ema eeccaty EEanuehs mo A Method for Including Fluid Structure Interaction in an ANSYS Response Spectrum Seismic Analysis ‘Stove C. Austin and Huinam Rhee 1) ABB Combustion Engineering Nuclear Power, USA 2) Korea Power Engineering Compan, Korea ABSTRACT “The seismic response of structures located inside a Mud filled vessel subject to transient base ‘excitation isa complex problem, For problems where the ineracton between the Mud and the structure requires the use of a uid structure element, the ANSYS structural analysis tomputer code has Muid structure element available. However, his element ean only be ‘sed to perform unsymmetrc oF damped modal, harmonic and nonlinear transient analyss. For complex strotures performing a nonlinear transient analysis snot practical For finite element linear anelyss methods the ation of contained pls displaced water mass produces the cotrect response frequency. It Joes not include the uid structre interaction ‘tfets that reduce the response amplitude, Is use therefore produces conservative resus It {8 the objective ofthe paper to demonstrate a method for calculating more accurate and less conservative esl “This paper presets a mathod developed by ABB and KOPEC for including the Mid stuctare interaction effects, while pecorming the seismic analysis, using response spectrum methods. “The method consas of constructing uid structure interaction model and multiple stuctral ‘only models. Each structural only model duplicates one made from he fu structure medel Using the results ftom unit harmonic analyses, factors are calculate foreach result quantity for esch mode. These factors allow the results from spoctrum analyses using the structural ‘models to be modified t include the fluid siuctre inteaction effects. The result is @ significantly less conservative analysis than if uid structure interaction had. not been considered, INTRODUCTION Previous seismic spectsum analyses of reactor vessel imemals components had used auditional lumped mass tO model the effects of contained plus displaced water. This mass ‘only presentation produces very conservative resus, A less conservative and more accurate ‘method isto include the water effects using a Nuid structure interaction ( FSD alysis. All ‘Srctural analyses for the Korean Next Generation Reactor (KNGR) were to be performed ‘OCepih 1997, Canton Eng mc A Rig Reseusing the ANSYS code. The ANSYS code has FSI capabiliy. However, the FS] capability can not be used to perform a response spectrum analysis. As part ofthe first of a Kind engineering effort for KNGR, a method was developed for including the Nid structure imerseson effect ‘while performing the seismic analysis using response specrum methods. ‘The component used for developing the method was the KNGR Control Element Assembly (CEA) Shroud Assembly (IBA). This is a large cylindrical shaped assembly which # supported trom and mounted inside ofthe Upper Guide Structure (UGS ) bare. The wate in the annulus between the IBA and the UGS prosiuces the fui structure interact effect, MODELS, ‘The TBA is a cylindrical shaped assembly, approximately 160 inches tall and 147 inches in ameter. tis supported from a circular Mange nea ls top. Inside of the eyinder the IBA is Aivided into vertical cells by plates. Is mounted to and inside ofthe UGS rare. The gap ‘between the UGS barrel andthe IBA cylinder is 1.125 inches. The water inthis [125 inch annulus provides strong fluid structure coupling between the TBA. and the UGS. The water ‘rapped in the cells inside the IBA acts 28 addtional mass only, The UGS bare! is considered rigid compared to the IBA assembly and is included only 28 a fixed boundiy forthe FSt clemens, The IBA structure is modeled using plac stif63), soli (stif4S) beam (beam) and lumped mass (mass21) elements. The Fuld stucture interaction uses fig30 elements A half “symmetry mode is used with the approeriste boundary conditions, The X direct is parallel to the plane of symmetry, the ¥ direction is perendiculic tothe plane of symnety and the 2 direction is vertical, The excitation and response direstion is X, parallel the plane of ‘symmetry. The model consists of over 7000 elements. Three madels were ued, A complete ‘ode! including FST elements sed for FSI medal and harmonic analyses, Two models were used that did not included FSI elements. These models included addition lumped mass ‘lements on the IBA barrel. One model would be tuned to represent the beam mode as ‘determined by the FSI modal analysis. The