ML20199F411
| ML20199F411 | |
| Person / Time | |
|---|---|
| Site: | Waterford  |
| Issue date: | 01/29/1998 |
| From: | Dugger C ENTERGY OPERATIONS, INC. |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| Shared Package | |
| ML20070L320 | List: |
| References | |
| W3F1-98-0016, W3F1-98-16, NUDOCS 9802040008 | |
| Download: ML20199F411 (23) | |
Text
.
g Ent gy Oper:tions,Inc.
Of Mona. L 4 700C6-0751 Tel 504 739 M60 V er s o ans w-w 3 W3F1-98-0016 A4.05 PR ATTACHMENTS 1, 3, 4, 5, 6, 7,12 AN D 13 CONTAIN PROPRIETARY INFORMATION January 29,1998 U.S. Nuclear Regulatory Commission Attn.: Document Control Desk Washington, DC 20555 Gubject: Waterford 3 SES Docket No. 50-382 License No. NPF-38 Request Fcr Additional information (RAI) And Errata Pages Regarding Technical Specification Change Request NPF-38-193 i Gentlemen:
By letter dated March 27,1997, Waterford 3 proposed to amend O[erating License
-* NPF-38 to increase the Spent Fuel Pool storage capacity and increase the maximum fuel enrichment. The NRC review staff requested additional information, in their letter dated December 31,1907 regarding the proposed changes. This information is included in the enclosure entitle.d " Additional Information Regarding Technical Specification Change Request NPF-38-193."
Model input data, as described in RAI Item 1c and item 2, are being provided on the enclosed five 3.5" diskettes.
This submittal also includes (as Attachment 14) two errata change pages, one for Section 8.6.1.2 and one for Table 8.1, for Holtec International Report Hi-971628, which is an enclosure to W3F1-97-0061, dated March 27,1997. The errata change pages are being submitted to assure consistency between the License Amendment request and the calculations performed for the Spent Fuel Pool structural analysis.
In section 8.6.1.2, the value for Young's Modulus for concrete was revised from 3.85 x 10' psi to 3.834 x 10' psi to correct a transposition error (the correct value was used in the calculation). In Table 8.1, the moment numbers and resulting safety
[
f )
,eo204ooos 98o12, PDR m\
ADOCK 0500o382
\ @fo' g y mm w g
.- Q ,
)&(A[1d>
P PDR ,
\
k _ _ - - - - _ - - - - - - _
L L
l Request For Additional Information (RAl) and Errata Pages Regarding Technical Specification Change, Request NPF-38-193 W3F1-98-0016 .
Page 2 January 29,1998 factor for the Spent Fuel Pool east wall were revised, with the safety factor changing from 1.95 to 1.86. This change is inconsequential because the safety factor remains
' well above the limiting safety factor,1.11, which is for the west wall and is
= unchanged.
The information in this submittal has no effect on the previously provided
- determination of no significant hazards, Please note that Attachments 1,3,4,5,_6,7,12 and 13 of the enclosure contain information that is considered proprietary pursuant to 10CFR2.790. In this regard,
- EOl requests that these Attachments be withheld from public viewing. Please note that the respective Holtec International affidavits, pursuant to 10CFR2.790, precede the materialin these Attachments.
Should you have any questions or comments concerning the additional information, i please contact Roy Prados at (504) 739-6632. {
Very truly yours, C.M. Dugger
./ .I y Vice President, Operations Waterford 3 CMD/RWP/tmm
Enclosures:
' Affidavits Attachments Diskettes
Request For Additional Information (RAI) and Errata Pages Regarding Technical Specification Change Request NPF-38-193 W3F1-98-0016 Page 3 January 20.1998 cc: E.W. Merschoff, NRC Region IV C.P. Patel, NRC-NRR -
NRC Resident inspectors Office (w/o attachments)
J. Smith N.S. Reynolds Administrator Radiation Protection Division ,
(State of Louisiana)
American Nuclear Insurers J
e .
UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION In the matter of )
)
Entergy Operations, incorporated ) Docket No. 50-382 Waterford 3 Steam Electric Station )
AFFIDAVIT ,,
Charles Marshall Dugger, being duly sworn, hereby deposes and says that he is Vice President Operations - Waterford 3 of Entergy Operations, incorporated; that he is duly authorized to sign and file with the Nuclear Regulatory Commission the
- attached Additional Information and Errata Pages Regarding Technical Specification Change Request NPF-38-193; that he is familiar with the content thereof; and that the matters set forth therein are true and correct to the best of his knowledge, information and belief.
( - & '( ( _ W A~
Charles Marshall Dugger II Vice President Operations - Watedord 3 STATE OF LOUISIANA )
) ss PARISH OF ST, CHARLES )
Subscribed and sworn to before me, a N,otary Public in and for the Parish and State above named this > 9 s i day of _ k r. ~ . m._.m .1998.
! {
.~
- L.. , u.
