PERFORMANCE-BASED SEISMIC DESIGN OF REINFORCED CONCRETE BRIDGE COLUMNS

Transcription

1 PERFORMANCE-BASED SEISMIC DESIGN OF REINFORCED CONCRETE BRIDGE COLUMNS Dawn E LEHMAN 1 And Jack P MOEHLE 2 SUMMARY There is a new ocus in seismic design on the perormance o reinorced concrete bridges. In a perormance-oriented environment, a bridge is designed to meet speciied perormance levels. Recent earthquakes have provided preliminary data demonstrating that large losses can result rom inadequate perormance o bridge structures. Further research is required to better deine perormance indices and develop overall design methodologies to improve the understanding o bridge perormance. A research program was undertaken to improve methods to evaluate the perormance o reinorced concrete bridge columns using experimental and analytical methods. In the experimental investigation, columns with various longitudinal reinorcement ratios and aspect ratios were tested to characterize the response o modern bridge columns subjected to lateral loading. In the analytical investigation, methods to numerically assess engineering parameters to evaluate element damage were developed using the experimental results. The research results were used to develop a perormance-based seismic design ramework or reinorced concrete bridge columns. INTRODUCTION Perormance-based design may be deined as design to reliably achieve perormance objectives. Each perormance objective is deined by a single pairing o a structural perormance level and a seismic demand level. The structural capacity or each perormance level is related to a speciic state o damage or required repair and is quantiied using one or more engineering limit states. For reinorced concrete bridges supported by columns, key aspects o structural perormance include cracking, spalling, and cross section atigue. Development o perormance-based seismic design provisions or reinorced concrete bridges requires engineering approaches that consider these aspects to deine the state o structural perormance. Recent earthquakes, including the Loma Prieta and Northridge earthquakes, have provided preliminary data to study the seismic perormance o bridges. In these earthquakes, damage was primarily ocused in older bridge construction; damage to modern reinorced concrete bridges was predominantly in the orm o cracking and minor spalling. As demonstrated by damage to older construction, damage to important bridges that results in delayed operation may be o signiicant economic cost. Although past perormance has indicated that modern bridges perormed well, the uture perormance o modern bridges is not known. Current seismic design standards or reinorced concrete bridges do not provide adequate perormance design requirements. Most current codes speciy seismic design orce levels and standard details or key structural components. Likewise, previous seismic research has ocused primarily on strength and detailing rather than structural perormance. Although uture earthquakes may result in damage to reinorced concrete bridges, it is possible that the damage may be limited; this result may indicate that current detailing requirements have resulted in overly conservative column design or the selected design earthquake. Although previous research on the response o reinorced concrete bridge columns is extensive, these studies are not adequate to develop all aspects o perormance-oriented design. Previous experimental research primarily 1 2 Acting Assistant Proessor, University o Washington, Seattle WA Director o the PEER Center and Proessor, University o Caliornia, Berkeley CA

2 has emphasized improving the design and understanding o reinorced concrete bridge columns subjected to signiicant plastic cyclic displacement demands. Development o perormance-based design methods requires urther experimental and analytical investigations to evaluate intermediate damage levels and to develop analytical models and appropriate design methodologies. Recognizing the shortcomings o current inormation, a research program was designed to develop improved methods to evaluate the perormance o modern bridge columns over the range o typical geometries and range o perormance levels. The objective o the research program was to characterize and quantiy the seismic perormance o reinorced concrete bridge columns. The research objective was achieved by designing an experimental and analytical investigation that would characterize the seismic perormance o modern bridge columns at various damage states. Results rom the experimental and analytical investigation were used to develop and evaluate the important aspects o