Parametric Nonlinear FEM Study of Load Transfer Efficiency in Jointed Plain Concrete Pavements

J. Civil Eng. Mater.App. 2025 (September); 9(3): 131-144 ························································································· 131 Journal of Civil Engineering and Materials Application http://jcema.comJournal home page: Received: 16 July 2025 • Revised: 02 September 2025 • Accepted: 24 September 2025 doi: 10.22034/jcema.2025.553798.1183 Parametric Nonlinear FEM Study of Load Transfer Efficiency in Jointed Plain Concrete Pavements Goytom Gidey Desta 1*, Ashenafi Aregawi 2, Berhane Grum 2 1 Civil Engineering, Engineering and technology college, Adigrat University, Adigrat, Ethiopia 2 School of Civil Engineering, Mekelle University, Mekelle, Ethiopia *Correspondence should be addressed to Goytom Gidey Desta, Civil Engineering, Engineering and technology college, Adigrat University, Adigrat, Ethiopia Email: [email protected] Copyright © 2025, Goytom Gidey Desta. This is an open access paper distributed under the Creative Commons Attribution License.Journal of Civil Engineering and Materials Applicationis published by (ISNet); Journal p-ISSN 2676-332X; Journal e-ISSN 2588-2880. 1. INTRODUCTION ointed Plain Concrete Pavements (JPCPs) are widely used in highway systems due to their durability and load-bearing capacity [1,2]. The performance of JPCPs depends on efficient load transfer across transverse joints, which control cracking and accommodate thermal and moistureinduced movements [3,4]. Inadequate load transfer leads to differential slab deflections, resulting in J ABSTRACT This study develops a validated three-dimensional nonlinear finite element (FE) model in ABAQUS to evaluate the influence of key design parameters—slab thickness, dowel diameter, joint width, base modulus, axle load, and dowel–concrete interaction—on load transfer efficiency (LTE) in jointed plain concrete pavements (JPCPs). The model incorporates realistic dowel–concrete interface behavior, Concrete Damage Plasticity (CDP), and multilayer base–subgrade support. Simulation results indicate that increasing slab thickness from 0.25 m to 0.35 m enhances LTE by 0.70 %, while enlarging dowel diameter from 32 mm to 38 mm increases LTE by 2.56 %. Conversely, widening the joint from 8–12 mm reduces LTE by 1.24 %. Base stiffness exhibits the strongest influence; increasing base modulus from 100 MPa to 10 000 MPa raises LTE by approximately 13.5 %. Heavier axle loads (8–15 tons) decrease LTE by about 4.12 %. Dowel– concrete friction and dowel support modulus have minimal effects (less than 1 %), indicating their limited contribution under typical design conditions. The results indicate that LTE is primarily governed by loading, slab geometry, and base stiffness, whereas interface friction and dowel support modulus have only marginal effects. The validated model demonstrates strong agreement with analytical predictions (deviation <2.5 %), ensuring reliability for mechanistic pavement design applications. Keywords: Load Transfer Efficiency (LTE); Jointed Plain Concrete Pavement (JPCP); Finite Element Method (FEM); Dowel–Concrete Interaction J. Civil Eng. Mater.App. 2025 (September); 9(3): 131-144 ························································································· 132 distresses such as faulting, corner cracking, and joint deterioration [5,6]. Transverse joints act as structural discontinuities that transfer loads between adjacent slabs through dowel action, aggregate interlock, and subbase support [7– 9]. Aggregate interlock contributes significantly when joint openings remain narrow (<0.8 mm); however, its influence decreases rapidly as the joint widens [10,11]. Hence, dowel action becomes the dominant mechanism, with parameters such as dowel diameter, spacing, alignment, and subbase stiffness strongly influencing LTE and overall pavement performance [12,9]. Researchers have identified Load Transfer Efficiency (LTE) as a key indicator of the structural integrity and serviceability of JPCPs, providing valuable insight into joint performance and the effectiveness of design features aimed at optimizing pavement longevity and ride quality [7,8,13,14]. Achieving high LTE mitigates load-related distresses by reducing stress concentrations and limiting relative slab movement [7,8]. Field observations indicate that LTE values above 70% generally correspond to satisfactory pavement performance [9,13]. Foundational analytical studies, notably Friberg’s