Seismic fragility assessment of continuous RC rigid-frame bridges with irregular geometry: insights and recommendations Authors: Seyed Hadi Rashedi, Alireza Rahai Published: 2026 ABSTRACT AbstractThis study presents a detailed seismic fragility and collapse performance assessment of geometrically irregular continuous reinforced concrete rigid-frame (CRCR) bridges based on a representative bridge in Iran. A regular configuration was first established as the reference, after which a systematic set of bridge models was generated by varying the span-length ratio, pier-height ratio, and pier skew angle to investigate their individual and combined effects on seismic response. Detailed three-dimensional nonlinear finite element models were developed in OpenSees and analyzed under a comprehensive set of far-field and near-field earthquake ground motions. Seismic fragility curves were constructed based on pier drift demand corresponding to four damage states ranging from minor damage to collapse, defined by the attainment of the complete damage state. The results indicate that geometric irregularities significantly amplify seismic vulnerability, particularly at higher damage states. Across a range of span-length ratios, bridges with pronounced pier-height irregularity exhibit up to 50–60% higher probabilities of exceeding severe damage at the same intensity level compared to the regular configuration, while the median seismic intensity associated with collapse is reduced by approximately 55% in highly irregular cases. Among the investigated parameters, pier-height irregularity was identified as the most influential factor governing fragility and collapse behavior, followed by pier skewness, whereas span-length irregularity showed a comparatively smaller effect. The resulting fragility functions and collapse metrics provide quantitative insight into the seismic performance of CRCR bridges and support fragility-based assessment and performance-based seismic design in earthquake-prone regions.Keywords: Bridge structure, Seismic performance assessment, Irregularity, Nonlinear structural analysis, Seismic fragility analysisSubject terms: Engineering, Natural hazards, Solid Earth sciences Introduction Bridges are vital parts of transportation systems. In regions exposed to seismic hazards, the structural integrity of bridges takes on added significance, as their failure during an earthquake can have catastrophic outcomes. Among the various bridge types, continuous reinforced concrete rigid-frame (CRCR) bridges stand out due to their prevalence and importance in many transportation structures1. The performance of these bridges, which have continuous decks and are supported by piers made of reinforced concrete (RC), is well-known for their ability to handle heavy loads and to function effectively over long spans. However, they are not protected against the destructive forces of earthquakes, and their vulnerability to seismic activity increases as irregularities in geometry and structural details are added. Irregularities in bridges can take diverse forms, and these may encompass variations in span lengths, pier heights, or other geometric parameters, in addition to deviations in reinforcement detailing. They may be intentional or a result of design constraints, construction obstacles, or even the modification of existing bridges. Regardless of their origin, these irregularities can drastically impact the seismic response of bridges. In recent years, researchers have explored the significance of evaluating the vulnerability of bridge systems by considering a variety of factors. Analytical responses are examined in bridge components such as piers, abutments, and superstructures. In addition, consideration is given to the development of fragility curves for different bridge classes and their seismic performance levels. In the following, some of the studies and research that have been carried out on the vulnerability and fragility of these bridges have been presented2. Seismic fragility analysis is significant in bridge design as it helps assess the vulnerability of bridge structures to earthquakes3. A performance-based seismic fragility assessment of bridges takes into account seismic wave characteristics, bridge structure specifics, potential damage mechanisms, performance criteria, and uncertainties related to seismic events and structural responses. It provides valuable information about the probability of different levels of damage or failure that a bridge may experience under various earthquake intensities4–9. The earthquakes that occurred in 1989 at Loma Prieta, 1994 at Northridge, and 1995 at Kobe marked a significant turning point in the exploration of fragility. A number of researchers collaborated to develop empirical fragility curves that were calculated by the actual damage caused by these three earthquakes10,11. In the field of seismic fragility analysis for bridge structures, several methodologies have been developed to evaluate vulnerability. In general, these approaches can be categorized into three main groups: ones based on expert opinion12, ones based on empirical evidence3,13, and others based on analytical methodologies14,15. Among various methodologies, analytical fragility curve methods have gained widespread acceptance7. In a study by Mosleh et al.16, they conducted a seismic performance analysis on a multi-span bridge using SAP2000 software17. They developed fragility curves from near-field and far-field ground motions to assess damage probabilities, suggesting future research should include additional structural elements for comprehensive fragility curves. Zhang et al.18 focus on steel fiber-reinforced concrete bridge piers, offering insights into damage analysis, structural response under seismic conditions, and probabilistic seismic demand calculations. They provide valuable fragility curves for damage probabilities. Stefanidou et al.19 emphasized seismic vulnerability assessment with a focus on soil-structure interaction, encompassing steps like finite element modeling and limit state thresholds. Pang and Wu20 underscore the importance of assessing the entire bridge system, creating fragility curves for components, and highlighting uncertainty in ground motion. Ren et al.21 and Zhao et al.22 investigated seismic fragility and vulnerability analysis methodology, utilizing cloud approaches, nonlinear time-history analysis, and OpenSees software23,24. They emphasize the consideration of uncertainty in structural parameters. These resources are essential for researchers and engineers working on seismic bridge design and reinforcement in susceptible regions. Zhang et al.25 introduced a method for seismic fragility analysis of bridge piers, employing moment-based techniques to simulate seismic accelerations, calculate structural response moments, and determine failure probabilities through fourth-moment reliability indices. Fragility curves are generated for various PGA values and damage levels, underscoring the role of design response spectrum parameters. In other recent studies, researchers focused on a seismic fragility analysis of various CRCR bridges, covering structural damage and seismic performance under different conditions and offering findings and design recommendations1. The researchers observed that the traveling wave effect significantly impacts the fragility of long-span CRCR bridges in mountainous areas, with low piers having a higher likelihood of damage than high piers26. Additionally, they analyzed the combined effects of actual site conditions and collisions on the fragility of high-pier, long-span CRCR bridges, which was closely tied to the pier heights in the bridge structure27. Also, using a probabilistic framework, including seismic hazards and fragility curves, the studies assess the structural response and quantify the expected annual losses of rigid-frame bridges28. A substantial body of research has explored the seismic damage mechanisms and vulnerability of CRCR bridges exposed to marine environmental conditions29,30. Jiang et al.31 considered factors such as pile-soil interaction and hydrodynamics in their analysis. Liang et al.32 explored time-varying seismic fragility during a bridge’s service period. Zhang et al.33 emphasized the importance of fuzzy damage criteria and conducted a case study on a Chinese coastal bridge. Liang et al.34 also introduced optimized analysis methods for assessing bridge collapse resistance and robustness, with a specific focus on a four-span offshore CRCR bridge, addressing durability damage and bond-slip effects. Cui et al.35 examined seismic fragility analysis for high-pier CRCR bridges in deep-water areas, addressing hydrodynamic pressure impact and damage states with a focus on aiding design and assessment. These studies collectively stress the significance of various factors in assessing the seismic performance and vulnerability of marine bridges. Furthermore, several aspects that have the potential to exert influence on CRCR bridges, such as material, seismic excitations, long-period seismic waves, etc., were also given proper consideration36–41. For irregular bridges with varying pier heights and different structural configurations, fragility curves help in understanding the specific vulnerabilities of different components of the bridge1. They provide insights into the likelihood of damage occurrence and the severity of damage for different parts of the bridge under seismic loading. Irregularity in bridge structure, often subject to intense and highly damaging near-fault ground motions, results in a significant reduction in the shear and compressive capacities of their piers, especially when located in near-fault zones, as demonstrated by studies such as Wilson et al.42, Jeon et al.43, Theodoulidis et al.44, Deshan et al.45, and Wei et al.46. In order to investigate the effects of irregularities in the structure of CRCR bridges, some studies have been conducted. Akbari47 examined the seismic fragility of irregular CRCR bridges by varying pier height as a design parameter and highlighted the role of damage and limit states in ensuring seismic safety. The demand-capacity ratio approach reveals that the shorter piers on the irregular bridge are severely damaged during a high-intensity earthquake, while the higher piers maintain elastic. Dong et al.48 presented a theoretical model for assessing seismic vulnerability in irregular CRCR bridges, highlighting bridge piers’ susceptibility to earthquake damage. Chen et al.49 discussed seismic vulnerability and fragility analysis in long-span irregular CRCR bridge structures and emphasize variable seismic damage probabilities among structural components, underscoring the importance of evaluating systemic seismic performance. In the following, Shan et al.50 discussed seismic fragility analysis of non-circular irregular CRCR bridge piers in China, emphasizing unique challenges, proposing a methodology, considering ground motion angles, and providing fragility curves. Overall, these studies