Shake Table Test fo Lage Indiect-Ai-Cooling Towe Stuctue of Fie Powe Plant Pat I Junwu DAI, Yongqiang YANG & Xuan WNG Key Laboatoy of athquake ngineeing and ngineeing Vibation, China athquake Administation, China SUMMARY: Fo undestanding the seismic behavio of exta-lage scale cooling towe with dimension of 220 metes high and 188 metes in diamete, both of the dynamic nonlinea finite element analysis and shake table tests fo a 1:30 (length atio) model wee caied out to simulate the eathquake impacts. In the model design, a new model mateial simulation method is developed. Specially teated lead sand was used as one of the main aggegates of the model constuction mota. Both of the dynamic NF analyses fo the pototype towe stuctue and its 1:30 model countepat wee caied out to compae each othe in consideing of the similitude law. The eathquake esistant capacity of the towe as well as its citical element of the suppot leg columns wee veified and studied caefully. This pape povides useful efeence to the seismic design pactice fo the exta-lage towe stuctues. Key wods: shake table test, lage indiect-ai-cooling towe, model stuctue, pototype stuctue 1. BACKGROUND With the apid demand of the fie powe plant, exta-lage indiect-ai-cooling towe (1000MW) will be selected to be constucted in the high seismic isk aeas of China such as mid-noth and west-noth egions. The dimension of the huge towe stuctue can each up to 220 metes high and 188 metes in diamete. It s constucted with X type R/C column suppoted hypeboloid shell and the X column s length-width atio can each up to 1:40. It s eally a challenge but vey necessay and ugent to know the seismic behavio and design weak points of the huge towe unde stong eathquake attacks. In 2005, S. Saboui-Ghomi and M.H.K. Khaazi took a study on the einfoced concete column suppoted hypeboloid cooling towe stability assessment fo seismic loads. In thei study, finite element analyses have been pefomed to obtain the stess concentation, nonlinea behavio, stability o safety facto of the R_C_ towe due to eathquake loads. Outcomes of thei study show that consideable plastic hinges wee ceated in the X shape long columns of the R/C hypeboloid cooling towe due to seismic loads, which esulted in a significant decease in the stability safety facto. Accoding to W.S. Guo s intoduction, R.Hate and U.Montag pefomed a study on compute simulations and cock-damage evaluation fo the duability design of the wold-lagest cooling towe shell (200m high and 152m span) at Niedeaussem powe station (1000MW gade). But as we all know, Gemany is not located in the seismic egion and the Niedeaussem powe station is not exposed to sevee eathquake isk. The study on the 200m high and 152m span cooling towe can t
povide useful efeence to the seismic design fo the wold lagest 220m high and 188m span cooling towe in China. This pape studies the seismic behavio of the wold-lagest R/C hypeboloid cooling towes with vey long X shape suppoting columns. Both of the dynamic nonlinea finite element analysis and shake table tests fo a 1:30 (length atio) model wee caied out to simulate the eathquake impacts. A new model mateial simulation method is developed to fulfill the goal of shaking table test. Specially teated lead sand is used as one of the main aggegates of the model constuction mico aggegate concete. Both of the dynamic nonlinea finite element analyses fo the pototype towe stuctue and its 1:30 model countepat wee caied out to compae with each othe in consideing of the similitude law. The eathquake esistant capacity of the towe as well as its citical element, the suppot X-type columns wee veified and studied caefully. 