Estimating the Fundamental Frequency of Masonry Towers Through Experimental Data and Physics-Guided Predictive Models

Historic masonry towers are a significant component of architectural and cultural heritage and require reliable assessment tools to support seismic evaluation, structural monitoring, and conservation. This study investigates the prediction of the fundamental natural frequency of isolated masonry towers using empirical formulations, calibrated power-law regressions, Artificial Neural Networks, and physics-guided approaches. An experimental database of 27 towers with different geometries and boundary conditions was analysed to identify the parameters most strongly influencing the first vibration mode. Traditional empirical equations were compared with measured frequencies, while data-driven and physics-constrained models were developed and validated through cross-validation procedures. The results indicate that effective height and minimum base dimension are the dominant predictors of the first natural frequency. Among the investigated approaches, the proposed physics-constrained model provided the best balance between predictive accuracy, robustness, and interpretability. The proposed framework offers a practical tool for rapid seismic assessment and structural health monitoring of heritage masonry towers