Immersive virtual reality for extending the potential of building information modeling in architecture, engineering, and construction sector: systematic review

ABSTRACT The field of architecture, engineering, and construction (AEC) is continually striving to use resources efficiently and manage complex processes. Now more than ever, there is a strong need for digital transformation in AEC. The improvement in the accessibility of consumer-based head-mounted displays (HMD) is attracting different entertainment and research fields to immersive virtual reality (VR) applications. Building Information Modeling (BIM) is known as a promising technology in AEC. The full potential of BIM is not yet employed to empower this field, however, and this could be a result of some barriers still to be surmounted by BIM in both technological and management perspectives. One of these barriers is the communication and collaboration between design, construction, operation, and maintenance phases. VR can fill this gap by providing additional capabilities for BIM which either were not available before or were not possible to employ in practical ways. In this paper, we systematically review recent research around the application of VR in BIM and discuss the results using the PRISMA flowchart. We discuss the most commonly used technologies, software, and evaluation methods and the various applications of VR in the reviewed papers. Finally, we extend the discussion by summarizing the potential future work in this area.


Introduction
The fourth industrial revolution is seeking to enhance and accelerate the traditional developmentworkflows using automation, smart technologies, and modern communications. This revolution has put the spotlight on the necessity for intelligent systems across a very broad spectrum of different industries by employing the internet of things (IoT), artificial intelligence, and big data (Kozlovska, Klosova, and Strukova 2021). The implementation of industry 4.0 in architecture, engineering, and construction (AEC) is leading to the definition of Construction 4.0 (Forcael et al. 2020). Construction 4.0 can be defined as a framework that aims at a conjunction of three broad categories: industrial production, cyber-physical systems, and digital-computational technologies (Irizarry 2020). Examples of industrial production are prefabrication and offsite manufacturing, where cyber-physical systems focus primarily on sensors, IoT, robots, and drones. Digitalcomputational technologies consist of several topics such as building information modeling (BIM), artificial intelligence, augmented/virtual reality, and cloud computing. Focussing on the VR is one of the promising technologies that can improve the process of design, construction, application, and maintenance in many AEC projects. Although various studies use the term 'VR' to describe all digitally created environments, we use this term here to refer to the immersive virtual experience commonly developed for Head Mounted Displays (HMD). In recent years, the improvement in the quality and usability of HMDs, as well as the more affordable cost for consumers, has led to more people experimenting with and using VR (Liu 2019). Many different applications of VR are currently being discovered and used, from entertainment to research and education (Damiani et al. 2018;Radianti et al. 2020). VR can be used as a movie theater or gaming environment to immerse the user in an unseen experience. Furthermore, it can serve as a teaching environment with additional functionality, such as a virtual physics lab (Pirker et al. 2019), without the barrier of providing expensive equipment. However, it seems that in many areas, the inherent potential of VR is yet to be properly exploited (Safikhani, Holly, and Pirker 2020). The use of VR is not only the difference between interaction and UI compared to the traditional ones, but it is a fully immersive experience that can define new eras ahead. In the AEC industry, VR can be used as a tool for collaboration, project management, and education to improve, facilitate, and close the gaps in the implementation of BIM in projects (Pratama and Dossick 2019;Davila Delgado et al. 2020). Furthermore, it can improve the level of awareness and communication in project stakeholders by promoting iterative decision-making (Kamari and Kirkegaard 2019). However, as Cho et al. (2019) show there is a big gap in VR interest between academia and industry in the AEC domain. This indicates potential benefits and the necessity of employing VR in AEC education and industry. In addition to the benefits of accessing BIM-based VR experiences, it can add complexity to the project as it needs additional education of personnel and requires more advanced technology to implement (Shaoze Wu, Hou, and Zhang 2020). Davila Delgado et al. (2020) found out in their systematic study that the main limitation of implementing AR and VR in the construction sector is its cost, as they need specific pieces of equipment. In this case, defining use cases and their scope can reduce the need to use unnecessary or complicated solutions. Understanding the state of the art of this technology, as well as the limitations of its use, can also provide a basis for future research and development.
In this review paper, we discuss recent developments and research in the field of VR and BIM. To better understand the extent to which virtual reality is already being used and researched in the field of AEC/BIM, we would like to discuss the current status, use cases, and technologies used, and then also focus on relevant future research. In this context we thus focus on the following research goals: (RG1) to identify potential use cases and applications of VR-enabled BIM in AEC industries, (RG2) to provide an overview of technologies used to develop and implement VR experiences, and (RG3) to give an overview of potential future research and development possibilities in this area. To fulfil these objectives, we conducted a systematic review in three IEEE, ACM, and Scopus databases to find relevant papers. We then classify the selected papers into four categories based on their application field and discuss them accordingly. We have discussed RG1 in the section 'Immersive Virtual Reality for Building Information Modeling', RG2 in the section 'Results', and RG3 in the section 'Limitations and Future Work'.

Related works
In their systematic review, Wen and Gheisari (2020) highlight the complexity in AEC industry and how VR can improve communication between stakeholders. They follow a systematic approach to summarize the results of 41 peer-reviewed papers and describe the application of VR to support communication in the AEC domain. Most of the reviewed work provides evidence that VR applications are able to improve communication efficiency in AEC. Further, they identify four future research directions of VR applications to facilitate communication in AEC: (1) real-time data transfer between BIM and the game engine; (2) conjunction with AR applications; (3) improvement of the realism of VR; and (4) simulations of avatars' emotions and physical behaviors.
A recent review paper by Shaoze Wu, Hou, and Zhang (2020) explores the application branches of BIM-XR in the AEC industry. Their review considered the Scopus and Web of Science databases to find articles. They classified XR applications in BIM into four main categories: (1) task guidance and information gathering, (2) design review and refinement, (3) process planning and control (4) upskilling of AEC staff. Accordingly, they proposed a systematic application framework to demonstrate the use of XR in BIM. The proposed framework can be used as a research plan for future studies.