2. Length Ratio 1:30 Model Similitude Design The fist step of shaking table test is the pope model design. Geneally, the length atio of the model to the pototype stuctue should be detemined accoding to the load capacity and dimension of the available shaking table. Fo the cooling towe, the most difficult issue is the installation of atificial mass to the model stuctue. In some test cases, the atificial masses have to be hanged out along the shell wall like hanging sacks. It caused a new poblem that is the hanging way may change the eal dynamic esponse of the model stuctue and the test esults eo due to the stiffness and damping change. Fo avoiding the poblem of the atificial hanging way, the authos develop a new method to solve the mass loss and the hanging poblem simultaneously. 2.1. Descibe of the Pototype Cooling Towe The huge pototype R/C hypeboloid cooling towe has a total height of 220 m, a span of 188 m in diamete on the foundation, a span of 169 m in diamete at the tansition of columns to shell, a span of 107 m at the thoat section and a span of 110 m in diamete at the top. The total elevation fom the gade fo the X shaped column is 28.7 m. The columns have a dimension of 1.6 m by 0.9 m and the thickness does not vay thoughout the height. They wee built on the concete suppoting pies with the dimension of 4.0m High by 4.0m wide by 3.5m thick. The thickness of the shell vaies fom 1.7 m close to the columns top end to 0.45 m at an elevation of 38.6 m. Fom thee, it deceases to 0.4 m at the elevation of 165.3 m, and the keep 0.4 m to the elevation of 214.6 m, then it incease to 0.65 m at the top. The cooling towe is built on a ing stip foundation, which is 4.0 m below gade and with a width of 14.0 m and an aveage height of 2.0 m. A concete stiffening ing (o tansient ing) with a thickness and width of 0.40 m and 1.7 m is built togethe with the uppe towe shell at the top of the X shaped columns. As well, a concete stiffening ing (o top ing) with a thickness and width of 0.45 m and 1.8 m, espectively, has been built at the top of the cooling towe. Fig. 1 shows the elevation plan of the R/C cooling towe. The mateial popeties of the cooling towe including the concete and einfocement ae shown in Table 2.1.
Table 2.1. Mateial popeties of the R/C cooling towe Mateial Yield point (MPa) Ultimate point (MPa) Concete -- 45 Reinfocement 335 445 220m 110m 107m 28.7m 195m 0.0 Figue 1. levation of the pototype cooling towe 2.2. Design and Constuction of the Model Stuctue with Length Ratio 1:30 All stuctual elements including the X shaped suppoting columns, towe shell and the column suppoting pies ae scale down to 1:30 of the pototype towe in geometic dimension. The vital coesponding dimensions ae shown in table 2.2. As in known, in the dynamic shaking table test fo small atio model stuctues, due to the dimension scale down, the mass missing will cause significant inetial foce loss and bing inevitable eo to the test esults. Fo compensating of the mass loss, atificial mass usually has to be used in the test. Howeve, due to the special shape and stuctue of the hypeboloid shell towe, it s difficult to add the atificial mass on the model s shell duing the dynamic eathquake simulation test. Fo solving this poblem, in design and constuction of the towe model, a
kind of specially teated lead sand was used as one of the main aggegates of the model mico-concete. The mico-concete s equivalent mass density eaches up to about 6700Kg/m 3 by mixing with the lead sand, almost 3 times of the common mico-concete s mass density. The simila design fo einfocements in all stuctual elements is contolled by the einfocement atio. As esult, the numbe of einfocement bas is deceased significantly fo consideing of the constuction convenience. Fo example, the longitudinal einfocement bas numbe in X shaped column deceases fom the pototype 6036 to the model 26+22. Coespondingly, in the dimensional analysis of the similitude law fo dynamic test, the equivalent density atio, length atio as well as the efficient