Comfort-oriented stiffness sensitivity assessment of long-span cable-stayed bridges based on vehicle-bridge coupled vibration analysis Authors: Fangyu Wu, Yonghui Fan, Ningbo Ye, Jiang Yuan, Hong Zhou, Qinfei Dai Published: 2026 ABSTRACT AbstractBridge stiffness is a key factor governing both structural serviceability and vehicle ride comfort in long-span cable-stayed bridges. This study proposes a comfort-oriented stiffness sensitivity assessment framework rather than an inverse stiffness-identification method. The objective is not to uniquely estimate bridge stiffness from ride comfort, but to quantify how prescribed stiffness reductions in major structural components influence bridge deformation and vehicle ride comfort under controlled vehicle-bridge interaction scenarios. Using the Linyu Yangtze River Bridge as a case study, a deterministic traffic flow is established through static-effect equivalence to code-based lane loading, and the influences of stiffness reduction in the main girder, stay cables, and bridge towers on bridge deformation and vehicle ride comfort are systematically investigated. The results indicate that pavement roughness has a dominant influence on vehicle-level ride comfort, whereas global bridge deformation is governed primarily by component stiffness. Different components exhibit different sensitivity characteristics: stay cable stiffness mainly controls the overall vertical deflection of the bridge, main girder stiffness has a more pronounced influence on girder-end rotation and comfort-related vehicle responses, and tower stiffness is closely associated with stability-sensitive global deformation. The proposed framework is therefore not intended to replace conventional load-carrying capacity evaluation, or stiffness identification. Instead, it provides a supplementary vehicle-bridge-interaction response layer that helps determine whether prescribed stiffness variations are mainly reflected in structural deformation, user-perceived vibration, or stability-sensitive response amplification. This information can support component prioritization, deck-serviceability diagnosis, and serviceability-oriented maintenance decision-making for long-span cable-stayed bridges.Keywords: Vehicle-bridge coupled vibration, Cable-stayed bridge, Vehicle ride comfort, Stiffness reduction, Stiffness sensitivity assessment, Deterministic traffic flowSubject terms: Engineering, Mathematics and computing Introduction With the rapid development of transportation networks, long-span bridges have become essential components of river-crossing and sea-crossing infrastructure1,2. Structural stiffness is a fundamental parameter governing bridge safety, and serviceability3–5. It not only directly affects the dynamic response of bridges under moving loads, but also strongly influences the driving safety and ride comfort of vehicles crossing the bridge6,7. Conventional bridge design codes usually define stiffness requirements on the basis of static load effects and material strength8, with limited consideration of vehicle-bridge dynamic interaction and the associated vibration problems. Therefore, investigating the sensitivity of bridge and vehicle responses to stiffness-related parameters from the perspective of vehicle–bridge coupled vibration is of considerable importance for improving the theoretical basis of dynamic stiffness design and achieving a better balance between economy and serviceability. As a key indicator of structural performance design, bridge stiffness has been widely studied from two main perspectives. The first concerns the evaluation of stiffness evolution and its influencing factors based on structural dynamic properties9–14. For example, Gonzales et al.15 measured bridge natural frequencies in different seasons to investigate the effect of ambient temperature on structural stiffness. Soyoz et al.16 analyzed modal data monitored over five years from a concrete bridge and employed neural-network techniques to determine the temporal variation in stiffness. Erduran et al.17 investigated the effect of bearing stiffness on bridge dynamic characteristics, and showed that it significantly affects both the magnitude and location of maximum bridge acceleration. The second perspective involves the development of new technologies and algorithms for stiffness identification18–20. McGetrick et al.21 proposed an algorithm for estimating overall bridge stiffness from vehicle vibration data, using acceleration measurements collected from instrumented vehicles as input. Zhang et al.22 developed a stiffness identification method based on microwave interferometric radar and rotational influence lines, and validated its effectiveness experimentally. Anastasopoulos et al.23 analyzed strain data from a steel bridge to derive temperature-insensitive strain modes and further explored methods for determining actual bridge stiffness from these strain patterns. Overall, although substantial progress has been made in bridge stiffness research, existing studies mainly focus on stiffness identification and temporal evolution of stiffness in completed bridges. Most of them emphasize bridge responses such as acceleration and displacement, while paying relatively little attention to the respective contributions of different structural components to the overall stiffness performance of the bridge. In addition, traditional stiffness-related serviceability assessments are rarely linked directly to vehicle ride comfort under realistic vehicle-bridge interaction conditions. To address these issues, this study takes the Linyu Yangtze River Bridge as the engineering background and develops a ride-comfort-informed stiffness sensitivity assessment framework within a vehicle-bridge coupled vibration analysis. Vehicle ride comfort of on-bridge vehicles is adopted as an additional serviceability indicator, and the influence of main girder, stay cables, and towers on bridge stiffness is systematically investigated. The aim is to provide a reference for stiffness-sensitive component identification and serviceability-oriented design of similar bridges. It should be emphasized that the present study does not attempt to uniquely identify the bridge components stiffness from ride comfort measurements alone. In this paper, “stiffness” refers to the equivalent component stiffness parameters represented in the finite-element model. The proposed framework is a forward parametric sensitivity assessment method, in which prescribed stiffness-reduction factors are introduced to examine their effects on bridge deformation and vehicle ride comfort under controlled traffic, speed, and pavement roughness conditions. Therefore, ride comfort is not used as a sole proxy for bridge stiffness, but as an additional serviceability-oriented response indicator within a vehicle-bridge interaction framework. Vehicle-bridge coupling dynamic analysis model Definition and scope of stiffness in this study In this study, the stiffness reduction factor is defined as the ratio of the reduced component stiffness to its original design-state stiffness. A reduction factor of 0.8, for example, represents a 20% reduction in the corresponding component stiffness in the finite-element model. These reduction factors are not intended to represent a unique physical damage mechanism, but are used as generalized degradation-inspired variables for sensitivity assessment. To avoid overinterpreting the prescribed stiffness reduction factors as direct damage-identification results, the stiffness reduction factors adopted in this study are interpreted as equivalent component-level stiffness variation scenarios. In practical bridge engineering, the same equivalent stiffness reduction may be caused by different deterioration mechanisms, such as cracking, corrosion, cable force variation, loss of composite action, or boundary-condition changes. Therefore, the relationship between the stiffness reduction factor and actual damage state is qualitative rather than one-to-one. In this study, stiffness reduction factors close to the reference state are regarded as engineering-relevant deterioration-inspired scenarios, whereas very low stiffness factors are retained only as exploratory lower-bound cases for identifying sensitivity trends and possible threshold-like response amplification. The engineering interpretation of the adopted stiffness reduction factors is summarized in Table 1. Table 1Engineering interpretation of stiffness-reduction scenarios.Stiffness reduction