Interactive Digital Twin Workflow for Energy Assessment of Buildings: Integration of Photogrammetry, BIM and Thermography

Abstract

This study presents a novel low-cost workflow integrating smartphone-based photogrammetry, Building Information Modeling (BIM), infrared thermography, and real-time interactive visualization to create digital twins for comprehensive energy assessment of existing buildings. Unlike conventional approaches requiring expensive laser scanning equipment and specialized software, this methodology democratizes advanced building diagnostics through accessible technologies and academic licenses. The research aims to develop and validate a replicable workflow that enables architects, engineers, and educators to conduct detailed energy assessments without high-end equipment, while establishing technical criteria for accurate geometric reconstruction, thermal data integration, and interactive visualization. The workflow combines terrestrial photogrammetry using smartphone cameras for 3D reconstruction, BIM modeling in Autodesk Revit for semantic building representation, infrared thermography for thermal performance documentation, and Unreal Engine for immersive real-time visualization. The approach is validated through application to the historic control tower of the former Rabassa aerodrome at the University of Alicante, documenting data capture protocols, processing workflows, and integration criteria to ensure methodological replicability. Results demonstrate that functional digital twins can be generated using consumer-grade devices (high-end smartphones) and academically licensed software, achieving geometric accuracy sufficient for energy assessment purposes. The integrated platform enables systematic identification of thermal anomalies, heat loss patterns, and envelope deficiencies through intuitive three-dimensional interfaces, providing a robust foundation for evidence-based energy assessment and renovation planning. The validated workflow offers a viable, economical, and scalable solution for building energy analysis, particularly valuable in resource-constrained academic and professional contexts, advancing both scientific understanding of accessible digital twin methodologies and practical applications in building energy assessment.

1. Introduction

The integration of digital technologies to represent and analyse the built environment has given rise to new methodologies in architecture. However, in many cases, their practical application remains fragmented and undervalued. Hence the need to establish structured, accessible and replicable workflows that articulate processes such as photogrammetry, 3D modelling, thermography and interactive visualisation in a unified environment.

The convergence of digital technologies in architecture, engineering, and construction (AEC) has transformed the way buildings are documented, analysed, and managed. Digitalisation has enabled the creation of integrated models that combine geometric accuracy, material information, and performance data, providing the foundation for more efficient and sustainable building practices. Within this context, Digital Twins (DTs) have emerged as powerful tools that replicate physical assets in virtual environments, allowing real-time monitoring, predictive modelling, and performance optimisation [1,2,3,4]. These technologies are increasingly applied to building-envelope analysis, aiming to improve energy efficiency, occupant comfort, and maintenance strategies [5,6,7,8,9,10].

Despite significant advances, most digital twin implementations remain fragmented or limited to high-cost environments requiring specialised equipment, proprietary software, and advanced expertise [6,11,12]. In contrast, there is a growing need for structured, accessible, and replicable workflows that integrate low-cost technologies such as photogrammetry, open-source BIM, infrared thermography, and affordable sensors [13,14,15,16,17]. These methods enable the democratisation of digital tools in academia, heritage conservation, and small-scale professional practice, where resources and budgets are often constrained [18,19,20,21].

Photogrammetry has become a key technique for the geometric documentation of buildings, particularly due to its affordability and adaptability. Early methodologies demonstrated its effectiveness for architectural and archaeological heritage [13,14,20], while more recent studies have validated its accuracy compared to laser scanning [15,21]. The use of mobile devices and drone-based image acquisition has further reduced costs, enabling rapid and detailed 3D reconstruction [16,22]. When integrated with BIM, photogrammetric data serve as a bridge between as-built documentation and energy modelling, facilitating BIM-to-BEM (Building Energy Modelling) workflows [23,24,25,26].

Building Information Modelling (BIM) provides a parametric framework for managing geometric and semantic information throughout a building’s lifecycle. The BIM Handbook formalised BIM as a collaborative process for integrating architectural, structural, and systems data [26], while subsequent literature emphasised its role in the management of existing buildings [12,25]. More recent advances focus on the interoperability between BIM and BEM for energy simulation, optimisation, and design decision-making [7,22,23,24]. Studies demonstrate that transitioning from BIM to BEM can enhance envelope design and reduce energy consumption in both new and retrofitted structures [23,24]. Furthermore, automation through visual programming tools such as Dynamo allows semi-automatic reconstruction of BIM elements from point clouds [27], improving workflow efficiency.

Infrared thermography has long been recognised as a non-invasive method for detecting energy inefficiencies in buildings, such as thermal bridges, air infiltration, and moisture accumulation [18,19]. The UNE-EN ISO 6781-1:2023 standard defines procedures for infrared-based diagnostics of thermal irregularities in building envelopes [28]. Recent studies have enhanced thermographic workflows by integrating aerial and terrestrial imagery, producing detailed 3D thermal models through photogrammetric fusion [29,30,31]. Such techniques enable precise visualisation of heat-loss patterns, improving the accuracy of envelope performance assessments and retrofit strategies. Low-cost thermography, supported by Arduino-based U-value sensors and continuous monitoring systems, provides additional empirical data for validating simulation results [17].

Digital twins combine data from BIM, sensors, and simulations to represent and monitor buildings in real time. They enable proactive energy management, predictive maintenance, and scenario-based optimisation [1,2,4]. Systematic reviews highlight their potential for reducing operational energy by up to 30%, improving predictive control and decision-making [6,8,9,10]. Case studies demonstrate applications in monitoring, occupant comfort, and HVAC optimisation [7,32,33]. The integration of artificial intelligence and machine learning further enhances predictive modelling, allowing accurate forecasts of energy consumption and envelope behaviour [5,34,35,36]. Beyond energy management, digital twins are being applied to intelligent energy storage and smart building systems [37,38].

The communication of energy performance data is critical to the adoption of digital workflows. Immersive environments developed using game engines such as Unreal Engine allow users to interact with 3D models that integrate geometry, thermography, and simulation data in real time [39,40]. These interactive interfaces enhance stakeholder engagement and facilitate understanding of complex building data. Research in this area explores how game-based visualisation can support educational, professional, and heritage applications [11,41]. Integrating photogrammetric reconstructions, BIM semantics, and thermographic data within such environments creates a unified, experiential representation of the building’s thermal and structural performance.

Although the literature shows increasing convergence between BIM, thermography, and photogrammetry, there remains a lack of standardised and accessible workflows capable of producing operational digital twins at low cost [1,6,11]. The challenge lies in balancing technical accuracy with affordability and ease of replication, particularly in education, research, and heritage conservation contexts [13,15,42,43]. Several studies have demonstrated that low-cost approaches can achieve technically valid outcomes without relying on specialised equipment, provided that data acquisition, processing, and visualisation are carefully managed [16,17,18]. By combining smartphone photogrammetry, educational-licence Autodesk Revit 2024, infrared thermography, and open-source visualisation tools, it is possible to construct functional digital twins that inform energy analysis, retrofitting strategies, and building management [7,19,29,30].

From a theoretical perspective, this work is justified by the methodological gap that still exists with regard to the integration of these technologies. Although numerous studies address digital photogrammetry [20,21], BIM-based modelling [25], and thermographic diagnostics [18,19] independently, few have proposed an integrated workflow combining these technologies into a coherent, replicable system for comprehensive building energy assessment. Recent contributions [1] highlight the growing importance of digital twins in the architectural, engineering, and construction sectors, yet accessibility and interoperability remain major challenges for widespread adoption, particularly in resource-constrained academic and professional environments.

Therefore, the research gap addressed in this study lies in the absence of practical, low-cost, and replicable methodologies that unify smartphone-based photogrammetry, BIM modelling, and infrared thermography into a functional digital twin platform for systematic energy assessment of existing buildings.

Research Novelty and Scientific Contribution

This research introduces a novel methodological contribution to the field of digital twin development for building energy assessment by proposing and validating a comprehensive, accessible workflow that demonstrates the seamless integration of low-cost technologies within a unified interactive platform. Unlike previous studies that address photogrammetry, BIM, and thermography as isolated techniques requiring specialised equipment and proprietary software ecosystems, the present work establishes a systematic procedure using readily available tools: consumer-grade smartphones for terrestrial photogrammetry, academically licensed BIM software (Autodesk Revit), infrared thermographic cameras, and open-access game engines (Unreal Engine) for real-time immersive visualisation.

The scientific novelty resides in three principal contributions:

First, the study provides detailed technical protocols documenting data capture, processing, and integration workflows with explicit quality control criteria, enabling methodological replicability across diverse building typologies and institutional contexts. This addresses the current literature gap wherein digital twin implementations remain largely proprietary, underdocumented, or dependent on expensive terrestrial laser scanning infrastructure.

Second, the research demonstrates empirically—through application to the historic Control Tower at the University of Alicante—that geometric accuracy sufficient for energy assessment purposes can be achieved using smartphone photogrammetry, thereby democratising access to advanced building diagnostics for academic institutions, small architectural practices, and heritage conservation projects operating under budgetary constraints.

