You are on page 1of 12
(61)US NCEE Conference on Earthquake Eaginering/EERI May 31 June 4,199, Seat, Washington Paper #269 DEVELOPMENT AND USE OF CAPACITY SPECTRUM METHOD eee Sigmund A. Freeman! ABSTRACT The Capacity Spectrum Method (CSM) was first introduced in the 1970's as a rapid evaluation procedure in a pilot project for seismic vulnerability of buildings at the Puget Sound Naval Shipyard. In the 1980's it was used as a procedure to find correlation between earthquake ground motion and building performance (ATC-10). The method ‘was also developed into a design verification procedure for the TriServices' (Army, Navy, Air Force) "Seismic Design Guidelines for Essential Buildings" manual. The procedure compares the capacity of the structure (in the form of a pushover curve) with the demands on the structure (in the form of a response spectrum). The graphical intersection of the two curves approximates the response of the structure. In order to account for nonlinear inelastic behavior of the structural system, effective damping values are applied to the elastic-linear response spectrum to imitate an inelastic response spectrum. The relationship between damped linear elastic response spectra (LERS) and inelastic response spectra (IRS) are presented such that elastic damping ratios (8) can be equated to ductility ratios (1) for various characteristics of hysteretic behavior. Recently, the CSM has been modified for use in the State of California Proposition 122 guidelines for evaluating and seismic upgrading reinforced concrete buildings (ATC-40), Three variations for estimating inelastic response by the CSM are proposed by the ATC- 40 document. The paper will present a fourth option that clarifies and simplifies the process. ‘The CSM can be used for a variety of uses such as a rapid evaluation technique for a large inventory of buildings, a design verification procedure for new construction of individual buildings, an evaluation procedure for an existing structure to identify damage states, and a procedure to correlate damage states of buildings to amplitudes of ground motion. ‘Principal, Wiss, Janney, Elstner Associates, Inc., 2200 Powell Street, Suite 925, Emeryville, CA 94608. INTRODUCTION ‘The Capacity Spectrum Method (CSM) presents an illustrative format that is applicable to any of the procedures that are being presented as performance-based engineering (PBE). These other procedures have names such as multilevel design, equal displacement method, direct displacement design, capacity-based design, secant method, inelastic response spectra, and substitute structure, It can also be simply applied as an evaluation for static code procedures. ‘The CSM, which started as rapid evaluation procedure (Freeman, et al. 1975), with its roots in John Blume's reserve energy technique (Blume, 1961) has gone through several periods of development (Freeman, 1978; ATC, 1982; Army, 1986; Freeman, 1987; Army, 1988; ATC, 1991; Freeman, 1992; Mahaney, et al. 1993; NIST, 1994; WIE, 1996; ATC, 1996) until it has reached its current stage. During the past 30 years, researchers and engineers have been trying to develop procedures for the seismic design of structures that would represent their inelastic response to earthquake ground motion. These efforts were made in recognition that the static code provisions (e.g., SEAOC, UBC, ‘NEHRP) were limited to a reasonable representation of linear-elastic response, but were dependent on special detailing requirements to provide a somewhat unknown degree of resistance to strong motion inelastic response. The static seismic design provisions generally provided a good basis for designing structures to resist earthquake loading; but it required a well thought out concept that was implemented by engineers that understood the limitations of the code provisions and used their experience and knowledge to provide designs that would perform well when subjected to major earthquakes. In the quest to develop inelastic design procedures, researchers and engineers were generally using the same principles of dynamics, mechanics, and materials, and similar, if not the same, experimental data, The primary differences were the thought processes and different paths for solving the same problem. As we review the various PBE methods we can identify the