Asl et al. [5] contributed to research on BIM, parametric modeling, and automation as a new trend in sustainable building design. Parametric modeling refers to a modeling process with the capacity to change the shape of geometry at the

same time as the dimension value is modified [13]. These new methods to evaluate the performance have opened opportunities to optimize solutions. They presented a case study using a newly developed integrated parametric BIM-based system to interact with ES and daylighting tools to optimize building energy performance. The objective of the case study was to allow architects to explore design alternatives and, at the same time, assess the building performance to search for the most appropriate design. Regarding competences, they stated that ES are complex and most architects do not have enough expertise and knowledge to manage them. To improve building performance, they claimed that architects need to have access to the ES results and the ability to understand and interpret the results. They also mentioned that BIM contains most of the data needed for ES and consequently, these simulations should be moved to the EDP to define the best energy efficiency solution to develop in the next phases. Even though the case study was simple (a residential building), the results demonstrated the potential of using a parametric BIM-based system towards more complex architectural modeling.

Andriamamonjy et al. [28] performed a combined review, scientometric and conventional literature review, to understand the BIM and ES integration as a potential strategy to improve interoperability. They concluded approximately the same as Asl et al., i.e. the available ES software requires manual input and a deep knowledge of the tools’ interface, which explains the limited use of ES by architects who lack expertise in the matter. Furthermore, they stated that in case of incomplete data, modelers often have to provide default values established on their knowledge and this default data are seldom documented. For these reasons, the outcome of the ES can vary from a modeler or energy expert to another since the results are dependent on their experience and judgment. Their article revealed some limitations of the BIM2BEM process. For example, they claim that to achieve total interoperability, several issues need to be solved, e.g., data loss, incompatibility, missing information, and the modelers’ limited competences.

Arayici et al. [29] research approach on interoperability also indicated that architects are able to perform BIM modeling, but they often have a poor understanding of the information they should share, and at which time, with energy and environmental experts. They adopted the design science methodology approach to encourage early collaboration by developing interoperability specifications. Design4Energy was used as a case study. Design4Energy is a research project funded by the European Union (EU) aiming to develop an innovative and integrated design methodology to predict the current and future energy demand of buildings. They mentioned that architects are also novices in the ES field, which makes it difficult for them to know what information is relevant to experts. Simply put, architects can do BIM modeling, but they have in general very little understanding of what information is relevant and when they should share it with the client and energy experts. The author’s approach contributed to the practice knowledge by understanding critical aspects of future projects’ interoperability developments in EDP.

Design Phases

Jabi [30] outlined that the majority of ES are achieved later in the design phases, when design modifications are difficult to integrate and very costly. Most of the time the ES is conducted to measure compliance with external regulations rather than to fundamentally change the design towards higher performance. For these reasons, Jabi developed and presented a software called DSOS. This tool integrates the use of a parametric design system and an energy analysis engine for the EDP. The software is limited by the amount of time to complete the analysis. Jabi et al. [31] maintained that the ES tools are not architect-friendly, and they are too complex for early design of architecture. The tools are not either compatible with the architects’ working methods, which are both more intuitive and iterative. They believe that we need a new way of thinking about modeling in EDP. A few years later, Jabi et al., presented a topologic tool kit based on the DSOS software. Topologic modeling refers to build lighter models that represent space instead of just building fabric, which reduces geometric complexity in ES. The authors concluded that the tool kit needs further investigation and user-testing before being released. Their contribution to parametric and topologic modeling research is an important school of thought. This approach can be appropriate for architects in EDP, where complete BIM is not feasible and not needed.

Picco and Marengo [9], concluded the same as Jabi et al.: EDP design is critical. According to them, if you make poor decisions at the early stages, it is almost impossible to correct them in the next design phases. Despite all the new ES tools that have been created in the last ten years, architects still have difficulty using basic tools supporting EDP decisions. They often simplify the model without paying attention to the consequences of these simplifications. Instead of focusing on developing new software, Picco and Marengo looked in detail at the impact of model simplification on simulation results. The strength of their research was to identify which parts of the model can be simplified without affecting the results. The methodology was based on a previous simplification method by Picco et al. [32] and three different case studies. The results showed that the simplification on the building geometry does not yield a clear difference in the results compared to detailed models if the thermal zone is correctly defined. They also mentioned the need for dynamic ES to increase energy efficiency, but this integration is slowed down by the complexity of ES and the required level of details. There are benefits of applying ES in the EDP. However, the results are limited by the difficulties that arise in the process to define the right level of details at the right phase without compromising the results.

Farzaneh et al. [11] found that previous efforts have focused on detailed performance aspects for a particular tool or a detailed design phase instead of addressing the complete design process. Therefore, there appears to be a need to improve the design process by identifying the essential information required at each phase. They proposed a BIM-BEM framework using a methodology based on theoretical and field investigations and one case study. The innovation of the proposed framework was not only to focus on the technical aspects, but also, to look at the modeling steps involved in the design process. Subsequently, Farzaneh et al. [27] presented a literature review about BIM2BEM during the design process using a systematic approach. They outlined that each phase requires specific activities with an adequate LOD to run the energy ES, and thereby reach the performance targets established at the beginning of the design process. The authors concluded that further research was needed to fill the academic and professional knowledge gap.

