Empirical Model for Driven Pile Capacity Prediction in Lagos Nigeria

Driven piles play a vital role in supporting structures in coastal regions with weak and compressible soils, particularly in Lagos, Nigeria. However, conventional pile design approaches commonly rely on empirical formulas and codes developed outside the country, often without calibration to local subsurface conditions. This mismatch can result in overdesign, unsafe foundations, or unnecessary construction costs. To address this, this study developed a localized empirical model for predicting the axial capacity of driven precast concrete piles using site-specific field and geotechnical data obtained from Ilubirin, Lagos. The research utilized data from full-scale static load tests conducted on 30 precast concrete piles installed across three test locations. Corresponding subsurface profiles, including pile geometry, average soil unit weight, and SPT N-values, were collected and analyzed. Using multiple linear regression techniques, an empirical model was formulated to correlate pile capacity with key influencing variables. The model demonstrated excellent predictive performance with a coefficient of determination (R²) of 0.944, indicating a strong fit between predicted and measured capacities. The model-predicted ultimate pile capacities ranged from 1265 kN to 3346 kN, closely matching the actual static load test results, which ranged from 1250 kN to 3727 kN. Statistical validation of the model using mean absolute percentage error (MAPE) yielded a low error margin of 6.13%, highlighting the accuracy and reliability of the developed prediction tool. Residual plots and regression diagnostics further confirmed that the assumptions of linearity and homoscedasticity were satisfied. This model provides a practical, cost-effective alternative for predicting driven pile capacities during the preliminary design phase in Lagos and similar tropical environments. It offers engineers a site-specific, data-driven solution that minimizes reliance on foreign design standards and reduces the need for multiple expensive static load tests. The study also reinforces the value of integrating local field data into engineering models for improved accuracy and relevance in geotechnical practice.