contribute to a comprehensive understanding of the effects of irregularities in CRCR bridge seismic vulnerability assessment and design improvement and provide necessary guidance to increase the seismic resilience of these bridges and configurations in earthquake-prone areas. While a substantial body of research has addressed the seismic vulnerability and fragility of CRCR bridges, the majority of existing studies have examined geometric irregularities independently. In real bridge systems, differences in span length, pier height, and pier skew angle often occur simultaneously and interact, producing coupled seismic response characteristics and localized damage patterns that cannot be captured by analyses focusing on a single parameter. More specifically, the combined effects of pronounced span-length asymmetry, pier-height irregularity, and skewed pier arrangements on the seismic fragility of CRCR bridges remain insufficiently explored. This limitation reduces the reliability of seismic performance prediction and post-earthquake damage assessment for irregular CRCR bridges in seismically active regions, which the present study addresses through the adoption of a comprehensive probabilistic fragility assessment framework. This study systematically evaluates the seismic vulnerability and fragility of geometrically irregular CRCR bridges using a comprehensive numerical framework. Span-length variation, pier-height non-uniformity, and pier skewness are explicitly incorporated, both separately and in combined scenarios, to account for their coupled influence on seismic response behavior. Detailed three-dimensional nonlinear finite element models are developed in OpenSees23, and seismic demands are evaluated using nonlinear time-history (NTHAs) and incremental dynamic analyses (IDAs) under a set of recorded earthquake ground motions selected in accordance with FEMA P-69551 recommendations, representing both near-field (NF) and far-field (FF) excitation characteristics. Bridge pier drift demand is adopted as the damage index, and four damage states—minor, moderate, severe, and complete—are defined based on established performance criteria52. Seismic fragility curves are subsequently derived within a probabilistic framework that accounts for uncertainties associated with ground motion variability and nonlinear structural response. The resulting fragility functions enable a quantitative comparison of different irregular bridge configurations and identify geometric irregularities that most significantly increase the probability of pier- and system-level failure, providing valuable insight for seismic performance evaluation and design of irregular CRCR bridges. Methodology This study adopts a numerical simulation–based framework to evaluate the seismic fragility of CRCR bridges with irregular geometry, using a real bridge in Iran as the case study. A reference bridge with regular geometry was first selected to represent typical CRCR bridges in the national roadway network, and a three-dimensional finite element model (FEM) was developed in OpenSees version 3.2.224, in which material nonlinearity, geometric nonlinearity, and the seismic response of the primary structural components were explicitly considered. Using the reference configuration as a starting point, geometric non-uniformities were introduced by developing multiple irregular bridge models with systematic variations in span-length ratios, pier-height ratios, and pier skew angles. NTHAs were subsequently performed for all configurations using ground motion records consistent with the regional seismic hazard, from which key engineering demand parameters were obtained to evaluate structural performance. Seismic fragility curves were then derived by linking selected ground motion intensity measures with the probability of exceeding predefined damage states, thereby allowing a quantitative evaluation of how geometric irregularities influence the seismic vulnerability of CRCR bridges. Case study bridge overview An existing bridge was adopted as the reference model to enable a comprehensive evaluation of the seismic fragility and vulnerability of CRCR bridges exhibiting geometric irregularities. The reference bridge is a prestressed CRCR structure characterized by a regular geometric configuration and situated in Iran. Figure 1 illustrates the overall configuration of the case-study bridge together with its three-dimensional numerical model, which together define the baseline structural system from which the irregular variants analyzed in subsequent sections are derived. Case studied bridge: (a) Configuration (b) Three-dimensional finite-element model. The bridge adopts a continuous three-span layout (Fig. 1), having a total length of 215 m with individual spans of 60 m, 95 m, and 60 m. Figure 2(a) depicts the bridge layout and principal geometric features of the superstructure, while Fig. 2(b) presents the main girder cross-sections. Figure 2(b) shows that the deck is formed as a prefabricated box girder with a variable depth along the span, decreasing from 5.4 m at the pier locations to 2.3 m at mid-span and near the abutments; this configuration is typical of long-span rigid-frame systems and controls the distribution of longitudinal stiffness. Structural layout of the case-study bridge: (a) schematic view of the bridge (unit: m); and (b) cross-sectional views of the main girder (unit: m). The substructure consists of symmetric RC piers with a uniform height of 27.7 m, resulting in a regular baseline configuration in terms of geometry and stiffness (Fig. 1). The pier geometry and the distribution of pier sections along their height are presented in Fig. 3(a), whereas Fig. 3(b) and Fig. 3(c) illustrate the detailed prismatic and non-prismatic pier sections, respectively. The dimensional properties of these prismatic and non-prismatic pier segments, including the corresponding section parameters, are reported in Table 1, which provides the quantitative geometric inputs required for consistent numerical modeling of pier nonlinearity. Pier geometry and reinforcement details: (a) arrangement of sectional locations along the piers; and (b,c) prismatic and non-prismatic pier sections. Table 1Dimensions of the piers in prismatic and non-prismatic parts.PartPiers section propertiesHeight(m)Sectiona(m)b(m)c(m)d(m)e(m)f(m)g(m)H110(3–3)2.5102.651.50.50.52.1H217.7(1–1) & (2–2)2.56.82.651.50.50.50.5 Dimensions of the piers in prismatic and non-prismatic parts. Longitudinal post-tensioning of the superstructure is achieved through a combination of isostatic and hyperstatic tendons, providing a prestressing force of approximately 158 tons. According to the bridge design specifications, the deck and piers are constructed using C40 and C30 concrete, respectively, based on 28-day compressive strength values. The piers are reinforced with S400 steel, incorporating longitudinal bars with diameters of 32 mm and 25 mm. Because the piers are symmetric, they share identical reinforcement layouts and sectional properties. The detailed longitudinal reinforcement characteristics (e.g., reinforcement ratios/areas by section) together with the transverse reinforcement ratios for the prismatic and non-prismatic regions are summarized in Table 2, which completes the structural definition of the pier sections shown in Fig. 3 and supports the subsequent nonlinear seismic analyses and fragility derivations. Table 2Details of longitudinal and transverse reinforcing in prismatic and non-prismatic parts of the piers.SectionLongitudinal reinforcement propertiesTransverse reinforcement (%)ABDEFSection area ()Reinforcement (%)Strong directionWeak direction(1–1)1063223321325132510259.052.710.650.50(2–2)1063223321325132510259.052.710.650.50(3–3)14432133213251325102517.051.720.650.50 Details of longitudinal and transverse reinforcing in prismatic and non-prismatic parts of the piers. Geometric irregularities and bridge models Using the case-study bridge as a reference (BM0), a systematic series of bridge models was developed to examine the influence of geometric irregularities on the seismic response and fragility of CRCR bridges. The complete categorization and structural characteristics of the reference model (BM0) and all configurations are reported in Table 4, which serves as the identification framework for the bridge models analyzed in this study. Figure 4 provides a schematic overview of the geometric deviations applied to the bridge models, highlighting variations in span length together with changes in pier height, while the effects of pier skew angles are shown with respect to the reference configuration. Table 4Structural configuration of bridge models.Dimensional and geometrical parametersModel IDSpan length ratio(L/l)Pier height ratio(h2/h1)Skew angle (pier #1)θ(degree)Skew angle (pier #2)θ(degree)Model IDSpan length ratio(L/l)Pier height ratio(h2/h1)Skew angle (pier #1)θ(degree)Skew angle (pier #2)θ(degree)BM00.631.000BM310.72.000BM10.51.000BM320.72.01515BM20.51.01515BM330.72.03030BM30.51.03030BM340.72.04545BM40.51.04545BM350.72.06060BM50.51.06060BM360.73.000BM60.51.500BM370.73.01515BM70.51.51515BM380.73.03030BM80.51.53030BM390.73.04545BM90.51.54545BM400.73.06060BM100.51.56060BM411.01.000BM110.52.000BM421.01.01515BM120.52.01515BM431.01.03030BM130.52.03030BM441.01.04545BM140.52.04545BM451.01.06060BM150.52.06060BM461.01.500BM160.53.000BM471.01.51515BM170.53.01515BM481.01.53030BM180.53.03030BM491.01.54545BM190.53.04545BM501.01.56060BM200.53.06060BM511.02.000BM210.71.000BM521.02.01515BM220.71.01515BM531.02.03030BM230.71.03030BM541.02.04545BM240.71.04545BM551.02.06060BM250.71.06060BM561.03.000BM260.71.500BM571.03.01515BM270.71.51515BM581.03.03030BM280.71.53030BM591.03.04545BM290.71.54545BM601.03.06060BM300.71.56060 Structural configuration of bridge models. Overview of the structural layouts considered for the bridge models. The selection of geometric irregularity parameters and their associated ranges was guided by prior numerical investigations and by national and international design standards, including Iranian bridge design guidelines published by the MRT-RITC53, as well as Caltrans54 and AASHTO55. Vertical irregularity is typically defined using the pier height ratio, expressed as the height of the tallest pier divided by that of the shortest, where ratios of approximately 1.3–1.5 are generally regarded as irregular, and values ranging from 1.5 to 3.0 are frequently documented in analytical studies and actual bridge designs56,57. Longitudinal irregularity is commonly quantified using the span length ratio, which represents the proportion between the shortest and the longest spans. Span length ratios lower than about 0.7–0.8 indicate irregular configurations, whereas ratios between 0.5 and 0.7 are commonly encountered in bridge practice in Iran as well as internationally. The presence of pier or deck skew angles represents a major contributor to plan irregularity. In common classifications, skew angles between 15° and 30° are treated as small, angles in the range of 30°–45° are