Notary Public My Commission expires a' 4, M
ADDITIONAL INFORMATION REGARDING TECHNICAL SPECIFICATION CHANGE REQUEST NPF-38193 Item 1 In the staff's first request for additional information (RAI), you were requested to provide detailed discussions as to how the dynamic fluid coupling was modeled in your analyses using the DYNARACK code. You provided a response in Reference 1. The staff understands that your approach may be one approximate way of modeling the fluid-structure interaction for the single-rack analysis. However, the staff has reservations as to whether it is applicable for accurately modeling the structure-fluid-structure interaction of the multi-rack analysis considering the fact that the fluid and structural responses of the multi-rack system are somewhat random and independent between rek to rack in a three dimensional space under safe shutdown earthquake (SSE) loading. You are requisted to provide the following:
(a) Detailed explanation of how the dynamic fluid coupling was considered for modeling the structure-fluid-structure interaction in the multi-rack analysis and technical justifications for using such modeling, ,
(b) What were the initial gaps between the racks used to calculate hydrodynamic masses? Were those initial gaps assumed to be constant during the rack movements under a SSE7 If constant gaps were assumed, justify your assumption. If the gaps were not constant during a SSE, how did you incorporate the changed hydrodynamic masses in your calculations?
(c) Submit a model drawing with complete inforniation (i.e., element and node number with their locations in the global coordinate system, spacing between racks, etc.). Also, submit the complete DYNARACK input data used for the 3-D multi-rack analysis in ASCll on a 3.5-inch diskette.
Response 1 The following provides an explanation of the DYNARACK methodology for inclusion of fluid coupling in the solution procers. This explanation is from both a technica! and historical perspective and is in addition to the response provided in our October 23, 1997 submittal.
The phenomenon of fluid coupling between rectangular planform structures was sparsely investigated until the 1980s. Fritz's classical paper (ca.1972; see Reference 1
l l
6.5.3 of Holtec International Report Hl 971628 submitted as part of Technical Specification Change Request NPF-38-193) was used in the earliest version of DYNARACK to model rack to surrounding fluid effects in the so-called single rack 3-D simulation. Enrico Feimi Unit 2 (ca.1980) and Quad Cities Units 1 and 2 (ca.1982) were licensed using this early version. The current version of the code continues to use the same Fritz fluid coupling terms. The Fermi 2 and Quad Cities 1 and 2 submittals were the first rerack applications wherein a rack module was analyzed using the 3-D time-history technique. The adoption of a nonlinear time-history approach helped quantify the motion of a rock under a 3-D earthquake event and as a byproduct, also served to demonstrate that solutions using the Response Spectrum Method (which assumes a linear structure) can be woefully unconservative. Practically all rerack licensing submittals from 1980 on utilized the 3 D time history method. While the nonlinear 3 D time-history method was a significant improvement over the Responso Spectrum Method (linear) approach, it was limited because only one rack could be modeled in any simulation. The analyst had to assume the behavior of the adjacent racks. Models, which postulated a pribri the behavior of the contiguous racks in the vicinity of the rack being analyzed were developed and deployed in safety analyses.
The two most commonly used models were the so-called " opposed phase" model and the "in-phase" model, the former used almost exclusively to predict inter-rack impacts until 1985. Holtec International Position Paper WS 115 (proprietary) included herein as Attachment 1, provides a summary description of these early single rack 3-D models.
The inadequacy of the single rack models (albeit nonlinear) to preoict the response of a grouping of submerged racks arrayed in close proximity bscame an object of prolonged intervenors' contention in the reracking of PG&E's Diablo Canyon units in 1986-87.
Holtec International, with active invigilation from the NRC, developed a 2-D multi-rack model for the Diablo Canyon racks; this model helped answer ir.tervention issues, permitting PG&E to rerack. NRC experts testified in support of the veracity of the 2-D multi-rack dynamic models at the Diablo Canyon ASLB hearingt in Pismo Beach, California in June 1987.
The Diablo Canyon intervention spurred Holtec International to develop what later came to be known as the Whole Pool Multi-Rack (WPMR) analysis. A key ingredient in the WPMR analysis is quantification of the hydrodynamic coupling effect that couples the motion of every rack with every other rack in the pool. In 1987, Dr. Burton Paul (Professor Emeritus, University of Pennsylvania) developed a fluid mechani:s formulation using Kelvin's recirculation theorem which provided the fluid coupling matrix (2N x 2N for a pool containing N racks).
For example, with reference to Attachment 2, where an array of N (N = 16) two-dimensional bodies (each with two degrees of freedom) is illustrated, the dynamic equilibrium of the i-th mass in the x-direction can be written as 2
i
N ,
(m, + M,, }( + E ;AI,( + N,y, = Q,,(t)
J'l In the above equation, mii s the mass of body i(i = 1,2...N), and g, is the x-direction acceleration vector of body I. M, and N denote the " virtual" mass effects of body J on body i in the two directions of motion. The second derivative of y with respect to time represents the acceleration in the y-direction.
