BIM and BEM Interoperability–Evaluation of a Case Study in Modular Wooden Housing

Abstract

Building information modelling (BIM) is the first step towards implementing Building 4.0, where virtual reality and digital twins are key elements. The use of unmanned aircraft systems (UAS/drones) to capture data from buildings is nowadays a very popular method, so a methodology was developed to digitally integrate the photogrammetric surveys of a building into BIM, exclusively with the use of drones. Currently, buildings are responsible for 40% of energy consumption in Europe; therefore, the interconnection between BIM and building energy modelling (BEM) is essential to digitalize the construction sector, increasing competitiveness through cost reduction. In this context, the BlueWoodenHouse Project aims, among other activities, to characterize the solutions/systems of building materials and monitor the temperature, relative humidity and CO₂, as well as energy consumption, of a single-family modular wooden house located in the north of Portugal, with 190 m² and three users. Thus, the experimental monitoring results, of this case study, were used to validate the numerical model developed in the DesignBuilder simulator, which includes the building envelope’s 3D geometrical data obtained by one of those aircraft, in order to demonstrate the usefulness of drones for the optimization of solutions, from the energy point of view.

1. Introduction

1.1. Background

In most developed countries, particularly in Europe, about 90% of the time is spent indoors, mostly at home. This value will tend to grow with the increase in teleworking, as recommended by the European Union [1]. As residential buildings are increasingly places for living and working, they must have suitable indoor air quality and thermal comfort requirements for their occupants.

The importance of indoor thermal comfort in residential buildings has been widely studied [2,3], as can be seen from the different studies existing in national and international literature, taking into account the impacts that lower thermal comfort has on health [4], productivity [5], and energy consumption [6,7,8].

In Portugal, as in the rest of southern Europe, the most used construction system for residential buildings is based on the intensive use of reinforced concrete and brick masonry, which results in high thermal inertia [9,10]. However, in recent years, alternative and more industrialized construction systems have proliferated throughout the world and started to gain market share in relation to more traditional solutions.

In this context, modular buildings have attracted greater attention from potential buyers, investors, and the construction industry itself, taking into account the numerous advantages of this construction method over traditional alternatives, namely: they need a smaller number of workers on site, construction processes are faster and safer, construction processes are more ecological and environmentally friendly and with more predictable completion times, there is less waste of resources, presenting numerous advantages in constructions in which units built are repeated [10].

Consequently, and since the research on building energy performance has been gaining more attention, it becomes useful to integrate and manage information from BIM (building information modelling) and BEM (building energy modelling) models [11]. Thus, it is beneficial to survey the 3D geometry of existing buildings (“as-is”) for subsequently performing energy analysis [12,13,14,15]. The use of new technologies in the study of buildings has increased, including drones [12,16,17] (also known as UAS, unmanned aircraft systems, and UAV, unmanned aerial vehicles). These devices bring various advantages, such as reduced human effort and more security, access to hard-to-reach places, reduced time of surveying, low cost, etc. [16,17,18,19]. These can be applied at different stages of a building’s life cycle [18,19,20], with different levels of adoption [19]. According to Albeaino and Gheisari [18], the main emphasis is on progress monitoring and on pre-construction activities such as site planning and site mapping.

1.2. Related Research

In their literature review about the use of UAVs for civil engineering purposes, Freeman et al. [19] noted the emerging application of using these aircraft to assist in the creation of digital models. Indeed, linking the potential of UASs with BIM along the construction’s value chain has already been discussed [21], for instance, for more efficient project management and quality control [20], and for construction safety management [22]. While the integration of these technologies has already been discussed in the literature, including using the BIM model afterwards for virtual reality purposes, the integration of UASs, BIM, and the IoT (internet of things) sensing technologies are not fully explored yet [23].

In order to create accurate geometric 3D models for energy studies of as-is buildings, reverse engineering methods must be applied [12,13,14,15]. For example, Rocha et al. [14] studied the thermal comfort of a bus station during the winter season. The study aimed to choose the best thermal comfort retrofitting solution possible. A laser scanner was used to acquire the geometrical data, which was used to create the BIM model before the BIM-to-BEM process. Through the BEM model, alternative scenarios were presented and analysed, not only in terms of the energy performance of the station but also an economical overview.

Gigliarelli et al. [24,25,26] used a laser scanner and terrestrial photogrammetry to establish an approach to analyse the energy performance of buildings of historical interest. Implementing heritage BIM (HBIM) models for this purpose is considered a challenge as it is necessary to simplify the building’s geometry to run simulations successfully on proper software. No sensors were used in this methodology to compare the simulations with reality, though. Nagy and Asharaf [27] went further by creating a real-time platform to visualise the energy performance of heritage buildings using smart sensing, applied to a case study in Egypt. Laser scanner and photogrammetric data were acquired to develop the HBIM model. The BEM model created afterwards was calibrated using real-time data from the sensors installed inside the building.