considered moderate and are often linked to pronounced dynamic effects, while angles exceeding 45° are categorized as large, reflecting severe skew conditions58. Based on available research and data from constructed bridges, the parameter ranges most frequently associated with irregular bridge configurations include pier height ratios of 1.5–3.0, span length ratios of 0.5–0.7, and skew angles in the range of 30°–45°. In this study, these established parameter ranges were employed to systematically evaluate the influence of geometric irregularities on seismic behavior. The bridge configurations were adopted, with minor adjustments, from a previous numerical investigation on CRCR bridges59, ensuring both realistic representation and comparability with earlier research. Based on this approach, the irregular bridge models were developed by varying three principal geometric parameters, including the span length ratio (L/l), the pier height ratio (h2 /h1), and the pier skew angle. The discrete parameter settings adopted in the study are documented in Table 3, where the span length ratio is evaluated at three levels, the pier height ratio at four levels, and the skew angle across five levels. These values cover regular, moderately irregular, and severely irregular configurations observed in practice. Table 3Dimension parameters.ParametersValuesSpan length ratio (L/l)0.5, 0.7, 1.0Pier height ratio (h2/h1)1.0, 1.5, 2.0, 3.0Skew angle of piers θ (degree)0, 15, 30, 45, 60 Dimension parameters. Through the systematic combination of the parameter values presented in Table 3, the complete set of bridge models was generated. As shown in Table 4, the reference bridge (BM0) is clearly identified, while the remaining models correspond to various irregular configurations derived from this baseline model. It is important to note that to ensure realistic structural behavior and a logical progression of plastic hinge formation across different configurations, all irregular bridge models (BM1 to BM60) were systematically redesigned based on their modified geometries. While the primary objective of this research is not to perform detailed design, but rather to investigate the pure effect of geometric irregularities, applying the exact same pier cross-sections and reinforcement details to configurations with significantly different pier heights or span lengths would not reflect practical design reality and could lead to unrealistic comparative results. Therefore, to achieve meaningful and comparable seismic performance, each bridge model was initially redesigned using CSiBridge software60 in accordance with relevant design codes. This process ensured that for every variation in pier height, span length, or skew angle, the piers possessed appropriate cross-sectional dimensions and reinforcement ratios to withstand the anticipated gravity and seismic demands. Consequently, the subsequent nonlinear analyses in OpenSees24 were performed on consistently designed models, allowing the observed differences in seismic response and fragility to be reliably attributed to the geometric irregularities themselves, rather than to inconsistencies in member capacity. Numerical modeling The seismic behavior of the bridge system was investigated using fully three-dimensional nonlinear finite element models developed in OpenSees24. This modeling framework inherently captures the coupled longitudinal, transverse, and torsional responses induced by geometric irregularities, including variations in pier height and pier skewness. Recorded earthquake ground motions were applied to the three-dimensional bridge models, and the seismic response was evaluated through NTHAs and IDAs, allowing the interaction effects among different response components to be explicitly represented. To evaluate the seismic performance and vulnerability of the bridge piers, IDA was conducted using a selected suite of earthquake ground motion records. Each ground motion record was incrementally scaled to higher intensity levels in order to track the structural response from the elastic range through pronounced nonlinear deformation. Pier displacement demand, expressed in terms of drift ratio, was extracted from the NTHAs and adopted as the engineering demand parameter for damage assessment. Based on these results, seismic fragility curves were derived to quantify the probability of exceeding predefined damage states as a function of seismic intensity. The resulting fragility functions provide a consistent basis for comparing different bridge configurations and for evaluating the influence of geometric irregularities on the seismic response and performance of the bridge system. Finite element–based numerical model In this study, a detailed finite element model of the bridge was developed, utilizing nonlinear force-based Beam–Column Fiber elements to simulate the piers. This formulation satisfies equilibrium exactly and captures distributed plasticity along the member height without relying on predefined plastic hinge lengths, thereby offering a more realistic representation of seismic damage progression compared to lumped plasticity models. To ensure accuracy while maintaining computational efficiency, each pier was discretized into five elements based on a convergence study, with five Gauss-Lobatto integration points per element providing critical sampling at the ends where moments are highest. This refined meshing strategy, combined with the fiber section approach, allows the inelastic behavior and curvature distribution to develop naturally based on moment-gradients and material response (Fig. 5). Structural FEM of the studied bridge. The pier–girder connection was modeled using Rigid Link Beam elements to represent a fully rigid, non-deformable joint. This modeling strategy ensures consistent transfer of axial forces, shear forces, and bending moments among structural components, while preventing artificial flexibility at component interfaces. An elastic behavior was assumed for the bridge deck in all analyses, given that the investigation primarily concentrates on the seismic performance and vulnerability of the piers. This assumption aligns with previous studies indicating that, for comparable bridge systems, the global seismic response is primarily controlled by pier behavior, although deck inelasticity may add to energy dissipation under intense ground motions. Concrete material behavior was represented using the Concrete0223 constitutive model to simulate both confined and unconfined concrete fibers61,62, whereas the longitudinal reinforcement was modeled using the Steel0223 material model. These constitutive models were chosen because of their established ability to capture stiffness degradation, strength deterioration, and hysteretic response under cyclic loading. Ground motion selection For the seismic performance evaluation carried out in this study, NTHA and IDA were conducted using recorded earthquake ground motions, which were chosen based on the selection criteria specified in FEMA P-69551. The selected dataset includes both far-field (FF) and near-field (NF) ground motions in order to capture a wide range of earthquake characteristics, including variations in magnitude, source-to-site distance, site conditions, and frequency content, as well as potential near-fault pulse (NFP) effects that may significantly influence nonlinear structural response. A total set of ground motion records was adopted. For each earthquake event, the two recorded horizontal components were retained and, where required, rotated to align with the bridge longitudinal and transverse axes. The resulting longitudinal and transverse acceleration time histories were then applied simultaneously as bidirectional uniform support excitations in the nonlinear analyses. For record characterization and scaling purposes, an equivalent scalar horizontal intensity was computed using the square-root-of-sum-of-squares (SRSS) combination of the two components. The seismological properties of the selected records, including earthquake event, recording station, magnitude (M), source-to-site distance, site classification, and peak ground acceleration (PGA) are summarized in Table 5. Table 5Properties of selected ground motions51.ID noEarthq. nameStation nameYearMTypeMechPGA (g)Vs30 (m/s)1San FernandoLA-Hollywood Stor FF19716.6FFReverse0.298316.462Friuli, Italy-01Tolmezzo19766.5FFReverse0.476505.233Imperial Valley-06Delta19796.5FFStrike slip0.422242.054Imperial Valley-06El Centro Array #1119796.5FFStrike slip0.528196.255Superstition Hills-02El Centro Imp. Co. Cent19876.5FFStrike slip0.442192.056Superstition Hills-02Poe Road (temp)19876.5FFStrike slip0.555316.647Loma PrietaCapitola19896.9FFReverse Oblique0.674288.628Loma PrietaGilroy Array #319896.9FFReverse Oblique0.669349.859Cape MendocinoFortuna Blvd19927FFReverse0.163457.0610LandersCoolwater19927.3FFStrike slip0.505352.9811LandersYermo Fire Station19927.3FFStrike slip0.288353.6312Northridge-01Beverly Hills- Mulhol19946.7FFReverse0.659355.8113Northridge-01C. Country-W Lost Cany19946.7FFReverse0.621325.614Kobe, JapanNishi-Akashi19956.9FFStrike slip0.67060915Kobe, JapanShin-Oosaka19956.9FFStrike slip0.32425616Kocaeli, TurkeyArcelik19997.5FFStrike slip0.24952317Kocaeli, TurkeyDuzce19997.5FFReverse0.479281.8618Chi-Chi, TaiwanCHY10119996.2FFReverse0.190258.8919Chi-Chi, TaiwanTCU04519996.2FFStrike slip0.012704.6420Duzce, TurkeyBolu19997.1FFStrike slip1.093293.5721Manjil, IranAbbar19907.4FFStrike slip0.715723.9522Hector MineHector19997.1FFStrike slip0.42272623Gazli, USSRKarakyr19766.8NFReverse1.113259.5924Imperial Valley-06Bojoti Corner19796.5NFStrike slip0.981223.0325Imperial Valley-06Chihuahua19796.5NFStrike slip0.371242.0526Nahanni, CanadaSite 119856.8NFReverse1.634605.0427Nahanni, CanadaSite 219856.8NFReverse0.629605.0428Loma PrietaBRAN19896.9NFReverse Oblique0.679476.5429Loma PrietaCorralitos19896.9NFReverse Oblique0.805462.2430Erzincan, TurkeyErzincan19926.7NFStrike slip0.629352.0531Cape MendocinoCape Mendocino19927NFReverse1.819567.7832Northridge-01LA-Sepulveda VA Hospital19946.7NFReverse1.198380.0633Northridge-01Northridge-17645 Saticoy St19946.7NFReverse0.572280.8634Kocaeli, TurkeyYarimca19997.5NFStrike slip0.39429735Chi-Chi, TaiwanTCU06719996.2NFReverse0.269433.6336Chi-Chi, TaiwanTCU08419996.2NFReverse0.154665.237Denali, AlaskaTAPS Pump Station #1020027.9NFStrike slip0.446329.438Imperial Valley-06El Centro Array #619796.5NFPStrike slip0.634203.2239Imperial Valley-06El Centro Array #719796.5NFPStrike slip0.580210.5140Irpinia, Italy-01Sturno (STN)19806.9NFPNormal0.39338241Superstition Hills-02Parachute Test Site19876.5NFPStrike slip0.578348.6942Loma PrietaSaratoga-Aloha Ave19896.9NFPReverse Oblique0.609380.8943Cape MendocinoPetrolia19927NFPReverse0.887422.1744LandersLucerne19927.3NFPStrike slip1.071136945Northridge-01Rinnaldi Receiving Sta19946.7NFPReverse0.994282.2546Northridge-01Sylmar-Olive View Med FF19946.7NFPReverse1.038440.5447Kocaeli, TurkeyIzmir19997.5NFPStrike slip0.28381148Chi-Chi, TaiwanTCU06519996.2NFPReverse0.450305.8549Chi-Chi, TaiwanTCU10219996.2NFPReverse0.077714.2750Duzce, TurkeyDuzce19997.1NFPStrike