ML20235W764
| ML20235W764 | |
| Person / Time | |
|---|---|
| Issue date: | 10/14/1987 |
| From: | Murphy A NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Gallagher L NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| References | |
| NUDOCS 8710190016 | |
| Download: ML20235W764 (20) | |
Text
_ _ _ _ _ _ _ _ _ _, _ _ _ _
f Y SS OCT 14 gg 2
MEMORANDUM FOR:
L. Gallagher PMPDAS/RES FROM:
Andrew J. Murphy, Acting Chief SSEB/DE/RES
SUBJECT:
1987 NRC ANNUAL REPORT Enclosed are write-ups for the Structural and Seismic Engineering Branch research programs to be included in the 1987 Annual Report.
If you have any questions, please contact H. Graves at 443-7862.
Andrew J. Murphy, Acting Chief SSEB/DE/RES
Enclosure:
As stated cc:
A. Eiss, RES R. Hoskins, RES DISTRIBUTION: RESReading HGraves GArndt AMurphy RBosnak GArlotto SSEB/DE/R S
.7kES SSEB/DE/RES HGraves t rn t AMurpa &
10//[/8 10//h87 10/a/ 87 0Db hhfjA PDR
l 3
SEISMIC AND FIRE PROTECTION RESEARCH Earth Sciences
- The primary. goal of the NRC research program in the geological sciences is to be able to define the potentia: for earthquakes at nuclear power plant sites and in the regions around the sites, and to determine the possible effects
. earthquakes would have on the plants and their safety systems.
A major focus-of' the NRC. research programs in geology, seismology and geo-physics continues to be identifying and defining potential earthquake sources or source zones in the Eastern United States and using that information in as-sessing seismicL hazards with respect to nuclear power plants.
Many unknowns exist regarding these issues, including a strong basis for seismic zonation,
' source mechanisms, characteristics of ground motions. and site-specific response.
The NRC is addressing these uncertainties. through research that encompasses sustained -seismic monitoring, neotectonic investigations, exploring the earth's crust at hypocentral depths, and conducting ground motion studies.
The backbone of the NRC program in the eastern United States has been the seismic networks deployed throughout the eastern and central United States.
The NRC is currently funding seismograph networks in the following regions:
Northeastern United States, Virginia, Charleston, South Carolina, the Southern Appalachian region, the New Madrid region, and Ohio and Indiana, Eastern Kan-sas, and Oklahoma.
The NRC has negotiated an inter-agency agreement with the U.S. Geological Survey (USGS) to jointly support the establishment of the
. eastern portion of a national seismographic network.
Northeastern Neotectonics As part of the seismic research program to improve our ability to estimate L
. seismic hazard in the eastern United States, Columbia University and Penn-sylvania State University have been investigating for the past several years, seismically active regions in the northeast for evidence of Quaternary surface l
a 2
or near surface tectonic deformation.
Methods of accomplishing this have evolved from classical field geological techniques utilized in earlier studies of this region, to newer cus developed for neotectonic investigations in the New Madrid, Charleston, and California regions.
These include:
search for paleoseismic evidence of prehistoric earthquakes to extend the seismic record j
beyond the limited historic record, correlation Quaternary marine and fluvial terraces, and utilization of new, relatively shallow high resolution explora-tory techniques and equipment such as ground penetrating radar.
Columbia University has completed its studies of surface structures in the epicentral area of the 1983 Goodnow earthquake in the Adirondack mountains. No relationship between surface faults and the earthquake was found.
Columbia researchers are now investigating seismically induced liquefaction features near Cape Ann, Massachusetts that are known to have occurred during the 1755 MMI VIII earthquake. They plan to use the data obtained to identify other such features that many have been caused by prehistoric earthquakes.
They are also investigating the 125th Street fault in New York city for evidence of young displacements, the Lancaster, Pennsylvania seismic zone, the Lower Hudson Val-ley, Eastern Newark Basin seismic zone, and the New Jersey Coastal Plain. The paleoseismic investigations, by providing isotopic dates of large prehistoric i
earthquakes, have the potential of providing deterministic guidance for cal-culating return periods of large earthquakes in the northeastern United States.
