Development of Seismic Force Requirements for Non-structural Components in Taiwan

Transcription

1 Develoment of Seismic Force Requirements for Non-structural Comonents in Taiwan Fan-Ru Lin 1, Juin-Fu Chai 2 and Jing-han Lee 3 1 Assistant Researcher, Center for Research on Earthquake Engineering (NCREE), Taiei, Chinese Taiwan 2 Researcher, Center for Research on Earthquake Engineering (NCREE), Taiei, Chinese Taiwan 3 M.D,Deartment of Construction Engineering, Taiwan University of Science and Technology(NTUST), Taiei, Chinese Taiwan frlin@ncree.org.tw, chai@ncree.org.tw, ABSTRACT : Taiwan is located in the circum-pacific seismic zone. Therefore, the seismic design methods and standards for structures have significantly rogressed and been reformed in recent decades. However, earthquake resistant design of equiment and non-structural comonents is still not oularized in Taiwan. To investigate the alicability of the seismic design standards to non-structural comonents, the rescribed requirements for the high-tech equiment according to Taiwan seismic design guideline and SEMI S2 were comared in this aer. In addition, a ractical formula was roosed for the seismic design of non-structural comonents in Taiwan area. The floor resonse sectra at the target location within an interesting building were develoed in order to determine the required seismic design force for the attachments of nonstructural comonents. Meanwhile, in this aer, the seismic force requirements for several non-structural comonents secified by Taiwan guideline, NEHRP and IBC were comared for the follow-u research. On the other hand, NCREE shall take the resonsibility to establish an aroriate environment for the increasing requests of non-structural seismic qualification in Taiwan. The inut time histories for shaking table test in comliance with the requirements of AC-156 could be generated numerically. The synthetic strong motions which were comatible with the secified resonse sectrum were simulated by using the hase sectrum of a recorded floor resonse because it kees the non-stationary characteristics of an actual resonse. In addition, the excitations of the shaking table at NCREE and the associated test resonse sectrum were achieved and discussed in this aer. KEYWORDS: Non-structural Seismic Force Requirements, floor resonse sectrum, seismic qualification testing, AC-156, high-tech equiment 1. INTRODUCTION Seismic design for nonstructural comonents and equiment is little regarded by architects and structural engineers in Taiwan. There are some reasons resulting in the henomenon. For examle, only the seismic erformance of structural comonents is requested in Building Act, and lots of nonstructural comonents are installed after the usage building licenses issued. Nowadays there are no articular seismic guidelines for non-structural comonents in Taiwan, so that Taiwan Building Code (TBC) and Outline Secifications for Public Construction are alied to the seismic design of mechanical and electrical comonents. Nevertheless, the anchorage and bracing requirements of Outline Secifications are considered for the functionalities of equiment rather than the seismic demands. Through strong earthquake exeriences in recent years, eole gradually realize that although the structural damages may not harm the building function, several imortant facilities and buildings will be forced to be out of oeration due to the non-structural roblems. In order to meet more and more requirements for imroving seismic erformance of nonstructural comonents, there are three arts discussed in this aer. First, the seismic design secification for nonstructural comonents in Taiwan is comared with NEHRP23 (FEMA45) and IBC26 briefly. Then, an equation to calculate the reasonable

