ML20127J769
| ML20127J769 | |
| Person / Time | |
|---|---|
| Site: | Trojan File:Portland General Electric icon.png |
| Issue date: | 01/15/1993 |
| From: | Beckjord E NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Murley T Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML20127J684 | List: |
| References | |
| NUDOCS 9301250251 | |
| Download: ML20127J769 (45) | |
Text
. - - _ _ _ _ -
, . /po sseg#e UNITE D STATES
' 'df A" e NUCLE AR REGULA*f 0RY COMMISSION
? J' cAswievoton,0. c. eosts JAN 151993 MEMORANDUM FOR: Thomas E. Hurley, Director Office of Nuclear Reactor Regulation FROM: Eric S. Beckjord, Director Office of Nuclear Regulatory Research
SUBJECT:
INTER!H PLUGGING CRITERIA FOR TROJAN HUCLEAR PLANT In rny January 5,1993 letter to you on this subject I stated the position of this Office on the interim plugging criteria (IPC) for outer diameter stress corrosion cracking at tube support plate intersections, and gave the best estimate of the leak rate expected following a main steam line break (MSLB) for a pressure differential of 2600 psi across the tubes, and gave the best estimate of core damage frequency for the MSLB (or stuck-open safety valve).
The purpose of this letter is two-fold: (1) to provide a description of the leak rate calculations, and a description of the core damage frequency derivation, and the results of a study of the riSLB transient and resulting differential pressure across the steam generator tubes; and (2) to present RES plans for additional work en steam generator tube non-destructive examinations (NDE) that RES intends to do this year. Although Trojan has decided on permanent shutdown, I think both topics have applications and importance to other operating PWRs.
As previously stated the best estimate of the leak rate expected following a MSLB at the end of the Trojan Plant cycle 14. for a differential pressure of 2600 psi, is 145 gallons per minute (gpm). The uncertainty in the anhlysis is approximately an order of magnitude, with the predicted leak rate ranging up to 1350 gpm based on crack growth rates determined from examinations of the tubes removed from Trojan. Enclosure 1 describes the data, analysis, '
assumptions, and calculations to obtain these values. I must emphasize that this analysis is specific to Trojan and was focussed on short-term operation.
While the methodology is general, the results are Trojan specific.
The best estimate of core damage frequency for the sequence initiated by HSLB or a stuck-open safety valve has been revised somewhat to 1 E-6 per year. .
Enclosure 2 describes the data, analysis, assumptions, and calculations. .s The leak rate calculations above are based on the assumption of a steam generator tube differential pressure of 2600 psi. We have reviewed the transient analysis of the MSLB sequence for a Westinghouse RESAR plant
, representing Trojan. Calculations were done for three sizes of break. With no operator action, the steam generator tube differential pressure approaches 2200 psi after 30 minutes. However, operator procedures for the MSLB call for operator action to throttle back the pumps. The operators have 15 to 20 minutes available to take this action, and the result of the action is to limit the steam generator tube differential pressure to less than 1800 psi. A 9301250251 930119 i DR ADOCK 0500 4
7 ,.
Thomas E. Murley JAN 15 593 further result of the opet tor actions would be to reduce the leakage rate indicated above by a significant amount. Enclosure 3 describes the MSLB :
transient ' analysis for steam generator tube differential pressure, finally, Enclosure 4 describes our view of the current state of the art in eddy current inspection techniques arid the near term improvements in those techniques that can be anticipated. '
Our research program is addressing some of the key aspects of steam generator tube degradation, but our pe' mary emphasis remains on tube inspection techniques. However, we ar6 4 c'ous to work with you and your staff in developing a broad scope p*. .or addressing steam generator tube integrity issuts. ~
bh EricS.Beckjord,thrector Office of Nuclear Regulatory Research
Enclosures:
As stated cc: J. Taylor J. Sniezek F. Miraglia J. Partlow J. Roe W. Russell J. Richardson J. Strosnider L. Kokajko J. Fouchard Q
0 1
,.._...[..._..,. , . _ _ - . . _ , . . _ . . . _ - _ . _ _ _
9 .
ENCLOSURE 1 EVALUATION OF STEAM GENERATOR TUBE LEAK RATE UNDER MSLB CONDITIONS The evaluation described herein was >erformed to provide a 'best estimate' of the primary-to-secondary leak rate tint would be expected due to a Main Steam Line Break (HSLB) and the resulting leakage from steam generator tubes at the Trojan nuclear power plant without considering specific results of the In-service Inspections that have been performed. Tae term 'best estimate' must be qualified at the outset. An attempt has been made to provide an estimate that is not intentionally biased in either a pessimistic or optimistic asnner.
However, for many aspects of the evaluation, the data bases and general analysis models were not sufficiently complete or rigorous to support a "best estimate' in a statistical context. In those instances, engineering iudgement was necessary to complete the analysis. An attempt is made in the following discussion of the analysis methods and models to describe the data and the validation of the models, and to clearly identify where engineering judgement was invoked. Further, the sensitivity of the analysis results to the judgements made is evaluated to the extent practical.
To protect proprietary information used in this evaluation, neither the plant-specific flaw data nor the fit parameters developed for this evaluation are included in this document.
- 1. Introduction The overall approach is to estimate the leak rate for one steam generator, given that the MSLB event occurs, as the sum of the leak rates from four sources. That is, L sp ,, = L R, + LR, + LR, + L R, where, LR 3 - cumulative leak rate from tubes that were not leaking before HSLB, but develop a stable leak under MSLB LR, cumulative leak rate from tubes that were not leaking before HSLB, but develop rupture under MSLB LR, = cumulative leak rate for small leaks under normal operation, but l that increase under MSLB (take as 10 times the permissible LR I
under N0P)
LR, cumulative leak rate for small leaks under normal operation, but
! that result in tubt ruptures under MSLB (take as 0 for first l
analysis).
l The analysis methods and models for estimating LR and LR, are described in 3
l Section 2.
4 4 Setting LR, equal to 10 times the permissible leak rate under normal operation is based on engineering judgement and on leak rate models and experimental data reported in NUREG/CR-2336 (Ref.1). The basis for this judgement, and an evaluation of information that contradicts the judgement, is provided in Section 3.
Setting LR, equal to 0 is based on the engineering judgement that through-wall cracks sufficiently long to rupture under MSLB loading would leak enough to be '
detected during normal operation, and that the likelihood that several such long cracks would go undetected is sufficiently low that it would be a small contributor to the overall leak rate.
Other potential sources of leakage are not treated in this analysis. This ts based on the judgement that the primary form of degradation in the Trojan steam generator tubes is due to 0D500 confined to the tube-to+19be support plate region, or extending slightly beyond the tube support plate.
Section 4 describes the results of the present analysis and discusses those results in the context of the continued operation of the Trofan nuclear power plant.
Section 5 describes the sources of uncertainty in the analysis.
- 2. Methods for Estimating LR and LR, 3
Three models are necessary to estimate LR and LR : (1) a failure pressure a leak rate model estimationcapable of distinguishing model; and between (3) a flaw size model le,ak and, or distribution, rupture; (2) includin number of flaw locations. These models are comhined, as described below, to provide an estimate of the leak rate for one steam generator for each leak type.
Failure Pressure Models As part of the Steam Generator Tube Integrity Frogram, the Pacific Northwest Laboratories (PNL) developed an empirical cor. elation between crack depth and length and
- burst
- prer.sared (' ef. 2). However, it is not possible to distinguish batween a stable leak and a tube rupture from this correlation.
To resolve this problem, a model developed by Battelle-Columbus for predicting the failure of p'pe with axial vacks (Ref. 3) ras selected for this analysis.
The model was comsared to the burst t'est data. from tubes with EDM slots. The comparison was limited to the Trojan tube diameter of 0.875 inches with wall thickness of 0.05 inches. The data used in this comparison are reported in Tables F-1 and F-4 in NUREG/CR-0718 (Ref. 4).
The results of the comparison to the model reported in Ref. 2 indicate that the crack depth correction in the original model was 'over-correcting" for crack depth, giving very low predictions of burst strength for the deeply notched tubes. A non-linear correction was incorporated in the model to adjust the burst test predictions to give good agreement with the test results I for the EDM slots. The equations for predicting tube leak and rupture ar>
given in Figure 1.
