ML20237K402
| ML20237K402 | |
| Person / Time | |
|---|---|
| Site: | Diablo Canyon  |
| Issue date: | 06/17/1987 |
| From: | Degrassi G BROOKHAVEN NATIONAL LABORATORY |
| To: | Office of Nuclear Reactor Regulation |
| References | |
| CON-FIN-A-3841 OLA-S-001B, OLA-S-1B, NUDOCS 8709040333 | |
| Download: ML20237K402 (20) | |
Text
{{#Wiki_filter:_ _ _ - _ - - _ _ _ _ ___- - E NN g4 jp _ y 7$ 97 3,gg 4[/7/f7 y,,. _n 3 - lB EVALUATION OF THE STRUCTURAL ADEQUACY OF THE DIABLO CAN j 26 P4 :10 DENSITY SPENT FUEL RACKS IN ACCOMMODATING MULTIPLE RACK IMPACTS DURING THE POSTULATED HOSGRI EARTHQUAKE Technical Evaluation Report G. DeGrassi Structural Analysis Division Department of Nuc. lear Energy Brookhaven National Laboratory May 1987 \\ Prepared for: U.S. Nuclear Regulatory Commission Office of Nuclear Reactor Regulation Contract No. DE-ACO2-76CH00016 Fin No. A-3841 l 1 8709040333 870617 PDR ADOCK 05000275 g PDR IVRC SUh W 1-B
) htlCth REGGATOH COM1140N g{g( I6 oocw no. 60 4WOL A omcw En. no. b4b o*D AW in the matter of n 1 Staff _ l _IDINTIDID _ V -~__ RECEIVED _ App licat.t _ _RUECTED _ latervenor -...
- t. cst'g Ott's Contisctar DATE Witneu Othat -
'. O-Y YM MO I Reporter ( i l f 1 l s l j l I l A Y ~
l b i 1 1 1 1 l l \\ TABLE OF CONTENTS i
1.0 INTRODUCTION
1 2 2.0 DESIGN BASIS ANALYSIS BACKGROUND 3 3.0 MULTIPLE RACK IMPACT CONCERN 4 j 4.0 HULTIPLE RACK STUDIES 6 5.0 ADDIT 1"NAL SINGLE RACK STUDIES 7 6.0 EVALUATION 7 6.1 MULTIPLE RACK STUDIES 6.2 ADDITIONAL S;50LE RACK STUDIES 7 l 1 8 l 6.3 EVALUATION
SUMMARY
l 1 11 l l
7.0 CONCLUSION
S l 12 REFERENCES i l l l l l l l 1 ii I l l
s . l.0 INTRODUCTION This' technical evaluation report (TER) covers the Brookhaven National Laboratory (BNL) evaluation of various analytical studies submitted by-Pacific Gas and Electric Company (PG&E) in support of their original seismic analysis of the proposed Diablo Canyon high density spent. fuel racks. A concern regardi'ng the adequacy of the design basis single rack models in predicting loads that could develop during possible_ multiple rack impacts had been raised by BNL. The PG&E studies were performed at the request of NRC and BNL to quantitatively demonstrate that the conservatism used in the design basis models provide an adequate design basis to accommodate multi-rack impact effects. a e0 1
r 2.0 DESIGN BASIS ANALYSIS BACKGROUND The. proposed reracking of the two spent fuel pools at Diablo Canyon involves the installation of sixteen free standing high density racks of various sizes -as shown in Figure 1. Each rack consists of an array of square tubes, each designed to accommodate one fuel assembly. Figure 2 shows a typical rack configuration. Since the racks are not anchored to the pool floor or walls, during a earthquake, the recks would be free to slide and tilt. If the earthquake is cf sufficient intensity, the racks could impact each other or the pool walls. In order to demonstrate the structural adequacy of the fuel racks during an earthquake, the Licensee developed three-dimensional, non-linear mathematical models of single racks and performed dynamic analyses to determine maximum rack response. Details of the dynamic analyses were provided in the Reracking Report (Ref.1) and the Seismic Analysis Report (Ref. 2). Several case 6 of individual racks were analyzed using models ~ representing various rack configurations, different rack locations within the pool, and empty to full fuel load conditions. Due to the uncertainty in the value of coefficient of f riction between the rack feet and pool floor, each case was run with an upper limit coefficient of 0.8 and a lower limit of 0.2. The Licensee indica'ted that the following assumptions used in the design basis single rack models are conservative <a: o Each adjacent rack module was assumed to move in a manner equal and opposite (out of phase) to the rack module being analyzed. This assu=ption was incorporated in the model by utilizing a reference impact plane midway between adjacent racks. o The fluid coupling coefficients were based on the conservative assumption that adjacent rows of rackc are an infinite distance away (the distance is measured perpendicular to the horizontal ground e " s ' This neglects the cross-coupling effect of the adjacent r e .a -' racks and results in hAgher displaeuments and impact. scus4 o The impact., g coefficients were eet at a value significantly higher (generally over 10 times) than the calculated values to produce conservative impact forces. The results of the design basis analyses are presented in Ref. 2. The maximum impact forces between racks and fuel, racks and racks, and racks and walls are presented in Table 1 along with corresponding allowable loads. These results show substantial margins between predicted and I allowable loads. j 2