2 seismic design force is recommended for the equiment of high-tech industries in Taiwan. Finally, the synthetic time histories for shaking table test were generated numerically in comliance with the requirements secified by AC-156. In addition, the excitations of the shaking table at NCREE and the associated test resonse sectra would be discussed in this aer briefly. 2. PRESCRIPTIVE SEISMIC DESIGN FORCE REQUIREMENTS 2.1 Background Previously, seismic force requirements for nonstructural comonents in Taiwan mostly referred to Uniform Building Code (UBC). From 1982 to 24, the seismic force was calculated by the equation: F =ZIC W, where F was the seismic design force, Z and I were the seismic zone coefficient and occuancy imortance factor, C was the comonent coefficient, and W was the comonent weight. Values of Z and I were the same as for structural design. The value of C was 2. or.75 decided according to the ductile or brittle fracture of comonents, resectively. From 25 to date, referred to IBC 2, the nonstructural requirements are substantially revised in TBC, and the seismic force is calculated by the following equation: a F ) h =. 4S DS I ( 1+ 2hx / hn W (2.1) Ra where S DS is the design sectral resonse acceleration at short eriods, I is the comonent imortance factor for equiment or comonents, a is comonent amlification factor, and R a is allowable comonent resonse reduction coefficient. In Eq. (2.1),.4 S DS is regarded as the Effective Peak Ground Acceleration (EPA) at the free field, and.4s DS (1+2h x /h n ) reresents the acceleration at the oint of attachment of comonent with a height h x resect to the ground level. Furthermore, the values of F shall not be taken as less than.3s DS I W but needn t be greater than 1.6S DS I W. Horizontal Sesimic Design Acceleration (g) Ra for Taiei Basin (TBC) Ra for general site (TBC) R (NEHRP 23) R Values in NEHRP 23 (a) at Roof level (b) at mid-story level (c) at ground level Fig. 1: Influence of different Resonse reduction coefficients (R a or R ) on F h There are two main differences in nonstructural requirements between TBC and NEHRP23. The first one is the vertical seismic force, F v. In TBC, F v is defined by 1/2F h for general sites and 2/3F h for near-fault regions, while F v being defined by.2s DS W in NEHRP23. The other difference is about the values of comonent resonse reduction coefficients, R in NEHRP23 and R a in TBC. Due to the uncertainties of installation quality and ductility caacity of comonents and attachments, the values of comonent resonse reduction coefficient R in TBC are defined within the range of 1.25 to 3.5, while the values being defined between 1.5 and 5. in NEHRP23. Furthermore, in TBC, in order to control the damage level under the design base

3 earthquake, the value of R is reduced to the allowable comonent resonse reduction coefficient R a for nonstructural comonents according to the same secifications for structural system. The effect of R a or R on F h is shown in Fig. 1, where S DS is defined by.8, and a and I are all equal to 1.. At roof level, the values of F h in Taiei Basin are greater than those calculated by NEHRP23 with a ratio of 1.33 to 1.79, it seems a little conservative to design nonstructural comonents by TBC. At ground level, most F h aren t affected by R a or R because F h with R equal or greater than 2.5 are controlled by the lower bond.3s DS I W. 2.2 Comarison of TBC with NEHRP 23 and IBC 26 Recently, nonstructural requirements in the United States change a lot in IBC26 (ASCE 7-5). To show the differences among TBC, NEHRP 23 and IBC 26, the horizontal seismic design forces for nonstructural comonents at ground or roof level of a regional hosital in Taiei are calculated according to different codes. The comonents include HVAC, elevator comonents, electrical comonents, and iing systems. The results are shown in Fig. 2, where S DS is.6, and I is equal to 1.5 due to the occuancy factor of regional hosital buildings. Piing system are classified in accordance with ASME B31, high-deformability materials with joints made by welding or brazing, limited-deformability materials with joints made by threading or grooved coulings, and low-deformability materials. It can be observed from Fig. 2: (1) Design forces for sring-isolated HVAC equiment at both ground and roof level are much larger than other comonents because of the larger value of a for sring-isolated HVAC equiment. In IBC26, though a of vibration isolated