Study on Cross Girder to Rail Bearer Connections of Plate Girder Bridges in Sri Lanka Railways

J. Civil Eng. Mater.App. 2025 (March); 9(1): 25-33 ························································································· 25 Journal of Civil Engineering and Materials Application http://jcema.comJournal home page: Received: 01 January 2025 • Revised: 24 February 2025 • Accepted: 15 March 2025 doi: 10.22034/jcema.2025.527093.1160 Study on Cross Girder to Rail Bearer Connections of Plate Girder Bridges in Sri Lanka Railways Vanushan Kunanathan *, Kamal Karunananda Department of Civil Engineering, The Open University of Sri Lanka, Sri Lanka. *Correspondence should be addressed to Vanushan Kunanathan, Department of Civil Engineering, The Open University of Sri Lanka, Sri Lanka. ; Email: [email protected] Copyright © 2025, Vanushan Kunanathan. This is an open access paper distributed under the Creative Commons Attribution License. Journal of Civil Engineering and Materials Application is published by (ISNet); Journal p-ISSN 2676-332X; Journal e-ISSN 2588-2880. 1. INTRODUCTION ri Lanka has an extensive railway network of around 1350 bridges and culverts, which serves as the backbone of the nation's transportation system. According to the bridge register of Sri Lanka Railways (formerly, Ceylon Government RailwayCGR), More than 90% bridges are made of steel, highlighting the predominant use of steel in railway bridge construction in Sri Lanka, possibly due to its high strength-to-weight ratio, components are prefabricated in factories and are quickly assembled onsite, and good fatigue resistance than concrete [1]. Now, the Sri Lanka railway network consists of 9 major lines with a total length of about 1262 km covering all 09 provinces of the country as S ABSTRACT The connection between the cross girder and the rail bearer stands as one of the most critical structural details of plate girder bridges in Sri Lanka railways. Despite numerous modifications and advancements implemented by Sri Lankan railway engineers over time, this connection remains a repeated point of failure. This paper examines the underlying causes of these failures, in a plate girder railway bridge which is located in the coastal railway line of Sri Lanka, through numerical modelling and experimental validation. A global finite element model of the bridge was developed using SAP2000 V.23 to analyze the overall structural behavior, while a detailed local finite element model of the cross girder-rail bearer connection was developed in Abaqus V.6.14 to capture localized stress distributions and the failure mechanisms. To validate the numerical findings, vibration analysis was performed using smartphone-based measurements, where natural frequencies extracted from experimental data were compared with numerical predictions. The results of the numerical simulation indicate that the connecting plate in the cross girder–rail bearer connection is the most critical component in the selected bridge. Stress Triaxiality Factor (STF) analysis revealed that the plate undergoes high triaxial tensile stress states, particularly at its central region, indicating a high vulnerability to brittle failure with void growth. These results highlight the necessity of strengthening the connecting plate of the cross girder to rail bearer connection to enhance the structural integrity and the overall durability of the plate girder bridges. Keywords: cross girder to rail bearer connection, plate girder, Railways, finite element modelling, smartphone sensor J. Civil Eng. Mater.App. 2025 (March); 9(1): 25-33 ························································································· 26 mentioned in Table 1. The coastal line of Sri Lanka Railways is about 158 km in length, from Colombo (Maradana) to Beliatte and runs on the lower part of the west coast of the country [2]. This line was the second railway line in Sri Lanka and was opened in 1894 by the British colonial government after the Colombo–Badulla Main Line [3], [4]. Currently, it consists of 61 steel bridges. Many of these bridges were constructed in the 20th century and are still in service, exceeding 50 years of life [5]. However, with increasing train loads (magnitude and frequency) and the natural effects of aging, most of the steel railway bridges have started to deteriorate their condition, and even cracks have formed in some of the bridge members. Table 1 summarizes the lengths and number of bridges along each railway line. Table 1. Details of major railway lines in Sri Lanka Line no. Line Length (km) No. of bridges 1 Mainline (Colombo–Badulla) 291 159 2 Northern line (Polgahawela–Kankesanthurai) 339 352 3 Batticaloa line (Maho–Batticaloa) 212 119 4 Puttalam line (Ragama–Puttalam) 133 76 5 Coastal line (Maradana–Matara) 158 61 6 Trincomalee line (Gal Oya–Trincomalee) 70 46 7 KV line (Colombo fort-Avissawella) 60 42 8 Matale line (Peradeniya - Matale) 34 21 9 Mannar line (Madawachchiya - Thalaimannar) 106 63 A critical element in bridge stability is the cross girder to rail bearer connection (CG-RB). The rail bearers are placed parallel to the main girders right below the