slip0.655281.86 Properties of selected ground motions51. For the NTHAs, the selected records were scaled in accordance with ASCE 7–1663. A single scale factor was determined for each record based on an equivalent horizontal response spectrum obtained by combining the two horizontal components using an SRSS-based definition. The scaling was performed such that the mean equivalent horizontal spectrum of at least 11 scaled records meets or exceeds the target design spectrum of the code over the period range of 0.2 T to 2.0 T (where T is the period of the main mode). The obtained scale factor was then applied identically to both horizontal components, which were subsequently imposed simultaneously along the longitudinal and transverse directions of the bridge. The compatibility between the scaled ground motion spectra and the target design spectrum within this period range is illustrated in Fig. 6, where the mean spectrum of the scaled records closely follows the target spectrum and the individual spectra collectively envelope it, confirming compliance with code-based spectral matching requirements. Target design response spectrum and response spectra of scaled ground motions. In this study, the nonlinear analyses explicitly accounted for concurrent two-component horizontal input motions; this procedure was consistently adopted for conducting IDA across all bridge models investigated. For the IDA, the same set of NF and FF ground motions was employed following the FEMA P-69551 procedure. The earthquake intensity was parameterized using the first-mode spectral acceleration, Sa (T1), where T1 denotes the fundamental period of the bridge system obtained from the three-dimensional modal analysis of the corresponding model, and each record was incrementally scaled to multiple intensity levels. At each IDA step, the same scale factor was applied to both horizontal components, which were then imposed simultaneously along the longitudinal and transverse directions of the bridge. The corresponding structural responses were traced until the onset of severe nonlinear behavior or numerical instability, enabling construction of IDA curves describing the progression of response from elastic behavior to highly nonlinear and near-collapse states. Fragility curve Seismic fragility analysis was employed to quantify the probability of damage in the studied bridge structures under increasing earthquake intensity levels. Fragility curves provide a probabilistic framework to evaluate structural vulnerability by expressing the likelihood that the structural demand (D) exceeds its capacity (C) for a given ground motion intensity measure (IM)64. The fundamental definition of seismic fragility is expressed as: where Pf denotes the probability of reaching or exceeding a specified damage state conditioned on the seismic intensity measure. In performance-based seismic assessment, fragility curves, limit states, and damage indices are closely interrelated. Limit states represent discrete levels of structural performance degradation and are commonly classified as slight, moderate, extensive, and complete damage. These states correspond to progressively increasing levels of physical damage and loss of functionality in bridge components and the overall structural system. In RC bridge structures, various engineering demand parameters have been employed as damage indices, including lateral displacement, drift ratio, curvature ductility, rotational ductility, and energy dissipation capacity65,66. Among these, displacement- and drift-based indices are widely adopted due to their clear physical interpretation and strong correlation with observed damage in bridge piers. Numerous experimental and numerical studies have demonstrated that RC bridge piers govern the global seismic response, making pier-level demand parameters particularly suitable for fragility assessment14,52,67–69. In this study, pier drift ratio was adopted as the engineering demand parameter (EDP) for fragility assessment. For each pier, drift ratios in the longitudinal and transverse directions were computed from the relative pier-top to pier-base displacements normalized by the effective pier height. Under simultaneous two-component horizontal excitation, the directional drift ratios were combined at each time step using the SRSS measure, and the peak SRSS drift ratio was extracted for each pier. The governing EDP for a given analysis was defined as the maximum of these peak values across all piers. Damage state thresholds were established using the median SRSS drift-ratio capacities reported by Mackie et al52. for the corresponding limit states. These limits are characterized by a median drift ratio and an associated lognormal standard deviation and are summarized in Table 6. Table 6Maximum column drift damage states for RC bridges52.Damage limit stateDescriptionMedian drift ratio(%)Lognormal standard deviation MinorNegligible damage with initial cracking0.230.30ModerateCover concrete spalling1.640.33SevereLongitudinal reinforcing bar buckling6.090.25CompleteColumn failure6.720.35 Maximum column drift damage states for RC bridges52. Damage states consistent with prior RC bridge fragility studies and code-based performance descriptions were adopted70. IDA was conducted to obtain peak drift demands over increasing intensity levels, and fragility curves for each damage state were then developed by relating the selected intensity measure to exceedance of the corresponding drift thresholds, thereby accounting for record-to-record variability and nonlinear response. Model validation and comparison The accuracy of the nonlinear dynamic finite element analysis (FEA) framework implemented in OpenSees24was verified using shaking-table experiment data from a scaled CRCR bridge introduced in Zong et al71.. To reproduce the laboratory configuration, a two-span numerical model consistent in geometry with the physical specimen was constructed. The material parameters for concrete and reinforcing steel were defined using experimentally obtained data from Zong et al71., and the scaled specimen was formulated under similitude theory so that its geometry, material behavior, and dynamic characteristics matched those of prototype CRCR bridges. Within this modeling context, comparative evaluations were performed between test results and finite element analysis predictions, examining modal characteristics as well as peak displacement and acceleration responses. The experimental program involved a scaled CRCR bridge model composed of a continuous box girder supported by three RC piers. Figure 7 presents the geometry, sectional dimensions, and reinforcement arrangement of the specimen, showing the full bridge layout, the pier section details, and the variation between the largest and smallest main-girder section depths. The bridge model was rigidly mounted on a three-table shaking system and instrumented with a comprehensive measurement layout comprising 31 accelerometers, 76 strain gauges, and three laser displacement sensors. Two earthquake records, namely the El-Centro (EC) and Jinjiang (JJ) ground motions (Fig. 8), were applied under different PGA levels, as summarized in Table 7. It should be noted that the EC and JJ ground motions shown in Fig. 8are the exact input motions used in the experimental program by Zong et al71.. These records were intentionally scaled to lower peak accelerations in the original tests to accommodate the limitations of the shaking table facilities and to ensure the 1:15 scaled model remained within operational safety limits. Consequently, the EC record presented here is a scaled version of the classical record and not the original full-intensity motion. This scaling does not affect the validity of the numerical model validation, as the primary objective is to compare the numerical predictions with experimental measurements under identical input excitations. Geometric characteristics and dimensions of the scaled bridge model: (a) schematic view of the two-span prototype bridge (unit: cm); (b) pier cross-sections and reinforcement details (unit: cm); and (c,d) sections corresponding to the maximum and minimum depths of the main girder (unit: cm)71. Acceleration input time-history applied in the analysis: (a) EC wave; and (b) JJ wave71. Table 7Set of test cases considered71.Case No.Earthquake waveInput peak acceleration (m/s2)XY1EC0.50.02EC1.00.03EC0.00.54EC0.01.05JJ1.00.0 Set of test cases considered71. Under seismic loading, bridge piers are expected to exhibit nonlinear elasto-plastic behavior, particularly under moderate to strong ground motions. To capture this response, the piers were modeled using nonlinear beam-column fiber elements, which allow for the explicit representation of distributed plasticity along the member length through fiber-section discretization. This modeling strategy allows the representation of stiffness loss and energy dissipation mechanisms typically observed in RC under cyclic loading. In the CRCR bridge model, the pier–girder joints were modeled using Rigid Link Beam elements to ensure proper transmission of internal forces and bending moments without introducing unintended flexibility at the interface. The concrete constitutive response was represented using the uniaxial Concrete02 material model23,72, which accounts for nonlinear behavior in compression together with linear softening in tension. To reflect the effect of transverse reinforcement, confined and unconfined concrete fibers were modeled separately61,62. The model parameters include compressive strength (), strain at peak compressive stress (), crushing strength (), strain at crushing strength (), tensile strength () and tension softening stiffness (), as summarized in Table 8. Consistent with previous experimental and numerical studies, a value corresponding to 20% of the maximum compressive strength was used for the crushing strength59,73. The reinforcing steel bars were represented using the Steel02 material model, derived from the Giuffré–Pinto formulation together with Menegotto74 hysteresis rules, which allows a gradual transition from elastic to plastic response and accounts for post-yield hardening effects23. The adopted steel model parameters, including the post-yield stiffness ratio and transition control factor, are listed in Table 859,73. Table 8Constitutive parameters used in nonlinear FE modeling of concrete and reinforcing steel.