shear-transfer theory [15], established the theoretical basis for LTE in doweled joints. These early models, grounded in classical elastic theory [16], assumed linear material behavior, idealized boundary conditions, and simplified dowel–concrete interaction. While suitable for early design applications, such simplifications limit predictive accuracy under modern traffic and environmental loading conditions [17,18]. Consequently, contemporary JPCPs subjected to heavier traffic loads and complex material–structural interactions have revealed the limitations of linear-elastic assumptions [19-21]. Recent investigations demonstrate that perfectly bonded or linear-elastic models fail to capture essential nonlinear behaviors such as dowel slip, interface separation, and stiffness degradation under repeated loading [12,15,16]. As a result, advanced nonlinear finite element (FE) frameworks have been developed to better simulate dowel–concrete contact, subbase interaction, and stress redistribution under realistic field conditions [17,22]. In summary, the performance of JPCPs is primarily governed by LTE across transverse joints, which depends on the interaction between dowels, concrete, and the subbase. The limitations of traditional analytical models emphasize the need for advanced nonlinear numerical methods capable of reproducing the multi-mechanism behavior observed in field pavements. 1.1 Numerical and Field Investigations of LTE Three-dimensional nonlinear finite element modeling (FEM) has become a widely used tool for simulating realistic joint behavior in JPCPs. These models incorporate key mechanisms such as dowel– concrete interaction, progressive concrete damage, and heterogeneous support conditions [23–26]. Field and laboratory investigations confirm that doweled and hybrid-reinforced systems can achieve LTE values exceeding 90% [3,4]. FEM analyses have identified dowel placement, joint width, interface bonding, slab thickness, and base stiffness as the principal parameters influencing LTE [15,21,27]. Higher base stiffness (≥300–350 MPa) enhances LTE and reduces slab cracking [28,29], while the use of fiber-reinforced polymer (FRP) or stainless-steel dowels has been shown to improve long-term performance [3,4,30]. Furthermore, recent advances in displacement and vibration sensing technologies allow real-time LTE evaluation consistent with traditional field-testing methods [31]. Despite these advancements, commonly used FEM tools such as ANSYS, EverFE, and DIANA still rely on embedded-contact or linear-elastic assumptions originally developed for reinforced concrete rather than JPCPs [16,17]. These simplifications overlook critical nonlinear effects—such as dowel slip, bearing deformation, and progressive concrete damage—leading to systematic overestimation of LTE [12,32]. For instance, Zang et al [12] reported a 15–20% overprediction of LTE using linear models, while Zewdie et al. [15] observed a 25% reduction in dowel shear capacity when nonlinear interfaces were considered. Huang et al. [33] found that LTE decreased from 85% to 68% under a 40 kN load due to cracking, whereas Davids et al. [21] and Bilal et al. [38] demonstrated that dowel looseness and joint widening significantly reduce LTE. Collectively, these findings highlight the necessity of nonlinear, multi-mechanism modeling to more accurately replicate field performance. 1.2. Research Gap and Study Objectives Although substantial progress has been made in understanding joint behavior, the accurate prediction J. Civil Eng. Mater.App. 2025 (September); 9(3): 131-144 ························································································· 133 of LTE in JPCPs remains constrained by simplified dowel–concrete interface modeling, coarse meshing, and idealized boundary assumptions [39-41]. Classical and empirical design frameworks—such as those proposed by AASHTO and Huang [1,25]— typically employ linear-elastic or fully bonded assumptions that cannot reproduce the nonlinear load transfer and degradation observed in field conditions. Even modern FEM models often neglect localized slip, bearing deformation, and stiffness loss, thereby limiting their ability to predict actual pavement response [15–18, 42]. To address these limitations, the present study develops a nonlinear finite element framework in Abaqus, featuring explicit surface-to-surface contact between dowel and concrete, integration of the Concrete Damage Plasticity (CDP) model, and refined meshing