factorInterpretation in this studyPossible engineering relevanceUse in this paper1.0Reference stiffness stateBaseline calibrated modelBaseline case0.9-0.8Mild equivalent stiffness reductionEarly-stage deterioration, minor cable force variation, slight degradation of composite actionEngineering-relevant sensitivity range0.7-0.6Moderate equivalent stiffness reductionMore evident deterioration requiring inspection attentionEngineering warning range0.5-0.2Severe hypothetical stiffness reductionNot representative of normal service states; correspond only to extreme deterioration, artificial lower-bound cases, or near-limit degradation scenariosExploratory lower-bound sensitivity analysis Engineering interpretation of stiffness-reduction scenarios. It should be emphasized that the purpose of Table 1 is not to establish a deterministic mapping between a specific damage mechanism and a stiffness reduction factor. Instead, it provides an engineering interpretation framework for relating the numerical sensitivity analysis to possible deterioration levels. Direct diagnosis of actual damage mechanisms would require additional information such as field monitoring data, and material tests. Spatial model for vehicle vibration analysis In vehicle-bridge coupling dynamic analysis, vehicles are typically idealized as a combination of a rigid vehicle body, suspension system, and tires24,25. Each component possesses six spatial degrees of freedom26. Since the longitudinal extension-compression vibration of vehicles traveling at a constant speed has little influence on the lateral and vertical vibrations of either the bridge or the vehicle, this degree of freedom is generally neglected. Based on d’Alembert’s principle27, spatial dynamic models were established for both two-axle and three-axle vehicles. The two-axle vehicle model is shown in Fig. 1. The vehicle body has five degrees of freedom, excluding longitudinal extension-compression, and the two axles include a total of four wheels. Each wheel is assigned two degrees of freedom in vertical and lateral directions, yielding a total of 13 degrees of freedom for the complete vehicle model. The three-axle vehicle model is shown in Fig. 2, and the complete model has a total of 17 degrees of freedom. Two-axle vehicle model. Three-axle vehicle model. Both vehicle models incorporate the stiffness and damping characteristics of tires. The corresponding parameters and definitions are summarized in Table 2. Table 2Parameters and definitions for vehicle models.ParametersDefinition Vehicle body mass. Mass moments of inertia for rolling, pitching, and yawing of the vehicle body. Mass of the i-th wheel. Lateral suspension stiffness (per side) for front, middle, and rear wheels. Vertical suspension stiffness (per side) for front, middle, and rear wheels. Lateral suspension damping (per side) for front, middle, and rear wheels. Vertical suspension damping (per side) for front, middle, and rear wheels. Lateral tire stiffness (per side) for front, middle, and rear wheels. Vertical tire stiffness (per side) for front, middle, and rear wheels. Lateral tire damping (per side) for front, middle, and rear wheels. Vertical tire damping (per side) for front, middle, and rear wheels. Longitudinal distances from front, middle, and rear wheels to vehicle mass center. Parameters and definitions for vehicle models. Bridge vibration analysis model To improve computational efficiency in the vehicle-bridge coupling analysis, the bridge is discretized as a spatial frame structure combined with lumped masses and spring-damper elements. The dynamic equilibrium equation of the bridge at any instant can be expressed as where , , and are the mass, damping, and stiffness matrix of the bridge, respectively; is the displacement vector of the bridge; and is the instantaneous vehicle loads acting on the bridge. The structural damping matrix is used to characterize energy dissipation in the bridge. It is commonly formulated from the global mass matrix and stiffness matrix . In this study, Rayleigh damping is adopted: The coefficients α and β are determined from two prescribed damping ratios corresponding to two distinct natural frequencies: where and are the i-th and j-th natural frequencies of the bridge, and and are the corresponding modal damping ratios28. In this study, the Rayleigh damping coefficients α and β are determined from two dominant natural frequencies associated with the principal vibration direction under consideration. Specifically, for vertical vehicle-bridge coupling analysis, the first two dominant vertical bending frequencies of the bridge were selected, whereas for lateral coupling analysis, the corresponding dominant lateral frequencies were used. The rationale is that the dynamic response of long-span cable-stayed bridges under moving vehicles is governed primarily by low-order global modes, which dominate deck displacement, tower-girder interaction, and vehicle ride comfort. Therefore, selecting the two modal frequencies most relevant to the investigated vibration direction ensures that the prescribed damping ratio is matched within the frequency range that contributes most to the vehicle-bridge coupled response. Vehicle-bridge coupling model When a vehicle travels on the bridge deck, the wheels are assumed to remain in continuous contact with the bridge surface. Accordingly, the vertical motion of the wheels and that of the bridge deck satisfy a set of coupled compatibility constraints. These constraints constitute the time-varying boundary conditions in the decoupled iterative solution of the vehicle and bridge equations, which are referred to as coupling conditions29,30. The coupling conditions between the vertical vibration quantities of the wheelsets in the vehicle coordinate system, the bridge response and deck irregularities in the bridge coordinate system can be expressed as: where , , and are the vertical displacement, velocity, and acceleration of the i-th wheel, respectively; , , and are the corresponding bridge displacement, velocity, and acceleration at the i-th wheel position; and , , and are the pavement roughness, its velocity, and acceleration at the same position. In the numerical implementation, the vehicle and bridge subsystems are solved in a decoupled iterative manner within each time step. Starting from the bridge response and wheel-bridge contact state obtained at the previous time step, the vehicle equations and bridge equations are solved alternately, and the wheel-bridge interaction quantities are updated iteratively until convergence is achieved. In this study, convergence at each time step is judged by the relative change in the wheel-bridge coupling response quantities between two successive iterations, including the contact-point displacement and the corresponding interaction state, being smaller than the tolerance . The time step is adopted, and the iteration is terminated once the convergence requirement is satisfied or the maximum number of iterations, set to 50, is reached. To improve convergence efficiency and numerical robustness under multi-vehicle loading conditions, the Aitken iteration acceleration strategy is employed throughout the solution process. Pavement roughness model Pavement roughness is one of the primary excitation sources in vehicle-bridge interaction analysis, and it directly affects both bridge dynamic response and vehicle ride comfort. In this study, the bridge deck irregularity is modeled using the power spectral density (PSD) function specified in ISO 8608. The one-sided PSD of the road profile is expressed as where is the PSD of the road profile elevation at spatial frequency n, is the roughness coefficient at the reference spatial frequency, and ω is the waviness exponent. In this study, the reference spatial frequency is taken as , and the waviness exponent is taken as ω =2. To investigate the influence of pavement conditions on vehicle-bridge coupled vibration, three representative pavement roughness levels were considered, namely ISO 8608 Class A, Class B, and Class C. These roughness classes correspond to the dynamic analysis cases D1, D2, and D3, respectively, in the subsequent case study. The adopted roughness coefficients were for Class A, for Class B, and for Class C. Evaluation standards for vehicle ride comfort on bridges Vehicle ride comfort refers to the ability of a vehicle to ensure that vibrations and impacts generated during operation do not cause discomfort, fatigue, or adverse health effects to