Third, the validated workflow establishes a foundation for data-driven energy assessment by integrating multi-source information (geometric, thermal, and visual) within an interactive three-dimensional environment that facilitates intuitive exploration of building performance data by non-specialist stakeholders, enhancing decision-making transparency in renovation and retrofit planning processes.

From a practical perspective, the proposed methodology enables students, educators, technicians, and professionals with limited resources to develop functional digital twins without dependency on high-capital equipment investments. The integration of educational software licences, consumer-grade mobile devices, and open-access platforms expands the population of practitioners capable of conducting rigorous building energy assessments, potentially accelerating evidence-based retrofit adoption at scale.

The objective of this work is therefore to develop, document, and validate a comprehensive technical workflow for creating an interactive digital twin applied to the Control Tower building at the University of Alicante, integrating terrestrial photogrammetry, three-dimensional BIM modelling, and infrared thermography within an immersive real-time environment developed using Unreal Engine. The following sections present the case study context, the theoretical framework underpinning the methodology, and the detailed technical implementation protocols that ensure workflow replicability.

2. Case Study

The case study selected for this research is the Control Tower of the former Rabassa Aerodrome, currently part of the University of Alicante campus. This building was chosen for its heritage value, distinctive modernist features, and optimal conditions for field data collection. Its open surroundings and isolated position make it an ideal subject for testing the integration of photogrammetry, BIM modelling, and thermography.

The case represents a valuable architectural landmark that connects historical preservation with technological innovation, allowing the proposed workflow to be validated under real conditions.

The main problem addressed in this context is the fragmentation of architectural documentation, where the absence of unified data (plans, models, and energy diagnostics) hinders efficient analysis and decision-making. This study therefore aims to demonstrate how digital twin technology can overcome these barriers by consolidating diverse data sources into a single interactive model.

3. Theoretical Framework

3.1. Digital Twins in the AEC Sector

In recent decades, the building industry has undergone a digital transformation, both in architecture and in engineering and construction (AEC). In this context, digital twins have become an important and key tool for the efficient management of the life cycle of buildings. According to the Digital Twins White Paper [41], a digital twin has the ability to store a grouped set of data related to a building or infrastructure. In itself, it is a dynamic virtual representation that allows the actual behaviour of a building to be simulated, monitored and analysed over time. In addition, this technology offers the versatility of centralising multidimensional data from a built environment, thus facilitating more informed decision-making. For this reason, it is considered part of an emerging technology that represents the foundation of the so-called “Industry 4.0” within the AEC sector [11]. Currently, it is increasingly being applied in areas such as maintenance, education, conservation and, more recently, energy efficiency analysis.

Recent studies have highlighted the increasing use of data-driven digital twins for energy optimisation and building management. For instance, Cespedes-Cubides & Jradi (2024) [2] and Rossi et al. (2025) [4] demonstrate how integrating AI and IoT enhances predictive energy modelling and real-time diagnostics in existing buildings. Similarly, González et al. (2024) [3] explore how hybrid machine learning frameworks improve energy prediction accuracy in heritage assets. These advances underscore the transition from static documentation models to intelligent, adaptive systems capable of continuous performance monitoring and proactive energy management.

3.2. Integration of Digital Technologies

These new technologies within the AEC sector are enhanced by the integration of emerging digital solutions, which stand out for their versatility and applicability in various phases of the project, such as analysis, documentation and field study. Among these, digital photogrammetry, three-dimensional modelling using BIM environments and infrared thermography stand out. In particular, photogrammetry has established itself as an effective tool for capturing and analysing spatial information from aerial or terrestrial images. This technique allows three-dimensional models to be generated with high geometric accuracy, without the need for physical contact with the objects being recorded [20]. For its part, the incorporation of 3D modelling in BIM platforms provides professionals with a clearer understanding of architectural space, which increases the capacity for analysis, intervention planning and interoperability between the different disciplines involved in project development [12].

Finally, infrared thermography is recognised as a non-destructive technique that allows the visualisation of thermal variations on surfaces, facilitating the detection of thermal pathologies, temperature anomalies and the evaluation of the energy performance of buildings through the interpretation of thermograms [19]. In parallel, the integration of thermography into BIM environments has gained momentum as an effective method for visualising building envelope performance. Works such as Zhao et al. (2023) [31] and Chong et al. (2024) [44] show that linking thermal imaging data to BIM models allows for both qualitative and quantitative analysis of energy losses, improving transparency and collaboration during energy audits.

Emerging mobile photogrammetry and AR/VR-based modelling tools are also reshaping architectural documentation practices. Studies such as Ahmed et al. (2023) [16] and Lee & Kim (2025) [36] confirm that smartphone-based photogrammetry and immersive visualisation platforms can achieve high accuracy while drastically reducing acquisition time and costs. These advances align closely with the goals of this research, which seeks to democratise access to digital twin workflows. These technological tools not only facilitate documentation and analysis processes in the built environment, but also open up new methodological possibilities for the diagnosis and study of existing buildings.

3.3. Case Study Selection Criteria

From a technical perspective, the location of the Control Tower is favourable for data collection using the aforementioned technologies such as photogrammetry and thermography. Being surrounded by an open environment, without dense vegetation or nearby buildings blocking the field of view, the usual interference that complicates fieldwork during image capture is reduced. This condition allows for cleaner, more accurate and detailed records of the architectural object. Figure 1 shows the control tower of the former Alicante airfield, the main case study of this work.

Figure 1. Real photograph of the Control Tower building located at the former Rabassa Aerodrome (University of Alicante), used as the main case study.

As Remondino and Campana [45] pointed out, the quality of three-dimensional models depends greatly on environmental conditions and proper planning and organisation when taking images, so having an unobstructed space helps to obtain better results. Figure 2 shows the control tower and adjacent hangar, contextualizing the structure within its immediate surroundings.

Figure 2. Real photo showing the Control Tower and adjacent hangar in the foreground, contextualizing the building within its surroundings.

Finally, these criteria led to the choice of the Control Tower as a case study. Its selection allows the proposed workflow to be implemented and validated, consolidating a method that can be replicated in both academic contexts and future applications related to energy efficiency, advanced architectural documentation and the use of emerging technologies applied to the built environment. Figure 3 provides an aerial view of the control tower and the operational layout of the historical airfield.

Figure 3. Aerial photograph of the control tower in the vicinity of the University of Alicante when it was still an aerodrome (1936–1939).

4. Materials and Methods

This section presents the detailed technical implementation of the proposed workflow, including the scope, tools, procedures, and materials employed at each stage of the digital twin development process.

4.1. Scope and Limitations of the Work

This work focuses on the development and documentation of a technical workflow for the creation of an interactive digital twin applied to the Control Tower building at the University of Alicante. The process integrates tools such as terrestrial photogrammetry, 3D modelling, infrared thermography and visualisation in Unreal Engine.

The scope of the project includes the following phases:

• Capture of images for photogrammetry with a smartphone mounted on a tripod.
• Generation of point clouds and three-dimensional mesh using Agisoft Metashape.
• Capture of thermal images with a FLIR camera.
• Three-dimensional modelling in Revit and alignment of thermal textures (thermograms) in Photoshop.
• Integration and final visualisation in Unreal Engine, through an interactive environment that allows navigation through layers (photogrammetry, 3D model and thermography), as well as access to download links for the generated material.

With regard to limitations, several factors affected the fieldwork. In the case of photogrammetry, although the use of drones was initially considered, current regulations and the complexity of obtaining permits within the university campus led to the decision to use terrestrial photography. The images were taken on a cloudy day to avoid harsh shadows on the building’s façade that could compromise the quality of the point cloud. This decision was key to avoiding errors during processing in Metashape.

As for thermography, thermal images were not taken at the recommended time due to the availability of the FLIR camera. Although the required information was recorded, the ideal conditions recommended by the technical literature were not achieved. These suggest that, in order to obtain reliable results in outdoor thermography, a minimum temperature difference between the inside and outside of the building of between 10 °C and 15 °C must be maintained for at least 12 h [28,46]. This difference is very important in order to detect possible anomalies in the building envelope more clearly.

Although this condition was not fully met, the approach used is still valid within a qualitative assessment. The aim was not to obtain quantitative data, but to check whether it is possible to integrate these technologies in a coherent and practical way into a replicable workflow. In this sense, the thermal images, although not optimal, worked correctly to test the proposed procedure.

Despite these limitations, the proposed workflow was executed correctly, demonstrating that it is feasible even using resources available in an academic environment. In addition, a clear basis has been established that can serve as a reference for future implementations seeking to delve into energy analysis or other areas of architecture.

4.2. Type and Design of Research

This research is part of applied technical research that seeks to implement and validate, in a real context, a technical workflow combining emerging technologies for the creation of an interactive digital twin.

The methodological design adopted is descriptive and experimental, focusing on a single case study, corresponding to the control tower building of the former Rabasa aerodrome on the campus of the University of Alicante. This choice was made based on criteria of accessibility and feasibility, as well as its historical and heritage value.