To test the modified model e. gainst another data set, the burst test data from tubes with SCC cracks, reported in Table 2 of NUREG/CR-2336 (Ref. 1), were compared to predictions of those test data. Figure 2 is a plot of the predicted burst test pressures versus the measured pressures for both the EDH slot data and the SCC data. The pred h tions for the EDH slot data, where the crack dimensions are well defed, are generally within 110% of the burst test data. However, the predictio: for the SCC data are generally lower than the burst test results, figure 2 m.cludes +10% and -20% lines to provide a reference to assess the error in the prediction. It should also be noted that the modified crack depth correction results in a significant over estimate of failure pressure for very short cracks. Therefore, the modified model should not be used for crack lengths less than approximately 0.2 inches. However, this is of no practical significance for tiis evaluation since the failure pressure for such short cracks is in excess of 8000 psi, and approaches the full burst strength of uncracked tubes.
The equations presented in Figure 1 provide a method to estimate the failure pressure for a through-wall crack as a function of crack length, and the failure pressure for a surface crack as a function of crack depth and length.
The overall failure )rediction model uses the surface crack model to determine the pressure at whic1 the crack would extend through-wall. Then, the resulting through-wall length is compared to the through-wall failure curve to determine if the f ailure is a stable leak or a rupture. If the length is less than the critical through-wall length for the Main Steam Line Break pressure of 2600 psi, then a stable leak will result. Conversely, if the length exceeds the critical through-wall length, a rupture is predicted.
Figure 3(a) illustrates the application of the failure pressure model for the Trojan steam generator tube geometry and MSLB pressure. From this figure it can be seen that the critical through-wall crack length is 0.83 inches for the MSLB pressure of 2600 psi. A crack that is 82.2% of the wall thickness deep and 0.83 inches long would rupture under MSLB conditions. Cracks of that depth but shorter length would not penetrate the wall thickness under the HSLB pressure. However, there are a range of crack depths greater than 82.2% but shorter than the critical length that would result in stable leaks under the HSLB pressure, for example, a crack 90% of the wall thickness deep and approximately 0.4 inches long would result in a stable leak under the MSLB pressure. The full range of leak-rupture responses was considered in the leak rate evaluations.
The pressure differential across a tube under MSLB conditions also has a significant effect on the leak versus rupture behavior of a cracked tube, figure 3(b) illustrates the crack sizes that would leak for pressure differentials of 1200, 1600, and 1800 psi. Note that for these pressures, * /'
l tube rupture is not feasible for realistic crack sizes. Also note that only I
very deep cracks (>90 percent of the wall thickness) will even lead to leak at the lower pressures.
Leak Rate Models l Two models for leak rate were needed; one for stable leaks and another for ruptures. The leak rate model used for stable leaks was based on a leak rate estimation computer program written by PNL and described in Reference 1.
i 1 _. _ _
Results from that computer progran for various crack lengths were fit to an exponential equation which was used to estimate the leak rate for steam generator tube leaks. Figure 4 plots the predicted flow rate versus crack length for the computer program results, and the exponential fit to those data. The model does not account for the close proximity of the tube support plate to the crack, which would tend to restrict the flow. The degree of conservatism introduced by this model could not be quantified at tills time.
However, for relatively low flow rates, the effect is probably small, but could become appreciable for large flow rates. Other factors that contribute to the computer program predicting leak rates larger than the eessured values are discussed in Reference 1.
The model for tube ruptures is essentially the same as that used for tube leaks. However, the crack opening area model was modified to account for the restraint that would be imposed by the tube support plate. Because of the small clearance between the tube and the tube support plate, the plate would restrict crack opening under tube rupture conditions. The rupture flow model accounts for this effect by limiting the crack opening displacement to n(Dia..,. - Dia.,,.), or approximately 0.06 inches. The crack opening is then taken as a rectangle with area given by crack length times crack opening (0.06 x length). The PNL computer program was modified to incorporate this crack opening area model, and the results for various crack lengths were fit to a straight line. The computer program results and the resulting curve fit are shown in Figure 5.
As with the leak flow model, and except for limiting the crack o>ening, the rupture flow model does not further restrict the flow based on tie proximity of the tube sup> ort slate. Depending on the crack size, and the associated distortion of t1e tu>e within the tube support plate due to the crack osening, this effect could be significant. However, to estimate the degree to witch the tube support plate could restrict flow in a more rigorous manner was beyond the scope of this evaluation.
Flaw Size Distribution The approach taken in this evaluation was to determine flaw depth and length information from the destructive examination of the tubes removed from the Trojan steam generators. This information was then used in determining flaw depth and length distributions used in the leak rate evaluations. It was assumed that the flaw depths were uniformly distributed from 0 to 100% of the wall thickness. As discussed below, the flaw length data were fit to a 2-parameter Weibull distribution.
The Westinghouse PROPRIETARY report submitted by Portland General Electric in support of operation of the Trojan plant for cycle 14 (Ref. 5) contained flaw depth and length information, as well as tube burst test results, for tubes removed from the Trojan steam generators at the end of cycle 13. The Westinghouse report provided 'aounding' flaw sizes, as well as information on sizes for the micro-cracks that made up the bounding flaws.
Two concerns were identified in using the >ulled-tube data. First, there was l a concern that the pulled-tube data would )las the distributions toward larger l crack sizes. However, the data reflected destructive examination of three i
1
tube suppo'rt plate locations for each tube, with varying degrees of cracking at each location. Thus, the crack size distributions determined from there data are judged to be reasonable for cracked tubes, and are judged to be a random sample of the tube-to-tube support plate locations in the Trojan steam generators.
The second concern was that the data sample was too small to provide a reasonable crack length distribution. However, the Weibull distribution appears to result in a good fit to the data, with reasonably narrow confidence bounds.
Using the failure models discussed above and the burst test results, it was determined that the ' micro-crack' sizes reported in the WCAP report were more realistic than the bounding crack sizes in terms of predicting tube integrity and the potential leak rates. However, those crack sizes had to be adjusted to account for the additional SCC crack gmwth anticipated during operation.
Stress corrosion crack growth rate data specific to the Trojan steam generator environment and tubing material were not available. However SCC growth rate data for mill-annealed Alloy 600 in a NaOH environment (4 g/l concentration) were available in Figure A.9 of NUREG/CR-5117 (Ref. 2). The crack growth rate curves in Figure A.9 are reproduced here in Figure 6. Note that data for a 100 g/l NaOH solution are also presented in the figure. However, it was judged that this environment was unrealistically severe, and the data produced in the 4 g/l solution were used in lieu of more detailed information concerning the Trojan application.
To use the SCC growth rate data, it is necessary to estimate the stress intensity f actor, Ki , for the crack sizes of interest. For the range of crack sizes being considered and for the normal operation pressure loading, the K 's for a crack 0.5 inches long and 50% of the wall thickness deep are below 20 HPadm (-18 ksidin) implying growth rates of approximately 0.05 m/hr,or less. The K, was estimated using equations for axial flaws in pipe and reported in NUREG/CR-1319 (Ref. 8).
At each of the crack locations being considered there were several micro-cracks and each of these cracks could be propagating. Since it was not possible to perform a detailed analysis of each of the micro-cracks, it was judged appropriate to increase the growth rate for the dominant crack by a f actor of 2 to approximately account f or the potential growth of the adjacent micro-cracks. Thus, the growth rate used was 0.1 pm/hr.
The size of each crack was increased by adding a length given by 2
- growth-rate
- time.
The factor of 2 accounts for growth of each crack tip, and the time is the number of effective full-power hours during cycle 14 For the 12 month cycle 14, the number of hours was 8640. As noted earlier, the resulting distribution of crack lengths was fit to a 2-parameter Weibull distribution.
The remaining parameter needed to complete the evaluation is the number of potential flaw locations to be considered. In determining this number it was 8
-- , , . - , , , . , - , , - - n-, - - -
noted that the Weibull distribution effectively accounts for the probability that certain tube-to-tube support plate locations may have very small saximum crack lengths; the depth distribution takes into account the fact that some locations may have only very shallow cracks. Therefore, the number of locations to be considered is the number of tube-to-tube sup> ort plate intersections. The evaluation was restricted to the first tiree tube support plates on the hot leg side of the steam generator. Thus, the actual number of locations to be considered is thus given by (3
- no, tubes) - (3
- no. plugged tubes) - (no. sleeved locations). ,
for the 'D' steam generttor at Trojan, this number is approximately 7900. It should be noted that the evaluation results would not be significantly affected by' choosing a different steam generator at Trojan.