strategies [6,43]. This modeling approach enables realistic simulation of dowel– concrete interaction, stress redistribution, and load transfer degradation. Validation against analytical and experimental benchmarks ensures the reliability and applicability of the developed framework. A systematic parametric analysis is conducted to examine the effects of slab thickness, dowel diameter, joint width, and subgrade stiffness on LTE and stress distribution. The outcomes of this analysis provide mechanistic insight into doweled joint behavior and support the performance-based design of durable concrete pavements. While the proposed model effectively captures nonlinear load transfer mechanisms, certain limitations remain. The analysis assumes static loading and uniform support, without explicitly considering thermal or moisture gradients. Field instrumentation was not incorporated, and validation relied on published datasets. Nevertheless, the developed framework establishes a robust mechanistic foundation for LTE evaluation and lays the groundwork for future studies incorporating environmental and dynamic effects. 2. MATERIALS AND METHODS 2.1 Material Properties The 3D finite element (FE) model represents two concrete slabs separated by a transverse joint, with mid-depth dowels spaced at 300 mm. Concrete (f′c = 40 MPa; ft = 3.5 MPa) was simulated using the Concrete Damage Plasticity (CDP) model [16,18], while the base and subgrade layers were assumed linear elastic, consistent with common mechanistic modeling approaches [6,16]. Seven cylindrical dowels were symmetrically distributed across the joint at 300 mm center-to-center spacing. To balance computational efficiency with stress accuracy, a slab width of 2.1 m was adopted, reflecting the estimated 0.7 m stress penetration from a corner wheel load [16,44]. Base and subgrade layers extended 0.4 m beyond the slab on each side to ensure complete stress distribution. A 10 mm joint width was provided to accommodate thermal expansion and contraction. Each dowel was embedded halfway into the loaded slab, with the remaining half interacting frictionally or sliding relative to the adjacent slab, depending on the dowel–concrete interface behavior [6,16,45]. Concrete was modeled using the nonlinear CDP model to capture both tensile cracking and compressive plasticity [16, 36]. The CDP input parameters were defined as a dilation angle (φ) of 15°, an eccentricity of 0.1, a biaxial-to-uniaxial compressive-strength ratio (Fb/Fc) of 1.16, and a tensile-to-compressive meridian ratio (k) of 0.666, representing concrete behavior under combined stress conditions [16, 46]. These represent typical concrete behavior under combined stress conditions. General material properties for all pavement layers and dowels are summarized in Table 1. The base and subgrade layers were assumed isotropic and linear elastic [47], while dowels were modeled as thermoelastic–plastic to account for pseudo-temperature effects [48]. J. Civil Eng. Mater.App. 2025 (September); 9(3): 131-144 ························································································· 134 Table 1. Material Property and Geometric Parameters (a) Concrete Slab, Base, and Subgrade) Component Length (m) Width (m) Thickness (m) Young’s Modulus, Ec (MPa) Poisson’s Ratio Density (kg/m³) CTE (°C⁻¹) Slab 4.6 2.1 0.25 29.7 × 10³ 0.19 2400 1.08×10⁻⁵ Base 10.01 2.9 0.20 310 0.3 2100 — Subgrade 10.01 2.9 0.25 30.3 0.4 2040 — (b) Steel Dowel Bars) Component Length (m) Diameter (m) Yield Stress (MPa) Young’s Modulus, Ec (MPa) Poisson’s Ratio Density (kg/m³) CTE (°C⁻¹) Dowel 0.5 0.032 345 200 × 10³ 0.3 7830 1.22×10⁻⁵ Notes: CTE = Coefficient of Thermal Expansion. SI units are used. Table 2. CDP Model Parameters for Concrete Compression and Tension (a) Concrete Compression Behavior) Stress (Pa) Crushing Strain Damage (dc) Crushing Strain (for dc) 1.50 × 10⁷ 0.0000000000 0.0000000 0.0000000000 2.02 × 10⁷ 0.0000747307 0.0000000 0.0000747307 3.00 × 10⁷ 0.0000988479 0.0000000 0.0000988479 4.03 × 10⁷ 0.0001541230 0.0000000 0.0001541230 5.00 × 10⁷ 0.0007615380 0.0000000 0.0007615380 4.02 × 10⁷ 0.0025575590 0.1954020 0.0025575590 2.02 × 10⁷ 0.0056754310 0.5963820 0.0056754310 5.28 × 10⁶ 0.0117331190 0.8948650 0.0117331190 (b) Concrete Tension Behavior) Stress (Pa) Cracking Strain Damage (d_t) Cracking Strain (for d_t) 1.99893 × 10⁶ 0.0000000000 0.0000000 0.0000000000 2.84200 × 10⁶ 0.0000333300 0.0000000 0.0000333300 1.86981 × 10⁶ 0.0001604270 0.4064110 0.0001604270 8.62723 × 10⁵ 0.0002797630 0.6963500 0.0002797630 2.26254 × 10⁵ 0.0006845900 0.9203890 0.0006845900 5.65760 × 10⁴ 0.0010867300 0.9800930 0.0010867300 Note: These define nonlinear concrete response under compression and tension. J. Civil Eng. Mater.App. 2025 (September); 9(3): 131-144 ························································································· 135 2.2. Loading Conditions A standard single-axle dual-wheel (SADW) system with an 80 kN total load (two 20 kN wheels spaced 300 mm apart) was applied, positioning the outer wheel above the corner dowel and 10 mm from the joint edge [15, 49], as illustrated in Figure 1. Figure 1. Abaqus solid elements (1) and corresponding load transfer distribution under dual-wheel loading (2). FE model representation of the rectangular loading area (3). A uniform pressure of 830 kPa was applied over a rectangular contact area of 0.187 × 0.129 m to simulate realistic wheel–slab interaction [15, 30]. A minor pseudo-temperature increase (285–290 K) was introduced to induce initial contact pressure between dowels and concrete [50]. 2.3 Finite Element Modeling 2.3.1 Element Type A three-dimensional solid finite-element model was developed using first-order, reduced-integration C3D8R brick elements with hourglass control in ABAQUS [37]. Solid continuum elements were selected to capture complex stress and displacement fields more accurately than truss, beam, or shell elements [37]. Reduced integration minimized computational cost and mitigated volumetric or shear locking in bending-dominated regions [37]. 2.3.2 Interface Modeling The slab–base interface was modeled with hard contact and a friction coefficient of 0.95, while the base–subgrade interface was tied to enforce identical displacements [39]. A schematic of these contact interfaces is shown in Figure 2. Dowel–concrete interaction (DCI) was captured through two complementary mechanisms to realistically simulate load transfer across the joint [15, 51, 52]. 1 2 3 J. Civil Eng. Mater.App. 2025 (September); 9(3): 131-144 ························································································· 136 Figure 2. Finite element contact interfaces adopted for the analysis. Longitudinal Bond: Mechanical interlock and static friction along the dowel length were modeled using elastic springs positioned along four diametrically opposite lines of the dowel [35]. Transverse Interaction: Bearing stresses, normal contact, and Coulomb friction were represented with surface-to-surface contact elements [35]. Initial contact pressure was applied via a minor thermal expansion. For epoxy-coated dowels, a friction coefficient of 0.05 and a uniform clearance of 0.012 mm were used [35]. 2.3.3 Mesh Development Eight-node C3D8R elements were used with aspect ratios below 2. Mesh refinement was applied in critical zones, including dowel holes and load areas, to accurately capture stress gradients [53]. The meshsensitivity results presented in Table 3 confirmed less than 1.3 % variation in stress, validating numerical convergence [37]. Figure 3 and Figure 4 show the finite element mesh and the meshed dowel bar, respectively. Table 3. Mesh Sensitivity Analysis Mesh Type Number of Elements Maximum Tensile Stress (Pa) Execution Time Mesh I 27,795 2.36 × 10⁶ 8 hours Mesh II 31,873 2.39 × 10⁶ 15 hours Mesh III 41,174 2.37 × 10⁶ 1.5 days Note: Variation < 1.3 %, confirming convergence. Mesh II adopted as optimal. J. Civil Eng. Mater.App. 2025 (September); 9(3): 131-144 ························································································· 137 Figure 3. Finite element mesh of a jointed plain concrete pavement (JPCP). Figure 4. Meshed FE model of a dowel bar Figure 5. Mesh sensitivity analysis (Meshes 1–3) and the spring-force distribution along half of the dowel [15]. 1 2 3 J. Civil Eng. Mater.App. 2025 (September); 9(3): 131-144 ························································································· 138 2.3.4. Boundary Conditions The bottom surface of the subgrade was fully fixed to represent bedrock support [39]. Nodes along the base and subgrade perimeter were horizontally restrained (ux, uy). Slab faces along the longitudinal direction were restrained in x; the transverse joint face was left free, and the opposite slab face was fully constrained in all translational degrees of freedom [54], as illustrated in Figure 6. Figure 6. Boundary conditions applied in the finite element (FE) model. 