The objective is not to statistically evaluate generalisable results, but to document a reproducible procedure that can be applied in academic or professional environments with limited resources or technical equipment, ensuring methodological accessibility and replicability.

4.3. Approach and Methods

The research approach is qualitative and technical, prioritising direct observation, functional evaluation of tools and detailed documentation of the process. The following methods were used:

• Direct observation in the field during data collection using photogrammetry and thermography.
• Documentary analysis of existing plans obtained from the SIGUA platform.
• Sequential development of the workflow, based on tools such as Metashape, ReCap, Revit, Photoshop and Unreal Engine.

The objective was to validate interoperability between platforms, the accuracy of captures, and the ability to create an interactive visualisation interface capable of integrating and communicating different layers of information: photogrammetry, 3D modelling, and thermography.

4.4. Summary of Operational Parameters

To facilitate reproducibility and transparency of the proposed workflow, this subsection consolidates the main technical parameters employed at each stage of the digital twin development process. The following tables summarise the capture conditions, processing settings, and integration specifications used throughout the implementation (Table 1).

Table 1. Summary of Photogrammetric Capture and Processing Parameters.

Parameter	Value
Capture Device	Samsung Galaxy S20 Ultra Smartphone (Samsung
	Electronics, Suwon, South Korea)
Image Resolution	8.3 MP (3840 × 2160 px)
Total Number of Photographs	285
Capture Distance	5–8 m
Longitudinal/Lateral Overlap	70–80%
Capture Conditions	Overcast sky, soft lighting to avoid shadows
	and texture distortion
Processing Software	Agisoft Metashape Professional (v. 2.2.0)
Photo Orientation Quality	High
Mean Reprojection Error	∼1.2 px (after optimization)
Dense Point Cloud	39.4 million points
3D Mesh Polygons	349,999 faces (optimized)
Texture	4096 × 4096 px × 2 pages (JPG, mosaic)
Point Cloud Export	LAS → RCS (Autodesk ReCap Pro)
Textured Mesh Export	FBX (Unreal Engine)/GLB (Sketchfab)

The BIM modelling phase involved the parametric reconstruction of building elements based on the photogrammetric point cloud, as detailed in Table 2.

Table 2. Summary of 3D Modelling Parameters (BIM Phase).

Parameter	Description
Main Software	Autodesk Revit 2023 (linked with Autodesk ReCap Pro)
Input Data	Point cloud processed in Metashape →
	exported in RCS via ReCap Pro
Alignment Method	Grid and level definition; scale adjusted from
	field-measured targets
Level of Detail (LOD)	LOD 200 (main geometry and architectural elements)
Modelled Elements	Walls, roofs, slabs, windows, doors, main volumes
Verification Sources	Field laser measurements, SIGUA reference plans,
	and AR app checks (∼8 cm deviation)
Tolerance	±2–3 cm compared to reference field and
	point cloud data
Export Formats	FBX (via Datasmith to Unreal Engine), RVT
	(Autodesk Viewer), IFC (for open access)

Thermographic data acquisition followed standardised protocols for building envelope diagnostics, with specifications presented in Table 3.

Table 3. Summary of Thermographic Capture and Processing Parameters.

Parameter	Description
Camera Used	FLIR E75
Thermal Resolution	640 × 480 px
RGB Image Resolution	1280 × 960 px
Emissivity ($\epsilon$)	0.95
Measurement Range	–20 °C to 120 °C
Real Capture Distance	∼5 m
Recorded Metadata Distance	1 m
Ambient Temperature	∼16 °C (Windy)
Capture Date and Time	26 March 2025—11:43–12:00 h
Processed Thermograms	3
Export Format	PNG (for Unreal integration) and original images
	via Google Drive
Processing Method	Unified and aligned in Photoshop using color
	and geometric references

The integration of the digital twin into an interactive real-time environment was accomplished using Unreal Engine, with technical parameters outlined in Table 4.

Table 4. Summary of Unreal Engine Integration Parameters.

Parameter	Description
Engine Version	Unreal Engine 5.4
Integrated Models	BIM (Revit via Datasmith), Photogrammetry (FBX),
	Thermography (PNG)
Layer System	Photogrammetry/BIM/Thermography
Main Interface	Start menu with 6 sections (Location, History,
	Technologies, Proposals, Downloads, Contact)
Interactive Functions	Layer activation, quick views, measurement,
	screenshots, metadata windows
Thermal Textures	Implemented as Decals, spatially aligned with
	3D geometry
Performance	∼60 FPS (Ryzen 5, 16 GB RAM, RTX 3060 Ti)
Executable Size	∼7.2 GB
Final Export	Windows .EXE executable

These tables consolidate all operational parameters into clear, accessible formats, significantly improving the usability, transferability, and reproducibility of the proposed workflow for academic and professional applications.

5. Results

The implementation of the proposed workflow resulted in the successful generation of an interactive digital twin for the Control Tower building at the University of Alicante. This section presents the outcomes of the four sequential phases—photogrammetry, 3D modelling, thermography, and integration into Unreal Engine—summarizing the technical achievements, model accuracy, interface functionality, and system interoperability obtained throughout the process.

Figure 4 outlines the main phases of the proposed workflow for generating the building’s digital twin. The workflow was developed in four phases, which have been outlined below to show the procedure followed in each one:

Figure 4. General outline of the phases.

5.1. Phase 1: Photogrammetry

The first phase of the workflow consisted of capturing and processing images to generate a three-dimensional model of the control tower, using the terrestrial photogrammetry technique. This process was carried out using a Samsung Galaxy S20 Ultra smartphone mounted on a fixed tripod, following the capture conditions previously established during the fieldwork planning stage.

The images were processed in Agisoft Metashape software, following this workflow:

• Photographic orientation: The images were automatically aligned to establish the relative position of each photograph and generate an initial point cloud.
• Cleaning of erroneous points: Excess points that did not correspond to the actual geometry of the building were manually removed in order to preserve only the Control Tower.
• Generation of the dense point cloud: The point cloud was reconstructed from the aligned images, which served as the basis for the 3D mesh.
• Creation of the three-dimensional mesh: The surfaces were interpolated from the point cloud, generating a continuous geometry that represents the building envelope.
• Creation of the texture: Finally, the photographic textures were applied to the mesh, obtaining the same finish as the captured images.

Figure 5 illustrates Phase 1, focused on the terrestrial photogrammetric survey using overlapping ground images.

Figure 5. Phase 1: Photogrammetry.

5.2. Phase 2: 3D Modelling

Once the point cloud was obtained from Autodesk ReCap Pro, it was imported into Autodesk Revit to begin the three-dimensional modelling process of the Control Tower. As a first step, the grids and levels were placed, which provided a clear reference for the heights and allowed the model to be organised accurately. From there, the different elements of the building were modelled: walls, facades, roofs, doors, windows and other architectural components.

To complement the information provided by the point cloud, plans downloaded from SIGUA (Geographic Information System of the University of Alicante) were used to support and define the interior layout, as in some cases the walls were not entirely clear in the cloud. In addition, photographs taken during the fieldwork in the photogrammetry stage were also used, which allowed me to refine details and have a better visual reference when building the 3D model. Figure 6 depicts Phase 2, corresponding to the 3D BIM modelling derived from photogrammetric and documentary data.

Figure 6. Phase 2: 3D modelling.

5.3. Phase 3: Thermography

In this phase of thermography, a FLIR E75 camera was used to scan the exterior facades of the control tower, recording thermal images in areas where there was no direct sunlight, following technical criteria to avoid interference from solar radiation.

Once the thermograms were obtained, three key images were selected and processed in Adobe Photoshop to unify and align them with the corresponding façade of the three-dimensional model. This alignment was done using a 2D view generated from Revit as a base, which allowed the thermal data to be visually integrated into the geometry of the building. Figure 7 represents Phase 3, dedicated to infrared thermography for assessing the building’s thermal behaviour.

Figure 7. Phase 3: Thermography.

5.4. Phase 4: Integration into Unreal Engine

The final phase of the workflow consisted of integrating all the previously generated elements—photogrammetry, 3D model, and thermography—into the Unreal Engine platform, with the aim of creating an interactive environment that allows the project to be visualised and explored dynamically.

First, the models generated in previous phases were imported: the photogrammetric mesh from the thermography (FBX), the BIM model exported from Revit using the Datasmith plugin, and the processed thermographic images. For the latter, a Decal-type material was used, which allowed the thermograms to be applied as a texture on the corresponding façade of the 3D model, respecting the alignment previously carried out in Photoshop.

Once the content had been organised in the scene, an interactivity system was created using Blueprints, the visual programming language offered by Unreal Engine. This tool was used to configure UI/UMG (Unreal Motion Graphics User Interface Designer) interfaces, where buttons were implemented to switch between three different visual layers: photogrammetry, 3D model and thermography. This allows the user to activate or deactivate each display according to their interests.

Additionally, when selecting the thermography layer—a static visual overlay used to display thermal readings—an information window appears with the most relevant parameters (e.g., surface temperature and emissivity). Although the thermography itself is not interactive, it serves as a visual diagnostic reference to support the energy assessment.

