An Integrated Scan-to-BIM Approach for Buildings Energy Performance Evaluation and Retrofitting

Abstract

Energy retrofitting is paramount to reduce the use of energy in existing buildings, with benefits to the environment and people's economy. The increasing use of novel technologies and innovative methodologies, such as Terrestrial Laser Scanning (TLS) and Building Information Modelling (BIM), is contributing to optimise retrofit processes. In the context of energy efficiency retrofitting, complex semantic 3D BIM models are required that include specific information, such as second level space boundaries (2LSBs), material energy performance properties, and information of the Heating Ventilation and Air Conditioning (HVAC) system and their layout. All this information is necessary for energy analysis of the existing building and planning of effective retrofitting strategies. In this paper, we present an integrated (semi-)automated Scan-to-BIM approach to produce BIM models from point clouds and photographs of buildings by means of computer-vision and artificial intelligence techniques, as well as a Graphical User Interface (GUI) that enables the user to complete the models with information that cannot be retrieved by means of visual features. Information about the materials and their performance properties as well as the specification of the HVAC component is obtained from a database that integrates information from BAUBOOK, OKOBAUDAT and ASHRAE. The Scan-to-BIM tool introduced in this paper is evaluated with data from an inhabited two-storey building, delivering promising results in energy simulations.

Automating building element detection for deconstruction planning and material reuse: A case study

“…After capture, each cloud was spatially sub-sampled to 10 mm for more efficient processing and to reduce local variation in point density. This resolution matches the subsampling step of other contemporary Scan-to-BIM studies [36,67].…”

“…Reality capture has been extensively studied in AEC as a method of digitizing historic sites, verifying as-built details, and tracking new work against a known digital model [31][32][33][34]. Verification is often specifically applied in the mechanical, electrical and plumbing (MEP) domain, given the large amounts and difficult accessibility of relevant components [35,36]. Photogrammetry (taking real-world measurements from 2D imagery) provides the common contemporary method of structure from motion (SfM), which estimates three-dimensional points using the overlap and parallax between a series of still images.…”

“…Son 2017 [63] convexity analysis concrete structural elements producing new models Xu et al 2018 [64] random forest with LSSHOT descriptor scaffolding producing new models Ochman 2019 [48] ILP walls, floors, rooms producing new models Yang 2020* [60] PCA, cross section fitting steel structural elements producing new models Guo et al 2021 [62] feature-based global registration steel structural elements deformation monitoring Kim 2021 [65] deep learning, connectivity relations concrete structural elements producing new models Valero et al 2021 [36] histogram analysis, voxel clustering, polygon matching slabs, walls, MEP producing new models Kaufmann et al 2022 [66] RANSAC, H-obb concrete slabs, walls, columns producing new models Wang 2022 [67] deep learning MEP producing new models smaller continuous planes), and have more opportunities for selfocclusion. This makes full process automation difficult, leading to the need for manual pre-segmentation [60] or working within a specific preknown structural system [62].…”

Methodology for retrospectively developing a BIM model from point cloud scans using ongoing building project as case study

“…By employing laser light to capture data on various objects' height, longitude, and latitude in three coordinates, 3D laser scanning generates spatial data consisting of millions of points, typically in the form of unstructured point clouds (Tan et al, 2020). Subsequently, these point clouds can be processed to extract significant information, generate precise representations, and are employed to identify objects, quantify volumes and other distances, ultimately resulting in the creation of semantically rich 3D models (Valero et al, 2021;Sadeghineko & Kumar, 2020b). Extant studies have mainly focused on the integration of BIM and 3D laser scanning in existing, heritage or historical buildings (Rocha et al, 2020;Alshawabkeh et al, 2021;Liu et al, 2023), with a lack of attention given to the adoption of this technology for new and ongoing large-scale construction projects.…”

Quantifying the Confidence in Models Outputted by Scan-To-BIM Processes

“…At a higher level, it applies tools that make use of that information to provide diagnostics of why something might be happening, predict the possible future outcome, and decide the action based on the objectives. Optimising construction project execution (Akula et al, 2013;Bueno et al, 2018), building energy usage (Valero et al, 2021), and space utilization (Pan et al, 2022) are some of the numerous use cases of digital twins in the built environment. A built environment digital twin is commonly built from a Building Information Model (hereafter `BIM model'), which contains geometric and some semantics (such as element materials) that can be used to support the envisioned use cases (I.…”