Despite being useful for on-site data acquisition for further BIM integration, laser scanners demand significant human action, thus, drones can replace them in many cases [21]. For instance, Maiolatesi et al. [28] performed an energy analysis to assess how neighbouring buildings influenced energy simulations by comparing two BEM models (one with and the other without the surrounding buildings) with the actual energy consumption of the building under study. A drone was used to acquire external geometry and a laser scanner detected the internal geometry of the apartments. These were then imported into BIM before following a BIM-to-BEM framework.

Gorzalka et al. [29] present an approach to automatically generate a BEM model of an existing building from UAV photogrammetric point clouds. Python was used in this process, creating a CityGML file to be simulated on BEM software (Modelica in this case). The building under study was monitored for 21 days, both the interior and the exterior. Then the simulated interior temperatures of the model were compared with those from the monitoring campaign so that further calculations could be made after calibration.

The drone-exclusive framework proposed by Benz et al. [30] assesses the as-is building’s energy performance based on photogrammetric and thermographic data, which is used for a BIM-to-BEM approach calibrated by sensors installed inside and outside the building. The thermographic data allowed for the estimation of U-values, and the BEM model was obtained by converting the IFC file from the BIM software (Graphisoft ArchiCAD) to an IDF file, which was used for the energy simulation on TRNSYS.

1.3. Research Significance

According to KPMG International, the level of drone adoption in the construction sector is around 26% worldwide, with 45% starting to use them [31]. This value is still far from the Portuguese reality, for example, which is only 4% [32]. As such, this article intends to demonstrate the usefulness of drones for optimising solutions from the energy point of view.

The development of digital models of existing buildings, in particular of models that are suitable for simulating energy performance, has long been a common albeit laborious task. These models are expected to become increasingly relevant with the greater prominence of digital twins in academic and industrial applications. Therefore, reducing the cost and time required to obtain suitable as-is models of buildings is a relevant research goal.

This paper explores the use of surveying and modelling tools to obtain building information and energy models. A detailed workflow is presented in the next section, using drones to perform photogrammetric surveys of buildings, BIM to establish a suitable information model, and BEM to develop energy simulations. This workflow is tested with a case study about a building in the north of Portugal. The case study is presented in Section 2, and the results are described in Section 3.

2. Materials and Methods

2.1. Methodology for Photogrammetric Surveys of Buildings with Drones

In recent decades, there has been an increasing interest in studying as-is buildings’ energy performance. This analysis can be improved by integrating BEM with BIM. BIM models can include information that is then traditionally introduced manually in a BEM model, thus, providing cost and time savings [11]. Therefore, to conduct an energy analysis of an existing building, it is interesting to collect data on its 3D geometry prior to that evaluation [12,13,14,15].

This work will follow a previously created methodology that digitally integrates the data acquired by the photogrammetric survey into BIM and BEM environments solely using a drone [33,34,35,36]. This method’s quantitative approach will then be further extended through a case study located in the north of Portugal.

2.1.1. Reconnaissance and Preparation

It is essential to follow a methodology to initiate the photogrammetric survey process with UASs. The first step is to consult the local legislation and plan the flight in terms of trajectories and local conditions [37,38,39,40], in addition, if possible, to seek useful information about the building to be surveyed, to aid the BIM modelling process [14,37,38,40,41].

Applicable Legislation

The first procedure when planning a drone survey is to ascertain if the aircraft can or cannot fly in that area. The local airspace legislation may restrict the flight height or, in a worst-case scenario, completely prohibit it if the operation is close to airports, for instance. In Portugal, these areas are defined by the Portuguese Civil Aviation Authority (ANAC) [42].

Countries under the European Union Aviation Safety Agency (EASA) jurisdiction must follow common rules, including drone flights. The Commission Delegated Regulation (EU) 2019/945 [43] states that the aircraft is classified depending on the UAS characteristics (not only, but in particular, its weight and maximum speed). Then, the Commission Implementing Regulation (EU) 2019/947 [44], divides them into categories and subcategories according to the operational risks they pose, which will influence the distance the drone can be from people and buildings. If they present a high operational risk (i.e., from the “specific” and “certified” categories), security measures and a safety plan must be adopted after being previously approved by the competent authorities.

The Commission Implementing Regulation (EU) 2019/947 also states that the UAS, the remote pilot, and the drone operator (the “owner”) must all be registered on the national aviation authority website and that only EASA-certified remote pilots can legally fly these aircraft, as long as they pass the qualification courses and exams [44].

