A RATIONAL APPROACH TO SEISMIC QUALIFICATION TESTING OF NONSTRUCTURAL BUILDING COMPONENTS

Transcription

1 13 th World Conference on Earthquake Engineering Vancouver, B.C., Canada August 1-6, 2004 Paer No A RATIONAL APPROACH TO SEISMIC QUALIFICATION TESTING OF NONSTRUCTURAL BUILDING COMPONENTS Jeffrey A. GATSCHER 1, Phili J. CALDWELL 2, Robert E. BACHMAN 3 SUMMARY A rational rocedure is develoed for alying equivalent nonstructural static force design requirements to generate a dynamic qualification test method for seismic validation of equiment for commercial (nonnuclear) service. The equiment qualification required resonse sectrum (RRS) is derived based on the National Earthquake Hazard Reduction Program s (NEHRP) lateral force rocedure in conjunction with the building design earthquake resonse sectrum. This aroach accounts for above grade elevation equiment installations, with or without knowing the building dynamic characteristics. Provisions are included to extend the qualification methodology to families of roducts by emloying tye-testing rationalization techniques. A well-defined ass/fail accetance criterion is established that utilizes the equiment imortance factor to define ost-test accetability. This rocedure is the first recognized seismic qualification test rotocol for shake-table testing to address secondary system and nonstructural building comonent requirements as defined by the rimary model building codes being emloyed in the United States. INTRODUCTION Seismic qualification of nonstructural building comonents to meet the requirements secified in model building codes introduces significant challenges to equiment manufactures. Nonstructural comonents are defined as the entire grou of miscellaneous building structural subsystems, architectural elements, mechanical and electrical equiment, and elements ermanently attached to rimary structures. The challenge resents itself on three areas: (1) translation of the building codes nonstructural lateral force requirements into dynamic test arameters, (2) imlementation of an equiment qualification strategy that is comatible with given requirements and (3) romoting awareness for shared resonsibility between equiment manufactures, oerators and installers in order to maximize the robability that equiment will be intact and functional after a seismic event. All of these issues require close examination and are the focal oints of this aer. 1 Senior Staff Engineer, Schneider Electric, Nashville, TN, USA. gatscher@squared.com 2 Staff Engineer, Schneider Electric, Seneca, SC, USA. caldwel@squared.com 3 Structural Engineer, Bachman Consulting, Sacramento, CA, USA. rebachmanse@aol.com

2 Regarding the first issue, resent model building code requirements for nonstructural comonents are secified using rescritive lateral force rocedures. This imlies that for any given nonstructural comonent, the seismic requirement is rescribed by a force magnitude, which is simly an equivalent static acceleration coefficient times the comonent weight. Therefore, the seismic demand requirement for any nonstructural comonent is a static lateral force value that is deendent uon several design factors and arameters as secified in code rovisions. The magnitude of this lateral force coefficient is then the comonent design requirement and a manufacturer can use analytical methods to verify that the comonent s seismic caacity exceeds the rescribed demand. However, design assurance validation based on analytical techniques is less reliable than dynamic shake-table testing, esecially for oerating equiment. In addition, there is a growing marketlace interest in roviding design assurance that seismic caacity exceeds demand based on results of shake-table qualification testing, since this is the only ractical way to verify equiment functionality after the seismic event. Therefore, from the manufacturer s ersective, the desire is to achieve seismic qualification via dynamic shake-table testing in order to satisfy the lateral force rocedures that are secified in the model building codes. The key issue is defining a building code secific test standard that is universally acceted in industry such that all nonstructural comonent suliers are rating their equiment offerings using a common yardstick. Regarding the second issue of imlementing an effective nonstructural qualification strategy, manufacturers need the ability to seismically qualify a wide variety of equiment tyes. Within each major tye category, there may exist several roduct families, each of which may offer configurable variations. Each equiment tye will have fundamental structural differences in design construction and thus offer various levels of seismic resistance. In other words, some equiment latforms may be less resistant to lateral forces than other tyes and need to be qualified to lesser acceleration levels. In addition, testing every ossible configuration of a given roduct latform is neither ractical nor feasible. Therefore, it is necessary to select test secimens that adequately reresent the entire equiment roduct line family for seismic qualification. This necessitates the establishment of a generic rationalization criterion that can be used to determine test unit configuration requirements for the roer reresentation of roduct latform families. The third issue concerns romoting general awareness of seismic risk mitigation and the concet of shared resonsibility between the rimary stakeholders. Mitigating seismic risk and achieving the lowest ossible risk associated with seismic erformance of nonstructural equiment deends on the concet of shared resonsibility between equiment oerators, installers, and manufacturers. It is the resonsibility of equiment manufacturers to conduct seismic qualification testing rograms in alignment with the intentions of code writing authorities by using standardized test rocedures. The equiment manufacturer determines that the equiment will be functional following a seismic event via shake-table testing. However, roer anchoring of nonstructural comonents is absolutely critical. The manufacturer can design equiment based on successful shake-table test results, but if the equiment is not attached to the building structure in accordance with the minimum standards recommended by the structural engineer of record, the equiment might overturn or shear the attachment devices and slide off its foundation. Such movement may damage either the building structure or other comonents. History has time-after-time told us that imroerly anchored equiment will not erform satisfactorily during the seismic event and will likely cause catastrohic damage. It is the equiment secifier/installer s rimary resonsibility to determine that the nonstructural comonent is rigidly suorted and will not leave its foundation during a seismic event. This oint cannot be overstated. Incorrectly anchored or even worse unanchored equiment is a recie for disaster and is not comatible with a seismic risk mitigation hilosohy. The roblem faced by equiment manufacturers is to correctly interret and relate the model building code requirements for nonstructural comonents into dynamic qualification test requirements for seismic validation of equiment designs in a uniform fashion that can be imlemented in industry. Based on the