occupants31. For a stationary random process, the autocorrelation function of the acceleration response can be expressed as The power spectral density of acceleration can be written as Accordingly, the root mean square (RMS) spectral value of acceleration in one-third octave bands can be calculated as where is the RMS acceleration spectral value in the jth (j = 1, 2, 3, …, 20) one-third octave band with center frequency ; and are the lower and upper cutoff frequencies of the band; and is the constant-bandwidth acceleration auto-power spectral density function. The comfort reduction boundaries (RMS acceleration values) vary with the center frequency of the one-third octave bands, exposure duration, and vibration direction, as shown in Fig. 3. These boundaries reach their minimum values within the frequency ranges to which the human body is most sensitive. For vertical vibration, this range is 4–8 Hz, while for lateral vibration, it is 1–2 Hz. Comfort reduction boundaries for vehicle vibration. (a) Vertical vibration; (b) lateral vibration. To characterize whole-body vibration in different directions using a single representative quantity, frequency-weighted vibration assessment can be performed. The weighted acceleration can be calculated as where is the RMS value of single-axis weighted acceleration, and is the weighting factor for the j-th one-third octave band. For vertical vibration (from feet to head), the weighted vibration value is evaluated according to the limit values in the 4–8 Hz frequency range shown in Fig. 3a. For horizontal vibration (from back to chest or from right to left), the evaluation is based on the limit values in the 1–2 Hz frequency range shown in Fig. 3b. In practical applications, weighted evaluation and comfort spectrum evaluation may yield different results, with the former generally being more conservative. Under the most unfavorable conditions, where the measured vibration exhibits a broadband spectrum and its one-third octave characteristics are comparable to the whole-body vibration exposure limit curve, the weighted vibration level may exceed the one-third octave level in the most sensitive frequency band, leading to overly conservative results. Under the most favorable conditions, where all vibration energy is concentrated within a single one-third octave band, the two methods give the same result. Therefore, the present study adopts the one-third octave comfort spectrum as the evaluation criterion. It should be noted that the bridge-crossing process is generally a short-duration exposure event, and the actual exposure time on the bridge may in some cases be shorter than 1 min. In the present study, the 1-minute exposure limit was adopted as a unified evaluation criterion for two reasons. First, it provides a standardized and conservative reference for comparing the effects of different stiffness parameters under a consistent comfort framework. Second, for short-duration bridge-crossing scenarios, the use of the 1-minute limit is on the safe side, since shorter exposure durations generally permit higher comfort thresholds. Validation of the vehicle-bridge coupling model Based on the vehicle dynamics model, bridge finite element model, and coupling relationships described above, a Vehicle-Bridge Dynamic Analysis Software (VBDA) was developed. This software uses the Newmark direct integration method to solve the dynamic equilibrium equations of the bridge, while the Runge-Kutta method is used to solve the vehicle equations. Benchmark example. Benchmark example. Since the benchmark example adopted from Ref32. concerns a vertical vehicle-bridge interaction problem, the present validation is carried out for the vertical coupling subproblem of the developed VBDA framework under the corresponding reduced-dimensional assumptions. Therefore, this section is intended to verify the correctness of the program implementation for the vertical vehicle-bridge interaction response, rather than to claim a full validation of all three-dimensional coupling effects considered in the subsequent case study. The benchmark example is taken from Ref32. The bridge parameters are as follows: span length , elastic modulus , flexural moment of inertia , linear mass , and Poisson’s ratio ν =0.2. To ensure consistency with Ref32., the benchmark was implemented as a reduced-dimensional vertical vehicle-bridge interaction case within the VBDA framework. Specifically, only the vertical bending response of the bridge and the corresponding vertical vibrations of the vehicle were retained, while the lateral, torsional, and other spatial coupling degrees of freedom were not activated. The vehicle parameters are as follows: , , , , , , , , . The bridge and vehicle models are shown in Fig. 4. Figure of the vehicle-bridge model in the numerical example. The computational parameters are as follows: vehicle speed , modal damping ratio ξ =0, and a travel distance of 60 m from the time the first wheel enters the bridge under self-weight equilibrium conditions, corresponding to a total analysis duration of . The time step for dynamic integration is set to . 2.Comparison of results. Comparison of results. The VBDA software was used to calculate the time history results of bridge mid-span displacement and vehicle vertical acceleration, which were then compared with the results in Ref32. The comparison is presented in Figs. 5 and 6. Comparison of vertical displacement time histories at bridge mid-span. Comparison of vehicle vertical acceleration time histories. As shown in Figs. 5 and 6, the results obtained from VBDA agree well with those of Yeong-Bin Yang for the benchmark vertical interaction case, thereby verifying the correctness of the numerical implementation of the vertical vehicle-bridge coupling model in the present framework. The validation provides a sound numerical basis for the subsequent application of VBDA to the Linyu Yangtze River Bridge in “Case study analysis” and “Sensitivity of bridge responses and vehicle ride comfort to component stiffness variations” sections. Case study analysis Project overview The Linyu Yangtze River Bridge is a three-tower cable-stayed bridge with a total length of 1080 m, as shown in Fig. 7. The tall towers adopt rigid tower-girder connections, while the short towers are equipped with vertical support systems. This is consistent with the boundary conditions in its finite element model. Meanwhile, the corresponding longitudinal and transverse restraints at the bridge abutments are represented by the simplified boundary conditions adopted in the bridge finite element model. It should be noted that the present study focuses on the global vehicle-bridge coupled dynamic response within a linearized structural analysis framework. Therefore, the nonlinear behavior of bearings and expansion joints under dynamic loading is not explicitly simulated. Layout of Linyu Yangtze River Bridge (unit: m). The girder adopts a steel-concrete composite structure. The main spans are constructed in steel, while the side spans use concrete girders, as shown in Fig. 8. The steel box-girder segment of the bridge is 744 m, with a cross-sectional width of 49 m and a girder height of 4.0 m at the centerline. Standard girder cross-section layout (unit: m). The bridge towers are centrally positioned on the deck, featuring single-column configurations in the transverse direction. The tall towers reach a total height of 184 m, while the short towers extend to 136 m. For this three-tower configuration, the central tower and the side towers play different structural roles in the global force transmission and deformation coordination of the bridge system. In particular, the central tower is more directly involved in maintaining the overall stability of the main-span system, while the side towers primarily provide boundary restraint and local stiffness coordination with the side spans. Traffic flow simulation on the bridge The characteristics of vehicles traveling on bridges, including vehicle type, axle load, headway, and traveling speed, are inherently random. For newly constructed bridges, these random parameters are usually estimated by referring to measured traffic data from similar bridges and then generating stochastic traffic flows through statistical simulation. However, such stochastic traffic simulation depends strongly on the availability and quality of field measurements, and the obtained results may vary with the selected samples and statistical assumptions. Therefore, although stochastic