From BIM to BEM—Modern Thermal Simulations Using a Building Information Management Model: A Case Study

Abstract

This article raises awareness of Building Information Management (BIM) and its significance for Construction 4.0. BIM is often mistakenly understood only as a 3D model of a building object, but its true potential lies in the information associated with the model (e.g., mechanical and physical properties, costs, etc.). Models can subsequently be used in the building energy management (BEM) at all stages of the building object’s life cycle. This article focuses on the possibility of creating a model using available libraries in the Czech Republic provided by manufacturers and suppliers of building materials and the subsequent use of the model for energy modelling. The results obtained from computational modelling are then compared with real values measured on a timber construction located in Ostrava (Czech Republic). These results show that properly configured BIM modelling allows faster data processing while maintaining the quality of outputs and results. Additionally, there is potential to eliminate common pitfalls in the design and subsequent processing of thermal assessments of building objects.

1. Introduction

At the beginning of the 21st century, the conception of the information modelling of buildings (BIM) was developed by the U.S. company Autodesk for industrial building applications. BIM in building engineering can now be seen as a favoured and well-known approach, as a method of interpreting large datasets, used in all life cycle stages of buildings—from their preparation, design, and construction to building management.

The main contribution of BIM is in integrating all data related to a building into one unified digital model, which facilitates communication between the particular participants of the building life cycle and suppresses the risk of error. BIM makes related processes more effective and reduces costs, with the possibility of the simulation and prediction of the behaviour of a building under real operation. Its functions overlap these domains, offering wide possibilities of applications, such as building sustainability planning and their complete life cycle analyses [1,2,3].

In the early phase of investment project preparation, designers and architects must frequently perform detailed simulations to predict the ecological demands of such projects. The design process requires highly effective compatibility between CAD systems, special simulation software, and energy evaluation tools to support the easy transfer of data models and the results of the analyses. This technical synergy provides consistent and exact outputs for the complex evaluation of the environmental demands of objects and the further optimisation of their design parameters.

The design approach based on BIM in the building industry relies on digital 3D models, utilised throughout the entire life cycle of the project. Numerous studies have highlighted the ability of such approaches to improve the effectiveness of building projects. BIM supports digital creation, exchange, and the transfer of all project information. Specifically, 6D BIM [4] connects internal and external energetic parameters. Architects often use 6D BIM to evaluate selected energy parameters to optimise the building energy design. However, in research directed toward 6D BIM coupled with building energy management (BEM), questions remain, namely (i) the energy simulation process in the BIM environment typically suffers from data loss during transfer, and (ii) the simulation tools are rather complicated and require users to have advanced programming knowledge and further specialised skills. Especially for high public investment projects/contracts, the proper utilisation of 6D BIM seems necessary in the early design phase.

The exploitation of BIM during the lifecycles of buildings is an inevitable trend in the building industry in the last few years. Numerous potential research areas have been stimulated by BIM, particularly the integration of BEM to establish energy simulations in large projects [5].

Several software packages have been developed over the last few decades to support the analysis of energy consumption in buildings, such as DOE-2, EnergyPlus, ESP-r, HASP, and DeST [6]. Of all these software tools, EnergyPlus, with its algorithms for dynamical simulations, modular architecture, and excellent scalability, can be seen as the standard for energetic simulations, the integration of thermal technical computations, as well as advanced algorithms for climate control in buildings, providing a key basis for the design of effective and environmentally optimal buildings [7,8]. Consequently, the design of such buildings needs to improve data exchange between BIM systems and BEM models. The strong motivation for this is the fact that the European building industry consumes approximately 40% of total energy production, and one of the most important factors influencing this consumption is the quality of the thermal envelopes of buildings [9,10].

The European directive [11] states that all new buildings are to be designed and constructed as carbon neutral from 2028, thus necessitating that contemporary building design methods must be substantially changed. The testing of new ecological building materials and components must be accompanied by innovative technologies to guarantee energy efficiency in the early project phase [12]. The design of high-energy performance buildings requires the creation of a shareable working environment, where architectonic and energy solutions can communicate easily and exchange all of the required information.

During the last two decades, BIM has been established as a tool with the potential to significantly transform the building industry [13]. With the increasing complexity of buildings, the control of building projects has also become more challenging [14]. In this context, BIM, as a multi-functional method, can support the whole building design process and contribute to constructing ecological buildings [15]. Its power resides in the very exact and detailed digital reproduction of all architectonic building elements. Such working practice brings benefits not only for the environment but also for the future users of the buildings because the possible energy savings cause a distinctive reduction in operation costs. At the virtual conference, Sustainable Places 2020, nine EU-supported research teams presented their results and innovations; all of those projects had their attention set on recent challenges related to the digital transformation in civil engineering [16]. Moreover, the building sector was evaluated as the least productive sector in Europe because of the difficulties with implementing digital innovations [17].

The principal aim of BIM, as an integral part of the building sector, is to improve the cooperation between various interested parties [18,19] and to increase the efficiency at the building sites and between various project phases [20]. BIM, as the key tool in cyber-physical systems, also improves the quality of the building objects during their whole life cycles, in accordance with the paradigm Industry 4.0. It can be taken as the necessary means for the interconnection of digital technologies with the building industry [21]. Its tools for the management of digital information make it possible to optimise and control the complete building process [22]. BEM, together with the BIM conception, then forms the basis for the certification of energy performance (EPC) of buildings. Such certifications must be user-friendly, reliable, and profitable from the economic perspective, for them to obtain the investors’ confidence [22]. Three categories of BEM can be distinguished: (i) the physically based (white box) ones [23], (ii) the purely data-driven (black box) ones [23], and (iii) the hybrid (grey box) ones [24].