This would be a major step in assessing seismic hazards in the eastern United J
States.
i A study performed by the Pennsylvania State University focused on the Lan-caster, Pennsylvania and Moodus, Connecticut seismic zones. The Lancaster area reveals a N-S trending structural and seismic zone that cuts across the strike of the major Appalachian structures.
The zone is favorably oriented to be activated by the prevailing ENE compressional stress.
The N-S trend is out-lined by epicenters, aftershocks and focal plane of the 1984 earthquake, and geologic and geomorphic trends, which include diabase dikes, faults, springs, drainage, and lineaments.
3 At Moodus, no specific seismogenic structure has been found, but there are some L
characteristics that may be reiated to the seismicity.
Lineaments at Moodus are more numerous and shorter than in adjacent areas, possibly be-cause the crust is more broken up and hence more easily activated by local stresses.
Stream water samples from the seismic zone have higher than normal pH, which is consistent with influx of water from the subsurface along fractures.
A recent deep borehole has found a water-filled zone at 3200 feet, below the Honey Hill fault.
Above this zone, stresses are higher than below.
This may explain the shallowness of the hypocenters, which seem to be concen-trated in the more highly stressed surface layer.
j A third subject for this study were travertine deposits and their possible use as inc'cators of recent fault' movement.
It had been found previously in Virginia that travertines have formed on the downstream side of faults that cut limestone strata.
Limestone crushed by the faults would be more easily dis-solved and carried to the surface into streams, where travertine is deposited in locations where the saturated water is aerated such as riffles and falls.
In the present study, it has been found that near Lancaster very recent travertine deposition occurs downstream (either east or west) from the surface projection of the fault that is assumed to be associated with the seismicity.
Thus, there is substantial evidence for the association of travertines with recent fault movements.
Southeastern United States Charleston Studies.
The NRC has funded over the past few years studies by the USGS and the University of South Carolina of soil deformed by liquefaction during the 1886 earthquake and of similar, but older, features (paleoliquefac-tion features) that were apparently formed by prehistoric earthquakes of about the same size. These investigations suggest recurrence intervals between 1,000 j
and 2,000 years for earthquakes of the same size as, or greater than, the Charleston event. To support the NRC position that the Charleston seismic area is unique or to demonstrate that such an assumption is not valid, the NRC has encouraged expanding the area of investigation to determine whether or not l
there are paleoliquefaction features elsewhere on the Atlantic Coastal Plain.
4 The USGS has identified paleoliquefaction features beyond the immediate Charleston earthquake area and postulates that either a much larger earthquake occurred in the Charleston epicentral area, or earthquakes of similar size occurred prehistorically at other locations along the southern Atlantic coast.
In 1986, the NRC awarded a research contract to Ebasco Services Incorporated to look for paleoliquefaction structures throughout the Atlantic Coastal Plain.
Based on detailed investigations in the Charleston region by the USGS, the University of South Carolina and Ebasco, Ebasco researchers have applied the results of those studies to identifying ~ other areas in the Atlantic Coastal Plain that have the potential for evidence of large prehistoric earthquakes if
-l they have occurred.
These sites are:
southeastern New Jersey - northeastern 1
Delaware, the eastern part of the Central Virginia seismic zone, and north-eastern North Carolina.
Techniques of investigations and criteria developed at Charleston will be used to investigate these sites.
A. study of the Charleston seismicity by Law Engineering Testing Co., has em-ployed 2-and 3-dimensional stress models to clarify causes of the seismicity and to complement the sparse stress data in this region.
The stress models take topography, density and plate boundary stresses into account to derive the stress distribution over the area.
There are two major structures that should influence local stresses, namely the Appalachians and the continental shelf edge. The models show that these features indeed generate large stresses.
{
However, when ridge-push forces are taken into account, the stress near the l
Blue Ridge of the Appalachians is enhanced whereas the shelf edge stress is largely cancelled.
This corresponds to the observed seismicity which is high in the Appalachians and low near the shelf edge.