traffic flow is suitable for describing the probabilistic variability of actual traffic operations, it is not always the most practical approach for design-oriented parametric analysis. In highway bridge design, the lane loads specified in current codes are established through statistical analysis and equivalence treatment of extensive traffic survey data. Therefore, they can be regarded as representative design-level traffic actions. Based on this idea, the present study adopts a deterministic traffic flow derived by reversing inference from code-based lane loading according to the principle of static-effect equivalence. The purpose of this deterministic traffic flow is not to reproduce the full stochastic characteristics of real traffic at the individual-vehicle level, but rather to construct a representative, repeatable, and computationally efficient loading scenario for vehicle-bridge coupled vibration analysis and stiffness sensitivity assessment. For the Linyu Yangtze River Bridge, the vertical static displacements influence line at the mid-span of the left main span and the longitudinal displacement influence line at the tower top were first calculated under code-based lane loading, as shown in Fig. 9. Then, based on the principle of equivalent static bridge effects, the corresponding vehicle types, axle loads, vehicle spacing, and vehicle quantities were determined by reverse inference, thereby establishing the deterministic traffic flow for the subsequent coupled dynamic analysis. Displacement influence line of the Linyu Yangtze River Bridge. (a) The vertical displacement influence line at mid-span of the left span; (b) the longitudinal displacement influence line at the bridge tower top. The Highway Class I loading specified in the current Chinese bridge design code is approximately equivalent to the Vehicle-Super 20 loading. Accordingly, two equivalent two-axle vehicles with gross weights of 52.0 t and 20.6 t were selected in this study to represent the heavy and light design vehicles, respectively. Their gross weights, axle-load distributions, and wheelbase arrangements are close to those of the corresponding heavy and light vehicles in the Vehicle-Super 20 loading model, as summarized in Table 3. In the deterministic traffic flow, one 52.0 t heavy vehicle is placed in each lane, while the remaining vehicles are 20.6 t light vehicles. The inter-vehicle spacing is taken as 20 m and 40 m. These equivalent vehicles are used not only in the static-effect equivalence procedure, but also as the moving vehicles in the subsequent vehicle-bridge coupled dynamic analysis. In other words, the vehicle models adopted in “Case study analysis” and “Sensitivity of bridge responses and vehicle ride comfort to component stiffness variations” sections are the spatial two-axle vehicle models defined in “Spatial model for vehicle vibration analysis”, with their total weights, axle-load distributions, and wheelbases specified according to the equivalent vehicles listed in Table 3. Table 3Parameter comparison between equivalent two-axle vehicles and Vehicle-Super 20 loading standard.Comparison itemsThe heavy vehicleThe light vehicleVehicle-Super 20Equivalent two-axle vehiclesVehicle-Super 20Equivalent two-axle vehiclesTotal weight (ton)56.152.020.420.6Front axle load (ton)27.524.17.15.9Rear axle load (ton)28.527.913.314.7Wheelbase (m)8.88.444.2 Parameter comparison between equivalent two-axle vehicles and Vehicle-Super 20 loading standard. Because the influence lines for the vertical displacement at the mid-span and the longitudinal displacement at the tower top are similar in loading efficiency, a unified deterministic traffic flow can be adopted, as shown in Fig. 10. The resulting bridge displacement responses under this deterministic traffic flow are presented in Table 4. For the six-lane bidirectional loading case, the deterministic traffic flow yields a tower top longitudinal displacement of -0.120 m and a left-span mid-span vertical displacement of -0.320 m, while the corresponding live-load design effects are − 0.118 m and − 0.311 m, respectively. The corresponding D2/D1 ratios range from 1.01 to 1.02, indicating good agreement between the deterministic traffic flow effect and the code-based live-load design effect. Schematic diagram of vehicle distribution in deterministic traffic flow. Table 4Bridge displacements under deterministic traffic flow.Traffic loading patternDisplacement itemLive load design effect D1(m)Deterministic traffic flow effect D2(m)D2/D1Six-lane bidirectionalLongitudinal displacement at the tower top-0.118-0.1201.01Six-lane bidirectionalVertical displacement at the left span mid-span-0.311-0.3201.02 Bridge displacements under deterministic traffic flow. To further quantify this agreement, the root mean square error (RMSE) is calculated. The RMSE is defined as where and are the live load design effect and the deterministic traffic flow effect of the i-th benchmark response, respectively, and n is the number of benchmark responses. Based on the two displacement responses listed in Table 4, the RMSE is 0.0065 m. Compared with stochastic traffic flow simulation, which is more suitable for probabilistic evaluation of realistic traffic variability, the deterministic traffic flow adopted in this study is better suited to the present objective of comparative parametric analysis. Its main advantage lies in providing a standardized and repeatable loading scenario, so that the influence of stiffness variation in different structural components can be identified without being masked by random fluctuations in traffic composition. On this basis, the dynamic analysis cases listed in Table 5 were established to investigate the vibration performance of the Linyu Yangtze River Bridge under different pavement roughness conditions and vehicle speeds. In each case, all vehicles were modeled as two-axle spatial vehicle models within the framework described in “Spatial model for vehicle vibration analysis” section, and the coupled interaction between the bridge and all moving vehicles was solved using VBDA. Table 5Vehicle-bridge coupled dynamic analysis conditions.Case designationCase descriptionD0Static analysisD1Dynamic analysis with ISO 8608 Class A road roughnessD2Dynamic analysis with ISO 8608 Class B road roughnessD3Dynamic analysis with ISO 8608 Class C road roughness Vehicle-bridge coupled dynamic analysis conditions. Dynamic response of the bridge under traffic flow In this study, ANSYS APDL was used to establish the finite element model of the bridge and to perform the corresponding static and modal analyses. The VBDA software can directly import the finite element model from ANSYS APDL, generate the deterministic traffic flow, and conduct the vehicle-bridge coupled dynamic analysis. Modal analysis Modal analysis A finite element model of the Linyu Yangtze River Bridge was developed using ANSYS APDL software. The towers and girders were modeled using BEAM188 elements, while the stay cables were simulated using LINK10 elements. Modal analysis was conducted to determine the first ten natural frequencies and corresponding mode shapes of the bridge structure, with results presented in Table 6. Table 6Modal analysis results of Linyu Yangtze River Bridge.ModeFrequency(Hz)Mode shape description10.2110Lateral bending of tall tower20.2555Antisymmetric vertical bending of main girder30.3711Lateral bending of short tower40.3713Lateral bending of short tower50.4382Antisymmetric lateral bending of main girder60.4481Symmetric vertical bending of main girder70.5446Antisymmetric vertical bending of main girder80.6058Symmetric lateral bending of main girder90.6282Symmetric vertical bending of main girder100.7791Torsional mode of main girder Modal analysis results of Linyu Yangtze River Bridge. (2)Dynamic response analysis Dynamic response analysis The time-history curves of the mid-span displacement of the left main span under different deterministic traffic flow scenarios are shown in Figs. 11 and 12. The three pavement conditions considered correspond to ISO 8608 Class A, Class B, and Class C roughness levels defined in “Pavement roughness model” section. Time history of vertical displacement at the mid-span of the left main span under deterministic traffic flow with different speeds: (a) 40 km/h; (b) 60 km/h; (c) 80 km/h; (d) 100 km/h. Time history of lateral displacement at the mid-span of the left main span under deterministic traffic flow with different velocities: (a) 40 km/h; (b) 60 km/h; (c) 