Interoperability between BIM and BEM is the key step for decreasing costs and reducing the time needed to create models in the design phase. The potential of BIM is in making this creation more effective, thanks to the stored important data, such as building geometry, structure typology, and physical properties [25]. Iterative development of effective proposals in a shared environment is supported [26]. BEM tools are useful for understanding different ways to reduce energy consumption [27], thanks to computational energy simulations, analysis of real energy demands, and multi-criteria optimisation of project design [28]. Apart from evaluating BIM as a multi-disciplinary tool [29], the interoperability between BIM and BEM still contains unpleasant digital problems. The rather weak compatibility between BIM models and applications for energy simulations, repeated manual operations, and non-standardised processes result in data loss and incorrect interpretations, as identified by Sanhudo et al. [30]. Such difficulties force frequent substantial revision of the model, which is unnecessary additional work for the designer. Following these findings, Kamal and Memari [31] focused on developing standards needed for the above-mentioned interoperability.

Multiple file formats are currently used for data transfers from BIM to BEM, e.g., HTML, XHTML, bcXML, IFCXML, IFC, and gbXML. The last two formats mentioned are the most widely used open standards for such data exchange. IFC allows transferring complex data about a building project, but it is rather complicated and creates large files, while gbXML offers a more flexible and direct approach to energy analysis. Both standards transfer information about materials, HVAC (heating, ventilation, and air conditioning) systems, and thermal zones, but only gbXML also contains location data [32].

The communication between BIM and BEM depends on data exchange files, which are instrumental for information transfer. Both formats, IFC and gbXML, are supported by most software tools for energetic simulations [33]. Specifically, the IFC is an ISO standard (International Organization for Standardization) covering a large class of domains of civil engineering [34,35], supported by the buildingSMART platform [36]. Hitchcock [37] suggested the application of IFC for coupling the software tools for building design with tools oriented on energy evaluations (since 2002). The upgraded versions of IFC include numerous BIM standards, namely IFC4 (in use since 2013) includes approximately 800 entities, 258 attribute sets, and 121 data types. However, apart from good possibilities for analysing the thermal performance of buildings, numerous relevant software tools still have difficulties with data import and utilisation of the IFC.

A concurrent format, gbXML, was developed in 2002, thanks to the financial support of the U.S. Department of Energy and the companies Autodesk, ASHRAE, and Bentley Systems, as an open standard for BIM with good abilities to share information about buildings. This format can characterise all attributes of buildings, thanks to its 346 elements and 167 simple types; this is appreciated especially by experts in the area of environmental demands of buildings [38]. In the last several years, due to the official statistics of organisations managing standards connected with IFC and gbXML formats, 45 software packages support data transfer based on IFC, whereas only 15 packages support that based on gbXML. The mechanisms and forms of both IFC and gbXML descriptions of relevant information on environmental demands of buildings were analysed by Lin et al.; the mapping of entities between both formats [39] forms a base for mutual data conversion.

Numerous software applications for thermal numerical simulations, such as Green Building Studio, PKPM-Energy, HY-EP, DesignBuilder, or OpenStudio, rely exclusively on the gbXML format for data import. Nevertheless, the interaction process has many disadvantages, difficulties, and errors. For example, an exported file from the Autodesk Revit platform contains only information about the building geometry, whereas information on the thermal technical characteristics of specific building envelope layers is missing. Using the choice Rooms/Spaces, the conversion is more appropriate but the data on HVAC systems are still not available [3,40]. The mechanism of conversion from BIM to BEM has been discussed by many authors. More specifically, Yang et al. [41], using their analysis of existing methods of conversion from BIM to BEM, developed a method for transferring BIM models to 2D graphics, followed by an automatic reconstruction of information from BIM. Unfortunately, all geometrical calculations are performed in 2D; therefore, roofs, outer shading, and various special structures are not effectively included. Liang [42] tries to improve this to simulate energy consumption in green buildings, utilising selected advantages of the IFC format. The complex evaluation of the process of data import and the setting of software parameters for energy simulations in the above-introduced six software tools has been performed by Wang et al. [43], yet further improvements are still required [31]. As emphasised by Pezeshki et al. [29], both BEM and BIM rely on replaceable formats IFC and gbXML, which are able to store graphic information together with certain attributed data. However, in most (not quite simple) cases, BIM tools cannot export such information so that BEM can analyse and apply it [44]. The BIM/BEM interoperability study of Porsani et al. [45], summarising the main shortcomings of the process, contributes to the projects which use modelling from Revit and also from ArchiCAD, describing the utilisation of various formats, including IFC [46,47].

Modern energy modelling techniques introduce advanced approaches, such as artificial intelligence (AI) and machine learning (ML), for more accurate predictions of building energy demands. Ahsan et al. (2024) [48] and Merabet et al. (2021) [49] state that AI can significantly improve the efficiency of thermal comfort management and building energy performance. However, these techniques are often demanding in terms of input data and computational capacity. Moreover, their application requires extensive expertise, which limits their adoption in standard construction practice.