The terms Ma are functions of the shape and size of the bodies (and the container boundary) and, most important, the size of the inter-body gaps. Mc are analytically derived coefficients. Q,, represents the so-called generalized torce* that may be an amalgam of all externally applied loads on the mass i in the x-direction. The above equation for mass i in x-direction translational motions can be written for all degrees of freedom and for all masses. The resulting second order matrix differential equation contains a fully populated mass matrix (in contrast, dynamic equations without multi-body fluid coupling will have only diagonal non-zero terms).
The above exposition explains the inclusion of fluid coupling in a multi-body fluid coupled problem using a simplified planar motion case. This explanation pre vides the building blocks to explain the more complicated formulation needed to simulate freestanding racks Dr. Paul's formulation is documented in a series of four reports written for PG&E in 1987, which are included as Attachments 3,4,5, and 6. The Paul multi-body fluid coupling theory conservatively assumes the flow of water to be irrotational (inviscid) and assumes that no energy losses (due to form drag, turbulence, etc.) occur. NRC personnel reviewed this formulation in the course of their audit of the Diablo Canyon rerack (ca.1987) and subsequently testifieriin the ASLB hearings on this matter, as stated above.
While the ASLB, NRC, and NRC consultants (Brookhaven National Laboratory and Franklin Research Center) all endorsed tne Paul multi-body coupling model as an appropriate and conservative representation, the theory was still just a theory.
Recognizing this perceptual weakness, Holtec International and Northeast Utilities undertook an experimental program in 1988 to benchmark the theory. The experiment consisted of subjecting a scale model of racks (from or,a to four at one time in the tank) to a two-dimensional excitation on a shake table at a QA qualified laboratory in Waltham, Massachusetts.
g The Paul multi-body coupling formulation, coded in QA validated preprocessors to DYNARACK, was compared against the test data (over 100 separate tests were run).
The results are documented in Problem 10 of Holtec International Report HI-91700,
- which was previously provided in the Waterford 3 October 23,1997 submittal. As discussed in more detail in the response to RAI Item 2 below, Holtec International Repo t Hl-97100 was revised (revised pages included as Attachment 7) to provide additional confirmatory information for DYNARACK. The experimental benchmark work 3
validated Paul's fluid mechanics model and showed that the theoretical model (which noglects viscosity offects)is consistently bounded by the test data. This experimentally verified multi-body fluid coupling is the central underpinning of the DYNARACK WPMR solution that has been employed in every fuel rack license application since 1989 (See table 6.2.1 of Holtec international Report HI-971628). The DYNARACK 3 D WPMR solution has been found to predict much greater rack displacements and rotations than the previously used 3 D single rack results (see Attachment 8).
In general, the advance from linearized analyses (response spectrum) in the late 1970s to the single rack 3 D analyses until the mid 1980s and, finally, to the 3 D WPMR analysis in the past ten years has, at each technology evolution stage, led to an increase in the computed rack response. The stresses and displacements computed by the DYNARACK 3 D WPMR analysis for the Waterford 3 racks, in other words, are greater (and more conservative) than the docketed work on similar instances from 15 years ago. The conservatism's built into the WPMR solution arises from several simplifying assumptions explicitly intended to estaNish an upper bound on the results, namely:
a) In contrast to the single rack 3 D models, the fluid forces on every rack in the pool consist of the aggregate of fluid coupling effects from all other racks located in the pool. No assumptions on the motion of racks need be made a priori; the motion of each rack in the poolis a result of the analysis.
b) The fluid coupling terms are promised on classical fluid mechanics; they are not derived from empirical reasoning. Further, the fluid drag and viscosity effects, collectively referred to as " fluid damping", are neglected. in short, while the transfer of fluid kinetic energy to the racks helps accentuate their motion, there is no subtraction of energy through damping or other means.
c) in the Waterford 3 rack simulations, the dynamic model for the fuel assemblies in a rack assumes that all fuel assemblies within a rack move in unison. Work in quantifying the effect of discordant rattling of fuel assemblies within a rack has shown that the " unified motion" assumption exaggerates the rack response by 25% to 60%, depending on the rack geometry details and earthquake harmonics.
d) The rack to-rsck and rack-to-wali gaps are taken as the initial nominal values.
During th3 earthquake, these gaps will change through the time-history duration.
The fluid coupling matrix should be recomputed at each time-step with the concomitant gap distribution. The inversion of the mass matrix at each time-step (there are over four million time-steps in a typical WPMR run) would, even today, mandato use of a supercomputer. Fortunately, neglect of this so-called nonlinear fluid coupling effect is a conservative assumption. This fact is rigorously proven in a paper by Drs. Soler and Singh entitled " Dynamic Coupling in a Closely Spaced Two Body System Vibrating in a Liquid Medium: The Case of Fuel Racks", published in 1982 (Attachment 3 of the Waterford 3 October 23,1997 4
v submittal). The only docket where recourse to the nonlinear fluid coupling was deemed essential was Vogt!e Unit 2 (in 1988) where the margin inherent in the nonlinear fluid effect, published in the above mentioned paper, was reaffirmed.
Nonlinear fluid coupling is not employed in this present application which imputes over 15% margin in the computed rack response.