The Charleston area also emerges as a region of higher stress, while areas of minimal seismicity, such as eastern North Carolina and southwestern Georgia are characterized by low l-Stress.
It appears, therefore, that stress computations can be a valuable tool i
for analyzing the seismic potential of certain areas.
l The University of South Carolina has performed shallow stratigraphic drilling in the seismic areas near Charleston and Bowman, South Carolina.
The borehole dets have shown the presence of a trough near the coast which may be bounded by
5 faults.
These possible faults and similar linear subsurface features found at Bowman are characterized by NW and NE.
Seismicity seems to be concentrated at intersections between these trends.
Relative travel time residuals from tele-seismic data are being analyzed to gain additional information on the crustal structure in this area.
Virginia Piedmont l
VPI has also reprocessed and reinterpreted a seismic reflection profile ac-quired by the USGS near route I-64 and shot new reflection profiles in the Virginia Piedmont in the Central Virginia seismic zone and in a non-seismic area along the Roanoke River.
This has led to a reinterpretation of the base-ment structure under the Piedmont and Coastal Plain.
The data confirm a pre-viously suspected upwarp of the Moho underneath the Coastal Plain that coin-cides with a major gravity high.
The Piedmont and western portion of the Coastal Plain are now interpreted to be underlain by Grenville basement.
The eastern portion of the Coastal Plain may be undertaken by a basement equivalent l
to the Carolina State Belt, the two basements being separated by an extension l
of the Taconic Suture.
Comparisons of crustal configuration between seismic and non-seismic areas show that the seismic areas have a greater number of subsurface reflectors.
This j
may indicate that the crust is more sheared and segmented than the more massive o
crust in surrounding areas.
It is also significant that the seismic zone over-lies the shallow portion of the Moho.
The new reflection profiles have pro-4 vided excellent reflections from the Moho, but data processing has not been completed yet.
Studies of the crustal structure in southeastern Tennessee by Georgia Institute of Technology have located a sedimentary basin that is associated with the relatively high seismicity of that area.
Magnetic and gravity data, travel time residuals, and refraction data were used to derive the interpreted crustal structure.
The sedimentary basin parallels the New York - Alabama lineament, which probably represents an ancient strike-slip fault.
Thickness and l
t
6 seismic velocity of the crust differ on the two sides of the lineament. North-west of the basin a relict Precambrian rift is postulated, which is truncated by the New York - Alabama lineament.
Epicenters relocated on the basis of new velocity information correlate well with the main NE and NW trend lineaments in the area.
A 2-dimensional stress model shows high compressive stress in the seismic area.
The stress is derived from local topography and density anoma-lies in the crust.
Central United States Southern Illinois Earthquake of June 10, 1987 A magnitude 5 earthquake occurred at about 7:50 p.m. E.D.T. on June 10, 1987 near Lawrenceville, Illinois.
It was felt over 15 states and southern Canada.
Geologically, the earthquake occurred in the vicinity of the Wabash Valley Fault System near its intersection with the LaSalle Anticline.
Earthquakes of this size have occurred in this area of southern Illinois histo'rically.
NRC has funded geologic research of faults in this fault system but no avidence of recent activity was found.
The earthquake triggered instrument! at several nuclear power plants in the region including Dresden, at 300KM from the epicenter, Cook at 390KM, Quad o
Cities at 400KM, Palisades at 420KM, and Prairie Island at 770KM.
It was felt at other nuclear plants, but instruments were not triggered.
Immediately after the main shock, NRC contractors from Memphis State, Univer-l sity of Kentucky and St. Louis University deployed a temporary network of seismographs.
Excellent records of this event and aftershocks were obtained and are still being analyzed.
l l
New Madrid / Anna, Ohio Purdue University has analyzed seismicity, geologic and geophysical data, and borehole information of the midcontinent region between New Madrid, M0
i i
7 and Anna, OH.
The improved database that has been developed over the past 10 years has led to the conclusion that two hypotheses can explain most of the midcontinent seismicity.
The dominant mechanism is reactivation of existing zones of crustal weakness that are favorably oriented with respect to the NE-SW direction of the maximum compressive stress in this region.