Alizadehsalehi, Hadavi, and Huang (2020) combined a review paper with a case study on the conversion of BIM models in VR and mixed reality environments. They classified the literature according to the XR platforms. They also discussed different options for wearable XR and compared them according to the content and interactivity. They classified the software used in the literature into BIM platforms and translators. Accordingly, they visualized the relationship between them. The leading BIM platforms in their classification are Autodesk tools such as Revit and Civil 3D. The translator software consists of game engines (e.g. Unity) and 3D modeling tools (e.g. Autodesk 3Ds Max). They summarized the strengths and weaknesses of each XR technology in a table, taking into account the perspective of the authors or case studies.
In their previous work Alizadehsalehi, Hadavi, and Huang (2019) present a BIM/MR Lean-based project delivery management model. Based on a literature review, they design a model, analyze advantages and identify challenges. They highlight the potential of the applied tools and procedures to display the diverse needs and interests of all stakeholders before construction. The benefits of this approach include time/cost reduction and improved quality. Zhang et al. (2020) present a mixed quantitative-qualitative review on VR applications in AEC. They perform bibliometric analysis using VOSviewer on 229 journal articles retrieved from scopus. Building on these results they analyze trends, form relevant categories and discuss in-depth research gaps and needs regarding VR for the built environment. Further they point out the following future research directions: (1) user-centered adaptive design, (2) attention-driven virtual reality information systems, (3) construction training systems incorporating human factors, (4) occupant-centered facility management, and (5) industry adoption. Noghabaei et al. (2020) conducted a trend analysis on the adaptation of VR and Augmented Reality (AR) in AEC industries. They believed that the slow adaptation of these technologies in an AEC project is related to the lack of a proper study to investigate the cost-benefit ratio for these technologies. They conducted a survey with 158 industry professionals as participants to assess the current status and their expectations for the future of VR/AR. This study shows that residential and commercial projects are the main area where virtual environments are being implemented. In addition, industry experts expect a sharp increase in applying these augmented reality technologies over the next 5 to 10 years. Based on their findings, the older generation showed more confidence in the future of VR/AR. Sidani et al. (2021) focused on several research questions considering the implementation, application, and evaluation of VR in BIM. They screened 16 papers from 2013 to 2018 using the PRISMA statement strategy. According to the research fields, they classified the results into six categories: Collaboration, Construction Design, Construction Management, Construction Safety, Education, and Facility Management. According to their findings, the workflow from BIM to VR consists of four stages: (1) Create BIM representation of the project (2) preparation of the BIM project for VR (considering model conversion and optimization) (3) implementing BIM model in VR environments mainly using game engines (4) exchange non-geometric information between BIM and VR. They did not find a holistic assessment methodology in the reviewed papers. Depending on the application, however, case studies and questionnaires were mainly used as the evaluation methods.
A review paper by Alizadehsalehi, Hadavi, and Huang (2021) discussed the recent development of VR in the AEC industry and education and measured the potential of this technology to improve student learning outcomes. This review shows that VR can be useful in education (e.g. safety training), design (e.g. understanding of complex design), and project management (e.g. project schedule control). They conducted a survey of 14 Master of Science students on the benefits of using VR devices, considering (1) learning ability (2) interoperability (3) visualization (4) similarity to the real world (5) preconceived interactions (6) creativity (7) motivation and (8) comfort. The results of this study show that students believed they had a better understanding of concepts using VR technology. Accordingly, they concluded that VR can increase students' creativity, improve visualization of complex project designs, and increase comprehension of concepts in the course.
A critical review by Peng  focused on construction engineering Education and training. They reviewed 66 journal papers from Scopus and Web of Science databases. They observed that VR technologies move from traditional desktop-based applications towards mobile versions due to enhancement in immersion and interactions. According to their findings, students can participate more fully and experience greater motivation by considering immersive VR, 3D game-based VR, and AR as training aid tools. VR technology can help students to understand a building better in spatial terms through its inheritance immersive feature. They classified the future research direction for VR in Education and training of construction engineering into five topics: (1) Considering the state of the art of educational paradigms in VR experience (2) Development of VR kits with particular consideration for educational purposes (considering cost, size, mobility, and immersive interactions) (3) Distance learning in VR (4) Hybrid visualization through Mixed Reality approach (5) Rapid as-built scene generation using photogrammetry and laser scanning technology.
While these review articles cover general aspects of XR technologies, this article focuses particularly on immersive VR applications in BIM. We discuss the advantages and limitations of implementing VR in AEC projects. Readers can use this article to find out the current trend in this topic and possible future applications.

Methodology
In assembling the literature for this review paper, we examined the paper databases of Scopus (one of the largest abstract and citation databases of peer-reviewed literature, www.scopus.com), IEEEXplore (as a common database for computer science and technology-related articles, ieeexplore.ieee.org), and the ACM Digital Library (https://dl.acm.org/). The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Matthew J. Page et al. 2021) flow diagram of the screening procedure is demonstrated in Figure 1. The search in Scopus is set to 'Article title, Abstract, Keywords'. We use the general search field for the ACM Digital Library and IEEEXplore. The search terms that we consider in all databases are ('Virtual Reality' OR 'VR') AND ('BIM' OR 'Building Information Modeling'). Using these search terms in February 2021, we found 116 articles in ACM, 44 in IEEE, and 455 in Scopus (including journal and conference articles). In the course of the identification step, we select 402 unique papers (duplicates were removed) in the range from 2016 to 2021 (as the time frame was not limited in the main search). By assessing the eligibility of the papers and manually screening them, we then identified 73 studies to be included in this review paper (11 papers removed due to using a language other than English, 57 papers due to lack of access, 12 review papers, and 249 papers were not relevant either on explanation or not giving enough information related to implementation).
The analysis of the number of papers per year related to VR and BIM shows us a positive growth rate except for 2020 (see Figure 2). This result means that more AEC researchers are interested in VR-related topics. The total number of VR-related papers in 2020, however, is less than it was in 2019, which may be due to the global pandemic and the restrictions in conducting VR studies under lock-down conditions. However, even in this case, the number of screened papers in 2020 is close to that of 2019.
