ML20086A112

TMI-1 Steam Generator Linear Elastic Fracture Mechanics & Load Characterization, Technical Evaluation Rept

ML20086A112
Person / Time
Site:	Three Mile Island
Issue date:	09/28/1983
From:	Subudhi M Battelle Memorial Institute, PACIFIC NORTHWEST NATION
To:	NRC
Shared Package
ML19283C007	List: ML20086A104 ML20086A112
References
NUDOCS 8311140084
Download: ML20086A112 (6)

Text

. . ..

.~' -

h h D

THREE MILE ISLAND UNIT 1 STEAM ENERATOR LEFM AND LOAD CHARACTERIZATION TECHNIC AL EVALUATION REPORT M. Subudhi and P. Bezler INTRODU'CTION In November 1981, primary to secondary side ^1eaks'were d5scovered in both of the once-through stean generators (OTSG) at Three Mile Island Unit 1 (TMI 1). Subsequent inspections and failure analysis performed by the utility and their consultants revealed widespread circumferential intergranular stress corrosion cracking (SCC) on,the Inconel 600 tubes in the upper portions of both OTSG's.

• Extensive utility investigations canbined with laboratory investigations performed at Brookhaven National Laboratory (BNL) have confirmed that the integranular SCC resulted from the presence of both oxygen and metastable sulf ur compounds in the TMI-1 coolant. Prop.er future control of coolant water chemistry should prevent the reoccurence of SCC.

Although the root cause of attack was established and can-be correct'ed, the tubes in the OTSG experienced degradation. Thru wall cracks and part thru -

, wall cracks exist in many of the tubes in the region of the upper tube sheet (UTS). To bring the OTSG to an operable condition extensive inspections to -

l locate the defects and corrective actions to repair the system have been

pe rfo rmed. The corrective actions consist of tube plugging and stabiliza- *

[ tion of tubes showing unacceptable defects and explosive forming of tubes ,

showing' defects in the UTS. The repaired and remaining tubes still exhibit detectable defects. To support the contention that these tubes were serviceable a detailed. linear elastic fracture mechanics evaluation (LEFM) w'as performed to demonstrate that defects of permissible size are not

- susceptible to unstable failure.

~

8311140084 831103 - - .

CF ADOCK 05000289)

CF

1 h k_.

The present technicai evaluation. report is based on the' BNL review of GPUNC topical report 008, Rev. 3 (8/19/83), GPUNC technical design reports TDR-388, (5/11/83) and TDR-417 (9/19/83) and GPUNL consultant reports " Stress Intensity Factor and C0D calculations" and "The Residual Crack Opening Displacenents in Tubes and Pipes" by F. Erdcgan (Lehigh). The purpose of the review was to determine if the proper loadings were considered in the .LEFM studies and if these studies support the GPUNC contention that cracks of permissable size will not lead to unstable failure.

DISCUSSION GPUNC topical report 008, Rev. 3 (8/19/83)" ~ssessment of TMI-1 Plant Safety for return to service after Steam Generator Repair", provides a summary description of the GPUNC studies and asserts that the unrepaired portions of the OTSG tubes will satisfy the design criteria for the life of the plant.

The corrosion studies indicate that the cracking mechanism has been arrested and will not reactivate in low sulfur primary coolant water chemistry. The Analyses show that cracks below a minimum range of length and through wall thickness will not propagate to failure under the combined effects of flow induced vibration, thermal cycles, and mechanici loads. In addition, any

. defects that 'are large enough to propagate thru wall and are not detected during the 100% ECT inspection will be detected by leakage monitoring programs during the test program.

The loads considered in the LEFM studies include those due to differential thermal expansion of the shell and tubes during cooldown, pressure, tube

. pretension and fluid induced vibration bending loads. Tube axial tensile load is greatest during cooldown but can vary from compression to tension during .

operation. In the LEFM analysis a load cycle with a peak tensile load of 1107 lbs. corresponding to a 100*F/hr cooldown, taken from the generic design basis document, was used to represent this low cycle load function. Detail ed -

load determinations considering tube sheet flexibility, actual TMI-1 thermal cycles and measured tube preloads support the contention that the generic cooldown load cycle bounds the cooldown load cycle corresponding to the operating guideline limiting shell to tube temperature differential. of 70*F maximum. The measured tube preload was obtained from TMI-1 tubes which had -

{: 0 parted from the UTS and jumped down to a presumed tension free condition. The high cycle cmponent of loading, simulating fluid induced vibrations, was taken to correspond identically with measured tube deflections at 97% full power in TMI-2.

~~

The stability of the crack growth and the size of cracks after 40 years of plant operation were calculated using the EPRI developed computer code BIGIF.

l Since this code did not include in its element library a model to simulate the I cracked steam generator tubes, the stress intensity (K) data necessary to ,

i charact'erize the cracks was provided as input. This data was developed by Prof. Ercogan of Lehigh University. Other parameters pertinant to the analysis include the crack growth rate (da/dn)?and the thrisi1old stress intensity factor ( Kth). Conservative estimates for thse parameters were l derived from compiled experimental data for Inconel 600. The BIGIF code was I then used to predict the locus of initial crack sizes that would exhibit stable crack growth over a period of 40 years. The predicted locus was well within the range of detecbion by ECT inspection and was used to define the size of permissible cracks.

3

~

A pertinant contention established. by Crack 0pening Displacement (C0D) l . considerations was that a thru wall crack would result in detectable leakage. ,

1 NSAC-EPRI and FAA developed the values of tube leak rates as a function of

crack arc length with tubes subjected to 1107 lbs. and 500 lbs. tensile loads.

