Comparison of Westergaard and FE Solutions for Different Temperature Gradients in Concrete Pavements

Abstract: The classical Westergaard formulations offer convenient closed-form estimates for rigid pavement responses, but they rely on idealized assumptions that often diverge from actual field conditions. These assumptions include modeling the slab as infinite and continuous with complete contact to the subgrade, excluding edge lift-off during nighttime upward curling, and assuming linear, symmetric temperature gradients, despite the prevalence of nonlinear thermal profiles in practice. Additionally, Westergaard’s approach lacks explicit modeling of joints, slab separation, and spatially variable support. It reports responses at mid-span rather than along wheel paths, where field measurements are typically taken. These simplifications can misrepresent curvature, misplace elevation extrema, and overlook critical mechanisms such as partial loss of support. To overcome these limitations, finite element (FE) analyses using I-SLAB were conducted across a wide range of parameters, including slab geometry, foundation stiffness, joint load and moment transfer characteristics, and nonlinear thermal gradients. I-SLAB directly models contact conditions, enabling realistic simulation of slab uplift and re-contact, and accounts for spatial variability in subgrade support without relying on superposition. Importantly, outputs are extracted along the wheel path, providing elevation, curvature, and stress profiles that align with field measurement locations and capture coupled effects of temperature, support variability, and joint mechanics—factors that closed-form solutions cannot represent. In this paper, a field comparison using LTPP Section 55-3009 evaluated Westergaard predictions against I-SLAB results under elevated temperature gradients. The FE model successfully reproduced upward curling and edge lift-off along the wheel path, closely matching observed profile shapes and curvature inflection points. In contrast, Westergaard’s assumptions of full contact and mid-span evaluation led to an underestimation of curvature and misplacement of elevation extrema. Overall, the I-SLAB FE approach demonstrated superior agreement with field observations, highlighting its advantages for modeling realistic pavement responses over classical closed-form solutions.