80 km/h; (d) 100 km/h. A comparison of Figs. 11 and 12 shows that, at the same vehicle speed, increasing pavement roughness level from ISO 8608 Class A to Class C has a more pronounced effect on the lateral displacement at the bridge mid-span than on the vertical displacement. In addition, the maximum displacement and sectional internal forces at critical bridge locations under deterministic traffic flow are examined, and the results are shown in Figs. 13 and 14. Displacement results in critical bridge locations under deterministic traffic flow. (a) Longitudinal displacement at tower top; (b) vertical displacement at mid-span of left span of main girder; (c) lateral displacement at mid-span of left span of main girder; (d) torsional displacement at mid-span of left span of main girder. Internal force results in critical bridge locations under deterministic traffic flow. (a) Bending moment at mid-span of left span of girder; (b) bending moment at tower base; (c) cable tension of outermost cable in sector 2; (d) cable tension of outermost cable in sector 3. The results indicate that the dynamic response of the bridge is closely related to its natural frequencies. The moving vehicle loads act as an excitation with generalized deflection frequencies. When the deflection frequency approaches a certain vertical natural frequency of the bridge, a first-type resonance condition may occur, leading to pronounced bridge vibration. For the Linyu Yangtze River Bridge, the first order generalized deflection frequencies at vehicle speeds of 80 km/h and 100 km/h are 0.250 Hz and 0.431 Hz, respectively. These values are close to the first-order vertical bending fundamental frequency (0.2555 Hz) and second-order vertical bending frequency (0.4481 Hz), respectively. The first-order vertical bending mode exhibits an anti-symmetric configuration, while the second-order mode displays a symmetric configuration. This explains why the dynamic vertical displacement response of the main girder at 100 km/h is smaller than that at 80 km/h. For the girder vertical displacement, girder-end rotation, and tower top longitudinal displacement, the influence of pavement roughness is negligible and can be essentially disregarded. It should be noted that the above response characteristics are consistent with the Rayleigh damping parameters selected in this study. The dynamic response of the Linyu Yangtze River Bridge under moving vehicles is mainly governed by low-order global modes, as confirmed by the modal analysis in Table 6 and the dynamic response results discussed above. In the Rayleigh damping model, the modal damping ratio varies with frequency, so calibrating the damping coefficients from low-order frequencies tends to produce relatively larger damping in the higher-frequency range. For long-span cable-stayed bridges, this helps suppress spurious local high-frequency response components and numerical noise while preserving the dominant low-frequency global response relevant to bridge displacement and vehicle ride comfort evaluation. Dynamic response of the vehicles under traffic flow Theoretical and numerical analyses indicate that the most pronounced vertical vehicle vibrations occur when the deterministic traffic flow travels at 80 km/h. Light vehicles 23, 26, and 27, as well as heavy vehicles 3, 9, and 15, were selected to investigate their comfort curves under different pavement roughness conditions, where the roughness levels correspond to ISO 8608 Class A, Class B, and Class C, as defined in “Vehicle-bridge coupling model” section, and the results are shown in Fig. 15. Comfort curves for different vehicles under varying pavement roughness conditions. (a) Vehicle 23; (b) vehicle 26; (c) vehicle 27;(d) vehicle 3; (e) vehicle 9; (f) vehicle 15. The results show that, when vehicles 23, 26, and 27 travel at 80 km/h on Class C road surface, their vertical vibration comfort exceeds the 1-minute limit. On a Class B road surface, all three vehicles exceed the 25-minute limit but remain below the 1-minute limit. In contrast, when vehicles 3, 9, and 15 travel at 80 km/h on a Class C road surface, their vertical vibration comfort exceeds the 25-minute limit but remains below the 1-minute limit. On a Class B road surface, all of them exceed the 1-hour limit but remain below the 25-minute limit. These results also show that, under the highway traffic scenario considered in this study, light vehicles exhibit more pronounced vibration responses than heavy vehicles. Therefore, for comfort-oriented assessment of highway cable-stayed bridges, the comfort indices of light vehicles may be taken as the governing indicators. However, this conclusion is scenario-dependent and should not be generalized to industrial heavy-haul bridges or special heavy-load traffic conditions. In such cases, heavy-vehicle running safety may be more critical than light-vehicle ride comfort in governing the required bridge stiffness. The 25-minute and 1-hour limits are discussed here only to indicate the relative severity of vibration exposure, whereas the 1 min limit remains the governing criterion for comparative comfort assessment in this study. Sensitivity of bridge responses and vehicle ride comfort to component stiffness variations Sensitivity analysis of vertical component stiffness To investigate the sensitivity of bridge responses and vehicle ride comfort to component-level stiffness variations, the deterministic traffic flow described in “Case study analysis” section was adopted as the representative highway vehicle loading scenario. The analysis was conducted by systematically reducing the equivalent stiffness of the main girder, stay cables, towers, and the combined girder-tower systems. The stiffness reduction factors adopted in this section should be interpreted according to the engineering framework defined in “Definition and scope of stiffness in this study” section. Specifically, stiffness reduction factors from 1.0 to 0.6 are mainly used to represent engineering-relevant deterioration-inspired scenarios, ranging from mild to moderate equivalent stiffness degradation. In contrast, the lower stiffness factors from 0.5 to 0.2 are retained as exploratory lower-bound cases. These severe cases are not intended to represent routine bridge service conditions, but are used to reveal sensitivity trends, response amplification, and possible stability-sensitive behavior of the vehicle-bridge coupled system. Following the analysis framework established above, the bridge responses considered herein include the vertical displacement at the mid-span of the left main span, the vertical rotational angle at the girder end, and the one-third-octave band comfort spectrum of highway vehicles. The bridge responses and comfort evaluation results obtained by varying the vertical stiffness reduction factor of the main girder are shown in Table 7. Table 7Variation of vehicle vibration comfort evaluation indices with girder stiffness reduction.Stiffness reduction factorGirder-end rotation angle (‰)Mid-span displacement (m)Comfort requirement compliance (1 min)1.00.1137− 4.72 × 10−1Yes0.90.1340− 4.73 × 10−1Yes0.80.1420− 4.98 × 10−1Yes0.70.1614− 5.18 × 10−1Yes0.60.1878− 5.37 × 10−1Yes0.50.2170− 5.53 × 10−1Yes0.40.2785− 5.69 × 10−1Yes0.30.3987− 6.14 × 10−1Yes0.20.6315− 6.61 × 10−1Yes Variation of vehicle vibration comfort evaluation indices with girder stiffness reduction. Based on the results in Table 7, as the vertical stiffness reduction factor of the main girder decreases, the girder-end rotation angle increases from 0.1137 to 0.6315‰. When the stiffness is reduced to 20% of its initial value, the rotation angle increases by 455%, indicating that the main girder stiffness is highly sensitive to girder-end rotation. The absolute value of mid-span displacement increases more moderately with decreasing girder stiffness. When the stiffness is reduced to 50% of its initial value, the displacement increases by only 17%, suggesting that the reduction in main girder vertical stiffness has a relatively minor impact on the mid-span displacement. Under all analyzed conditions, vehicle vibration comfort satisfies the 1-minute limit requirements. These findings suggest that, from the perspective of the adopted comfort criterion alone, the investigated bridge retains a substantial response tolerance within the analyzed stiffness-reduction range. However, this should not be interpreted as a direct basis for economic stiffness optimization, since serviceability, stability