rails, and span between adjacent crossgirders with typical spans from 3m to 5m. These members are assumed to be simply supported or continuous, depending on their connections to cross girder. Further, the rail bearers carry the weight of rails, fastenings, the weight of sleepers, and their self-weight. The cross-girder spans right angles to the main girders and carries the weight of rails, fastenings, weight of sleepers, weight of rail bearers, and its self-weight. The cross girders are subjected to maximum live load and impact load when both the adjacent rail bearers are loaded. In this study, a plate girder bridge in the coastal railway line was visually inspected, and defects were identified. Figure 1 illustrates the key structural components of a typical plate girder railway bridge, including the main girders, cross girders, rail bearers, sway bracings, and wind bracings. These components work together to distribute loads from train traffic and maintain the bridge's lateral and longitudinal stability. As seen in Figure 2, the connecting plate shows visible signs of excessive corrosion and material loss. This degradation is common in coastal environments and significantly weakens the structural integrity of the CG-RB connection. Frequent replacement of this component has been necessary during routine maintenance. Figure 1. Components of a plate girder railway bridge J. Civil Eng. Mater.App. 2025 (March); 9(1): 25-33 ························································································· 26 Figure 2. Corroded connecting plate removed from the bridge Due to the bridge location on the coastline, prolonged exposure to the saline atmosphere makes corrosion an unavoidable form of degradation. Figure 3 illustrates typical failure modes such as corrosion-induced section loss and cracking in the CG-RB members. These visible defects emphasize the need for stress-based finite element analysis and justify the selection of this region for both local and global modeling efforts. Figure 3. Corrosion and cracks in the members of CG-RB connection The dynamic interaction between railway loads and bridge structures induces significantly higher stress than that estimated through static analysis alone, necessitating the application of dynamic amplification factors to accurately represent service conditions [2], [6]. Stress concentrations commonly localize at critical junctions, particularly where cross girders connect to rail bearers or main girders, making these regions highly susceptible to fatigue cracking [7], [8]. The three-dimensional behavior of plate girder bridges plays a crucial role in the development of these local stresses, as demonstrated through finite element modelling and empirical measurements, which revealed that fatigue cracks frequently initiate at cross-girder to rail bearer connections [8]. Continuity of rail bearers across cross girders can significantly reduce mid-span bending moments by allowing more effective load redistribution and enabling the rail bearers to function as continuous beams. Additionally, the structural behavior of plate girders is influenced more by axle spacing than axle load magnitude, emphasizing the importance of span configuration in bridge response [9]. Studies incorporating static and dynamic testing on bridges with high axle loads have shown that a considerable portion of longitudinal forces, nearly 50% is transferred to the girders, confirming their critical role in load-bearing behavior under train-induced loading conditions [10]. Vibrations induced by moving trains, braking actions, and track irregularities affect the performance, durability, and serviceability of railway bridges, with dynamic responses being highly sensitive to parameters such as span length, damping, boundary conditions, and material stiffness [11], [12]. Resonance effects are of particular concern, as they arise when train loading frequencies coincide with the natural frequency of the bridge, potentially leading to excessive vibrations [11]. Past research studies indicate that vehicle speed and axle configuration significantly influence vibration J. Civil Eng. Mater.App. 2025 (March); 9(1): 25-33 ························································································· 26 amplitudes, while stiffer or heavier bridges tend to experience lower peak accelerations and delayed resonance responses. The use of analytical and experimental methods, especially finite element modelling validated by field measurements, remains essential for accurately predicting bridge performance under dynamic conditions and informing effective design and maintenance strategies [12], [2], [13]. In recent years, the integration of mobile technology into civil engineering practice has opened new pathways for cost-effective structural health monitoring (SHM). Smartphones, equipped with micro-electromechanical systems (MEMS) accelerometers, have emerged as practical tools for capturing structural vibration data. These devices are increasingly being used for validating finite element models (FEMs) and assessing the dynamic performance of structures under service conditions. Their