Centrifuge Test: The Soil Plug Effect on Pipe Piles in Near-Realistic Pile Driving Penetration Process

Large-diameter open-ended steel pipe piles are increasingly used for offshore wind foundations, yet the influence of installation history on soil plugging and bearing capacity, and the applicability of current design codes, remain inadequately understood. This study conducted 400 g-ton centrifuge model tests under high-net penetration and low-net penetration cyclic installation modes at three relative penetration depths. Regarding penetration characteristics, the low-penetration mode causes residual resistance to increase rapidly and stepwise with depth, reflecting a strong cumulative effect, whereas the high-penetration mode mainly enhances peak resistance at the end of each stroke while accumulating residual resistance more slowly. Regarding bearing characteristics, the pile-head reference load under low-net penetration was consistently higher than that under high-net penetration at the same relative penetration depth, and the difference increased with penetration depth. The equivalent back-calculated plugging coefficient, λp,eq, increased with relative penetration depth in both modes and was consistently larger under low-net penetration. Comparison using the JGJ 94-2008 framework further indicated that the inferred plugging contribution is sensitive to the adopted code-based shaft and tip resistance parameters. Overall, the results show that installation history is an important factor affecting penetration resistance evolution and subsequent axial pile response. However, because soil plugging was not monitored directly, only one complete centrifuge test was performed for each penetration mode, and the loading procedure represents a quasi-static displacement-controlled analog rather than impact hammer driving, the findings should be interpreted as installation-history-dependent trends rather than field-calibrated design relationships. Further replicated centrifuge testing and field-scale validation are required before quantitative design or code-calibration recommendations can be proposed.