I This was done considering the linear and non-linear change. in C0D to predict the leakrate versus crack arc length. GPUN applied the NSAC method for predicting leakrate to other loads using comparative assumptions. The -

predicted minimum leakrate at cooldown for a single TMI-1 specific core tube

with zero preload and a pre-critical thru wall crack was double the .

, administrative limit of 6 GPH.

,- EVALUATION j

Several elements entered into the LEFM analysis. These were the development of the . loading function, definition of the input parameters, delection of the analysis method and implementation of' the analysis.

Considering each: \

i 3

Q j' '(

Loading function - the critical loads, those producing the njaximum tensile state, were a combination of a generic cooldown thermal cycle and actual pressure. and fluid. induced vibration l oads. The generic cooldown lo.ad was shown by detailed calculation to bound the cooldown load corresponding to the administra'tive limit cooldown rate. The pressures confonn to actual system pressure extremes while the vibration loads are based on measured extrenes for TMI-2.

~

Input Parameters - The material parameters were conservatively selected

' from available and applicable experimental data for inconel 600. The crack specific data (stress intensities) were developed by a recognized authority in the field. _m Analysis Method - The BIGIF computer code is a recognized state of the art LEFM analysis method. The C00 evaluations are extensions o,f acceptable NSAC-EPRI and FAA analyses.

Analysis Implementation - A developer of the BIGIF analysis method was engaged to guide the LEFM analysis. A recognized authority participated in the C0D evaluations. -

In summary the LEFM and C0D analyses were performed in a professional fashion using state of the art methods, experimental data and the servic'es of recognized consultants.

One oversight in the LEFM evaluation was the failure to perfonn a separate analysis to consider the effect of the residual stress generated by the

~

explosive fonning process. The stress state, as developed by GPUNC, varies .

from tension on the ID tube surface to yield in canpression as the wall neutral axis is approached from the ID, to yield in tension as the neutral .

axis is crossed, to conpression at the OD surface. This stress state exists at the juncture of the undeformed and deformed tube regions, two or eight 'l inches above the lower surface of the upper tube sheet. If added to the' state

- of stress produced by the LEFM loads the resultant stresses could acccelerate initial crack growth, impede radial growth and encourage circumferential-growth as the crack tip enters the compression zone and finally accelerate G

4 ..

b

'o (.f radial growth as the crack tip enters the tensile yield zone. 'The crack j characti zation in this region will clearly differ from that predicted for the naninal tube, possible exhibiting greater arc extent at thru wall penetration.

j If this is true these transition regions may be susceptible to parting j f a il ure.

i f Another oversight by GPUNC is the lack of an investigation of the tubes

! which have experienced jump down. These tubes are assumed to be free of j tensile preload and consequentially are more susceptible to the compressive loads ' associated with heat up than the nominal tubes. In particular design j calculations by B&W indicate that a tube with 100 lbs. preload will experience a compressive load of 775 lbs. during no rmal '^heatuji. A Eube with no preload would then experience a conpressive lo'ad of 875 lbs. However, a calculation  ;

! by the GPUNC consultants MPR Associates indicates that the critical buckling ,

! load for the tubes is 779 lbs. Apparently tubes which have exhibited jump ,

1 down experience compressive, loads in excess of the buckling load and should be

assumed to buckle. Since 'the loading is displacement induced these tubes will l not fail but rather will develop transverse displacements (bowing) to j accanodate that fraction of the loading above the, critical value. GPUNC l should address this condition to determine the assoc.iated tube stresses, the ',

I . impact, if any, on the LEFM studies and the vibrational characteristics of ,

i these tubes. If the bowing or lateral displacements are not large the tube stresses will renain compressive throughout the cross section although r

exhibiting moment distributions. The tubes will ~ exhibit lower natural  ;

f frequencies than the nominal tube. However, the magnitude and significance of I j this difference is not known but felt to be minor. Lastly, at heat' up, when -

the tubes are bowed, flow is low and little excitation of the tubes would be expected. .

CONCLUSIONS

. I

1. .The LEFM and C00 analyses were performed in a professional fashion

! using state of the art technology and the services of recognized authorities. i l The analyses are acceptable and support the GPUNC contentions concerni.ng crack -

p'rogagation and growth.

1 5

.~.

r -

- o (.. ..

2. The residual st'ress fields in the formed tubes were no't considered in the LEFM analyses. Since these stresses could alter crack propagation the i developed crick charact.erization may not apply to these zones (i.e., locus of

! crack size for 40 year life). However early thru wall cracking in these zones should be detected by the leakage monitoring program. Additionally, a tube parted in this zone should remain trapped in the upper tube sheet minimizing the consequences of failure.

I -

3. Tubes which have experienced jump down may exhibit bowing deflections during heat up. These deflections will induce bending stresses in and alter the vibrational characteristics of these tubes. Owing to the high axial com-pression, it is anticipated that the stresses in thibowed tubes will remain conpressive and therefore this loading should not accelerate crack propaga-tion. Considering vibrations, the bowed tubes should exhibit lower natural frequencies than the naninal tubes. However, since the fluid flow and con-sequentially the excitation, force are minimal during heat up, the fluid in-duced vibrations of these ' tubes should be below that exhibited by the nominal tubes at full power.

0 0

0 o

G e

g ,

Mme

_ _ _ _ . _ _ _ _ _ _ _ _ _