Generative algorithms and multi-criteria optimisation, as highlighted by Benaddi et al. (2024) [50] and Ascione et al. (2025) [51], enable the optimisation of building envelopes in terms of energy, and environmental and economic sustainability. While these methods provide detailed analysis and optimisation capabilities, they are not always compatible with standardised tools, such as BIM.

Furthermore, advanced simulation methods have been explored, including using OpenStudio, which utilises data transfer via gbXML. Lin et al. (2016) [39] mention that gbXML offers a flexible framework for data exchange between BIM and BEM but they point out its limited ability to capture complex physical processes, such as moisture transport. Our study contributes to this discussion by demonstrating how gbXML can be applied in timber structure simulations under specific local conditions. In this paper, the communication between BIM and BEM is demonstrated in the case study, accenting the interpretation and validation of results obtained by BEM. After the introductory Section 1, Section 2 presents the methods of both computational modelling and measurements of the experimental building object, located in Ostrava (Czech Republic). The results are shown in Section 3, followed by their discussion in Section 4. The brief evaluation of present results in Section 3 is supplemented by references to problems, which have not yet been resolved, interpretable as challenges for further research.

The available literature frequently mentions the limited interoperability between BIM and BEM systems, which leads to data loss and inefficiencies in energy simulations. This study, therefore, focuses on testing the use of the gbXML format to improve interoperability under conditions specific to Central Europe. At the same time, it contributes to a deeper understanding of the impact of construction solutions on the accuracy of energy simulations for timber structures—an issue that has so far been studied mainly at a general level, without specific case studies in local conditions. The findings of this study have the potential to improve design and simulation processes in practice, thereby contributing to reducing the energy demands of buildings, which aligns with current European directives.

The primary objective of this article is to assess the effectiveness of the gbXML format as a tool for data transfer between BIM and BEM environments. The study focuses on identifying key limitations and challenges associated with the interoperability of these platforms while also exploring ways to optimise this process. Through a case study of a timber structure, the extent to which gbXML can enhance the accuracy of energy simulations and facilitate design processes in building energy performance is analysed.

The main focus is on the following:

(1)

Verifying model-derived data and their validation using experimental measurements from a real timber structure.

(2)

Formulating recommendations to improve interoperability and efficiency in data transfer between BIM and BEM platforms, which could be applied in both practice and future research.

(3)

Bridging theoretical knowledge about gbXML with practical insights from a real case study. The study’s results can serve as a foundation for further research aimed at developing data exchange standards and supporting sustainable building design.

2. Methods

As introduced above, this case study aims to compare results obtained from a computational simulation of time-dependent thermal energy changes, forcing the temperature redistribution in particular rooms of the experimental wooden building, with real values obtained by measurements in situ. This should contribute to the realistic evaluation of the possibilities and limits of applying BEM in building practice.

2.1. Computational Simulations

The energy performance of buildings is studied by appropriate software tools, simulating the behaviour of a building or its part using the methods of numerical and computational analysis. An appropriate model of energy flows allows us to understand the behaviour of a building as a thermal system even before its construction, including its energy demand under various climatic conditions, thus, it is able to answer crucial questions related to building orientation, optimal size and location of rooms, particular bearing structures and insulation layers, as well as about the design and control of technological devices. Software packages like Design Builder, EQuest, or SW Stabil are typically used for such global modelling tasks, accompanied by specialised software tools for detailed considerations regarding heating, ventilation, and air conditioning in buildings and their parts. These software tools, working with certain engineering simplifications, express the principles of classical thermodynamics, namely that on energy conservation, supplied by appropriate constitutive relations, using certain partial differential equations with various types of boundary, contact, and initial conditions [52]. The SW Stabil software [53], developed and implemented at the Brno University of Technology, Faculty of Civil Engineering, based on the combined finite volume approach with the method of discretisation in time (coming from the theory of Rothe sequences), is for its simplicity preferred in this case study. Alternative methods try to apply some (frequently incomplete) knowledge of fundamental solutions related to special expected solution decomposition, such as multiplicative Fourier one (method of lines) [54,55]. Yet other approaches, which are not discussed in detail here, have been developed, e.g., for the analysis of special types of buildings or their parts [56], analysis of potential thermal bridges [57], implementation of phase change materials in insulation layers [58], and proper incorporation of further significant physical processes, as moisture or (more general) mass transfer [49,50,51,52,53,54,55,56,57,58,59,60,61].

A strong motivation for the development of SW Stabil was the need to evaluate the dynamic response (i) of particular rooms in buildings to thermal loads under both winter and summer climatic conditions, as required by the European technical standard EN ISO 52016-1 [62], and (ii) of the whole building as a closed structural object, as required by the Czech technical standard ČSN 730540-2 [63] (valid from 2011). The considered model applies the system approach, taking into account the inner and outer thermal relations between structural parts (as elements and subsystems of such thermal system), using certain 1D simplifications. Such relations express particular types of heat transfer by conduction, convection, and radiation. Consequently, relying on the time decomposition and backward Euler modifications of nonlinear terms, due to an appropriate initial condition, in any positive time $\tau$ (assuming sufficiently short, typically equidistant, time steps) we have to solve only a sparse system of linear algebraic equations with a finite number of unknown values of absolute (Kelvin) temperature $t\left(\tau \right)$.