In summary, the WPMR analysis utilizes a fluid coupling formulation that is theoretically derived (without empiricism) and experimentally validated. The assumptions built into the DYNARACK formulation are aimed to demonstrably exaggerate the response of all racks in the pool simulated in one comprehensive model. The DYNARACK fluid coupling model is not approximate or empirical at all, as suggested in the RAl. A mathematical explanation of the manner in which fluid coupling is considered in the solution is provided in response to item 1(a) below.
Response is DYNARACK, developed in the late 1970s and continuously updated since that time to incorporate technology advances such as multi-body fluid coupling, is a Code based on the Component Element Method (CEM). The chief merit of the CEM is its ability to simulate friction, impact, and other nonlinear dynamic events with accuracy. The high-density racks designed by Holtec International are ideally tailored for the CEM-based Code because of their honeycomb construction (HCC). Through the interconnection of the boxes, the HCC rack essentially simulates a multi flange beam. The beam characteristics of the rack (including shear, flexure, and torsion effects) are appropriately modeled in DYNARACK using the classical CEM " beam spring".
However, the rack is not rendered into a " stick" model, as implied by the staff's RAl.
Rather, each rack is modeled as a prismatic 3 D structure with support pedestal locations and the fuel assembly aggregate locations set to coincide with their respective center of gravity axes. The rattling between the fuel and storage cells is simulated in exactly the same manner as it would be experienced in nature: namely, impact at any of the four facing walls followed by rebound and impact at the opposite wall. Similarly, the rack pedestals can lift off or slide as the instantaneous dynamic equilibrium would dictate throughout the seismic event. The rack structure cari undergo overturning, bending, twist, and other dynamic motion modes as determined by the interaction between the seismic (inertia) impact, friction, and fluid coupling forces. Hydrodynamic loads, which can be quite significant, are included in a comprehensive manner, as explained below.
As discussed in the foregoing, the fluid coupling offect renders the mass matrix into a fully populated matrix. In modeling the fuel rack as a multi-degree of freedom structure, the following key considerations are significant:
5 J
4 a) Over 70% of the mass of the loaded rack consist of fuel assemblies, which are unattached to the rack, and resemble a loose bundle of slender thin-walled tubes (high mass, low frequency).
b) In honeycomb construction (HCC) racks, as shown in a 1984 Nuclear Enaineerina and Desian paper (Attachment 9), the rack behaves like a stiff elongated box beam (End Connected Construction racks, built 20 years ago and now obsolescent beheve as a beam and bar assemblage).
Since the Waterford 3 racks under inertial loading have overall structural characteristics of a multi-flange beam, it is computationally wasteful (and, as we explain later, numerically hazardous) to model such a structure as a plate assemblage. The DYNARACK dynamic model preserves the numerical stability of the physical problem by representing the rack structure by an equivalent flexural and shear resisting
" component element"(in the terminology of the Component Element Method in Reference [3).
A detailed discussion of the formation of the fluid mass matrix is presented below.
The problem to be investigated is shown in the figure in Attachment 2, which shows an orthogonal array of 16 rectangles which represent a unit depth of the 16 spent fuel racks in the Waterford 3 Spent Fuel Pool. The rectangles are surrounded by narrow fluid filled channels whose width is much smaller than the characteristic length or width of any of the racks. The spent fuel pool walls are shown enclosing the entire array of racks.
The dimensions of the channels are such that an assumption of unidirectional fluid flow in a channel is an engineering assumption consistent with classical fluid mechanics principles.
, Ench rectangular body (fuel rack) has horizontal velocity components U and V parallel to the x and y exes, and that the channels are parallel to either the x or y axes. The pool walls are also assumed to move.
It is conservatively assumed that the channels are filled with an inviscid,
~
incompressible fluid. Due to a seismic event, the pool walls and the spent fuel racks are subject to inertia forces that induce motion to the rectangular racks and to the wall.
This motion causes the channel widths to depart from their initial nominal values and causes flow to occur in each of the channels. Because all of the channels are connected, the equations of classical fluid mechanics can be used to establish the fluid velocity (hence, the fluid kinetic energy) in terms of the motion of the spent fuel racks.
For the case in question, there are 40 channels of fluid identified. The figure in Attachment 10 shows a typical rack (box) with four adjacent boxes and fluid and box 6
_ _ _ _ - - . - . - - )
-._ c . --
I velocities identified. The condition of vanishing circulation around the box may be expressed as P = h v,ds = 0 or -
s/2 6/2 l(u,-u)h*+
r ,
(v,-vdh=0 1
-el) -6/2 where the subscripts (L, R, B, T) refer to the left, right, bottom, and top channels, respectively; 4, y are local axes parallel to x and y, and u, y are velocities parallel to 4, y.