Assessment of Site Conditions—Obstacles and Weather Conditions

Even when it is permitted to fly a drone in a given geographic area, other issues must be considered, including overflight restrictions from places such as schools, public parks, hospitals, power plants, prisons, or military facilities. Applications such as Guardian by Altitude Angel (available as a website and a mobile application) [45] can be used to assess this information prior to the flight. Furthermore, flight planning approaches must take into account constraints such as:

−

The UAS battery and the existing airspace regulations [46];

−

The existence of obstacles and the correct GPS signal for proper georeferencing [17];

−

Wind speed and direction [47];

−

Precipitation, which causes noise in the images due to water droplets and consequently erroneous results [47];

−

Adverse weather conditions such as storms and the presence of other manned or unmanned aircraft in the vicinity [48] present relevant safety risks;

−

Attacks from nearby birds [49];

−

Proximity to electromagnetic fields, especially if it is intended to fly in the vicinity of high voltage cables, because electromagnetic interference may disturb the correct functioning of the equipment [40].

In addition to the equipment issues, remote pilots should consider their piloting conditions. I’M SAFE (Illness, Medication, Stress, Alcohol, Fatigue, and Emotion) [50] is a self-assessment checklist that pilots can follow to determine their readiness to fly safely.

Flight Planning

When an obstacle-free flight is possible, it is suggested to plan the survey with applications capable of automating the operation, such as the DroneDeploy [51] or Drone Harmony [52]. Applications such as these can set, for instance, the flight path, the desired percentage of image overlay, and the image capture angle of the cameras.

Rakha and Gorodetsky [17] analysed the most commonly used flight heights, distances to façades, and overlap percentages in the literature. They conclude that, empirically, a minimum overlap percentage of 70% in grid-shaped flights and 90–95% in elliptical-shaped flights is recommended. Besides the flight planning constraints mentioned previously, which need to be taken into account for a safe operation, the following protocol is a suggestion for data acquisition with a UAS:

−

Flight height should be higher than the largest building in the area, and it is advisable not to exceed this height by 1.5 times [12];

−

Distance to the façade—there are several alternatives, but the suggestion is that the remote pilot will suit this distance according to the building’s surroundings (presence of walls or other buildings, for example);

−

Flight plan can be circular, elliptical, grid-shaped [17], or Y-shaped (multiple flights overlapping each other) [12];

−

Camera angle—camera angles outside the 5°–60° range should be avoided to prevent radiation reflections [47]. For photogrammetric purposes, an angle of 45° is suggested [37];

−

Overlap of about 90% is suggested for adjacent images [17,53].

Project Elements

As suggested by Shibasaki [40], the collection of information about the building through documents such as design elements, construction records (such as the work diary), photographic or video records, and as-built drawings are useful for the correct geometric survey of the building.

These design elements are important for geometric modelling and for the correct acquisition of the properties of the materials to be later integrated into the BIM model [41]. Indeed, the proper acquisition of these properties is an essential step for quantitative approaches, such as the quantification of the thermal properties of the materials, i.e., the coefficient of thermal transmission (U) and the thermal resistance (R), fundamental for the analysis of the energy performance of a building [54,55]. It should be taken into account that these properties affect the thermal comfort conditions of a building. Commonly, these properties are considered constant for the same element, which is not always true because the maintenance conditions of a building may change throughout its life cycle [55].

2.1.2. Image Collection Using Drones

In the phase after the initial study of the selected building, the equipment must be calibrated and, if necessary, the site prepared with methods to support georeferencing [38,39,40,56]. After following the previous steps, the drone operation is set to begin.

Equipment Calibration

In order to start the flight and consequently collect images with the UAV, calibrations must first be carried out for the safe and correct operation of the device. This step includes the calibrations of the cameras of the vision systems and infrared positioning sensors, the compass, and the inertial measurement unit (IMU)—accelerometer, gyroscope, thermometer, and barometer. One must follow the aircraft’s user manual for a correct procedure.

Georeferencing Support Methods

It is possible to adopt targets as a way to visually mark on-the-ground auxiliary points for georeferencing purposes (GCPs—ground control points) and evaluate the accuracy of the photogrammetric reconstruction in comparison to the reality (checkpoints) on appropriate photogrammetric software [37,57]. These must be clearly visible from the air and evenly distributed in the area that will be surveyed, as well as ensuring they are firmly placed so that they cannot be easily moved [58]. These markers are commonly used when a high degree of absolute precision is needed, such as for topographic surveys [57,58,59,60,61].