elastic modulus atio can be set as the basic vaiables, and othe vaiables such as acceleation, fequency and time etc. could be deived fom the dynamics fomulation easily, shown in table 2.3. Table 2.2. Dimension scaling of the cooling towe stuctue No. Citical index Pototype towe Dimension of the stuctue 1:30 model towe 1 Oveall height 220m 7333mm 2 Diamete at thoat level 107 m 3567mm 3 Diamete at the foundation top 188 m 6267mm 4 Sealing stuctue edge 195 m 6500mm 5 Height at the top of X shaped column 28.7 m 957mm 6 Max./Min. thickness of the towe shell 1700mm/400mm 57mm/13mm 7 Height of the column suppoting pie 4000mm 133mm 8 Thickness of the ing foundation 2000mm 67mm 9 Absolute oveall height 226m 7533mm Table 2.3. Similitude elationship used in dynamic test and finite element analysis fo model stuctue Physical Similitude atio paametes Lowe excitation Medium excitation Lage excitation Length l l l quivalent modulus 0 f 2 0 1 1 2 f0 i f 2 0 i 2 f0 Density Stess 0 1 i
Time t l 0 0.5 t l 1 0.5 t i l 0.5 Defomation l l l Velocity 0 0.5 1 0.5 i 0.5 Acceleation a 0 1 a l l a i l Fequency l 1 0 0.5 l 1 1 0.5 l 1 i 0.5 Stain gauge S31~S36 Stain gauge S21~S26 Stain gauge S11~S16 Stain gauge S01~S06 Figue 2. X shaped column and the suppoting pies, ing foundation
Figue 3. Constuction of the 1:30 model cooling towe stuctue to be tested on shaking table 3. Finite lement Analyses fo Both of the Pototype and 1:30 Model Towe Due to the fact of thee is no accuate nonlinea similitude law available fo dynamic test and the uncetainty of most laboatoy tests, the esults obtained fom the shaking table test only can be used qualitatively and vey difficult to be used diectly to compae with the pototype stuctue design and analysis quantitatively. Theefoe, fo establishing the elative accuate quantitative elationship fo the linea and nonlinea dynamic esponse between the pototype and the model stuctue, it s vey necessay to do the nonlinea dynamic analysis fo both of the pototype stuctue and its dimension scale-down model countepat simultaneously. It s expected to though the compaison of the esults between the numeical analyses and the shaking table test, veify and enhance the eliability of the nonlinea analyses fo the 1:30 model stuctue, and then compae with the analytical esults fo the pototype stuctue, tansfe the dynamic test esults fo the 1:30 model to the pototype stuctue quantitatively. To pefom the seismic esponse analyses fo both of the pototype and its 1:30 model towe, two softwaes wee used in the study. The fist applied softwae is the commecial softwae ANSYS Vesion 14.0. The modal analyses, dead load analyses as well as the linea analyses fo the eathquake spectum esponse of the both stuctues wee caied out. The LS-DYNA vesion 14.0 was used to pefom the dynamic nonlinea analysis fo both of the pototype and model towe stuctue unde stong eathquake excitations (will be pesented in the confeence togethe with the shaking table test esults). The stuctual membes including column suppoting pies, X shaped columns, towe shell wee all modeled with solid elements, as shown in fig.4. The einfocement atio at diffeent height is taken into account in the column meshing scheme fo both of the pototype and model stuctue. The diffeence is in the finite element meshing fo the towe shell. That is fo the pototype towe, the meshing size fo the shell is contolled by the height of constuction template each laye (1.3m high, totally 150 layes), but fo the 1:30 model stuctue, fo enhancing the calculating speed effectively, most uppe shell is meshed with 1/6 of the height of the coesponding pototype elements except fo the shell close to the columns top (4 layes), with 1/30 of the size of the pototype elements. The feedom coupling was used to make sue the effective connection between the X shaped columns and thei suppoting pies.