3 above stated reasons, it is clear that secondary systems and nonstructural building comonents should be the subject of a ractical and comrehensive seismic qualification testing rotocol in order to eliminate subjective interretations and non-uniform test methods. The aroach resented herein rovides the foundation needed to standardize equiment qualification ractices by establishing methodologies and accetance criteria, thereby consistently meeting the nonstructural seismic requirements as stated in the model building codes that are utilized in the United States. INCREASED IMPORTANCE OF SECONDARY SYSTEMS In site of their name, secondary systems are far from being secondary in imortance. It is now widely recognized that the survival of secondary systems and nonstructural building comonents is essential to rovide emergency and recovery services in the aftermath of an earthquake. However, the evolution of model building codes in the United States has been a rocess that fundamentally focused on the rimary structural system the building. Consideration for secondary systems and nonstructural comonents has historically been given much less attention than that given to the rimary structural systems. Nonetheless, in the last decade, nonstructural comonents have been given a renewed interest from code writing authorities and have been strongly influenced by the National Earthquake Hazards Reduction Program, BSSC [1] and BSSC [2]. This focus has led to the enactment of significant changes in this area. It is uniformly acceted that damage to secondary systems reresents a threat to life, may seriously imair a building s function, and may result in major direct and indirect economic losses. During the 2001 Nisqually earthquake that hit the Seattle/Tacoma region, it was estimated that the majority of the estimated $2 billion dollar loss was associated with damage to nonstructural comonents, Filiatrault [3]. In regard to the economic imact caused by the failure of nonstructural comonents, evidence from ast earthquakes has reeatedly shown that the cost associated with the loss of the nonstructural comonents themselves, the loss of inventory and the loss of business income may easily exceed the relacement cost of the building that houses those nonstructural comonents, EERI [4] and Naeim [5]. The rincial contributor for recent building code changes effecting the nonstructural comonent requirements is related to the increased availability of seismic exerience data. The United States Geological Survey and the California Geological Survey have made significant rogress in the widesread field installation of seismic monitoring instrumentation and in collecting seismology records over the ast two decades. This earthquake exerience data has been reduced, analyzed and the results have been incororated into the NEHRP source rovisions for nonstructural comonents, Gillengerten [6]. Advancing the state of the art by enriching and effectively bringing some science to the emirically driven design requirement definition for secondary systems and nonstructural comonents is a major ste forward for the model building codes. It is now u to industry to kee in ste and standardize on a uniform equiment qualification test rotocol that can be generic in its alication, and yet secific enough in its imlementation, such that seismic hazard risk mitigation becomes a relatively straightforward design assurance activity that all equiment manufacturers can ursue. THE SEISMIC RISK MITIGATION PHILOSOPHY Building owners, develoers and industrial enterrises imose uon their subcontractors and suliers the requirements detailed in the revailing building code being emloyed at the building site location. These building code requirements are catured by the equiment secifying agent and then imosed uon both equiment manufacturers who are sulying equiment to these locations and to the equiment installer who erforms the required installation tasks. Mitigating seismic risk and achieving the lowest ossible