similarities and see that the differences are in initial assumptions and the format used to reach solutions. All valid methodologies should give similar results. Differences will reflect variations of assumptions such as those related to properties and behavior of the materials and characteristics of earthquake motion. ‘The interesting part about comparing the various procedures is that they tend to verify each other and point out potential weaknesses in each of them. BACKGROUND The Capacity Spectrum Method (CSM) was originally developed as a rapid evaluation method for a pilot seismic risk project of the Puget Sound Naval Shipyard for the U.S. Navy (Freeman, et al, 1975). It was later used as a procedure to correlate earthquake ground motion with observed building performance (Freeman, 1978 and ATC, 1982) and was further developed for the TriServices "Seismic Design Guidelines for Essential Buildings" (Army, 1986) as part of the two-level approach to seismic design. The TriServices version was reformatted for the proposed U.S. Postal Service seismic evaluation program (ATC, 1991) and is included in Appendix B of the "Standards of Seismic Safety for Existing Federally Owned or Leased Buildings" (NIST, 1994). The CSM was recently updated for a proposed revised edition of the TriServices manual on dynamic analysis (WIE, 1996) and modified for the State of California Proposition 122 guidelines for evaluating and seismic upgrading reinforced concrete buildings (ATC, 1996). CAPACITY SPECTRUM METHOD ‘The Capacity Spectrum Method can be used to approximate the performance level that a building will be subjected to during an earthquake, The method can be used at various levels of accuracy ranging from that required for rapid evaluation of a large group of existing buildings to that required for a detailed structural evaluation of the design of an essential building to verify that the code designed structural system will satisfy specified performance goals for major postulated earthquakes. The concept is rather simple, The lateral force resisting capacity of a building is represented by a force- displacement curve such as obtained by a pushover analysis. The demands of earthquakes are represented by response spectra curves. When these two curves are plotted on the same set of coordinates, the relationship between demand and capacity are readily apparent. If the capacity curve can break through the demand envelope, the building survives the earthquake. The intersection of the two curves approximates the response and performance levels of the structure for that earthquake. Once a capacity curve is developed for a structure, it can be quickly compared to a variety of earthquake response spectra. The capacity curve can range from very rough and approximate to reasonably accurate and detailed. The demand curves can range from simple idealized smooth design spectra to spiked response spectra obtained from recorded earthquake ground motion. The demand earthquake is displayed as a family of damped response spectra to represent various ductility demands, For example, 5% damping will represent a linear elastic response and 20% damping might represent a response at a displacement ductility in the neighborhood of three. Capacity Curve—Pushover The capacity curve is determined by statically loading the structure with realistic gravity loads combined with a set of lateral forces to calculate the roof displacement a, and base shear coefficient (Cy = VW) that defines first significant yielding of structural elements. The yielding elements are then relaxed to form plastic hinges and incremental lateral loading is applied until a nonlinear static capacity curve is created, The curve is created by superposition of each increment of displacement and includes tracking displacements at each story (ATC, 1982). This procedure is sometimes referred to as the pushover analysis, There are several levels of sophistication that may be used for the pushover analysis, ranging from applying lateral forces to each story in proportion to the standard code procedure to applying lateral story forces as masses times acceleration in proportion to the first mode shape of the elastic model of the structure, For added sophistication, at each increment beyond yielding, the forces may be adjusted to be consistent