Cross-Laminated Timber (CLT) walls are crucial components of modern buildings, consisting of multiple layers of timber bonded together. However, as combustible construction materials, their potential fire risk remains a significant concern. The behaviour of CLT components during a fire is complex and requires careful consideration of both (a) the temperature-dependent behaviour of the timber layers and (b) potential chemical reactions (e.g., pyrolysis) within the wood. For (a), the Eurocode EN 1995-1-2 provides guidelines for assessing the fire resistance of timber structures. For (b), this paper applies a pyrolysis model within a Heat Transfer (HT) analysis framework to predict the CLT structural response under fire conditions. This paper introduces and demonstrates a One-Way Coupled (OWC) fire-structure simulation, which combines Computational Fluid Dynamics (CFD) and Finite Element Method (FEM) domains. Inspired by the standard fire test for CLT walls (Osborne et al. 2012), here, CFD is used to reproduce the ISO-834 standard fire as a preliminary demonstration case. Using the thermal data obtained from the CFD model as the boundary condition, a subsequent heat transfer analysis using Abaqus is able to predict pyrolysis and heat transfer behaviour, but it fails to represent the initial temperature distribution (which because of water evaporation is not considered) and capture post-failure behaviour. Additionally, a Structure Response (SR) analysis of the CLT wall under various mechanical loads indicated failures at different times during the fire. However, due to the lack specific information about experimental set-up in the literature, such as mechanical loads and material properties, future studies are planned to verify the model against experimental data.

The maintenance of buildings, underpinned by the digital twin technique, becomes integral to heritage conservation efforts. To achieve efficient modelling with minimal manual intervention, automated component recognition based on semantic segmentation of point clouds is imperative. Confronted by the challenges of the paucity of requisite datasets and the inherent geometric diversity of historical buildings, a two-step strategy including feature extraction and classification is proposed. First, an improved SHOT descriptor is proposed to extract discriminative features by defining a specific local reference system and concatenating support fields at different scales. The extracted features are then classified with a learning-based network, avoiding a feature learning process that relies on sufficient data. Experiments on real-world heritage point clouds yield 93.7% accuracy and an 80.0% mean-intersection-over-union (mIoU) when descriptors with radii of 0.3 m and 0.9 m are combined, surpassing computationally expensive deep learning networks and data-intensive unsupervised learning. A slight decrease in segmentation performance with random removal of points indicates the high robustness of the proposed method against data missing and sampling density changes. Additionally, a geometric modelling process with an error of less than 10% is introduced to achieve a direct transition from point cloud to model, contributing to the establishment of digital twins for heritage structures.