2.3.5. Finite Element Analysis Procedure Nonlinear 3-D analysis was performed in five stages: one initial step, three static-general steps, and one static Riks step with automatic time increments (initial size 0.001) [37]. The NLGEOM option accounted for large displacements due to geometric, material, and boundary nonlinearities [37]. The first step applied boundary and temperature loads, followed by staged introduction of contact, friction, and wheel load [37], aligning with best practices in nonlinear pavement modeling [15, 35, 37]. 2.4. Model Validation The developed finite-element model (FEM) was validated against Friberg’s analytical formulation for dowel deflection and bearing stress [9, 23]. According to the analytical approach, Friberg’s closed-form solution is applicable when the parameter βL > 2, where L represents the dowel embedment length within one slab [9]. In this study, βL = 3.185, confirming the validity of using Friberg’s equation for comparison with the finiteelement analysis (FEA) results [9, 23]. Deviations of less than 2.5 % between analytical and numerical results confirmed the model’s numerical accuracy and the adequacy of the adopted boundary and contact conditions [9, 23]. Furthermore, the model was evaluated against AASHTO performance criteria [2], consistently predicting load-transfer efficiency (LTE) values above 70 % and relative joint deflections below 0.25 mm [2, 38]. The validation procedure included computations of load transfer, modulus of subgrade reaction, radius of relative stiffness, effective dowel count, and load carried by the critical dowel, following Eqs. (1)–(14) [15, 30, 55]. The model was subjected to a dualwheel loading configuration, as illustrated in Figure 1, and the resulting FEA outputs were compared with Friberg’s closed-form analytical solution for compressive bearing stress [9, 56]. 2.4.1. Validation - Load Transfer and Dowel Response Step 1 — Load Transferred Across the Joint It is assumed that 45% of the applied wheel load is transferred across the transverse joint: Pt=0.45 Pw (1) For an applied wheel load of Pw=20 kN: Pt=0.45×20=9.0 kN (2) Step 2 — Modulus of Subgrade Reaction The modulus of subgrade reaction is determined using the empirical relation: J. Civil Eng. Mater.App. 2025 (September); 9(3): 131-144 ························································································· 139 K=( ESG 20.15)0.779 (3) Given ESG =310 MPa =44,925.2 psi: K=405.75 psi/in =109.99 MPa/m (4) Step 3 — Radius of Relative Stiffness The radius of relative stiffness is computed as: lr=(Ech3 12(1−μ2)K×19.7)0.25 (5) Using the model parameters Ec, h, μ, K, the result obtained is: lr=0.7014 m Step 4 — Number of Effective Dowels Dowels located within one radius lr are considered effective. For a dowel spacing of s=0.30 m: n=Int (lr s)+1=Int (0.7014 0.30 )+ 1=3 (6) Using load distribution coefficients x1=0.572 and x2=0.144: neff =1+x1+x2=1.716 (7) Step 5 — Load on the Critical (Corner) Dowel The load on the corner dowel is given by: Pc1 =Pt neff =9.0 1.716=5.24 kN (8) Contributions from adjacent dowels are: x1Pc1 =2.77 kN, x2Pc1 =0.76 kN (9) Thus, the total load carried by the corner dowel (including adjacent effects) is approximately: Pcorner, total ≈8.0 kN Step 6 — Dowel Support Stiffness, Analytical Comparison, and Model Validation From the finite element analysis (FEA): σ0,FEA =7.524 MPa, y0,FEA =0.000165098 m (10) The corresponding dowel support stiffness is: K=σ0 y0=7.524 0.000165098 =4.557×1010 N/m3 (11) The relative dowel stiffness is: β=(Kd4 4EdId)0.25 =13.74 m−1 (12) Using Friberg’s analytical closed-form solution for identical conditions: y0,Analytical =0.000161568 m, σ0,Analytical =7.363 MPa ( 13) Table 4. Comparison of Dowel Deflection and Stress: FEA vs. Friberg Analytical Solution Quantity FEA Analytical Absolute Difference Relative Difference y0 (m) 0.000165098 0.000161568 3.53×10−6 2.14% σ0 (MPa) 7.524 7.363 0.161 2.14% Friberg’s analytical solution and the finite element model exhibit excellent agreement. The maximum dowel deflection and peak bearing stress deviate by only ≈2.14%, confirming the high numerical accuracy of the developed model and validating the boundary conditions, contact definitions, and dowel– subgrade interaction assumptions for the studied configuration. 2.5. LTE Computation Deflection-based LTE is computed using the standard formula: LTE (%)=δloaded δunloaded×100 (14) Where δloaded and δunloaded are the maximum deflections of the loaded and adjacent slabs, respectively [11,14,57].