4 risk associated with seismic erformance of nonstructural comonents deends on the concet of shared resonsibility between equiment oerators, installers, and manufacturers. At the surface, this concet aears simle and strait forward to imlement. However, in reality the resonsibility boundaries between the rimary stakeholders are not clearly understood by the resective arties. The manufacturer must roerly design and detail the equiment s force resisting members and subassemblies and also rovide the means for tie-down attachment to the rimary suort structure. The equiment manufacturer s rimary resonsibility is to determine that the equiment will be functional following a seismic event via shake-table testing using seismic qualification testing rograms that are in alignment with the intentions of code writing authorities. It is the resonsibility of the building owner s structural engineer of record to roerly design and detail the equiment s anchorage system. The building s structural engineer is rimarily resonsible to determine that the equiment is rigidly suorted and will not leave its foundation during a seismic event. The equiment installer must roerly install the equiment based on the installation recommendations made by both the equiment manufacturer and anchorage system manufacturer. It should be noted that in addition to anchorage details, the structural engineer of record should evaluate otential seismic interaction of the equiment with other nearby features including distribution systems attached to it. Imact and differential dislacement on attached systems have been major contributors to nonstructural comonent failures in ast earthquakes. Attached conduit, tubing and interface elements require adequate slack and flexibility. Adequate clearance should be rovided to revent imact with other features. These issues are system-level nonstructural seismic concerns that are not exlicitly catured by shake-table testing erformed by equiment manufactures. The rest of this aer is dedicated to describing a generic test rocedure that can be used to satisfy the nonstructural manufacturers ortion of the shared resonsibility with the overall goal to maximize the robability that the equiment will be intact and functional after a seismic event. EQUIPMENT QUALIFICATION TESTING PROTOCOL Objectives Square D / Schneider Electric established the goal of develoing a generic dynamic qualification test method for seismic validation of equiment for commercial (non-nuclear) service, which would be consistent with the lateral force design requirements for nonstructural comonents. The aroach taken was to: Develo the technical basis for these requirements/methods and relate them to relevant interretations of the 1997 and 2000 NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, BSSC [7] and BSSC [8]. Account for above grade elevation equiment installations with or without knowing the dynamic characteristics of the rimary suort structure (i.e., rimary structure dynamic roerties not necessary, but if available, may be used). Define and establish a verifiable ass/fail accetance criterion for the seismic qualification test based uon the equiment imortance factor. Develo a generic rationalization criterion that is used to establish test unit configuration requirements for reresenting roduct line families. Utilize recognized and acceted rocedures for the develoment of the time history waveform used during the dynamic shake-table test. Gain national accetance for the resulting seismic qualification test rotocol and technical NEHRP interretations by at least one credible model building code organization. Promote concet of shared resonsibility between equiment oerators, installers, and manufacturers comatible with a comrehensive seismic risk mitigation hilosohy

5 Standards Develoment Background The develoment of standardized equiment qualification test rotocols is an activity that has been erformed for several industries in the United States. The nuclear ower industry has a rich history of advancing seismic qualification test rocedures. The rimary organization driving that effort has been the IEEE (Institute of Electrical and Electronics Engineers), and the IEEE has guided the rocess for seismic qualification of equiment for nuclear ower generating stations for many years, IEEE [9]. A more recent IEEE undertaking was directed at establishing recommended ractice for seismic design of ower distribution substations, IEEE [10]. The telecommunications industry develoed its own seismic qualification rocedure, which was strongly influenced by work erformed at Bell Laboratories, Bellcore [11]. However, none of these existing rocedures relate secifically to the seismic demand requirements for secondary systems and nonstructural comonents as defined in the US model building codes. Because of the broad definition of nonstructural comonents covered under building code rovisions, the aforementioned test rocedures do not adequately address these more generic nonstructural seismic requirements. The model building codes indicate that hysical test data (e.g., shake-table testing) can rovide basis for the earthquake-resistant design of a articular tye of nonstructural comonent, but do not offer any guidance on how to erform testing and more imortant on how to interret seismic code requirements. Simly stated, building code authorities recognize that shake-table testing rovides design assurance verification of meeting code seismic requirements, but has conceded to not define a uniform test standard to accomlish it. The International Code Council (ICC) organization offers an Evaluation Service grou (ICC ES) that is resonsible for erforming technical evaluations and interreting UBC and IBC code rovisions. The evaluation rocess culminates with the issuance of reorts or listings on code comliance to show that roduct and systems, which ass established hysical testing criteria, meet code requirements. This was the rocess that was followed and successfully comleted by Square D / Schneider Electric for establishing a generic accetance criterion for seismic qualification testing of nonstructural building comonents, ICBO [12]. The develoment of this nationally acceted test rocedure (as described herein) is a first attemt to meet an industry need for erforming shake-table testing of nonstructural comonents. The authors view this rocedure as an initial line in the sand until future seismic research efforts can lay the necessary foundation for more recise test methods and rocedures. Test Secimen Rationalization Criteria Testing every single configuration of a given equiment roduct line may not be feasible, nor economically justified. Therefore, it may be necessary to select test secimens that adequately reresent the entire equiment roduct line. The following general criteria are offered as an illustrative examle for establishing test secimen (unit under test or UUT) configuration requirements for reresenting an equiment roduct line. It is recognized that industry secific roduct tyes will likely offer unique rationalization challenges that may deviate from the general rules rovided here: Structural Features: A rationale shall be rovided exlaining that the selected UUT s structural configuration is one offering the least seismic withstand caacity comared to other otions that are available within the roduct line being qualified. The UUT s force-resisting systems shall be similar to the major structural configurations being sulied in the roduct line. If more than one major structure is a configurable otion, then these other structural configurations shall be considered in the equiment roduct line extraolation and interolation rationalization rocess. Mounting Features: A rationale shall be rovided that exlains that the selected UUT s mounting configuration is one offering the least seismic withstand caacity comared to other mounting otions that are available within the roduct line being qualified. The configuration mounting of the UUT to