3.0- MULTIPLE RACK IMPACT CONCERN The design basis analytical models considered rack to rack impacts through the use of impact springs (gap elements) at the cornere of the girdle bars and baseplates. Since only a single rack was included in each model, the motion of adjacent racks had to be predefined. It was assumed that maximum impact loads would occur if the adjacent racks move completely out of phase with the rack being analyzed. This assumption was incorporated into the model by defining a plane of symmetry midway between adjacent racks creating a " mirror image" effect around the analyzed rack. Thus whenever displacements exceeded half the gap between racks, the model would simulate an impact with the. adjacent rack moving at equal velocity in the opposite direction. A typical rack model showing inter-rack impact elements is shown in Figure 3. For rack models representing racks adjacent to pool walls, the gaps on the wall side were set to the actual distance between the rack and wall. BNL questioned the adequacy cf the design basis models in predicting maximum impact loads which could result from complex multiple rack interactions. The out-of phase adjacent rack assumption is conservative in the sense that it maximizes the relative velocities and resulting impact forces between any two adjacent racks. In reality, it is extremely improbable that adjacent racks would respond to an earthquake in an out of phase manner. However, the multirack concern is,related to the fact that the design basis single rack model constrains the motion of a rack within a tight predefined boundary and does not allow for possible simultaneous impacts of more than one set of racks or one rack.to a wall. During an ear thquake, there is a greater likelihood that a row of racks would move nearly in phase with each other. If the seismic motion is strong enough, the possibility exists that a row of racks could experience large sliding motions and pile up against a pool wall. Such a " pile up" could generate higher multiple rack impact forces than predicted by the single rack model. Due to the complexity and non-linearity of the system, it was difficult to judge whether the conservatism of the design basis model was sufficient to predict impact loads that would bound potential multiple rack impact loads. To investigate this concern, BNL suggested that a parametri'c study be performed to quantitatively verify the conservatism of the single 4 rack model. Recognizing the complexity of developing a three dimensional model of all sixteen racks in the pool, it was judged that a two dimensional model would be appropriate and sufficient for this parametric study. ~ i i 3
4.0 MULTIPLE RACK STUDIES PG&E performed a number of parametric studies to demonstrate the conservatism of their methodology for obtaining the design basis impact loads. The results of these studies are documented in a proprietary report (Reference 3). A nonproprietary version of the report (Reference 4) was also provided. These studies utilized analytical models of each rack which were similar to those used in the design basis analysis. However, in order to reduce the complexity of multiple rack models, two-dimensional planar motion was assumed. A total of seven cases were analyzed. They included two single rack models and five cultiple rack models. The Case 1 single rack model used parameters which were consistent with the design basis model parameters. This case was used to develop benchmark loads for comparison with the other ca s e s. The Case 2 single rack model and the five multiple rack models used revised parameters which the Licensee considered less conservative but more realistic then the parameters used in the 3-D design basis model. As discussed in Section 2.0, the design basis models had used impact spring values which were generally an order of magnitude stiffer than calculated values. They also used low fluid coupling coefficient values which neglected