Drones equipped with RTK (real-time kinematic) or PPK (post-processed kinematic) may dismiss the use of GCPs, as they will not substantially influence the absolute accuracy of the georeferencing [57,60,62]. Even if a survey is GCP-free, it is still possible to produce accurate 3D results [61,62]; however, issues with the orientation, scale or absolute position of the reconstruction may occur as a consequence [62].

If no real geographical coordinate points are required (i.e., the absolute accuracy of the project), points that appear in multiple images (known as tie points) can replace the GCPs to improve the relative accuracy of the photogrammetric project (i.e., how close geometries in the reconstruction compares with the reality, for example, a distance) [58,62].

A high degree of absolute accuracy will often be unnecessary for photogrammetric models that will be incorporated into 3D BIM modelling software [61]; thus, physical markers may be dispensable [59]. If needed, tie points of the building (e.g., the corner of a window) can be a valid replacement to improve the quality of the photogrammetric model [38,62]. It is recommended to use between five and ten points in the reconstruction process and assess its precision [62,63].

Image Collection

With the preflight stages completed, the photogrammetric drone survey may begin. One must remember that it is crucial to have all the up-to-date documents, identifications, and insurance to fly the drone legally when performing the survey to avoid possible wrongdoings [33,34,36].

2.1.3. Processing and Analysis of the Collected Images

After the image acquisition, the process of digitally rebuilding the building is performed with photogrammetric data reconstruction software, and the generated point cloud should be cleaned [37,38,40]. This process strongly influences the subsequent BIM modelling [37,39,56].

Photogrammetric Model (Point Cloud and 3D Mesh Model)

At this stage of the methodology, a representation of the current state (“as-is”) of the surveyed building will be created based on the georeferenced images collected during the aerial survey [37,38]. Proper software based on structure-from-motion (SfM) algorithms must be used at this stage to provide satisfactory results regarding accuracy [37]. SfM uses the acquired images’ EXIF metadata to automatically calibrate, match, and align them so that a sparse point cloud can be created. Subsequently, the cloud can be densified by applying a multi-view-stereo (MVS) algorithm [37,60,64].

The creation of a 3D BIM model depends on the accuracy of the software used in this process; therefore, the best available software must be used at this step [65]. Pix4Dmapper and Agisoft Metashape, which have been tested in previous UAS-based point cloud reconstruction operations [64], are among the SfM-based software suggestions.

Point Cloud Cleaning

Point clouds generated by the processes described above typically contain unwanted points from the background or the surroundings of the objects of interest (e.g., buildings, trees, sky, etc.), which should be removed [38,40,66]. This process of cleaning the unwanted points which do not belong to the model is performed manually [37] through specific cleaning tools in the software, such as crop and delete [33,67].

Due to this cleaning process, the point cloud will require less computer power, preventing software lag, as the files will be smaller. Also, as a result of this process, it will be easier to perceive the building itself [66,67], which will ease the modelling process afterwards [66].

2.1.4. Integration and Modelling in BIM Environment

In order to incorporate the point cloud created from the photogrammetric data into BIM environment, it is first necessary to export it from the SfM software used. Although Autodesk Revit is one of the world’s most used BIM software [15,37], it has some particularities regarding its point cloud importation procedure; nevertheless, the subsequent modelling process itself is similar to other BIM solutions. This manual BIM modelling stage is performed by using the imported point cloud as a guide [14,28,37,39,66].

Import to a BIM Authoring Software

Before importing the point cloud into the BIM authoring software, one must export it in a format that can be properly integrated for further modelling. According to Opincar [65], the LAS (LASer) format is recommended because it is suitable for point clouds created from photogrammetric methods and LiDAR data [65], as it can support any three-dimensional x, y, z dataset [68]. This format can be exported from several applications, including Agisoft Metashape and Pix4Dmapper [33]. Note that this binary file format was created and is maintained as a public standard by the American Society for Photogrammetry and Remote Sensing (ASPRS) [68].

When Autodesk Revit is used, some intermediate processes are required to convert the point cloud data to a readable format by this software. The LAS file must be imported into Autodesk Recap first and then exported as either an rcs or rcp file, so that Autodesk Revit can successfully read the point cloud [37,69].

The procedure described above was validated by Martinez-Carricondo et al. [37] during their reconstruction work using a UAV-based photogrammetric survey to create an HBIM model of a historical building in Níjar, Almería, Spain.

BIM Modelling

After importing the point cloud into the BIM authoring application, its orientation should be adjusted, and the building’s levels should be set [37]. The model must be oriented correctly to obtain accurate energy simulations [70].

The point cloud can then be used to guide the modelling process for the visible building elements, such as external walls, roofs, doors, and windows [37,71]. These elements’ correct dimensions and locations are critical for the later energy analysis [71]. The material’s properties in the BIM model are a key factor for this analysis; therefore, those properties must be considered during the modelling process [41].