structural damage. Engineering design criteria or use within the perormance-based seismic design ramework or reinorced concrete bridges were developed using the analytical and experimental results. PERFORMANCE-BASED SEISMIC DESIGN FRAMEWORK Perormance-based seismic design may be deined as design to reliably achieve targeted perormance objectives. Figure 1 illustrates the dierences between perormance-based seismic design and current seismic design provisions (e.g., AASHTO or bridges and UBC or buildings). Those codes deine a single level o seismic hazard and a single level o perormance that is generally understood to be lie-saety. Furthermore, those codes use indirect methods such as base shear strength and linear-elastic analysis to deine the perormance state, which can be expected to be relatively inaccurate. Perormance objectives other than the lie-saety are not evaluated explicitly; whether these perormance objectives are achieved depends somewhat randomly on the seismic environment characteristics o the structural and nonstructural components. Where current seismic design provisions speciy design requirements or a single hazard and perormance level, perormance-based seismic design can speciy perormance or a range o hazard levels. Perormance can be deined in terms o structural component parameters (e.g., spalling), structural parameters (e.g., stability), or unctionality. Categorizing structures as ordinary or important, the example perormance based-seismic design ramework has multiple perormance objectives or ordinary structures important structures. Current Seismic Design Perormance-Based Seismic Design Seismic Hazard Level Structural Perormance Level 10% Probability o being exceeded in 50 years Lie Saety Seismic Hazard Level Ordinary Important Minimum Hazard Level Intermediate Hazard Level Structural Perormance Level Fully Operational Delayed Operational Stability Maximum Hazard Level Figure 1 Perormance Objectives or Current and Perormance-Based Seismic Design Perormance-based seismic design o bridges requires that the engineer complete the ollowing tasks: select perormance objective(s), deine perormance level using engineering limit state, deine site hazard level at site, perorm structural design and evaluation using engineering approaches and quality assurance. The ollowing sections summarize recommendations or the tasks o deining and quantiying the perormance objectives within the context o a perormance-based seismic design ramework or the design o reinorced concrete bridges. Details o appropriate engineering approaches may be ound elsewhere [Lehman 1998]. Perormance Levels Structural perormance may be deined by the required repair eort. To date most codes and documents advocating perormance-based design o bridges have adopted a two-level design ramework. Logically, a the two levels correspond to the seismic hazard level or which structural repair is not required and the seismic hazard level corresponding to severe damage without collapse (structural stability), however this is not always the case (e.g. ATC 32). The perormance-based seismic design ramework recommended herein adopts three perormance levels. The three perormance levels outlined in Table 1 are designated as Fully Operational 2

3 Perormance Level, Delayed Operational Perormance Level, and Stability Perormance Level. Each perormance levels is deined by the expected bridge serviceability, required repair eort, and uture perormance. This ramework can be easily condensed to a two-level ramework. However, the distinct disadvantage o the two-level approach is that the level or which repair is required is not speciied. For a bridge meeting the Fully Operational Perormance Level, repair is not required and the bridge is expected to be ully serviceable immediately ollowing the earthquake. The uture seismic perormance will be essentially unaected which requires negligible damage accumulation. A bridge meeting the Delayed Operational Perormance Level requirements is expected to have sustained some damage during the earthquake. The bridge should provide limited service to emergency vehicles. Closure o the bridge should be limited to several days provided suicient resources are available. In uture, more signiicant events, the bridge perormance is expected to be close to the original perormance. A bridge meeting the Stability perormance level is expected to have sustained signiicant damage. As a result, partial or ull replacement o bridge elements (including columns and restrainers) may be required and the bridge may remain out o service or several weeks or months. The uture perormance o the structure is