with the changing deflected shape, For tall buildings the effects of the higher modes of vibration may be considered (Paret, et al., 1996). It is assumed that the structure can take a number of cycles along the capacity curve and behave in a hysteretic manner. The stiffness is assumed to reduce to an equivalent global secant d ‘modulus measured to the maximum excursion along the capacity curve for each cycle of motion. The ‘Ag vs V/W coordinates (Figure 1) are converted to spectral displacements (S,) and spectral accelerations (S,), respectively (Table 1 and Figure 2), by use of modal participation factors (PF,,,) and effective modal weight ratios (a,) as determined from dynamic characteristics of the fundamental mode of the structure (WIE, 1996 and ATC, 1996). These values change as the displaced shape changes. An equivalent inelastic period of vibration (TI) at various points along the capacity curve are calculated by use of the secant modulus (ie., T, = 2n/S,/S,; g). Now the capacity spectrum curve can be plotted with the same coordinates as a response spectrum. Table 1, Conversion of V and A, to S, and S, v Ay s, Se T Point | (kips) | (in) | VW | PFid, | @ Gin) | Gee) A_| 2200 | 251 | 0209 | 131 | 0828 | 0254 | 192 | 088 B 2600 3.60 0.247 1.28 0.800 0.309 2.81 0.96 c 2800 | 5.10 | 0266 | 135 | 0.770 | 0346 | 378 | 1.06 D 3000 | 1090 | 0285 | 139 | 0.750 | 0380 | 784 | 145 Note: 5, = VW= a; and S, = Ay/PF,$,. PFs and a’s change because the mode shape is changing as yielding occurs. 4000 s Re 2000 Base Shear, V (kips) 5 10 15 Roof Displacement, (x (inches) Figure 1. Capacity curve (refer to Table 1 for values of Pt. A to Pt. D.) Spectral Acceleration, Sa (g) Spectral Displacement, Sd (inches) Figure 2. Capacity spectrum curve. (Points on curve correspond to Pt. A - Pt. D in Figure 1. Values of S, and S, in Table 1.) It should be noted that the capacity curve need not be exact in order to be usefull. A reasonable approximation of the elastic limit and the inelastic limit will give a general idea of how the building will respond to various earthquake demands. This was the basis of the original rapid evaluation procedure. As the pushover analyses become more detailed, it is useful to denote yielding, and cracking benchmarks along the capacity curve. The pushover should be continued to the largest displacement practicable until degradation of the overall system occurs or limits of structural stability occur, In cases where a target displacement is set as a goal, itis generally worthwhile to push a little further to establish a better confidence level. ‘After the capacity curve has been plotted, itis useful to approximate an equivalent bilinear capacity representation that establishes an effective yield point (y=1) and an effective peak inelastic limit (S4=7.8 inches). Points of displacement ductility ratios (11) can be marked along the post-elastic that willbe useful in a later phase of the CSM. This process is shown in Figure 3 for Type C, r 10 behavior (referred to later in Table 3). The six points on Figure 3 represent the following: Ba «5 ul 14 1S 1616 it 1 15° 20-25 0 3.03.7 s, 0) a2). G31) 35 3738 Sa 23° 35 46 S58 69 78 Spectral Acceleration, Sa (g) 2 4 6 ee Ole ia Spectral Displacement, Sd (inches) Figure 3. Bilinear idealized capacity spectrum. (Constructed from Fig. 2 for use in Figs. 5 and 6.) Demand Curves - Response Spectra ‘The demand curve is represented by earthquake response spectra. It is presented at various levels, of damping. For example, the 5 percent damped response spectrum is generally used to represent the demand when the structure is responding linearly-elastic (LERS). Higher damped response spectra are used to represent inelastic response spectra (IRS) to account for hysteretic nonlinear response of the structure. Response spectra have traditionally been plotted with S, vs T coordinates or tripartite log coordinates. In order to more visually illustrate the relationship between accelerations and displacements the S, vs T coordinate system for the response spectra are converted to a set of coordinates defined by S, and S, (i.e., S, = S,g (T/2)?). When the spectral values are plotted in this acceleration-displacement response spectrum format (ADRS), the period can be represented by lines radiating from the origin (Mahaney, et al., 