(a) Concrete02Concrete region (MPa) (Mpa) (Mpa) (Mpa)Unconfined concrete34.60.00206.90.00353.1950Confined concrete37.70.00257.50.01903.2950(b) Steel02ParameterSymbolValuePost-yield stiffness ratio 0.044Elastic-plastic transition coefficient 15.5 Constitutive parameters used in nonlinear FE modeling of concrete and reinforcing steel. As an initial validation step, the modal properties of the FEA model were examined by comparing the first five natural frequencies with the experimentally identified values. The comparison presented in Table 9 shows that the relative errors for all considered vibration modes remain below 4%, indicating that the FEA model accurately captures the global stiffness and mass distribution of the scaled bridge specimen. This agreement establishes confidence in the numerical representation prior to time-domain response comparisons. Table 9Natural frequency of the bridge model.Mode no.Vibration mode descriptionTest frequency (Hz)FEA frequency (Hz)Error (%)1First longitudinal8.1158.0370.962First transverse15.01215.1220.733First vertical bending21.23521.5591.524Second vertical bending29.80328.6403.905Second transverse34.22633.3412.58 Natural frequency of the bridge model. The numerical model was further assessed by comparing peak pier-top displacements obtained from nonlinear time-history FEA with the corresponding experimental measurements. Figure 9 illustrates the experimental and FEA peak displacement responses for different pier locations and loading cases, while the quantitative results are summarized in Table 10. The FEA predictions closely follow the experimental trends and magnitudes, with relative errors ranging from 2.8% to 9.2%. In the comparative analyses, longitudinal displacement is identified as D1, transverse displacement at the pier tops is defined as D2, and transverse displacement at the top of the central pier is designated as D3. Experimental and numerical peak pier-top displacements. Table 10Comparison of peak pier-top displacements.Case no.PGA (m/s2)DirectionTest (mm)FEA (mm)Error (%)10.50D11.201.099.1721.00D12.202.056.8230.50D22.802.924.2830.50D32.892.993.4641.00D22.102.257.1441.00D32.602.776.5451.00D12.522.452.78 Comparison of peak pier-top displacements. Peak accelerations at the pier tops were additionally employed as a primary validation metric to assess the dynamic amplification behavior of the finite element model. Figure 10 compares the experimental peak acceleration responses with those predicted by the FEA, and the corresponding numerical values are reported in Table 11. The FEA results capture the experimentally observed amplification at pier tops, with relative errors ranging from 2.5% to 8.2%. Experimental and numerical peak pier-top accelerations. Table 11Comparison of peak pier-top accelerations.Case no.PGA (m/s²)Pier no.DirectionTest (m/s²)FEA (m/s²)Error (%)10.501X0.610.593.2821.002X1.581.542.5330.502Y0.920.866.5241.002Y1.491.424.7051.002X1.581.458.23 Comparison of peak pier-top accelerations. Overall, the comparisons between experimental measurements and numerical predictions demonstrate a good level of agreement across all considered response metrics, including natural frequencies, peak pier-top displacements, and peak pier-top accelerations. The observed discrepancies are predominantly below 10%, which is generally accepted for nonlinear dynamic analyses of RC bridge structures under seismic loading. This level of accuracy confirms that the adopted numerical modeling strategy is capable of reliably capturing the dominant seismic response characteristics of CRCR bridges. Consequently, the validated model provides a sound and credible basis for the subsequent nonlinear dynamic analyses and seismic fragility assessment of irregular CRCR bridge configurations presented in this study. Numerical results This section presents an evaluation of the seismic response characteristics of the studied CRCR bridge models based on NTHAs and IDAs. The numerical results provide a foundation for quantifying the evolution of pier drift demand under increasing seismic intensity and for establishing a reference response against which the effects of geometric irregularities are examined in subsequent sections. The seismic response of the reference bridge with regular geometry (BM0) is first investigated to establish a baseline for comparison. The IDA results shown in Fig. 11 illustrate the relationship between pier drift demand and the first-mode spectral acceleration, Sa (T1). At low intensity levels, the structural response remains predominantly linear, exhibiting limited dispersion among ground motion records. As the seismic intensity increases, stiffness degradation and material nonlinearity progressively develop, leading to a noticeable increase in response variability associated with record-to-record uncertainty. At higher intensity levels, a rapid escalation in pier drift demand is observed, indicating the development of extensive nonlinear deformation and the onset of severe damage or collapse mechanisms. IDA curves of the case study CRCR bridge (BM0). The relatively smooth and stable trend of the IDA curves for BM0 reflects a well-balanced distribution of stiffness and strength in the regular bridge configuration. This response behavior provides a reliable benchmark for subsequent comparisons and enables a clear identification of demand amplification and fragility shifts induced by geometric irregularities in the following analyses. Effect of span length and pier height ratios Variations in span length ratio modify the longitudinal stiffness distribution and dynamic characteristics of CRCR bridges, which in turn influence seismic force demand, deformation patterns, and damage concentration in the piers. Similarly, non-uniformity in pier height introduces significant stiffness and strength irregularities in CRCR bridge systems, leading to pronounced differences in seismic demand distribution and damage localization among individual piers. The numerical results presented in this section quantify the effects of span-length asymmetry and pier-height irregularity on pier drift demand and seismic fragility under increasing earthquake intensity, with emphasis on comparative response trends among regular and irregular bridge configurations. The IDA results shown in Figs. 12, 13 and 14 demonstrate that, for all investigated configurations, the initial portion of the IDA curves exhibits an approximately linear trend corresponding to the predominantly elastic response of the bridge system. In this elastic range, the initial slope of the IDA curves is primarily governed by the global stiffness of the structure, such that stiffer bridge configurations exhibit steeper slopes. Bridges with uniform pier heights (h2/h1 = 1.0) consistently show higher initial stiffness, whereas increasing pier-height ratios result in flatter initial IDA branches, reflecting reduced elastic stiffness and increased flexibility. IDA curves of the: (a) BM1 (); (b) BM6 (); (c) BM11 (); and (d) BM16 (). IDA curves of the: (a) BM21 (); (b) BM26 (); (c) BM31 (); and (d) BM36 (). IDA curves of the: (a) BM41 (); (b) BM46 (); (c) BM51 (); and (d) BM56 (). As illustrated in Fig. 12 for bridges with pronounced span-length asymmetry (L/l = 0.5), increasing the pier height ratio from 1.0 to 3.0 leads to a substantial amplification of pier drift demand at comparable seismic intensity levels. At moderate intensity levels, the median pier drift demand for bridges with h2/h1 = 3.0 is approximately 1.5 to 2.0 times that of the regular-height configuration, indicating strong stiffness concentration and deformation localization in the shorter pier. At higher intensity levels, this effect becomes more pronounced, with collapse occurring at spectral acceleration levels that are approximately 45–55% lower than those associated with bridges with uniform pier heights. Similar response trends are observed in Fig. 13 for bridges with moderate span-length irregularity (L/l = 0.7). Although the overall seismic demand is slightly reduced compared to the L/l = 0.5 configurations, increasing the pier-height ratio still results in a marked reduction in seismic capacity. For example, bridges with h2/h1 = 3.0 exhibit collapse-level spectral accelerations that are on the order of 55% lower than those of bridges with uniform pier heights, confirming that vertical irregularity remains the dominant factor governing seismic vulnerability even when span-length asymmetry is less severe. In contrast, Fig. 14 shows that for bridges with uniform span lengths (L/l = 1.0), the detrimental effects of pier-height irregularity are partially mitigated in the elastic and early nonlinear response ranges. Nevertheless, as seismic intensity increases, significant amplification of pier drift demand is still observed for configurations with large pier-height ratios. This indicates that pier-height irregularity alone can substantially degrade seismic performance, even in the absence of longitudinal geometric irregularity, particularly at intensity levels associated with severe damage and collapse. Across all span-length configurations, increasing seismic intensity leads to a noticeable increase in dispersion of the IDA curves, reflecting enhanced record-to-record variability. This effect is especially pronounced under NF ground motions, where pulse-like characteristics induce abrupt increases in pier drift demand and accelerate the transition from moderate damage to collapse. Compared with FF records, NF excitations consistently produce higher drift demands at equivalent intensity measures, with the most severe amplification observed in bridges with large pier-height ratios and short effective stiffness paths. Overall, the results presented in Figs. 12, 13 and 14 indicate that while span-length irregularity influences the longitudinal distribution of stiffness and seismic demand, pier-height irregularity is the governing parameter controlling drift amplification, damage concentration, and collapse capacity. The combined effects of reduced elastic stiffness, increased nonlinear deformation demand, and the amplified response induced by NF ground motions lead to a pronounced increase in seismic vulnerability for CRCR bridges with highly non-uniform pier heights. Effect of skew angle of piers Pier skewness alters the load transfer mechanism and induces coupled longitudinal–transverse–torsional responses in CRCR bridges, which can significantly influence seismic demand distribution and damage progression. The IDA results presented in Fig. 15 illustrate the effect of pier skew angle on pier drift demand for bridge configurations with different pier-height ratios under increasing seismic intensity. IDA curves of the: (a) BM4 (); (b) BM9 (); (c) BM14 (); and (d) BM19 (). For all skewed configurations, the initial segments of the IDA curves exhibit an approximately linear trend corresponding to the elastic response of the bridge system. In this range, the initial slope of the IDA curves is primarily controlled by the effective lateral stiffness, which is reduced as the skew angle increases due to torsional deformation and reduced alignment between inertia forces and principal stiffness directions. Consequently, bridges with larger skew angles show slightly flatter initial IDA branches compared to non-skewed configurations, indicating reduced elastic stiffness. As seismic intensity increases, the influence of skew angle becomes more pronounced. As shown in Fig. 15, increasing the skew angle to leads to a systematic amplification of pier drift demand across all pier-height ratios. At comparable intensity levels, skewed