According to Arayici et al. [29], the focus of the AEC industry has been on new tools and methods for improving the quality and structure of information over the project’s life cycle. In line with Farzaneh et al., they indicated, that the design process needs increased flexibility and a shared understanding between professionals about which information to share, at which stage, and using which tools and data models for BIM. The authors stated that “we cannot rely on adopting new digital tools such as Grasshopper or Dynamo to improve information using parametric modeling with BIM in the professional practice; this is not sufficient”.

Jabi et al. [31] presented a new way of thinking in the initial process, consisting of building the lightest possible model using minimum time and effort to be able to provide relatively accurate results. Their claim supports research by Farzaneh et al. about the importance of adequate LOD. The level of precision of a simulation needs to be compatible with the advancement in the design process. In other words, it would make no sense to build a high precision model when many parameters can still be changed. Precision and time are key aspects to consider in this endeavor. The accurate representation of the building is thus tightly related to time and design phase. There is a need to minimize the simulation time in order to make it compatible with design phase. One way to achieve this goal is to identify and follow the right LOD during the elaboration and development of the design. The authors noted that the full potential of BEM and LCA cannot be exploited at EDP, partly due to a low LOD of BIM.

Level of development

Building practitioners that are able to integrate performance targets using the right LOD in the EDP will necessarily improve the overall performance of the design. Devoting more time in the EDP in combination of the LOD use and its requirements can improve the level of information and detail in BIM objects. This will lead to fewer interoperability errors since the IFC information will increase.

The lack of ES tools compatible with a low LOD has been outlined as one of the main technology gaps. A model defined at LOD 100 will not allow to conduct, e.g. energy modeling required for a LEED certification. However, recent tools such as Insight (building performance analysis software integrated in Revit) can identify how the building’s energy use will influence the outcome and propose an interpretation of the results to adjust EDP design and obtain a better chance to meet the certification requirement [33].

With an early contribution about tools development, Andriamamonjy et al. [28] - [34] claimed that performing ES in the EDP with an architectural BIM and a LOD 200 can yield a reduction of energy use of about 20% from a reference model. Furthermore, they noted that the lack of ES tools compatible with a low LOD emphasizes a technology gap. According to these authors, developing solutions to bridge this gap will be beneficial for building design.

The research of Gomes et al. [12], indicated that LOD 300 is sufficient and compatible with current whole building LCA practice. As a methodology, they selected a case study to support the analysis of the variations between manual quantity assessment from design documents and automatic extraction of data at a suitable LOD level for LCA purposes. The results indicated that automatic extraction could facilitate the process by saving time, reducing data collection effort, and minimizing the risk of errors relative to the manual process. Nevertheless, the authors concluded that there is a possibility that LOD levels lower than 300 can be used during design and preliminary study phases.

Therefore, Nilsen and Bohne [35], suggest that the designers should develop competences with the requirements needed at each LOD level and use average values and best practice where the LOD level is low. Their research assessed LOD to improve the calculation of the environmental impacts at different EDPs more efficiently, using the methodology of LCA. They performed a literature review and a case study to evaluate BIM4LCA in the EDP. Finally, the authors speculated that investing more time in the EDP and using the appropriate LOD can contribute to improve the detail in BIM and reduce errors in BIM4LCA.

Eurac research about regenerative design in digital practice [8] demonstrated that ES tools are designed to make calculations, design and feedback possible, but these workflows remain outside conventional practice, meaning achieving only minimum standards that reduce impact instead of resulting in a positive impact. Moreover, the data analysis is still missing as basic knowledge about sustainable building design. While data are abundant, very little of this data is used to effectively inform the early design of the buildings since there are many other issues to consider by the architects, e.g. concept, context, functions, esthetic, noise, regulations, materials, etc. [36].

The main area we identified to improve the use of data with BIM tools are related to three main subtopics:

1) BIM and automation

Automation means the process of using a computer to perform an operation without involving a human brain [37]. In a review, Kamel and Memari [38], identified automation of modeling processes as one of the major benefits of the application of BIM in ES of buildings, which save time, cost and reduce human errors compared to the conventional energy modeling process. They presented a review of research studies focusing on different types of BIM schemes and ES tools capable of reading these files. In addition, they reviewed and analyzed the results of three case studies. Their results showed that many areas still require manual interaction as opposed to automatic data input, e.g. the current BEM tools cannot read the BIM files with minimum errors even if the required data are provided within the file. Anecdotal evidence from architectural practice also reveals that architect firms sometimes perform steady-state energy calculations by manually extracting geometrical data from digital models. In order to improve the transition from manual extraction of data to full automation, we identified LOD 100 as the right level to conduct a steady-state energy calculation towards building code compliance.