limited; however, in its damaged state, the bridge is expected to survive an atershock o lesser intensity. Table 1 Perormance Levels Perormance Level Fully Operational Delayed Operational Stability Required Repair Eort Serviceability Minimal Damage Fully Serviceable Repairable Damage Delayed Service Unrepairable damage Unserviceable Future Perormance Original Level Slightly Reduced Relative to Original Minimal Level (Atershock) Seismic Hazard Levels Ideally, seismic hazards should be considered in terms o ground motion, including temporal and spatial variation, pounding, liqueaction, lateral spreading, and landslides. Although all types o hazard levels, especially ground shaking and lateral spreading, may be critical or bridge design, seismic hazards are generally deined only in terms o ground shaking. This approach will be adopted herein. For the ramework under discussion, three perormance levels have been deined; each perormance objective requires a pairing o a perormance level and a minimum seismic hazard level. Thereore, three seismic hazard levels are deined. Ideally, the return period or the seismic hazard level will depend on the seismicity o the region and the site and is deined to match an acceptable level o uniorm risk. Although deining a single return period or each hazard level may expedite the design process, it may be more realistic to consider the risk or each perormance level. However, deining a uniorm level o risk or each perormance level depends on the seismicity o the region, economic actors, and structural importance o the bridge. In the context o the perormance-based design ramework proposed (Figure 1), the expected perormance or ordinary structures is expected to be ully operation or the minimum level event, delayed operational or the intermediate level event and lie sae or the signiicant level event. Thereore, the earthquake levels are deined in a broad sense, without speciic reerence to uniorm risk or hazard levels. Perormance Objectives A perormance objective is the pairing o a perormance level and a seismic hazard level. Discrete perormance objectives are deined or each seismic hazard level. Five perormance objectives are deined within the perormance-based design ramework (Figure 1). Three are deined or Ordinary bridge structures; two are deined or Important bridge structures. Ordinary bridges are structure is expected to meet the objectives o the Fully Operational perormance level or the Minimum hazard level, the objectives o the Delayed Operational perormance level at the Intermediate hazard level, and the objectives o the Stability perormance level at the Signiicant hazard level. Important bridges are expected to meet the objectives o the Fully Operational perormance level at the Intermediate hazard level and the objectives o the Fully Operational perormance level at the Signiicant hazard level. Closure o Important bridges is not permitted or the hazard levels speciied. EXPERIMENTAL PROGRAM Seismic design o reinorced concrete bridges requires that yielding elements withstand the expected cyclic deormation demand. Sophisticated numerical modeling may be required to ully characterize the cyclic response 3

4 o the structure. The inelastic response o reinorced concrete elements, such as columns, joints, and beams is complex, and even the most sophisticated modeling o an element can require simpliication. Thereore, analysis and design methods must be evaluated using experimental results. Many parameters can inluence the inelastic cyclic behavior o a cantilever bridge column. Previous research has demonstrated that or modern columns, the parameters that have the most inluence on the response include spiral reinorcement ratio, the column shear demand, the axial load ratio, the column aspect ratio, and the quantity o longitudinal reinorcement are the most signiicant. However, current code requirements restrict the range o these ive parameters and as a result there are typical ranges o each parameter ound in modern construction. For example, codes speciy a minimum spiral reinorcement ratio. In capacity design, column shear demands are inluenced primarily by aspect ratio and longitudinal reinorcement ratio; most seismic codes limit shear stress ratios to 12 c. Axial load levels due to dead load are typically less than 0.1 c A g, where c = concrete compressive strength and A g = the gross area. Column aspect ratios depend on bridge geometry and column diameter (typically sized to satisy maximum axial load ratios) and typically vary between 1 and 10. On average, longitudinal reinorcement quantities all between 2% and 4% o the gross cross-sectional area. A study