comonents and distribution systems are all equal to 2.5, the associated design acceleration for isolated HVAC equiment is still larger because the lower value of R is defined for sring-isolated comonents. (2) Excet electrical comonents,the design forces for other comonents at the roof level determined by TBC are conservative with resect to other codes. (3) Excet sring-isolated HVAC equiment and low-deformability iing, the design forces for other comonents at the ground level are the same, because they are all controlled by the lower bond.3s DS I W,. Meanwhile, the design force for sring-isolated HVAC equiment is controlled by the uer bond 1.6 S DS I W. (4) In IBC 26, the lower seismic demands can be determined for iing system in accordance with ASME B31 (category 5 and 7 in Fig. 2). Horizontal Sesimic Design Acceleration (g) NEHRP23 TBC IBC Categories of Nonstructural Comonents (a) at ground level (b) at roof level (c) difference between at ground and roof level 1. Nonvibration Isolated HVAC System 2. Sring Isolated HVAC System 3.Elevator Comonents 4.Electrical Comonents 5.High-Deformability Piing (ASME B31) 6.High-Deformability Piing 7.Limited-Deformability Piing(ASME B31) 8.Limited-Deformability Piing 9.Low-Deformability Piing Fig. 2: Comarison of horizontal seismic design acceleration according to different Codes 3. SEISMIC DESIGN FOR EQUIPMENT IN HI-TECH FABS 3.1 Seismic Demand Secified by SEMI S2 and TBC

4 In addition to TBC for general nonstructural comonents, some secial or imortant comonents, like hi-tech equiment, need secific seismic requirements to ensure their seismic erformance. For instance, the equiment in roduction line of semiconductor industry in Taiwan shall comly with SEMI S2, a standard develoed by Semiconductor Equiment and Materials International. Based on UBC 1997, the simlified seismic design force secified by SEMI S2 is shown as follows: a Fh = Ca ( hx hn ) I W = 1. 32W (3.1) R This is based on the assumtion that the site is located at seismic zone 4 in United States, with soil rofile tye S D and 5 km from a seismic source tye A, and the HPM (Hazardous Production Materials) equiment is attached at the mid-story level with shallow anchor bolts. In SEMI S2, because of the high fundamental frequency, the tyical semiconductor equiment is assumed as a rigid structure. There are some roblems to adot Eqn. 3.1 for semiconductor industry in Taiwan. First of all, the format is based on UBC 1997, and it is much different from that secified by TBC which is develoed on the basis of IBC2,. Secondly, for a tyical hi-tech fab in Taiwan, the roduction line in cleaning rooms are usually located at the to level of internal structure, which is searated from the outside truss structure. Therefore, equiment shall be assumed to be mounted at the roof level instead of the mid-story level. In order to determine the most suitable seismic force for semiconductor facilities in Taiwan, two alternative simlified equations are discussed in this section. One is the rescritive equation based on the format secified by TBC, and the other is modified by real floor resonse sectrum of hi-tech fabs. Based on the critical assumtion that the HPM equiment is mounted by short anchored bolts at the roof level of a hi-tech fab in Tainan Science Park with 2 km from Hsinhua fault, the rescritive seismic force can be defined as follows: Tainan Science Park: 1. Fh = W = 1. 33W (3.2) 1.33 Comared with Eqn. 3.1, though the original equation and assumtions are not the same, the resulted seismic demands are quite similar. For HPM equiment in Hsin-Chu Science Park with no active fault nearby and smaller S DS than that at Tainan Science Park, the rescritive seismic force can be also defined as follows: Hsin-Chu Science Park: 1. Fh = W =. 