- 3. Estimation of LR, As noted in Section 1, the increase in leak rate under MSLB conditions due to cracks that were leaking under normal operation (LR,) is estimated as 10 times the leak rate under normal operation. This estimate is based on the leak rate model and associated test data resorted in NUREG/CR-2336 (Ref.1). However, arguments have been put forward t1at the factor of 10 is non-conservative, and that the leak rate increase cannot be estimated as a simple factor times the normal operation leakage (see Ref. 6).
Nonetheless, upon review of the argument for a larger factor of increase, it was judged that the factor of 10 does provide a reasonable estimate of the leakage increase. This judgement is based on three observations.
(1) PNL Test Data -- Some of the test data reported in NUREG/CR-2336 (Ref.1) for low flow rates (-0.1 gpm) are likely to be incorrect in that the flow measuring device had a lower sensitivity on the order of 0.1 gpm so the test results for those tests are artificially high.
However, other test data did not suffer from this problem, and they also are adequately represented by the leak rate model. Thus, based on the laboratory data, the leak rate model is capable of estimating the leak rate increase under HSLB conditions.
(2) Pulled-Tube Leak Rate Data and Industry Test Data -- None of the tubes removed from Trojan had leaking cracks. There have been a few tubes removed from other plants that contained through-wall cracks, and the test data from those cracks indicate increase in leak rate in excess of the factor of 10. In all cases the leak rates increased from extremely small values to values less than 1 gpm. Thus, even though the increase was greater than a factor of 10, the absolute leak rate was still a very small number.
Other laboratory test data indicate increase in leak rate going from normal operation pressure to HSLB pressure that is generally bounded by a factor of 10.
(3) Theeretical Hodel -- A theoretical model to estimate leak rate through steam generator tube cracks has been put forward by Belgian engineers,
and is discussed in EPRI NP-6864-L, Rev.1 (Ref. 7). The codel is germane to relatively small, tight cracks. For those cracks, the leak rate increase could be greater than the facter of 10. However, the leak rate would increase from a very small value to a larger but still sen11 value, which is qualitatively consistent with the test data from the pulled-tubes noted above. But, as noted, the leak rate under MSLB conditions would likely remain a small value.
As noted earlier the leak rate under normal operation conditions is taken as
. This leak rate could be the the permissible cumulative leakfrom leak rate rate many (0.09 small gpm for Trojan)d cracks, an if so the increase could be greater than the f actor of 10. However, based on engineering judgement and on service experience which ho not shown many small leaking cracks in a single generator, the possibility of many swall leaking cracks is judged to be unlikely. Thus, for the purposes of this evaluation, the factor of 10 increase is judged ta .be a reasonable estimate.
- 4. Results The models and analysis enethods discussed in Section 2 were combined into a computer program for estimating the " expected" value of leak rate for the Trojan 'D' steam generator. it was judged that these results would be either representative or bounding for the other Trojan steam generators.
The estimation of LR, is given by J.
LR3 = N p, (1) f (l) T(w (J) dl where 1, is the minimum crack size considered (.15 in); 1 is the maximum crack length for stable leak (.83 inch); p,(1) is the probability, for a crack of length 1, that the crack depth exceeds the critical depth for the crack to propagate through the tube wall given a main steam line break; f(1) is the probability density function for the crack length, assumed to be a Weibull fuction; R,,,,(1) is the leak rate as a function of the crack length; LR, is the mean, or expected, leak rate from stable cracks; and N is the number of tube-to-tube support plate locations to be considered.
The crack depths are assumed independent of the crack lengths, and uniformly distributed from 0% to 100% through wall. Thus, for example, the critical depth for a crack of length .S3 inch is .822 of the tube wall thickness, and the probability that a crack of length .83 inch exceeds the critical depth is taken as 1.0 - 0.822 0.178.
The formula for LR, is analogous:
where 1, is an upper cutoff for the integral, beyond which it is negligible.
(taken as 1.4 inch), and R,,,,,,,(1) is the rupture flow rate for a crack of length 1.
I l
J, LR=3 N p, (1) f (J ) R,g,(1) dl ,
Four different f(1) functions were used, corresponding to different crack growth rates.
The evaluation considered four pressure differentials: 2600 psi, 1800 psi, 1600 psi, and 1200 psi. The 2600 psi case was taken as the base case, but, as discussed in Enclosure 3, 1200 psi is believed to represent a 'best-estimate
- pressure differential if operator actions are considered. Figure 7 and Table 1 provide the results of the evaluation.
For the conditions and models used in this evaluation, at 2600 psi pressure is essentially differential the expected value of LR is 144 gpm, and LR, the pressure zero. However,consideringappropriakeo)eratoractions, differential is a> proximately 1200 psi, witch gives an expected value of LR of 13 gpm, with L h being zero. As discussed in Section 3, LR is estimated to be 10 times the permissible leak rate limit for Trojan, or ,10
- 0.09 0.9 gpm. Thus, the expected leak rate under MSLB conditions, LR,.. is 145 gpm at the conclusion of cycle 14 for a pressure differential of 2600 psi, and 14 gpm for a pressure differential of 1200 psi. Smaller leak rates would be expected at earlier points in the cycle.
- 5. Evaluation of Sensitivity and Uncertainty for Leak Rate Calculations l,ncertainties in engineering analyses are comonly attributed to:
(1) Physical randomness in system parameters; (2) Statistical uncertainty due to the use of limited information to establish the characteristics of these parameters; (3) Model uncertainties that are due to simplifying asst.mptions in the analytical and prediction models, simplified methods and idealized representations of reality; (4) Vagueness in the definition of certain parameters; and (5) Ambiguity and vagueness in defining the relationships between the parameters of the problems, especially for complex systems. >
The focus for this analysis will primarily be on the first three aspects listed above. A two step approach is taken for the analysis:
(1) Identification of critical parameters for the uncertainty analysis; (2) Qualitative evaluation (tendency, e.g., conservative or non-conservative) of uncertainty;
This section combines steps (1) and (2) for identifying and qualifying the
- uncertainties in the leak rate analysis. Quantification is beyond the scope of the present investigation and is the subject of a longer tern evaluation of '
leak rate predictive methodologies by RES.
Identification and qualitative evaluation of critical parameters: 1 (1) riaw Distribution and Crack Growth Rates ,
The >rimary variable affecting the results is the flaw length distribution, and )y association the flaw growth rate. The flaw distribution was taken as ;
the population of flaws from the 1991 pulled tube examinations that existed at the end of cycle 13 (Ref. 5). As described previously in Section 2, two concerns were identified in using the pulled-tube data. First, there was a concern that the pulled-tube data would bias the distributions toward larger crack sizes. However, the data reflected destructive examination of three tube support plate locations for each tube, with varying degrees of cracking at each location. Thus, the crack size distributions determined from these data are judged to be reasonable for cracked tubes, and are judged to be a random sample of the tube to-tube sup> ort >1ste locations in the Trojan steam generators. The second concern was t1at tie data sample was too small to provide a reasonable crack length distribution. However, the Weibull '
distribution appears to result in a good fit to the data, with reasonably narrow confidence bounds.
An additional concern regarding the distribution of the crack sizes was the fact that the data did not support a definitive relationship between growth in length versus depth. Therefore, for pur>oses of the Weibull fit, the length and depth distributions were assumed to ae independent. However, from a fracture mechanics point of view, once the cracks have initiated, the length and depth growth rates for the *microcracks' should be specifically related to each other. The complicating effect of the microcracks linking to fom .
macrocracks adds further uncertainty to the process. Quantifying this uncertainty is beyond the scope of this investigation. However, the simplified analysis described below attempts to provide an somewhat conservative estimate of the overall growth rate.
Figure 7 illustrates the sensitivity of the expected leak rate to an increasing flaw length distribution under the condition of MSLB for several different differential pressures. In Figure 7, the 'no growth' condition reflects the length distribution without adjustment for crack growth during cycle 14. The remaining conditions reflect increases in flaw size due to y' crack growth rates of 0.01, 0.05, 0.1, and-0.5 m/hr, respectively. The "s
' bounding' case reflects the distribution of bounding flaw lengths reported in Ref. 5. The sensitivity to the flaw size distribution is due to a larger proportion of flaws reaching a size that would either leak or rupture under MSLB conditions.