portability, affordability, and capability to capture structural vibrations, particularly when used in direct contact with the structure, make them particularly suitable for preliminary investigations, field-based monitoring, and educational applications, especially in scenarios where traditional instrumentation is either unavailable or impractical [14]. While limitations such as noise interference, low sampling rates, and reduced sensitivity at high frequencies can affect data fidelity, the convenience and affordability of smartphone-based sensing make it a viable alternative in scenarios where traditional sensors are impractical or prohibitively expensive. These devices support both contact-based methods and non-contact approaches, such as video-based motion tracking, enabling diverse applications ranging from laboratory experiments to in-field monitoring [14]. Furthermore, the vehicle-bridge interaction (VBI) technique, which interprets vertical accelerations of a moving vehicle to estimate a bridge's dynamic properties, has been effectively implemented using smartphones alongside standard sensors [15]. This approach enables the estimation of natural frequencies and offers a non-intrusive means of assessing structural health. Recent studies have shown that networks of smartphones can be deployed on long-span bridges to extract modal frequencies and mode shapes with reasonable accuracy [16]. These advancements affirm the growing potential of smartphones in dynamic structural assessment, especially in applications demanding mobility, affordability, and rapid deployment. This study aims to identify the critical structural members of the bridge through numerical modeling and to validate the developed model using experimental data. Validation is intended to be achieved by conducting a smartphone-based vibration analysis of the bridge under live load conditions. 2. METHODOLOGY The methodology followed to achieve the aim of this study is mentioned in detail below. 1. Literature review of past publications on the railway bridges along the coastline and discussions with Sri Lankan Railway personnel. 2. Conduct a field visit to visually inspect and assess the current condition of the bridge. • Identify the defects and perform an elementary survey to measure the dimensions of the members. • Observe the types of locomotives in operation on this bridge and find out the weight of each locomotive. • Collecting material samples for experimental evaluation. 3. Conducting tests to identify the mechanical properties of the material used for different bridge members. • Mechanical properties are yield strength, ultimate tensile strength, and modulus of elasticity 4. Develop a global numerical model of the bridge using SAP2000 V.23 software. • Identify the most critical cross girder to rail bearer connection of the bridge 5. Validating the numerical model using smartphone-based vibration analysis. • Measure acceleration data from the bridge using a smartphone under the live load conditions. • Calculate the natural frequency from the measured acceleration data and compare it with the natural frequency obtained from the numerical model to assess its accuracy. 6. Develop a local model of the critical connection using Abaqus V.6.14 software. J. Civil Eng. Mater.App. 2025 (March); 9(1): 25-33 ························································································· 26 • Determine the stresses on the various elements at the cross girder–rail bearer connection and identify the most critical member. 7. Result analysis and interpretation. 2.1. Material Testing Beginning in the late 1990s, high-strength structural steels such as S355 have been progressively adopted as standard materials in bridge construction due to their superior mechanical properties and enhanced performance under load [17], [18]. Past research study on SLR bridges has highlighted that S355 steel has been used for bridge constructions since 2004 [19]. Material testing conducted [7] on a coastal railway bridge built during the same period with the same type of steel supports this information. To validate this, material samples were obtained from the bridge during routine maintenance. A connecting plate having a thickness of 12mm was tested using a digital Rockwell hardness tester with ASTM E18-15 standards as shown in Figure 4. The average hardness measured was HRB 61.3, indicating the steel is likely S355. Figure 4. Specimen on digital Rockwell hardness testing machine The mechanical properties of the steel used in the bridge are presented in Table 2. Table 2. Mechanical properties of S355 steel [7] Mechanical Properties Value Yield strength (MPa) 355 Ultimate tensile strength (MPa) 433 Strain at break (mm/mm) 0.36 Fracture strength (MPa) 287 2.2. Finite element modelling of the bridge (Global numerical model) The selected Panadura railway bridge is a semi through continuous plate girder bridge, consisting of a total length of 194.24 m with four equal spans and consists of various members such as main girders, cross girders, rail bearers, sway bracings, and wind bracings. It is supported on the piers using elastomeric bearings. Therefore, one end support was fixed while other supports were pin-supported considering the continuity of the bridge. (Figure 5). J. Civil Eng. Mater.App. 2025 (March); 9(1): 25-33 ························································································· 25 Figure 5. 