95W (3.3) Alternative Equation Based on Normalized Equiment Design Resonse Sectrum From Eqn. 3.1 to Eqn. 3.3, we can see the dynamic characteristics of structures of Hi-tech fabs aren t clearly manifested. To obtain more real interaction between structure and equiment, the amlification factor A FP is roosed in this study on the basis of time history analysis. The time histories of floor resonse of tyical Hi-tech fabs located at Hsin-Chu Science Park corresonding to the sectrum comatible design earthquake (Chai, 25) are used in this study. Reasonable numerical models were built according to the structural characteristics of the selected fabs, including the equivalent late element for diahragm, the RC elements and steel trusses, as well as the mass distributions of heavy equiment (Chai, 25). The fundamental frequencies for the selected hi-tech fabs in both X- and Y- directions were from 1.9 to 3.3 Hz, which were quite close to those obtained by ambient measurements. In addition, according to the fundamental frequency, the tyical

5 Hi-tech fabs can be taken for short-eriod structures both in Tainan and Hsin-Chu Science Parks. As shown in Fig. 3, based on the simulated inut ground motions a g (t) that is consistent with the maximum otential earthquake with a return eriod of 475 years, the time histories of floor resonse R es1 (T s ;t) at each reresentative location of the tyical Hi-tech fabs can be determined, where T s is the fundamental eriod of the selected Hi-tech fabs. Based on each time histories of floor resonse, the associated normalized floor resonse sectrum (2%-daming ratio) can be determined while scaling the eak floor resonse R es1 (T s ; t) to 1g. Then, the amlification factor A FP can be defined by the normalized floor resonse sectrum, which is a function of the dimensionless frequency F /F s, where F and F s are the fundamental frequencies of the equiment and the selected Hi-tech fab, resectively. The amlification factors can be determined for the four selected hi-tech fabs in both X- and Y- directions, and the mean and mean+1σ of the eight amlification factors are shown in Fig. 4. R e (T s,t ; t ) HPM equiment (T, 2%) R es1 (T S ;t ) R es2 (T S ;t ) Hi-Tech Lab SDOF system (T s, 5%) (T s, 5%) a g (t) Design Sectrum Comatible Time Histories Fig. 3 Procedure of Analysis Fig. 4 Comarison of the amlification factor On the other hand, both the equiment and the suorting structure can be modeled by SDOF systems with fundamental eriods of T and T s, and daming ratios of 2% and 5%, resectively. The resonses of equiment and the suorting structure are assumed in elastic ranges. Then, based on the same inut ground motion as used for the dynamic analysis of multi-degree models, the time histories of the resonse R es2 (T s ; t) and the associated sectrum S ad (T s ) of the SDOF structure can be determined. Similarly, based on the time histories of resonse R es2 (T s ; t), the associated floor resonse sectrum S ap (T s, T ) can be determined, and then the amlification factor A FP can be defined by A FP = S ap (T s, T ) / S ad (T s ) (3.4) Figure 4 shows the mean and mean+1σ of the amlification factors determined under SDOF assumtion. It can be found from Fig. 4 that the amlification factor determined by MDOF system can reflect the high mode effect significantly. Based on the amlification factor, the simlified equation to determined seismic design force for attachments of equiment is roosed as follows: A S FP ad F h = I W (3.5) R a where the value of A FP for flexible equiment is defined by 2., and 1. for rigid equiment. Herein, a rigid equiment is defined as the associated ratio F /F s is equal to or greater than 8., and a flexible equiment is defined as the associated F /F s is between 2. and 8.. Equiment with F /F s smaller than 2. has great ossibility to be resonant with the suorting structures, and hence is recommended to be analyzed exressly. Generally seaking, a rigid equiment can be defined as its fundamental frequency is higher than 16.6Hz. In this case, the frequency of a rigid equiment may be greater than 16. Hz since F s of the selected tyical Hi-tech