2.1.5. Energy Simulations on BEM Software

In order to carry out energy simulations, first it is necessary to analyse the formats of data transfers between the BIM modelling applications and the programs where the energy simulations will take place [72,73]. In this section, some suggestions are presented to carry out energy simulations from a photogrammetric survey with a UAV, briefly addressing the main data transfer formats between BIM and BEM and procedures to carry out the simulations.

The photogrammetric data acquired during the survey is fundamental, as the geometrical data serves as the basis for the BEM model. The building’s physical properties are all linked to its geometry, whereas the systems and the building’s operational data are inserted manually and edited directly in the BEM software [66,74].

From BIM to BEM—Data Transfer Formats

Several file formats are available to transfer information between BIM and BEM software to perform energy simulations, such as IFC (Industry Foundation Class), gbXML (Green Building eXtensible Markup Language) [11,70,71,72,73], CityGML (City Geography Markup Language) [29,56] and IDF (Input Data File) [11,30,74]. IFC and gbXML are considered the most commonly used BIM-to-BEM formats [11,72,75]; nonetheless, gbXML has been proven to have superior performance for this process, as it can correctly transfer the (simplified) geometry, materials, and thermal zones of the model [11,75].

gbXML is a data file that stores information according to a hierarchy capable of integrating thermal data with geometric data of a given building [70]. However, the interoperability between BIM and BEM software is not seamless; therefore, interoperability issues in this BIM-to-BEM process must be checked and corrected [11,66,72,73,76]. In particular, it is recommendable to simplify the building geometry [24,26,70,75,77], to avoid collisions between elements, since they create openings in the BEM model [70].

From BIM to BEM—Energy Simulations and Software

Before exporting to BEM software, “spaces” must be created in the BIM model covering the entire building volume, which is a key step for energy simulations [70,72,75,77]. A space corresponds to a closed volume that is thermally simulated and to which thermal properties are attributed [70]. These spaces are of vital importance as they will create the energy analytical model [72,74]; therefore, if gaps or geometric flaws exist in this space, the model may not even be exported correctly [66,72].

“Rooms” should also be created, as useful information for energy simulations is linked to them, such as the occupancy of the building [77,78].

There are many energy simulation software, but not all are compatible with BIM applications [70]. DesignBuilder, for instance, is a BEM application that accepts gbXML files from a BIM authoring software with a building’s geometry as input. The properties needed to perform the simulation can be added in DesignBuilder to avoid information losses in the transference process [66,78], or through a plug-in between BIM and BEM applications [75,77].

When building a BEM model, one must ensure that the model meets the information requirements to proceed with the energy simulations. These requirements include the building geometry, material properties, HVAC systems, lighting features, local weather data, and operations data (occupancy, schedules, etc.) [70,74,79].

Figure 1 summarizes the process described previously, applied to the case study presented hereafter.

2.2. Case Study

2.2.1. Experimental Monitoring

The case study is a single-family wooden house located in the north of Portugal, with a covered area of 190 m², in full use, that is, inhabited by a family of three people (two adults and one child). The energy sources associated with this house are: (i) natural gas for cooking; (ii) electricity for heating, cooling, ventilation, and domestic hot water; and (iii) heat pump for sanitary water heating.

The monitoring plan included measuring at 15 min intervals the temperature, relative humidity (RH), and CO₂ concentration (ppm), using 7 HOBO MX CO2 sensors distributed over different rooms of the house (kitchen, living room, social zone, room 1, room 2, room 3 and bathroom 3). A HOBO UX100 sensor (which measures temperature and RH) was placed on the exterior, with an acquisition every 15 min, and the results of this sensor allowed the creation of the weather file used in the numerical simulations, based on the weather file of Esposende, where the building is located. Figure 2 shows the floor plan of the house as well as the monitoring positions and AC equipment. The sensors were placed outside the direct area of influence of the AC equipment to avoid disturbances.

Monitoring started in November 2021 and this work presents an analysis of the data focused on the air temperature obtained between December 2021 and November 2022 (1 year of the experimental campaign).

2.2.2. Numerical Simulation

In this work, the numerical simulation was carried out using DesignBuilder, an energy simulation software from EnergyPlus. For this purpose, the numerical model was also divided into different thermal zones that represent an approximation of the housing divisions, as seen in Figure 2.