Given the extreme sensitivity of the expected leak rate to the flaw size distribution, it becomes important to examine the judgement that a growth rate of 0.1 m/hr is a reasonable estimate to account for the linking of several smaller flaws.
i e r ,,-..en n ,--,en..,~.v.,--,,-,,w,----- ,---..m,-c m,n,. wn~m.+ w . cn ,~-,-e- ,,---r ,-,m... , n. wn,--*,- .-w ,-.-.~m-mm,
o .
The original basis for the 0.1 m/hr growth rate was discussed in Section 2.
This value is supported by evaluations of inservice inspection data where flaw size increases were evaluated (see Ref.10). That evaluation suggests a through-thickness growth rate of 0.048 pm/hr, which gives a crack length growth rate of 0.096 m/hr.
The most compelling argument for the growth rate comes from service experience at Trojan. The mean crack length deduced from the examinations of the pulled tubes is 0.165 in., and the longest crack was 0.35 in. (Ref. 5). To determine an average growth rate requires an estimate of the period of time during which the cracks initiated and propagated.
Reference 9 stated that metallography performed on a TSP intersection from a Trojan steam generator tube in 1986 showed no corrosion attack. However, there had been an upset in secondary water chemistry during Cycle 8 (July 1985 through April 1986), which suggests a possible initiation time for the ODSCC cracking. Taking the end of cycle 13 (1991) as the evaluation point for the extent of the cracking, this gives a total of 5 years for initiation and growth of the cracks. The best estimate growth rate is determined by dividing the mean crack length by the 5 year operating time, ignoring any down-time during that period. This yields a length growth rate of 0.033 inches / year (0.096 m/hr), which is very similar to the growth rate deduced from the inservice inspection data.
To evaluate the uncertainty in the growth rate, the service data were examined looking for another operational occurrence that could have resulted in the crack initiation and growth. Based on operational data it was determined that the secondary chemistry was adjusted in 1988 resulting in a caustic secondary environment. The uncertainty in growth rate then can be estimated by dividing the larger crack length (0.35 in.) by the shorter period of operation starting in 1988 and ending in 1991. Ignoring any down-time, this gives a growth rate of 0.33 pm/hr.
Sustained growth rates appreciably higher than this value are judged to be unlikely simply because they would be expected to have resulted in leaks during service. While it is possible that through-wall cracks could develop without resulting is significant leakage owing to the oxides and corrosion products in the cracks and crevice between the tube and tube support plate, it is unlikely that all such cracks would be blocked; a few leaking cracks would be expected to occur if the high growth rates existed.
The uncertainty in growth rates can be expressed as an uncertainty in leak rate. Table 1 and Figure 7 present the leak rate results for the lower extreme growth rate, being no growth, and the best estimate growth rate. The leak rate for the 0.33 m/hr growth rate was estimated, based on the detailed results for the other growth rates, to be 1350 gpm. Thus, the expected leak rate would lie in the range from 33 gpm to 1350 ppm, with the best estimate leak rate being 145 gpm, assuming a MSLB pressure differential of 2600 psi.
Lower pressure differentials would result in lower expected leak rates as illustrated in figure 7.
O e
(2) Leak Rate Modeling The leak rate model used for this analysis accounted for both-leakage and -
potential rupture events which are-partially constrained by the presence of the tube support plate (see Section 2). The leakage estimations from this model do not take into account specific sources of uncertainty (Ref. 2) resulting from:
o #-d1 Crack Openings - The pressures and crack lengths of interest are such that the expected pressure induced openings of the cracks are -,
typically on the order of a few thousandths of an inch. Hence, second order factors can come to govern the leakage through the crack.
o Residual Stresses - Residual stresses from fabrication of the tubes may cause 'the cracks to remain closed until some threshold pressure needed to initiate crack opening is applied, in other cases, the residual stresses may cause the cracks to stay open even when the internt1 pressure is zero, o Crack Plugging - Small amounts of deposits could_readily restrict the flow through narrow crack openings, Deoosits could originate from impurities in the water, or the formatun of corrosion products on the faces of the cracks, o Crack Roughness - The leak rate model includes a term to predict the etfects of rough crack surfaces on the leakage through the narrow slit of an open crack. Coefficients for this contribution can only be roughly estimated for the ODSCC cracks found at Trojan. On a micro-scale, the ODSCC cracks are intergranular and therefore relatively
' rough" as compared _to a smooth surface. The actual roughness is subject-to consider &ble variation and could cause unexpected variations in the leak rate.
o Crack Area - The leak rate model assumes a simple through-wall flaw of given dimensions. However, the 00500 cracks found at Trojan are generally semi-elliptical in character, with the OD length typically much greater than 10 length for through-wall- penetration.
All of the above considerations, with the possible exception of residual stresses, add conservatism to the magnitude of the calculated leak rates, As noted above, residual stresses could act either to open or close the cracks.
Tube rupture leak rate predictioc are also deemed conservative due to contributions from two sources:
o Tube Support Plate- Restriction - The leak rate model was modified to account for the mechanical ' restraint" from the support plate that would be imparted te-cracks within the TSP region. -However, quantification of the reduction in the leakage that would result from the flow restriction in the TSP gap was beyond the scope of this investigation, o Tube OD Deposits at the' TSP - The modification of the leak rate model to allow crack opening within the TSP assumed a gap between the tube 'and the TSP of 0.010 inches. However, Ref. 9 reported that this gap more
- a m --
Itkely would range from 0.00 to 0.003 inches based on the presence of black, hard, tenacious deposits in the TSP crevices. These deposits would act to further restrict flow from leaking cracks at the TSP intersections.
(3) Other Sources of Leakage - Other sources of leakage are not treated in this analysis, e.g., leaks from pits, free span or tubesheet cracks. This is based on the fact that the primary form of degradation in the Trojan steam generator tubes is due to ODSCC confined to the TSP intersections or 'vtending .
slightly beyond the ISP's. This assumption obviously acts in a nor, conservative direction.
(4) MSLB Differential Pressures - This leak rate analysis was originally performed assuming a 2600 psi differential pressure across the steam generator tubes resulting from MSLB. This is approximately the pressure that was used by the vendor (Westinghouse) for MSLB in Ref. 5. Recent analyses by the Reactor and Plant Systems Branch (RPSB), RES, have shown the MSLB differential pressure to be a strong function of time into the transient and operator i action regarding interruption of high pressure safety injection flow. RPSB has shown that the MSLB differential pressure can be significantly lower than the 2600 psi head of the safety injection wn;s, provided appropriate operator action is taken early in the transient, kconpanying this pressure reduction would be a significant reduction in leak rou (see Figure 7). This reduction I is based on both the pressure driving force for leakage being significantly less and the critical crack length for rupture being longer than that calculated for 2600 psi. The RPSB analysis is described in Enclosure 3.
C. Sumary and Conclusions An evaluation methodology for estimating the leak rate for the Trojan steam generators was developed and applied. The methodology is based on flaw '
distributions deduced from destructive examination of tubes removed from the Trojan steam generators prior to the start of cycle 14. Other data nd analysis methods necessary to the analysis were obtained from the oun literature. The methodology is general but requires knowledge of tie flaw distribution that is specific to a steam generator at the beginning of the time period in question; at the beginning of a fuel cycle for example.
I The results of the evaluation for the Trojan 'D' steam generator indicate a range of leak rates under MSLB at 2600 psi from 33 gpm to 1350 gpm depending on the assumed crack growth rate. The best estimate leak rate for this condition is 145 gpm. The expected leak rate is strongly dependent on the flaw length distribution, and therefore is strongly dependent on the flaw growth rate used to adjust the beginning of cycle flaw distribution. A rationale for choosing the flaw growth rate is presented based on service experience.
l I
- f
- 7. References-(1)'- U.S. Nuclear Regulatory Comission, " Steam Generator Tube Integrity
' Program,-Phase II. Final Report,' NUREG/CR-2336,l August.1988.
(2) U.S. Nuclear _ Regulatory Comission, ." Steam Generator Tube: Integrity-Program / Steam Generator Group Project,- Final- Project Summary Report,"
NUREG/CR-5117, May 1990.-
(3) Kiefner,-J.F., Maxey, W.A...Eiber, R.J., and Duffe;r .A.R., " Failure Stress-Levels of Flaws in Pressurized Cylinders,' Proaress in Flaw -
Growth and Fracture Touchness-Testina. ASTM STP 536, American Society:
for Testing and Materials 1973, pp. 461-481.