6 fabs are larger than 2. Hz. The comarison of amlification factor and the normalized equiment resonse sectrum according to different studies is shown as Fig. 5, which is under the assumtion that the equiment is mounted at roof level, and the suorting building is classified as a short-eriod structure. From Figs. 4 and 5, we can see that using design sectra as adoted by revious studies may underestimate the effect of higher modes of Hi-tech fabs. Following the same conditions in section 3.2, the design force for HPM equiment is calculated by Eqn. 3.6: Tainan Science Park: Fh = 1.5 W = 1. 11W 1.33 (3.6a) 1..7 Hsin-Chu Science Park: Fh = 1.5 W =. 79W 1.33 (3.6b) where S ad is equal to S DS because the selected fabs are recognized as short-eriod structures at both Tainan and Hsin-Chu Science Park, and HPM equiment is still regard as a rigid comonent. Comaring to Eqn. 3.2 and 3.3, the seismic demand based on Eqn. 3.8 is reduced to 83%. But for flexible equiment with F /F s <8., the seismic demands would be amlified by 1.66 times. Amlification Factor (S ap/sad) NCEER STUDY 1999NEHRP This Study Normalized Frequency (F /F s) Fig. 5: Comarison of normalized equiment resonse sectrum according to different studies 4. SEISMIC QUALIFICATION TESTING OF NONSTRUCTURAL COMPONENTS AC156, issued by ICC Evaluation Service, Inc., is one of the seismic qualification testing guidelines for general tyes of nonstructural comonents that are aroved by IBC 26 and ASCE 7-5. The current edition of AC156 (27) is alicable to nonstructural comonents and systems with fundamental frequencies greater than or equal to 1.3 Hz. Either triaxial, biaxial or uniaxial testing can be erformed in comliance with the secified requirements. Other testing guidelines used for nonstructural comonents include GR-63-CORE and IEEE 344, which are develoed for telecommunication equiment and Class 1E equiment in nuclear ower generating stations, resectively. Fig. 6 shows the comarison of three nonstructural seismic qualification guidelines. In accordance with AC156 and GR-63-CORE, the associated RRSs (Required Resonse Sectrum) for the region with highest S DS value (S DS =1.136 g) in Taiwan are comared in Fig. 7. The shae of the RRS corresonding to GR-63-CORE almost enveloes the one in comliance with AC156, and the ratio is u to 2.75 in the range of 2 5 Hz for the horizontal directions. In addition, the RRS secified for vertical direction in GR-63-CORE is much larger than that secified in AC156 because GR-63-CORE adots the same RRS for both vertical and horizontal directions. Furthermore, comared with other equiment, the weight of telecommunication equiment is relatively low, and hence, it should be noted that the tested equiment with heavy weight may be damaged owing to the extremely inertial load during the qualification testing in comliance with the requirements of GR-63-CORE.

7 AC 156 GR-63-CORE (Zone 4) IEEE-344 Resonant Frequency Search Test Amlitude of Swee.1g±.5g.2g.2g Frequency Range 1.3~33.3Hz 1~5Hz 1~33Hz Swee Rate 2 Oct. /1 min. 1 Oct. /1 min. equal or less than 2 Oct. /1 min. Multifrequency Seismic Simulation Test RRS Daming Ratio 5% 2% 5% Frequency Range 1.3Hz~33.3Hz 1.Hz~5.Hz 1.Hz~33.Hz Amlified Rigion 1.3~8.3Hz 2~5Hz - A flex (Roof floor) 1.6S DS 5.g - A rig (Roof Floor) 1.2S DS 1.6g - RRS of Vertical direction A flex=2/3 S DS A rig=2/3(.4s DS) same as horizontal direction - TRS Daming Ratio 5% 2% equal or larger than 5% Tolerance uer limit 13% 13% (1~7Hz) - Tolerance lower limit 9% (conditional) 1% 9% (conditional) Inut equal or larger than Motion Peak Acceleration ZPA value of RRS Total Duration 3 sec., at least 2 sec. of strong motion - the designated motion is about 3 sec. equal or larger than ZPA value of RRS at least 15 sec. of strong motion Fig. 6 Requirements for seismic qualification testing in AC156, GR-63-CORE and IEEE-344 Sectral Resonse Acceleration (g) AC156 Horz. Horz.+3% Horz.-1% AC156 Vert. Vert.+3% Vert.