otbec would be tunel to the shell mode. These ‘mass21 elements allowed the ation ofthe tuned mas oa inthe horizontal directions onl. ANALYSES Modal Analysis ‘The ANSYS fhuid30 clement uses unsymmetsic matrices. Therefore, the mola analysis for ‘the FST model wed the ANSYS unsymmetic method option In arr to mininize differences {de to selection of master degrees of freedom the subspace method was Used ‘or ll analyses, ‘The FSI results are summarized in Table 1. Based onthe FSI rerults, modes | and 4 are the ‘modes which will produce dynamic amplification for a X direction excitation. If an input spectrum has very large peaks above 33 HZ, which would be pial of loss of coolant accident (LOCA) spect, this method coold be extended to include higher modes using the same procedures used for modes | and 4. However, since modes above 33 HZ. will be higher shell modes iti unlikely that they would be excited by uniform motion in the X direction ‘Therefore, use of modes I and 4 pls static missing mass to account fr the higher modes is Justified. The mass2i elements udded tothe TBA shell were adjusted until te Fequency and ‘mode shapes match the FST modal results. The beam model is adjusted to mach FSI mode and the shell mode is ajusted to match FSI made 4. The coresponding medss for the beam and shell models are highlighted in Table 1 ‘Gopi 1997. Contuton gain ne AUR Re vanHarmonie Analysis ‘The harmonic analyses were performed atthe exact beam and shell Fequencies forthe FSI and non-FSI models. 1G acceleration is applied paalal othe pane of symmetry. The rel ‘and imaginary solutions are combined in Post! by square root of the sum of the squares (SRSS} to produce the harmonic result. The results ave summarizad in Tables 2, and 3. The columns iabeled TOP, MID and BOT provide the maximum stress intensity a the element centroid atthe top sueface, midplane and botiom surface. The TOP and BOT results are ‘emirane pis bending while the MID result is membrane only. Fo the higher stressed parts of the model, the webs and tubes, the difference between membrane plus bending. and ‘membrane only isnot significant. The column labeled NODAL provides the maximutn sess intensity athe element nodal lections for membrane plus bending. The use of FST provides & very large reduction in the stesses due to the beam response node. Although the input ‘requencis for the harmonic analyses are sot the exact frequesey of the mde of interest, ‘he other modes of the mode are still active. Por the PSI model the X divetion modes are ‘well separated. For the Beam model mode 6 and 7 we close but the paicipaton facto for mode 7, the desired beam mode, is significantly larger than for mode 6, a complex shell mode. For the Shell model mode | the desired shell mode, and mode 4 ae fury close with mode 4 having the large: participation factor. The mode shape of Shell model mode 4 maiches the ‘mode shape of FSI mode 7, which has a frequency of 33.67HZ. The possible effect of Shell ‘model mode 4 on the calculsed harmonic ests was investigated and corected by use of spectrum analyses, Response Spectrum Analysis “The response spectrum results ae used fo confitm thatthe harmonic analyses ae valid. They se also used to provide an example of how the method would be used for « production analysis and to give an estimate of the expected stresses in tho IBA. Spectcum analyses were performed for the beam and shel! modes using the Yengawang (¥GN) 384 and YON § & 6 "Reactor Vessel (RV) flange Operational Basis Earthquake (OBE) 2% spectrum ‘The spectrum analysis method allows for the result for each made to be individually caleulated, thus avoiding the problem of having more than one mode contributing tothe result, ‘The harmonic results were converted tO match the response spectra reslls using amplification formule fox hatmonie excitations Output = Inpue2*Dampingy | G harmonic produces 25 G response for 29% éxmpin. Comparing the converted harmonic results to the spectrum reals, the Beatn mode results ‘how very good agreement, with the Nodal results being essential identical, The Shell mode resuls indicate that the harmonic results are consistently greater than the spectrum results ‘This is most likely due to more than I mode canebting tthe harmonic rests, as discussed shove. Since the FSI model doesnot have modes that are close together itis conluded that the FSI harmonic analyses can be used to predict the te response ofthe IBA dve to response spectrum input “The eats show tha the harmonic