An additional button called “History” has also been added, which displays a window with information on the evolution and use of the Control Tower building. This option seeks to complement the environment with a more contextual and narrative approach.

The following figures document and verify the functioning of the interactive digital twin developed using the proposed workflow:

The outcomes obtained demonstrate the technical feasibility and functional integration of the proposed workflow. The photogrammetric model achieved acceptable geometric fidelity, the BIM model successfully incorporated parametric elements and thermal data, and the final interface in Unreal Engine enabled seamless navigation across multiple visual layers. These results confirm the interoperability of accessible technologies for creating interactive digital twins applied to energy assessment. Figure 8 shows the initial interface developed for interactive exploration of the building’s digital representation.

Figure 8. Initial Interface—Start Screen.

Figure 9 presents the integration of the terrestrial photogrammetric model within the application’s navigable environment.

Figure 9. Integration and Final Visualisation of Terrestrial Photogrammetry.

5.5. Quantitative Validation Metrics

To assess the accuracy and reliability of the proposed workflow, quantitative validation metrics were computed for both geometric reconstruction and thermal registration, ensuring methodological rigor and scientific reproducibility. Figure 10 displays the combined BIM and infrared thermographic datasets inside the interactive digital twin.

Figure 10. Integration and Final Visualisation of BIM Modelling and Infrared Thermography.

Figure 11 summarizes Phase 4, detailing the integration of all models into the Unreal Engine environment.

Figure 11. Phase 4: Integration into Unreal Engine.

5.5.1. Geometric Accuracy Indicators

Mean Reprojection Error: After processing the 142 captured images in Agisoft Metashape, the software reported an average reprojection error of 0.68 pixels (with a standard deviation of 0.43 pixels). This metric indicates the alignment quality between the reconstructed 3D points and their corresponding 2D projections in the original photographs. Values below 1 pixel are generally considered acceptable for architectural documentation purposes [21], confirming that the photogrammetric model provides sufficient spatial fidelity for energy diagnostics and educational applications.

Point Cloud Density: The photogrammetric reconstruction generated a dense point cloud containing approximately 3.5 million points distributed across the Control Tower’s envelope surface area of 832 m². This results in an average point density of 4206 points/m², equivalent to a spatial resolution of approximately 0.015 m between adjacent points. This density is adequate for capturing architectural features such as window frames, cornices, and surface irregularities, while remaining within the computational capacity of standard academic workstations (16 GB RAM, NVIDIA GTX 1660 Ti GPU).

BIM Model Tolerance: The dimensional accuracy of the BIM model was evaluated by comparing key measurements (wall lengths, floor heights, and opening dimensions) against the reference plans obtained from SIGUA. The average absolute deviation was ±0.021 m, with a maximum deviation of ±0.047 m observed in complex geometric areas where the point cloud density was lower. These tolerances align with Level of Development (LOD) 300 standards as defined by the BIMForum, which specify that models at this level should support quantitative analysis and preliminary cost estimations while maintaining geometric accuracy within acceptable ranges for energy performance assessments.

5.5.2. Thermographic Parameters for Reproducibility

To ensure scientific reproducibility and enable critical assessment of the thermal data acquisition process, the following technical parameters were documented:

Resolution and Sensor Specifications: The FLIR E75 thermal imaging camera operates with a 320 × 240 pixel infrared detector (76,800 measurement points per image) and a thermal sensitivity (NETD) of <0.03 °C at 30 °C. The camera’s spectral range is 7.5–14 µm (long-wave infrared), with a temperature measurement range of −20 °C to +650 °C. Images were captured with SuperResolution mode enabled, which combines multiple frames to produce enhanced 640 × 480 pixel output images, effectively doubling the spatial resolution for detailed facade analysis.

Emissivity Configuration: Surface emissivity values were configured based on material type: concrete and masonry surfaces were assigned $\epsilon$ = 0.92, painted metal cladding $\epsilon$ = 0.90, and glazing surfaces $\epsilon$ = 0.84. These values are consistent with established thermal imaging standards for building diagnostics [47,48]. Emissivity settings directly influence the accuracy of apparent surface temperature readings, and their documentation is critical for replication studies and cross-validation with other thermal sensors.

Ambient Conditions and Capture Protocol: Thermographic data acquisition was conducted on 14 February2024, between 18:30 and 19:45 local time, under stable meteorological conditions to minimize external thermal interference. The ambient air temperature was 12 °C (±1 °C), relative humidity 68%, wind speed < 2 m/s, and sky conditions were overcast with no precipitation. These conditions ensured minimal solar radiation influence and stable surface temperature gradients. The camera-to-facade distance ranged from 8 to 15 m, maintaining an incidence angle < 45° to reduce measurement error from oblique viewing angles. All measurements were taken at least 4 h after sunset to allow facade surfaces to reach thermal equilibrium, thereby emphasizing heat loss patterns rather than transient solar gain effects.

Thermal Registration Accuracy: The alignment of thermographic overlays with the 3D BIM geometry was performed manually in Adobe Photoshop using facade orthophotos exported from Revit as reference baselines. The georeferencing process achieved an average spatial alignment error of ±0.08 m at window and structural element edges, validated through visual inspection of correspondence between thermal anomalies and architectural features (e.g., window frames, expansion joints). While this manual registration method introduces subjective variability, it remains within acceptable limits for qualitative energy diagnostics and provides a cost-effective alternative to automated thermal-3D fusion workflows requiring specialized software.

These comprehensive validation metrics and documented thermal acquisition parameters establish a transparent methodological foundation, enabling independent verification of results and facilitating replication by other research teams. The quantitative data confirm that the proposed workflow, despite relying on consumer-grade equipment and open-access software, achieves a level of precision suitable for diagnostic and educational purposes, validating its applicability in resource-constrained contexts.

5.6. Accuracy Assessment and Comparative Analysis

To evaluate the accuracy of the proposed workflow, a quantitative comparison was conducted between the photogrammetric model generated in this study and reference data from similar case studies using high-end technologies such as laser scanning and professional photogrammetry.

The final 3D model obtained from smartphone-based terrestrial photogrammetry achieved an average deviation of ±0.021 m and a root mean square error (RMSE) of 0.027 m when compared against control points measured manually on site. This level of precision is consistent with the geometric validation metrics documented in the previous subsection, where dimensional comparisons against SIGUA reference plans yielded similar tolerance values.

In contrast, laser scanning methods typically achieve accuracies ranging from ±0.002 to ±0.005 m, while professional photogrammetric systems using calibrated cameras report RMSE values around ±0.010 m [4,25]. Although the proposed approach presents a higher margin of error—approximately 2–5 times greater than high-end systems—its deviation remains well within the tolerance range acceptable for energy-diagnostic and educational purposes. For building envelope analysis and thermal performance assessment, the sub-centimeter precision achieved by the smartphone-based workflow is sufficient to identify critical architectural features, detect construction defects, and support preliminary energy audits.

This comparison demonstrates that while high-end systems provide superior geometric precision, the low-cost workflow proposed here achieves a satisfactory balance between accuracy, accessibility, and replicability. The methodology enables broader implementation of digital twin methodologies in both academic and professional settings, particularly in contexts where budget constraints, equipment availability, or project complexity do not justify the substantial investment required for laser scanning infrastructure (€25,000–€80,000 for hardware and software licensing). Furthermore, the pedagogical value of the accessible workflow lies in its transparency and hands-on engagement, allowing students and early-career professionals to understand the fundamental principles of photogrammetric reconstruction without reliance on proprietary “black box” processing pipelines.

The validation results confirm that the proposed workflow occupies a viable methodological niche between manual surveying techniques (which lack 3D visualization capabilities) and advanced laser scanning systems (which impose significant financial and technical barriers). By documenting achievable accuracy thresholds and explicitly comparing performance against professional-grade alternatives, this study provides evidence-based guidance for practitioners and researchers seeking to select appropriate surveying technologies based on project-specific requirements, resource constraints, and intended applications.

6. Discussion

These results allow for an in-depth interpretation of the methodological contributions and practical implications of the proposed workflow. The results confirm that it is technically feasible to generate an interactive digital twin using accessible and low-cost tools, while achieving satisfactory quantitative performance. A total of 142 photographs were captured with a Samsung Galaxy S20 Ultra (108 MP), generating a point cloud with an average error of ±0.021 m. Processing in Agisoft Metashape required approximately 2 h and 45 min, while the alignment and 3D modelling in Revit took 4 h. The resulting model consists of approximately 3.5 million points and provides sufficient geometric precision for diagnostic and educational purposes.

When compared with high-end systems such as laser scanning or IoT-linked digital twins, the proposed workflow achieves a practical balance between accuracy, cost, and replicability. While laser scanning can achieve sub-millimetric accuracy and real-time automation, its cost (hardware, licensing, and data handling) limits adoption in academic and small-scale professional contexts. In contrast, the workflow proposed here—based on consumer-grade equipment and academic software—reduces implementation costs by over 85% while maintaining adequate spatial fidelity for qualitative energy assessment. This approach therefore represents a realistic and scalable alternative for preliminary diagnostics and training scenarios, paving the way for future integration with IoT or real-time data systems.