2) BIM2BEM and interoperability

In their review and case study, Kamel and Memari [38] also indicated that most issues slowing down the automation process stem from interoperability, which is the degree to which two programs are completely compatible [39]. This interoperability makes the information from BIM applications understandable while providing the possibility to share digital information between the disciplines, which reduces the need for traditional documents. The authors concluded that there is still a long way to reach the objective of performing the BIM2BEM interoperability process with the click of a button.

Andriamamonjy et al. [28] - [34] also identified that the lack of interoperability arises when semantic data (method to structure data to represent it in a logical way) has multiple definitions across different disciplines or when incompatible proprietary information models are used by the collaborators. As outlined by Kamel and Memari [38], even though the use of a standardized data structure such as Industry Foundation Classes (IFC) constitutes a solution, it currently presents a recurrent inconsistency and lack of semantic, which results in information loss. IFC data model is used to describe architectural, building and construction industry data [40]. The format is a scheme widely accepted by the AEC industry to exchange BIM models. It is using layers (resources, core, interoperability and domain) to describe the geometry, material properties and relationships in a BIM model, as mentioned by Farzaneh et al. [11] and Andriamamonjy et al. [28] - [34] concluded data interoperability of BIM with other software packages is presently inefficient, which makes predictions of operational energy use difficult for building practitioners. Part of it is due to the absence of a standard format and differences in data structure among various types of analysis software.

Gerrish et al. [41] outlined that the challenge is the conversion from one tool to the other, which must be performed carefully in order to reduce the risks of model divergence and errors. They developed a method that links design documentation and building performance to identify the technological requirements for BIM and building performance connection in the real world. They supplemented their method with interviews to identify associated behavioral and methodological challenges. In line with Andriamamonjy et al. [28] - [34], they concluded that the BIM2BEM conversion must be supported by technology to assist the sharing and visualization of ideas in virtual 3D format. Both tools should also be able to support regional regulations and global standards. The major limitation for BIM2BEM integration is the data loss issue, as well as the complexity resulting from employing many tools, especially in the EDP. The limitation of Gerrish et al. research touches the question: “How can information describing the performance of a building be standardized to enable the automated application of tools to provide an accurate representation of energy use? And how could this be supported in context with the common data environment using BIM?” We identified the preliminary design with a specification of LOD 200 as the accurate design phase to complete dynamic energy calculations for the operational energy use and compliance to global standards.

3) BIM4LCA and datasets

Gomes et al. [12] performed research about BIM4LCA using the methodology of a case study. Their results confirmed the need to have embedded environmental parameters and calculation into the model instead of pure material quantities to provide the necessary feedback to accomplish sustainable buildings. In other words, they claim that LCA performance should be calculated through direct access to BIM information. Currently, most of the LCA actually needs other software in addition to BIM. For instance, Revit Architecture is connected to the life-cycle impact assessment tool ATHENA Impact Estimator to investigate negative impacts from the use of several building materials [42]. Similarly, ArchiCad 14 BIM software is connected to Grafisoft EcoDesigner to calculate the carbon footprint [43]. Gomes et al. concluded that this makes predictions of the embodied energy use for the AEC difficult since the second software data interoperability is ineffective and not accessible by the other professionals. One way to make it more effective is to display the embodied energy parameters of building component in the software interface at the developed design phase with a specification of LOD 300. Figure 4 shows an example of the analytical properties interface with data from Revit software where environmental properties can be added e.g., U-value, total climate impact, thermal resistance, solar heat gain and heat transfer coefficient, etc.

A collection of data, referred to as datasets, is one of the most valuable items for future design practices. At the recent AI in AEC conference in Finland, Hjelseth (2021) explained the need for data to be transformed into information by the team of designers, researchers, and scientists, to create a machine learning environment. He claimed that technology, process and people competences are fundamental to integrate a data management technology. Machine Learning is a branch of artificial intelligence (AI), which allows to automatically create models from data. This technology enables computers to learn on their own without having been explicitly programmed by humans to do so. To learn and develop, computers need data to analyze and train on. Data is thus the essence of Machine Learning, and it is seen as the technology that unlocks the full potential of Big Data [44]. Hagström (2021), outlined that it is essential to understand AI on

the basic level. She explained that, at the moment, the industry still has problems to understand and analyze data to create algorithms. One of the reasons for this misunderstanding is that data are missing, or their source is unknown.

Besides the three topics discussed in this section, another major issue needs to be addressed by future research, namely, the divergence between the simulated and real energy performance of a building in the operation phase. The discrepancy between the simulated energy performance assessments and the real energy use of a building in the operation phase obviously leads to incorrect BIM- based energy analysis results. Gerrish et al. [41] results showed that the application of BIM to manage building performance during both design and operation is another potential end-goal for the post-construction phase. Reducing the gap between predicted and actual building performance is an area where much effort has been targeted, although BIM as a performance management tool has not been adopted beyond research until now.