Development of Control Strategies for Earthquake Reproduction on Shaking Platform

Afonso Costa Henrique

Key information

Authors:

Afonso Costa Henrique (Afonso Henrique)

Supervisors:

Fernando Virgílio Pires de Oliveira (Fernando Virgílio Pires de Oliveira); Miguel Afonso Dias de Ayala Botto (Miguel Afonso Dias de Ayala Botto)

Published in

June 24, 2026

Abstract

Although computational models for earthquake simulation have grown highly sophisticated, they remain limited when addressing complex, nonlinear, or poorly understood structural behaviours. Numerical simulations and machine learning approaches alike depend on experimental data for validation, calibration, and training. Experimental testing therefore remains indispensable in earthquake engineering for model validation, understanding complex structural behaviour, and ensuring structural safety. This thesis analyses control strategies for reproducing earthquake ground motion on the uniaxial shaking table at Laboratório Nacional de Engenharia Civil, using a computational model of the system. The controller's objective is to reproduce the target ground motion as accurately as possible. Performance is evaluated via response spectra for three control strategies: (1) the currently used OLI (online iteration) method, a pseudo-open-loop offline adaptive control employing iterative drive-signal generation with closed-loop PI testing; (2) a tuned PID-F (proportional-integral-filtered-derivative) controller; and (3) an LQI (linear quadratic integral) controller. To tune the controllers, an identification method better suited to systems operating exclusively in closed loop, such as LNEC's shaking tables, is proposed and applied to identify an electrodynamic shaker. The resulting model is used to compare the three strategies. Results show that both PID-F and LQI outperform OLI, with the optimal controller achieving the best overall performance and representing the most promising solution for controlling the studied shaking tables.