Continuity within each channel gives an equation for the fluid velocity as w = w. . ( ) s where w represents the velocity along the axis of a channel, wm represents the mean velocity in the channel, s is along either 4 or y, and h is the rate of increase of channel width. For example h, = Up - U From the figure in Attachment 10, four equations for us , ut, va, vt in terms of the respective mean channel velocities, can be developed so that the circulation equation becomes a (Un. - Utm) + b (v% va) = 0 One such circulation equation exists for each spent fuel rack rectangle. We see that the velocity in any channel is determined in terms of the adjacent rack velocities if we can determine the mean fluid velocity in each of the 40 channels. Circulation gives 16 equations. The remaining equations are obtained by enforcing continuity at each 7
junction as shown in the figure in Attachment 11. Enforcing continuity at each of the 25 junctions gives 25 equations of the general form
[hn- L6 - o where w is the mid-length mean velocity in a connecting channel of length L and li is the relative normal velocity at which the walls move. The summation covers all channels that meet at the node in question. The sign indicator o = 11 is associated with flow from a channel either into or out of a junct!on.
There are a total of 26 + 16 = 41 equations which can be formally written; one circulation equation, however, is not independent of the other. This reflects the fact that the sum total of the 25 circulation equations must also equal zero, representing that circulation around a path enclosing all racks is equal to zero. Thus, there are exactly 40 independent algebraic equations to determine the 40 unknown mean velocities in this configuration.
Once the velocities are determined in terms of the rack motion, the kinetic energy can be written and the fluid mass matrix identified using the Holtec International QA.
validated pre-processor program CHANBP6. The fluid mass matrix is subsequently apportioned between the upper and lower portions of the actual rack in a manner consistent with the assumed rack deformation shape as a function of height in each of the two horizontal directions. This operation is performed by the Holtec International QA-validated pre-processor code VMCHANGE Finally, structural mass effects and the hydrodynamic effect from fluid within the narrow annulus in each cell between the fuel assembly and the cell wall is incorporated using the Holtec International QA validated pre-processor code MULTl122.
Response 1b The initial inter-rack and rack-to-wall gaps are provided in Figure 2.1.1 of Holtec International Report HI 971628.
These gaps, which directly figure in the computation of fluid mass effects in fluid coupling matrix, are assumed to apply for the entire duration of the earthquake. In reality, the gaps change throughout the seismic event and a rigorous analysis would require that the mass matrix be recomputed at every time-step. Besides being numerically impractical, such refinement in the solution would reduce the conservatism in the computed results, as discussed earlier.
The time variations in the inter-rack and rack-to-wall gaps are tracked for the duration of the earthquake. Closure of any gap at any location results in activation of the compression gap spring at that location. The loads registered in the gap spring quantify the collision force at that location. The fuel-to-storage cell rattling forces and 8
l
fack pedestal to pool liner impact forces (in the event of pedestal lif t off) are typical examples of collision forces that are ubiquitous in rack seismic simulations. The nonlinear contact springs in DYNARACK simulate these " varying gsp" events during seismic events using an unconditionally convergent algorithm.
In summary, the Whole Pool Multi Rack (WPMR) analysis is a geometrically nonlinear formulation in all respects (lift-off, sliding, friction, impact, etc.), except in the computation of the fluid coupling matrix, which is besed on the nominal (initial) inter-body gaps.
Rosconse ic Table 6.5.2 and Figures 6.5.6 through 6.5.11 of Holtec International Report Hl-971628 Identify the numbering and location of all of the gap spring elements for each model. A complete set of the input files are enclosed along with this response on four 3.5" diskettes. Flow charts, included as Attachment 12, from the DYNARACK user's manual are also included to assist in understanding the included input files. The flowcharts indicate the various preprocessor programs which are required for preparation of the input files, item 2 In the staff's first RAI, you wve requested to provide the results of any existing experimental studies that verify the correct or adequate simulation of the fluid coupling utilized in the DYNARACK analyses for the fuel assemblies, racks and walls. You provided a conparison study between the results of a DYNARACK analysis and an experimental test. The staff reviewed the comparison study and concluded that the study is not sufficient enough to demonstrate the adequacy of the DYNARACK code -
used to simulate the dynamic fluid coupling and the structure-fluid-structure interactions due to; (i) the conditions of the experimental setup (i.e., boundary conditions, dimensions and shapes of the structure, application of the load, etc.) are so different from the real conditions of the rack structures and (ii) very limited test data were obtained and presented. In order to further demonstrate the adequacy of your code, you are requested to provide a comparison study between the results of the DYNARACK predictions and the experimental tests shown in References 2 and 3.
Also,- submit the DYNARACK input data in ASCll on a 3.5-inch diskette for the comparison analysis.
Enponse 2 Holtec International Report HI-88243 was submitted in 1992 for D.C. Cook Unit 1 (Docket No. 50-381) and Unit 2 (Docket No. 50-316). This report by Dr. Burton Paul reports comparison of DYNARACK fluid coupling formulation with over 100 experiments '
carried out in an independent laboratory under a 10CFR50 Appendix B program.
9
. c _ _ _ . _ ._ _ _ _ .
These tests were performed with the sole purpose of validating the multi-body fluid coupling formulation based on Kelvin's recirculation theorem in classical fluid mechanics. These experiments, to our knowledge, are the only multi body fluid coupling tests conducted and recorded under a rigorous OA program. The participating bodies used in the tests were carefully scaled to simulate rectangular planform fuel racks. The tests were run under a wide range of seismic frequencies to sort out effects of spurious effects such as sloshing in the tank, and to establish that the fluid coupling matrix is independent of the frequency content of the impressed loading.