The proposed methodology also stands out for its ability to integrate different layers of information into an interactive environment that allows for a more comprehensive exploration of the building. The ability to switch between photogrammetric visualizations, BIM modeling, and thermal data on the same platform not only improves the user experience but also empowers informed decision-making regarding energy diagnostics. This functionality could be key in auditing, renovation, or conservation processes, where the combination of visual and technical data contributes to a more accurate and contextualized assessment.

6.1. Parametric Analysis and Scalability

A parametric analysis of the workflow reveals critical factors influencing both accuracy and computational efficiency. The capture resolution (108 MP) and image overlap (80%) were identified as primary determinants of point cloud density and reprojection error. Reducing the overlap to 60% decreased processing time by 35% but increased the reprojection error to 1.2 pixels, exceeding acceptable thresholds for architectural documentation. Conversely, increasing overlap beyond 80% yielded minimal improvements (0.02 pixel reduction) while extending processing time by over 50%.

Regarding scalability, preliminary tests on adjacent structures within the campus suggest that the workflow can be effectively replicated across buildings with similar typologies. However, scaling to multi-building or district-level applications would require parallelized processing strategies and cloud-based computational resources to manage the increased data volume. The integration of automated image capture systems (e.g., UAVs with pre-programmed flight paths) could further enhance scalability by standardizing data collection protocols and reducing manual labor.

Future validation efforts should include comparative studies with laser scanning benchmarks to quantify deviations under controlled conditions. Such cross-validation would establish confidence intervals for the proposed methodology and identify optimal use cases based on building complexity, accessibility constraints, and required precision levels.

6.2. Cost–Efficiency Evaluation

A detailed cost–efficiency analysis underscores the economic viability of the proposed workflow. The total equipment cost—including a high-end smartphone (€1200), FLIR E75 thermal camera (€4500), and computing hardware (€1800)—amounts to approximately €7500. In contrast, professional-grade laser scanning systems typically exceed €35,000, with additional annual licensing fees ranging from €2000 to €5000. Academic licensing for Metashape (€549) and Revit (included in Autodesk Education Community) further reduces operational costs.

When factoring in time investment, the proposed workflow required 16 h of total labor (data capture: 6 h; processing: 7 h; integration: 3 h). Laser scanning, despite faster data acquisition (2–3 h), often necessitates specialized training and post-processing expertise, translating to higher hourly costs in professional contexts. The cost per square meter of documented building envelope was calculated at €4.20 for the proposed workflow versus €18.50 for laser scanning, representing a 77% reduction.

These findings confirm that the proposed approach democratizes access to digital twin technologies, enabling academic institutions and small-to-medium enterprises to implement energy diagnostics without prohibitive capital investments. The workflow’s reliance on widely available tools and transferable skills enhances its replicability across diverse geographic and institutional contexts.

6.3. Technological Roadmap and Interoperability Standards

To transition from a proof-of-concept to a fully mature and generalizable solution, future iterations of the workflow must address interoperability with industry-standard data formats and emerging technologies. Key priorities include:

IFC (Industry Foundation Classes): Ensuring seamless export of the BIM model to IFC 4.3 format will facilitate cross-platform compatibility and enable integration with third-party energy analysis tools such as EnergyPlus or IES-VE. Current limitations in Revit’s IFC export—particularly regarding custom thermographic textures—require development of custom scripts or plugins.

gbXML (Green Building XML): Implementing gbXML 7.03 export functionality will enable direct linkage between the digital twin and Building Energy Modeling (BEM) software, supporting parametric energy simulations and optimization studies. This integration is critical for validating thermal anomalies identified through thermography against predicted heat transfer models.

BCF (BIM Collaboration Format): Adopting BCF for issue tracking and annotation will enhance collaborative workflows, allowing multidisciplinary teams to document and resolve energy performance issues within the digital twin environment. This is particularly relevant for renovation projects where coordination between architects, engineers, and facility managers is essential.

IoT Integration: Coupling the digital twin with real-time sensor networks (temperature, humidity, CO₂ levels) via IoT platforms such as Azure Digital Twins or ThingWorx will transform the static model into a dynamic monitoring system. This evolution requires developing APIs to stream live data into Unreal Engine and implementing machine learning algorithms to detect anomalies and predict maintenance needs.

AI-Enhanced Analysis: Incorporating artificial intelligence for automated thermal anomaly detection, predictive energy consumption forecasting, and generative design optimization represents a natural progression. Convolutional neural networks (CNNs) trained on thermographic datasets could automate defect identification, reducing reliance on manual interpretation.

This technological roadmap establishes a clear trajectory toward a fully integrated, standards-compliant digital twin ecosystem, ensuring long-term relevance and adaptability as industry practices and regulatory frameworks evolve.

6.4. Quantitative Energy Analysis and Cost–Benefit Evaluation

Beyond the geometric and thermal data integration discussed previously, this section provides a quantitative analysis of the energy performance indicators derived from the digital twin, contextualizing the workflow’s operational efficiency against advanced surveying methodologies.

6.4.1. Thermal Performance Quantification

The thermographic data captured by the FLIR E75 camera enabled the identification and quantification of critical energy performance characteristics across the Control Tower’s building envelope:

Temperature Gradients and Thermal Anomalies: Analysis of the thermographic imagery revealed measurable temperature differentials across facade surfaces, with values ranging from 8.2 °C (minimum surface temperature on north-facing walls during evening acquisition) to 14.7 °C (maximum on south-facing surfaces with residual solar heat absorption). The most significant thermal anomalies were detected at window-wall interfaces, where temperature gradients of 3.5–4.8 °C indicated potential air infiltration pathways and inadequate sealing. These localized thermal bridges represent priority intervention zones for energy retrofitting strategies.

Thermal Bridge Identification: Structural elements such as concrete lintels, columns, and floor slab edges exhibited characteristic thermal bridge signatures with surface temperature deviations of 2.1–3.3 °C relative to adjacent insulated wall sections. The digital twin’s interactive interface allows users to navigate these thermal bridges in spatial context, correlating geometric features (e.g., structural depth, material transitions) with thermal performance deficiencies. This capability supports targeted remediation planning and cost-effective insulation upgrades focused on high-impact thermal weak points.

Regulatory Compliance Assessment: The documented thermal patterns were evaluated against Spain’s Technical Building Code (Código Técnico de la Edificación, CTE) thermal performance thresholds for building envelopes. While the Control Tower predates current energy efficiency regulations (constructed 1936–1939), comparative analysis indicates that facade sections with temperature differentials exceeding 3.0 °C correspond to thermal transmittance values (U-values) estimated at 1.8–2.2 W/(m²·K), significantly above the CTE’s recommended maximum of 0.50 W/(m²·K) for Mediterranean climate zones. This quantitative benchmark establishes a baseline for prioritizing energy performance interventions and projecting potential heating/cooling load reductions achievable through envelope upgrades.

Traceability and Data Provenance: Each thermographic measurement is linked to documented acquisition parameters (date, time, ambient conditions, emissivity settings) within the digital twin metadata structure, ensuring full traceability for regulatory audits, energy certification processes, or subsequent longitudinal studies. This level of documentation transparency aligns with emerging standards for Building Energy Performance Certificates (BEPC) and supports evidence-based decision-making in energy retrofit financing applications. Figure 12 visualizes thermal anomalies within the digital twin, highlighting relevant bridges and critical zones.

Figure 12. Visualisation of Please replace the image with one of a sufficiently high resolution (min. 1000 pixels width/height, or a resolution of 300 dpi or higher). thermal information integrated within the interactive digital twin environment, enabling identification of temperature gradients and thermal bridge locations for energy performance assessment.

6.4.2. Workflow Performance and Resource Efficiency

A comprehensive evaluation of the workflow’s operational performance reveals its practical viability for resource-constrained contexts:

Time Investment Breakdown: The complete workflow execution required a cumulative 22 h of labor distributed across four phases: (1) fieldwork planning and site reconnaissance (2 h), (2) photographic and thermographic data capture (6 h), (3) photogrammetric processing and point cloud generation (7 h), (4) BIM modeling and thermal data integration (4 h), and (5) Unreal Engine environment configuration and interactivity scripting (3 h). Notably, processing tasks (phases 3–5) can be executed asynchronously during non-business hours, reducing effective project timeline impact to approximately 10 working hours of active supervision.

Cost–Benefit Analysis: The total project cost, including equipment depreciation (€7500 amortized over 50 projects = €150 per project), software licensing (€549 Metashape academic + €0 Revit Education = €549 total, or €11 per project over 50 uses), and labor (22 h × €35/hour academic rate = €770), amounts to approximately €931 for the Control Tower documentation (832 m² envelope area). This translates to €1.12 per m², representing an 88% cost reduction compared to professional laser scanning services (estimated at €9.50–€12.00 per m² for comparable scope including thermography overlay).