configurations exhibit median drift demands approximately 20–30% higher than their non-skewed counterparts, with the amplification becoming more significant in bridges with larger pier-height ratios. This behavior reflects the combined effects of torsional response, uneven force distribution among piers, and localized demand concentration in the shorter pier. At higher intensity levels, skewed bridges experience an earlier onset of severe nonlinear behavior and collapse compared to non-skewed configurations. For bridges with pronounced pier-height irregularity (h2/h1 = 3.0), the collapse-level spectral acceleration under skewed conditions is reduced by approximately 25–35%, indicating a notable decrease in seismic capacity attributable to skew-induced torsional effects. In addition, NF ground motions further exacerbate the adverse impact of pier skewness. Pulse-like NF records produce abrupt increases in pier drift demand, particularly in skewed configurations, accelerating the transition from moderate damage to collapse. Compared with FF excitations, NF motions consistently result in higher drift demands at equivalent intensity measures, highlighting the increased vulnerability of skewed CRCR bridges subjected to near-fault seismic excitation. The results presented in Fig. 15 demonstrate that pier skewness plays a significant role in amplifying seismic demand and reducing collapse capacity through the introduction of torsional response components and stiffness eccentricity. Although skew angle alone exerts a less pronounced influence than pier-height irregularity, its interaction with vertical irregularity and NF ground motions leads to a substantial increase in seismic vulnerability. These findings highlight the necessity of explicitly incorporating skew effects in the seismic assessment and design of CRCR bridges. Bridge structural seismic fragility analysis This section examines the influence of geometric irregularities on the seismic fragility of CRCR bridges by comparing the probability of exceeding predefined drift-based damage states52 under increasing seismic intensity. The fragility curves presented in Figs. 16, 17, 18, 19 and 20 are derived from IDA results and express the probability of exceedance as a function of the intensity measure Sa (T1, 5%). The analysis focuses on quantifying how span-length asymmetry, pier-height irregularity, and pier skewness modify both the median seismic capacity and the shape of the fragility curves, with particular emphasis on the Severe and Complete damage states that govern collapse-related performance. Fragility curves of the case study CRCR bridge (BM0) at the specified damage states. Compare the fragility curves of the: (a) BM1 (); (b) BM6 (); (c) BM11 (); and (d) BM16 () under different damage states. Compare the fragility curves of the: (a) BM21 (); (b) BM26 (); (c) BM31 (); and (d) BM36 () under different damage states. Compare the fragility curves of the: (a) BM41 (); (b) BM46 (); (c) BM51 (); and (d) BM56 () under different damage states. Compare the fragility curves of the: (a) BM4 (); (b) BM9 (); (c) BM14 (); and (d) BM19 () under different damage states. The fragility curves of the reference bridge with regular geometry (BM0), shown in Fig. 16, provide a baseline against which the effects of geometric irregularities are assessed. For BM0, the probability of exceeding Minor and Moderate damage increases rapidly at relatively low intensity levels, whereas the exceedance of Severe and Complete damage occurs over a narrower intensity range at higher spectral accelerations. At a representative intensity level of Sa (T1) ≈ 2.0 g, the probability of exceeding Severe damage for BM0 remains moderate, while the probability of Complete damage is still relatively low. This behavior reflects a gradual damage progression and a comparatively stable post-yield response for the regular bridge configuration. The effect of combined span-length asymmetry and pier-height irregularity on fragility is illustrated in Figs. 17, 18 and 19. For bridges with pronounced span-length irregularity (L/l = 0.5), shown in Fig. 17, increasing the pier-height ratio results in a substantial leftward shift of the Severe and Complete fragility curves. At a fixed intensity level of Sa (T1) ≈ 2.0 g, the probability of exceeding Severe damage increases from approximately 30–40% for the regular-height configuration (h2/h1 = 1.0) to about 50–60% for the highly irregular configuration (h2/h1 = 3.0). A similar amplification is observed for the Complete damage state, where the exceedance probability nearly doubles at the same intensity level. These results indicate that vertical irregularity significantly accelerates the transition from Severe damage to collapse when combined with strong longitudinal asymmetry. For bridges with moderate span-length irregularity (L/l = 0.7), shown in Fig. 18, the overall exceedance probabilities are reduced compared to the L/l = 0.5 cases; however, the influence of pier-height irregularity remains pronounced. At Sa (T1) ≈ 2.5 g, bridges with h2/h1 = 3.0 exhibit probabilities of exceeding Severe damage that are approximately 20–30% higher than those of bridges with uniform pier heights. Although the onset of Complete damage occurs at higher intensities than in the more asymmetric span configurations, the fragility curves clearly demonstrate that increasing pier-height ratio consistently increases the likelihood of exceeding critical damage states. For bridges with uniform span lengths (L/l = 1.0), presented in Fig. 19, the fragility curves shift toward higher intensity levels, indicating improved seismic performance relative to span-irregular configurations. Nevertheless, even in this case, pier-height irregularity leads to a measurable increase in exceedance probability at higher damage states. At Sa (T1) ≈ 3.0 g, the probability of exceeding Complete damage for h2/h1 = 3.0 is approximately 15–25% higher than that of the regular-height bridge, confirming that vertical irregularity alone can significantly degrade collapse resistance. The effect of pier skewness on seismic fragility is illustrated in Fig. 20, which compares skewed bridge configurations () with pronounced span-length asymmetry. Skewness leads to a systematic increase in exceedance probability across all damage states, with the most pronounced effects observed for Severe and Complete damage. At a representative intensity of Sa (T1) ≈ 2.0 g, skewed bridges exhibit probabilities of exceeding Severe damage that are approximately 20–30% higher than those of corresponding non-skewed configurations. This increase is further amplified for bridges with large pier-height ratios, indicating a strong interaction between torsional response induced by skewness and deformation localization associated with vertical irregularity. Across all configurations, the fragility curves for Severe and Complete damage states exhibit broader transition regions compared to those for Minor and Moderate damage. This behavior reflects increased sensitivity to ground motion characteristics and is strongly influenced by the inclusion of NF and pulse-like ground motions in the record set. NF excitations tend to increase exceedance probabilities at a given intensity measure, particularly for irregular and skewed bridges, by inducing concentrated energy input and abrupt drift escalation. As a result, the adverse effects of geometric irregularities on fragility are consistently more pronounced in the presence of near-fault seismic excitation. Critical results comparison The fragility results presented in Figs. 21 and 22 provide a comparative assessment of the influence of pier-height irregularity on the probability of exceeding critical damage states under different geometric configurations. Unlike the previous sections, which focused on individual bridge models, this section emphasizes cross-comparisons among configurations in order to identify dominant trends and critical combinations of geometric parameters governing seismic vulnerability. Figure 21 compares the fragility response of bridges with combined span-length and pier-height irregularities by examining the Severe and Complete damage states for different span-length ratios (L/l). For each span-length ratio, the pier-height ratio is systematically increased from 1.0 to 3.0, allowing direct evaluation of the sensitivity of damage exceedance probability to vertical irregularity. The results indicate that, for all span-length ratios, increasing the pier-height ratio consistently leads to a leftward shift of the fragility curves, corresponding to higher probabilities of exceeding both Severe and Complete damage at a given intensity level. However, the magnitude of this shift strongly depends on the degree of span-length asymmetry. Comparison of different geometric configurations under two damage states: (a) Severe (); (b) Severe (); (c) Severe (); (d) Complete (); (e) Complete (); and (f) Complete (). Comparison of skewed configurations under two damage states: (a) Severe (); and (b) Complete (). For bridges with uniform or moderately irregular spans (L/l = 1.0 or 0.7), the increase in exceedance probability due to pier-height irregularity is noticeable but relatively gradual. At representative intensity levels associated with collapse-sensitive behavior, the probability of exceeding Complete damage increases by approximately 15–25% as the pier-height ratio increases from 1.0 to 3.0. In contrast, for bridges with pronounced span-length irregularity (L/l = 0.5), the same increase in pier-height ratio produces a much sharper amplification in vulnerability. In these configurations, the probability of exceeding Severe damage at a fixed intensity level increases by roughly 30–40%, while the probability of Complete damage nearly doubles. This comparison highlights a strong interaction between longitudinal asymmetry and vertical irregularity, whereby stiffness concentration and demand localization in the shorter pier significantly accelerate damage progression. Figure 22 further investigates this interaction by focusing on the most critical span-length ratio (L/l = 0.5) and introducing pier skewness of . The results clearly show that the presence of skewness amplifies the adverse effects of pier-height irregularity across both Severe and Complete damage states. As the pier-height ratio increases from 1.0 to 3.0, the fragility curves shift further toward lower intensity levels compared to the non-skewed counterparts shown in Fig. 21. At representative seismic intensities, the probability of exceeding Severe damage in skewed configurations is approximately 15–30% higher than that of corresponding non-skewed bridges, while the probability of Complete damage exhibits an even more pronounced increase. The comparison between Figs. 21 and 22 indicates that skewness acts as a vulnerability multiplier when combined with geometric irregularities. While pier-height irregularity alone significantly increases the likelihood of exceeding critical damage states, its interaction with span-length