1993). An example of demand spectra is shown in Figure 4. Spectral Acceleration, Sa (g) 0 2 4 6 8 10 12 Spectral Displacement, Sd (inches) Figure 4. Family of demand response spectra in ADRS format for Bye = 5%, 8%, 12%, 18%, 26%, and 40%, seismic zone 4, soil site Sp. One controversial item of the CSM, as well as for other inelastic methods, is the relationship between inelastic response spectra (IRS) and equivalent linear elastic response spectra (LERS). In the example shown in Figure 4, the top curve represents the LERS 5 percent damped response spectrum (f = 5%). The damped curves for f = 8 percent to 40 percent are used to represent IRS. where an effective damping value (B,q) represents a displacement ductility demand (i.., » = d/4,). There is a fairly good consensus on the spectral reduction factors (SR) used to reduce 5-percent damping to the higher percentages of damping (refer to Table 2). These are based on the Newmark- Hall studies (Newmark and Hall, 1982). The one-sigma set of values are for application to design spectra based 84.-percent cumulative probability and median set of values are for 50-percent probability. The next step is to relate damping () to ductility (1). One method is to relate Newmark- Hall's equations for B andy (Newmark and Hall, 1982) where the elastic response is reduced by 1//2u-1 in the constant acceleration portion of the response spectrum and by 1/1 in the constant velocity portion. This procedure was used for the proposed update of the TriServices manual (WJE, 1996), Another method is based on idealized hysteretic parallelogram loops (Type A behavior) and the derivation of energy dissipated by damping (ATC, 1996). To represent less idealized behavior (ie., more realistic behavior) values for Types B and C were also developed for ATC-40 (ATC, 1996). Variations were also developed for various slope ratios of the inelastic tangent stiffness to the clastic stiffness. Examples of Bye Vs 1 are shown in Table 3, including values for slope ratios (r) of 0 and 0.10. Also included are similar relationships obtained from Priestley's direct displacement- based design method (Priestley, et al., 1996). 2 3 Table 2. Spectral reduction factors: SR, for constant acceleration, short period, range and SRy for constant velocity, long period, range Ba | One-Sigma Median SR, | SRy | SR, | SRy 5 1.0 10 1.0 1.0 ile 80) 90] 91 8 10 72 0) |e 78) 83 Is| 58 70 64 73 20] 46 60 | 55 66 25 38 53 48 60 30]. 32 ATE a2 35 35 - 38 52. 40} ~ ~ 33 50 Table 3. Displacement ductility (1) vs. effective damping (B,.). Bur (percent of critical damping) WIE 1996 ATC 1996 (r= 0 and 0.10) From Fig. | TY of Pric " 1-Sigma | Median “ - r ci 88 a o | 10} o | 10 | o 10 10 5.0 5.0 5 5 5 s 5 tL bl 125 7S BS 18 16 13 12 9 9 - 150 | 10 12 | 24 | 23 | as | a7 Poa | on a 20 | 14 16 | 33 | 29 | 25 | 22 | 16 | 14 3 3.0 21 26 39 33 29 25 19 16 ae 4.0 26 35 40 34 29 25 20 16 19 6.0 40 33 29 25 20 16 21 8.0 40 31 29 24 20 = 22 Capacity and Demand ‘When both the capacity spectrum and the demand response spectrum are defined with the same set of coordinates, they can be plotted together. ‘The Capacity Spectrum Method can be summarized as follows: If the capacity curve can extend through the envelope of the demand curve, the building survives the earthquake. The intersection of the capacity and appropriately damped demand curve represents the inelastic response of the structure 2 To illustrate the Capacity Spectrum Method in the ADRS format the idealized capacity curve of Figure 3 is superimposed on the response spectra of Figure 4. This example is included in ATC-40 (ATC, 1996) showing three procedures, The procedure shown in Figure 5 is a simplified version of those shown in ATC-40. In Figure 3, effective damping (Bq) relating to displacement ductilities (1) are noted on the capacity curve (¢.8., Byq = 11% at = 1.5). The demand response spectra for B = 5%, 8%, 12%, and 18% from Figure 4 are also shown in Figure 5. Note that the elastic limit (ie., Sy 2.3 inches) of the capacity curve does not reach the 5 percent LERS curve; therefore, the elastic demand exceeds the elastic capacity and the structure will displace into the inelastic range. Also note that the inelastic capacity at 5.8 inches, (Byy = 15 percent) is less than the demands of the Bu = 15 percent demand curve (S, = 6.4 in. as interpolated between the 12% and 18% demand curves) and the capacity at 6.9 inches (Byy = 16%) is greater than the demands at the 16 percent demand curve (S, = 6.1 inches). This tells us that the response will be somewhere between S, = 6.1 and 6.4 inches. Thus, the common intersection can be estimated by averaging 5.8, 6. 