(4) U.S. Nuclear Regulatory Comission, ' Steam Generator Tube Integrity; Program Phase I- Report," NUREG/CR-0718, September 1979. - -
(5) Westinghouse Energy Systems, " Trojan Nuclear Plant Steam Generator-Tube-Repair Criteria for Indications at Tube support Plates, WCAP-13129, Rev. 1,-~0ecember 1991, WESTINGHOUSE PROPRIETARY CLASS 2.
(6) R.P. Mcdonald, Alabama Power Company, " Alternate Steam Generator TubeL Repair Limits," Presentation.to the Advisory Cemittee on Reactor Safeguards, November 6, 1991.
(7)' Electric Power Research Institute,-"PWR Steam Generator Tube Repair Limits: Technical-Support Document for Expansion Zone PWSCC in Roll Transitions (Revision 1)," EPRi NP-6864-L, Rev. 1, December 1991, LICENSABLE MATERIAL.
(8) U.S. Nuclear Regulatory Comission,-" Cold Leg Integrity Evaluation, Final -Report," NUREG/CR-1319, February 1980.
(9) Electric Power Research Institute " Examination of Trojan Steam Generator Tubes, Volume 1:: Examination Results,'EPRI TR-1101427, Volume
- 1. November, 1992. LICENSABLE MATERIAL.
(10) C. Rau, "FaAA Overview TNP- Cycle ~ 14 RestartI- FaAA Project-SF16772,* L Letter to-W.F. Peabody Portland General Electric Company, December 14,s 1991. ,
i
Table 1 Results from Leak Rate Evaluation EXPECTED LEAK RATE (gpm1 A Press. No 0.01 0.05 0.10 0.50 Bounding Growth' m/hr ym/hr pm/hr #m/hr Size' 2600 33 36 57 145 8342 11782 1800 --
14 21 44- 945 --
1 1600 -- 9- 14 28 502 --
1200 -- 5 7 14 161 --
lid 1LTt 1 Flaw size distribution based on flaw sizes reported in Ref. 5 as ' max.
length at max. depth' 2 Flaw size distribution based on flaw sizes reported in Ref. 5 as the flaw size that bounds the microcracks J
B C
s , ~ <, _ . - -
i 4 FIGURE 1 EQUATIONS FOR FAILURE PRE 55URE AND LEAK RATE MODELS Throuch-Wall Cracks 5#
P" =
M, R, where, P,, is the pressure at which a through-wall crack would rupture
& is the flow stress (taken as 10 ksi)
R is the nean radius of the tube t is the tube wall thickness and N' = 1 + 1.255 - 0.0135 R, t R,' t '
with c as one-half the through crack length.
Surface Cracks P" =
5#
N, R, (Corr )
where 1-y= M, t *
.e5 1 .*
t
< o.
C "' - 0.0135 C "*
- ' = 1 + 1.255 R, t R,' t '
l
-'-2 e , , ,
,g Corr o -0.344226 + 1.72565 .a ,
3,gg,4-_ ,a b
t 4 h t 4 h
N with
,,nc-4 where a and c are the surface flaw depth and length, respectively, leak Rate -- Stable Leaks R,,,,(1) C, e"'
Pressure Differential C, C, l 2600 psi 0.0080654 10.195 1800 psi 0.0106676 7.3969 1600 psi 0'.0101051 -6.7644 1200 psi 0.0102778 -5.5519 Leak Rate -- Ruoture within Tube Sueoort Plate Applies only-to 2600 psi pressure _ differential. Ruptures are not predicted for lower pressure differentials for reasonable crack sizes. .
R,,,,,,, ( 1 ) = 56. 4 7 2 5 ' ( 7 )
i d
4 4
G.
1
--.v, ,,e.g--r- s e- .,,v , s. .~ - .- - , - ~
r ,
a\,
! 1 . ,cr
\
.t Ns !- i j T a
's 4)
I ?
\, '
lo j
N:
N g <3 i
! <:) j i
- eTl v o, .
4 f
s N <] >.
e u- ;
e 3
s - p -
s '
e
- e $i- I 4Nq <d g A +!' g-gct
<1 5 s p' e ss 6,A . u) c
- =
- ' D.
L- ,
N_ a r- a
!\ 00 '-
'a I s i\ r t N
s i e P oij "
x-83m - ~
n
,m oo<1 e-
.s
- d O u) -i!L >
l \
2 m C
\ g i ss ow-
- l. !
j \, \ _
Q 130
- i x a ::E -
c
- s..,
f m I s t t t t '\
8 8 o.
o-o.
co to e cv .
(!Sd) eJnSSeJd 1SJng pelo!pe>d e
S-
_g n .
-o a 0
o^
e 4,000 E o ,
TWC Cdtical Length = 0.83 in.
- o 2 0
- s e
o o MSLB Press. = 2000 psi g 3,000 , - y. ..
._ .N 5 ........9.....
e . . . . . . . . . . . . . . .......... ..... .. ...........
j hvemv%
E 2,000 - - - - - - - - - - - - - -
N .- ----.. -
% %rrr....,,,
, , , , , , ,;,n 3,a i i i i i i 0.2 0.4 0.6 0.8 1 1.2 1.4 Flaw Length (in.)
TWC aA=0.822 aA = 0.9 6 O figure 3(a) Application of Failure Pressure Model to Trojan Steam Generator Tubes Under MSLB Conditions 4,000 0 0
3,503 -
O O
Ta 3,000 g O 2600 psi o
$u C O
.........0..................\
9 .... ....................................
D 2,500 -O m 1800 psi M O c 1600 Psi O
O'. 2,000 -
on o -
Ei A / N ig1,soo u.
* ~
/
1200 psi ,2
.,,,,,,,,,,'I 3 I Il l l llE Ifli, i i i i i i i 300 0.2 0.4 0.6 0.8 1 1.2 1.4 Flaw Length (in.) ,
'iWC aA=0.9 aA= 0.95 O O Figure 3(b) Effect of Differential Pressure on Flaw Size That Would Leak Under MSLB Conditions
}
1 MO
-- . D _ __ .
500 m
fQ
. . . - - . . _ . . _ . . . - . - - . . . . . ~ . .
O. .
v CD .
G ey . - . . , . . . . . . _
(U T _
3 '
O 200 -- --
li.
100 -- - -- - - - - - - - - - - - - - - - -
rh r rh r' --
I I I I O m v a 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 Through-Crack Length (in.)
Figure 4 Flow Rate vs. Crbck length for Cracks in Tube Free Span (R,,,,(1))
IJi.J m -
1 l
I -
90 80 - - - - - - --- - - - - - - - -
^ 70 --- -- -- - ---- - - -
E -
O_
CD v 60 -
e e -
(G -
CC 50 - - - - - - - - - - --- - - - - - - -
3: -
_o_
l.L 40 - -- - -=
30 - --- -- - - - - - - - - - - - - - -- - - - - - - - - - - - - - -
I ' ' ' I 20 0.4 0.6 0.8 1 1.2 1.4 1.6 Through-Crack Length (in.)
Figure 5 Flow Rate vs. Crack length for Ruptures at Tube Support Plate Intersections (R.,,,,,,(1))
\'
A A da A da g Smt) - U mt) 10 to -
Alloy 600 MA -NaCH 100 91
'i l
Afloy 600 MA- NaOH 4 01 ANoy 690 MA-NaOH 100 $
q -
Alloy 600 HT-NaOH 100 91 1
..,*** ....~~-~ Alloy 600 lHT-NaOH 4M
- ,.** ...--~ ,
I ,.
- , Apoy 690 HT.NaOH 100 M
,. J 0.1 .
0.1 -
- ;l .:
- -AAoy690 HT.NeOH 1 a , , ,
m 4M'
- ! m 0 10 20 30'40 50 60 Kg (MPa G) 0 10 20 30.40 50 60 Kg (MPs G)
MA: Mill Annesled HT: Heat Treated 18h at 700'C 38712170.5M 4
~
Figure 6 Stress Corrosion Tests in Deaerated Sodium Hydroxide at 350*C on Fracture Mechantes Type Specimens -
see Figure A.9 in Reference 2 s~
4 --e
O Leak Rate (gpm) ~
I A 117a2gf342g .
6 1
i-r ;
il%?ASW, 2 ..
, 502 .
hh R
/ ,, . . . :?fe liiiii
/ t,
li !L* M -- -
t 200 / ;
$6{ m. .
ih6 g gg;;: *..,
~
p '
h5~ !!!
$!! [I!L .
