Effect of Installation Method on the Uplift Capacity of Single Driven Steel Piles: Full-Scale Field Study

This study examines the pullout behavior of steel piles used as foundations for solar panels in Shahrebabak city, Iran. The site's geotechnical characteristics were assessed through 15 boreholes using the wash boring method. A total of 40 tests were conducted, evenly divided between UNP 120 and UNP 100 piles, to record penetration rates under constant energy during pile driving. The load-settlement behavior of all piles was systematically compared. Initial results showed no consistent correlation between uplift load and penetration rate. The calculated tensile loads were compared with predictions from the Tomlinson equation, based on standard penetration tests (SPT), yielding consistent results. To enhance the accuracy of predicting pile uplift, two computational estimation models were developed using the LinearRegression class from the Python library scikit-learn, with penetration rate and average SPT values as inputs. The models demonstrated strong predictive performance, with a mean squared error (MSE) of 0.123 Ton and an R-squared (R²) of 0.94 for UNP 120 and an MSE of 0.118 Ton and an R² of 0.928 for UNP 100. Given these strong performance indices, geotechnical engineers can reliably use these models for accurate estimation of pile uplift.