o previous research results [Taylor 1993] indicates that although a wide range o the parameters have been studied, columns with aspect ratios o 4 or greater and longitudinal reinorcement ratios between 2% and 2.5% have received limited attention. These results indicate that longitudinal reinorcement ratio and aspect ratio, which may vary signiicantly in the ield, had not been thoroughly investigated in the laboratory. Thereore, an experimental research program was designed to investigate the inluence o these parameters on response and ailure o modern bridge columns in the context o perormance-based seismic design. The test program was designed to model the behavior o a ull-scale reinorced concrete bridge column assembly, measure local and global response quantities, and acilitate comparison with previous research studies. The experimental research study was developed to establish the eects o column aspect ratio and longitudinal reinorcement ratio on seismic behavior. Two test series were developed to individually study each ocus parameter and are shown in Figure 2; the two study parameters are identiied or each column. The column designations are indicated; each designation has three or our numerals. The latter two numerals denote the percentage o longitudinal steel; the irst one or two numerals indicate the column aspect ratio, e.g., or column designation 815, 8 Figure 2 Test Matrix indicates an aspect ratio o 8 and 15 indicates a reinorcing ratio o 1.5%). Test series I consisted o three columns that varied in longitudinal reinorcement ratio; the aspect ratio o each column was 4 to 1. The center column, representing an average bridge column, was reinorced with 1.5% steel longitudinally and was denoted Column 415. Column 407, shown to the let o Column 415, had hal the amount o longitudinal steel (0.75%); Column 430, shown to the right o Column 415, had twice the amount o longitudinal steel (3.0%) and was detailed with bundled bars. The three specimens o second test series, which will be denoted Test Series II or the remainder o the report, are depicted in the center column o the matrix. The aspect ratios o the three specimens o Test Series II varied between 4 and 10; the columns were reinorced with 1.5% longitudinal steel. Thereore, Columns 815 and 1015 had an aspect ratio o 8 and 10, respectully. Details o the geometry and reinorcement are shown in Figure 3. The column diameter was 2 eet to model a 6-oot diameter prototype column. The three columns o Test Series I had lengths o 8 eet. Columns 815 and 1015, had column lengths o 16 eet and 20 eet, respectively. The columns were reinorced longitudinally with No. 5 bars. The longitudinal reinorcement was embedded into Figure 3 Specimen Details the joint to a depth o 21.5 inches, approximately 34-bar diameters. The column spiral reinorcement ratio was 0.7%. The spiral was 1/4 inches in diameter smooth wire and spaced at 1-1/4-inches. The spiral reinorcement was continuous throughout the column height and joint depth. Footing ties were 1/4 inches in diameter spaced at 4 inches on center. 4

5 Material Properties Table 2 shows the speciied, expected and actual strengths o the longitudinal steel, spiral steel, and the concrete. The speciied strength is the minimum permissible strength. The expected strength is used in capacity design to predict the upper-bound demand rom inelastic action o adjacent elements. The actual strength is the strength measured rom the actual materials used in the test specimens. Details o the testing procedures and the measured stress-strain responses or each material can be ound elsewhere [Lehman 1998]. Table 2 Material Properties Material Speciied (ksi) Expected (ksi) Actual (ksi) Yield Ultimate Yield Ultimate Yield Ultimate Longitudinal Steel Spiral Peak Conined Peak Conined Peak Conined Concrete Varies N/M Loading Axial and lateral loads were applied to the top o the column. Figure 4 depicts the experimental coniguration. The applied axial load o 147 kips was approximately 7 caa g, where ca= the actual concrete compressive strength. The axial load ratio chosen corresponded to average axial load ratios ound in single column bent bridge construction. The axial load was applied through a spreader beam using a post-tensioning rods placed on either side o the column. The lateral displacement was applied using a servo-controlled hydraulic actuator that was attached to the top o the column. The imposed displacement history included three cycles at each displacement level. The primary displacement levels were monotonically increasing to provide an indication o damage accumulation. Both pre-yield and post-yield