The evaluation of the effect of both direct and diffuse solar radiation is rather delicate. Due to the building’s location and orientation, including all slopes of exterior building surfaces, the continuous movement of the sun in daily and annual (quasi) cycles is considered, together with the reference climatic year database for appropriate conditions (or sufficiently close and comparable ones, e.g., those recorded at the nearest airport which are processed statistically, if available). More specifically, the air temperature of the outer environment during a typical year is needed, as well as the subsoil temperature for the estimate of heat losses through the building’s foundations. Both the thermal insulation and the thermal accumulation properties of all structural parts and their components can be considered. Repeated simulations for various alternatives are helpful for designers of HVAC systems who must decide what thermal power will be needed to be brought to particular rooms (i) to keep the stable temperature inside the formal restrictions based on technical standards and (ii) to optimise the thermal comfort of all users [64]. The simulation model may be driven by the mentioned requirements (i) and (ii), controlling the supplied heat continuously. The effects of ventilation, air flow in rooms, user occupation of rooms, operation of household appliances, artificial lighting, etc., must be included in a simplified way [65]; let us be reminded that, e.g., the proper evaluation of the thermally driven flow of air (taken as a compressible gas mixture) refers to the numerical analysis of the system of Navier–Stokes equations, whose solvability and solution smoothness is one of the to date (for nearly 25 years) unsolved mathematical Millennium Prize Problems [66]. An additional nontrivial problem can be identified in the multi-criteria design optimisation, supported by advanced computational simulations, of both building structures and HVAC systems [61].

Advanced methods, such as those utilising real-time data (Pezeshki et al., 2019 [29]) or dynamic simulation algorithms (Dimitriou et al., 2016 [32]), offer a higher simulation accuracy. However, they are often not practically applicable for standard projects due to their high costs and implementation complexity. In contrast, our study demonstrates that a standardised approach using gbXML is suitable for efficiently connecting BIM and BEM, even when using readily available tools such as EnergyPlus or SW Stabil.

Another key approach, utilising the IFC format for simulations, is thoroughly discussed by Bahar et al. (2013) [35]. While the IFC allows for the transfer of complex building data, its complexity often requires manual adjustments, which can lead to data loss. In our approach, we have noticed specific advantages of using gbXML for data transfer in the context of timber structures, particularly in enabling quick zone configuration and seamless geometric data transfer.

The model presented in the following case study was created in the Graphisoft ArchiCAD 27 software environment, which was followed by the export of data to the format gbXML, acceptable by all relevant computational modelling and simulation programmes and components of SW Stabil version 2.3, as mentioned above. Computational analysis of the experimental wooden object in Ostrava (see Section 2.3 for more details) takes into account the initial temperature distribution from January 2018; in the following years, the heating system was activated.

The above-described system approach, implemented in SW Stabil, can be applied separately, without the obligatory formulation of an initial and boundary value problem for (at least) one partial differential equation, corresponding (from the point of view of classical physics) to the requirement of energy conservation, supplied by an appropriate set of (mostly algebraic) constitutive relations [67,68]. However, for the reader’s convenience, we shall present simple equations of such type in the following sections.

2.2. Physical and Mathematical Background

For the simplicity of all formulas, we shall work everywhere with the 1D model reduction. The thermal energy conservation (avoiding other physical processes) can be then read as

ρc ∂t/∂τ = ∂q/∂x

(1)

for a certain thermal flux rate q [W·m⁻²], which is a function of τ in general (observe the formerly introduced temperature t [K] and time τ [s]), and is usually supplied by the empirical linearised Fourier law.

q = λ ∂t/∂x,

(2)

Thus, (1) and (2) yield

ρc ∂t/∂τ = ∂(λ ∂t/∂x)/∂x.

(3)

Here, the following material characteristics occur: the material density ρ [kg·m⁻³], the thermal conductivity λ [W·m⁻¹·K⁻¹], and the thermal capacity c. One can take these characteristics as (nearly) constant and time-independent, thus, the thermal diffusivity a = λ/(ρc) can be frequently found in engineering considerations. We can introduce x belonging to a certain finite interval [0, d]; the natural generalisation will be discussed later. This approach can also be applied to the 2D, 3D, etc., models, with minor modifications of (1), (2), (3): namely ∂/∂x must be replaced by the gradient operator, and the material characteristics are allowed to receive a matrix form to handle an anisotropic medium.

However, (1), (2), and (3) are able to describe thermal conduction on [0, d] only; one unknown field t dependent on both x and τ is present here. Thermal convection can be seen, namely on the exterior surfaces, using the evaluation formula for the surface heat flux rate:

q = h (t − t*);

(4)

Here, t* refers to the temperature of the outer environment, and h denotes the thermal transfer factor assigned to the related surface. Thanks to (2) and (4), this can generate the Neumann boundary conditions for all outer building surfaces or their parts. In particular, for the zero-valued h, we come to t = t*, which leads to the Dirichlet boundary conditions; this enables us, i.e., to decompose [0, d] for (1), (2), (3) to a finite number of subintervals, typically, into constructive and insulation layers in walls, without any change in previous considerations. This is significant for evaluating the total heat passage coefficients U, namely for windows, door(s), etc., required by most technical standards for building structures [69]. Apart from the boundary conditions, the natural Cauchy initial condition, i.e., the prescribed values of t everywhere on [0, d], must be implemented. In numerous simplified computations, the total flux rate Q is also needed instead of the local one, e.g., Q = qS for (4), where S means the related surface area, as transferred via gbXML, cf. Section 2.1.