-1% GR 63Core Frequency (Hz) 1 1 Fig. 7 RRS of AC156 and GR-63-CORE Based on the incororation with NCREE laboratory in roceeding some seismic qualification shaking table tests of nonstructural comonents, the alicability of AC156 can be further understood. The synthetic inut motions for shaking-table tests can be generated on the basis of the hase sectrum of an observed ground motions and the modified Fourier amlitude, such that the associated resonse sectrum is comatible with the RRS secified by AC156. Based on the hase sectrum and a trial Fourier amlitude A m (ω), the time histories of excitation can be recovered and then the associated sectral resonse acceleration S am (T ) for any nonstructural eriod T can be determined. Because the resonse will be dominated by the excitation with resonant frequency, the modification factor for the Fourier amlitude at a circular frequency of ω =2π/T can be defined by MF m ( ω= 2π / T ap am ω ) = S ( T ) / S ( T ) (4.1) P where S ap (T P ) is the sectral acceleration demand at eriod T according to RRS. Then, the Fourier amlitude A m+1 (ω) on the next iteration can be defined by A m+ 1( ω) = MFm ( ω) Am ( ω) (4.2) This iteration rocess will be continued until MF m (ω) is equal to 1. at all frequencies, and hence the final recovered excitation will be comatible with the RRS. Figure 9 shows the synthetic inut motions that are generated by keeing the hase sectra of Chi-Chi earthquake recorded in CHY9 and TCU54 stations, and it can be observed that the associated horizontal and vertical resonse sectra are comatible with the RRS well. CHY9 TCU54 Fig. 9 Synthetic time histories and resonse sectra comatible with AC-156 RRS For the testing in comliance with AC156, in order to study the influence of heavy weight on TRS (test resonse sectrum), a set of lead blocks with 28 tons was used first in the ilot test to model the tested equiment with a weight of 3 tons. In contrast to the case without any secimen, it can be found that the TRS in x-direction is affected significantly by the heavy weight of secimen. As shown in Fig. 1, the measured TRS

8 corresonding to heavy secimen exceeds the tolerance (3% of RRS) over the range of 1~5Hz and 15~2Hz, and the results may be caused by table and secimen weight, resonance roblem, force feedback, and stability of table erformance. The most effective solution to this roblem is to carry out a model measurement test with secimen attached to the table before executing the formal qualification test. The model test can incororate the comensation in a time history test to remove the dynamic effects of the control system by adjusting the drive signal. However, it needs to be considered that the secimen may be damaged during the model test due to the large amlitude (usually 35 gal) and long durations (aroximately 6 minutes). Sectrum Acceleration (g) With Secimen Without Secimen RRS RRS x 13% RRS x 9% Frequency (Hz) (a) X direction (b) Y direction (c) Z direction Fig. 1 TRS and RRS in the testing for AC CONCLUSIONS In the first art of this aer, seismic requirements in TBC, IBC 26 and NEHRP23 are comared briefly. In the second art, the equation used to determine the reasonable seismic design force based on normalized floor resonse sectrum is recommended for the equiment of hi-tech industries in Taiwan. In the third art, the testing guidelines used for seismic qualification of nonstructural comonents are comared and the synthetic time histories which are comatible with the RRS secified by AC-156 were generated. Potential roblems of execution of shaking table tests for seismic qualification are addressed as well. REFERENCES Ministry of the Interior (25). Taiwan Building Code: Seismic Design Secifications and Commentary of Buildings, Ministry of the Interior, Chinese Taiwan (in Chinese). Chai, J.-F., Loh, C.-H., Jean, W.-Y. and Liao, W.-I (25). Develoment of Floor Design Resonse Sectra for Hi-tech FAB, Research Reort, NSC Z (in Chinese) Semiconductor Equiment and Materials International (2). Semi S2-2: Environmental Health and Safety Guidelines For Semiconductor Manufacturing Equiment. ICC Evaluation Service inc. (27). AC156: Accetance Criteria for Seismic Qualification by Shake-table Testing of Nonstructural Comonents and Systems, ICC Evaluation Service inc., U.S. Telcordia Technologies, Inc.(22). NEBS Requirements: Physical Protection. GR-63-CORE. Issue 2. Chester, NJ: Telecordia Technologies, Inc., U.S. IEEE-SA Standards Board (24). IEEE 344 : IEEE Recommended Practice for Seismic Qualification of Class 1E Equiment for Nuclear Power Generating Stations, IEEE-SA Standards Board, U.S.