asymmetry and torsional response induced by skewness leads to a disproportionate increase in collapse-related exceedance probability. This behavior reflects the combined effects of reduced effective stiffness, torsional–translational coupling, and intensified deformation demand in the shorter and skewed piers. In addition, the collapse safety of the studied CRCR bridge configurations is quantified using the collapse margin ratio (CMR) and the adjusted collapse margin ratio (ACMR) in accordance with the FEMA P-69551 framework75. In this study, collapse is explicitly defined as the attainment of the Complete damage state, corresponding to column failure based on the pier drift limits reported in Table 6. As defined in Eq. (2), the CMR is computed as the ratio of the median collapse intensity, , obtained from IDAs, to the reference intensity level corresponding to the maximum considered earthquake, ​. To account for record-to-record variability and period-dependent spectral characteristics of the selected ground motions, the spectral shape factor (SSF) is incorporated, and the ACMR is calculated using Eq. (3). The SSF is a key parameter in the FEMA P-69551 methodology that accounts for the effect of spectral shape on collapse capacity. This factor accounts for the fact that ground motions with different spectral shapes can have different effects on structural collapse capacity, even when scaled to the same first-mode spectral acceleration. Furthermore, to explicitly evaluate the influence of geometric irregularity on seismic response and collapse performance, the ratio of the ACMR of each irregular bridge configuration to that of the corresponding regular model is employed as a comparative metric. The resulting collapse metrics for the investigated bridge models are summarized in Table 12, while Fig. 23 visualizes the normalized ACMR ratio against the fundamental period, and Fig. 24 provides a complementary view of the median spectral acceleration, Sa (T1), associated with the Severe and Collapse damage states for each bridge configuration. Table 12Summary of collapse assessment results for CRCR bridge structures.Model IDPier height ratio(h2/h1)Span length ratio(L/l)Period(sec) (g) (g)CMRSSFACMRACMR/ACMRBM0BM01.000.631.052.230.772.901.273.681.00BM11.000.501.142.150.782.761.273.510.95BM61.501.182.030.782.601.303.380.92BM112.001.201.990.792.521.303.280.89BM163.001.221.850.752.471.303.210.87BM211.000.700.992.320.743.141.253.931.07BM261.501.042.190.723.041.253.801.03BM312.001.062.140.722.971.273.771.02BM363.001.092.020.702.891.273.670.99BM411.001.000.882.590.693.751.224.581.24BM461.500.932.390.723.321.224.051.10BM512.000.962.330.763.071.253.841.04BM563.000.992.180.732.991.253.741.02BM41.000.501.481.720.652.651.373.630.98BM91.501.551.650.662.501.373.430.93BM142.001.591.620.662.451.373.360.91BM193.001.641.600.672.391.373.270.89 Summary of collapse assessment results for CRCR bridge structures. ACMR ratio of studied CRCR bridge structures. Median spectral acceleration at Severe and Collapse from IDA for all bridge configurations (all ground-motion records). The results presented in Table 12 indicate that the reference bridge (BM0) with regular geometry defines the baseline collapse safety level, with a normalized value equal to unity. Bridge configurations characterized by pronounced span-length asymmetry exhibit normalized collapse safety values consistently below the reference level, and this reduction becomes progressively more pronounced as the pier-height ratio increases. In the most unfavorable configurations, the normalized values fall well below unity, indicating a substantial loss of collapse resistance relative to the reference bridge. In contrast, bridges with uniform or moderately irregular span lengths generally maintain normalized collapse safety values close to or exceeding unity, even when vertical non-uniformity is introduced. This behavior demonstrates that longitudinal regularity significantly enhances robustness and can partially offset the adverse effects of pier-height irregularity. As illustrated in Fig. 23, the variation of normalized collapse safety with the fundamental period reveals a clear and systematic trend. Bridge configurations with shorter periods tend to exhibit normalized values greater than or approximately equal to unity, reflecting higher effective stiffness and improved resistance to collapse. Conversely, as the fundamental period increases due to combined geometric irregularities, the normalized collapse safety decreases and remains consistently below the reference level. Skewed bridge configurations are generally associated with lower normalized values than their non-skewed counterparts at similar periods, highlighting the additional reduction in collapse resistance caused by torsional response. In the following, Fig. 24 provides an intensity-based interpretation of the collapse safety trends identified in Fig. 23 by presenting the median spectral acceleration, Sa (T1), associated with the Severe and Collapse damage states for each bridge configuration. Configurations that exhibit normalized collapse safety values close to or greater than unity in Fig. 23 are consistently associated with higher median Sa (T1) thresholds in Fig. 24, indicating a delayed onset of severe damage and collapse under intensity measures that are directly relevant to the bridge’s fundamental period. In contrast, bridge models characterized by normalized values well below the reference level in Fig. 23 display markedly reduced median Sa (T1) levels, confirming that lower normalized collapse safety translates into earlier exceedance of critical limit states at lower period-dependent seismic intensities. This correspondence is particularly evident for configurations with pronounced span-length asymmetry and increasing pier-height irregularity, where reductions in normalized collapse safety are accompanied by substantial drops in the median Sa (T1) thresholds for both Severe and Collapse damage states. The effect is further amplified in skewed bridges, which plot lower in Fig. 23 and simultaneously exhibit the lowest median Sa (T1) values in Fig. 24. Together, Figs. 23 and 24 provide a consistent and complementary description of collapse vulnerability, linking reductions in normalized collapse safety to physically meaningful decreases in period-dependent seismic intensity thresholds and highlighting the compounded adverse effects of geometric irregularities on the likelihood of exceeding collapse-related limit states. Also, Fig. 25 shows the median spectral acceleration corresponding to collapse for each bridge model, separated into NF and FF records. In general, for most configurations, the median collapse spectral acceleration under NF records is lower than that for FF records, indicating a reduction in the intensity threshold required for collapse and thus an increase in vulnerability under NF excitations. This reduction in the threshold is more pronounced in irregular models; such that the same configurations that exhibited lower normalized collapse safety in Fig. 23 also show the lowest median collapse spectral acceleration values in this figure. On the other hand, Fig. 25 shows that in more regular configurations, the gap between the NF and FF median spectral acceleration values is larger, but with the intensification of geometric irregularities, this gap decreases and the values for the two fields converge; this convergence indicates that the effective stiffness degradation and concentration of deformation demand in irregular models have generally reduced the collapse capacity and increased the structure’s sensitivity to both types of records, such that the response gap between the NF and FF becomes smaller. Comparison of median collapse spectral acceleration for NF and FF ground motions. Discussion The results of this study provide a comprehensive understanding of how geometric irregularities influence the seismic response, fragility, and collapse safety of CRCR bridges. By systematically varying span-length ratio, pier-height ratio, and pier skew angle, and by evaluating both deterministic response measures and probabilistic performance indicators, the analysis reveals the governing mechanisms through which geometric irregularities amplify seismic vulnerability, particularly at collapse-related limit states. A primary finding indicates that pier-height irregularity consistently emerges as the dominant contributor to seismic vulnerability, irrespective of span configuration. While span-length asymmetry modifies the longitudinal stiffness distribution and dynamic characteristics of the bridge system, vertical non-uniformity directly governs deformation localization by concentrating stiffness and strength demands in the shorter pier. This phenomenon arises from the significant stiffness disparity between adjacent piers, where the shorter, considerably stiffer pier attracts a disproportionately higher share of the seismic lateral force. Consequently, this results in deformation concentration in the short pier, which experiences much higher curvature ductility demand and earlier plastic hinging compared to the taller, more flexible pier. This mechanism is reflected in both the IDA results, where an increased pier-height ratio leads to flatter elastic slopes and earlier nonlinear escalation, and in the fragility analysis, where Severe and Complete damage states exhibit pronounced leftward shifts. Ultimately, the acceleration of damage localization hastens stiffness degradation and strength deterioration in the short pier, leading to earlier collapse; this explains why increasing the height ratio consistently reduces seismic capacity, highlighting the critical role of pier-height irregularity in controlling post-yield behavior and stability. Span-length asymmetry functions as a vulnerability amplifier by exacerbating the adverse effects of vertical irregularity, fundamentally altering the longitudinal mass distribution and stiffness continuity of the superstructure. When spans are unequal, the dynamic characteristics of adjacent spans differ, leading to out-of-phase vibrations during seismic excitation; this asynchronous movement imposes additional deformation demands on the connecting piers, effectively amplifying drift demands and modifying the distribution of inertial forces. Consequently, bridges with pronounced longitudinal asymmetry exhibit lower effective stiffness and greater demand redistribution among piers, which accelerates damage progression when combined with height non-uniformity. This interaction is evident in fragility comparisons, where bridges with a span ratio of 0.5 show substantially higher probabilities of exceeding Severe and Complete damage states at a given intensity level than those with more regular spans. Conversely, the results demonstrate that maintaining longitudinal regularity provides a stabilizing effect, significantly improving collapse resistance even in the presence of pier-height irregularity. This finding underscores the importance of considering longitudinal