3D bridge model developed using SAP2000 software 2.3. Assigned loads The M11 is the heaviest locomotive operating on this bridge, with a total weight of 120 tons distributed across six axles. Therefore, M11 was selected to determine the critical cross girder to rail bearer connection and the impact loads from the passenger compartments are negligible when compared with the engine. The model was run in 19 various load combinations when the train was on the bridge. Due to the dynamic effects associated with moving trains, the actual service load on the bridge would be higher than the static load. Therefore, a dynamic factor of 1.5 was applied to the static load to estimate the service load [6]. The load configuration assumes that the total weight is transmitted from the axles through the wheels and ultimately to the rail bearers. Additionally, the force exerted by each wheel was applied as a point load, as shown in Figure 6, with axle spacing provided in millimeters. Accurate representation of axle spacing and load magnitude is essential for determining critical loading scenarios and stress concentrations in bridge components. Figure 6. Load geometry of M11 locomotive Figure 7 identifies the regions experiencing the maximum stress on the bridge for different load combinations. These stress patterns were correlated with data in Table 3, which lists axial forces at the CG-RB connection. Figure 7. Maximum Von-Mises stress location on the bridge The maximum stress corresponding to each loading case and load combination was extracted and tabulated in Table 3 below. J. Civil Eng. Mater.App. 2025 (March); 9(1): 25-33 ························································································· 26 Table 3. Maximum stress on the bridge with location Location of Locomotive (m) Axial load at the CG-RB connection (kN) 1.4 Gk + 1.6 Qk Gk + 1.5 Qk Gk + Qk 0 -39.1 -7.4 -24.8 10 66.9 58.7 43.5 20 83.8 72.7 54.8 30 69.8 60.5 45.6 40 -15.3 -11.2 -10.9 50 21.6 20.7 13.3 60 40.6 36.9 25.8 70 51.1 47.2 32.2 80 -21 -18 -13.8 90 -32.2 -28.4 -20.8 100 38.3 35.4 24.2 110 70.9 65.4 44.8 120 51 47 32 130 -33.9 -29.2 -22.2 140 -46.7 -40.4 -30.5 150 33.2 31 20.8 160 105.2 93.3 68.1 170 58.2 50.1 38.1 180 58.1 48.7 38.6 It is clear from Table 3, that the maximum stress occurs when the locomotive is positioned at 160 m and the natural frequency obtained from the finite element bridge model was 0.161 Hz. Therefore, these corresponding stress values can be used to develop the local modelling of the critical CG-RB connection. 2.4. Vibration analysis In this study, vibration measurements were obtained using smartphones equipped with the ‘Phyphox’ application, which utilizes the built-in accelerometer to record acceleration data. Four measurement locations were strategically selected on the bridge based on anticipated zones of significant vibration, enabling a comprehensive assessment of the structural response during testing. The selected test locations are illustrated in Figures 8, 9 and 10. Figures 9 and 10 illustrate the physical setup of smartphones fixed on the rail bearer and cross girder, respectively. The devices were mounted in direct contact with the bridge members using high power double tape, ensuring minimal movement and accurate vibration capture during train passage. Figure 8. Plan view of test locations J. Civil Eng. Mater.App. 2025 (March); 9(1): 25-33 ························································································· 27 Figure 9. Smartphone fixed on Rail bearer Figure 10. Smartphone fixed on cross girder The test configurations are shown in Table 4 below. Table 4. Test configurations Span Locomotive class Location Member Smartphone Span 1 S-8 A Main girder iPhone 6s+ B Rail bearer 1 iPhone 8 Span 2 M-4 C Cross girder iPhone 8 D Rail bearer 2 iPhone 6s+ Measurements were recorded in three directions: x, y, and z as shown in Figures 11, 12, 13 and 14. The data for each direction were collected over time, resulting in acceleration vs. time graphs that display the variation in acceleration as the vibrations occurred. Figure 11. Acceleration vs time graph for location A J. Civil Eng. Mater.App. 2025 (March); 9(1): 25-33 ························································································· 28 Figure 12. Acceleration vs time graph for location B Figure 13. Acceleration vs time graph for location C Figure 14. Acceleration vs time graph for location D The power spectral density (PSD) was determined by applying the Fast Fourier Transform (FFT) to the acceleration data, and the PSD vs. frequency was generated in MATLAB Online software. The dominant frequency in the plot is the natural frequency of the bridge. The PSD vs frequency plots for the locations A, B, C and D are shown in Figures 15, 16, 17, and 18, respectively. Figure 15. PSD vs frequency graph for location A Figure 16. PSD vs frequency graph for location B