The co-occurrence network of keywords shows the most frequently used keywords for the selected papers (see Figure 3). We used VOSviewer (https://www.vosviewer.com/) for constructing and visualizing keyword networks. The graph provided shows which keywords are commonly used together. Moreover, it provides insight into overlapping topics, which can be seen in colored and nearby positioned clusters. As expected, according to our search terms, main research arguments, BIM and Virtual Reality are the most frequently used keywords. Augmented reality is often mentioned in the selected publications next to VR, as it serves in related use cases. VR as a revolutionary visualization tool for Historic Building Information Modeling (HBIM) is associated with it. Construction, security, and information modeling are other terms commonly used for this research area.

Immersive virtual reality for building information modeling
In the reviewed papers, three main categories can be found that define the application of VR in BIM for this study: (1) education and training, (2) design and data exchange, and (3) project management and collaboration. A clear distinction between these categories is not always given, and in some studies, there are combinations of two or three categories. In these cases, we select the most considerable one to make it easier for the reader to get an idea of that publication. In addition to these main categories, we also discuss a specific use case of VR in Historic Building Information Modeling (HBIM). This use case may also be considered a distinct category, but due to similarity in the implementation and application with other categories, we do not discuss it as a separate category.

Application and implementation
We summarize the paper classifications according to their primary application in Tables 1 and 2. Here, we discuss the relevant studies for each of the aforementioned categories.

Education and training
In this section, we focus on the potential of VR-enabled BIM for education and training. The power of immersive experiences in VR can create new opportunities in education that were previously difficult or impossible to achieve. VR can increase student motivation and understanding of learning content (Fonseca et al. 2017). In addition, as an interactive and immersive environment, it can facilitate learning processes by providing more realistic learning content (compared to traditional desktop or presentation experiences) in an almost zero-risk environment. As we have already pointed out, the fourth revolution in industry requires the automation of workflows. One of the prerequisites for the development of automation technologies in the AEC industry is adequate worker training. VR can provide a learning situation in which workers can learn with less risk of injury or damage to the end product (Peterson et al. 2021).
Safety training: One of the topics that is gaining increased attention is safety training with VR. The virtual environment brings the opportunity to learn safety aspects on construction sites (Hilfert, Teizer, and König 2016). The results of a recent study by Bhagwat, Kumar, and Delhi (2021) showed that students prefer interactive mobile VR simulations as a training module over traditional ones. In this case, however, construction professionals chose the non-interactive VR experience over the interactive one because it is simple, inexpensive, and saves time. One of the solutions to make the VR experience more affordable is using mobile/web-VR devices (Martinez, Ferris, and Wadley 2021;Getuli et al. 2021). However, this leads to a limited computational budget (Kreuz 2020). A study by Yan et al. (2017) explores the possibility of applying fire evacuation simulation  (2021) (2017) in web-VR. They simplified the simulation by making lightweight scenario objects, path planning, collision detection, and fire evacuation system. To improve the impact of safety training applications, designers can implement procedural or automatic data-driven scenarios. The study by Mo et al. (2018) explores this approach by defining scenarios from the 232 accident reports over 27 years. Their study demonstrates the improvement in accuracy and efficiency of scenario designs.
Design training: In this section, we focus on the potential of VR-enabled BIM for the design of AEC projects and exchanging data. Recent studies have explored this potential and discuss the challenges. Student understanding of architectural, structural, and engineering design can be significantly enhanced by implementing VR visualization into the training process (Ahmed 2020). The visualization of construction projects in VR can simulate a similar condition that students have previously only been able to experience on the construction site, e.g. visualization of reinforcement structures and formations hidden in the concrete slab (Dinis et al. 2017b(Dinis et al. , 2018. Kandi et al. (2020) conducted a study on the impact of VR educational games on improving students' design review skills. The researchers asked undergraduate students to find noticeable design errors in the course of a VR learning game. This study shows significantly higher student performance (number of errors found) when using VR than physical construction drawing. Improving the understanding of 3D space and spatial skills is the result of using VR as a gamified strategy ). The environment developed by Sanchez-Sepulveda, Marti-Audi, and Fonseca-Escudero (2019) offers the possibility to virtually recreate the urban areas of the city of Barcelona. Students can make decisions about how to recreate this area by interacting with elements in the VR environment. Their findings on the impact of VR technology in urban planning education were consistent with Kandi et al. The results show that the younger participants adapted slightly better to the new technologies, but they did not achieve a significant difference in their results. Accordingly, and as shown by the results of Dinis et al. (2017a), we can conclude that VR is a suitable tool for civil engineering education. Moreover, students can quickly learn the interaction for this virtual experience regardless of their age and background. In addition, VR technology can be used as a training tool for survey practitioners. A study by Wang et al. discusses the positive impact of VR on student decision-making (Chen Wang, Li, and Kho 2018). They considered only the Google Cardboard as a 360-degree display and simple interaction in the form of a walk-through. The influence of a more advanced VR setup with more interactions with the environment can be studied as complementary results.

Design and data exchange
VR can add another dimension to the building design process (Roupé et al. 2016). The benefits of the immersive experience in VR can reduce the need for physical maquettes of the future building and improve the results of the design iteration. Perceived visualizations in VR are more accurate   (2018), and Patti et al. (2017) and provide more motivation and interest for users compared to 2D drawings (Calderon-Hernandez et al. 2019), and place fewer demands on cognitive load compared to 3D visualizations on desktop monitors (Hermund, Klint, and Bundgård 2018).