The references cited by the staff predate the above study by a decade. The University of Akron tests were performed under the sponsorship of the predecessor company of U.S. Tool & Die, Inc., in the time when racks were still being analyzed using the Responso Spectrum Method.
Reference [1] cited by the staff was reviewed, which includes the results of Reference l
[2]. We note that we have no knowledge of the quality assurance procedures (if any) l employed. Nevertheless, we have carefully considered the relevance of the results to the real problem of dynamic analysis of spent fuel racks. With respect to the staff's j Reference [1], we note that the theoretical model developed in that paper is exactly that used in the Holtec International WPMR analysis when the Holtec International mass matrix is reduced to a single rectangular solid block surrounded by four rigid (pool) walls. That is, the work by Scavuzzo is a special case of a Holtec International WPMR analysis for a spent fuel pool containing a single spent fuel rack.
The Holtoc International WPMR fluid mass matrix for many racks in the pool is obtained by applying the same classical principles of fluid continuity, momentum balance, and circulation, to a case of many rectangular bodies in the pool with multi-connected narrow fluid channels.
4 The experimental work performed by Scavuzzo, et al Reference (1), does not attempt to model a free standing rack since many rack structures of that vintage were not free-standing. The experimental test is equivalent to a single spring-mass-damper subject to a forced harmonic oscillation while submerged, if one accepts the fact that the fluid model used by Scavuzzo is a limiting case of the more general Holtec International
- formulation, then the good agreement of theory with experiment for the single " rack" modeled experimentally serves as additional confirmation that the Holtec International theoretical hydrodynamic mass model, which is identical to tho Scavuzzo model (for a single rack) is reproducible by experiment.
Nevertheless, in conformance with the staff's request, we are using the data supplied by Scavuzzo to simulate the experiment using the pre-processor CHANBP6 and the solver DYNARACK. The results of this comparison have been incorporated as Problem 12 in the Holtec International validation manual for DYNARACK (see the revised pages 10
l l
of Holtec international Report HI 91700, included as Attachment 7) as an additional confirmatory test The requested DYNARACK input data for the comparison analysis is enclosed on a 3.5" diskette.
Item 3 I in the staff's first RAI, you were requested to provide technical explanation as to the somewhat inconsistent and irregular displacement predictions shown in a table on page 6-24 of Holtec International's Report HI-971628. You provided possible attributes, but did not fully address the staff's concerns. Provide detailed qualitative and quantitative technical discussions based on the results obtained (i.e., pedestal force, hydrodynamic force, etc.) and other studies (i.e., parametric sensitive study, etc.). Also, provide predicted maximum displacements in x., y , and z-directions for the cases of EVENT (DBE and OBE) and COF (0.8,0.2 and RANDOM) for all racks in the spent fuel pool, the cask pit and the refueling canal.
Response 3 The predicted results are typical of a kinematically non-linear problem and contain no inconsistencies or irregularities. The September 25,1997 RAI requested a discussion of the factors causing large differences in the (DBE/CBE) ratios of displacements for the Cask Pit vs. the Spent Fuel Pool. The response indicated that the higher rack aspect (width / length) ratios and the larger rack to wall gaps are the principal factors which explain the increased (DBE /OBE ) displacement ratios for the Cask Pit. These attributes of the physical problems explain the results quite well.
Please realize that rack displacements reported in the table on page 6-24 of Holtec International Report Hl-971628 reflect the maximum displacements polled from:
a) All racks in the simulation ,
b) At every time instant throughout the entire time history duration ,
c) At all corners of the top of the racks.
As requested by the staff, a more complete listing of the maximum displacements for every rack under each of the simulations is provided in the tables included as Attachment 13. The values from_the table on page 6-24 of Holtec International Report HI 971628 were extracted from the data contained in Attachment 13. Each of the displacement values provided in Attachment 13 represents the bounding displacements for all four corners of the rack and for all time instants. These values are excerpts from 11
the archived DYNARACK solver summary output. The complete raw data reported by DYNARACK contains the displacements for every corner for every rock in each simulation at every time instant. However, the complete output is not provided, since each simulation output displacement file is very large (contains several mega-bytes of data).
The predicted rack corner displacements are determined by DYNARACK through superposition of the displacaments resulting from rack bending, sliding, twisting, and tipping (if present). The analysis is highly non-linear due to the physics of the problem wherein:
a) The great bulk of the structure undergoing the seismic excitation is free to rattle in a confined space.
b) The fluid coupling forces between the racks, between racks and walls, and between stored fuel and rack cell walls interacts with friction and impact forces, c) The impact spring elements at the pedestals, rack boundaries, and fuel masses (which are only loaded upon elimination of the gaps modeled at these locations) are inherently non-linear.
