Integral bridge abutment and backfill behaviour

Abstract

Integral bridges do not contain the bearings or expansion joints found in conventional bridges. Interaction between the abutments and backfill due to thermal movements of the bridge deck leads to the accumulation of soil strain — known as strain ratcheting — which results in increased lateral earth pressures along with settlement at the bridge approach. Uncertainty surrounding strain ratcheting is reflected through design restrictions. This research investigated strain ratcheting and ways it can be limited through smarter design. Geotechnical centrifuge modelling was the main research method used in this project. The reference configuration consisted of a full-height abutment atop a spread footing, supported on and retaining a dense granular sand, with thermal cycles simulated at the deck level using a high precision mechanical actuation system. The actuator was developed for this project, operating in displacement control and capable of applying a 12 kN force in 25 µm increments. The standard loading pattern simulated 120 cycles of ±20 mm representing seasonal thermal movements of a 150 m long concrete bridge over its design life, compliant with the UK design document PD 6694-1 (BSI, 2020). Sensors included an LVDT and load cell measuring applied movement and axial force in the bridge deck, strain gauges to capture abutment bending moments, a Tekscan sensor measuring lateral earth pressures, and a camera to track soil and structural movements. A total of seven tests were carried out with alterations made to the thermal loading, abutment stiffness, backfill density, and backfill constitution. Four tests were initially compared to investigate the mechanisms driving strain ratcheting, in which the behaviour under different cycle magnitudes, backfill densities, and loading histories was considered. Strain ratcheting was found to be dependent upon densification and dilation within the backfill, driven by settlement at the backfill-abutment interface. A similar development in deck axial load before and after large thermal cycles, despite smaller soil movements in the latter, suggests fabric build-up or the readjustment of particles close to the active stress state may also contribute to ratcheting behaviour. Other findings include foundation movement limiting earth pressures, the significance of daily thermal cycles on settlement rather than pressure generation, and the logarithmic increase in deck axial force. A second comparison was made between tests with different relative stiffness configurations that combined a simulated abutment thickness of 1 m or 0.4 m with a backfill relative density of 85% or 40%. The flexible abutments accommodated thermal movements with greater bending than their stiff counterparts, leading to a 46% and 76% decrease in the net soil resistance and peak bending moment, respectively. Furthermore, the increase in the lateral earth pressure coefficient followed a power series relationship with seasonal displacement cycles. At the end of testing, a rotational pattern of soil displacement vectors was observed behind the top half of the abutments and up to 0.83H into the backfill of the reference configuration, with a settlement trough extending 4 m back with a peak of 110 mm. In a final centrifuge investigation, the ability of EPS geofoam to mitigate earth pressure escalation was examined by simulating the placement of a 1.2 m thick block of material between the abutment and backfill of the reference configuration. The compressible inclusion reduced pressure build-up considerably, leading to the deck axial force and peak abutment bending moment reaching around a third and two-thirds of those values found in the reference structure, respectively. Furthermore, the EPS geofoam did not significantly alter the backfill settlement and minimal permanent deformation was observed under realistic loading. Field monitoring was undertaken alongside the centrifuge work to relate findings to real-world behaviour. Over six years of data was analysed from the Van Zylspruit Bridge, a 90 mlong integral bridge in South Africa, with specific aspects of the response explored through separate experiments on a 1.5 m tall abutment model. Only a small increase in earth pressure was observed over the monitoring period, reflecting the seasonal thermal movements of ±5 mm being much smaller than design estimations. Furthermore, the backfill water content stabilised at a residual value of 4% which small-scale modelling suggested may also have contributed towards the lack of ratcheting. Additional tests indicated that strain ratcheting occurs irrespective of bridge deck drying shrinkage and the season of construction. Discrepancy between the high earth pressure generation found in centrifuge modelling and the limited increase recorded in field monitoring agreed with the wider literature. Based on a comparison of 11 studies, this difference was justified by the small backfill shear strains often experienced in the field. Furthermore, data from three integral bridges showed that thermal movements were overestimated by a minimum of 73% due to the high return period typically used when calculating the design temperature range. This work concludes by presenting key implications on integral bridge design, which can benefit from: 1) using temperature ranges with a lower return period to calculate thermal movements; 2) adopting a fixed increase in the earth pressure coefficient below a certain abutment rotation threshold to simplify routine bridge design; 3) providing a design earth pressure distribution for flexible abutments, such as that proposed; and 4) placing compressible inclusions between the abutments and backfill to mitigate pressure escalation.