Local basement inhomogeneities are a second mechanism that may explain seismic activity of low magnitude.
The Anna, OH seismic zone has been investigated in detail, and there is no evidence for a structural connection of the area with the NE extension of the New Madrid rift.
The newer data do support an extension of the Gren-ville front which locally trends N-S and is found just E of the area.
The seismicity may be related to intersecting trends in the basement or to lithologic differences (inhomogeneities) in the basement.
Meers Fault Studies The initial NRC-funded investigations of the historically aseismic Meers Fault in Oklahoma have been completed. These investigations have shown, with several lines of evidence, that about 26KM of the fault has undergone recent displace-ment, the latest of which probably occurred 1,100 to 1,200 years ago.
Cumu-lative displacements of up to 5 meters of reverse offset and a much larger left lateral strike-slip offset were recorded.
A new contract was led to Geomatrix Consultants in May 1987.
The purpose of the contract is to characterize completely the Meers Fault for seismic assess-ment and to detennine if there are other such faults that may have been reacti-vated in the Quaternary within the Frontal Fault System.
Another fault in this system, which is east of, and en echelon with the Meers Fault, is being investigated under an NRC grant to the l'niversity of Arkansas.
Geologic and geomorphic evidence regarding this fault, the Washita Valley Fault, suggest Quaternary displacement.
Like the Meers Felts, this fault is not known to be associated with historic seismicity.
l l
l
8 These studies are extremely important not only to assess the seismic hazard l
posed by these faults but also to test the validity of the assumption used frequently in the licensing process, that the lack of associated seismicity is an important criteria indicating that a fault is not capable within the meaning of Appendix A 10 CFR Part 100.
Pacific Northwest A major unknown concerning the Pacific Northwest is the nature of the Juan de Fuca subduction zone and its potential of generating a great earthquake. There is geological and geophysical evidence that subduction is taking place at a fairly rapid rate, but there have been no large thrust earthquakes histori-cally, like those that characterize other subduction zones around the Pacific Ocean. The issue is whether subduction is occurring seismically or whether it is taking place seismically but the historic record falls within the recurrence interval of large subduction zone earthquakes. The NRC is providing funding to the U.S. Geological Survey to investigate paleoseismic evidence that might have been induced by prehistoric large earthquakes.
Evidence has been found in marsh and shallow marine deposits within bays and tidal estuaries along the coast of Washington and northern Oregon that suggest the occurrence of several great earthquakes during the Holocene, the last occurring about 400 years ago.
A second major issue in the northwest is the nature of ground motion from a o
subduction zone earthquake.
Along with the geologic investigations the NRC is finding a USGS study in the Santiago, Chile region location of a magnitude 7.8 subduction zone earthquake in 1985.
The study consists of analyzing all data from this event and its aftershocks to be able to determine the character-istics of strong subduction zone earthquake ground motion for use in nuclear licensing activities in the U.S. Pacific Northwest.
Soil Response to Earthquakes A research program to validate dynamic stress models that would be capable of predicting soil settlement resulting from seismically induced liquefaction
i l
i l
9 continues.
The objective of the research, being conducted by the Anny Corps of Engineers, is to evaluate various seismic settlement models identified in a previous phase of this project and reported in NUREG/CR-3880.
During 1987, l
two two-dimensional plain strain centrifuge experiments simulating massive structures such as nuclear power plants were conducted at Cambridge University in England.
The experiments were continued until liquefaction failure of the supporting soil was achieved.
The data is being analyzed and compared to the predictions of the validated two-dimensional effective stress model TARA, Je-veloped during the course of this research program.
In a related development, Japanese investigators have compared the results obtained from the TARA code with DIANA, an effective stress code developed by Dr. Zienkiewicz at the University of Swansea.
While both solutions converged to the same answers on earthquake motions, pore pressures and displacement, the TARA code utilized only a tiny fraction of the computer time taken by the DIANA code.
It may be noted that the TARA code was developed and validated from data obtained from centrifuge testing using soil mechanics principles while DIANA is a more theoretically exact and rigorous mathematical formulation, designed to predict the rerponse of soil to earthquake motion.