6.1, and 6.4, which coincides with about S, = 6.3 inches at Bx = 15.5% damping. In other words, when the sample structure is subjected to the sample earthquake, S, = 6.3 inches, Bye = 154%, and S, = 0.36g. By referring to Table 1, this can be translated back to a roof displacement of 8.6 inches (A, = S, x PF,) and a base shear coefficient of 0.27 (V/W = @ §,). This description for Figure 5 may seem complex from the text, but visually it can be quickly estimated at 6 to 7 inches by observing that the solution is between demands of 12% and 18% damping and capacities of 15% and 16% damping. The important observation is that the inelastic 3) capacity limit at 7.8 inches is reasonable beyond the demand of 6.3 inches and that the displacement ductility demands are about three. 12 Spectral Acceleration, Sa (g) 0 2 4 6 8 10 12 Spectral Displacement, Sd (inches) Figure 5. CSM for idealized capacity (Fig. 3) and zone 4, soil S, demand (Fig. 4). J Demand Byr: 5%, 8%, 12%, 18. Capacity Bq: 5%, 11%, 14%, 15%, 16%. 9 Generally, the design response spectra are smooth in shape (Fig. 4), such as those in building codes (¢.g., UBC and NEHRP); however, response spectra derived from actual earthquake records are irregular and contain spikes at predominant response periods, These spikes tend to fade away at higher damping values. An example is shown in the ADRS format in Figure 6. In the CSM solution for the sample building, the structure just barely exceeds yield. The demand S, lies between 5% and 10% damping at about 2.5 inches. Spectral Acceleration, Sa (g) 0 2 4 6 8 10 12 Spectral Displacement, Sd (inches) Figure 6. CSM for idealized capacity (Fig. 3) and Loma Prieta Arguello Drive recorded motion. Response spectra for 0%, 5%, 10%, and 20% damping. COMPARISONS WITH OTHER METHODS The similarities between CSM and other approximate inelastic procedures become more apparent ‘when the other methods are described in the ADRS format. For the equal displacement method, visualize the elastic capacity line in Figure 5 extending, at a period of T = 0.88 sec., to intersect the 5 percent demand curve (S,= 5.5 inches). S_= 5.5 inches crosses the capacity curve (inelastic) at a period of about 1.3 sec. (ie., the secant period). T = 13 sec. crosses the 5 percent demand curve at about S,=8 inches, However, a secant stiffhess esponse would respond at a higher effective damping because of inelastic action. The equal displacement rule would estimate Byy = 18 percent (j., where S, = 5.5 inches on the capacity spectrum intersects the 18% damped demand spectrum). In other words, the displacement increases ue to stifiness reduction and period lengthening; but reduces due to energy dissipation and increased damping. The decrease in displacement balances the increase for a net gain of zero. Thus, the equal displacement method indicates a Buy = 18 percent vs. the CSM value of 15% percent. 10 Inelastic response spectra are generally shown in S, vs. T and/or S, vs. T formats. ‘The period TT represents the linear elastic period. The effective secant period will equal approximately the square Toot of 1 times T for an elastic-plastic capacity model that can be adjusted by the slope ratio for a bilinear model. (i.e., Typ approximately equals Tu"), Ifthe inelastic response spectrum is plotted on an ADRS format using S, v5. Typ S, V8. Tuy, oF S, vs. S, (When both plots are available), the similarities between inelastic response spectra and [yg spectra should be apparent. Secant methods, substitute structure methods and direct displacement-based design (Priestley, tal, 1996) also have similarities to the CSM that are apparent when compared on an ADRS format. Rather than using S, vs. T and S, vs. T plots separately, they can be combined on one plot showing S, S,, and T. SUMMARY AND CONCLUSIONS ‘The CSM and the ADRS format have been shown to be a useful tool evaluating existing buildings for seismic performance, verifying designs of new construction for performance goals, and correlating