DesignBuilder was chosen for the building energy simulation (BES) because it is an interface for EnergyPlus, which enables the simulation to be completed within the DesignBuilder interface and utilises a complex system of modules and algorithms to estimate the energy needs for heating, cooling, ventilation, hot water production, lighting, and appliance electrical needs. The surface heat balance manager, air heat balance manager, and building systems simulation manager represent the main three modules interacting with the integrated solution manager, which simultaneously handles all module interactions. The thermal comfort was evaluated using the adaptive model for thermal comfort EN16798-1 [80], which offers acceptable indoor temperature limitations as a function of the exponential weighted running mean of the outdoor temperature.

The walls, roof, and floor are composed of oriented strand board (OSB) plates with thermal insulation. The exterior walls have the following components from the exterior to the interior face: slate tiles, air layer (0.02 m), vapour permeable membrane, OSB (0.012 m), mineral wool (0.09 m), OSB (0.012 m), vapour barrier and, finally, plasterboard as an internal finish.

The floor is in contact with the ground and consists of a concrete slab (0.2 m), followed by waterproofing and vapour barrier, OSB (0.012 m), mineral wool (0.09 m), OSB (0.012 m), and a floating wood floor (0.012 m). Finally, the roof, which is sloped and not accessible, has slate tiles, vapour permeable membrane, air layer (0.02 m), OSB (0.012 m), mineral wool (0.18 m), OSB (0.012 m), vapor barrier, and as an interior finish a suspended ceiling system with plasterboard.

Table 1. Materials used in opaque elements and their properties.

Material	Thickness [m]	λ [W/m·K]	ρ [kg/m3]	Cp [J/kg·K]
Walls
OSB	0.012	0.120	600	1700
Mineral wool	0.09	0.035	80	1030
Plasterboard	0.013	0.350	750	840
Floor
Concrete	0.200	2.000	2400	940
Mineral wool	0.090	0.035	80	1030
OSB	0.012	0.120	600	1700
Floating wood floor	0.012	0.130	400	1300
Roof
OSB	0.012	0.120	600	1700
Mineral wool	0.018	0.035	80	1030

presents the physical and thermal properties of the materials used in the walls, floor, and ceiling. All the windows are double-glazed, with 6 mm and 4 mm glass, with a low emissivity coating, plus a 16 mm argon layer with an aluminium frame with thermal break. The thermal conductivity U_w is 1.1 W/mK, with a solar transmittance of 0.83 and a solar heat gain coefficient of 0.63. The shading devices are metallic external venetian blinds.

An occupancy schedule based on an approximation of the user habits was considered, together with the number of users, the type of indoor activity, and the structure of the typical working week, i.e., considering that the weekdays are different from weekends. Cooling and heating set point temperatures were also defined (see

Table 2. Occupancy, and AC parameters for DesignBuilder simulations.

	Hourly Time	Presence Rate	Heating Setpoint (°C)	Heating Setback (°C)	Cooling Setpoint (°C)	Cooling Setback (°C)
Weekdays	19:00–7:00	1.0	18	14	26	30
	8:00–18:00	0.6
	9:00–17:00	0.4
	10:00–16:00	0.0
Weekends	19:00–7:00	1.0	18	14	26	30
	8:00–9:00	0.9
	10:00–15:00	0.5
	16:00–18:00	0.6

). In this way, the AC turns on and off based on indoor temperatures and the defined schedules. Heating and cooling are activated during working hours and during unoccupied periods the system turns on only under “extreme” circumstances, according to the setback temperatures (see Table 2).

A specific lamp power of 3 W/m²/100 lx was assumed during the occupancy schedule. The lighting levels adopted were bathroom—75 lx, bedrooms—200 lx, living room and kitchen—300 lx, and hall—100 lx. Finally, as previously mentioned, the weather file for Esposende, Braga was edited using the temperature and RH measured on the outside sensor.

3. Results and Discussion

3.1. From UAS to BEM

3.1.1. Photogrammetric Survey with a Drone

The building under study is located in Esposende, Braga, in the north of Portugal. As a way of acquiring data from the building, the methodology presented in Section 2 was followed. This continuous workflow of UAS-to-BIM and BIM-to-BEM methodologies had as its final objective the comparison and validation of the experimental monitoring results with those obtained by energy simulation through DesignBuilder.

First, one has to verify which airspace rules are in force in the place under study, as drone flight might face restrictions. That is not the case for this building as it is located outside restricted or prohibited geographical areas for the flight of these aircraft stipulated by ANAC (the Portuguese Civil Aviation Authority), so the flight is authorization-free up to 120 m in height [42,81]. However, authorization is required to capture images by AAN (the National Aeronautical Authority) to avoid violating the personal data protection policy, due to being able to capture images with the UAS camera [82,83].