Even the surface radiation by the Stefan–Boltzmann law can be incorporated using such a unified approach, replacing both t and t* in (4) by their 4th powers and h by εσ; here, σ = 5.670374419… × 10⁻⁸ W⋅m⁻²⋅K⁻⁴ is the famous Stefan–Boltzmann constant, and ε denotes the prescribed surface emissivity. The effect of designed ventilation and air exchange loss through imperfect joints and various other gaps can be evaluated as

Q = Vc° (t − t*)

(5)

where the temperature t corresponds to specific rooms, and they are considered constant there (dependent only on τ); c° denotes the thermal capacity of air, and the exchanged air volume V must be estimated in a reasonable way. The simple analogy of (5) can also be formulated for the heat exchange between two different rooms.

Finally, all thermal fluxes through solid walls, floor, roof, and windows and doors, including the effects of ventilation and solar irradiation, can be considered. The above-mentioned thermal flux rates contribute to the balance equation for every room, incorporating all relevant thermal gains and losses through integrating Q over appropriate time intervals. However, the time-dependent values of t for (6) are not known in advance, unlike t*; one must include them in all computations using the appropriate interface relations.

Notice that the modified version of (4) for the surface radiation always disturbs the originally linear system of partial differential equations, whose discretised version (for any fixed time step) in SW Stabil is expected to generate a system of linear algebraic equations. Fortunately, this contribution to the thermal flux rates is rarely dominating, thus, the backward Euler scheme, as given in Section 2.1, seems to be acceptable in practical calculations; otherwise, an inexact (less effective) Newton algorithm is available [66]. Another type of nonlinearity comes from the heating or air conditioning real-time control of thermal stability with prescribed temperature bounds; this is a non-trivial problem of the design and choice of some adaptive algorithms where some elements of artificial intelligence are welcome [67].

The computational simulations in this study took approximately 5 h and 33 min on a computer with 16 GB of RAM. The minimum hardware requirements were estimated at 4 GB of RAM, indicating that the method is computationally efficient and accessible even on standard workstations. The software (SW Stabil) is optimised for thermal flow simulations and employs a combined approach of the finite volume method and time discretisation. This optimisation enables robust data processing without the need for high-end hardware.

In terms of scalability, the method shows potential for application on larger or more complex models. However, as the model size or the number of simulated zones increases, higher computational power and memory requirements can be expected. Simulations of buildings with highly intricate structures, a greater number of zones, or more detailed physical processes may require more advanced hardware and longer computation times. Nevertheless, the method has demonstrated its applicability for medium-complexity projects while maintaining accuracy and computational efficiency.

2.3. An Experimental Timber Building

The object was built in 2012, thus reverse BIM modelling was necessary. It is located in the urban neighbourhood Ostrava-Poruba. The original 2D technical drawings were available in PDF format, including a complete listing of structures; this was provided by VŠB-Technical University Ostrava, Faculty of Civil Engineering, Department of Building Constructions. All used building materials are compatible with original documentation. From the constructive point of view, the object can be characterised as a passive house of compact shape and optimal orientation to cardinal points. The house disposition was suggested to satisfy the standards and expectations of young families. All living rooms are oriented to the south or west. The total built-up area is 97 m². Nowadays, the object is a training centre with three classrooms and one office. The technical background is located on the ground floor, with disabled facilities, including toilets. Three mansard classrooms are supplied by suspended soffit from gypsum wallboard. Figure 1 shows two illustrative photos of the object. Figure 2 shows the representation of the individual envelope structures of the building (external walls, Figure 2a; roof, Figure 2b; floor, Figure 2c). For simplification, the structural system is not included in the illustrations. In

Table 1. Overview of U-Values and Thicknesses of Building Components.

Building Component	Thickness d mm	Thermal Transmittance U W·m−2·K−1
External wall	510	0.10
Roof	650	0.10
Floor	700	0.12
Window and door elements	-	0.71

, an overview of the thermal transmittance (U-values) and the thicknesses of individual building components used in the calculations can be observed. Three-dimensional ground plan, created in the Graphisoft ArchiCAD 27 software environment, can be seen in Figure 3. Figure 4 presents two views of the object from the project documentation for comparison, supplied by the ArchiCAD setting of particular zones, which can be seen in Figure 5.

Individual zones can be observed in Figure 2 and Figure 4, distinguished by colour coding. The grey represents hallways and staircase areas, while the blue colour represents living spaces, and the green colour represents technical and sanitary facilities. The red colour represents room No. 5 on the 2nd floor, which is referred to as the monitoring room where temperature values were measured during the winter months.

The object is founded on a floating reinforced 300 mm thick concrete base, embedded in gravel and insulated by an extruded polystyrene layer of thickness 140 mm. The building’s skin is made of prefabricated wooden-based panels; a combination of various insulation systems and thickness of layers is used. The building walls are built of 300 mm high I-shape beams and filled in by fibrous wooden insulation. On the interior side, the walls are supplied by auxiliary structures, coated by plates from fibrous gypsum material, and filled in by fibrous wooden insulation. The exterior facade (piecewise insulated using the contact system, piecewise ventilated) is also insulated by wood fibre plates. All inner walls and wooden ceilings are filled in with mineral insulation. The heat passage coefficient of all windows is 0.71 W·m⁻²·K⁻¹; for the front door, it is 1 W·m⁻²·K⁻¹. The entrance to the object is covered with a marquee. Windows are supplied by exterior sunblinds. The terrace and balcony, located on the south facade, contain a construction for anchoring solar panels. The counter-shape roof uses favourable properties of the fibrous wooden insulation, applied both between the roof spars and over them, of a total thickness of 410 mm.

