ML20207M208
| ML20207M208 | |
| Person / Time | |
|---|---|
| Site: | Oyster Creek |
| Issue date: | 12/29/1986 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML20207M191 | List: |
| References | |
| NUDOCS 8701130118 | |
| Download: ML20207M208 (9) | |
Text
___ _ _ _ _ _ _ _ _ _
[ o
/n UNITED STATES
'F 's NUCLEAR REGULATORY COMMISSION g,' . ,1 WASHINGTON, D. C. 20555 l \, /
SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION -
RELATED TO CORROSION OF THE DRYWELL SHELL LINE BREAK IN THE ISOLATION CONDENSER GPU NUCLEAR CORPORATION OYSTER CREEK NUCLEAR GENERATING STATION DOCKET NO. 50-219
1.0 INTRODUCTION
The licensee, GPU Nuclear Corporation (GPUN), first observed water coming from the drains that are connected to the sand cushion between the drywell and the surrounding concrete in 1980 but could not identify the source of leakage. The T licensee continued its investigation to locate and repair the source of the '
leakage in 1983. It was observed that the severity of the leakage was related to whether the unit was in operation or refueling. The source of leakage, the drain line gasket (30 in. x 7 in.), was identified and repaired by the licensee in the 1986 cycle 11 refueling outage.
By letter dated December 18, 1986 GPUN presented a summary of its investigations to date, the data obtained, the assessment of that data including a safety evaluation, and its intended future actions. The GPUN and NRC staffs met on December 1, December 10 and December 19, 1986 to discuss the above stated matters as the work by the licensee progressed.
To determine if the watcr observed coming from the drains had an adverse effect on the steel drywell, the licensee made a series of ultrasonic thickness measurements on the drywell plates. Initial surveillance measurements, utilizing a digital reading ultrasonic thickness measurement instrument, were made of the Oyster Creek drywell in the April /May time frame. The initial measurements indicated containment plate conditions and thicknesses were consistent with the original design except for areas at the approximate elevation of the interior drywell floor directly opposite the exterior sand cushion and extending over several bays. These early readings indicated apparent thinning due to loss of material on the exterior of the drywell down to thicknesses of about 0.95 in, compared to the as-fabricated thickness of 1.154 in. These early measurements led to an attempt to qualify the technique for painted wrfaces and then to a much more extensive series of measurements.
The more extensive ultrasonic testing (UT) surveys confirmed the general corrosion wastage mentioned above and further indicated there were local UT thickness measurements with indicated shell thickness as low as 0.383 in. In order to confirm the adequacy and accuracy of the UT measurements, to understand further the source of the highly localized UT readings, and to assess drywell plate conditions below the level of the interior concrete floor, the licensee decided to take drywell plate core samples in seven locations. These samples were obtained early in December.
8701130118 861229 2 PDR ADOCK 0500 P
The staff, in the following section of the safety evaluation, has reviewed the information presented by the licensee to support interim operation of the Oyster Creek Nuclear Generating Station for an additional operating cycle. The staff's evaluation addresses the credibility of the UT measurements, a review of the metallurgical assessment of the corrosion by the licensee, a review of the stress analysis considering the locally reduced shell thicknesses, and a review of the corrosion process itself.
2.0 EVALUATION 2.1 UT Testing / Metallurgical Assessment Because the licensee was concerned that repeated exposures of the drywell steel to water in the sand region could result in degradation of the drywell, measurements of the drywell portion of the containment shell were made to verify its thickness during the 1986 outage using a digital UT thickness instrument. Approximately 1,000 UT readings were eventually taken utilizing the ultrasonic thickness gauge device (0-meter).
To further characterize the drywell plates, additional examinations were per-formed using the "A-scan" UT technique. The characteristics of a reflector can i
be better determined by the "A-scan" technique since all reflections may be observed on a cathode ray tube display rather than as a single digital readout.
In addition, additional measurements were performed by EPRI and GE personnel.
Advanced UT techniques were also utilized to characterize corrosion conditions.
The UT results were utilized to select samples for metallurgical evaluation.
From the UT results it appears that general wastage of the shell, to a varying extent, is present in Bay 17 through Bay 1 and Bay 9 through 11 in the shell plates adjacent to the sand cushion. Localized UT reflectors ranging from 0.350 in. to 0.800 in, were found in Bays 5 and 15 adjacent to the sand cushion.
Further UT thickness measurements were taken in trenches in Bay 17 and Bay 5.
The Bay 17 measurements showed that plate thinning was present below the inside floor level on the shell in the sand pocket region but was not more severe than the thinning measured above the floor. No thinning was observed in the Bay 5 trench but other localized reflectors were identified. Thickness measurements taken at other locations of the drywell shell other than those adjacent to the sand cushion were normal, i.e., plate thickness were similar to those
.C.own in the original design drawings.