sensitivity, and other code-governed design requirements may become critical earlier. The bridge responses and comfort evaluation results obtained by varying the vertical stiffness reduction factor of the stay cable are shown in Table 8. Table 8 Variation of vehicle vibration comfort evaluation indices with cable stiffness reduction.Stiffness reduction factorGirder -end rotation angle (‰)Mid-span displacement(m)Comfort requirement compliance (1 min)1.00.1137− 4.72 × 10−1Yes0.90.1143− 4.80 × 10−1Yes0.80.1087− 5.09 × 10−1Yes0.70.1101− 5.25 × 10−1Yes0.60.1082− 5.64 × 10−1Yes0.50.1196− 6.25 × 10−1Yes0.40.1068− 6.87 × 10−1Yes0.30.1042− 7.83 × 10−1Yes0.20.1266− 9.57 × 10−1Yes Variation of vehicle vibration comfort evaluation indices with cable stiffness reduction. The results indicate that, as cable stiffness decreases, the girder-end rotation angle fluctuates within the range of 0.1042–0.1266‰ without exhibiting a clear monotonic trend, indicating that cable stiffness has only a limited influence on local deformation at the girder-end rotation angle. Conversely, the mid-span displacement exhibits a pronounced increase with decreasing cable stiffness. When the stiffness is reduced to 50% of its initial value, the displacement increases by 32%, and when reduced to 20%, displacement doubles, confirming that cable stiffness is the dominant factor governing the overall vertical deflection of the bridge. Vehicle ride comfort criteria are satisfied under all operational conditions within the 1-minute threshold limits. The bridge responses and comfort evaluation results obtained by varying the stiffness reduction factor of the girder-cable are shown in Table 9. Table 9Variation of vehicle vibration comfort evaluation indices with girder-cable stiffness reduction.Stiffness reduction factorGirder-end rotation angle (‰)Mid-span displacement (m)Comfort requirement compliance (1 min)1.00.1137− 4.72 × 10−1Yes0.90.1308− 4.95 × 10−1Yes0.80.1490− 5.49 × 10−1Yes0.70.1646− 5.87 × 10−1Yes0.60.1877− 6.92 × 10−1Yes0.50.2102− 7.58 × 10−1Yes0.40.2690− 8.72 × 10−1Yes0.30.3888− 1.11Yes0.20.5750− 1.48Yes Variation of vehicle vibration comfort evaluation indices with girder-cable stiffness reduction. The results show that when the girder-cable stiffness is reduced to 50% of its initial value, the mid-span displacement increases to − 0.758 m, corresponding to an increase of 61%. When the stiffness is reduced to 20%, the displacement reaches − 1.480 m, corresponding to an increase of 214%. These amplifications are substantially larger than those observed in the single-component reduction cases of the main girder alone or the cables alone, indicating a strong synergistic effect between girder stiffness and cable stiffness, whereby their simultaneous reduction produces significantly amplified deformation. In engineering terms, the synergy may be interpreted as the excess deformation amplification produced by simultaneous girder-cable stiffness degradation compared with the corresponding single-component degradation cases. Even when the girder-cable stiffness is reduced to 20% of the initial value, the one-third-octave band comfort spectrum still satisfies the 1-minute threshold limits. However, this result should be interpreted with caution. The mid-span displacement, approximately L/287 for a 425 m main span, already far exceeds conventional serviceability expectations and may induce potential vehicle safety risks. It should be noted that the very low stiffness reduction factors in this study, such as 0.3 and 0.2, mainly as extreme degradation scenarios for parametric sensitivity exploration, rather than as representative conditions for routine engineering design or normal bridge service states. When the girder-cable stiffness reduction factor decreases to 0.2, the resulting mid-span displacement reaches 1.48 m, which is already far beyond conventional serviceability expectations. Under such extreme deformation states, the bridge response may become increasingly sensitive to geometric nonlinearity and second-order effects. Since the present study is conducted within the adopted vehicle-bridge coupled numerical framework without an explicit large-deformation geometric nonlinear analysis, the results at such extremely low stiffness levels should be interpreted primarily as indicating the trend and severity of stiffness degradation, rather than as direct engineering design recommendations. The bridge responses and comfort evaluation results obtained by varying the vertical stiffness reduction factor of the tower are shown in Table 10. Table 10Variation of vehicle vibration comfort evaluation indices with reduced tower stiffness.Stiffness reduction factorGirder-end rotation angle(‰)Mid-span displacement(m)Comfort requirement compliance (1 min)1.00.1137− 4.72 × 10−1Yes0.90.1200− 4.69 × 10−1Yes0.80.1121− 4.59 × 10−1Yes0.70.1728− 5.05 × 10−1Yes0.60.1931− 5.24 × 10−1Yes0.51.1316− 1.09Yes Variation of vehicle vibration comfort evaluation indices with reduced tower stiffness. It should be noted that, for the present three-tower cable-stayed bridge, the central tower and the side towers play different roles in the overall force transmission and deformation coordination of the bridge system. Owing to its location within the main-span system, the central tower is more directly involved in maintaining the global longitudinal and lateral stability of the bridge, whereas the side towers mainly provide boundary restraint and stiffness coordination with the side spans. In this study, the tower stiffness reduction is treated in a unified parametric manner to evaluate the overall sensitivity of the tower system. Therefore, the results in Table 10 should be interpreted as reflecting the global tower system sensitivity, rather than a fully decoupled comparison between the central and side towers. The results in Table 10 show that a 30% reduction in tower vertical stiffness causes only moderate changes in structural response, with the mid-span displacement increasing by 7% and the girder-end rotation angle increasing by 52%. However, when the stiffness is reduced to 50% of its initial value, the mid-span displacement increases sharply by 131%, while the girder-end rotation angle increases by 895%, indicating a substantial deterioration in deformation control and suggesting that the bridge system may be entering a stability-sensitive response regime. Although the corresponding displacement of -1.09 m is smaller than the − 1.48 m obtained in the extreme girder-cable reduction case, the abrupt amplification observed here suggests that tower stiffness plays a more critical role in maintaining the global stability-related deformation control of the bridge system. The threshold-like response suggests that the role of tower stiffness in the present bridge is not merely proportional stiffness contribution, but also global stability control of the tower-girder-cable system. Once the tower stiffness decreases beyond a certain level, the restraint provided by the tower to the main-span system is markedly weakened, leading to a rapid amplification of deformation. From a mechanical perspective, this behavior may be associated with the increasing sensitivity of the structural system to geometric nonlinearity and second-order effects, including P-Delta-type amplification under the combined action of dead load and traffic load. Sensitivity analysis of lateral component stiffness The same deterministic traffic flow adopted in the vertical stiffness analysis was used in the present lateral stiffness investigation. By systematically reducing the lateral stiffness of the main girder, the combined girder-cable system, and the bridge towers, the effects of different structural components on vehicle ride comfort were examined. The physical interpretation of the adopted stiffness reduction factors is the same as that described in “Sensitivity analysis of vertical component stiffness”, namely, generalized degradation scenarios ranging from mild performance deterioration to severe stiffness-loss states. In the lateral stiffness analysis, the bridge response indices considered include the girder-end lateral rotation angles and the lateral displacement at the mid-span of the main span. Similar to the vertical stiffness analysis, highway vehicle ride comfort is evaluated using the one-third-octave band comfort spectrum with 1 min exposure limit as the governing criterion. The results obtained by varying the lateral stiffness reduction factor of the main girder are shown in Table 