The ArchiCAD software makes it possible to export the model into the format gbXML, to be ready for SW Stabil-based modelling, but several particular operations must be still performed manually: thermal blocks with building zones have to be set and connected mutually, and all particular constructions should be checked carefully, as described in Section 1.

3. Results and Discussion

The development of temperature was measured and evaluated by SW Stabil to examine the potential of computational prediction of the thermal behaviour of such a residential object, as introduced in Section 2.3, using the approach described in Section 2.2. Figure 6 shows the time development of temperature in one selected room (number 5) in January 2018, i.e., in the period crucial for the controlled heating under the Central European climatic conditions, in 15-min intervals. There were two obtained datasets, (i) empirical data and (ii) the results of numerical simulation using the gbXML-based transfer from the digital twin of the structural object.

From a thermal-technical perspective, the building is composed of materials with high thermal capacity, particularly considering botanical fibres such as wood. Throughout the study period, the entire building, including classrooms, remained unheated, allowing internal room temperatures to redistribute following the thermal fluxes driven by external boundary conditions—air temperature, solar irradiation, wind pressure, and natural ventilation. Since no artificial heating was active, the influence of neighbouring rooms was minimal, and additional internal heat sources, such as occupants or artificial lighting, were excluded.

Empirical data reflect real thermal conditions inside the building, whereas model-based data aim to replicate these conditions using predefined physical parameters and environmental influences. The study considered boundary conditions based on a 20-year temperature average for the given locality. The period under review spanned 31 days, yielding 2976 measurements per dataset.

The empirical data exhibit dynamic temperature fluctuations with sharp local transitions, likely caused by external factors such as (i) door openings or (ii) ventilation system changes. These short-term variations highlight momentary responses to ambient conditions essential for understanding the thermal behaviour of timber buildings. In contrast, computational simulation results show more moderate temperature changes due to numerical modelling simplifications. Short-term variations (i, ii) are not captured in detail, but long-term trends are well aligned between both datasets.

To further validate the computational approach, temperature profiles for Room No. 5—located in the heavily insulated timber house—were analysed for January 2019. This period was characterised by low temperatures and limited solar irradiation under typical Central European winter conditions, with no artificial heating. The simulation considered temperature development over an entire year, including all seasons, winter, spring, summer, and autumn, ensuring quasi-periodic trends over multiple years, accounting for initial temperature distributions and potential controlled heating scenarios.

Given the inherent uncertainties in climatic data, a sensitivity analysis was conducted. Original local climatic data spanning 26 years—including environmental air temperature and global and diffuse solar irradiation—were compared with officially averaged reference-year data. Comparative simulations using 26 distinct datasets were performed to evaluate deviations, including variables such as air exchange rates influenced by occupant behaviour.

When applying reference-year data, the Pearson correlation coefficient between measured and simulated temperatures for January 2019 was 0.975, indicating strong alignment. To assess variability, standard deviations between specific yearly data and reference-year values were computed. Using an expansion factor of 2, a 95% confidence interval was established, resulting in a ±2.1 K band around the temperature curve. Importantly, real measurements remained within this band, supporting the validity of the computational simulation approach and its applicability for long-term thermal predictions in timber buildings.

These findings emphasise the necessity of refining material parameters, particularly those related to the moisture content in wooden structures, and improving environmental modelling of solar gains and air infiltration. While computational models effectively capture long-term trends, their reduced sensitivity to short-term fluctuations suggests potential applications in optimising energy control strategies, particularly in heating system design.

Key Findings

The following findings should be emphasised:

• Empirical temperature-in-time development demonstrates the quasi-cyclic running, corresponding to daily warming-up, followed by nights of cooling-down.
• Similar trend in both datasets is indicative of the applicability of the simplified computational model to the prediction of basic thermal technical processes in building structures.
• Most perceptible differences between both datasets correspond to the less smooth course of empirical data.
• An approximate difference between both datasets is 0.7 K; maximal values are greater than 1.5 K.
• From time to time, the peaks in temperature development from computational simulation occur sooner or later than in the empirical data. This can indicate some model restrictions in representing the dynamics of thermal transfer or reactions to unexpected changes in the environment.

The following groups of reasons for the above-specified differences in datasets can be distinguished:

