Purpose This study investigates the thermo-mechanical response of heritage masonry elements under environmental fluctuations, focusing on the Monastery of Batalha, Portugal. The aim is to move from descriptive monitoring to predictive modelling of temperature-driven displacements. By capturing the delayed thermal response of limestone due to thermal inertia, the research supports the development of passive alert systems that can forecast structural behaviour in advance. The ultimate purpose is to enhance preventive conservation strategies and improve resilience of heritage structures in the face of increasingly severe climate-induced environmental cycles. Design/methodology/approach Two fibre Bragg grating sensor systems were installed in the cloister of the Monastery of Batalha to continuously monitor displacement and temperature over two years, generating more than 900, 000 data points. The methodology included (1) initial linear regression analyses, (2) correction to isolate temperature-independent displacement, (3) evaluation of time lags to capture thermal inertia effects and (4) development of predictive regression models trained on lag-adjusted data. Models were validated using training/testing data splits and evaluated with R2 statistics for predictive accuracy across multiple sensors and monitoring weeks. Findings Raw linear regressions showed moderate correlations (R2 up to 0. 51). Introducing optimal weekly sensor-specific lags revealed significant improvements, reflecting variable thermal inertia effects. The predictive models achieved high accuracy, with R2 values above 0. 93 for crack monitoring (FBG₁) and 0. 75–0. 95 for joint monitoring (FBG₂). These models successfully reconstructed missing displacement data and demonstrated the capability to predict responses under projected temperature cycles, enabling earlier detection of critical displacements relevant for structural health assessment and conservation decision-making. Originality/value The study pioneers the integration of fibre optic sensing with lag-based predictive modelling for heritage masonry. Unlike conventional monitoring, which is reactive, the proposed approach forecasts displacement behaviour by accounting for thermal inertia and environmental variability. The methodology transforms high-resolution monitoring data into a predictive tool, offering a scalable framework for heritage conservation worldwide. Its originality lies in enabling passive, data-driven alert systems that anticipate structural responses to climate-driven thermal cycles, directly supporting long-term resilience of monuments exposed to environmental change.
Bourgeois et al. (Fri,) studied this question.