Decision making: VR can be used as a tool for decision-making in the design phase while looking at the different design alternatives. These alternatives can be just geometric aspects of the building, or they can be cost-related (Bademosi, Tayeh, and Issa 2019) or environmentally friendly decisions. Furthermore, recent studies have shown that VR can even help non-AEC professionals develop a greater interest in sustainable design solutions (Kamari, Paari, and Torvund 2021). Allowing end-users to make design decisions can improve their satisfaction and reduce the number of design changes (Balali, Zalavadia, and Heydarian 2020). Engaging non-specialist in early design phases of a project can be beneficial when it comes to issue related to energy and green design (Wiberg et al. 2019). One of these issues is modeling zero emission neighborhoods. A zero-emission neighborhood aims to reduce its direct and indirect greenhouse gas emissions to zero over the analysis period (Wiberg et al. 2019). Visualization can play an important role in reducing the complexity of this process for planners, architects, researchers, and citizens. A study by Wiberg et al. (2019) demonstrates the use of VR in visualizing greenhouse gas emissions associated with transportation and materials use. The question that should be answered in the case of end-user involvement is to what extent VR can represent the spatial perception of the real project. A study by Cha et al. (2019) points to this question and finds that users can distinguish between many aspects of spatial features. Similar findings are also reported by Daniel Paes, Arantes, and Irizarry (2017) that indicate an immersive virtual environment can improve the overall spatial perception of the users. Another example of the use of VR visualization in the design phase is crowd simulation scenarios in public buildings such as hospitals. Crowd simulation in VR can help users perceive occupancy. This understanding in terms of user experience can be used as design input (Ventura et al. 2018).
Planning smart cities is a complex task that requires multiple considerations. VR can facilitate this process by improving the testing environment in the design phase. A recent study demonstrates one of these aspects in the case of smart city planning for wheelchair users (Götzelmann and Kreimeier 2020). In this study, by implementing an actual wheelchair and a dedicated physical platform combined with HMD, users can explore the planned building or urban environment in VR.
Interactive visualization: Comparing the influence of VR on visualization of AEC projects and traditional 2D drawings indicates the positive effect on users perception accuracy and memory (Calderon-Hernandez et al. 2019;Roupé et al. 2016). Moreover, from a neurological perspective, immersive VR representation of architectural visualization is less cognitively demanding than traditional 3D visualization on desktop monitors (Hermund, Klint, and Bundgård 2018). Using the idea from the other types of extended reality also showed positive potential in improving VR applications. Prouzeau et al. (2020) took some ideas from augmented reality (AR) use in BIM and implemented them in VR to achieve three goals: 1. improve remote work for BIM workflow 2. real-time and interactive visualization 3. communication and collaboration with experts without programing skills. Their work showed several potentials that are usually expected from an AR environment. The test users in their study mentioned that VR could be helpful in the design phase because it can show the impact of different equipment configurations. A study by Tariq et al. (2019) on the application of VR for visualization of BIM projects shows positive response of professionals and students towards this technology.
MEP structures: In addition to demonstrating parts of the AEC projects that will be visible in the design phase and after construction, VR can also provide insight into the hidden facilities such as MEP (mechanical, electrical, plumbing) structures (Jing-Ying Wong et al. 2020;Ortega et al. 2020). Besides the educational aspect of this feature, it can also be helpful in the design phase as an additional view angle for the project. Before introducing VR and its application in BIM, lighting design was usually based on the previous experience of the designer or simulation results (Natephra et al. 2017). This approach is powerful but cannot provide clear insight into the user experience. Algorithmic design: Algorithmic design can be defined as a set of algorithms that generate architectural environments (spaces and forms) through the use of rule-based logic (Terzidis 2004). This approach is gaining more attention in architectural firms in recent years (Caetano, Santos, and Leitão 2020). The conventional way to present this data is through the use of CAD or BIM models. In this case, VR can improve the visualization to optimize the workflow. Castelo-Branco, Leitão, and Brás (2020) proposed a live coding solution for Algorithmic Design in VR. The approach looks promising, but the implementation lacks a viable UI that improves productivity. Using more intractable elements and native VR interaction such as controllers or hand gestures instead of keyboard typing can improve this approach.
Data exchange: The difficulty in exchanging data between BIM and VR limits the use of VR in the AEC industry. To facilitate this process, several researchers have proposed workflows in the literature (Rahimian et al. 2019;Kado and Hirasawa 2018;Nandavar et al. 2018). Khalili (2021) had proposed an XML conversion workflow. The workflow can convert the model from BIM to Unity VR environment. The converted model can include both architectural (e.g. material, texture) and engineering data (e.g. beam loads). The approach looks promising, but the interaction system and user interface need to be improved after the conversion. In addition, an automated system with realtime workflow can improve the usability of VR applications. The approach of  and Du et al. (2017) introduced real-time data synchronization between BIM and VR using metadata interpretation and communication protocol. According to their findings, the system provides better functionality in preserving design data, non-disruptive synchronization of design changes, and better consistency with the original material information compared to the other alternatives. In parallel to exporting models with semantic data, the conversion efficiency can also improve the performance of the VR application. The approach proposed by Chen et al. (2020) attempted to improve VR performance by reducing the number of polygons in the converted model. An example project using this workflow showed an 88% improvement in the VR application's FPS (frames per second). However, even in its simplified version, the application performance was not acceptable for VR (about 32 FPS), resulting in an unpleasant experience for users. The other solution to optimize data is to use LOD (level of detail) . LOD is widely used in the gaming industry and specifically for 3D games due to the need for constant FPS in runtime. This approach was not considered to any significant extent in BIM modeling because realtime rendering was not a priority. Since VR requires more computational power, converting data considering LOD can improve performance by reducing the processing and memory requirements for the objects that are out of focus or far away from the camera.