displacement levels were imposed. The pre-yield Figure 4 Experimental Set-up displacement levels are deined to include a displacement level prior to cracking, two levels between cracking and yielding, and a level approximately corresponding to the irst yield o the longitudinal reinorcement. The post-yield displacement levels are deined to include all subsequent cycles. For the post-yield displacement levels, a small displacement cycle was imposed ollowing the three main cycles. Imposed displacement histories were determined or each column according to the column aspect ratio rom nominally identical displacement ductility histories. As a result, the three columns o Test Series I were subjected to the same displacement history. Details o the loading regime may be ound elsewhere [Lehman 1998] EXPERIMENTAL IMPLICATIONS FOR PERFORMANCE CRITERIA To quantiy the structural capacity (or perormance) or each perormance level, the experimental observations must be quantiied to deine the perormance levels. As described previously, the perormance levels are deined in terms o the required repair eort. Thereore, the damage states must relate to the repair levels. The repair levels o interest are the no repair, repairable, and beyond repair without collapse. Observations during testing o the ive columns suggest that the sequence o damage was similar o the ive columns. This section provides a general description o the progression o damage listing each category o damage chronologically. A brie description o the visual indications is provided. Speciic occurrences o each stage o damage are provided or each column may be ound elsewhere [Lehman 1998]. Cracking Typically, cracking was not detected during the initial displacement level but was initiated during the subsequent cycle. The crack spacing in the lower portion o the column stabilized ollowing the yield. 5

6 Initial yielding o longitudinal steel Yielding o the extreme longitudinal reinorcing bar was noticeable in the orce-displacement response or in the physical response. Yielding was detected using strain gauges that had been placed on the longitudinal steel prior to construction and were monitored during testing. Spalling Initial spalling occurred above the column-ooting interace. With continued loading, the spalling region increased in elevation, around the circumerence, and into the column core. Exposure o spirals and longitudinal steel Complete loss o the concrete cover exposed the spirals and longitudinal steel. Visual extension/racture o spiral and longitudinal bar buckling Subsequent loading resulted in a permanent displacement o the lower column spirals. Longitudinal bar buckling was visually evident. Spiral racture The spirals located within the buckled length continued to extend as the bar continued to buckle until the spiral ractured. The lateral stiness decrease as a result o spiral racture which permitted the other longitudinal bars to buckle over a longer length. Longitudinal bar racture Fracture o the longitudinal bars occurred ater bar buckling. Typically, racture o one or more longitudinal bars resulted in strength loss signiicant enough to cause column ailure. Previous research suggests that requirement or a limited repair depends on the crack widths. Recent studies [Elkin 1998] indicate that the extent o spalling, core crushing and damage to the longitudinal reinorcement dictates the ability to repair a bridge column primarily responding in a lexural mode. Similar work suggest diiculty in repairing reinorced concrete cross-sections damaged due to atigue o the core concrete and longitudinal reinorcement without replacement eort. Engineering parameters including tensile strain in the longitudinal reinorcement, compressive strain in the cover concrete, and strain-based low-cycle atigue damage index are used to evaluate the damage states and thereby required repair eort. Perormance levels are then deined using engineering limit states expressed using the applicable engineering parameters. Cracking Crack widths and crack patterns may be used to indicate i repair is required. Large residual crack widths (rom 1-2 in.) may be required to be illed with epoxy or other material to restore the tensile strength. The residual crack widths measured during testing were used to postulate maximum permissible displacement ductility demand to ensure minimum crack widths. Although strain demands should provide a more uniorm assessment o crack widths, observations o the post-yield response o the longitudinal strain gauges indicated the measurements were not reliable when the yield plateau was reached. In addition, since yielding o the