Manual P&ID review is time-consuming, while existing image-based or hard-coded automation methods lack efficiency. We propose a hybrid framework that extracts P&ID data from DXF files and converts natural language rules into FOL-based Python verification modules. By restricting the LLM to code generation, the framework ensures deterministic rule verification via formal logic. Evaluated on four industrial drawings (140 objects, 587 checks), our approach achieves a 97.5% F1-score, compared to 55.9% for the image-based LLM baseline. This enables non-experts to automate design, conditional, and engineering rules, advancing QA in digital twin environments.
Automated Level of Development (LOD) assessment in Building Information Modeling (BIM) is hindered by semantic ambiguity in specifications and geometric complexity of objects. Existing rule-based and vision-based methods lack contextual reasoning or struggle with fragmented, occluded geometry. This study proposes a hybrid framework combining a three-layered knowledge graph—built top-down via LLM requirement extraction and bottom-up via visual population—with a geometric pipeline reassembling meshes and exposing occluded components. A multimodal LLM cross-references graph-derived criteria with multi-view geometry, producing explainable assessments. A four-condition ablation study achieves 83.47% accuracy while reducing false negatives, confirming feasibility for explainable BIM quality assurance.
Construction is increasingly digitalized, and organizations are integrating operational technology, such as robots, into site operations, introducing cyber risks. This paper presents a web-based crowdsourcing platform to collect and analyze human knowledge and to provide recommendations to guide cybersecurity decision-making in the construction sector. Collected insights on risks and mitigations via open-ended questions are converted into recommendations using semantic segmentation, topic modeling, and large language models. The platform was implemented in collaboration with a UAE-based contractor on a lift-assist robot. The resulting recommendations were mapped to the foundational requirements of ISA/IEC 62443, showing alignment with six of the seven requirements.
Current digital twins for hydrogen energy systems mainly rely on physics-based simulations. Still, data-driven approaches are not fully deployed, limiting the ability to support systematic evaluation of control strategies under diverse operating scenarios. This paper addresses this gap by proposing a hybrid digital twin framework for hydrogen-based testing and experimentation facilities, integrating physics-based and data-driven models, which combines time-series and contextual data. The framework enables scenario-based experimentation and performance-oriented control evaluation. While no experimental results are yet presented, the proposed approach establishes a scalable foundation for performance-based feedback loops, supporting future development of adaptive, energy-efficient, and low-carbon control strategies.
This extended abstract addresses the persistent gap between geometric BIM exchange and analysis-ready structural models. Although IFC supports analytical entities, exchanged models often contain geometry without the grids, axes, and connectivity definitions required for finite element analysis (FEA). A geometrydriven workflow is presented for reconstructing local axes, inferring structural grids, correcting alignment, and generating FEM models directly from IFC-derived geometry. A multistorey case study demonstrates that the reconstructed analytical model can be used for numerical analysis and produces structurally meaningful response fields. The contribution is a practical abstraction workflow that improves BIM–FEA interoperability without requiring manually authored analytical models
Title-blocks in engineering drawings contain essential metadata, but automated interpretation remains difficult because of historical scans and heterogeneous layouts. This paper presents a vision-language model (VLM)-based workflow for title-block detection and structured information extraction using the open-source Qwen2.5-VL backbone. The workflow combines promptable visual grounding on full drawings with schema-constrained extraction from cropped title-block images. The study focuses on three representative fields: scale, drawing name, and route. Results show promising performance for both tasks and indicate that open-source VLMs can provide a simplified alternative to conventional multi-stage pipelines for architecture, engineering, and construction (AEC) document analysis.
Bridge conceptual design is challenged by heterogeneous boundary conditions, strict engineering constraints, and large solution spaces. This paper presents an AI-assisted pipeline for automated bridge conceptual design, focusing on early-stage span layout planning. The proposed approach integrates ontology-based semantic knowledge representation with reinforcement learning to enable automated decision-making while preserving engineering consistency. By embedding engineering constraints into the learning environment, the method supports efficient exploration of feasible span configurations. The results indicate that the proposed pipeline provides a robust and extensible foundation for AI-supported bridge planning in realistic engineering scenarios.
Digital Twins increasingly rely on real-time IoT data, yet the end-to-end architectural implications of alternative integration approaches remain underexplored. This paper presents a controlled comparative analysis of three IoT data pipelines connecting The Things Network (TTN) to a Unity-based Digital Twin for mixed reality visualisation: cloud-centric pipelines using Amazon Web Services and Microsoft Azure, and a lightweight publish–subscribe pipeline based on direct MQTT delivery. By isolating the data integration layer, a benchmarking approach evaluates data transmission behaviour, persistence, integration complexity, scalability, and cost. The findings provide practical architectural insight into how pipeline design choices shape Digital Twin behaviour.
Waste generation in deep renovation is driven by onsite activities and their temporal sequencing and scheduling, yet most existing circularity assessment approaches remain scenario and execution agnostic. This paper proposes a schedule-aware BIM-based approach that integrates activity-zone modelling with precedence-constrained scheduling to enable time-based estimation of material flows and waste generation. Implemented through the Circular Intelligent-based Waste Engine (CIWE), the approach supports activity-level waste estimation and onsite circularity assessment. A multi-residential renovation case in Lille, France is used to compare three renovation scenarios. Results show that renovation sequencing significantly influences waste intensity, recovery potential and circularity outcomes.
Construction projects, which are prone to delays and budget overruns, struggle to transform fragmented data streams into coherent intelligence. We propose a novel approach to monitoring that combines onsite data capture, agentic workflows and knowledge graph architecture to provide a foundation for evidence-based progress tracking. This approach is built on an ontology for structured schedule management that unifies concepts from domain-specific ontologies and Autodesk Construction Cloud. Autonomous AI-driven agents enable knowledge evolution by interacting with a knowledge graph. This lays the foundation for a living digital twin that adapts as new data emerges, supporting consistency checking and cross-system integration.
