, cladding oxidation regime.

One data point excluded from the data reference by Karb(

(Rod #A2.3) ruptured at an engineering burst stress of 2.5 KPSI and a burst temperature of 1015 c.

The ramp rate for the data set is reported to be between 6 C/sec and 19 C/sec. This data point falls above the 28 C/sec correlation in this high temperature regime.

3.1.4 Applicability of Isothermal Data To derive the low temperature dependence of the correlation, isothermal (O C/sec) data have been used.

Examination of four of these experiments shows that hold-times of these tests ranged from 49 to 250 seconds.(13)

Such tests present data which are not consistent in that hold-time is an additional variable. Very low temperature ramp rate data rather than isothermal tests should be used to develop the correlation.

3.1.5 Ramp Heating Rate Accuracy Cladding heatup ramp rates reported in the data base are those experienced during heatup prior to the time-of-burst. As the burst time is approached, instantaneous heatup rates are

. 1630 296

consistently much less than that experienced during the preliminary heatup ramp.(5,8) This observation has not been addressed in the data analysis.

3.1.6 Burst Temperature in Direct Heating Experiments Observations in Sections 3.1.1 through 3.1.3 and Section 3.1.5 suggest that the burst temperature / engineering stress correlation may be greatly improved by increasing the accuracy of the temperature data.

Experiments designed'to accomplish this by reducing temperature variation (e.g., direct heating experiments) tend to produce large rupture strains. However, rupture temperature / pressure data tend to fall above the correlation curve in the range of that indicated by the systematic biases. The DRAFT NUREG study should investigate these data to determine if such experiments are valid for establishing a rupture criterion.

3.1.7 Other Burst Criteria Hagrman has recently evaluated several cladding stress at failure / burst criteria and has determined a true tangential stress versus temperature relationship to be most appropriate.

Heating rate and strain rate do not affect this criterion.

Hagrman used some data in the development of this correlation which were from direct electrical heating experiments and which covered a wide range of heating rates.

The DRAFT NUREG report should investigate this criterion and compare its applicability and accuracy with respect to the DRAFT correlation.

1630 297 9

i 3.2 Burst Strain Reference to the Chung and Kassner data (3'15) to quantitatively establish the temperature ranges for location of j

}

superplastic peaks and valleys appears valid, although explanations

[:

related to the physical phenomena of phase transformation and Ii Zircaloy oxidation would provide a more appropriate basis.

This I

li type of infonnation (phase transformation temperature, Zircaloy i.

e oxidation rate effects, Zircaloy creep tests) should be incorporated i

to justify the peak and valley positions with respect to i

temperature. The present approach limits accuracy of these positions to that achieved in this particular experiment.

h Scaling the peaks observed by Chung and Kassner to the present 11 data set assumes a quasi-linear relationship exists between burst h

strain results f rom direct heating experiments versus those from internal rod simulator heating experiments.

Setting peaks to enclose all the data assumes (1) that LOCA conditions have been accurately simulated in the experiments and (2) that peak values at given temperatures are accurate and represent real upper bounds of swelling at rupture. The data which fall below the derived bounding curve are then ignored; therefore, the interpretation process used in the DRAFT REPORT to address the " bewildering" scatter in the more realistic data base is unsuccessful. A more fundamental examination of the data which identifies experimental uncertainties and uncontrolled independent variables should be incorporated in correlation development for the final report.

i 1630 298 3.3 Assembly Flow Blockage The axial distribution of cladding rupture in the ORNL MRBT cannot be considered unbiased in that the location of burst is influenced by the performance of the internal heater element.(5,8) The suppressing effects of spacer grids exist in reactor bundles and should be considered in the data extrapolation.

It does not appear that the exact scaling factor, SF, has been used in the comparison to vendor models in that the derived blockage curves appear the same for each comparison.

1630 299 4.

Summary and Recommendations 1.

The experimental program should not be limited in scope of temperature and pressure conditions, but should continue to focus on providing a test matrix for qualifying LOCA predictive models.

2.

Data in the high temperature / low pressure burst regime for slow-ramp conditions are required.

3.

Data qualification with respect to measurement techniques, accuracy, and interpretation of each of the experimental data sets should be included in the final report.

4.

Apparent discrepancies in data within experimental data sets used in the report should be identified, discussed, and resolved.

5.

Statistical evaluation of the rupture correlation demonstrates bias toward underprediction of the burst temperature, especially at low cladding heatup rates.

Such an analysis should be included in the final report.

6.

The difference in temperature measurement techniques for the experiments comprising the data base should be identified. An attempt should be made to examine these differences and to quantify resultant temperature uncertainties.