Laser scanning and photogrammetry: BIM information can be generated either before the construction of the building or even for the exciting ones. One of the common methods for converting the data from an exciting building into a BIM project is laser scanning and point cloud technology (Dinis et al. 2020;Pavelka, Matoušková, and Pavelka 2019;Pavelka and Michalík 2019). They can produce the model and material information of the building at a reasonable time and cost. In addition, they can visualize the current stage of a construction site in the construction phase (Vincke et al. 2019). However, in order to use them as input data for VR, data optimization is crucial. One of the specific use cases for Laser Scanning and Photogrammetry is in architectural heritage (HBIM), which has gained popularity in recent years, e.g.  and Garagnani (2017). In HBIM, 3D survey data is transformed into a digital representation of architectural heritage using informative point cloud models. HBIM allows different disciplines to remotely access historical building data and achieve a new level of collaboration and data sharing (Banfi 2016). In this way, VR can help visualize and interact with these immersive environments (Banfi, Brumana, and Stanga 2019a). This information can be presented in VR either as 360 • panoramas (e.g. Banfi, Previtali, et al. 2019) or as 3D-textured meshes (Banfi, Brumana, and Stanga 2019b). 3Dtextured meshes are more interactive than panoramas, but they require more computing power in runtime, especially in VR implementation. In this case, simplifying the scanned model and defining LOD for each mesh can improve the application's overall performance ).
Consideration of LOD can even be used as a tool to focus user attention on specific points in virtual environments. A study by  shows that defining higher LOD values can guide participants through a virtual reality narrative.

Project management and collaboration
BIM is a well-known and common solution for project management and planning. It can provide insights into AEC projects in different phases and times. Traditionally, updating the project phase was done manually according to on-site observations and comparing actual and target states. This process is very labor-intensive, especially for large projects, since it required every small part of the project to be considered. Recent developments in VR, BIM, machine learning, and image processing can facilitate this process. By monitoring the actual phase of the project and combining image processing and machine learning, it is possible to update the BIM model accordingly (Dinis et al. 2020;Rahimian et al. 2020).
On-site implementation: VR can be used as an interactive tool to compare the as-built and asplanned state Kun-Chi Wang et al. 2018). Here, one of the gaps in the use of VR in construction is the technical implementation on site (Nasrazadani et al. 2020). This gap is because VR usually requires a predefined area for comfortable interactions and the setup of sensors and computers (Nasrazadani et al. 2020). Moreover, on-site personnel are less well trained and educated to work with BIM (Nasrazadani et al. 2020). A research by Nasrazadani et al. (2020) propose a solution to fill this gap by implementing a VR-based BIM simulation in a 30-foot trailer onsite. They equipped this trailer with a VR setup for visualization and an omnidirectional treadmill for walking in the virtual environment. The bidirectional connection between VR and BIM provides the capability to read, annotate, and modify any object in the environment in real-time. In addition, the environment can host a virtual meeting with multiple users for project collaboration.
Collaboration: As a key requirement in AEC projects, improving the collaboration between project stakeholders can enhance their understanding of project and construction performance ). This kind of collaboration requires real-time interaction and face-to-face conversation. However, it is difficult to achieve this when the various stakeholders work in a visually disconnected and non-immersive environment .
The traditional concept of face-to-face communication in projects can be extended or replaced by in-VR communication (A type of communication using VR user interface and corresponding interactions).  developed a cloud-based multiuser VR application called CoVR to achieve this goal. Their study shows that users in CoVR perform better on assigned tasks than users in the traditional desktop application. In addition, CoVR improves interpersonal interactions and communication during construction tasks. A recent study has shown that the results of in-VR communication are comparable to traditional face-to-face communication in most aspects, in particular in discussion quality (level of effectiveness and satisfaction experienced), communication richness (detailed responses and vivid messages), and openness (enjoyableness and open-mindedness). The accuracy and appropriateness of communication were higher in face-toface communication, however, due to the weak human-human interaction in the current generation of VR .
Virtual collaboration can be an effective tool in problem definition and solution-finding. A study by T. H. Wu et al. (2019) compares virtual and traditional BIM in this way. Their results show that the virtual experience, either in VR or in a game-like PC application, can lead to higher performance in problem finding. As a joint walk-through, it can improve the efficiency of communication among stakeholders and motivate them to have the same vision for the project Du et al. 2016).
In the case HBIM, in addition to visualization and interaction with cultural heritage, risk management is a crucial component of its preservation. Given the importance of communication between heritage managers and conservators, VR can play an effective role. An example of this application is a project by Lee et al. (2019) that demonstrates an integration of architectural heritage risk management information in VR. Thus integration can simplify the process of information exchange between heritage managers and conservators.
Disaster management: BIM can be used for disaster management because of the detailed visualization of building information. Implementing BIM in a VR environment can help planners manage crises. Planners can simulate crisis conditions, such as an earthquake or fire scenario, by visualizing safe and dangerous zones on the evacuation route in VR (Kanak et al. 2020;Getuli et al. 2020). VR can be used to reduce safety accidents in underground pipelines and improve the operational and maintenance efficiency of these pipelines (Chih-Ming Wu et al. 2020). In accident reconstruction, this technology can also be used to immersively recreate the sequence of events (Patti et al. 2017).
Workspace planning: One of the effective factors that crucially influence the succession of a project is site planning . In this case, VR can play an essential role in the planning of construction workplaces. It can be used both as a learning environment for workers and as a planning tool for construction managers (Yu et al. 2018). It offers several advantages to construction companies to set up a virtual construction site for information sharing between disciplines. It can visualize the execution methods on the construction site to help site engineers better understand the procedures (Quan 2019). In addition, the 4D simulation workflow in VR can provide a supportive environment for constructability analysis meetings (Boton 2018). However, a proper workflow is required to update the virtual environment according to the current state of the design (Vincke et al. 2019). Although users in immersive VR experiences can better grasp the visualization of a construction project, a study by Muhammad et al. (2019) has shown that traditional 2D methods of visualization and interaction are less time-consuming and easier to use. These results may reflect that being more time-consuming is either an inherent property of immersive VR or that the current interaction design and user interface are not adapted to be comparable to traditional representations. We believe that future studies in this field are needed to clear out the issue.
Building maintenance: The maintainability of a building is one of the features that should be considered in the pre-and post-construction phases. This has not traditionally been considered in design, however, because no technology or tools were available for it (Akanmu, Olayiwola, and Olatunji 2020). With the advancement of BIM, it is more convenient to include maintainability considerations. In addition, it can make the accessibility of various facilities easier and safer. VR can help improve the visualization of facilities and verify their maintainability in advance. An example of incorporating VR and BIM for building maintenance is a study by Akanmu, Olayiwola, and Olatunji (2020). They developed an automated system to consider the facility manager's input regarding the accessibility of building components for maintenance. Accordingly, real-world tasks, such as collision detection in MEP systems, can be integrated into a VR environment as a collaboration and visualization tool. Furthermore, this functionality can practically solve the problem of accessibility of MEP systems in case of repair or replacement of a component (Zaker and Coloma 2018).