Core samples of the drywell shell were taken at seven locations. The locations selected were places on the shell that were believed to be representative of general wastage, contained localized UT reflectors, or were just above the wastage. A summary of the samples, their locations and thickness measurements are presented in Table 1. Sand behind the metal shell samples were also taken for chemical and microbiological analyses. The results of the chemical analyses are discussed in Section 2.2.
Metallographic evaluation of the samples by the licensee and GE confirmed that the wall thinning resulted from general wastage of the carbon steel plates.
The results of evaluations showed the samples were generally free of pitting and uniformly reduced from corrosion caused by water containing aggressive
TABLE 1 .
CORE SAMPLE THICKNESS EVALUATION 1
Sample No. Location Type of Sample Pre-removal Thick. Post-Removal Thick. (Ave;)
1 19C - 11'3 5/8" Wastage .815" (avg.) .825" ,
, 2 ISA - 11' 5 1/4" Pitting '.490" (min.) 1.170" center 1.17 (avg.) only 170 - 11' 3 3/4" 3 Wastage .840" (avg.). .860" 4 19A - 11' 3 3/8" Wastage .830" (avg.) .847" 5 IIA - 11' 3 Wastage .860" (avg.) .885" 6 11A - 12' 2 3/4" Above Wastage 1.170" (avg.) 1.19" center only 7 19A - 12' 1" Above Wastage 1.140" (avg.) 1.181" center only I.
I
/
, , __ _, r_ ~ . . .
p
't i .
anions. The biological results were inconclusive. The staff agrees with the c licensee that the corrosion mechanism is water with aggressive anions.
il Meta 11ographic evaluation of sample 15A confirmed that the local deep UT '
l indications in the sample were non-metallic inclusions typical of those might i be expected in a low carbon steel. The staff believes the small laminations were present since the_ manufacture of the plate. The. staff concurs that the l.
physical measurements of wastage samples adequately confirm that the UT i thickness measurements are accurate.
~
l
! 2.2 Corrosion Evaluation Examination of the core samples removed from the drywell shell 'shows a general aqueous corrosion attack on the outer surface of the shell in specific areas.
The licensee stated that informal undocumented communications indicate that
-water was observed in the torus room floor during and following construction.
It was also reported that during the application of the Firebar,D material to the outside of the drywell vessel, copious quantities of water were observed coming from the Firebar and running down the outside of the drywell. It is presumed the water with contaminating constituents from the insulation entered, into the sand bed.
l The drywell shell was analyzed as a free standing structure able to resist all required loads without interaction with the surrounding radiation shield ,
concrete. The shield concrete was set back away from the vessel to provide a
- gap that would limit shell deformation due to concentrated jet forces. At all elevations above the sand layer the gap was created during construction by applying a compressible, inelastic material (Firebar-D and Fiberglass) to the exterior of the vessel prior to pouring concrete. The vessel was then expanded to create the required gap needed for thermal and pressure expansion. During the expansion it was noted that the gap material had entrapped moisture due to incomplete curing of the Firebar D and introduction of water from external '
sources. To perform the expansion'of the vessel and create the required gap the vessel was pressurized to 40 psig and the temperature raised to 140'F.
This served as another opportunity to introduce impurities and water to the sand cushion. .
1 During the 1980 Oyster Creek plant outage, water was found leaking from various locations from the concrete surrounding the drywell. In addition, it l was reported that water was coming from the sand cushion drain lines into the torus room. Leakage from the sand bed drain lines was also observed during the 1983 refueling. These again were periods where water and impurities could be introduced into the sand cushion. ,
To verify the UT thickness measurements, assess the condition of the outside of l the shell, and assess the drywell below the level of the interior concrete floor, the licensee took core samples of the shell and the backup sand cushion.
Examination of the ~ exterior face of shell core samples removed from the locations that were thought to be general wastage were verified. The corrosion films were magnetic, dark in color and when subjected to chemical and other analyses showed high nitrates, chlorides and sulfates. Leachate samples of the Firebar D and of sand samples from four of the removad plugs showed high concentrations of chlorine and sulphates. The Firebar leachate contained 573 pg/gm of chloride while the chlorides from the sand leachate were in the range of 6.5 to 93 pg/gm.
I l .
Sand is generally considered to nave good drainage properties if not compacted too tightly. This property would allow the bulk of the water introduced into the sand bed during construction to flow out thru the drain lines but would leave the bed moist and contaminated. Expansion of the vessel to create the gap, though minor at the sand bed level, would compact the sand bed irregularly because of non-uniform original sand compaction and different shell movement ,
around the periphery and thus change the local permeability of the bed. Wetting of the sand bed during the subsequent refuelings would leave aqueous solutions of contaminated water in contact.with the shell. With the above scenario one plausible explanation is to assume that the major corrosion activity at the outside of the shell was initiated during the 1980 refueling outage since this is documented as the first time when the sand cushion is known to have been thoroughly wet after the drywell was expanded. The exterior of the drywell was coated with red lead primer which would provide general protection to the steel surfaces against atmospheric corrosion; however, this would not provide much protection against the environmental conditions in the sand cushion.