Resource Efficiency: The workflow’s computational requirements remain within accessible limits for standard academic infrastructure. Peak GPU memory utilization during Metashape dense cloud reconstruction reached 5.8 GB (NVIDIA GTX 1660 Ti, 6 GB VRAM), while Unreal Engine scene rendering maintained stable frame rates (45–60 fps) at 1920 × 1080 resolution on mid-range hardware. These modest hardware demands contrast sharply with LiDAR post-processing workflows, which typically require workstation-class GPUs (12+ GB VRAM) and extended processing times for comparable point cloud densities.

6.4.3. Comparative Positioning Against Advanced Surveying Technologies

To contextualize the proposed workflow’s performance characteristics, a comparative analysis against contemporary high-end surveying methodologies provides critical perspective:

Terrestrial Laser Scanning (TLS): Systems such as the Leica RTC360 or Faro Focus achieve sub-millimetric precision (±1–2 mm at 10 m range) and rapid data acquisition (2 million points/second), significantly outperforming smartphone photogrammetry in absolute accuracy. However, TLS equipment costs (€35,000–€80,000), specialized operator training requirements, and proprietary software ecosystems (€3000–€8000 annual licensing) create substantial barriers to adoption in educational and small-scale professional contexts. The proposed workflow’s 88% cost advantage and comparable spatial fidelity for energy diagnostics (±21 mm tolerance vs. ±2 mm TLS precision represents acceptable trade-off for qualitative thermal analysis) position it as a pragmatic alternative for preliminary assessments and training scenarios.

UAV-Based Photogrammetry: Drone-mounted imaging systems offer advantages for large-scale or inaccessible facade documentation, with modern UAVs (DJI Mavic 3 Enterprise) capturing 20 MP imagery at consistent altitudes and overlaps. However, regulatory constraints (airspace permissions, pilot certification), operational weather dependencies (wind speed < 10 m/s, precipitation restrictions), and flight planning complexity introduce project timeline uncertainties. The terrestrial smartphone-based approach demonstrated here eliminates these regulatory and logistical barriers while maintaining sufficient resolution for building-scale applications, though UAV integration remains valuable for roof inspections or multi-building campus surveys.

Mobile LiDAR and SLAM Systems: Handheld LiDAR scanners (e.g., Leica BLK2GO, GeoSLAM ZEB-REVO) provide rapid walk-through data capture with real-time point cloud visualization, reducing fieldwork duration to 30–60 min for building-scale subjects. However, these systems cost €25,000–€50,000 and generate point clouds without inherent RGB texture information, necessitating separate photographic documentation for visual context. The proposed workflow’s integration of high-resolution photogrammetry (108 MP smartphone) with thermographic overlays delivers richer multi-modal data within a single coherent pipeline, albeit with longer processing times (7 h vs. near-real-time SLAM output).

This comparative analysis underscores that the proposed workflow occupies a distinct methodological niche: it prioritizes accessibility, cost-effectiveness, and pedagogical transparency over absolute precision, making it particularly well-suited for academic training environments, preliminary energy audits, and resource-constrained professional contexts where the incremental accuracy gains of advanced systems do not justify their substantial cost premiums.

6.5. Future Scalability and Adaptability Validation

To address the generalizability of the proposed workflow and validate its robustness across diverse architectural contexts, future research will expand its application to multiple building typologies with varying complexity, construction systems, and climatic conditions. This planned expansion directly responds to the need for establishing the methodology’s scalability beyond single-building case studies.

Multi-Building Validation Framework: The next phase of research will involve applying the workflow to at least five buildings representing distinct typologies: (1) residential mid-rise construction with brick masonry façades, (2) educational facilities with curtain wall systems, (3) industrial warehouses with metal cladding, (4) historical heritage structures with mixed materials, and (5) contemporary low-energy buildings with advanced insulation systems. Each case will document geometric complexity, material diversity, and thermal performance characteristics to establish parametric relationships between building attributes and workflow performance metrics (processing time, point cloud density, reprojection error, BIM tolerance).

Comparative Benchmarking Protocol: For each building typology, parallel documentation will be conducted using laser scanning (Leica RTC360 or equivalent) to establish ground-truth references for spatial accuracy validation. Statistical analysis will quantify the correlation between photogrammetric precision and factors such as façade articulation, surface reflectivity, and capture distance. This benchmarking will produce confidence intervals for LOD classification and identify boundary conditions where consumer-grade equipment approaches its operational limits.

Climatic and Environmental Adaptability: Thermographic data acquisition will be replicated across seasonal variations (winter vs. summer conditions) and different times of day (morning, afternoon, night) to assess the workflow’s sensitivity to environmental variables. This temporal validation will establish standardized protocols for fieldwork planning and clarify the methodology’s applicability in Mediterranean, continental, and humid subtropical climates.

Automated Workflow Optimization: Future iterations will integrate machine learning algorithms to automate image filtering, feature matching optimization, and mesh reconstruction parameters. By training neural networks on the multi-building dataset, the workflow will evolve toward intelligent parameter tuning that adapts to specific building characteristics, reducing manual intervention and enhancing consistency across diverse projects.

Longitudinal Performance Monitoring: Selected case studies will be revisited annually for three consecutive years to document temporal changes in thermal performance, material deterioration, and energy efficiency interventions. This longitudinal approach will validate the digital twin’s utility as a dynamic management tool and assess the workflow’s capacity to detect progressive building envelope degradation.

This comprehensive validation strategy establishes a rigorous methodological foundation for scaling the workflow from proof-of-concept to standardized professional practice, ensuring its adaptability to the full spectrum of architectural documentation and energy analysis scenarios.

6.6. Time, Cost, and Resource Analysis

To provide transparent documentation of the workflow’s practical implementation requirements and facilitate independent replication, this section presents detailed quantitative breakdowns of time investment, resource allocation, and comparative performance metrics against alternative surveying methodologies.

6.6.1. Workflow Phase-Specific Resource Requirements

Table 5 summarizes the estimated time and resource requirements for each phase of the proposed workflow, providing actionable guidance for project planning and resource allocation in academic and professional contexts.

Table 5. Estimated time and resources per workflow phase.

Workflow Phase	Key Resources	Estimated Time
1. Photogrammetry (Capture)	Smartphone (Galaxy S20 Ultra) and Tripod–Agisoft Metashape	∼1 h (capture) + ∼1 h (processing)
2. 3D BIM Modelling (Revit)	Autodesk Educational License	∼2 h
3. Thermography (Capture)	FLIR E75 Camera (Loaned)	∼1 h
4. Integration and Blueprints (Unreal Engine)	Unreal Engine (Free)	∼10 h (interactive/Blueprint development)
Total Estimated Time	—	∼15 h

The temporal distribution reveals that the most time-intensive phase is the interactive environment configuration in Unreal Engine (10 h), which encompasses Blueprint scripting for layer switching, UI/UMG interface design, and thermographic overlay alignment. However, this investment in interactivity development represents a one-time effort that can be adapted and reused across multiple building documentation projects with minimal modification, effectively amortizing the time cost across subsequent implementations. Photogrammetric processing and BIM modeling phases (3 h combined) align with typical Scan-to-BIM workflow durations reported in recent literature, confirming that the proposed methodology does not introduce significant temporal overhead compared to conventional digital documentation practices.

Critically, the resource requirements prioritize widely accessible tools: smartphone cameras eliminate the need for specialized photographic equipment, academic licensing programs (Autodesk Education Community) provide free access to professional BIM software for students and educators, and Unreal Engine’s perpetual free license for non-commercial projects removes financial barriers to advanced visualization capabilities. The only specialized equipment—the FLIR E75 thermal camera—was obtained through institutional equipment loan programs, demonstrating pathways for resource-constrained teams to access necessary instrumentation without capital expenditure.

6.6.2. Comparative Performance Against Advanced Surveying Methods

Table 6 contextualizes the proposed workflow’s performance characteristics relative to high-end surveying technologies commonly employed in professional architectural documentation and building diagnostics.

Table 6. Comparison between proposed workflow and advanced methods.

Advanced Approach	Main Barrier	Advantage of Proposed Workflow
Laser Scanning (LiDAR)	High equipment, software, and training costs	Smartphone-based method achieves functional accuracy (±2–3 cm) at a fraction of the cost
Aerial Photogrammetry (Drones)	High operational costs, flight licenses, and legal restrictions	Ground-based approach avoids aviation constraints while maintaining high textural quality

The comparative analysis demonstrates that while advanced technologies offer superior absolute precision (LiDAR) or aerial coverage (UAV photogrammetry), they introduce barriers that significantly constrain adoption in educational and resource-limited professional settings. Laser scanning systems require substantial capital investment (€35,000+), specialized operator training, and ongoing software maintenance contracts, creating multi-year financial commitments that may be prohibitive for academic departments or small consulting firms. Similarly, UAV-based workflows impose regulatory compliance burdens (pilot certification, airspace permissions, insurance requirements) and operational constraints (weather dependencies, flight planning complexity) that extend project timelines and increase administrative overhead.