Research to expand the current l
resear.:h project to consider modeling three-dimensional effects and update the two-dimensional TARA code is planned to start in FY 88.
j l
Component Response to Earthquakes Seismic Category I Structures Program:
l The static testing of two large reinforced concrete models representing a por-tion of a nuclear power plant building (i.e., shear wall and floor segment) were performed th',s year. This is part of static and dynamic test series which began 510 3, and will conclude in 1988.
The purpose of this test series is to investigate the large differences observed when analytical predictions of building response are compared with experimental data.
Based on a preliminary evaluation, it appears that the 1987 static test data contradicts previous dynamic test observations.
That is, an excellent comparison of analytically derived stiffness was obtained from the recent tests.
Subsequent program I
activities will center on investigating the rationale for the differences ob-tained from the static and dynamic tests.
The overall goal of this program is
10 to assess- (1) the ability. of-Category I structures other than the containment to sustain ~ earthquake motions in excess of their original design bases, and (2)
. the effect.that the changed building response has on the criteria used in the.
-design of piping and equipment.
EPRI/NRC Piping and Fitting Dynamic Reliability Program (PFDRP)
This cooperative EPRI/NRC research program was initiated in 1985 with three main objectives:
To identify failure mechanisms and failure levels of piping components o
and systems under dynamic loadings, To provide a data base that will improve our prediction of piping syster.
o response and failure due to high level dynamic loads.
To develop an improved and defensible set of piping design rules for in-o-
closion into the ASME Code.
'The majority of the experimental work of the PFDRP was completed by the end of FYB7. A major milestone was reached in June 1987 when ETEC completed the "Sys-tem 1" series of design-level and high-level seismic input tests of a pres-surized 6-inch carbon steel piping system.
The piping system was well-instru-o mented and the recorded response data will provide valuable benchmarks for j
future evaluation of linear and nonlinear piping analysis methods.
Of immedi-ate interest is that for the first time a failure of a pressurized prototypical piping system was achieved under very high seismic-like loads. An input scaled roughly 25 times higher than normal SSE design limits produced hanger and valve operator failures, ratcheting in elbows, but not leakage.
The input was then scaled even higher and excessive ratcheting at an elbow resulted in rupture.
ETEC also began construction of the stainless steel " System 2" in FY87 and will complete testing of this by January 1988.
i
.y
[
11 j
Over one half o' fee forty piping component tests were completed by ANC0 engineers by the end of FY87.
The rest will be completed by January 1988.
Fatigue ratcheting specimen tests and waterhamer tests are also currently underway.
The-results of the pipe component and pipe system experiments have shown sup-prisingly consistent general trends:
1.
1he results show that typical elastic piping design evaluations using the current ASME Code are very conservative for dynamic inertial loads.
Mar-gins to failure of 15 to 30 were usually observed.
2.
Dynamic failure is dependent upon cyclic effects, even at input levels of incredible earthquake size.
3.
Ratcheting and wall thinning led to the dynamic failure of pressurized piping.
4.
Cross-sectional collapse (as assumed by Eq. 9 of the ASME Code) did not occur.
It seems that dynamic load reversal prevents collapse.
5.
Failure locations were determined by loading and geometry, and independ-ent of weldment locations.
i 6.
" Loss-of-flow" failures did not occur.
Swelling occurred in the pres-surized - piping and crimping was minimal in the unpressurized piping.
j j
7.
Extensive testing at OBE and SSE levels produce no detectable permanent deformation or damage.
Even at the 5 SSE level, deformations were very localized and small.
Items 1 through 4 above indicate the need end justification for the future ASME Code criteria changes that will result from the PFDRP.
Items 6 and 7 show the need to rethink piping " functionality" concerns.
l 12 General Electric is now heavily engaged with the tasks of identification, de-velopment, and evaluation of alternative piping design rules.
These will pro-duce proposed revisions to the ASME Boiler and Pressure Vessel Code for the dynamic laod design criteria for Class 1, 2 and 3 piping components. Several piping consultants review and support this effort and both ASME and PVRC standards groups are monitoring the progress. These propsed criteria revisions will be completed when the PFDRP ends in the Spring of 1988.