Apart from the national rules in force, European regulations have to be followed too [43,44]. It is worth remembering that, besides the recommendations stated in 2.1.2, the European drone rules also mandate that the drone, the remote pilot, and the operator (the “owner”) must all be registered on the national aviation authority website [44], in this case, ANAC’s [42].

After verifying that a drone could fly at that location, the weather conditions were assessed. The building’s surroundings were checked before the survey for possible obstacles through the Guardian by Altitude Angel application, which identified that cables from a nearby transformer station passed above the house. Thus, Google Street View images were analysed, and a site visit was made before the drone operation to create a suitable survey plan.

After this verification, the survey operation was planned in the Drone Harmony Mobile application with a flight height of 15 m (for safety reasons, above the cables), with the triple grid mode, 90% overlap between images, and a 45° drone camera angle.

The survey was carried out on 12 July 2022, around 10 a.m., with sunny weather conditions and no wind, taking about an hour to capture 246 photographs for the 3D reconstruction of the building. The operation was conducted by an EASA-certified drone pilot (one of the authors), as only certified pilots can fly a drone legally [44].

The chosen aircraft, a DJI Mavic 2 Enterprise Dual, was calibrated according to its user manual [84] prior to the survey. This 899 g quadcopter is equipped with a 1/2.3″ CMOS sensor, 12 MP resolution, and since it also has an uncooled VOx microbolometer sensor, thermographic information can also be acquired, although this was not used for this study.

Figure 3 shows the drone during the aerial photogrammetric survey of the building as previously scheduled.

3.1.2. 3D Reconstruction and BIM Model (UAS-to-BIM)

In this work, Agisoft Metashape was the SfM-based photogrammetric software chosen to generate a 3D point cloud of the surveyed area from the multiple images taken. If desired, it is possible to create 3D polygonal meshes from point clouds (called 3D mesh models), which have a better representation and completeness by comparison.

In Figure 4, it is possible to observe the dense point cloud created in the software used (with approximately 55.66 million points) and the 3D mesh model of the north façade (with roughly 4.56 million faces and 2.28 million vertices).

No physical GCPs were used throughout this operation as the project did not require a high degree of absolute precision (the real coordinates on the planet); however, seven checkpoints from the building envelope were marked in the photogrammetric software to assess the relative accuracy of the reconstruction. These points estimated the following errors: RMSE_XY = 1.43 cm (RMSE_X = 0.99 cm and RMSE_Y = 1.03 cm), RMSE_Z = 1.09 cm and RMSE_TOTAL = 1.80 cm. These represent the distance between the physical position of the points calculated by the drone’s onboard GNSS receiver (global navigation satellite system—GPS + GLONASS in this case) and the reconstruction’s estimated location for these specified points. The first represents the longitude (X) and latitude (Y) root mean square error (RMSE), the second is correlated with the altitude (Z) and the last indicates the total error, also known as the vector sum [37]. These RMSE values are in accordance with some other recent photogrammetric studies with UAVs which also used Agisoft’s software, such as [37,85,86].

Before exporting the point cloud, unnecessary points for the BIM model were manually removed (e.g., trees, other buildings, the garage, etc.) through Agisoft Metashape’s point removal functions. This process reduced the number of points to roughly 1.94 million and, then, the project was exported in LAS format.

Autodesk Revit was the selected BIM software for this study, as it is one of the most used BIM software in the world [15,37]. In order to import the point cloud to Autodesk Revit, a conversion from the previous format to a file readable by this software was performed in Autodesk ReCap, as Agisoft’s software cannot export to those formats [87]. After converting the LAS file into an rcp file, the cloud was imported into Autodesk Revit. Subsequently, the rotation and orientation were adjusted. Once this process was finished, the BIM modelling process of the dwelling (namely its envelope) started, using the point cloud as a reference.

As mentioned in Section 2.1.5, “spaces” and “rooms” must be created in Autodesk Revit so that energy simulations can be conducted afterwards. After creating the energy analytical model, the gbXML file was exported for further developments in the BEM software chosen to proceed with this research. Figure 5 shows the finalized BIM model of the dwelling under study.

3.1.3. BEM Model and Energy Simulations (BIM-to-BEM)

The energy simulation software used was DesignBuilder, which is capable of importing gbXML files from Autodesk Revit with the building geometry and, later, capable of inserting the necessary properties for the simulation, avoiding interoperability issues between this software [66,78]. The model imported into DesignBuilder can be seen in Figure 6 and the materials used in this case study and their respective properties were inserted, assembling the various constructions of the model.

3.2. Experimental Results

Figure 7 shows the air temperature (T) evolution measured inside the seven thermal zones: living room, kitchen, room 1, room 2, room 3, bathroom 3, and social zone. It is also possible to compare the indoor conditions with the respective exterior temperature fluctuation.