• Environmental influences: Local structural manifestations, such as thermal bridges, behaviour of residents, or variability of exterior weather conditions, cannot be incorporated into a reasonable deterministic model. Solar gains through windows, unexpected ventilation, etc., can significantly influence temperature development, but their reliable modelling is also quite difficult.
• Material properties: Timber structures are rather complicated systems with heterogeneous and anisotropic material properties. Model assumptions on thermal conductivity and capacity, as well as on moisture content, rarely perfectly correspond to real conditions.
• Model simplifications: Numerical models rely on a set of simplifications, to come to robust and effective computations. For example, all assumptions on (quasi-)stationary states neglect intermediate phenomena such as ventilation caused by door or windows opening; detailed analysis of such forced air flow in rooms is not realistic.
• Measuring Equipment and Calibration: The recorded data are influenced by the accuracy and resolution of the measuring equipment used. Even minor errors in measurement or calibration could have contributed to deviations. For example, the effects of local thermal bridges or solar gains were not measured uniformly throughout the experiment.
• Impact of Moisture in Structures: Timber structures are sensitive to changes in moisture, which affects their thermal properties, such as thermal conductivity and capacity. The simulation assumed constant material properties, whereas, in reality, moisture levels could fluctuate significantly, depending on temperature, seasonal changes, and microclimatic conditions. In the model, these variables could not be fully accounted for.
• Influence of Internal Heat Gains: The simulation did not consider internal heat gains from occupants, electrical appliances, or lighting. These factors could have influenced the temperatures measured, especially during the day when internal heat gains are more significant. While these simplifications were intentionally made to focus on fundamental thermal processes, they may have contributed to discrepancies between the data.
• Limitations of the gbXML Format for Data Transfer: The gbXML format, used for transferring data from BIM to BEM, does not convey all details regarding the thermal-technical properties of individual construction layers. For example, information about the actual composition of insulation layers or surface materials may be simplified during export, affecting the accuracy of the simulation. These limitations have also been identified in previous studies (e.g., Lin et al., 2016) [39].
• Neglecting the Thermal Inertia of Adjacent Structures: The influence of neighbouring structures, such as foundations, roofs, or partitions, was modelled as in the case of an isolated structure, even though their thermal inertia and interaction with surrounding structures can significantly impact the resulting indoor temperatures. This simplification may have led to minor discrepancies between empirical and simulated data.
• Imbalance in Input Climatic Data: The climatic data used were based on a reference climatic year, a standard approach in energy simulations. However, these data represent average conditions over a long period of time and may not reflect precisely the actual meteorological conditions at the time of measurement. This difference could have affected the results, especially if the real conditions showed deviations in extreme temperatures or solar radiation intensity.
• Local Thermal-Technical Variations in Construction: Timber buildings can exhibit local variations in material properties due to differences in insulation quality or minor construction imperfections (such as air leaks or uneven insulation thickness). These localised differences, which were not accounted for in the model, could have led to more significant deviations between simulated and measured temperatures in specific parts of the building.

The model exhibits good ability to express the overall thermal dynamic behaviour of a structure. The quasi-cyclic character of the data and clear compliance of both datasets indicate the model’s robustness. On the other hand, short-time fluctuations and certain non-consistent delays in datasets suggest the need to improve assumptions on thermal inertial behaviourbehavior or the reaction of wind influences.

This leads to the following consequences:

• Thermal performance of a timber structure: Data from detailed measurements demonstrate that the temperature inside the house is relatively stable despite windy winter conditions. Such stability manifests the effectiveness of the building enclosure for the thermal comfort maintenance for residents.
• Parameters of numerical models: Detected differences highlight the need for precise valuation of input parameters, with expectable consequences of revision of physical formulations, mathematical simplifications, and computational algorithms. Further calibration, like for a building in winter with controlled heating or for a building under hot summer conditions, could contribute to the model’s validity. Nevertheless, designing experiments and the numerical analysis of related inverse problems is a stand-alone, non-trivial research area [68,69].
• Exploitation in building practice: Reliable predictive models are crucial for the optimisation of energy consumption of buildings. Identifying physical processes and influences should suppress the differences between empirical data and data from numerical simulations, resulting in remarkable progress in designing HVAC systems and energy control strategies [70,71].

As emphasised by Wang et al. (2019) [43], one of the main limitations of energy modelling is the lack of accurate input data, which can lead to discrepancies in simulations. In our case, when comparing simulated and empirical data, such deviations were observed as expected. Another factor is the manual setup of thermal zones, as noted by Yang et al. (2022) [41], which can be time-consuming. Nevertheless, gbXML offers flexibility and simplicity, making it widely applicable in construction practice.

One of the primary objectives of this study was to demonstrate the use of the gbXML format for connecting BIM and BEM platforms. During the data transfer process, we identified several issues that affect the efficiency and accuracy of simulations:

• Inconsistent Processing of Geometric Data: The gbXML format has limitations when handling complex geometric models, which can lead to issues during import into BEM software. For example, we observed incorrect definitions of some thermal zones, requiring manual corrections and thus increasing the time required for processing.
• Mismatch in the Definition of Boundary Conditions: Boundary conditions, such as thermal loads or solar gains, were represented in a simplified manner within the gbXML format. This may affect the accuracy of simulations, particularly for buildings with complex shading elements or unusual structural features.
• Differences in Supported Standards: Different software platforms implement the gbXML format differently, making smooth data transfer more challenging. This discrepancy was identified when working with exports from ArchiCAD to SW Stabil, where certain parameters had to be manually verified.

Interoperability between BIM and BEM platforms is a key factor for efficient data transfer and accurate energy performance simulations of buildings. Although gbXML is a widely accepted standard for integrating these two environments, its practical implementation encounters several challenges, such as inconsistencies in processing geometric data, limited export capabilities for complex material properties, and variability in how different software platforms interpret the data. These challenges can result in additional time and manual effort, impacting the overall quality of the simulation results.