the cross-coupling ef fect of adjacent rows of racks. Cases 2 through 7 used impact springs which were generally 1.5 times their calculated values and fluid coupling coefficients which were based on an assumed lateral gap of 7.5 inches on each side of the racks. The multiple rack models represented single rows of four racks in the cast-west direction. That direction was chosen because the Licensee determined that the east-west horizontal Hosgri earthquake represents the most conservati 1 horizontal seismic input. Case 3 represented an interior row of fully loaded racks. Case 4 represented the same interior row with three fully loaded racks and one empty rack. Case 5 represented an exterior row of racks with a 37 inch spacing between one rack and the adjacent wall. For that case, fluid coupling coefficients were developed by conservatively assuming that the adjacent vall is 37 inches away from all racks in the row. Cases 6 and 7 represented interior rows and were developed to investigate the effects of variable gaps resulting from f abrication and installation tolerances. Consistent with the design basis methodology, each case was analyzed twice using the bounding coefficient of friction values of 0.2 and 0.8. The Hosgri east-west and vertical time histories were applied simultaneously to each model and maximum responses were determined. The results of the analyses are presented in Reference 3. Due to the proprietary nature of the material, quantitative results are not. presented in this report. The results indicated the following: 1 4
l ) 1. The maximum rack-to-fuel impact load predicted by the Case 1 single rack model enveloped the corresponding multiple rac2 model 3 impact loads by a substantial margin. The m'ximum rack-to-rack impact load predicted by the Case 1 model 2. a enveloped the multiple rack model rack-to-rack impact loa,ds. 3. The rack-to-wall impact loads predicted by the multiple rack models were not enveloped by the corresponding Case 1 model maximum impact load. However, the Case 1 maximum rack-to-rack load enveloped all multirack model rack-to-wall loads in all cases except for Case 5. j i 4. The Case 5 multirack edge row model rack-to-wall impact load exceeded both the Case 1 rack-to-wall and rack-to-rack maximum loads. ) I ~ 5. A comparison of the Case 1 and Case 2 single rack model results indicated the same trend. Case 1 maximum rack-to-fuel impact load enveloped Case 2 by a large margin,, Case 1 maximum rack-to-wall impact load did not envelope the Case 2 rack-to-wall loads but the Case 1 maximum rack-to-rack load enveloped both the Case 2 rack-to-rack and rack-to-wall loads. The results generally appeared to support the Licensee's position on the conservatism of the design basis methodology but observations on the rack-to-wall impacts raised an additional question on the conservatism of the high impact spring rates used in the design basis models. BNL recommended that an additional three-dimensional single rack analysis be performed to investigate the sensitivity of impact spring and fluid j coupling variations on rack loads. l I l l i l 1 i 5 1
t. 5.0 ADDITIONAL SINGLE RACK STUDIES PG&E performed an additional three-dimensional single rack analysis of a fully loaded-10 x 11 corner rack using " realistic" impact spring and fluid coupling parameters consistent with the 2-D multirack models. This case was selected for reanalysis as the overall most critical rack considering high strest ratios, fuel-to-rack impact loads and rack-to-rack inpact loads (Run No. Acorn 10 from Reference 2). To be consistent with the design basis methodology, two cases were analyzed using the bounding coefficient cf friction values of 0.2 and 0.8. The results of the study are documented in Ref erence 5. A comparison of maximum values of impact loads and stress ratios predicted by the design basis model and the " realistic" parameter model is presented in Table 2. The stress ratios are the ratios of calculated stress to ASME Level A Code allowables for compression (R ), shear I 3,R ) and combined compression plus bending (R. (R ), bending (R 4 S 2 R ) ns defined in Reference 2. The allowable value of stress ratio at 6 the areas listed in Tabic 2 for the Hosgri earthquake (ASKE Level D) is 2.0. The results of this study indicated the following: 1. The design basis rack-to-fuel impact loads enveloped the " realistic" model loads by a substantial margin. 