/
{i hY . u;q h f$ f iNM!?ih[ "ljff f lii j l >[ .
ll ;$ _ ((lli.3%
/ j J IE
~
/- .
i
, a . -
c :f.f N!lf l}fg-fth!
N!hy' h'i : 2600 f IS.NM.,cu,;f{il51j.{N~.s hjg;
,/ [ .j
? U. hs-
/ q q t; ; }, &%uwt - , .y.,g tu -
1800 j/ i
,,/
50 ;
.); i-
- 1600 1200 Bounding 0.50 0.10 0.05 0.01 No Growth I
Figure 7 Expected Leak Rate for Evaluation Cases
ENCLOSURE 2 FREQUENCY OF NON-ISOLABLE LOSS OF SECONDARY SIDE INTEGRITY AND ASSOCIATED CORE DAMAGE FREQUENCY The ever.t of concern here is a non-isolable loss of secondary side integrity which leads to essentially complete depressurization of the secondary side, leading to a large pressure differential of approximately 2600 psi across the steam generator tubes. Moreover, the non-isolable loss of secondary side integrity must be outside of containment, so that a possible core damage sequence would provide a radioactive release path which bypasses containment, and so that any possible loss of primary coolant through steam generator tube leaks / ruptures would be to the atmosphere, and not to the containment sump, where it could be recirculated to the primary. The event of concern may be decomposed into the following three events:
E,: A main steam line break (MSLB) between the containment and the main steam isolation valves (MSIVs)
E,: A stuck open secondary safety / relief valve leading to more or less compiete depressurization of the secondary E: 3 A MSLB downstream of the MSivs, coupled with an MSIV failure.
The frequency of E,, a MSLB between the containment and the MSIV, is estimated in the prioritization memorandum for GI-163 (Heltemes to Gillespie, Sept. 28, 1992), as 7E-4 per reactor year, based on an unpublished estimate using an expert judgement approach. This appears to be a reasonable estimate.
An attempt was made to compare this value to the value given in PRAs. One difficulty encountered is that PRAs generally distinguish between isolable and non-isolable MSLBs, but do not distinguish non-isolable MSLBs which are inside containment from those which are outside containment but upstream of the MSIVs. For comparison, however, the Diablo Canyon PRA (PL&G study), South Texas PRA (PL&G study), and the German Risk Study for Biblis give the following mean frequencies for a non-isolable MSLB:
! South Texas PRA: 4.65E-4/ reactor-year Diablo Canyon PRA: 4.65E-4/ reactor-year l
1 Biblis PRA: 1.6E-4/ reactor-year The fact that PL&G performed both the South Texas PRA and the Diablo Canyon PRA accounts for the same value used in both these studies. Since at least a portion of this frequency corresponds to MSLBs inside containment, it is clear that estimates of the frequency of E, from thcse values would be somewhat lower than that used in the prioritization, but an estimate will not be attempted, since it would require estimating the relative fraction of breaks inside containment from those outside, but up to the MSIV. One should note however that the studies tend to identify these frequencies as MSLBs inside l
l l
containment, but it is clear from the context in which they are used that they include the portion of the pipe outside containment and up to the MSIV. The German Risk Study does give an estimate of the frequency of a MSLB between the containment and the MSIVs. This estimate is IE-7 per year. However, it is not applicable to Trojan. In 1985, the Biblis plant replaced the piping between the containment and MSiv by 20MnMcNi 55 and 15 MnNi 63, and the estimate of MSLB frequency is for this new piping.
In any event, the frequency estimates for a non-isolable MSLB are predominantly for breaks inside containment, so that they overestimate the -
frequency for a break between the containment and the MSIVs.
The event E,, a stuck-open secondary safety / relief valve leading to nearly complete depressurization of the secondary system has already occurred.
NUREG-0844 notes, on p. 3-13, that a stuck-open safety valve at Davis Besse Unit 1 in March 1984 led to a complete blowdown of the affected steam generator and increased the pressure differential across the tubes from an initial value of 1300 psi to a maximum value of 2220 psi.
NUREG-0844 estimates lE-2 per reactor-year as the frequency of the event E,,
at B&W plants; however, at Westinghouse plants the estimate is an order of magnitude lower because of the reduced frequency of challenges to the secondary safety valves in Westinghouse plants, as compared to B&W plants.
We note that on September 25, 1984, there was a precursor to the event E, at Trojan. A main steam safety valve lifted and stuck open. However, it reseated when the main pressure reached 890 psig.
As for the event E , the Diablo Canyon PRA (see Table 6-33 of the PRA),
3 estimates the frequency of a MSLB outside containment (i.e., downstream of the MSIVs) at 6E-3 per reactor-year. Taking IE-2 per demand as an estimate of the failure probability of an MSIV to ciose, one obtains 6E-5/yr as an estimate of the frequency of event E,.
Of the three contributors to the event of interest, E, appears to be the dominant contributor, with a frequency of IE-3 per year. The sum of the three events is estimated roughly to have a frequency of 1.5E-3 per year, accounting for what appears to be a somewhat conservative bias in the estimate 7E-4/yr for event E i .
CONDITIONAL PROBABILITY Of CORE DAMAGi GIVEN LOSS OF SECONDARY-SIDE INTEGRITY We consider here a non-isolable loss of secondary side integrity, outside of containment. The RWST capacity is 428,000 gallons. Our best estimate leak rate is 146 gallons if the pressure differential across the steam generator tubes reaches 2600 psi. However, it is unlikely that the pressure differential would be much above the pressure differential for normal operation, if proper operator action is taken. Consider first the case of a 145 gpm leak corresponding to a 2600 psi pressure differential across the steam generator tubes. Even with no credit for depressurization of the primary, the time for depletion of the RWST is about 49 hours5.671296e-4 days <br />0.0136 hours <br />8.101852e-5 weeks <br />1.86445e-5 months <br />. With depressurization, the leak rate will be considerably less. The NUREG/CR-4550 m
_ . _ . _ . _ _ . _ _ . _ __ ._ _ .. _.- _ _~ _. _ _ _ ._
d _.4. h study- for/Sequoyah,-performed in support:cf NUREG-ll50, states -(NUREG/CR-4550,.
vol. 5, Rev.1,5Part 1, 4.4-26) that the -likelihood =of not being able to -
Emitigate a 100-opm leakiwhere the RWST depletion timetis about.50 hour5.787037e-4 days <br />0.0139 hours <br />8.267196e-5 weeks <br />1.9025e-5 months <br />s-is
-negligibly- small. - Of course, with primary system depressurization the leak
' rate-can be madezconsiderablyf smaller allowing more time for recovery.: .
Recovery actions < include (1) cooling down and _depressurizing the primary to atmospheric -(2)'depressurizing to reduce the leak rate and refilling the i RWST, (3) closing a stuck-open MSIV, if this is .the.cause of. the loss of 1
-secondary side integrity,lor (4) possibly gagging a stuck-open: safety valve-while it:is blowing down,-although the severe environmental-conditions limit the possibility of success here.-
-In addition, the transient may be coupled with hardware failures, such as _ ECCS-failures or failure of the auxiliary feedwater system (AFWS). A rough overall estimate of 'the conditional probability of core damage given a loss of.
secondary side integrity is 7E-4. Human errors lmay_ contribute 3E-4, failure of the AFWS lE-4,- and other hardware failures,another 3E-4, leading to 7E-4.
t The core damage frequency from this sequence ,is therefore
- 1.5E-3/yr x 7E-4 = IE-6/yr.
If the peak pressure differential is lower than 2600 psi and'the corresponding leak rate is lower than~145 gpm, the estimated core. damage: frequency would be reduced, but there is a lower. limit, since even a MSLB without steam generator '
tube leaks has a conditional probability of core damage. However, the core damage frequency may be reduced by a factor of three or more, taking into _
account the fact that the-estimated leakage will be less' than 145 gpm, with '
. operator action to prevent the high pressure rise in the primary.
m
?
E o
{
A A
l'
. . . , w .- _ ,- - .~ ,+. . .,. .
I ENCLOSURE 3 EVALUATION OF RCS TEMPERATURE AND 6P ACROSS SG TUBES UNDER MSLB CONDITIONS As noted in the January 5,1993, memorandum from E. Beckjord to T Marley, MSLB calculations were to be performed in which ECCS (HPI and charging) and feedwater flow would be controlled in accordance with amergency operating procedures, in order to determine the maximum pressure expected due to repressurization. Upon review of previous analyses, it was decided that ,
additional break sizes should be explored to determine the effect of that parameter on the 4.0 across the SG tubes.