Technologies used
The required technologies to implement VR in the BIM project can be divided into software and hardware tools. The software contains BIM-based tools (such as Autodesk Revit and ARCHICAD), a data exchange tool (to prepare BIM data for VR simulation package), and a VR simulation package (commonly game engines, such as Unity and Unreal Engine). However, according to the used software and project requirements, some of these tools can be neglected or included in the other tools.
The most commonly used BIM software in the literature studied is Autodesk Revit, which may result from accessibility, free trial and 3-years student license, and availability of several plugins for this tool. Most studies used game engines to use the exported BIM model and create the VR environment and implement interactions, e.g. Unity: Getuli et al. (2020), Balali, Zalavadia, and Heydarian (2020); Unreal Engine: Nasrazadani et al. (2020), . In some studies, however, the authors used Autodesk Revit plugins only to do the VR simulation, e.g. Enscape: Kamari, Paari, and Torvund (2021). Although using game engines for VR simulation may provide more flexibility and customization options, these plugins can enable rapid implementation of VR walk-throughs. Figure 4(a) shows the percentage of the game engine used in the selected literature. Based on what we observe in the literature, the most used game engine is Unity (65.8%), followed by Unreal Engine (17.8%). The rest of the studies (16.4%) used other tools or plugins to provide the VR experience of BIM.
From the hardware aspect of required technologies, accessing the VR environment is mainly possible via HMDs. Due to the ease of use and simplicity of interaction with these devices, they have attracted more attention in recent years. Several manufactures currently provide HMDs (such as HTC, Oculus, Valve, and HP) with different display quality, controller layout, and tracking systems. Although these have different controller layouts, they support standard key-bindings for interactions provided by game engines. In addition, from the tracking system aspect, they can be classified into two categories: using an external tracking system and using internal tracking systems. Some of HMDs support both systems, however, in order to provide more accurate and flexible tracking options.
The analysis of the reviewed papers shows that the authors used HTC Vive/Vive Pro and Oculus Rift devices in about 81% of the studies (see Figure 4(b)). Both devices require a PC as a computational source, and they can connect to the PC mainly via USB and HDMI cables. There is also the possibility of wireless connection for these devices, which was not mentioned in the papers. Both devices support a pair of controllers for VR interactions. For the tracking system, they use two different technologies, but both require a predefined environment setup. About 16% of the reviewed articles used mobile VR, and only a tiny part of the studies used other technologies like CAVE (Cave Automatic Virtual Environment).

Discussion
Application: As we have established consistently throughout the literature, VR-enabled BIM in AEC industries has diverse and influential applications. These applications can enhance conventional use cases of BIM or even create new opportunities for various stakeholders. One of the main applications of VR in the AEC industry can be an educational and training tool for professionals and students. Although both professionals and students find VR helpful in education and training, implementation is a point of contention. Professionals prefer less expensive and faster solutions. Students, on the other hand, prefer a more interactive experience. We can interpret this as the interactive VR experience not providing sufficient benefits compared to the cost from the perspective of the professionals. In the context of achieving acceptance by the professionals, the current interaction types, hardware, and software need to be studied in depth for future improvements. In divergence from to the professional perspective, interactive VR improves understanding, decisionmaking, and creativity on the part of students in various aspects of AEC projects. Although the impact of VR-based training looks promising, the long-term impact and applicability of this type of training in future projects should be further investigated. The benefits of using VR in decision-making are not limited to the training phase and can play a significant role during the design, construction, and maintenance of the project. In addition, VR can facilitate the participation of non-specialists in design decisions, which can improve communication among stakeholders and reduce future design changes. Interactive, straightforward, and easy-to-use interfaces for non-professionals can improve their efficiency in working with VR-enabled BIM and their willingness to continue using it. Furthermore, incorporating real-time collaboration capabilities among all project stakeholders can improve the functionality of each task. In this case, using a unified solution for implementing VR in the BIM project and exchanging data between different disciplines is essential. This solution can thus be considered as hardware and software sections. For the hardware section, the enhancement in the technology of HMDs as well as in site implementation can facilitate the procedure. On the other hand, using tools that enable users to implement VR interactions easier and exchanging data between different tools is required from the software perspective.
Technology: Considering the need for accessible and efficient implementation of VR experience in BIM, using game engines is gathering considerable attention in recent years. Unity and Unreal Engine are the most used ones in the literature among the different available game engines. The Unity game engine was released as a free tool in 2005, and many companies and researchers have built their applications with it. Given Unity's history, it is clear that a majority of studies (65.8%) use it as a VR implementation platform. The other advantage of Unity is the wide range of plugins and pre-built assets for objects, materials, and even VR interactions. The SteamVR, Oculus XR, OpenVR, and VRTK plugins are some of the commonly used tools that simplify the implementation of VR interactions in Unity. These plugins provide several predefined examples for different conditions, making them easier and more efficient to use. In Unity, developers can use the C# scripting language for general application mechanics and gameplay as well as Bolt visual scripting tool (https://assetstore.unity.com/packages/tools/visual-scripting/bolt-163802). The visual scripting tools offer the possibility to implement gameplay and mechanics even for non-programmers. On the other hand, the Unreal Engine was released in 1998, but not as a free tool at that time, which resulted in fewer researchers using it as a visualization tool, even for non-VR applications. The fourth version of the Unreal Engine changed plans and was released as a free and publicly available game engine. Since then, more studies have been attracted to the Unreal Engine. The use percentage of use of this tool (17.8%) in the selected studies is much lower than that of Unity. Similar to Unity, a variety of assets and plugins are available in Unreal Engine to facilitate VR interaction implementation, e.g. SteamVR, Unreal Standard VR Template, VR Expansion Plugin. The scripting language in Unreal Engine is mainly C++ and Blueprints (a visual scripting tool, https://docs.unrealengine.com/en-US/ProgrammingAndScripting/Blueprints/index.html). The native implementation of Blueprints in the game engine makes the scripting workflow efficient and straightforward. In addition, the Datasmith plugin for Unreal Engine supports data exchange between multiple BIM tools (e.g. Autodesk Revit, SketchUp pro) and formats (e.g. IFC) with Unreal Engine faster and more structured. The rest of the studies (16.4%) used other tools for VR visualization. The most common in this remnant group is the Enscape plugin, a commercial tool for real-time rendering and VR.