To enhance interoperability, both technological and process-oriented improvements are necessary to ensure more efficient and consistent data transfer between BIM and BEM tools. The following recommendations outline the key measures that could help mitigate these limitations and improve the quality of energy simulation outcomes:

• Implementation of Data Validation in the BIM-BEM Workflow: A critical step toward improving interoperability is the development of advanced algorithms for data consistency checks during the export and import of gbXML files. These algorithms should automatically verify geometric accuracy, completeness of data attributes, and compatibility with the target BEM platform. This would reduce the risk of manual errors and improve simulation accuracy.
• Expanding the Functionality of the gbXML Format: Since gbXML is primarily designed for transferring geometric and thermodynamic properties, further expanding its data model could allow for the inclusion of more detailed parameters, such as moisture dynamics, precise specifications of material layers or interactions with ventilation systems. These enhancements could be standardised through broader collaboration between BIM and BEM tool developers.
• Integration with Advanced Simulation Tools: The introduction of advanced simulation tools capable of working natively with gbXML data while minimising the need for additional manual adjustments could significantly increase the efficiency of the BIM-BEM workflow. For example, integrating gbXML with dynamic simulation tools such as OpenStudio or EnergyPlus could provide more accurate results with less user intervention.
• Standardisation of Export and Import Processes: Improving standards for data export and import would ensure more consistent data transfer across different software platforms. This could include the introduction of a common framework for interpreting gbXML data across various tools, eliminating the current variability in how the format is implemented by different developers.
• Collaboration among Software Developers: Achieving seamless interoperability requires closer collaboration between BIM and BEM software developers. This collaboration could involve sharing open data formats, reference projects, and the development of application programming interfaces (APIs) to facilitate integration across different platforms.

4. Conclusions

The aim of this submitted case study was to introduce a synergetic interconnection between the information modelling of buildings (BIM) and their energy modelling (BEM), using an experimental timber structure in Ostrava. The comparison of measured and computed temperature fields formed a basis (i) for identifying key factors that control the thermal dynamics and (ii) to evaluate the accuracy of computational simulations.

Validating the ability of simplified numerical models to predict long-term trends of temperature development at a rather high precision level can be seen as a non-negligible contribution of this article. Whereas computational simulations provide a robust basis for strategic decisions in the domain of energy efficiency of buildings, non-negligible differences between measured and computed data, in particular temperature values, have been found, caused by imperfect implementation of material properties into the model or undervaluation of some environmental influences. These drawbacks stimulate further development of advanced adaptive algorithms and dynamic simulation tools under real in situ conditions.

BIM has significant potential for the automation of processes, especially the integration of data between particular phases of the design and construction of a building. Applying data formats such as IFC and gbXML should improve their interoperability and reduce the risk of data loss during their transfer. Other improvements of BEM simulation tools require deeper analysis of all physical processes significant for the heat transfer problems in buildings and related mathematical formulations of direct, sensitivity, and inverse problems, accompanied by the design of appropriate experiments for parameter setting in engineering applications. This calls for a remarkable upgrade of cooperation between researchers from seemingly extraneous areas, such as building design, control theory, distributed computing, and artificial intelligence.

Our study provides a practical contribution to the ongoing discussion on BIM and BEM interoperability. While advanced techniques offer higher accuracy and flexibility, our work demonstrates that standardised approaches, such as gbXML, are easily implementable in practice and provide sufficient accuracy for most applications. This contributes to the adoption of these methods in everyday design practices.

Although this study focuses on a case study of a timber structure, its value lies in demonstrating the practical application of the gbXML format to enhance interoperability between BIM and BEM. The methodology described in this study is universal and can be applied to other types of buildings, such as concrete or steel structures, provided that appropriate input data are given and material properties are adjusted.

Furthermore, the results highlight the importance of properly defining thermal zones and ensuring input data accuracy, which is crucial for any building simulation. This approach can also be utilised to optimise the energy performance of buildings under various climatic conditions, and its implementation is feasible within a broad range of design and simulation tools.

Finally, our work demonstrates that commonly available tools can be used efficiently and are user-friendly, providing practical benefits for designers and researchers involved in building energy modelling. These findings align with current challenges in sustainable building design and lifecycle optimisation.

The results of this study can be applied across multiple areas of construction practice and energy policy:

• Building Design Optimisation: Accurate simulations of heat transfer between different zones and structures provide designers with valuable insights into the efficiency of proposed insulation materials and their impact on the overall energy balance of a building. For instance, our simulations revealed that minor adjustments in insulation composition or thermal bridge design could significantly reduce heat losses. These insights can be seamlessly integrated into the building design phase, even using commonly available BIM software tools.
• Enhancing Decision-Making Processes: Energy simulations based on our methods can be used to assess various design scenarios, such as evaluating the impact of investing in better windows or shading elements. This helps investors and architects determine which changes will yield the greatest benefits in terms of long-term operational costs and emissions reduction.
• Supporting Sustainable Building Policies: Our findings can serve as a foundation for developing energy standards and regulations, particularly concerning timber structures and their thermal/technical properties. The results can also be utilised in the design of subsidy programmes focused on energy efficiency and renewable energy sources or as a rationale for implementing stricter requirements for insulation properties in building materials.
• Application in Climate Change Adaptation: The BIM-BEM integration methodology described in this study allows for the modelling of buildings while considering the different climate scenarios, which is crucial for designing adaptive structures. For example, when predicting higher summer temperatures, our findings can support the design of more efficient cooling and shading strategies for buildings.