The monitoring results showed a clear alignment between the different indoor temperature conditions throughout the monitoring period. It is also possible to observe that, in general, the temperatures inside the different rooms are very close, identifying, occasionally, moments when the temperature inside the living room is higher.

The thermal comfort assessment was carried out based on the adaptive model proposed by the EN 16798-2 [88] standard, which defines the operating temperature limits that lead to thermal comfort as a function of the exponentially weighted average temperature of the outside air. In this context, it was considered that the air temperature is a good approximation of the operating temperature since the air velocity is low and there are no relevant radiative effects. Figure 8 shows the representation of temperature records and comfort limits, separately for each thermal zone and considering that the building is in category II (normal level of expectation). In order to facilitate the interpretation of the results presented in Figure 8, the percentage of time in discomfort was calculated in each thermal zone, obtaining 2.1% in room 3, 2.1% in the bathroom, 3.1% in room 2, 3.1% in room 1, 5.2% in the living room, and 6.3% in the kitchen.

3.3. Numerical Results

As an example of the application of the methodology described above, the numerical results obtained for the temperature profiles during the winter and summer seasons, and a comparison with the experimental results of the monitoring are presented. Figure 9 and Figure 10 show a comparison of the temperature profiles obtained for different thermal zones, during the two seasons analysed.

The results, for the summer season, show a good agreement between the experimental (monitoring) and numerical results, where the absolute error between the two temperature profiles does not exceed 6.7%, on average, i.e., approximately 1.5 °C.

Finally, the results, for the winter season, presented in Figure 10, also show a good agreement between the experimental (monitoring) and numerical results, where the absolute error between the two temperature profiles does not exceed 9.3%, on average, i.e., approximately 2 °C.

3.4. Discussion

This work demonstrates the usefulness of existing drone-based hardware and software solutions to digitize the built environment geometry for BIM-to-BEM purposes through aerial photogrammetry. These solutions are particularly convenient as they support highly automatable workflows. They are often more agile and easier to use than alternative solutions, such as laser scanners [21].

In the proposed workflow, the model geometry for the BEM simulations is acquired during the survey to recreate the as-is building in a BIM environment, which is then exported to the energy simulation software. While some simulation properties can be manually entered and changed directly in the BEM software, the building’s geometrical properties are obtained directly from the BIM model [66,74]. Hence, the importance of the survey.

However, since the UAS is only used to acquire the external geometry of the building, further adjustments must be performed to the digital models after the survey is performed. As described in 2.2.1, the building under study has sensors in different locations. The thermal zone created in the BIM model can be divided into the BEM software according to the sensor locations. These new thermal zones represent an approximation of the rooms for further analysis.

The numerical results of the building energy simulation showed a good agreement between the monitoring results of the experimental campaign, and the numerical simulation for the winter and summer seasons. These numerical results validate the use of the BEM model for building energy simulations where different scenarios can be explored, in order to improve energy efficiency, avoiding time-consuming and expensive work.

4. Conclusions

The use of drones in the construction sector has been growing, bringing with it several useful advantages. Although drone adoption in this sector is already widespread worldwide, in Portugal its incorporation is still far behind. As such, this article aims to highlight the usefulness of adopting these aircraft for the optimization of solutions, from the energy point of view, of a dwelling in the national territory.

The method used in this paper follows a continuous UAS-to-BIM and BIM-to-BEM workflow, starting at the aerial photogrammetric survey of the building under study, and ending with the simulations in a BEM authoring software, in this case, DesignBuilder. Prior to these simulations, one must first create a BIM model based on the point cloud created using SfM software through photogrammetry, as this model will then be exported to the BEM application. Agisoft Metashape and Autodesk Revit were the selected SfM and BIM software for this case study, respectively.

After exporting the model as a gbXML file format from Autodesk Revit to DesignBuilder, the various simulation parameters were entered, referring to the materials used, as well as the occupation and use times of the relevant equipment for the thermal study. Seven sensors were also placed inside the different rooms of the building and a sensor was placed outside to measure temperature and relative humidity.

The comparative analysis between the monitoring results and the numerical simulation for the winter and summer season show a good agreement.

In summary, this work analyses and discusses the BIM model which facilitates the definition of the geometric and parametric model to be used in the BEM, and validates the use of the BEM model for energy simulations where different scenarios can be explored, avoiding time-consuming and expensive work.

As scan-to-BIM solutions based on the combined use of reality capture solutions continue to improve, it is expected that this methodology can be further automated. Solutions based on laser scanning have been developed and tested in earlier initiatives developed by this research team [89,90]. These tools will be applied to point clouds generated from photogrammetric surveys in the future.