a) Pototype towe b) 1:30 Model towe Figue 4. Finite element meshing fo the pototype towe and its 1:30 model stuctue 3.1. Compaison of the Modal Analysis Between Pototype and Its 1:30 Model Towe Fo undestanding the dynamic chaacteistics of the cooling towe stuctue, the modal analyses wee caied out fo both of the pototype towe and its 1:30 model stuctue espectively. Although thee ae some diffeence fo the element meshing between the pototype and its scale-down model, fom fig.5 and table 3.1, it can be seen that the vibation mode is quite simila with each othe. As example, only the fist 3 modes and the 31 st mode wee listed in fig.5 fo compaison. Table 3.1. Compaison of the modal analyses esults fo pototype towe and its 1:30 model Mode Fequency of the 1:30 model towe Fequency of the 1:30 model towe (Hz) (Hz) Similitude atio 1 10.01911 0.648679 15.445412 3 10.62728 0.689223 15.419225 5 10.66332 0.700281 15.227199 7 11.58211 0.765836 15.123478 9 14.09556 0.928196 15.185974 31 21.26193 1.373466 15.480494
a) Pototype towe b) 1:30 Model towe Figue 5. Compaison of pat vibation modes of the pototype towe and its 1:30 model stuctue
3.2. Response Spectum Analysis fo the Pototype and Its 1:30 Model Towe Fo estimating the oveall seismic capacity of the cooling towe, the fist step should be check the esults of the esponse spectum analysis (RSA) unde the seismic design intensity 8 in Chinese eathquake intensity scale. Fo consideing the most disadvantage case, this pape chooses the most disadvantage site class IV (soft soil with shea velocity no lage than 150m/sec., with pominent peiod 0.65sec, which is almost the same with the fist vibation mode s natual peiod 0.65sec of the pototype towe.) as one of the conditions to detemine the input spectum. In the RSA the dead load (weight) is also taken into account. Fo compaing, the analytical esults fo defomation and stess fo both of the pototype and its 1:30 model ae abstacted in the fig.6 and table 3.2~3.3., espectively. quivalent defomation contou quivalent Von-Mises stess contou a) Pototype towe b) 1:30 Model towe Figue 6. Compaison of the RSA esults fo pototype and its 1:30 model stuctue
Table 3.2. Maximum defomation esponse (esponse spectum + dead load)mm Load case Pototype 1:30 model Solution atio Design atio Solution/design atio Dead load only 17.3483 0.077027 0.00444 0.033333 0.133201 Hoizontal RS 78.9246 2.47151 0.031315 0.033333 0.939446 Vetical RS 5.92809 0.168633 0.028446 0.033333 0.853394 Standad Q. RS 27.6923 0.866988 0.031308 0.033333 0.939238 Dead load plus RS 40.1235 1.12556 0.028052 0.033333 0.841572 Table 3.2. Maximum Von-Mises stess esponse (esponse spectum + dead load)n/mm 2 Load case Pototype 1:30 model Solution atio Design atio Solution/design atio Dead load only 6921.01 475.816 0.06875 0.537 0.128025 Hoizontal RS 17130.1 8479.2 0.494988 0.537 0.921766 Vetical RS 1589.83 749.591 0.471491 0.537 0.87801 Standad Q. RS 6017.37 2978.31 0.494952 0.537 0.921699 Dead load plus RS 8829.34 3796.04 0.429935 0.537 0.800623 4. CONCLUDING RMARKS In summay, due to the papes limitation, this pape hee only povides basic eseach activities and peliminay esults fo both of the pototype cooling towe and its 1:30 model vey biefly. The peliminay study shows that thee huge of analytical and test eseaches should be caied out fo undestanding the accuate eathquake esponse of the exta-huge cooling towe. The esults fo both of the nonlinea dynamic analysis and the shaking table test as well as the compaison analyses will be epoted in the following papes and in the oal pesentation of the 15WC. ACKNOW LDGMNT The authos appeciate the financial suppot fom the athquake Scientific Reseach Funds Pogam (No. 201208013) and the China National Natual Science Foundation Pogam (No.51078336). RFRNCS J.Y. LI, C.L. RN, Z.L. HUANG. (2007). Test and F Analyses fo natual ventilation cooling towe. Mechanics Quately. 28(3), 443-447. M. WANG, Z.L. Huang, L.Q. LI. (2006). F Analysis fo natual ventilation smoke-cooling towe. Mechanics and Pactice. 28(4), 64-67. Aksu T. (1996). A finite element fomulation fo column suppoted hypeboloid cooling towes. Computes & Stuctues. 59(5), 965-974. Castiau T. and Gauios R. (1991). The design of cooling towes in extemely sevee eathquake conditions. ngineeing Stuctue. 13.