6 the shake-table shall simulate mounting conditions for the roduct line. It would be imractical and uneconomically justified to test every ossible anchorage system available in the marketlace (wedge, undercut, sleeve, shell, adhesive and various cast-in-lace tyes). Thus seismic testing of equiment is tyically conducted using the smallest diameter tie-down bolt size (or minimum weld size) that can be accommodated with the rovided tie-down clearance holes (or base structural members) on the equiment. If several mounting configurations are used, they shall be simulated in the test. Subassemblies: A rationale shall be rovided exlaining that the selected UUT s subassemblies are reresentative of roduction hardware and offer the least seismic withstand caacity of the UUT comared to other subassembly otions that are available within the roduct line being qualified. The major subassembly comonents shall be included in the UUT. These comonents shall be mounted to the secimen structure at locations similar to those secified for roosed installations. The comonents shall be mounted to the structure using the same tye of mounting hardware secified for roosed installations. Substitution of nonhazardous materials and fluids is ermitted for verification of equiment or subassemblies that contain hazardous materials or fluids, rovided the substitution does not reduce the functional demand on the equiment or subassembly. Mass Distribution: A rationale shall be rovided exlaining that the selected UUT s mass distribution is one contributing to the least seismic caacity of the UUT comared to other mass distribution otions that are available within the roduct line being qualified. The weight and mass distribution shall be similar to the tyical weight and mass distribution of the equiment being reresented. Weights equal to or heavier than the tyical weight shall be accetable. Equiment Variations: A rationale shall be rovided exlaining that the selected UUT s overall variations contribute to the least seismic withstand caacity of the UUT comared to other variations that are available within the roduct line being qualified. Other equiment variations, such as number of units/comonents in roduction assemblies, indoor and outdoor alications, etc., shall be considered in the equiment roduct line extraolation and interolation rationalization rocess. Imlementation of the roduct line selection criteria requires a configuration feature/otion matrix that mas available structural design features/otions against seismic withstand erformance. This can be accomlished by characterizing the structural dynamic effect offered from each available configuration feature/otion using analytical and/or emirical methods (finite element analysis and/or modal hammer survey). This structural demand matrix can also be used to determine test requirements for new roduct develoment activities by ranking seismic withstand erformance against reviously qualified configurations. Model Building Code Provisions The generic accetance criterion for seismic qualification testing of nonstructural comonents, ICBO [12], was written to satisfy the requirements as secified in the 1997 and 2000 NEHRP rovisions. The NEHRP rovisions form the source document that most model building codes and standards have adoted, including the UBC, IBC, ASCE 7, NFPA 5000, and others. It should be noted that general commentary regarding the suitability of current code requirements for seismic qualification is beyond the scoe of this aer. Section 6 Architectural, Mechanical, and Electrical Comonents Design Requirements of the 1997 and 2000 NEHRP secifies the seismic design requirements for secondary systems and nonstructural building comonents. The lateral force requirement is defined by:

7 where: F a S DS R I z h W F 0.4 a = DS 1 2 R I S + z h W, (1a) Seismic design force centered at the comonent s center of gravity and distributed relative to comonent s mass distribution Comonent amlification factor Design earthquake sectral resonse acceleration at short eriod Comonent resonse modification factor Comonent imortance factor Height in structure at oint of attachment of comonent Average roof height of structure relative to the base elevation Comonent oerating weight. Additional stiulations are defined for allowable maximum and minimum values for F : Maximum F = 1. 6 S I W (1b) DS DS Minimum F = 0. 3 S I W (1c) Close examination of equation 1a is needed to convert the lateral force requirements into shake-table test arameters. The height factor ratio (z / h) accounts for above grade-level equiment installations within the rimary suorting structure and ranges from zero at grade-level to one at roof-level, essentially acting as a force increase factor to recognize building amlification as you move u within the rimary structure. The comonent amlification factor, a, also acts as a force increase factor by accounting for robable amlification of resonse associated with the inherent flexibility of the nonstructural comonent, ranging from one for rigid comonents (no amlification) to two-and-a-half for flexible comonents (maximum amlification). The site-secific ground sectral acceleration factor, S DS, varies er geograhic location (taken from the design value mas) and site soil conditions. The S DS factor is used to define the general design earthquake resonse sectrum curve and is used to determine the design seismic forces for the rimary building structure. Figure 1 shows the building design sectrum at the ground s surface contained in Section General Procedure Resonse Sectrum of the 1997 and 2000 NEHRP. The S DS factor ranges from relatively low values (0.18 g s in Florida) to high values (2.46 g s in New Madrid hazard area) deending on the seismicity of a given location. The ratio of R over I (R / I ) is considered to be a design reduction factor to account for inelastic resonse and reresents the allowable inelastic energy absortion caacity of the equiment s forceresisting system. The comonent imortance factor, I, reresents the greater of the life-safety imortance of the comonent and the hazard exosure imortance of the structure. The I factor is used to reduce the ermitted inelastic resonse. In other words, the higher the erformance desired, the lesser the ermitted inelastic resonse. The I factor is also used to indicate that verification of comonent oerational functionality at design earthquake levels is required. Values for the (R / I ) ratio range from 0.67 to 2.33 (while fixing I = 1.5). Hence, equation 1a defines the lateral force design requirement imosed on any nonstructural comonent for any given equiment location within a building structure and for any given building location in the United States.