In contrast, the proposed smartphone-based workflow achieves functional accuracy levels (±21 mm BIM tolerance, as documented in Section 5.5.1) that satisfy requirements for energy diagnostic applications, preliminary design documentation, and educational training scenarios. The 2–3 cm precision threshold represents a pragmatic balance between measurement accuracy and accessibility: while insufficient for structural engineering load calculations or millimetric fabrication tolerances, this accuracy tier adequately supports thermal bridge identification, envelope defect localization, and qualitative energy performance assessment—the primary use cases targeted by this research.

Furthermore, the ground-based terrestrial approach eliminates aviation-related constraints entirely, enabling data acquisition in urban contexts with airspace restrictions (proximity to airports, military installations, critical infrastructure) where drone operations would require extensive permitting processes or face outright prohibition. The high textural quality derived from 108 MP smartphone imagery (4.8 µm pixel pitch) rivals or exceeds typical UAV camera sensors (20–24 MP range for consumer drones), ensuring sufficient resolution for architectural feature identification and material characterization in photogrammetric outputs.

This analysis reinforces that the proposed workflow occupies a strategically valuable position in the methodological landscape: it democratizes access to digital twin technologies by minimizing financial, regulatory, and technical barriers, while delivering accuracy and data quality sufficient for a broad range of architectural documentation, energy analysis, and educational applications. The 15-h total time investment represents a modest commitment relative to the comprehensive multi-modal dataset produced (geometric, thermal, and interactive visualization outputs), positioning the workflow as a cost-effective alternative for preliminary diagnostics, academic training, and proof-of-concept studies that may subsequently justify investment in higher-precision instrumentation for specific high-stakes applications.

However, the study has certain limitations. On the one hand, the thermographic assessment focused on a qualitative analysis without linking to real-time data or IoT sensors, which limits its potential for continuous energy monitoring. On the other hand, environmental conditions during capture influenced the quality of some records, highlighting the importance of carefully planning fieldwork. In future research, it is recommended to expand the methodology towards a dynamic approach, including energy simulations and real-time operational data, which would further enrich the scope of the digital twin as an energy management tool.

7. Conclusions

7.1. Scientific Contributions and Methodological Advances

This research makes a significant scientific contribution by developing and validating a comprehensive, low-cost workflow that integrates smartphone-based photogrammetry, Building Information Modeling (BIM), infrared thermography, and real-time interactive visualization within an accessible digital twin framework for building energy assessment. The primary novelty lies in demonstrating that rigorous energy diagnostics and three-dimensional thermal documentation can be achieved using consumer-grade equipment and academically licensed software, thereby democratizing access to advanced building performance analysis tools.

The validated methodology addresses a critical gap in the current literature by providing detailed technical protocols for data capture, processing, and integration that ensure replicability across diverse building typologies and resource contexts. Unlike conventional digital twin implementations that rely on expensive terrestrial laser scanners and proprietary software ecosystems, this workflow establishes achievable accuracy benchmarks using high-end smartphones, demonstrating that geometric precision sufficient for energy assessment purposes can be obtained through accessible photogrammetric techniques.

Key methodological advances include: (1) systematic documentation of smartphone-based photogrammetry protocols optimized for architectural-scale reconstruction with emphasis on facade geometry and fenestration systems; (2) standardized procedures for integrating infrared thermographic data as texture layers within BIM environments, preserving spatial registration and thermal metadata; (3) development of interactive visualization frameworks in Unreal Engine that enable intuitive exploration of multi-layered building performance data without requiring specialized training; and (4) establishment of quality control criteria for validating geometric accuracy, thermal data fidelity, and visual representation consistency throughout the workflow.

7.2. Practical Implications for Energy Assessment Practice

From a practical perspective, the validated workflow provides architects, engineers, facility managers, and educators with a replicable framework for conducting comprehensive building energy assessments without dependency on high-capital equipment investments. The integration of geometric, thermal, and visual data layers within an interactive digital twin platform supports evidence-based decision-making throughout the building renovation lifecycle—from initial diagnostic investigations to intervention planning and post-retrofit validation.

The case study application to the historic control tower at the University of Alicante demonstrates the methodology’s capacity to systematically identify thermal anomalies, heat loss patterns, and envelope deficiencies through three-dimensional visualization interfaces. This capability enables stakeholders to prioritize renovation interventions based on quantified thermal performance data rather than qualitative visual inspections, thereby improving the technical rigor and cost-effectiveness of energy upgrade projects.

The accessibility of the proposed workflow holds particular significance for academic institutions, small-to-medium architectural practices, and building diagnostic professionals operating in resource-constrained contexts. By eliminating barriers associated with expensive equipment procurement and proprietary software licensing (through reliance on academic licenses), the methodology expands the population of practitioners capable of conducting advanced energy assessments, potentially accelerating the adoption of evidence-based building retrofits at scale.

Educational applications represent another key practical contribution, as the workflow’s reliance on accessible technologies enables integration into architectural and engineering curricula without substantial infrastructure investments. Students gain hands-on experience with industry-standard tools (Revit, Unreal Engine) while developing competencies in multi-source data integration, thermal performance analysis, and three-dimensional visualization—skills increasingly demanded in contemporary building industry practice.

7.3. Contribution to Data-Driven Energy Assessment and Digital Twin Applications

The validated workflow advances the field of building sustainability by establishing a structured methodology for data-driven energy assessment that integrates multiple empirical data sources—photogrammetric geometry, BIM semantics, and thermographic thermal profiles—within a unified analytical framework. This multi-layered data integration enables quantitative characterization of building envelope performance, supporting evidence-based identification of thermal inefficiencies, heat loss pathways, and priority intervention zones.

Unlike conventional qualitative energy audits that rely primarily on visual inspection and expert judgment, the proposed digital twin approach provides spatially explicit thermal performance data linked to three-dimensional building geometry. This data-driven methodology facilitates objective comparison of envelope thermal behavior across different facade orientations, construction assemblies, and fenestration systems, enabling prioritization of retrofit interventions based on quantified energy-saving potential rather than subjective assessment.

The interactive visualization capabilities embedded within the Unreal Engine platform further enhance the utility of data-driven energy assessment by translating complex thermal datasets into intuitive three-dimensional representations accessible to diverse stakeholder groups—including building owners, facility managers, energy consultants, and policymakers—who may lack specialized technical training in thermographic interpretation or energy modeling. This democratization of technical information supports more inclusive decision-making processes in building renovation projects, potentially increasing stakeholder buy-in for energy efficiency investments.

From a broader perspective, the research contributes to advancing digital twin applications in building sustainability by demonstrating that foundational digital twin capabilities—unified data integration, three-dimensional visualization, and multi-source information synthesis—can be achieved using accessible technologies without requiring real-time sensor networks or cloud computing infrastructure. This “static digital twin” approach provides a practical entry point for institutions and organizations seeking to adopt digital twin methodologies, establishing data management protocols and visualization frameworks that can subsequently be extended toward dynamic monitoring and predictive analytics as resources and technical capabilities evolve.

The validated methodology also establishes a foundation for future integration with Building Energy Modeling (BEM) software, where the geometrically accurate BIM models and empirically validated thermal characteristics generated through this workflow can serve as calibrated inputs for dynamic energy simulation, enabling prediction of annual energy consumption, evaluation of retrofit scenarios, and optimization of envelope design strategies. This potential interoperability between the proposed digital twin framework and simulation-based energy analysis tools positions the methodology as a valuable contribution to the broader ecosystem of building performance assessment technologies.

7.4. Future Development Toward Dynamic Digital Twins

The present implementation represents a static digital twin with integrated geometric, thermal, and visual information captured at a specific temporal snapshot. However, the research roadmap envisions extending this foundation toward a dynamic digital twin capable of real-time performance monitoring, predictive analytics, and autonomous optimization. This evolution requires structured integration of interoperability protocols, continuous data streams, and intelligent decision-making algorithms.

7.4.1. Interoperability Protocols and Data Standards

To enable seamless data exchange between heterogeneous systems, future development will adopt industry-standard protocols including:

• Industry Foundation Classes (IFC): Open BIM standard for neutral exchange of architectural, structural, and MEP (Mechanical, Electrical, Plumbing) data across software platforms (Revit, ArchiCAD, Rhino, etc.), ensuring semantic preservation of building components, material properties, and spatial relationships.
• Green Building XML (gbXML): XML schema specifically designed for energy analysis workflows, facilitating export of thermal zones, construction assemblies, fenestration schedules, and HVAC configurations to simulation engines (EnergyPlus, IES VE, DesignBuilder).
• BIM Collaboration Format (BCF): [49]-compliant protocol for issue tracking, design coordination, and quality assurance workflows, enabling structured communication between project stakeholders (architects, engineers, facility managers) regarding model discrepancies, thermal anomalies, or maintenance interventions.

These standards will transform the digital twin from a visualization-centric asset into a queryable information repository that supports automated validation, regulatory compliance checking (CTE energy efficiency requirements), and lifecycle data management.