Three steamline break sizes were analyzed (1.4 ft', 0.6 ft', 0.15 ft') without operator action. The 1.4 f t' and 0.6 ft' cases were repeated with operator action. A typical Westinghouse RESAR plant was used. However, the ECCS was modified to include high capacity, high head charging pumps as is the case at some PWRs. This would further increase the possibility of higher AP across the SG tubes.
Figure 1.1 presents the primary and secondary pressures following a 1.4 ft' main steam line break. Following opening of the break, the faulted steam generator experiences a rapid depressurization. The secondary depressurization produces an accompanying reduction in secondary temperature as shown in Figure 1.2. This reduction in secondary temperature increases the energy extraction from the Reactor Coolant System. The increased energy removal rate causes the RCS to depressurize and a contraction of the RCS fluid volume which results in a complete loss of pressurizer level. Figure 1.3 presents the void fraction profile in the pressurizer showing a complete loss of level as indicated by the void fraction in Volume 8 approaching a value of 1.0 at about 200 seconds. During this time period the ECC and charging systems actuate and begin refilling the RCS with pressurizer level indication regained at about 300 seconds as shown in Figure 1.3 as the void fraction in Volume 8 begins to decrease below a value of 1.0 at this time. As the pressurizer continues to refill, with subcooling greater than 30' F and pressurizer level greater than 4% at 500 seconds, the RCS has achieved conditions which would allow the operator to throttle ECC and charging flow to pri. vent a continued repressurization of the RCS.
The 0.6 ft' main steam line break behaves similarly as the 1.4 ft' discussed above with the pressure, temperatures, and pressurizer responses illustrated in figures 2.1 through 2.3.
The 0.15 ft' break behaves the same except pressurizer level is never lost during the event. As such, once the RCS pressure begins to recover at 400 seconds in Figure 3.3, subcooling and level meet the minimum requirements to allow the operator to begin throttling injection flow to terminate the RCS pressure rise.
Figure 4 presents a summary of the primary-to-secondary pressure differential for the above spectrum of main steam line breaks. Also shown in Figure 4 are
the earliest operator action times which correspond to thr, earliest time in the event where RCS subcooling is at least 30' F and pressurizer level is greater than 4%. Note that as indicated in Figure 4, there is ap>roximately 15 to 20 minutes available for the operator to act to terminate tie RCS pressurization and limit the primary to secondary pressure differential to values less than 1800 psi for a broad range of main steam line breaks.
Figure 5.0 presents the core outlet temperature for these cases.
Table 1 is a sumary of events for tha five cases analyzed. Figure 6.0 is the RELAPS nodalization used.
' d' 5
- O k r P w g
, Sd ....
I
.....i.
e 0
(N
- 4*'
O ;
O - u O
O - i_ -
4 ; o %
O __ t.
o 10 i O i v- -
W . '_ -
8 ,,
i I -
t- .*
I b
' a
- I- >
I g I k i I
- :.lJ l
oO o . _. a::
oE
_ i_
i l " I-D t.
i C
8 O I" U O .
'_ i O -
E o -
l- .5 u
I O "
O _
._ O O , in 1
~
O -
LD ,
I -
Q. l t 4
- p 1
/ _
1 f _ -4 a *'I " ' ' ' ' '
e t t t I t 1 I i f O
e e e e e O O O O O r-
- - e ,- -
3 3 3 Y 3
a a u . O O O O O O d 'd d
d d 3
02nsse2d
( r. ) (w) 650 ,
sattemp- 192010000 ----- tempf-1920100001
, , , .. . 1a 600 -
- '2' i
~
1 i
a 550
- s n*
s y -
s -
. g - s -
g ~,
500 - -
s N %
450- - ~- _
^
35*
~___ '
400 ' ' ' ' ' ' '
O 500 1000 '
1500 2000
- Time . -
Figure 1.2 . RCS Subcooling.-vs Time' (1.4 .ftt)
I voidg-62OO70000 ----- voidg-62OO80000 7 voidg-62OO50000.' -----
voidg-62OO60000 .
voidg-62OO3OOOO -- ---
voidg 62OO40000 1.2 v idg-62OO 10000 ----
voidg-62OO2OOOO
. . i i i i . . . i i . . . i .. .. . i 1.0 , :. , ( -
\
/ \ i s g i s g \ g '. ,_
e '!'I I i \ s i _
3 0.8 l \ \\ \. i..
.U N k!ll' f
~!i.I '
!l e
\.
\ \
's
\
i
~
~
1 3 0.6 -lI ' I \ \\ \ \. -
s u
]!!!!
- l l
\s. \
\
\
\.'.
~
.- . ;! j . .I .\ \ s s -
O.4 4' l l \
\ N. -
. $ -iif f j \- \ i i
2 t \ s.,
- e. _-) -.,
$ \ '.\ s. _ ,
\ .'.s, a O.2 d). jl- \. \.
.s* , t. \ s. -
g g. i %
. t'I . ~
. 0.01- '------ -
Note: Volume 5 8..is located at bottos of. pressurizer ..
---0;2 ' ' ' '
'0 '500 1000 1500 .2000 Time -
Figure 1.3 Pressurizer - Level (void)' ' (1.4 ft') ,
.. . .. ,. n. -
c .. .
& W- 5 [ # ^
o i . . . . . . i i .- ..
10 l o - i.
O l O - i.
O _
i _ S~
O "
- T -
i to i
l_ e I -
i _
O- i i
I
- i _ e i
i e 8 >
- -i .
l o & ,W e_ o i
O :. E :.:
i " l-- 2 l~
~
i t.
~g 0 '
0 O O -
i - S o e
?
O Lf)
- i I-s O
e o --
_ e _
o- g O l D O -
e _ 1 t[) l- .n I -
l - !
g Q. ,
- , . 6 e -
f f f 9 f f 1 l t t tv
- I ~ t 1 I e e- e e e
?r_ O O O .O O
-- ,- - r
' ; .< .< ,. X O O O O O O O O O O d
eq l,-
6 l d e2nsse2d
m 650 , ,
tempf- 192010000 ----
sattemp-.192010000r4 vo -
e .
600 's -
~~,___- -
s,
- 're u .
3 .
.u _
.IV 550 - .
no s .
500 -
450-O 500 .1000 1500 2000 Time
-Figure 2.2 RCS Subcoolirig. vs Time (0. 6 ft )I - ,
...-i...
... ...i....i...i... .
N OnOn _.
O6O6 0000 _ _
OOOO 00 0 t N _ ..
OOOO ,-
OOOO
~ . .- -
NNNN
~
WWWW - ,,..- -
ca
-l l l l v OOOO -
~ 1
-V V O O .'
, u e.
'b'b'6'5 3
- - O t , ,
2.
- l I
, e _
F>y I i i ,
. E_ N-~.
g -
I-.-T
-OOOO _
' ~ l _
OOOO 0000 OOOO
.- J i
t I
s b LO n e OOOO OOOO NNNN
/
1 9
OOOO -
n Ec I l I l -
to +
DDQD ~ ~
OVVV
'6'6'6'8 l _
i .:
4-1 t
1 8
Q__,_ _
~ . __ _ ';
._~.-_-_-,--"----__!_
.J Ci;;::?"C ,"? ~i:FT'-i=vv W~ ~'
,,, 1 * O N O 00 W W N O .N d d' d d d d 6 d I.
uoT4De25 PToA .zazi.2nssoid
tg :.a..-
-w s.
e r h,. o i . . i ._ i i. i . . .. . . . .
- i _i N
~ ~
.O '
i O - _
O i O -
e {
r c O - ,' _ d O i o
e- m T - . .
L() e 2.
e I - _
O- '
I
- i .
e i i .
4 i ').
8 l - _
8oy -.
' O . .- a e
"F.t
, g- -
- o _
y O l 8 O - , -
-g O l '
O LO l
{
O O -g
-l .-
O O , tn O .
l _ T tf) ,
s l - , - u O- ,'
- t . . , , - he I
e I _
1 M I~ ' '
t t i t t t l'[t t l t t-O
.to to to to LD
.O- O O
-O .O r-
- i. '
X ':
M X
~
O O' O .O O O -
O -
O -
O,- O- -
v .l e - , .1 e v h .T"'" t" sansse2d e
= -- - - - - - - - - - - - - - - _ _ - - - - - -
I a) 620 tempf- 192010000 sottemp-192010000 su
"."{
__i___u i i
~~. .' _ ..