As we mentioned in the 'Results' section, most studies used HTC Vive or Oculus Rift as the HMD. This choice may be a result of higher screen resolution, enhanced interactions, and better performance. On the other hand, the proportion of mobile base VR usage cannot be ignored. The advantages of the mobile version are lower price and, most importantly, wireless connectivity, which means wider availability and a more straightforward setup. However, these advantages come at the price of lower computing power, resulting in lower fidelity and quality of visualization. These barriers limited VR interaction, only walk-through in some cases (Tariq et al. 2019), and mobile VR application in the literature. In recent years, all-in-one VR devices, such as Oculus Quest 2, are becoming more popular because they offer better display quality, higher processing power, and affordable cost. These devices can lead to a solution for implementing VR simulations wirelessly, even on-site.
Evaluation: The most commonly used methods in the reviewed papers for evaluation are either questionnaires, e.g. Ahmed (2020) and Castelo-Branco, Leitão, and Brás (2020), or implementation in real, e.g. Rahimian et al. (2020), and custom projects, e.g. Ortega et al. (2020). In the case of questionnaires, either standardized questionnaires, e.g. Abbas et al. (2019), or customized questionnaires, e.g. Quan (2019), were used. This approach can be helpful, but the effectiveness of the questionnaire design and items should be carefully evaluated (Safikhani et al. 2021); otherwise, the results can be influenced by some unwanted bias. The target groups varied from professionals, e.g. Wiberg et al. (2019), to students, e.g. Cha et al. (2019), to non-professionals, e.g. Getuli et al. (2020). To evaluate a newly developed method or tool, they usually used it in a project. In this case, they consider either an actual project or a customized project to evaluate the application with expert and non-expert target groups.

Limitation and future works
The reviewed publications show that there is much recent development involving BIM and VR. In some of the literature items, the software or HMD device used was not clearly mentioned. These were thus removed from our analytical results. The reviewed publications have diverse specific topics. Hence, the future work sections diverge into these perspectives. In the following, some of the relevant ideas and discussion points for future research are identified.
Education and training: In an educational environment in particular, it is crucial to improve the setup to be accepted, practical, and comfortable. After successfully probing VR prototypes in a classroom, it is essential to address found limitations and improve the prototype. The implementation of assessment tools can help to quantify students' skill gain (Kandi et al. 2020). Furthermore, it can be helpful to include user proposals in order to minimize negative aspects. An intuitive tutorial can help the users to adapt in a more straightforward manner with the environment. In addition, more realistic detail (e.g. lighting, models) can improve their understanding and engagement during the experience. Chen Wang, Li, and Kho (2018) recommend developing detailed 3D structural modeling to further support immersion in the virtual world. Sánchez-Sepúlveda et al. (2018) point out the importance of improvement in the visualization method for the user interaction on the user experience. The area of improvement can be the quality of the graphics and reality of the geometries, as well as the interaction with space and objects. Hafsia, Monacelli, and Martin (2018) propose the use of a robotic force feedback platform to increase immersion. In this case, ergonomics should be taken into consideration as well. Video capturing techniques can be used for the evaluation of ergonomics and gestures in order to avoid any physical health risks.
Design and data exchange: Continuous progress in hardware, software, and mobile connectivity provide lots of optimization and enhancement possibilities for visualization, performance, and data transmission. One of the existing barriers is the lack of interoperability in 3D CAD software. Kado and Hirasawa (2018) describe the potential for 3D model-based design methods, including the possibility to use game engines as an interface for BIM data. They highlight the benefit of bidirectional updates of 3D models between architectural 3D CAD software and the VR application, respectively, the game engine. Their plans include further two-way co-operations between CAD software and other game engines. In the case of data transfer, automatizing and simplifying the model adaption process could be a point of interest (Raimbaud et al. 2018). Before the implementation of VR in AEC processes, suitable VR technologies should be considered. Ventura et al. (2018) highlight that processes, users, and requirements should be analyzed beforehand. Several studies mention 5G technology as a future enabler (Banfi, Brumana, and Stanga 2019b;Chew et al. 2020; Alizadehsalehi, Hadavi, and Huang 2020). Solutions on mobile devices in particular often depend on the way content is transferred. When a considerable volume of data needs to be transmitted, bandwidth can be a bottleneck that needs to be considered during the design. The current infrastructure development in the 5G context will allow a much higher bandwidth in the future. Accordingly, it will be possible to transfer more data (i.e. more detailed 3d models) and with better latency that allows faster transmission of more data, i.e. more complex structures to enhance the user experience. Regarding immersive visualization for construction sites, the import process to the game engine and the visualization performance can be optimized (Vincke et al. 2019). Further possible enhancements are proposed, such as the integration of facility management data and measurement tools. Khalili (2021) point out the barriers of data exchange between BIM and VR applications. First of all, data exchange in AEC projects is time-consuming, which can be a serious obstacle in megaprojects. Each software uses a specific data structure, which makes data exchange more cumbersome. In addition, the transferred data must be optimized for the VR application and be capable of interacting depending on the application. In their study, Khalili (2021) propose a novel method enabling better interoperability. They conclude that further limitations can be overcome, thus enabling more VR applications in practice. Defining a general template package for implementing VR-enabled BIM can improve the usability of the system by considering predefined interactions, a suitable user interface, and easy real-time data exchange. Since real-time data exchange needs to be consistent and performant, a procedural workflow for converting point cloud data into VR-enabled content can be helpful. Such photogrammetry and procedural modeling workflows can be found in commercial packages for the gaming industry, such as the Quixel Megascans library (https://quixel.com/megascans/). Balali, Zalavadia, and Heydarian (2020) report the benefits of 'end-user experience before the construction phase.' They highlight several possibilities in a virtual environment that can give users custom modifications, such as the ability to change objects or material textures instantly. Further, connecting multiple VR headsets to the same environment could enhance communication and collaboration.