Damaged-coating areas, and with time, all areas within the sand bed would be expected to experience general corrosion as long as the steel remained wet or until a protective oxide film built up on the steel surface as a result of the corrosion. The protective film did not appear, possibly because of the contaminants in the water, notably chloride.
From the conservative scenario stated, we therefore conclude that the major metal loss due to corrosion occurred in the time period from 1980 to the present.
We further conclude that the wastage in bays 11, 13, 17, and 19 is the result of an aqueous general corrosion process contralled by local oxygen available, the amount of moisture, temperature, and chlorides. The UT measurements of the plate confirmed by cored and pnysically measured samples showed an average corrosion loss 288 mils. This loss projected for 6, years, 1980 to present, will indicate a corrosion rate of about 50 mils per year. This corrosion rate will allow the safe operation of the plant for the next fuel cycle which will be 18 months in duration. If the corrosion was assumed to have started when the sand cushion was believed to have been initially wet and contaminated, the corrosion rate would be about 20 mils per year. Since there is uncertainty in the corrosion rate because of the inability to determine when the process -
began and there is a limited possibility that micro-organisms may have contri-buted to the corrosion process, the staff requires a UT shell examination no later than September 30, 1987 to confirm the corrosion rate.
2.3 Structural Integrity Review As a result of the reduction of thickness of the steel shell from 1.154" -
to an average about 0.850" thickness with local spots to about 0.750" in some areas of the sand pocket region of the drywell, there is a major concern for the structural integrity of the drywell. Reduction in the thickness of the steel shell reduces the load-carrying capacity of the drywell. In order to assess the drywell structural capability, the licensee employed Chicago Bridge and Iron Company (C8&I) to perform a structural analysis of the steel shell. CB&I was the original designer, fabricator, and erector of the Oyster Creek containment structure. The CB&I analysis is contained in a draft preliminary report entitled "Drywell Analysis Sand Transition Zone-0yster l Creek Containment Vessel," dated December 16, 1986. The following is an evaluation of the CB&I Analysis as contained in the report.
Analysis Model -
c The Oyster Creek drywell consists of a 70-ft diameter spherical steel shell with a 33'ft diameter by 23 ft high cylindrical steel shell extending from
, the top. The lower spherical portion of the drywell is embedded in concrete to elevations of 8 ft - 11 ft 4 in. and a sand pocket extends from the' point of complete embedment upward 3 ft 3 3/4 in, to an elevation of 12 ft 3 in. The analysis model consists of the' spherical portion of the drywell which is assumed to be fixed at the point of embedment, and extending up to the sand pocket elevation 23 ft 6 7/8 in. In order to take into consideration the reduction of the shell plate thicknesses, a shell thickness of 0.70 in. was assumed for the region of shell adjacent to the sand pocket and a full shell thickness of 1.154 in. for the rest of the model. The boundary conditions for the various loading conditions at the top of the model were obtained from the analyses of primary membrane stresses for different loadings.
In view of the fact that the analysis is to determine the effects of reduction of shell thickness on the containment shell behavior at the point of discontinuity when subjected to various loadings and considering that such effects are limited only to the immediate vicinity of the discontinuity, the analysis model adopted by CB&I is judged to be appropriate.
Method of Analysis CB&I used the Kalnins Shell of Revolution Computer Code to make the analysis.
The computer code treats the shell structure as axisymmetric which can be subjected to axisymmetric and non axisymmetric loadings. In order to take the effect of the sand pocket into consideration, two analyses were performed:
one in which a sand spring constant for the radial displacement is applied to the shell in the sand pocket region, and the other analysis in which the presence of.the sand pocket is neglected. The Kalnins Shell of Revolution Computer Code has been widely used in the design of containment structures since late 1960's, and is, therefore, an established method of analysis and suitable for this application.
Loads and Load Combination The degraded steel shell of the drywell was analyzed for the following two cases:
a) Pressure = 35 psig, Temperature = 281*F b) Pressure = 62 psig, Temperature = 175 F The dead weight of the steel shell and of the appurtenance, operating basis earthquake (0BE) equivalent horizontal earthquake of 0.11g, and vertical earthquake of 0.05g were included. Since the evaluation is limited to the discontinuity portion of the steel shell, pressure and temperature should have the most significant effect. Therefore, it is judged that the loads and load combinations considered were appropriate for this analysis.
y Applicable Codes !
The Oyster Creek containment vessel was designed, fabricated, and erected by CB&I in accordance with the 1962 edition of ASME Code,Section VIII and Code Cases 1270N-5, 1271N, and 1272N-5. The allowable stresses used in this -
re-evaluation are consistent with the original Code used in the design.