A special feature of rack kinematics makes the proportionality of rack movement to the seismic excitation wholly inapplicable. It can be explained as follows:
During the dynamic simulations one or more of the rack module pedestals are often momentarily lifted from the floor. Under instances of large accelerations, only one pedestal may be in contact with the floor allowing the rack to rotate freely. This rotation would continue until opposing forces are realized from pedestals after restoring floor contact, from fluid coupling with adjacent racks or walls, or from rack impacts with adjacent racks or walls. Because of this phenomenon, comparisons of pedestal loadings or fluid hydrodynamic pressures between DBE and OBE conditions to substantiate / corroborate the displacement ratios under similar conditions are not appropriate and are not included in this response.
As the above description suggests, the ratios (DBE /OBE ) of displacements for the Cask Pit vs. the Spent Fuel Pool are not necessarily expected to be similar under these complex model characteristics. Expectations that the displacement ratios should be similar for these two conditions would be similar to the proposition that the expected ratio of displacements or loadings between OBE and DBE should be similar to the ratio of the earthquake spectral peak or ZPA values. This is also not the case because of the inherent features of all non-linear analyses.
As indicated in the Waterford 3 October 23,1997 submittal, the maximum displacements for OBE and DBE simulations (with all other conditions the same) may not even occur in the same rack or even in the same direction. Even if the same 12
)
Identical points and directions were chosen, with only the earthquake intensity varying, it is not expected that ratios would be consistent between the two pools as suggested, due to the different attributes (or parameters) discussed in the foregoing.
It should be noted that the DBE and OBE earthquake accelerogram sets used for all three pools are identical. Additionally, there were no modeling changes for any of the pools between earthquake simulations. Therefore, we submit that the reported results for all of the simulations are technically robust and devoid of any discrepancies.
itgm 4 In the staff's first RAI, you were requested to provide your plan regarding installation of support system to minimize displacement and impact force between the rack to-pool wall and rack-to-rack since your analysis shows that there are rack-to-pool wall and rack-to-rack impacts. You provided a response that you are not planning to install any support system to eliminate the rack impacts based on the results of the DYNARACK analysis. Provide technical discussions of why pur simple DYNARACK analysis is the most conservative approach and how it assures no further structural safety considerations are required. Justify your discussions by confirming your DYNARACK predictions with actual rack experimental test results.
Response 4 The high-density racks for Waterford 3 are designed for installation as freestanding structures. Numerous plants have installed racks in freestanding mode where the licensing basis analyses had indicated inter-rack and/or rack-to-wall impacts.
Examples from the past are: Byron 1 and 2 (1988, Docket Nos.50-454 and 50-455);
Braidwood 1 ano 2 (1988, Docket IJos. 50-456 and 50-457): St. Lucie Unit 1 (1987, Docket No. 50-335); Diablo Canyon 1 and 2 (1987, Docket Nos. 50-275 and 50-323);
and Ft. Calhoun (1994, Docket No. 50-285).
In all of the above dockets (which are typical of the industry) the racks are built as honeycomb construction (HCC) structures where all cells are connected along the entire length of the rack such that the effect of impact loading is limited to the local impacted region. In End Connected Construction racks, typical of the late 1970s, the impact loads produce primary or global stresses because the individual cells were connected only at the top and bottom of the rack.
Impact during seismic events is a natural corollary of a freestanding structure. At a minimum, during seismic events, the fuel assemblies rattle inside the storage cavity and the rack pedestals' compression forces change with time. Pedestal lift-off and impact are also more of a rule than an exception. Rack-to-rack impact is also observed in a significant number of cases. None of these impact forces would lead to an adverse 13 1
_ . _ - - _ - 1
1 effect on safety if their magnitudes are conservatively quantified and if their concequences to the rack structure are carefully examined. The Waterford 3 racks have been subjected to an exhaustive set of dynamic and stress analyses to ensure that the safety conclusions stated in HI 971628 are accurate. Where rack-to-rack impacts occur, there is no effect on the structure in the region of active fuel; the effects from the impact forces are accounted for in the subsequent dynamic response of the rack.
As to the veracity of the DYNARACK results, it should be noted that this Code has been subject to an in-depth NRC assessment in the 1983-84 period in the course of V.C. Summer (Docket No. 50-395) and Oyster Creek (Docket No. 50 219) rerack safety evaluations. In its 1983-84 evaluations, the NRC utilized its internal staff aided by university-based consultants. Later, in 1986-87, Diablo Canyon ASLB hearings led to another series of NRC investigations aided by the Brookhaven National Laboratory. In the 199193 period, another round of intensive reviews of DYNARACK occurred in the context of J.A. FitzPatrick (Docket No. 50-333) and D.C. Cook Uni;s 1 and 2 (Docket Nos. 50-315 and 50-316) reracks. These three intensive NRC assessments have been in addition to the standard reviews that occur on every docket.
On the FitzPatrick and Cook dockets, Holtec International provided conclusive demonstration that DYNARACK is capable of capturing abstruse nonlinear effects such as subharmonic resonance. The DYNARACK validation manual (Holtec International Report HI 91700), submitted as Attachment 1 in the Waterford 3 October 23,1997 submittal contains validation problems drawn from test data, classical text book problems, and from peer finite element codes.