8 Figure 1. Figure Design Resonse Sectrum, (figure rerinted from BSSC [8]). Derivation of Seismic Qualification RRS The goal of deriving the nonstructural seismic qualification RRS is to translate the building codes requirements for nonstructural comonents into the necessary arameters comatible with shake-table testing. Tyical resonse sectrum curves lotted using a logarithmic scale ram u, remain flat, and then ram down as illustrated in Figure 2. Correct determination of the A-B-C-D sectrum break oints is the most critical ste in relating lateral force code requirements to shake-table testing arameters. The shake-table RRS has its ordinate defined in terms of sectral resonse acceleration units (g s) and the abscissa in frequency units (Hz). It should be noted that for most mechanical and electrical equiment testing, it is common ractice to use frequency units instead of eriod units. Throughout the remaining discussion in this aer natural frequency and eriod terms will be interchangeable with the understanding that frequency is simly the inverse of eriod. Figure 2. Tyical shake-table required resonse sectrum curve lotted on logarithmic scale. RRS Acceleration Values In order to determine the nonstructural comonent s acceleration at oint of attachment to the rimary structure, the comonent weight is factored out of equation 1a resulting in the following formula for the NEHRP nonstructural comonent design lateral acceleration requirement:

9 A = 0.4 a R I S DS z (2) h As stated reviously, the ratio (R / I ) reresents the allowable inelastic energy absortion caacity of the equiment s force-resisting system and is considered a design reduction factor. During the seismic simulation shake-table test the UUT comonent will resond to the excitation and inelastic behavior will naturally occur. Therefore the ratio (R / I ) was set equal to one, which is indicative of an unreduced resonse. The authors have interreted equation 2 to mean that the imortance factor, I, does not increase the seismic test inut motion but does effect the requirement for the UUT comonent to demonstrate oerational functionality following seismic simulation testing and is used in AC156 to determine the ost test UUT requirements for functionality comliance. The key oint here is that the authors consider the shake-table inut motions indeendent of the (R / I ) ratio. For examle, if a nonstructural comonent is ground suorted (grade level installation), then the AC156 shake table inut motion should equal the ground motions defined by code rovisions for the base of a building. The ground motions are not increased by 1.5 because of the I designation for the comonent. However, what is exected with an I of 1.5 is essentially elastic resonse with verification of equiment functional erformance immediately following the earthquake event. Conversely, with an imortance factor set equal to one the exectation is inelastic resonse but overall structural stability and no exectation of comonent functionality. In addition, the authors do not consider that the ground motions should be reduced by the comonent s resonse modification factor, R. The comonent amlification factor, a, was taken from the formal definition of flexible and rigid comonents, BSSC [1] and BSSC [2]. By definition, for fundamental frequencies less than 16.7 Hz the nonstructural comonent is considered flexible (maximum a value), which corresonds to the amlified ortion of the RRS (segment B-C of Figure 2). For fundamental frequencies greater than 16.7 Hz the comonent is considered rigid (minimum a value), which corresonds to the zero eriod acceleration (oint D of Figure 2). Hence, the RRS is constructed using two normalizing acceleration factors. The flexible acceleration factor (values for break oints B and C of Figure 2) is defined by substituting (R / I ) equal 1 and a equal 2.5 into equation 2, resulting in: A z = SDS 1 +, (3a) h FLX 2 and the rigid acceleration factor (value for break oint D of Figure 2) is defined by substituting (R / I ) equal one and a equal one into equation 2, resulting in: A z = 0. 4 SDS 1 +, (3b) h RIG 2 where A FLX is limited to a maximum value of 1.6 times S DS, in accordance with equation 1b. It is noted again that the I factor is not used to increase the maximum value of the shake-table inut motion for reasons discussed earlier.