7.

The rupture temperature / pressure correlation extrapolation to the high temperature regime is not qualified by any data in the slow 1630

$00

ramp range. The correlation in this regime, especially for low ramp rates, should be re-examined.

8.

Isothermal data may not be appropriate and should be re-examined for applicability.

9.

Accuracy of cladding heat-up rates should be re-examined for each data set and identified in the report.

10.

Applicability of direct electrical heating experimental results for burst temperature / pressure prediction only should be re-examined.

11.

The derived rupture correlation should be quantitatively compared to llagrman's f ailure criterion.

12.

Reference to Chung and Kassner experiments should be deleted in determination of cladding strain. Temperature regimes of strain peaks and' valleys should be derived from fundamental Zircaloy property / performance and from the current, applicable data base.

13.

Magnitudes of peaks and valleys in cladding strain versus temperature curves should be derived from the current, applicable data base with attention focused on inconsistencies, uncertainties, and accuracies of the data sets.

14.

The derivation of the assembly blockage correlation should recognize the systematic locations of rupture in the ORNL MRBT data sets and the effects of assembly spacer grids.

1630 301 Table 3.1 Evaluation of the Rupture Temperature Correlation Preditive Capability Underprediction Overprediction Cladding Total Total Number RMS Number RMS Temperature Number RMS of Points Deviation of Points Deviation Ramp Rate of Points Deviation (OC)

(OC)

(OC/sec)

(OC)

Min.

Max.

0 10 40 24 27 23 13 24 8

10 20 29 21 23 24 6

7 C

20 30 63 18 45 21 18 12

[.

0 15 48 24 34 24 14 24 15 30 64

)9 46 21 18 12 0

30 112 21 80 22 32 17 Note:

Underpred*ction implies that the draf t NUREG correlation predicts burst to occur at a lower temperature than observed in the test.

U O

u O

N

References (1) R. R. Biederman, et al.; "A Study of Zircaloy-4--Steam oxidation Reaction Kinetics"; EPRI NP-734, Project 249-2 Final Report Part 2; April 1978.

(2) R. E. Westerman, et al.; "Zircaloy Cladding ID/0D 0xidation Studies";

EPRI NP-525, Project 251-1 Final Report; November 1979.

(3) T. F. Kassner, et al.; " Development of Embrittlement Criteria for Zircaloy Cladding"; paper presented at the 5th NRC Water Reactor Safety Research Information Meeting, Gaitherburg, Md; November 7-11, 1977.

(4) D. A. Powers and R. O. Meyer; " Evaluation of Simulated-LOCA Tests that Produced Large Fuel Cladding Ballooning"; NUREG-0536; March 1979.

(5) R. H. Chapman, et al.; " Bundle B-1 Test Data - Multirod Burst Test Program"; ORNL/NUREG/TM-322; June 1979.

(6) E. H. Karb; "Results of the FR2 Nuclear Tests on the Behavior of Zircaloy Clad Fuel Rods"; paper presented at the 6th NRC Water Reactor Safety Research Inforation Meeting, Gaithersburg, Md; November 7, 1978.

(7) R. H. Chapman; "Multirod Burst test Program Quarterly Progress Report for January-March 1976"; ORNL/NUREG/TM-36; September 1976.

(8) R. H. Chapman; "Multirod Burst Test Program Quarterly Progress Report for April-June 1979"; NUREG/CR-1023; November 1979.

(9) D. A. Powers; Personal Communication; November 15, 1979.

(10) R. H. Chapman; Personal Communication; November 30, 1979.

(11) D. G. Eisenhut; Letter to All Operating Light Water Reactors concerning NSSS vendor and fuel supplier evaluations of cladding heating-rate effects; November 27, 1979.

(12) P. Hof fman; " Influence of Iodine on the Strain and Rupture Behavior of Zircaloy-4 Tubing at High Temperature", in ASTM-STP-681; 1979.

(13) R. H. Chapman, et al.; "Effect of Creep Time and Heating Rate on Deformation of Zircaloy-4 Tubes Tested in Steam with Internal Heaters";

NUREG/CR-0343; October 1978.

(14) D. L. Hagrman; " Cladding Stress at Failure"; paper presented at the 7th NRC Water Reactor Safety Research Information Meeting, Gaithersburg, Md; November 8, 1979.

(15) H. M. Chung and T. F. Kassner; " Deformation Characteristics of Zircaloy Cladding in Vacuum and Steam Under Transient - Heating Conditions:

Summary Report"; NUREG/CR-0344: AWL-77-31; July 1978.

1630 303

_1,_

.