stiffness balance as a key design and assessment parameter for CRCR bridges, as the asynchronous span motion compounds the deformation concentration caused by stiffness disparity. Pier skewness introduces an additional layer of complexity by creating geometric eccentricity between the center of mass and the center of stiffness, thereby inducing a coupled translational–torsional response and reducing effective lateral stiffness. Under lateral excitation, this eccentricity generates a response where lateral displacement is accompanied by superstructure rotation, increasing deformation demands on exterior piers and causing uneven force distribution among piers. As the skew angle increases, the torsional stiffness contribution becomes more significant, resulting in longer periods and amplifying drift demands. Although skewness alone has a secondary effect compared to pier-height irregularity, its interaction with longitudinal and vertical irregularities leads to a disproportionate increase in seismic vulnerability. Consequently, skewed configurations consistently exhibit higher probabilities of exceeding collapse-related damage states and lower collapse safety margins than their non-skewed counterparts at comparable geometric ratios. This behavior reflects the acceleration of nonlinear deformation and instability due to uneven force redistribution and torsional demand concentration, particularly in systems already suffering from stiffness irregularity. The inclusion of NF and pulse-like ground motions plays a critical role in shaping fragility and collapse trends. NF excitations amplify deformation demands and broaden the transition regions of fragility curves, particularly for geometrically irregular bridges. This study provides a comparative assessment of NF versus FF effects, demonstrating that NF excitations reduce the median collapse intensity by 20–50% compared to FF records. Notably, the response gap between NF and FF records narrows as geometric irregularities intensify, indicating that highly irregular bridges become increasingly sensitive to record-to-record variability, which has important implications for probabilistic performance assessment and risk-informed decision-making. From an engineering perspective, the findings of this study have direct implications for the seismic design and assessment of CRCR bridges. The results suggest that limiting pier-height irregularity should be a primary objective, particularly in bridges with unavoidable span-length asymmetry or skewness. When geometric irregularities cannot be eliminated, preserving longitudinal regularity and mitigating torsional response can significantly enhance collapse safety. Mechanistically, pier height disparity creates stiffness concentration that forces the shorter pier to attract a disproportionate share of seismic demand, while span length asymmetry induces asynchronous vibrations that amplify deformation on the overstressed pier. Furthermore, pier skewness introduces geometric eccentricity, generating coupled translational-torsional response that unevenly distributes forces. This combination of mechanisms explains why configurations with combined irregularities exhibit disproportionately higher vulnerability, highlighting the limitations of assessment approaches focused solely on lower damage levels and underscoring the need for collapse-oriented evaluation frameworks. Conclusions This study investigated the seismic performance of irregular CRCR bridges through detailed numerical modeling, NTHAs, IDAs, fragility assessment, and collapse safety evaluation. A systematic set of bridge configurations was developed by varying span-length ratio, pier-height ratio, and pier skew angle to quantify the influence of geometric irregularities on structural response, damage progression, and collapse-related performance. The key findings of the study are summarized as follows: Increasing geometric irregularity, including larger pier-height ratios, greater span-length disparity, and increased pier skewness, consistently increases the fundamental period of CRCR bridges, reflecting reduced effective stiffness and more flexible dynamic behavior. Increasing geometric irregularity, including larger pier-height ratios, greater span-length disparity, and increased pier skewness, consistently increases the fundamental period of CRCR bridges, reflecting reduced effective stiffness and more flexible dynamic behavior. When tall piers, unequal span lengths, and large skew angles combine, the bridge system becomes significantly more flexible and dynamically sensitive, exhibiting pronounced stiffness concentration, torsional interaction, and demand localization under seismic excitation. This synergistic interaction explains why concurrent irregularities result in disproportionately higher vulnerability than the sum of their individual effects. When tall piers, unequal span lengths, and large skew angles combine, the bridge system becomes significantly more flexible and dynamically sensitive, exhibiting pronounced stiffness concentration, torsional interaction, and demand localization under seismic excitation. This synergistic interaction explains why concurrent irregularities result in disproportionately higher vulnerability than the sum of their individual effects. Nonlinear analyses indicate that increasing pier-height irregularity substantially reduces seismic capacity by concentrating drift demand in the shorter pier, whereas reducing span-length disparity improves load-carrying capacity and stabilizes force redistribution, thereby lowering drift demand and enhancing overall seismic performance. Nonlinear analyses indicate that increasing pier-height irregularity substantially reduces seismic capacity by concentrating drift demand in the shorter pier, whereas reducing span-length disparity improves load-carrying capacity and stabilizes force redistribution, thereby lowering drift demand and enhancing overall seismic performance. Fragility analysis indicates that geometric irregularities primarily increase the probability of exceeding Severe and Complete damage states, while their influence on the onset of Minor damage remains relatively limited. Fragility analysis indicates that geometric irregularities primarily increase the probability of exceeding Severe and Complete damage states, while their influence on the onset of Minor damage remains relatively limited. Near-field motions reduce median collapse intensity significantly, with the response gap narrowing as irregularities intensify. Near-field motions reduce median collapse intensity significantly, with the response gap narrowing as irregularities intensify. Collapse safety evaluation shows that bridge configurations with pronounced geometric irregularities exhibit normalized collapse safety values below the reference bridge, whereas longitudinally regular configurations maintain equal or higher collapse robustness even in the presence of moderate vertical irregularity. Collapse safety evaluation shows that bridge configurations with pronounced geometric irregularities exhibit normalized collapse safety values below the reference bridge, whereas longitudinally regular configurations maintain equal or higher collapse robustness even in the presence of moderate vertical irregularity. Threshold ranges associated with notable performance deterioration are identified as span-length ratios below approximately 0.6–0.7, pier-height ratios greater than about 2.0, and skew angles exceeding roughly 30°–45°, beyond which rapid increases in damage exceedance probability and collapse vulnerability are observed. These thresholds provide practical guidance for designers to identify configurations requiring special consideration or mitigation measures. Threshold ranges associated with notable performance deterioration are identified as span-length ratios below approximately 0.6–0.7, pier-height ratios greater than about 2.0, and skew angles exceeding roughly 30°–45°, beyond which rapid increases in damage exceedance probability and collapse vulnerability are observed. These thresholds provide practical guidance for designers to identify configurations requiring special consideration or mitigation measures. Limitations Despite the comprehensive analytical framework adopted in this study, several limitations should be acknowledged. The bridge deck was modeled as elastic, consistent with the observed dominance of pier response in governing global seismic behavior; however, nonlinear deck response may influence seismic performance in other bridge typologies or detailing scenarios. Soil–structure interaction effects were not considered, and all bridge models were assumed to have fixed-base boundary conditions, which may affect the estimated dynamic characteristics and collapse capacity, particularly for soft soil sites. In addition, only horizontal ground-motion components were included, while vertical ground motions—which may be significant for near-fault earthquakes—were not explicitly considered. Finally, the results are based on a specific CRCR bridge typology and selected ranges of geometric parameters, and caution should be exercised when generalizing the findings to other bridge systems. Recommendations Based on the findings of this study, several recommendations can be made for seismic design, assessment, and future research. From a design and assessment perspective, limiting pier-height irregularity should be considered a primary objective, as vertical non-uniformity was identified as the dominant driver of seismic vulnerability. Where geometric irregularities are unavoidable, maintaining longitudinal regularity and minimizing span-length disparity can significantly enhance collapse resistance. For bridges with notable skewness, torsional effects should be explicitly accounted for through improved detailing, supplemental damping, or seismic isolation strategies. For existing bridges exhibiting severe geometric irregularities, retrofit measures such as pier strengthening, jacketing of shorter piers, and energy dissipation devices are recommended. Future research should extend the present work by incorporating nonlinear deck behavior, soil–structure interaction, vertical ground-motion components, and experimental validation to further improve fragility-based and collapse-oriented seismic performance assessment of irregular CRCR bridges. Acknowledgements The authors express their appreciation for the valuable assistance provided by the Tehran Municipality (Tehran Urban Research and Planning Center) for the present research. Author contributions Seyed Hadi Rashedi : Writing – review & editing, Validation, Investigation, Writing – original draft, Visualization, Methodology, Resources, Software, Conceptualization. Alireza Rahai : Validation, Supervision, Resources, Project administration, Methodology, Investigation, Conceptualization. Data availability The data that support the findings of this study are available from the corresponding author upon reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request.