analysis fr the Shell made, utd to a lesser extent forthe [Beam mode, over predict the result. Therefore, the ratios of harmonic results for FST model to ‘Conia 199%, CombusinEngieing e.g Ress vitaShell and Beam models have tobe corected. The beam correction factors wers calculated 3s fellows: ‘Beam corrosion factor = (Converted Bean Harmonie / Beam Spectrum) (Beam Harmonic / FST Harmonic for Bear Mede) ‘Tables 4and 5 provide the final correction factors forthe beam and shell modes Missing Mass Analysis ‘The missing mass is calulated by adding together the percent mass active in the Beam and Shell modes and subtracting from 1.0. The mass values reported forthe FSI modal anlysis ‘re not coret due tothe unsyrmetcic matrices, therefore the percent masses ate take fram the Beam and Shell models. Since these models have more low frequency mods then the FSI ‘model the percent mass active in any one mode would be lower. Therefore, th calculation of missing mass is conservative. The missing mass analysis was pesformed wring a 1G unit acceleration in the direction parallel tothe plane of symmety, The calculated results are than scaled by the product ofthe percent missing mass and the 210 period acceleration (ZPA) of the response spectra, Combination of Results ‘The total esult is formed by combining the thee calculated results acording tothe equation: Toul ‘Beam mode result }*+ ( Shell mode result }* + (missing mass result)" RESULTS ‘Tables 4 and 5 provide the calculated IBA factors to adjust selsmic esponse spectra analyses fesulis to account for FST effects. Table 4 indicates that including fluid strctare interaction causes 2 significant reduction in the Beam mode response of the IBA. The shell mode ‘response, Table 5, is less significantly affected. A seismic response speciram analysis of the IBA would be performed in the following steps, + Perform response spectrum analysis of IBA using Beam model with inpt spectra ony at bean mode frequency ‘+ Perform response spectrum analysis of IBA using Shell model with input spectra only at shell mode frequency ‘Perform sti acceleration analysis using “missing mass” times ZA of resgonsespecea. + Covrect Beam and Shell mode results by use of cts in Table 1 ‘+ Combine results by SRSS of Beam and Shell Modes and acceleration analysis th resus from static npg 999. Contain gn ne Al Righes Retene var | | | | | | | i | | |Sample Result “Tables 6 and 7 provide the total results using seismic respense spectra calculated for YGN 3 & 4 and YON 5 & 6, The tables also provide, fr comparson, the sres as caleulted using added mass only for the Korean Standard Plant design. The KNGR design and for YON 5 & 6 and 3 82% methodology provide a 68% reduction in maximum sess reduetion for YON 3 & 4, TABLE | Modal Results rons Soe REA [za eae a0 S [ex] or 1 BEAM MODEL PARTICRATION FACTORS OSE [SE PONT fz ECT Tae om | tae on 3st [ar oo et | tr tar m0 SHELL. MODEL, ‘saRTicmon sucroxs NODE REDE BREE TOn|Y RECTION SEEN Sst [ ar — [st — oor caa—[ ar [ast [ sa Tt io Cay 19, Contin Ean Is A Rig Ree‘Teble? IBA HARMONIC RESPONSE ANALYSIS For Beam Mode, ' ‘poe : esl for compares ae mania ses see PST ‘Table IBA HARMONIC RESPONSE ANALYSIS. For Shell Mode 8 Clear 1G ps, 28 Daring, Hele Ae see SRT See eee asicvves | 3igr_[ eas Past aaa eer [Pr “Sinnwes [Tasa¢—[ aaa [soni [ost oe [ae soSinciwkes [ine] tise tae Par [ae a 25in bet [tae [1s | serosa [ee ep fangs 06 | rH worse [se 68 er eae [a a a eg [Cass 235 cupekaccia [ise 36 TT Vr capone Be HAT a ST HPT © conigh 99. Contin Eger I AI Rpts ReesTable 4 Bear Mode Correction Factors eT NOB Bink we [ware | oom | gare | onsi Siechaeb | n0560_| 00s | noaes [oan Tsinch wba] oo] 00s | ooemt_[ aa Sindh bane [oe 0009 | 00825_[-001sr Tp ange [oes 5 eppae | 00me_| oes | cares O05 fingers | 0060) | 10700 ousze | Otsee Tose dag 00257, ‘Table ‘Shell Mode Corection Factors a Finch wa [150 | Tar Tei a Stach web | 0905 | 0.8709 Oead [0-082 Sisiecniwsa | 07992_[-aatsr_[ 07866-[ 0.908 inch are] 03056 | 1.0046 | O40 _[-1710E Tings — 0.355 BSS Tap tnge [191 aod eppue [oss | OTE | Oe osm Gages | Tarae | 1.3504 | Tole8 [3.50 Tues disp 0983 © Copmih 19, Contin Eogiceing Is. ARMS Reed vayTable 6 IBA SEISMIC RESPONSE ANALYSIS CEA SHROUD ASSEMBLY YGN OBE 2% Damping YGN 384 OBE 2% Damping rea CORE Co] "esl Tr compensa asim ses ness in PSY Tete? CASESUC RESPONSE AWLIEIS Geass psseneLr ell Sso oe oh Betty Tat ST SSE Samy —<, CEE =e, = Ete Se Rest for compacts asin sesneses a 731 ‘Och 95. Combai Engng ne. AN Right Rene