Through over 2,000 applications to fuel rack structures, the stability and accuracy of DYNARACK have been subject to continuous testing over the past 20 years. No case of numerical instability or inaccurate response has been documented. In contrast, general purpose codes such as ANSYS have proved inadequate or unreliable, chiefly due to the demands on multi-body fluid coupling models and the numerical stability issues. A numerical instability problem, when ANSYS was applied by Holtec International to a fuel rack problem, was reported by Holtec International to the NRC in 1996 under a Part 21 notification. The status of this Part 21 is unknown.
The body of information on DYNARACK is quite extensive, having been deposited on over 40 dockets in over two decades. The code was tested against all applicable experimental work where the necessery data is available.
Item 5 In the staff's first RAI, you were requested to explain how the interface between the liner and concrete slab was modeled, and also, how the liner anchors were modeled.
You provided a response that the pool concrete structure was modeled without 14
R considering the liner system. Provide a physical description of the acwal anchorage system between the liner and concrete slab. Are plug welds used to anchor the liner plates to the plates embedded in the concrete slab? If plug welds are used, submit i
your calculations that demonstrate there are no axial and shear failures in the liner and j
plug weld at the temperature of 212'F used in your thermal analysis.
Resoonse 5 The anchorage system between the spent fuel pool liner and the concrete slab uses fillet welds at shop joints between the unwetted side of the liner and the liner anchors, consisting of W8 wide flanges surrounded by grout as shown in Section W on Drawing
- LOU 1564-G-847 provided in the Waterford 3 October 23,1997 submittal. At field joints , the unwetted side of the liner is shop welded to a bent plate (forming the leak detection collector channels) by fillet welda, with the bent plate shop-welded to the liner anchor (consisting of C8 channel flanges) by fillet welds, as shown in Section N on Drawing LOU 1564-G-847. The last piece of the liner plate is field butt-welded to the adjacent liner plate at the liner field joint. The weld strength of the 3/16",1 1/2" long fillet welds at every 6' attaching the liner to the anchors (W8 or C8) at both sides of the anchor is greater than the weld strength of the 3/16,4" long fillet welds at every 12" on only one side of the anchor used in the test performed for Arkansas Nuclear One and referencad in the Vvaterford 3 October 23,1997 submittal. Therefore, these fillet welds are adequate by comparison. No plug welds are used for the Waterford 3 pool liner.
References:
1, Scavuzzo, R.J., et al,
- Dynamics Fluid Structure Coupling of Rectangular Modules in Rectangular Pools", ASME Publication PVP-39,1979, pp. 77-87.
- 2. Radke, Edward F.,
- Experimental Study of Immersed Rectangular Solids in Rectangular Cavities," Project for Master Science Degree, The University of Akron, Ohio,1978.
- 3. Levy, S., and Wilkinson, John, "The Component Element Method in Dynamics,"
McGraw Hill,1976.
15
I LIST OF ATTACHMENTS 1.
- 3-D Single Rack Analysis of Fuel Racks,' Holtec International Position Paper WS 115, Revicion i dated January 20,1998 (Proprietary).
- 2. Planar View of A 16 Rack Array (figure).
- 3.
- Evaluation of Fluid Flow For in Phase and Out-of Phase Rack Motions," Hollec international Report No. HI-87113, dated April,1987 (Proprietary).
4.
- Estimated Effects of Vertical Flow Between Racks and Between Fuel Cell i Assemblies," Holtec International Repori No. Hl 87114, dated April,1987 (Proprietary).
- 5. " Study of Non Linear Coupling Effects," Holtec International Report No. Hl-87102, dated January,1987 (Proprietary).
- 6. " Fluid Flow in Narrow Channels Surrounding Moving Rigid Bodies," Holtec international Report No. Hl-87712, dated April,1987 (Proprietary).
- 7. Revision Pages for Holtec international Report Hl-91700 (Proprietary).
8.
" Chin Shan Analyses Show Advantages Of Whole Pool Multi Rack Approach,"
by K.P. Singh and A.I. Solar, from Nuclear Enaineerina International. dated March,1991.
- 9.
- Seismic Response Of A Free Standing Fuel Rack Construction To 3-D Floor Motion," Nuclear Enaineerino and Desian.1984.
- 10. Flows Around A Typical Cell (figure).
- 11. Flows At A Channel intersection (figure).
- 12. Flow Charts, Pages 3 and 4 of Holtec International Report HI-961465 (Proprietary).
13.
Tables 5.1,5.2 and 5.3 (6 pages) of Holtec International Report HI 971606, dated January 1997 (Proprietary).
.14. Errata Pages (2 pages) Holtec International Report No. HI 971628.
,p' *A ATTACHMENT 2 1
s .: . -_
2 6 4 0U 1 Y
A R
. R A
I C
A R
6 1
5 3 7i A F
O H
I V
. R rU ' A N
A L
6 P I
0 4 2 6 1 e
g nn oa sh nC o i 'l e
u b w s
4i o i"m n
Q d
t e ep e
n sSM p
o s M= "
ev 5
3 R de 1 9 1 T
- l ! i . :
6 ATTACHMENT 8 1
i J