cross section is progressive, local strain readings do not indicate cross section crack widths. In general, the measured response indicates that the residual crack widths were 1 inches or less or displacement ductility demands less than 1.5 and are 2 inches or less or displacement ductilities less than 2. To limit residual crack width, the displacement demand should be less than the twice the eective yield displacement. Limiting the displacement demand to the eective yield displacement will ensure crack widths that do not require column repair and essentially linear response. Spalling Cover spalling may result in a reduction in the lateral stiness o the cross section, durability, and lateral restraint on the longitudinal bar. Post-earthquake damage to the concrete cover can require concrete patching; core damage can require partial or complete replacement o the structural element. The experimental observations made during this study, which are relevant to the behavior o modern columns, are used to correlate physical damage and predicted response. Additional observations rom experimental research [Elkin 1998], which ocused on the repair o modern bridge columns, and experimental research by [Kunnath 1997] and [Calderone 1998], both o which ocused on the perormance o modern bridge columns, are included to substantiate the indings. An analysis was perormed or each test specimen using a discrete modeling technique developed by the authors using the measured material properties. Details o the model and analysis may be ound elsewhere [Lehman 1998]. Table 3 summarizes the results. The measured initial spalling displacement is provided in the second column. The predicted compressive strain in the extreme iber corresponding to the initial spalling displacement is recorded in the last column o the table. The results indicate that the strain corresponding to initial spalling o the cover is in the range o -08 to -1. For the provided data, the mean spalling strain is -09 with a standard deviation o 01. Although spalling is not uniquely related to strain demand, the results indicate that compressive strain may provide a reasonable estimate o initial concrete spalling. For design, a compressive strain o -07, is suggested. Due to the size o the sample set however, urther analysis is warranted. 6

7 Table 3 Spalling Strains Research Team Column Measured Spalling Displacement Predicted Spalling Strain Lehman and Moehle in in in in in. -08 Kunnath et al. A1 1.1 in. -1 Calderone et al. 328 in in in. -1 Mean -09 Standard Deviation 01 Fatigue and Cross Section Failure The response and ailure o reinorced concrete elements subjected to seismic loading can be inluenced by the imposed displacement history. Fatigue o the concrete cover may require removal and replacement o the damaged concrete. Cross section ailure, which may include longitudinal bar atigue and atigue o the core concrete, may require partial or ull replacement o a structural element. Experimental observations suggest that response o a longitudinal bar subjected to cycle loading depends on the lateral restraint provided. The lateral bar restraint, in turn, depends on the condition o the surrounding concrete, and the stiness and spacing o the spiral. Experimental evaluation o the use o such replacement techniques on modern bridge columns may be ound in the literature [Elkin 1998]. Herein, a preliminary study was undertaken to develop a new damage index to model the observed experimental response. Damage resulting rom cyclic loading is modeled using a two-phase model. The irst phase models damage to the lateral restraint on the longitudinal bar as damage to the concrete cover. The second phase models damage o the longitudinal steel. Both phases use a modiied ormat o the Coin-Manson equation as shown in Equation 1; the expression relates the number o complete cycles to ailure, N, to a normalize strain, ε r = ε/ε o. Miner's rule is employed to determine the damage index, DI (Equation 2). Failure is predicted when DI = 1. Equations 1 and 2 were used to develop the expressions or the concrete damage phase and the steel damage phase. The model was developed using the experimental results rom Kunnath et al. Details o the development o the model may be ound elsewhere [Lehman 1998]. N = a( ε ε o ) + c DI = ( 1 N ) Eqs. 1 and 2 Using Equation 1, two expressions were developed to estimate the number o cycles required to ully damage the concrete cover and the longitudinal steel at a particular strain demand. Equation 3 predicts the number o cycles required to completely remove the concrete cover, (N ) c, at a strain ratio o ε c /ε csp where ε c is