Seismic Component Fragilities Fragility data were developed for motor control centers, switchboards, panel-boards and DC power supplies.
A new and more reliable single parameter fra-gility descriptor called the " average spectral acceleration" was developed to l
replace the "zero period acceleration" used previously.
This information will be used in future seismic PRAs and margin studies to identify weaknesses and strengths in nuclear power plant seismic design, and to assist in seismically related licensing decision making.
A major additional task added to this program related to establishing informa-tion to support the resolution of USI A-46 dealing with seismic qualification of safety related equipment. Electrical cabinet damping and amplification data j
has already been developed.
A testing program for auxiliary and protective relays has been funded and will assist in evaluating the " qualification by similarity" concept in addition to determining the chatter fragility as a function of input frequency and relay adjustments.
1 i
Validation Validation of Seismic Calculation Methods i
Seismic probabilistic risk assessment (PRA) methods have been employed to l
clarify safety issues for nuclear power plants.
The randomness of the seis-f mic hazard, the uncertainties and variety of the data needed, and the in-i exactitude of the methodology raise questions of credibility with respects to the results of seismic PRAs. The objective of validation research is to obtain l
l 1
1 13 information that the NRC can use to develop criteria for judging predictions of the behavior of nuclear power plants subjected to large earthquakes and thereby improve the reculatory process.
The predictive methods to be validated are used in both probabilistic and detenninistic predictions.
The strategy is to engage in cooperative research program in order to stretch available resources.
The NRC is participating in the following three efforts:
1.
A soil-structure interaction (SSI) experiment at a site in Taiwan.
This effert is in collaboration with the Electric Power Research Institute and the Taiwan Power Company.
The objective of the experiment is to obtain data from a soft soil site that will test the fidelity of analytical pre-dictions of SSI effects.
Fourteen earthquakes have been recorded, three of which exceeded Richter magnitude 6.0.
A workshop will be held in December 1987 at which the results of the experiment will be evaluated.
2.
The Phase II experiments being performed at the Heisdampfreaktor (HDR) facility in Kahl, West Germany, in a collaboration with Kernforschung-szentrum Karlsruhe (KfK).
Results of the first series of tests are cur-rently being evaluated.
In those tests, the containment building was excited by a large eccentri-mass shaker and the responses of a piping loop were recorded for different support conditions. Support conditions ranged from a stiff system, typical of early U.S. practice, to a very flexible system and included innovative systems intended for replace snubbers.
Planning for the second series of tests, to be run in the Spring of 1988, in which the piping loop will be excited well into the inelastic range were completed in 1987.
The purpose of these experiments is to develop information about seismic margins.
3.
Tests of a 1/2.4 scale model of a PWR piping loop to be performed on the large shaker table in Tadotsu, Japan in collaboration with Japanese j
history of International Trade and Industry (MITI).
Final design of the experiment which will be carried out in April 1988 were completed in 1987.
The experiment involves modification of the Japanese scale model which will then be excited well into the inelastic range to develop information f
1 about seismic margins, j
l
D 4 v.
1, 14:
Seismic Design Margins
' Most seismic : design experts. agree that ~ nuclear power plants are capable. of withstanding earthquakes.much larger than their original' design basis without.
= compromising their ' ability to safely shutdown and remain in a safe shutdown j
(
condition.
However, only recently through the Seismic Design Margins Program, I
- have wel had: the tools to effectively and efficiently quantify the inherent overall seismic capability of nuclear. power plants and to provide results that can be used directly' for licensing decisions.
The successful completion of the Maine. Yankee seismic margins review (NUREG/CR-4826) in March 1987, and the
. issuance of. an NRR Safety Evaluation Report based on this review are major
. milestones in the seismic evaluation of nuclear power plants.
The Maine Yankee Atomic Power Station is a. Combustion Engineering ;three-loop pressurized water reactor located approximately 4 miles south of Wiscasset,
~ Maine.