11. Table 11Variation of lateral vibration comfort evaluation indices with girder stiffness reduction.Stiffness reduction factorGirder-end rotation angle(‰)Mid-span displacement(m)Comfort requirement compliance (1 min)10.1051.16 × 10−2Yes0.80.1061.23 × 10−2Yes0.60.1121.23 × 10−2Yes0.40.1131.29 × 10−2Yes0.20.1211.38 × 10−2Yes Variation of lateral vibration comfort evaluation indices with girder stiffness reduction. The results indicate that reducing the lateral stiffness of the main girder has only a limited influence on the lateral response of the bridge. When the stiffness is reduced to 20% of its initial value, the girder-end rotation angle increases by only 0.016‰, and the mid-span displacement increases by merely 0.0022 m. These increases are substantially smaller than those observed in the corresponding vertical stiffness reduction cases. Throughout the stiffness-reduction range investigated, both response indices increase approximately linearly, indicating that the lateral stiffness of the main girder has a relatively weak effect on the global lateral deformation of the present bridge. Under all analyzed conditions, the one-third-octave band comfort criteria remain satisfied. This is mainly because the magnitude of lateral displacement remains small, less than approximately 1.4 cm even in the most unfavorable case and is therefore insufficient to induce significant vibration excitation within the frequency range to which the human body is most sensitive in the lateral direction, namely 1–2 Hz. The evaluation indices obtained by simultaneously varying the lateral stiffness reduction factors of the main girder and the cables are shown in Table 12. Table 12Variation of lateral vibration comfort evaluation indices with girder-cable stiffness reduction.Stiffness reduction factorGirder-end rotation angle (‰)Mid-span displacement(m)Comfort requirement compliance (1 min)10.1051.16 × 10−2Yes0.80.1131.49 × 10−2Yes0.60.1141.52 × 10−2Yes0.40.1161.77 × 10−2Yes0.20.1212.81 × 10−2Yes Variation of lateral vibration comfort evaluation indices with girder-cable stiffness reduction. The results show that the combined reduction in girder and cable lateral stiffness produces a more pronounced effect on bridge lateral displacement than reducing girder stiffness alone. When the girder-cable stiffness reduction factor decreases to 0.8, the mid-span displacement increases to 0.0149 m, corresponding to an increase of 28.4%, whereas under the same reduction factor in Table 11 the displacement increases by only about 6%. This difference indicates that cable lateral stiffness is a more important controlling factor for displacement than girder lateral stiffness. By contrast, the girder-end rotation angle increases only slightly, reaching 0.121‰ at the stiffness reduction factor of 0.2, which corresponds closely to the magnitude observed in Table 11. This suggests that the girder-end rotation angle is governed primarily by the girder stiffness itself, while the effect of cable lateral stiffness on this local deformation index is negligible. Similar to the previous cases, the vehicle ride comfort criterion remains satisfied for all stiffness levels analyzed. The bridge responses and comfort evaluation results obtained by varying the lateral stiffness reduction factor of the bridge towers are shown in Table 13. It should be noted that when the tower lateral stiffness reduction factor decreases to 0.4, the numerical computation terminates because of excessive bridge deformation. Table 13Variation of lateral vibration comfort evaluation indices with reduced tower stiffness.Stiffness reduction factorGirder-end rotation angle(‰)Mid-span displacement(m)Comfort requirement compliance (1 min)10.1051.16 × 10−2Yes0.80.1191.41 × 10−2Yes0.60.1212.10 × 10−2Yes Variation of lateral vibration comfort evaluation indices with reduced tower stiffness. The results show that the lateral stiffness of the towers has a much stronger influence on bridge lateral response than the lateral stiffness of either the girder alone or the girder-cable system. When the tower lateral stiffness is reduced to 0.8, the mid-span displacement increases from m to m, corresponding to an increase of 21.6%. When the stiffness is further reduced to 0.6, the displacement rises sharply to m, corresponding to an 81% increase relative to the reference state. This abrupt amplification indicates that the structural sensitivity to tower stiffness increases markedly once the stiffness reduction exceeds a certain level. The threshold-like behavior observed between the stiffness reduction factors of 0.8 and 0.6 suggests that the tower system plays a critical role in maintaining the lateral stability and deformation control of the bridge. When the tower stiffness is reduced to 0.4, the computation is forcibly terminated, which indicates that the bridge system may be approaching a stability-sensitive regime or numerical instability under such severe stiffness degradation. From a mechanical perspective, this behavior implies that once the lateral restraint provided by the towers is weakened beyond a certain threshold, the tower-girder-cable system becomes significantly more sensitive to global instability effects and second-order deformation amplification. For all successfully completed calculation cases, the comfort requirement remains satisfied. This again indicates that, within the investigated lateral stiffness-reduction range, vehicle ride comfort is not the governing design constraint. Instead, excessive bridge deformation and potential stability-related problems are more likely to control the acceptable lower bound of lateral stiffness. Interpretation and limitations of conditional stiffness sensitivity results Although the ride comfort requirement is satisfied in all successfully completed cases, this does not imply that ride-comfort-based analysis is unnecessary. A binary pass/fail comfort assessment only indicates whether the adopted comfort limit is exceeded. The more important information lies in the comparison between comfort-related vehicle responses and conventional structural response indices. For example, some stiffness-reduction cases produce substantial bridge deformation while still satisfying the comfort criterion, indicating that structural serviceability or stability-sensitive deformation may govern the acceptable stiffness lower bound. Conversely, pavement roughness may significantly deteriorate vehicle ride comfort while producing only limited changes in global bridge displacement. Therefore, ride comfort should be interpreted as a supplementary response indicator rather than as a standalone control criterion. This finding has practical engineering significance. If both structural deformation and comfort-related vehicle vibration deteriorate simultaneously, component stiffness degradation may need to be examined together with deck-surface condition. If comfort deteriorates while structural deformation remains nearly unchanged, the deck surface, expansion joints, local irregularities, and vehicle excitation conditions should be checked first. If structural deformation becomes excessive while comfort remains acceptable, conventional serviceability and stability-related indices should govern the assessment. In this way, the proposed framework provides additional information for distinguishing different sources and consequences of serviceability deterioration. The stiffness reduction factors used in this study represent equivalent component-level stiffness variations rather than specific physical damage states. In actual bridge engineering, similar equivalent stiffness reductions may result from different deterioration mechanisms, including concrete cracking, steel corrosion, cable force loss, anchorage deterioration, bearing degradation, connection damage, or changes in boundary conditions. Conversely, the same physical damage mechanism may lead to different equivalent stiffness changes depending on its location, extent, and interaction with the global structural system. For this reason, the mapping between stiffness reduction factor and actual damage state is qualitative rather than deterministic. The severe stiffness-reduction cases, especially those with reduction factors of 0.5 or lower, are mainly introduced to reveal response trends, sensitivity amplification, and possible lower-bound warning behavior. They should not be regarded as representative conditions for routine engineering design or normal bridge service states. Temperature variation is