2. The design basis rack-to-rack impact loads were about 10% lower than the loads predicted by the " realistic" model. 3. The design basis model predicted no rack-to-wall impact while the " realistic" model predicted a 48 kip impact load. However, the design basis rack-to-rack load load enveloped the " realistic" rack-to-wall load by a good margin. 4. The design basis support foot impact load enveloped the " realistic" model load. 5. The design basis stress ratios enveloped all " realistic" model stress ratios except for the R1 value at the base of the rack. However, the magnitude of the " realistic" R1 value is less than 10% of the allowable value. l 6 4
l 6.0 EVALUATION 6.1 Multiple Rack Studies The result's of the multiple rack parametric studies generally support PG&E's position that the design basis single rack models have sufficient conservatism to accommodate multi-rack efects. The Case 1 single rack model with " design basia" impact spring rates and fluid coupling coefficients predicted maximum rack-to-fuel and rack-to-rack loads which enveloped the corresponding loads for all multiple rack cases. Rack-to-wall loads predicted by the multiple racks models exceeded the loads predicted by the Case 1 model for all cases. However, the Case 1 rack-to-rack loads enveloped all multirack model rack-to-wall loads except f or Case 5. Since the racks were originally evaluated for the larger of the rack-to-rack and rack-to-wall loads, this is a valid comparison for rack evaluation. Because the loads on the pool walls were shown to be higher than had been considered in the original design basis evaluations, the walls were reevaluated and shown to be capable of withstanding loads exceeding the strength of the racks. This indicates that the allowable load is controlled by the 175 kip rack-to-rack allowable. Therefore, for evaluation purposes, rack-to-wall loads are equivalent to rack-to-rack loads. The Case 5 edge row multirack model rack-to-wall loads were higher than the loads predicted by the Case 1 single rack model. However, two f actors should be considered in evaluating the significance of the higher loads. The first is the conservatism of the fluid coupling coefficients bared on a 37 inch rack-to-wall gap for all four racks in the row rather than the actual case of a single rack. While this conservatism is difficult to quantify, it is reasonable to expect the actual loads to be lower than predicted by this model. The second factor is the safety margin. The allouable load of 175 kips is well above the calculated value for Case 5 given in the proprietary report (Reference 3). The comparison of the Case 1 design basis single rack model results with the Case 2 " realistic" parameter single rack model results demonstrated the same trends seen in the multirack comparisons. The similarities observed in the rack-to-wall load comparisons prompted the request for the additional studies. It appeared that higher rack-to-wall impact loads could be more closely related to variations in model parameters rather than to multiple rack ef fects. Therefore an additional analysis to investigate the sensitivity of parameter variations was recommended to demonstrate that the design basis modeling assumptions were conservative. 6.2 Additional Single Rack Studies The comparison of the design basis three-dimensional single rack model r:: ult: to the " realistic" parameter model results provided valuable information on the sensitivity of parameter variations in the models. The 7
l l results demonstrated the substantial conservatism of the rack-to-fuel loads predicted,by the design basis model. This can be clearly attributed to the use of stif f er than calculated spring rates. However, the rack-to-rack and rack-to-vall results did not show the same clear trend but instead suggested more complex interaction effects. The fact that the design basis model predicted no rack-to-wall impact and a slightly lower rack-to-rack impact than the " realistic" model in spite of the stiffer impact springs may be an indication that the " realistic" rack model experiences greater rocking motion because of its " softer" support feet. At a meeting held on May 6, 1987, PG6E and its consultants were asked to provide their explanation of this rack behavior. They explained that the use of the " realistic" spring constant