the compressive strain and ε csp is the strain corresponding to spalling in the extreme iber o the conined core (see previous section). Equation 4 predicts the number o subsequent cycles to ailure, (N ) s, ollowing atigue-induced ailure o the cover (i.e., DI c = 1). Note that the expression correctly predicts that ailure will result i the column is subjected to one cycle or which ε s = ε su. ( N ) 33( ε ε ) ( N ) = 8( ε ) = c c csp 1 DI = ; DI 1 c i ( N ) ci c s i ( N ) ε Eqs. 3 and 4 s su 1 DI s = i DIc = 1; DI s = 1 Eqs. 5 and 6 i The concrete and steel damage indices, DI c and DI s respectively, are calculated using Miner's rule, as shown by Equations 5 and 6. The experimental results indicate that low-cycle atigue o the longitudinal reinorcement ollows complete spalling o the concrete cover. Mathematically, this is indicated when DI c = 1. Damage to the longitudinal reinorcement is modeled using Eq. 6. Cross-section ailure resulting rom ailure o the longitudinal steel corresponds to approximately DI s = 1. The dual-phase damage index is employed in two stages. The steel atigue index, (DI) s, is set equal to zero until the concrete damage index, (DI) c, is equal to one. Element ailure corresponds to a steel atigue index value o one, i.e., (DI) s =1. si 7

8 Value o Damage Index Value o Damage Index Value o Damage Index Column Column Column Value o Damage Index Value o Damage Index Column Column Concrete Damage Index Steel Damage Index Figure 5 Results using Dual-Phase Damage Index The dual-phase damage index was evaluated using the experimental results rom the present study. The analytical results are shown in Figure 5. The progression o damage predicted by the concrete atigue model is shown by a line marked with circles. The progression o damage predicted by the longitudinal steel atigue model is shown by a line marked with squares. The dual-phase index correctly predicts ailure or Columns , and 815. Failure o Column 430 is predicted a bit early. A damage index value o 0.92 corresponds to ailure o Column The Fatigue Engineering Limit States are speciied or the Delayed Operational perormance level and the Stability perormance level; each corresponds to the required repair eort speciied in 4. Fatigue ailure o the concrete is not permitted or a bridge designed to meet the Delayed Operational perormance state. For the Lie Sae Perormance Level, ailure o the concrete is permissible (i.e., DI c = 0); however, ailure o the longitudinal steel should be avoided (i.e., DI s < 0.9). Repair (i.e. concrete patching) will be required i DI c > 0. Table 4 Engineering Limit States or Dierent Levels o Repair Numerical Expression Physical Damage Repair Tensile Strain in Steel Cracking Epoxy Injection Compressive Strain in Cover Initial Spalling Patching Residual Drit Residual Drit Plumb Structure Dual-Phase Damage Index (DI) c =1 (DI) s = 1 Complete Spalling Concrete Replacement (DI) s = 0.9 Bar/Spiral Failure Replacement Only CONCLUSIONS In modern construction, the seismic perormance o reinorced concrete bridge structures depends on the response o the ductile hinge regions. A research program was undertaken to characterize the response o wellconined, circular cross section, concrete bridge columns. The research objectives included evaluating current design procedures and recommended perormance-based design procedures or reinorced concrete bridges in seismic zones. The research was executed in three stages. The existing literature was reviewed and used to guide the design o the experimental and analytical components o the investigation. The results o the research study were used to successully develop a perormance-based design ramework or reinorced concrete bridges. CITED REFERENCES [SEAOC 1995] Structural Engineers Assn. o Caliornia (SEAOC). Vision 2000 Committee. Perormance-Based Seismic Engineering o Buildings [Lehman 1998] Lehman, D.E. Perormance-Based Seismic Design o Circular Concrete Bridge Columns, PhD Dissertation, University o Caliornia, Berkeley 1998 [Calderone 1998] Calderone, A.C., Lehman, D.E. and Moehle, J.P. Behavior o Reinorced Concrete Columns Containing Varying Aspect Ratios and Varied Zones o Coninement. PEER Center Report. June 1998 [Kunnath 1997] Kunnath, S. K. Cumulative Seismic Damage o Reinorced Concrete Bridge Piers. NCEER ; Technical Report. National Center or Earthquake Engineering Research, Bualo, N.Y [FEMA 1997] NEHRP Guidelines or the Seismic Rehabilitation o Buildings. Federal Emergency Management Agency. FEMA 273. October 1997 [ATC 1996] Applied Technology Council. Improved Seismic Design Criteria or Caliornia Bridges: Provisional Recommendations. Report ATC