another important factor that may affect vehicle-bridge interaction responses. The present study does not explicitly consider temperature-induced stiffness variation. Instead, it focuses on the mechanical sensitivity of bridge and vehicle responses to prescribed component stiffness reductions under fixed environmental conditions. Therefore, before comfort-related vehicle responses are used for practical condition assessment, temperature effects should be separated or compensated using field monitoring data. Conclusions This study proposed a comfort-oriented stiffness sensitivity assessment framework for long-span cable-stayed bridges by integrating deterministic traffic flow simulation, vehicle-bridge coupled vibration analysis, and one-third-octave band ride comfort assessment. Taking the Linyu Yangtze River Bridge as the engineering background, the effects of stiffness reduction in the main girder, stay cables, and towers on bridge deformation and vehicle ride comfort were systematically investigated. Based on the results obtained, the following conclusions can be drawn. The deterministic traffic flow established through static-effect equivalence to the code-based lane loading reproduced the benchmark live-load effects with good agreement. The obtained D2/D1 ratios were approximately 1.01–1.02, and the RMSE value was 0.0065 m, indicating that the adopted traffic model can provide a representative and repeatable design-oriented loading scenario for subsequent vehicle-bridge coupled vibration analysis. The deterministic traffic flow established through static-effect equivalence to the code-based lane loading reproduced the benchmark live-load effects with good agreement. The obtained D2/D1 ratios were approximately 1.01–1.02, and the RMSE value was 0.0065 m, indicating that the adopted traffic model can provide a representative and repeatable design-oriented loading scenario for subsequent vehicle-bridge coupled vibration analysis. The dynamic responses of vehicles were strongly affected by pavement roughness, vehicle speed, and vehicle type, among which pavement roughness was the most influential factor for ride comfort. For bridge global displacement responses, however, the influence of pavement roughness was less pronounced than that on vehicle vibration. The dynamic responses of vehicles were strongly affected by pavement roughness, vehicle speed, and vehicle type, among which pavement roughness was the most influential factor for ride comfort. For bridge global displacement responses, however, the influence of pavement roughness was less pronounced than that on vehicle vibration. Different structural components govern different response indices. The stay cable stiffness has a stronger influence on the overall vertical deflection of the bridge, the main girder stiffness has a more pronounced effect on girder-end rotation and comfort-related vehicle responses, and the tower stiffness plays a critical role in stability-sensitive global deformation control. This indicates that component stiffness variations should not be evaluated using a single response index only. Instead, displacement, rotation, stability-sensitive deformation, and vehicle ride comfort should be considered together to identify the dominant response mode. Different structural components govern different response indices. The stay cable stiffness has a stronger influence on the overall vertical deflection of the bridge, the main girder stiffness has a more pronounced effect on girder-end rotation and comfort-related vehicle responses, and the tower stiffness plays a critical role in stability-sensitive global deformation control. This indicates that component stiffness variations should not be evaluated using a single response index only. Instead, displacement, rotation, stability-sensitive deformation, and vehicle ride comfort should be considered together to identify the dominant response mode. For the investigated bridge and controlled loading scenarios, vehicle ride comfort is generally not the governing constraint within most stiffness-reduction cases. In several severe stiffness-reduction scenarios, the comfort requirement remains satisfied, whereas structural displacement, rotation, or stability-sensitive deformation becomes critical. Therefore, ride comfort should not be used as an independent stiffness criterion. It should be regarded as a supplementary serviceability indicator that provides user-oriented vehicle-bridge-interaction information in addition to conventional structural response indices. For the investigated bridge and controlled loading scenarios, vehicle ride comfort is generally not the governing constraint within most stiffness-reduction cases. In several severe stiffness-reduction scenarios, the comfort requirement remains satisfied, whereas structural displacement, rotation, or stability-sensitive deformation becomes critical. Therefore, ride comfort should not be used as an independent stiffness criterion. It should be regarded as a supplementary serviceability indicator that provides user-oriented vehicle-bridge-interaction information in addition to conventional structural response indices. The proposed framework does not replace conventional load-carrying capacity evaluation, or stiffness identification. Its main engineering contribution is to provide a supplementary vehicle-bridge-interaction response layer within the conventional assessment framework. By comparing structural deformation indices with ride-comfort-based vehicle responses, the framework helps distinguish deformation-governed, comfort-sensitive, and stability-sensitive response characteristics. This information can support component prioritization, deck-serviceability diagnosis, and serviceability-oriented maintenance decision-making for long-span cable-stayed bridges. The proposed framework does not replace conventional load-carrying capacity evaluation, or stiffness identification. Its main engineering contribution is to provide a supplementary vehicle-bridge-interaction response layer within the conventional assessment framework. By comparing structural deformation indices with ride-comfort-based vehicle responses, the framework helps distinguish deformation-governed, comfort-sensitive, and stability-sensitive response characteristics. This information can support component prioritization, deck-serviceability diagnosis, and serviceability-oriented maintenance decision-making for long-span cable-stayed bridges. This study also has several limitations. The deterministic traffic flow adopted herein is intended as a representative and repeatable design-oriented loading scenario rather than a full probabilistic description of actual traffic operations. The stiffness reduction factors are equivalent component-level parameters and should not be interpreted as direct representations of unique physical damage mechanisms. In practical applications, the relationship between equivalent stiffness reduction and actual deterioration state should be established using field inspection, and structural health monitoring data. In addition, the present study focuses on mechanical sensitivity under fixed environmental conditions and does not explicitly consider temperature-induced stiffness variation. Although the proposed framework is applicable to similar long-span cable-stayed bridges, the quantitative conclusions obtained from the investigated bridge should not be directly generalized to bridges with substantially different span arrangements, tower-girder connection forms, cable layouts, boundary conditions, or traffic characteristics without bridge-specific recalibration. Author contributions Fangyu Wu: Methodology, Validation, Formal Analysis, Visualization, Software, Writing-original draft. Yonghui Fan: Conceptualization, Software, Validation, Writing-review & Editing. Ningbo Ye: Validation, Formal Analysis, Visualization, Software, Writing-Review & Editing, Funding Acquisition. Jiang Yuan: Funding Acquisition, Writing-Review & Editing, Validation. Hong Zhou: Funding Acquisition, Supervision, Writing-Review & Editing. Qinfei Dai: Funding Acquisition, Writing-Review & Editing. Funding This work was supported by Chongqing Natural Science Foundation General Project (Grant No. CSTB2025NSCQ-GPX0835). Data availability The datasets used and/or analyzed during the current study 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 datasets used and/or analyzed during the current study available from the corresponding author upon reasonable request.