resulted in a rack natural frequency which fell on the peak of the Hosgri vertical response spectrum. This resulted in maximum amplification of vertical rack response. The vertical response was probably overpredicted since the model did not include fluid coupling or damping in the vertical direction. They also stated that the higher vertical response can induce more significant rocking. This provided a reasonable explanation to the "realistir." model j rack behavior although the significance of the vertical-rocking interaction I is difficult to assess. From an engineering standpoint, however, it is most significant to note that an order of magnitude decrease in all of the impact spring rates resulted in only a 10% increase in rack-to-rack loads. l The design basis model support foot loads enveloped the realistic l l model loads. The design basis model rack-to-rack loads enveloped the realistic model rack-to-wall loads. The design basis model stress ratios I enveloped all " realistic" model stress ratios except for the R1 value at the base of the rack. Since R1 is related to the compression in the vertical direction, the higher value is consistent with the Licensee's explanation of the higher vertical response. In addition, the safety margin for R1 is very 1arge (R1 is less than 10% of allowable). 6.3 Evaluation Summary The original design basis analysis evaluated a number of critical rack configurations at dif f erent locations within the pool and with varying fuel load conditions. Maximum impact loads predicted by all of those analyses indicated substantial margins as shown in Table 1. The two-dimensional multirack studies demons 6cated that with few exceptions a single rack model j using parameters consistent with the design basis models predicts loads that envelop multirack models with more " realistic" parameters. The l three-dimensional parameter sensitivity study showed that with few exceptions the use of the design basis model with " conservative" impact I l spring and fluid coupling parameter predicts conservative results when compared with a similar model using more " realistic" parameters. l l l 8 I l
4 The safety significance of the exceptions noted above can be evaluated in terms of their potential impact on overall safety margins. The three areas where exceptions were noted involved rack-to-rack loads, rack-to-wall loads and direct compression stress ratios. The significance of each exception is discussed below. Rack-to-Rack Loads ~ The three-dimensional parameter sensitivity study indicated that the design basis models may slightly underpredict rack-to-rack impact loads. This appears to be related to a reduced rocking response resulting from the l stiffened support foot springs in the design basis model. The specific j case investigated showed a load increase of 76 to 85 kips between models in which the impact spring stiffnesses were decreased by an order of magnitude. A review of all design basis model results indicated a maximum rack-to-rack load of 105 kips as shown in Table 1. The allowable impact load is 175 kips. Considering the small increase in load resulting from the large change in stiffness coupled with the substantial safety margin, it can be concluded, with a high degree of confidence, that any increased i loads due to uncertainties in modeling parameters can be safely I a cco mmoda t ed. I Rack-to-Wall Loads 1 Both the two-dimensional multirack studies and the three-dimensional I parameter sensitivity studies indicated that the design basis models may l underpredict rack-to-wall loads. This underprediction may be a result of multirack effects combined with the reduced rocking response discussed j above. The design basis model predicted a maximum load of 63 kips as shown j in Table 1. The 2-D multirack studies predicted higher impact loads. However, the studies showed that rack-to-rack loads in all but one case envelop the rack-to-wall loads. Since the pool wall was shown to be stronger than a rack, the Licensee demonstrated that for evaluation j purposes rack-to-wsil loads are equivalent to rack-to-rack loads. Only one multirack study (Case 5) showed a rack-to-wall load higher than the single rack model maximum rack-to-rack load. However, considering the additional conservatism in fluid coupling coefficients for that particular case and the substantial margin between the predicted load and the allowable load, l it can be concluded that this result does not represent a safety concern. , Stress Ratios The three-dimensional parameter sensitivity study considered the most critical rack with lowest safety margins on stress. The comparison between design basis model and realistic model results demonstrated the conservatism of the design basis model in predicting rack stresses. There was only one exception which involved the direct compression striess in the 9