10 RRS Frequency Values Sectral shae of the AC156 seismic qualification RRS follows the general shae deicted in Figure 2 and is constructed using the acceleration variables A FLX and A RIG (equations 3a and 3b) and the frequency variables (f 0, f FLX and f RIG ) to define the RRS break oints as shown in Figure 3. Figure 3. AC156 required resonse sectrum normalized for nonstructural comonents. The frequency variables are directly related to the general design earthquake resonse sectrum curve used for design of the rimary building structure as shown in Figure 1. The f 0 value corresonds to the inverse of the design resonse sectrum transition eriod T 0, from increasing acceleration to constant acceleration. Therefore, f 0 is defined as: f S DS 0 = = T 0 2 T = (4a). S 0 When soil roerties are not known, Section of the 1997 and 2000 NEHRP (Site Class Definitions) ermits Site Class D to be assumed. In areas of high seismicity the ratio of S D1 / S DS on Site Class B (soft rock) is aroximately 0.4 and the ratio of S D1 / S DS on Site Class D (moderately stiff soil) is aroximately 0.6. Using the default otions and substituting into equation 4a we obtain: ( 0. 6) S D1 5 f = 8. 3 Hz (4b) 0 = Also in the Figure 3 AC156 sectrum, the frequency f FLX corresonds aroximately to the inverse of the design resonse sectrum transition eriod T S, from constant acceleration to constant velocity. It should be noted at eriods greater than T S (where T S = S D1 / S DS ), the acceleration resonse of rimary structures starts to reduce because the design ground motion acceleration resonse sectra beyond T S reduces by the ratio of T S / T (where T is the fundamental eriod of the rimary structure). Since this reduction in the design forces for the rimary structure is accounted for in the design force equations, it is justifiable to make a similar tye of reduction in the design forces and in-structure test motion sectra of non-structural comonents. However, in observing the actual in-structure resonse sectra of acceleration recordings measured at the roof levels of buildings with a range of fundamental eriods, this reduction in resonse tyically begins at eriods about 25% greater than T S. Therefore, the authors have assumed the transition

11 eriod, T FLX (1 / f FLX ), at the to of the rimary structure at which test motion in-structure sectra begin reducing by a ratio of T S / T has been lengthened by 25% to account for this observation. At the ground level, the test motion in-structure sectra transition eriod matches the ground sectra. At relative elevations between the roof and the ground, linear interolation is used to determine the transition eriod T FLX (1 / f FLX ). Therefore, f FLX is defined as: f FLX = 1 TS z h = S S DS D1. z h (5a) Assuming a Site Class D default and substituting into equation 5a for the ratio of S D1 / S DS, one obtains at the roof of the structure (where z / h = 1.0) the following for f FLX : 1 1 f = = = 1. 3 Hz (5b) FLX ( 0. 6)( ) ( 0. 6)( 1. 25) The f RIG value is defined as the frequency associated with the zero eriod acceleration (break oint D). NEHRP rovisions define rigid comonents as having a fundamental frequency greater than 16.7 Hz. A value was selected for f RIG to be at least one full octave greater than 16.7 Hz. Thus, f RIG was set equal to 33.3 Hz. This cutoff frequency value is also consistent with other industry acceted test rotocols, IEEE [9], IEEE [10], and Bellcore [11]. The values of f 0 equal to 8.3 Hz and f FLX equal to 1.3 Hz and f RIG equal to 33.3 Hz are currently included directly in AC156. Two additional oints should be noted about the Figure 3 AC156 seismic qualification RRS. First, between f 0 (8.3 Hz) and f RIG (33.3 Hz) there is a raming down of the sectra rather than a ste function at 16.7 Hz (ste function shown as dashed lines). The ste function found in the code rovisions is actually a simlification of the raming ortion of the design resonse sectrum (see Figure 1 between T = 0 and T 0 ). Second, the sectra are currently only valid for nonstructural building comonents with natural fundamental frequencies greater than f FLX (1.3 Hz). For nonstructural comonents with fundamental frequencies less than f FLX (1.3 Hz), site secific dynamic analyses are currently needed to develo suitable equiment qualification RRS and cannot be tested using AC156. Vertical Requirements Nonstructural building comonent earthquake effects are determined for combined horizontal and vertical loading. The required resonse sectrum for the horizontal direction is develoed based on the normalized equiment RRS and the formula for total design horizontal force, F, as described above. The required resonse sectrum for the vertical direction is based on a simle assumtion of two-thirds the ground level base acceleration, which is a common ractice for defining vertical ground motion. It should be noted while exlicit requirements for vertical ground motion are not currently secified in NEHRP rovisions for building structures, it was considered rudent to include a vertical comonent in the shaketable testing rotocol for nonstructural comonents. It is recommended that future research efforts be directed to better define the vertical nonstructural seismic requirements. The two-thirds assumtion uses the same frequency break oints as defined for the horizontal RRS. This assumtion may not accurately reflect vertical exerience data as noted in Paazoglou [13]. Both horizontal and vertical RRS are defined using a daming value equal to 5 ercent of critical daming.