However, the 1986 ASME Section III, Subsection NE Code is used to interpret the results of the analysis for the local discontinuity since the earlier Code did not address this. For the A-212-61T steel plates used, the following allowable stresses are used:
Primary Stresses (does not include thermal effects)
Allowable Stresses
- Ref.
General Membrane 1.1 x 17,500 = 19,250 psi 1272N-5 5(g)(1) local Membrane ** 1.5 x 1.1 x 17,500 = 28,875 psi NE-3221.2 Local Membrane 1.5 x 1.1 x 17,500 = 28,875 psi 1272N-5
+ bending 5(g)(2)
Surface Stress *** 3.0 x 17,500 = 52,500 Table NE-3217-1 Secondary Stresses (includes thermal effects)
Surface Stresses (Pl+Pb +Q) = 3.0x17,500 = 52,500 psi 1275-5 5.(f) and NE-3212.4 all actual stresses are either stress intensities per NE 3000 or unidirectional stresses per ASME VIII, and Code Case 1272N-5, whichever is greater a local primary membrane stress is defined as one which does not exceed 1.1.x1.1x17,500 = 21,175 psi for a distance greater than 1.0 (rt)b Ref. NE 3213.10 if bending moment at the edge is required to maintain the bending stress in the middle to acceptable limits, the edge moment is classified as Pb. Otherwise it is classified as Q.
f_J
, 1
. i Note: For primary stress evaluation -
loads include ,
(1) Internal pressure (2) Dead weight of steel (3) Dead weight of appurtenances (4) 11% Horizontal Earthquake - '
OBE equivalent (5) 5% Vertical Earthquake - OBE Equivalent essentially service Level A in 1983 ASME Code For Secondary Stress Evaluation - loads include all of'above plus meridional thermal gradient.
The allowable stresses used by CB&I as stated above are appropriate for the evaluation. In addition to stress analysis, the potential for buckling of the drywell shell in the sand entrenchment was investigated by Prof. A. Kalnins of Lehigh University. Both cases with and without the sand buckling were analyzed.
From the results of the analysis the margin of safety was found to be 6.12, which is larger than the acceptable minimum factor of 2.
Results of the Analysis The results of the analysis as given by CB&I can be summarized as follows:
I. With sand pocket a) P = 35 psig T = 201"F membrane stress = 25,251 psi (membrane + bending) stress = 32,147 psi b) P = 62 psig T = 175 F membrane stress = 17,889 psi (membrane + bending) stress = 24,935 psi II. Without sand pocket a) P = 35 psig T = 281*F l membrane stress = 8821 psi (membrane + bending) stress = 43,722 psi b) P = 62 psig T = 175 F i
membrane stress = 16,944 psi (membrane + bending) stress = 53,897 psi The allowable stresses are 28,875 psi and 52,500 psi, respectively, for the membrane and the (membrane + bending) stresses.
i
, . - ~ , - . - - , , ,._.-n ---n.- - - - - - - - - - . _ , _ . - - - - - - , - - - - - - - . - - . - . - . , - . - - - - - - - - - - - - . . - . , - - - . - - -
p ~
With the. exception of- the membrane plus bending stress for the case II.b, all the stresses were within the allowable. Even in the case where the stress was exceeded, the exceedance is less than 3% of the original Code. Furthermore, if the-1986 ASME Section III NE Code is applied, the calculated stress will be within the' allowable stress in the current Code.
Based on our review of the information'provided by the licensee, we conclude that the structural integrity of the drywell is adequate in terms of the-original design Codes with an assumed shell thickness of 0.700" in the sand cushion region.
3.0 CONCLUSION
Based on the factors discussed above and considering the compensatory measures outlined below, the staff concludes that the operation of the Oyster Creek Nuclear Generating Station for cycle 12 would be safe with the drywell steel plates not fully in confonnance with the original FSAR design criteria and would not present an undue risk to the public health and safety.
The staff has determined, however, that the following actions are necessary on the part of GPUN:
4
- 1. The licensee has committed "...to continue the UT shell thickness test programs at future outages of opportunity including forced outages otherwise requiring drywell entry during the next cycle." The staff re<1uests that the licensee provide the staff the plans for the intended inspection and, should there be no forced outages of sufficient duration prior to the mid point of the operating cycle, perform the inspection the no later than September 30, 1987. The staff feels the action stated is necessary to assure the corrosion rate assumed by the licensee is bounding.
- 2. The licensee shall, by June 30, 1987, provide for staff review and approval, its comprehensive plans for mitigating the corrosive attack on the drywell shell and any other intended long-term corrective actions it may deem necessary.
Principal contributors: R. Hermann, B Turov11n, and C. P. Tan.
Dated: December 29, 1986