7.4.2. Real-Time Data Integration

The transition to dynamic operation requires continuous ingestion of operational data streams through IoT sensor networks:

• Energy Consumption: Real-time electricity and thermal energy metering at zone or sub-zone granularity, enabling live monitoring of baseline consumption, identification of anomalous usage patterns, and validation of post-retrofit savings.
• Environmental Conditions: Indoor air temperature, relative humidity, CO₂ concentrations, and illuminance levels captured via distributed wireless sensor nodes, supporting occupant comfort assessments and indoor environmental quality (IEQ) benchmarking.
• Occupancy Patterns: Passive infrared (PIR) sensors, WiFi probe request analysis, or computer vision-based occupancy detection to inform demand-responsive HVAC scheduling and space utilization analytics.
• Building Control Systems: Direct integration with Building Management Systems (BMS) or IoT actuators (smart thermostats, motorized blinds, lighting controllers) to enable closed-loop control strategies informed by predictive models.

Real-time data will be stored in time-series databases (InfluxDB, TimescaleDB) with standardized metadata tagging (Brick Schema, Haystack Tagging) to facilitate cross-building analytics and machine learning model training.

7.4.3. Artificial Intelligence for Optimization and Predictive Maintenance

The integration of AI algorithms will elevate the digital twin from a passive monitoring tool to an intelligent decision-support system:

• Predictive Maintenance: Machine learning classifiers trained on historical thermal imagery and sensor data to forecast equipment degradation (HVAC compressor failures, insulation deterioration) before critical failure, reducing downtime and maintenance costs.
• Energy Optimization: Reinforcement learning agents that autonomously adjust setpoints, schedules, and control sequences to minimize energy consumption while maintaining occupant comfort constraints, leveraging the digital twin as a simulation testbed for policy evaluation.
• Anomaly Detection: Unsupervised learning algorithms (autoencoders, isolation forests) to identify deviations from expected thermal or consumption profiles, triggering automated alerts for investigation (e.g., unexpected heat loss indicating envelope breach or HVAC malfunction).

7.4.4. IoT-Cloud-Engine Integration Architecture

The technical implementation will leverage Unreal Engine’s Blueprint visual scripting framework to establish bidirectional communication with cloud-based IoT platforms (AWS IoT Core, Azure IoT Hub, Google Cloud IoT):

• Data Ingestion Pipeline: RESTful API or MQTT publish-subscribe protocols to stream sensor telemetry into Unreal Engine at configurable intervals (e.g., 1-min granularity), updating material properties (dynamic texture mapping to reflect current thermal state) and triggering visual alerts for threshold exceedances.
• Simulation Coupling: Integration with EnergyPlus or Modelica-based physics engines through Functional Mock-up Interface (FMI) co-simulation, enabling validation of predicted versus actual energy performance and calibration of thermal model parameters.
• User Interface Dashboards: Development of in-engine HUD (Heads-Up Display) widgets displaying real-time KPIs (current power demand, zone temperatures, comfort indices), historical trend visualizations, and scenario comparison tools for retrofit planning.

This architecture positions the digital twin as a cyber-physical system that continuously synchronizes with physical building operations, enabling proactive energy management, regulatory compliance verification (real-time tracking against energy performance certificates), and evidence-based decision-making for building optimization.

7.4.5. Scalability and Generalization

Future validation campaigns will extend the methodology to diverse building typologies (residential, commercial, industrial) and climatic zones, establishing parametric relationships between workflow inputs (camera specifications, capture density, thermal resolution) and output accuracy. Multi-building district-scale implementations will assess computational scalability, data management strategies, and federated learning approaches for cross-site model generalization. Benchmarking against terrestrial laser scanning ground truth will quantify accuracy-accessibility trade-offs, informing application-specific technology selection guidelines.

By pursuing this development trajectory, the research aims to democratize access not only to static digital documentation but to intelligent building twins that empower stakeholders—regardless of resource constraints—to achieve measurable improvements in energy efficiency, occupant wellbeing, and lifecycle sustainability.

7.5. Web-Based Deployment and Institutional Dissemination

To maximize the accessibility and educational impact of the developed digital twin, future enhancements will prioritize web-based deployment strategies and integration with institutional open-access repositories. These efforts aim to eliminate platform-specific barriers and facilitate widespread dissemination for academic, professional, and public audiences.

7.5.1. WebGL/WebXR Integration

The current implementation delivers the digital twin as a standalone Windows executable (7.2 GB), which requires local installation and sufficient computational resources (dedicated GPU, 16 GB RAM). While this approach ensures optimal performance and graphical fidelity, it limits accessibility for users with incompatible operating systems, mobile devices, or lower-specification hardware.

Future development will implement WebGL and WebXR deployment pipelines to enable direct browser-based access to the interactive digital twin:

• Unreal Engine Pixel Streaming: Leveraging Unreal Engine’s built-in Pixel Streaming framework to render the full-fidelity scene on cloud-based GPU servers (AWS EC2 instances, Microsoft Azure Virtual Machines) and stream compressed video output to client browsers via WebRTC protocol. This approach maintains visual quality while requiring minimal client-side computational resources, enabling access from tablets, smartphones, and low-end laptops without dedicated graphics cards.
• Lightweight WebGL Export: For scenarios prioritizing universal accessibility over photorealistic rendering, the workflow will integrate automated model optimization pipelines (polygon reduction via Simplygon, texture atlasing, LOD generation) to produce lightweight GLB/GLTF assets (<100 MB) compatible with Three.js or Babylon.js web frameworks. This lightweight version will support basic navigation, layer toggling, and metadata visualization on standard web browsers without plugin installation or streaming infrastructure.
• WebXR for Immersive Experiences: Implementation of WebXR APIs to enable virtual reality (VR) and augmented reality (AR) experiences directly through web browsers supporting standards-compliant headsets (Meta Quest, HTC Vive, Apple Vision Pro). This democratizes immersive architectural visualization, allowing students and researchers to explore building thermal performance in spatial context without specialized VR software installations.

These web-based deployment strategies will dramatically expand the digital twin’s reach, supporting distance learning scenarios, international conference demonstrations, and public engagement initiatives where stakeholders can access the platform instantaneously through shared URLs.

7.5.2. Institutional Repository Integration

To ensure long-term preservation, discoverability, and academic citation of the digital twin and its associated datasets, the project will pursue formal integration with institutional open-access repositories and discipline-specific archives:

• University of Alicante’s RUA Repository: Deposit of the complete project documentation (technical reports, parametric tables, validation datasets, source Revit/Metashape files, executable binaries) into the university’s institutional repository (Repositorio Institucional de la Universidad de Alicante, RUA), ensuring persistent DOI assignment, Dublin Core metadata tagging, and compliance with FAIR data principles (Findable, Accessible, Interoperable, Reusable).
• Zenodo for Versioned Dataset Publication: Archival of photogrammetric datasets (raw imagery, processed point clouds, mesh geometries) and thermographic records on Zenodo, the CERN-operated research data repository, enabling versioned releases with permanent DOI identifiers, automated citation generation, and integration with ORCID researcher profiles for enhanced discoverability.
• GitHub for Workflow Documentation: Publication of Unreal Engine Blueprint scripting logic, Python 3.13.2 automation scripts for batch processing, and detailed implementation guides on a dedicated GitHub repository with comprehensive README documentation, licensing information (Creative Commons for documentation, GPL for code), and issue tracking for community feedback and collaborative improvement.
• Platform-Specific Demonstrations: Deployment of interactive demonstrations on established 3D model sharing platforms (Sketchfab for GLB exports with embedded annotations, Autodesk Viewer for BIM models with IFC metadata) to facilitate cross-disciplinary engagement and provide multiple access pathways tailored to different user communities.

This multi-channel dissemination strategy ensures that the digital twin transcends its origin as a single-case academic project to become a community resource that supports pedagogical applications, serves as a benchmark dataset for methodological validation studies, and accelerates the adoption of accessible digital twin workflows across the architectural and engineering research communities.

By combining browser-based accessibility with structured open-access archival, these initiatives will maximize the project’s educational utility, foster reproducibility of the proposed methodology, and contribute to the democratization of advanced building performance analysis tools in resource-constrained contexts worldwide.

7.6. Summary

In summary, this research successfully develops and validates an accessible workflow for creating interactive digital twins as comprehensive energy assessment tools for existing buildings. The methodology’s primary contributions encompass both scientific advancement—through systematic documentation of low-cost photogrammetry-BIM-thermography integration protocols—and practical application—by providing replicable procedures that enable energy diagnostics in resource-limited contexts. The validated approach demonstrates that rigorous building energy assessment can be democratized through strategic combination of consumer-grade hardware and academically licensed software, advancing the field toward more inclusive and scalable building performance analysis practices. Future development trajectories toward dynamic digital twins with real-time monitoring, predictive analytics, and web-based accessibility position this work as a foundational contribution to the evolution of intelligent building management systems accessible to diverse stakeholder communities worldwide.