12 AC156 RRS Parameters When Building Characteristics are Known The above formulation for develoment of a generic test sectra is based on not knowing anything about the building in which the nonstructural comonent is attached (other than the installation height). However, there are many occasions in which the building dynamic characteristics are known, and in those instances the horizontal force requirements may alternatively be defined utilizing the results of a dynamic analysis rocedure for the seismic design of a secific building er Section 5.5 of 2000 NEHRP. The ground motion inut for the dynamic analysis shall be consistent with resonse sectra constructed in accordance with Section 4.1, using R = 1.0. A result of the dynamic analysis rocedure yields the following AC156 arameters for A FLX and A RIG : and where: A x a i A = 2. 5 A a (7a) FLX RIG x x i i A = 1. 0 A a (7b) Torsional amlification factor er Section of the 2000 NEHRP Acceleration at level i obtained from the dynamic analysis rocedure. Develoment of the Seismic Simulation Waveform To meet the AC156 required resonse sectra as defined above, the corresonding shake-table drive signals are derived using non-stationary broadband random excitations with energy content ranging from 1.3 to 33.3 Hz. The broadband shae of the AC156 RRS results from the generic aroach that code rovisions have taken regarding nonstructural requirements (e.g., any floor location within a building and any building location within the U.S.). Thus, the drive signal comosition is multile-frequency random excitations, the amlitudes of which are adjusted either manually or automatically based on a maximum one-third octave bandwidth resolution. The duration of the inut motion is 30 seconds, with the nonstationary character being synthesized by a linear inut signal build-hold-decay enveloe of seconds, resectively. Indeendent random signals that result from an aggregate of the maximum-onethird octave narrowband signals are used as the excitation to roduce hase incoherent motions in the two rincial horizontal and vertical axes of the shake-table. The AC156 requirement is to test in three orthogonal axes using either tri-axial, bi-axial, or uni-axial shake-tables. The use of bi-axial and uni-axial tables requires multile stages with UUT rotation between stages such that all three axes have been tested. In its resent form, AC156 defines the shake-table requirement in terms of the normalized RRS as described above. Develoment of a suitable time history drive signal that satisfies AC156 sectral requirements is left to the test laboratory to accomlish rior to erforming shake-table testing. The authors recognize that random multi-frequency time history waveforms may not resemble earthquake exerience data, however the very nature of a broadband sectrum, necessitates the use of multi-frequency random inut. Future research efforts are needed to better correlate multi-frequency random inut with exerience data to create hybrid waveforms that are comatible with broadband sectrum testing. The test resonse sectrum (TRS) is comuted using either justifiable analytical techniques or resonse sectrum analysis equiment using control accelerometers located at the UUT base. The TRS is calculated using a daming value equal to 5 ercent of critical daming. The TRS must envelo the RRS based on the maximum-one-third octave bandwidth resolution over the frequency range from 1.3 to 33.3 Hz, or u to the limits of the shake-table control system. The amlitude of each maximum-one-third octave bandwidth is indeendently adjusted in each of the rincial axes until the TRS envelos the RRS within the limitations of the test machine. The TRS should not exceed the RRS by more than 30 ercent over the amlified frequency range. It is recommended that the TRS be comuted with a maximum of

13 one-sixth octave bandwidth resolution. Provisions are defined in AC156 to allow the TRS to di below the RRS uon satisfying secific allowances. Pass/Fail Accetance Criterion The obvious goal of erforming seismic qualification testing is to demonstrate that the seismic caacity of a articular nonstructural comonent exceeds the seismic demand. In the ast, determination of a successful test outcome has been difficult and almost entirely deends on subjective interretation by the test laboratory. AC156 has addressed this issue by using the comonent imortance factor to define minimum requirements for a assing test. The nonstructural comonent imortance factor, I, reresents the greater of the life-safety imortance of the comonent and the hazard exosure imortance of the structure, BSSC [1] and BSSC [2]. In AC156 this factor directly accounts for the functionality of the nonstructural building comonent under test. Essentially, there are two criteria used to determine test accetance, one when I equals 1.0, and another when I equals 1.5: I = 1.0 Equiment design must ensure that the anchored UUT will not leave its mounting and cause damage to other building comonents or injury to ersonnel during the seismic event. Structural integrity of the equiment attachment system shall be maintained. At the comletion of the seismic simulation testing, the UUT does not ose a life safety hazard due to collase or major subassemblies becoming searated, and materials deemed to be hazardous have not been released into the environment in quantities greater than the exemted amounts listed in the code. Structural damage, such as local yielding, to UUT force-resisting members is accetable and structural members and joints comrising the UUT forceresisting system shall be allowed fractures and anomalies. Verification of ost-test functional erformance is not required. I = 1.5 The equiment is deemed to be essential to the continued oeration of a facility, and/or essential to maintaining critical life suort systems, and/or contain materials deemed to be hazardous, to humans or the environment, in quantities greater than the exemted amounts listed in the code. Equiment design must ensure that the anchored UUT will not leave its mounting and cause damage to other building comonents or injury to ersonnel during the seismic event. Structural integrity of the equiment attachment system shall be maintained. After comletion of the seismic simulation testing, the UUT shall satisfy all of the functional and oerational tests secified by the manufacturer, and materials deemed to be hazardous shall not have been released into the environment in quantities greater than the exemted amounts listed in the code. In addition, at the comletion of the seismic simulation testing, the UUT does not ose a life safety hazard due to collase or major subassemblies becoming searated. Structural damage, such as local yielding, to UUT force-resisting members is accetable and structural members and joints comrising the UUT force-resisting system shall be allowed fractures and anomalies. Verification of ost-test equiment functionality is required. CONCLUSIONS Manufacturers of nonstructural equiment using various different interretations of model building code requirements have ursued seismic qualification testing of nonstructural comonents for many years. Shake-table testing is the referred industry aroach for qualifying nonstructural equiment to meet the requirements contained in model building codes. However, building code rovisions do not define nor offer any guidance on how to correctly translate static lateral force requirements into shake-table testing RRS arameters. This situation results in multile code interretations and ultimately creates the occurrence of manufacturers claiming seismic qualification against building code requirements that were tested using vastly different shake-table RRS test levels.

