Experimental Characterization of Rotational Stiffness and Moment Capacity in Plain-Embedded Steel H-Pile to Bent Cap Connections

Abstract

Steel H-pile to concrete bent cap connections relying on plain pile embedment are standard in short-span highway bridges across the United States. Current practice treats these connections as pinned or fixed, yet actual behavior falls between the two. Rotational stiffness, and therefore the moment demand distribution in the bent, depends on embedment depth, but no systematic experimental basis existed for HP pile sections to quantify this relationship. This dissertation presents a full-scale experimental program to characterize the moment capacity and rotational stiffness of plain HP pile-to-concrete bent cap connections over a range of embedment depths. Eight specimens were subjected to lateral load tests to generate connection moment demand. The test matrix included HP10×42 piles at embedment-to-pile-depth ratios (Le/d) of 0.5 to 1.5, HP12×53 piles at Le/d = 0.5 and 1.0, and one strong-axis HP10×42 test at Le/d = 1.0. For HP10 and HP12 piles, a transition in governing failure mode was identified between Le/d = 0.75 and Le/d = 1.0. Connections at Le/d ≥ 1.0 failed by ductile steel flange local buckling, and those at Le/d ≤ 0.75 failed by brittle concrete bearing and crushing. No connection behaved as pinned or fixed. The measured degrees of fixity ranged from 50% at Le/d = 0.5 to 93% at Le/d = 1.5, although the measured rotational stiffness values are still very far from the theoretical fixity. The AASHTO LRFD minimum embedment of 12 inches coincides with the empirical failure mode transition boundary for HP10 and HP12 sections but is insufficient for larger pile sizes, or other pile types. A power-law regression, Kθ/EI = 0.78 (ft*rad)-1 (Le/d)2, was developed from the undamaged weak-axis specimens. The near-quadratic exponent is physically consistent with the concrete bearing couple mechanism, supporting the expression as a hand-calculation formula rather than a purely empirical fit. Prior lateral loading reduced rotational stiffness by 19 to 65% per loading event while moment capacity dropped by only approximately 13%, demonstrating that stiffness and capacity are governed by independent mechanisms. The experimental data were implemented in FB-MultiPier through a four-point nonlinear connection spring guideline, validated against all eight specimens and two full-scale pile bents from prior studies. This research provides the experimental, analytical and numerical basis for treating connection stiffness and moment capacity as two separate design checks, each with its own limit state, rather than the prevailing pinned-or-fixed assumption.