O vertical direction at the base of the rack. However, the predicted stresses were an order of magnitude lower than the allowable value. Since this is not a controlling design stress, the small increase is not a safety concern. The stress ratio results provided further confidence in the overall conservatism of the design basis models for predicting stresses. The R2 to R6 stress ratios in the 10 x 11 fully loaded corner rack were the highest design basis values observed for all racks. The corresponding " realistic" parameter model predicted lower stresses despite the slightly higher rack-to-rack impact load. 1 1 10
.4
7.0 CONCLUSION
S Based on the review and evaluation of the results of the original design basis analysis, the two-dimensional multiple rack studies and the additional three-dimensional parametric sensitivity studies described in this report, BNL concludes that the Diablo Canyon high density spent fuel racks will maintain their structural integrity during the Postulated Hosgri Earthquake. The additional studies have provided sufficient evidence to conclude with a high degree of confidence that the loads and stresses l resulting from multiple rack impacts can be accommodated within design allowables that are consistent with Regulatory requirements. e l 1 l 11 l
REFERENCES 1. "Reracking of Spent Fuel Pools, Diablo Canyon Units 1 and 2", enclocure to PG&E Letter No. DCL-85-306, dated September 19, 1985 ~ 2. Seismic Analysis Report, " Seismic Analysis of High Density Fuel Racks for Pacific Gas and Electric for Diablo Canyon Nuclear Power Station", Rev. 3, September 3,1986, A. Soler, Di #779. 3. " Additional Information on Rack-on-Rack Interactions", Enclosure to PG&E Letter No. DCL-87-070, dated April 7, 1987. 4. " Additional Information on Rack-to-Rack Interactions (Nonproprietary Version)," Enclosure to PG&E Letter No. DCL-87-072, dated April 9, 1987. "Three-Dimensional Studies of High Density Spent Fuel Racks (Acorn 10 5. and Acorn 12)," Enclosure to PG6E Letter No. DCL-87-082, dated April 23, 1987. 12
TABLE 1 Results-of Design Basis Single Rack Analysis. IMPACT LOCATION MAXIMUM PREDICTED LOAD-ALLOWABLE LOAD (kips) (kips) Fuel-to-rack-249 883(1) Raek-to-raek 105 175 Raek-to-wall 63 '175(2) (1) Allowable load.for 10 X 11' rack. (2). Allowable controlled by rack strength. Walls were qualified to loads exceeding 200 Kips. $U '13 ')
9 TABLE 2 Comparison of Results for a Fully Loaded 10 x 11 Corner Rack Design Basis Model vs. " Realistic" Parameter Model Impact Loads (kips) Design Basis Realistic Parameter Rack-to-Fuel 250 61 Rack-to-Rack 76 85 Rack-to-Wall 0 48 Support Foot 296 261 Stress Ratios Base R1 .122 .173 R2 .156 .080 R3 .231 .126 R4 .134 .089 R5 .323 .244 R6 .364 .265 Support Feet R1 .305 .277 R2 .347 .199 R3 .737 .362 R4 .506 .235 R5 1.265 .657 R6 1.436 .743 l 14
i i i 1 i I I N ~ I l iO-0" 3" 19'fg1 ${'h ~ R e t. Ref. 2%. i i 10 8" 3
- R e f.~'
j o .'N ' ie e k9i LN ' 9X9 g CF ic l r A, 8XZ+4X2j Dii M B^ ,e N Plus,1p 10 X 9 Tl Region 1 -{ ((,s 9 IOX9 e j m, s - I a 'G^ "C* '03 'A'z i t.. 8 X IO IO X IO 9 X IO Region 1 5 8 -- -m lo xfo .I R e f. 1 ( i4 N { ,g p e- 'O'
- f,
{ i.s* 6X9 IOX9+6 9 X IO i Region 1 e I e s C O I/ ' IO X IO ~ i ,e-
- e6 lG*
e e y ) 9, [ [/ 'E' T V f Cask Pit ' 6 X il 9x9 IO X 11 g ~ / ~ r [j 2 6'/16" l o ~ 4 4 Ref. L.e i I. e 10 - 0,, 64/E 9 7 '/16 108" -g y .- 9 9 /~6 2/ 2 '/ ~ f R e f. 3 5'- O ' l 1 Figure 1. Pool Rack Module Layout, Unit 1 15 s _-..___D
d J 'E N 4 t y 1 w 1 w i w 1 w, w,- ve 7 -, v- -,y ;- f 4 Ti t ' *- : - :- i :- > J l
- 1\\
s Girdle Bar e .o c e i 8 3 e e L S l ) { 4 s ~ ~ - - M i e i Figure 2. Typical Rack Module Elevation 16
3< Typ. Top impact Element /$ Ii t~V M f f ^ Rack Structure Typ. Bottom impact Element W M W[,- M 7 1 I- \\ rigure 3. Cap Elements to Simulate Inter-rack Impacts 17; _ _ _ _ - _ _ - _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ _ _ _ -}}