14 Resolution of this inconsistency in code interretation regarding qualification testing is addressed by generation of a generic test rocedure that has been nationally endorsed by the ICC Evaluation Service organization. This test rocedure document, ICBO [12], can be used to validate seismic withstand caacity for any nonstructural building comonent as defined by any model building code or standard that adots the NEHRP rovisions as the rimary source document (which includes 2000 IBC, 1997 UBC, 2002 ASCE-7, 2003 NFPA 5000 and others). The develoment of the seismic qualification RRS is based uon the existing nonstructural lateral force rocedure in conjunction with the building design resonse sectrum. This aroach accounts for above grade level equiment installations, with or without knowledge of the building s dynamic characteristics. A well-defined ass/fail accetance criterion is established that utilizes the equiment imortance factor to define ost-test accetability. In essence, this generic test rotocol establishes the seismic qualification shake-table test RRS for any nonstructural comonent for any given equiment location in a building and for any given building location in the United States. This rocedure is the first recognized qualification test rotocol to address secondary system and nonstructural comonent requirements as defined by the rimary model building codes being emloyed in the United States. The authors view this rocedure as an initial attemt to define a ractical seismic qualification standard that can be imlemented in industry. It is recognized that future research efforts will be needed to fine tune this standard in the areas of vertical requirements, time history waveforms, and consideration for statistical aroaches to nonstructural qualification using robability of failure measures. However, the rimary goal of this aer is to hel establish a standardized aroach to seismic qualification testing such that subjective interretations and non-uniform test methods can be eliminated, thus unambiguously satisfying the intent of model building code rovisions for secondary systems and nonstructural building comonents that have fundamental frequencies greater than 1.3 Hz. The final message to be communicated is about adoting the hilosohy of seismic risk mitigation. General awareness of what is needed to lower the risk of exeriencing nonstructural comonent failures following a seismic event is the first ste. However, mitigating seismic risk and achieving the lowest ossible risk associated with seismic erformance of nonstructural comonents deends on the concet of shared resonsibility between equiment oerators, installers, and manufacturers. None of these arties on their own can lower the total risk, but when each arty accets resonsibility for their ortion, the combined risk ownershi will ensure that the lowest ossible risk is achieved. This concet is aramount when maximizing the robability of nonstructural comonent survival and functionality following an earthquake is a riority. REFERENCES 1. BSSC, 1997 Edition: NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, Part 2 Commentary, Reort No. FEMA 303, Federal Emergency Management Agency, Washington, D.C., BSSC, 2000 Edition: NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, Part 2 Commentary, Reort No. FEMA 369, Federal Emergency Management Agency, Washington, D.C., Filiatrault, A., and Christooulos, C., ''Guidelines, Secifications, and Seismic Performance Characterization of Nonstructural Building Comonents and Equiment,'' Reort PEER 2002/05, Pacific Earthquake Engineering Research Center, Berkeley, CA, 102., Nonstructural issues of seismic design and construction, Earthquake Engineering Research Institute, Berkeley, California, Publication number 84-04, Naeim, Farzad, The Seismic Design Handbook, 2nd edition, Kluwer Academic Publishers, Boston, 830, 2001.

15 6. Gillengerten, J. D., and Bachman, R. E., Background on the Develoment of the NEHRP Seismic Provisions for Non-Structural Comonents, in Proceedings of the 2003 ASCE/SEI Structures Congress and Exosition, Seattle, Washington, BSSC, 1997 Edition: NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, Part 1 Provisions, Reort No. FEMA-302, Federal Emergency Management Agency, Washington, D.C., BSSC, 2000 Edition: NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, Part 1 Provisions, Reort No. FEMA-368, Federal Emergency Management Agency, Washington, D.C., IEEE Standard , IEEE Recommended Practice for Seismic Qualification of Class 1E Equiment for Nuclear Power Generating Stations, The Institute of Electrical and Electronics Engineers, Inc., IEEE Standard , IEEE Recommended Practice for Seismic Design of Substations, The Institute of Electrical and Electronics Engineers, Inc., Bellcore GR-63-CORE, Bellcore Network Equiment-Building System (NEBS) Requirements: Physical Protection, ICBO Evaluation Service, Inc., Accetance Criteria for Seismic Qualification Testing of Nonstructural Comonents (AC156), Paazoglou, A. J., and Elnashai, A. S., Analytical and Field Evidence of the Damaging Effect of Vertical Earthquake Ground Motion, Earthquake Engineering and Structural Dynamics, Vol. 25, , 1996.