ML031110344
| ML031110344 | |
| Person / Time | |
|---|---|
| Site: | Davis Besse  |
| Issue date: | 08/05/2002 |
| From: | Matthew Kirk Office of Nuclear Regulatory Research |
| To: | Nilesh Chokshi, Long S Office of Nuclear Reactor Regulation, Office of Nuclear Regulatory Research |
| References | |
| FOIA/PA-2003-0018 | |
| Download: ML031110344 (4) | |
Text
August 5, 2002 MEMORANDUM From:
Mark Kirk To:
Nilesh Chokshi Steve Long Re:
Status of ORNL Calculations of Burst Pressures for Cavities that are Larger than the Cavity Found at Davis Besse.
This memorandum provides the status of our analysis. Paul Williams (ORNL) is in the process of preparing a letter report that will contain all of the details; his report will be available next week.
- 1. The purpose of our analysis has been to assess how much larger the cavity in the Davis Besse head could have been before rupture of the cladding was likely at pressures at or near the operating pressure. Figure 1 shows the finite element model we have used in these calculations:
it features a global model of the head and a local model of the area near the wastage cavity.
This global/local approach allowed us to better refine the mesh in way of the cavity. In all of our analyses we have used
- a. A uniform cladding thickness of 0.24-in. (this was the minimum thickness reported by FENOC based on UT measurements made on a 1/2-in. square grid).
- b. Tensile properties at 6000F that represent a lower bound to all available 308 SS results available at this temperature.
- 2.
Figure 2 depicts the two patterns of wastage growth we have investigated. Our initial analyses used the "self similar" growth pattern. In performing these analyses we noted that the boundaries of the wastage cavity had expanded to interact with the boundaries of the local model. This caused us concern that the predicted burst pressures could be influenced by the modeling approach we had adopted. To help assess this we performed some calculations with a more idealized growth model (the ellipsoidal model) wherein we could better control the growth of the cavity, keeping it away from the boundaries of the local model.
- 3.
Failure pressures predicted using both growth models (see DRAFT ORNL Letter Report dated 7-15-02, "Stochastic Failure Model for the Davis-Besse RPV Head," by Williams and Bass for mode details on failure pressure prediction) are compared in Figure 3. Figure 3 also shows the relationship between failure pressure and area predicted for circular disks by a closed-form plasticity model developed in 1970 by Chakrabarty and Alexander (CA) (the error bars and error bounds are based on how well finite element analysis analysis was able to predict failure of burst disks reported by Riccardella, again see the DRAFT ORNL Letter report of 7-15-02 for further details). The close agreement between the burst pressures predicted for the two cavity geometries shown in Figure 2 and the closed-form CA formula suggests that cavity geometry is not a factor that exerts significant control on the burst pressure.
- 4.
Since cavity geometry does not appear to be a major variable, a first order appreciation for how much larger the cavity would have to be to rupture at pressures near the operating pressure can be obtained by thinking of these cavities as circles, as described in the following table. To express these results in terms of time they obviously need to be used together with an appropriate corrosion rate, and expressed in the context of an uncertainty analysis. It is my understanding Bill Cullen is currently in the process of establishing a contract with ANL to define corrosion growth rates. It is also relevant to point out that there is currently a disagreement
between ourselves and FENOC regarding the size of the original cavity (FENOC is using an initial area of 201/ in2). We are in the process of resolving this discrepancy.
Growth of radius Cavity Area Radius of an (from original cavity)
C[in]
equivalent circle [in]
needed to cause failure [in]
Original cavity 36 3.4 N/A Cavity with a 5% failure pressure at the set-point 165 7.2 3.9 pressure (2500 psi)
Cavity with a 5%
probability of rupture at 215 8.3 4.9 the operating pressure (2165 psi)
- 5. In Figure 4 we compare failure pressures predicted by SIA (FENOC's contractor) with the CA theoretical model. At least in terms of trends there is good agreement between SIA's analysis and the CA theory, and I think it is safe to say that SIA is predicting lower burst pressures than would our analysis under identical conditions. Clearly we could go into greater detail regarding why the SIA analysis is providing the answers that it is, and we are prepared to do that, if so requested.
- 6. It is, of course, always possible to do more analysis (if warranted). Some features of these analyses that could be improved (that is, made more realistic) include the following:
- a. The geometry of the original cavity could be made to agree with the dimensions taken from the dental impression made of the wastage cavity, once these dimensions become available.
- b. The dimensions and surface relief of the cladding could be made to agree with the dimensions taken from the dental impression made of the wastage cavity, once these dimensions become available. This improved cladding model could include modeling of the increased cladding thickness due to the J-groove weld (a feature which has currently been omitted)
- c. The stress-strain properties of the cladding could be made to agree with measurements made on the Davis Besse cladding, once these measurements become available.
- d. Nozzle 11, which becomes fully engulfed by the cavity by some of the large wastage areas, could be restrained against motion in the vertical direction (in our current model the nozzle 'floats" freely on the cladding) to better model the restraints imposed by structures above the RPV head. In the ORNL letter report of next week such an analysis will be reported. However we do not expect this change of boundary conditions to significantly change the outcome of this analysis, largely because the wastage area is already so large when nozzle 11 is corroded away from the RPV head.
Please let me know if you have any questions or concerns.
1 FIn i
Figure 1.
Finite element model used by ORNL.
Ellipsoidal Growth Pattems
,As
- 0
) 0.."""S) \\
Figure 2:
Wastage growth patterns: (left) self similar, (right) ellipsoidal.
Chakrabartv and Alexander (1970) Theory Compared to ORNL Results 100 CLad Thickness = 0 24 inches SS 308 bounding stress-strain curve 90 a80 z
7 0 6
6 0 r
F 50 u.
2 4 0 tL 30 95%
'.I
'As-Found' C
'(Center-li Self-Similar Growth i
j-E-As Found Ellipsoidal Growth
-- Operating Pressure Set Point i..... 5% (Theory) 95% Theory Chakrabarty TheoryJ ircular Diaphragms lembrane Failure Thery')
Error bars are 90% confidence intervals based on revised Log-Laplace stochastic model 20 10 00 IJ Cbakrabuny and 1 NM Alexander. 'Hydrostati. Builging of Cimrular Dsprzirgms."JAurn alof(Srain 4nahl1s. 5(3),(1970) 155-161 0
50 100 150 200 250 300 350 Footpnnt Area (in2) ptw 8-02-2002 Figure 3:
Burst pressures predicted from ORNL analysis.
Chakrabarty and Alexander (1970) Theory Compared to SIA Results 140 T Framatome SS308 Properties 95Vo c;
1084r
.. (ksi) 120 1
- CladThickness= 0 125 I a Clad Thickness = 0 297 I
Theory (t=O 125)
Theory (t=O 297)
5h (t=O 125)
...... 95% (t=0 125) i...-..5% {(1=0 297)
.95% (t1=O 297) xE
- 8 r
T 100 80-60 40-20-00
.4 J Chakraiubrty and J M Alxe'ndmrr
'l1drostatc Bulging of Cmtular Disphlgrn%,Jiiiriral of Sjrarii Aaiilih
.5(3) 11970) 155-161 Er bars wer 90% condence IUervas based on re-ised Log-Lallace stochastic model 00 200 400 600 Footprint Area (in) 800 1000 1200 Figure 4:
Comparison of SIA predicted burst pressures with theoretical predictions for circular disks.