Project management and collaboration: Suitably adequate means of communication are crucial for efficient collaboration, especially in a virtual environment. Abbas et al. (2019) point out the importance and future potential of non-verbal communication (e.g. facial expressions, body posture, gestures, and eye contact) to minimize misunderstandings and support efficient communication. Several studies highlight the future impact of VR in Smart City planning and maintenance (Götzelmann and Kreimeier 2020;Chih-Ming Wu et al. 2020;Chew et al. 2020). Bourlon and Boton (2019) point out the future potential of auto-generation of 4D VR simulations from a BIM model and a given schedule. A visualization of the working schedule in VR can be beneficial for a construction project to provide more accurate insight into the project state. The visualization of BIM Information can support, for instance, facility management before conducting a visit in person. Akanmu, Olayiwola, and Olatunji (2020) note that future work should evaluate VR support for reviewing designs for maintainability.

Conclusion
In this paper, we reviewed the literature concerning the application of VR in BIM for AEC projects. We conducted a systematic review in Scopus, ACM, and IEEEXplore for assembling the technical papers. The selected articles were screened in such a manner that only those papers that discussed immersive virtual reality were used, and no other types of XR or traditional non-immersive VR. We classified the articles according to their primary application into these categories: (1) education and training, (2) design and data exchange, (3) project management and collaboration.
Advantages: (1) The reviewed papers indicate that VR can improve the motivation and understanding of students through the interactive and immersive environment. Therefore, we can conclude, by implementing VR in the training process, we can provide a safe, collaborative and engaging experience for learners. In addition, VR can improve spatial skills and understanding between experts and students. (2) During the design procedure of an AEC project, VR can be beneficial as a visualization tool for different steps in the workflow. By defining a platform for exchanging data between technologies such as laser scanning/point cloud with VR, stakeholders can compare the as-planned and as-built conditions of the project in real-time. (3) VR can provide an insight into the hidden facilities such as MEP and the possibility to consider their maintainability in advance. (4) It can be employed as a testing environment for future smart cities or as a tool for maintaining and interacting with historical buildings. In addition, VR can be used to enhance the procedure of algorithmic design. (5) In the case of collaboration, VR can replace or extend face-to-face communication in the project, as several studies show that virtual collaboration can be an effective tool in problem definition and solution-finding. (6) Furthermore, it can help project management by improving workspace planning. Finally, VR can provide an opportunity to simulate disaster conditions in advance, providing the capability to manage these in the project.
Limitations and disadvantages: In addition to all the benefits, however, VR like any other technology has its limitations and disadvantages. (1) Professionals choose the non-interactive VR experience over the interactive one due to lower cost and easier/faster interactions. Some reports show that traditional 2D methods and visualization are less time-consuming and easier to use.
(2) In the case of communication, accuracy and appropriateness of communication are higher in face-to-face communication due to the weak human-human interaction in the current generation of VR. (3) Furthermore, the difficulties or timely process of exchanging data between BIM and VR limit VR usage in AEC industries. Real-time data transfer is the ultimate goal.
Opportunities for improvement and future research areas: (1) The cost of VR equipment can be reduced by considering the application of Mobile-VR. Recent developments in standalone VR devices, e.g. Oculus Quest 2, with high-quality displays, supporting hand recognition at a more affordable price, can be a solution for encouraging professionals to use interactive VR experiences. In this case, future studies on the influence of this equipment on the professional application of VR in AEC industries are needed. (2) The reviewed articles indicate that one of the missing topics to be investigated in future research is the influence of visualization and interaction type on the user experience. Over the years, traditional desktop experiences have evolved and improved to match user expectations and requirements. Accordingly, conversion from a traditional desktop application to immersive VR needs consistent application design guidelines (Safikhani, Holly, and Pirker 2020;Hepperle et al. 2019). These guidelines are crucial due to the results reported by professionals on their experiences in the VR environment. An adequate study that will lead to defining a guideline for designing VR environments, user interfaces, and interactions considering BIM application will thus be essential. (3) In order to study the user experience in VR, it is necessary to make a fully adequate assessment of the application scenario (Kim, Rhiu, and Yun 2019). In this case, the study of design guidelines can be extended by exploring different evaluation methods and devices to assess the user experience in VR. (4) To improve interactions in VR, studies on employing successful ideas from other types of XR in VR can be beneficial, e.g. Prouzeau et al. (2020). (5) As discussed, future improvements in the intervention of end-users in design procedures can enhance the general workflow and develop a more optimized VR experience. In addition, the utilization of procedural and data-driven scenarios, e.g. Mo et al. (2018), in the design of VR environments can enhance the user experience and simulation outcomes.
As a summary, the majority of reviewed articles reported the positive impact of VR on training delivery, design, or project planning. The immersive virtual environment can improve project flow and the productivity of stakeholders. In addition, the applications of VR are not limited to professional users, which means that non-professionals can also contribute valuable comments to the project. Although these results look promising, the limitations of VR (e.g. interactions, setup, and cost) should also be considered. Nevertheless, the way ahead appears to be clear, and that by a rapid improvement in VR, both from the hardware and software perspectives, it will be possible to use this technology more frequently in various AEC projects in the future.

Data availability statement
The data that support the findings of this study are openly available in the repository of "KityVR_review" at http://doi.org/10.17605/OSF.IO/SZVPC.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by Österreichische Forschungsförderungsgesellschaft and TU Graz Open Access Publishing Fund.