-It' started comercial operaticn in 1972.
The design safe shutdown earthquake.(SSE) has a horizontal acceleration of 0.1g.
1 The occurrence of two seismic events in the vicinity of the plant, one in 1979 and the other in 1982, prompted Maine Yankee to upgrade the capability of the.
plant to withstand a potential seismic event in excess of the original design-basis event.
Based on these upgrades and the inherent design capacity of the plant, Maine Yankee concluded that the plant structures, systems, and compon-
.o ents had sufficient strength to withstand a seismic event of at least 0.2g with ~ a Regulatory Guide 1.60 spectrum and still shut down without danger to j
the public health and safety.
To assure the NRC that the plant could with-stand earthquake motion greater than the design basis, the utility agreed to participate in the trial seismic margins review of the Maine Yankee plant.
For this review, it was agreed the seismic margin review earthquake level would be 0.39 with a 50th percentile Newmark Hall Spectra defined in NUREG/
CR-0098.
The Maine Yankee margins review followed the eight step process outlined in the guidance of NUREG/CR-4482.
The review involved the Maine Yankee Power Corporation, Yankee Atomic Electric, the NRC, LLNL as project manager and
y p
s.
' 15.
fragility and-system analysis. teams -(EOF inc., and Energy Incorporated, re-
.spectively).4 For. Maine. Yankee, two imn';rtant accident sequences were identified.
Both are initiated by a se%1cally induced loss of offsite power assumed to always occur at the eview earthquake level.
In one sequence there is a small loss-of-cool"..c accident'(LOCA of 3/8 in to 2 in. diameter equivalent area). assumed te occur because of seismically induced pipe breakage.
HCLPF capacities for components: which might cause other types of small LOCAs (pump seal or power-operated: relief valve LOCAs) were sufficiently high so they could be screened out.
The other accident sequence assumed no small LOCA.
The plant HCLPF capacity was determined to be 0.21g.-.This capacity is dominated by the small LOCA-accident ' sequence with the potential failure of the Refueling Water Storage Tank (RWST) being the dominant contributor.
Maine Yankee has decided to upgrade the anchorage system on this tank and by this design change the
' plant l HCLPF will be raised to 0.27g, a value comfortably above the design earthquake levels in question.
(It should be noted that a HCLPF value would generally be a factor of. 2 or more less than a PRA fragility median value).
During the review process, other important components were found for which a low HCLPF would result or insufficient data was available to determine the HCLPF. Maine Yankee either upgraded these items imediately, or agreed to make changes during the upcoming plant outages.
The analysis took credit for the upgrades listed below:
Description of Change o
Diesel fuel day tank anchorage upgrade o
Control Room cooler anchorage upgrade o
' Welding cart / gas bottle tiedown i
o' Security lighting tiedown
]
i o
Main control board alarm tiedown o
Strengthen a blockwall o
Upgrade anchors for two fans o
Install internal anchors for two transformers o
Replace lead-antimony safety class batteries l
l i
16 The Maine Yankee review demonstrated that the Seismic Design Mergins Program l
methodology can be successfully implemented and be used to solve seismic licensing issues.
Plant seismic vulnerabilities were found and upgraded as a result of the review.
The importance of both peer review and utility cooperation was clearly shown.
The experience of performing the review showed some areas where the method-oigy's guidance was lacking.
The resolutions of these procedural questions will serve as improvement to the review guidelines.
Areas of improvement in-clude:
1.
description of the seismic margins earthquake 2.
the integration with nonseismic failures 3.
use of equipment qualification data 4.
HCLPF calculation techniques 5.
sampling The next major task of the SDMP will be a trial plant review of a boiling water reactor (BWR).
Because of many advantages, this is planned to be done as a cooperative effort between EPRI and NRC.
EPRI has extended the scope of the current SDNP methodology, adding guidance for new considerations such as soil failure and relay chatter.
It has also made changes to make the methodolgy more utility-oriented.
An NPC panel of consultants has reviewed the EPRI methodology and the NRC is currently in the process of issuing an endorsement of its use in the trial plant review.