observed damage with recorded earthquake motion. The CSM can be used as a rapid evaluation Procedure to obtain rough estimates for large inventories of buildings (Freeman, et al., 1975) or as. a detailed procedure for new (Army, 1986 and WIE, 1996) and existing buildings (Army, 1988 and ATC, 1996). The procedure appears to be compatible with other approximate inelastic design and evaluation methods. Differences between the various methodologies have more to do with unknowns in material behavior and quantification of energy dissipation than in the methods of analysis. One area of interest that would be worthwhile obtaining a consensus is that of quantifying equivalent effective damping and inelastic response spectra to reduce differences expressed in Table 3. REFERENCES Axmmy, 1986, Seismic Design Guidelines for Essential Buildings, Departments of the Army (TM 5-809-10-1), Navy (NAVFAC P355.1), and the Air Force (AFM 88-3, Chap. 13, Section A), Washington, D.C. 1986. Ammy, 1988, Seismic Design Guidelines for Upgrading Existing Buildings, Departments of the Army (TM 5-809-10-2), Navy (NAVFAC P-355.2), and Air Force (AFM 88-3, Chap. 13, Sec B), Washington, D.C. 1988. ATC, 1982, Am Investigation of the Correlation between Earthquake Ground Motion and Building Performance (ATC-10), Applied Technology Council, Redwood City, California. 1982, ATC, 1991, U.S. Postal Service Procedures for Seismic Evaluation of Existing Buildings (Interim), Applied Technology Council, ATC-26-1 Redwood City, California. 1991. ATC, 1996, Seismic Evaluation and Retrofit of Concrete Buildings, Products 1.2 and 1.3 of the Proposition 122 Seismic Retrofit Practices Improvement Program, California Seismic Safety Commission, Report No. SSC 96-01, ATC-40, Applied Technology Council, Redwood City, California. Blume, J.A., N.M. Newmark, and Leo M. Coming, 1961, Design of Multi-story Reinforced Concrete Buildings for Earthquake Motions, Portland Cement Association, Chicago. nN Freeman, $.A., .P. Nicolet, and J.V. Tyrell, 1975, Evaluations of Existing Buildings for Setsmic Risk -A Case Study of Puget Sound Naval Shipyard, Bremerton, Washington, Proceedings of the U.S. National Conference on Earthquake Engineers, ERI, pp 113-122, Berkeley. Freeman, S.A. 1978, Prediction of Response of Concrete Buildings to Severe Earthquake Motion, Douglas McHenry Intemational Symposium on Concrete and Concrete Structures, SP-55, pp 589-605, American Concrete Institute, Detroit. Freeman, S. A., 1981, Rapid Dynamic Analysis, Proceedings, SEAOC 50th Annual Convention, Coronado, Freeman, S.A., 1987, Code Designed Stee! Frame Performance Characteristics, Dynamics of Structures Proceedings, pp 383-396, Structures Congress '87, Structural Division, American Society of Civil Engineers, Orlando, Florida, August 17-20, 1987. Freeman, $.A., 1992, On the Correlation of Code Forces to Earthquake Demands, Proceedings of the 4th U.S /Japan Workshops on Improvement of Building Structural Design and Construction Practices, ATC 15-3, August 1990, Applied Technology Council, Kailua-Kona, Hawaii. Freeman, S.A., 1995, A Review of Practical Approximate Inelastic Seismic Design Procedures for New and Existing Buildings, Proceedings, SEAOC 64" Annual Convention, pp. 311-331, October 19-21, Indian Wells, California Mahaney, J.A., TF. Paret, B.E. Kehoe and S.A. Freeman, 1993, The Capacity Spectrum Method for Evaluating Structural Response During the Loma Prieta Earthquake, 1993 National Earthquake Conference, Memphis, TN Newmark, N.M., and W.J. Hall, 1982, Earthquake Spectra and Design, Earthquake Engineering Research Institute, Oakland, California, NIST, 1994, Standards of Seismic Safety for Existing Federally Owned or Leased Buildings, U.S. Dept. of ‘Commerce, National Institute of Standards and Technology, NISTIR 5382, ICSSC RP-4, Gaithersburg, MD. Paret, T.F. , K.K. Sasaki, D-H. Eilbeck, and S.A. Freeman, 1996, Approximate Inelastic Procedures to Identify Failure Mechanisms from Higher Mode Effects, Paper No. 966, 11 World Conference on Earthquake Engineering (WCE). Priestley, MJ.N., MJ. Kowalsky, G. Ranzo, G. Benzoni, 1996, Preliminary Development of Direct Displacement-Based Design for Multi-Degree of Freedom Systems, Proceedings 65th Annual SEAOC Convention, pp. 47-66, October 1-6, 1996, Maui, Hawaii, 12

You might also like