Editorial

It is my privilege to bring you the Editorial of the November 2025 issue of our journal Structures and Buildings. In this issue, our readership will find six insightful research articles covering topics from artificial intelligence (AI) to structural strengthening. It is particularly pleasing to see familiar names of former colleagues in this issue. As usual, the diverse nature and origin of these excellent scientific contributions highlight the role of civil engineers in all aspects of the built environment, whatever the nature, size or geographical location of an infrastructure project.The use of AI has become common in civil engineering and higher education. With this in mind, Zhang and Su (2025) shed some light into the performance of multimodal large language models (MLLMs) (i.e. ChatGPT-4o and Gemini) at solving structural engineering problems using an image-input approach. Using 90 real visual-based fundamental mechanics questions from a module taught at the University of Hong Kong, the authors show that students performed better than the two MLLMs (average answer accuracies of 62% for students, 52% for ChatGPT-4o and 33% for Gemini). Overall, the performance of ChatGPT-4o is closely aligned with student performance (even exceeding student performance on the topics of kinematics and kinetics), whereas Gemini has a poor performance. The study also reveals the current limitations of LLMs at accurately extracting mathematical/mechanical information from structural images, as well as their limitations in solving calculations and ensuring the logical relationships expected from students. The article concludes with future research needs, which certainly give food for thought to those working in the higher education sector.Our readership interested in structural health monitoring (SHM) and machine learning will be interested in the article by Jesus and Mojtabaei (2025). They numerically investigate the applicability of the compressed sensing (CS) theory to reconstruct the static displacement field of a five-storey building, but using a small number of displacement samples. The five-storey reinforced concrete (RC) building was modelled in OpenSees software. Both modal and static linear analyses were then conducted on the building’s model. A CS reconstruction method was applied to reconstruct the displacement field under gravity load, followed by numerical tests to assess the reconstruction performance in terms of accuracy and computational efficiency. The work proves the effectiveness of CS at accurately reconstructing various measurement fields within buildings, thus emphasising its potential in civil engineering applications. The article is relevant, as very often the amount of data from existing structures is small, and therefore advanced techniques are needed to make the most of it.In the third contribution to this issue, Ahmad et al. (2025) propose a new artificial neural network finite-element analysis (ANN–FEA) method to predict the failure mode of RC beams. Based on a comprehensive dataset of 315 RC beams, the authors compare the predictions given by the trained ANN–FEA method with those obtained from Abaqus and SAP2000. A parametric investigation is also conducted, following the validation of numerical predictions given by the ANN–FEA method. The results show that the ANN–FEA predictions of brittle (shear) failure are closely aligned with the experimental and Abaqus results, whereas SAP2000 struggled to identify brittle failures. This was somehow expected given the different nature of the software. The article highlights the huge potential that machine learning has in assisting the design and assessment of structures.Recent extreme events around the world have produced extensive damage or even collapse of wind turbines, which in turn increases the risks of energy security. The insightful contribution by Ma et al. (2025) examines soil–structure interaction (SSI) effects on the seismic response and fatigue damage of a hybrid wind turbine tower. Non-linear time history analyses and fatigue analyses were conducted using Abaqus to evaluate the dynamic response and fatigue life of the tower (height = 180 m). The hybrid tower was simulated with an ultra-high performance concrete (UHPC) lower tower, and an upper steel tower. The analysis results show that the SSI effects decreased the natural frequency of the wind turbine tower. Compared to a common C80 (i.e. normal concrete) hybrid wind turbine tower, the UHPC hybrid wind turbine tower exhibited lower displacements and damage levels. Moreover, the fatigue life of the hybrid wind turbine tower increased by around 10% after considering the SSI effects. This is a very relevant article as countries move on to more environmentally friendly energy to meet their net-zero commitments. Moreover, the article should be very appealing to our readership working in the design of wind turbines, foundations and UHPC elements.Readers with an interest in bridge engineering are directed to the article by Lv et al. (2025). In this contribution, the authors numerically investigated the accuracy of the rubber tyre model using the discrete-element method when simulating vehicle loads on bridge decks. To achieve this, a bridge model, a rubber tyre model and a bridge deck pavement model were constructed based on the vehicle–bridge dynamic coupling theory using Abaqus. The simulated three-span continuous pre-stressed beam bridge had a total length of 90 m, and each span had a length of 30 m. A comprehensive parametric analysis was then carried out to investigate the bridge behaviour under different traffic load conditions and vehicle speeds (10, 20 and 30 m/s). A comparison of the vertical displacement of each pavement layer on the bridge deck showed that the upper layer displaces the most, but such displacement decreases gradually as the depth increases. The study also found that higher vehicle speeds resulted in smaller maximum vertical displacements, maximum equivalent stresses and maximum longitudinal stresses at the mid-span.The last contribution in this issue experimentally studies the effectiveness of internal metal straps at improving the shear behaviour of RC beams. Abdullah et al. (2025) tested five beams in four-point bending, with the metal straps placed at the shear span as a replacement for normal steel stirrups/links. The main variable in the test was the number of straps. It was found that 12 straps were effective at increasing the resistance of the beams by up to 34% over a control beam without straps, changing the failure mode from a shear-dominated to a more ductile flexure-dominated one. Response2000 software was successful at simulating the behaviour of the beams. Our readership working in the design and strengthening of RC structures will find this article particularly useful.I am certain that readers of Structures and Buildings will find the above contributions both interesting and useful. As usual, the editorial team is open to receive discussions on the above articles, as well as any comments or suggestions you may have.