Negotiations are ongoing with the plant owner of a candidate BWR.
Other smaller SDMP tasks now underway include a comparison of the two methods to calculate component HCLPF's; the Conservative Deterministic Failure Margins (CDFM) approach and the Fragility Analysis approach.
Another study is looking at ways to extend seismic margin review guidance to better analyze plant dam?ge states (providing radioactive release insights) and to improve the considera-tion of human factors and nonseismic failures.
O
17 It.should be noted that the SDMP is overseen by the NRC Working Group on Seismic Design Margins which is comprised of representatives from both the i
Office of Nuclear Regulatory Research (RES) and Nuclear Reactor Regulation (NRR).
This group is providing recommendations on the research program itself and how its products will integrate with other seismic policy action, such as the USI-A46 effort and plans to implement the Severe Accident Policy Statement.
l e
I i
i 1
I 1
'I 18 Reactor Containment Program-a Structural Tests Activity has. continued on a set of programs whose objectives are to provide the data ' base. required for the qualification of methods for predicting the response of LWR containment buildings during severe accidents (those beyond design basis. events) and extreme earthquakes.
Thit set of programs is examining the modes of containment failure that would result in the release of radioactive materials beyond the containment boundary.
These modes include
- structural failure of the containment building, leakage through or past the penetrations (electrical or mechanical), failure of containment isolation systems, or failure of the basemat by the molten reactor core.
A 1:6 scale model of a reinforced concrete containment was tested to failure in July 1986.
.The containment was designed and built in conformance with Boiler and Pressure Vessel Code of the American Society of Mechanical Engineers, just as are actual containments.
The model was 22 feet in diameter and 37 feet in height and included representative features such as four major penetrations (two airlocks and two equipment hatches) and several smaller penetrations that passed both separately and in clusters through the contain-ment wall.
The containment had a design pressure of 46 psig.
Pressure was increased in steps until failure occurred at 145 psig.
At that point a major tear, 20 inches long, developed in the liner.
Leakage through that tear overwhelmed the ability of the pressurization system.
Additional minor tears were also present in the liner but the concrete outer structure, although visibly cracked, did not show great distress.
Post-test analyses will focus on the measurements of strain and displacement taken at each discrete pressure step to evaluate the accuracy of pre-test pre-dictions made using different analytical techniques.
Nine organizations, in-cluding three from the U.S., three from the United Kingdom, and one each from
h i
L..
l l
L j'
6 19 France, Italy, and Germany performed pre-test predictions and will participate
)
in-the post-test evaluation.
Twelve hundred channels of data were recorded during the experiment and analysis of these data will permit assessment of the accuracy of the predictive methods.
Personnel Airlock Test l
A full-size personnel airlock, which was obtained from a cancelled nuclear power plant, was tested at Chicago Bridge & Iron Research and Development Center, Plainfield, Illinois under contract to Sandia National Laboratories.
i The work is part of the Containment Integrity Research that is sponsored by the U.S.NRC.
The objective of the tests was to obtain structural data on the behavior of an air 1cck, especially the sealing surfaces under severe accident conditions.
It was anticipated that leakage would not occur unless relative deformations between the sealing surfaces were developed and performance of the seal material was compromised.
The sealing surfaces could separate because of a mismatch in the out-of-plane displacements of the door and bulkhead, which resist internal pressure through bending action.
The per-formance of the seal material may be compromised in two ways:
(1) a loss of resiliency associated with thermal or radiation aging, and (2) degradation associated with exposure to very high temperatures.
Two of the four tests to be conducted were conducted on June 30 and July 2, 1987, with satisfactory results.
In the first test, the inner and outer doors were without gaskets and pressurized to 69 psig (1.15 times the design) at room temperature.
Leakage was measured at 45 scfm and 35 scfm on the inner and outer doors, respectively.
In the second test, the inner and outer doors had aged gaskets installed and pressurized to 69 psig (this pressure was held for 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br />), no leakage was measured.
The next test scheduled for early FY88 will involve higher pressure and temperature typical of severe accident condi-tions in BWR and PWR containments.
_ - _ _ _ = _ _ _