, 600 -
~
'm t
\ -
g S80 --
5'i 3 .
g _
g, .
8 560 -
5 :
~
540 -
5't 1
520 -
$6
~
500 ' ' ' ' ' ' ' ' ' ' '
O 500 1000 1500 2000 l
Time Figure 3.2 RCS Subcooling vs Time (0.15 ft) 2
~ - -
~
voidg-62OO70000 - - - - -
voidg-62OO80000 -
voidg-62OO50000 - - - - -
voidg-62OO60000 -
voidg-62OO3OOOO -- - -
voidg-62OO40000.-
1.2 , ,
voido-62OO
, ,s , ,
10000 , ,
voidg-62OO2OOOO 1.0 . -
s:
o % -
_i !
_. ,! \
s -
'! ! \ -
8 0.8 l-! ! \ -
Y -! i /'s 's -
g -i ! l N 4, -
u l 1 \ s -
O.6 -l I '
3 8 8 i
1
's g -
-s 8 i l I I \
~
_i \ -s 50.4 i -
4 / l \ 's. '
5 -i ! 1 \ ~.
~
3
" Ti 11/
/ I \
\
s
's
~
I -
! O.2 4
- \ l i \
\
's s.
's.
~
il I \
N ~
/
0.0 >t
\
. =+ - ~------------' - - - - -
~ ~
-0.2 ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' '
O 500 1000 1500 2000 Time Figure 3.3 Pressurizer Level (vold) (0.15 ftry
STEAM LINE BREAK }
l Primary-Secondary DELTA-P vs Time DELTA-P (Psi) 2400 -
200 -
c.<ii.ai i., meis.a vi . to.s rivi E.tli..t 00 nr.ter Act 10 9 Tim. (0.18 f t 2) 1600 -
l 1400 -
1000 800 -
l 600 -
400 - -
200 -
O O 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 l Time (Sec) (Thousands) l '
l DELTA P (1.4 ft2) DELTA-P (0.0 ff2) -*- DELTA-P (0.15 f f 2)
L For the 0.15 f t2 break, min subcooling &
PZR level was not lost. Earliest action time based on recovery of RCS pressure i
- Operator action to throttle injection based on min PZR level of 4% and RCS min subcooling of 30 F
,....y._, -
Figure 5.0 -
RCS Temperature vs Time CD -
tempf-192010000 (f) 600 , , . , , , , , ., , , , , , ,
,. , , , , , ; sta m
mi
~
550 -
gju 3 .
1.4 ft2 ._
2 _
g .
g .
" 500 -
.._ yya o.s rt ein seS l
k' h h l l l g 'g -
0 500 1000 1500 2000
- Time
. .. - _ . - - ___ = -
,t ,lil!!j ;ll , ; ,! l! t! :!r ii Jl : i lf i I 1
[
- - - , "* e s-c i
m s e
- s e
e
,:. 1
' : 3, - ___m
. ~- _ . .
s- .
' ~
'I':'
.'* ' 2 - _= t _^
6 _
6 9
l ;
3 I :\,% }
2 -
[
- _ I s
h,M -
3 n'
2 s_ o n
3 ' -
2 Y -
y n v#
- q.3 3
a .a t 1 i
2 m*** -
r
~3 e, ' ;
I I 3 .
l
- . [
a
. l .
I
,
- 4 _
R s a
- A r ;
Sv Eo v,
- 1 ; egg _
Rs , m ,P E * *
~
S *
- U *
- t O 1 1
- N
. _ -e s
H .
G "s
NI T w : . : . . .
M' S '
E -
W 2
S 7
~ ~
R -
- }
1 '
O ~ ah-F _
mR~
1 1 ' 3 3 2 n M ' ' i B
i 5 r t 5 a _
A ' ' . 3 .
h4E o,
Rv lv wa ' i G G o. "
E D Ar I
@ sWs '
v M
R #
U N u .o l
- 3 .
, gwg(Lg s.
GO su
[ ; ment of criteria for qualification of inspectors, procedures and equipment tirough performance demonstration. RES is updating Regulatory Guide 1.83 to provide these criteria. Pulling, examining and bursting of tubes may not be desirable in all cases. As a potential alternative, appropriate application of improved eddy current technology, enabling more accurate measurements of crack length and depth, i could minimize the dependence on destructive tube examinations. RES also intends to conduct a thorough review of the NDE basis for U-tube steam generator tube inspection in order to arrive at an independent evaluation of I the technology in CY 1993.
The commercial eddy current inspection vendors are using the currently.
available state of the art equipment, probes and technology for inservice
. - , _c _ ~ . _ _ - _ _ .. . . _ . - . - -
l inspection of steam generators. However, the present technology does not provide a high degree of reliability in detecting and sizing all of the defects present in steam generators today. Most operating steam generators have chemical deposits and degradation that makes the inspection much more difficult today than the inspections perforsed on new 1enerators. The generators have experienced the formation of magnetite on the tubes and in the crevices, the depositt ' Nallic copper on the surface of the tubes and denting of the tubes ; f tMed regions. In addition, the newer defect mechanismt such as inc u and a intergranularattack(MX),nularstresscorrosioncracking(IGSCC)ducedbythe produce smaller signals than those pro earlier defect mechanisms, such as wastage and fretting, and are harder to detect and size.
Analysis of the eddy current signals acquired during a steam generator tube inspection involves ' mixing" of the components of tie electrical signal.
'Hixing" is defined as the combination of the components of the eddy current signal at each frequency used. The ' mixing" that is presently used for steam generator inspections is directed at removing the effect of only one property variation, such as the tube support signal, or denting, and then evaluating the defect using the older method of pinse analysis. Usually, linear combinations of the signals at only two frequencies (or at the most three) are used for ' mixing". However, since there are many property variations, an equivalent number of equations are needed to allow discrimination and evaluation of flaw signals. Use of several frequencies enables superior detection and sizing capabilities relative to simple (two frequency) ' mixing."
The ORNL work is improving the fit of multi-frequency eddy current readings to steam generator inspection results. This work has shown that a non-linear polynomial fit of the readings from four frequencies to the properties gives considerably better results than the simple ' mixing" in current use by the industry.
The most sensitive eddy current measurement now in use by the industry is the rotating pancake probe. Very little ' mixing
- is done with this probe, and the industry use is limited due to the slow inspection speeds of the probe (about 0.2 inches per second).
ORNL has developed and is testing a probe that will overcome many of these problems. This probe, being manufactured by a commercial vendor, will consist of a 16 coil array incorporating improved reflection-type coils. Measurements and calculations have shown this coil to be much more sensitive than present pancake coils (higher level signals are produced). The 16 coil array probe, coupled with a new data acquisition and processing system provides better sensitivity than the lancake probe at a speed comparable to the in-field inspection speed of tie bobbin probe (about 12 inches per second).
Due to the large number of property variations present in a steam generator inspection, several million measurements are needed in order to adequately " train" the coils in the probe. This ' training' involves using the new probe in inspections of samples and standards incorporating many flaw types, sizes and other inspection variabler and constructing a data base that allows the correlation of a large number of readings to test properties. The results achieved thus far using the improved analysis, probes, procedures and equipment indicate major improvements over currently employed techniques for cetecting and characterizing flaws.
I l
f* l l * .
1
' i The improved data acquisition and analysis equipment and the new 16 coil array !
probes have been ordered and are being manufactured by a major U.S. supplier <
of eddy current equipment and inspection services. Following receipt of the {
equipment we plan to conduct extensive laboratory testing and evaluation usingrealisticsamplescontainingactualcracksandotnerinservicetesting l conditions. Metallographic data from actual pulled tubes can be added to this 1 data base if adequate calibrations and readings are taken. Methods to l transfer the data acquired from the laberatory standards to the field in-line calibration standards are also being develo>ed, further, since this probe acquires much more data than the norsal bobain coil inspection, new methods of :
data storage and display will also be developed and tested. These activities !
are part of the laboratory evaluation in preparation for field validation of ;
this technology by conducting actual steam generator inservice inspections. .
Thenewequipmentand16coilprobesareexpectedtobereceivedatORNLby January, 1993. The present alans are to begin field validation testing of the probes and equipment in the rail of 1993. We expect that improvements will be needed based on experience with the actual ins'ervice inspections and a complete validation will involve several inservice inspections which should be conducted in FY 1994.
4 i
i i
,