Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Integrating GIS and BIM for community building energy design Bai, Yunpiao 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_february_bai_yunpiao.pdf [ 2.78MB ]
Metadata
JSON: 24-1.0340680.json
JSON-LD: 24-1.0340680-ld.json
RDF/XML (Pretty): 24-1.0340680-rdf.xml
RDF/JSON: 24-1.0340680-rdf.json
Turtle: 24-1.0340680-turtle.txt
N-Triples: 24-1.0340680-rdf-ntriples.txt
Original Record: 24-1.0340680-source.json
Full Text
24-1.0340680-fulltext.txt
Citation
24-1.0340680.ris

Full Text

 INTEGRATING GIS AND BIM FOR COMMUNITY BUILDING ENERGY DESIGN   by  Yunpiao Bai  B.E.s., The University of Waterloo, 2014  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF  APPLIED SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Civil Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) December 2016  © Yunpiao Bai, 2016  ii   Abstract  Increasing urbanization has caused a corresponding increase in energy consumption from the design, construction and operation of the built environment. To achieve energy-efficient design in urban communities, the design phase needs to adopt reliable energy modeling approaches. However, current urban modeling approaches often use abstract and low level information to describe buildings because of the difficulties of collecting and managing building data on the large scale required of such urban communities. This abstraction of building data creates large uncertainties in the modeling and simulation of energy scenarios at the community level. An additional consequence is a general separation between community energy design (with low level building information) and building energy design (with high level building information). An important part of the solution to this challenge relies on the integration of information systems at the scale of both urban communities and individual buildings, which are based on Geographic Information System (GIS) and Building Information Modeling (BIM) respectively. Since current technologies do not sufficiently address the interoperability between GIS and BIM, the existing conversion between GIS and BIM does not satisfy the data requirements for community energy design. This thesis investigates this challenge and presents an approach that uses Semantic Web technologies, including OWL (Web Ontology Language) and RDF (Resource Description Framework), to integrate GIS and BIM data. In this approach, we first develop relevant design scenarios for energy consumption in buildings of the University of British Columbia (UBC) campus. Based on the scenarios and required information for the energy simulation, we create a suitable ontology to transform the data into a Semantic Web model. Then we conduct relevant queries on the transformed data to provide the required information for energy simulation of a UBC campus neighborhood that contains richer and more detailed building information that is extracted from the campus building information models. Finally, we visualize the simulation results in a three-dimensional environment and discuss how it supports designers and decision makers engaged in community planning and design.    iii   Preface  All the work in this dissertation is done by the author Y. Bai except the energy simulation results in Chapter 4.5.1, which are provided by Juchan Kim.  This research was not previously published in whole or in part.   iv   Table of Contents  Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iii Table of Contents ......................................................................................................................... iv List of Tables ................................................................................................................................ vi List of Figures .............................................................................................................................. vii Acknowledgements ...................................................................................................................... ix Chapter 1: Introduction ............................................................................................................... 1 1.1 Background and Problem Statement ................................................................................... 1 1.2 Research Objectives ............................................................................................................ 5 1.3 Research Methodology ....................................................................................................... 6 1.4 Overview of the Dissertation .............................................................................................. 9 Chapter 2: Related Background ................................................................................................ 10 2.1 Urban Energy Modeling ................................................................................................... 10 2.2 Overview of GIS and BIM ................................................................................................ 12 2.3 GIS and BIM Integration .................................................................................................. 15 2.4 Semantic Integration and Ontology-based Approach ....................................................... 19 2.4.1 Overview of Semantic Web ......................................................................................... 19 2.4.2 Applications of Semantic Web .................................................................................... 21 Chapter 3: Overview of Research Methodology ...................................................................... 24 3.1 Background on the PICS Research Project ....................................................................... 27 Chapter 4: Approach, Implementations and Results .............................................................. 31 4.1 UMI: An Environment for Urban Energy Modeling ........................................................ 31 4.1.1 UMI Workflow ............................................................................................................ 32 4.1.2 TLF: Standardized Building Property Template ......................................................... 32 v  4.2 Design of the Data Integration Approach ......................................................................... 36 4.3 Case Study: the UBC Vancouver Campus........................................................................ 37 4.3.1 Data: Geodatabase and IFC ......................................................................................... 37 4.3.2 Design Scenarios .......................................................................................................... 39 4.4 Implementation of the Approach ...................................................................................... 43 4.4.1 Ontology Development ................................................................................................ 44 4.4.2 Data Extraction and Transformation ............................................................................ 51 4.4.3 Data Query and Loading .............................................................................................. 59 4.5 Results Visualization ........................................................................................................ 61 4.5.1 Energy Simulation Results ........................................................................................... 61 4.5.2 Results Visualization ................................................................................................... 63 Chapter 5: Conclusion ................................................................................................................ 67 Bibliography ................................................................................................................................ 70         vi   List of Tables  Table 2.1   Key factors of building energy performance [3] ........................................................ 11 Table 2.2   Summary of data requirements for SCU [22] ............................................................. 12 Table 2.3   Mapping between IFC and CityGML classes [9] ....................................................... 16 Table 2.4   IFC-UBM-CItyGML mapping [33] ............................................................................ 17 Table 4.1   Data contents of the TLF [48]..................................................................................... 33 Table 4.2   Ventilation-related Properties ..................................................................................... 41 Table 4.3   Material attributes in the TLF ..................................................................................... 42 Table 4.4   Translation from Geodatabase to the campus-ventilation ontology ........................... 53 Table 4.5   Mapping between IFC and the building ontology classes .......................................... 55 Table 4.6   Mapping between the linked ontology and the TFL classes ....................................... 60    vii   List of Figures  Figure 1.1   An overview of research steps ..................................................................................... 7 Figure 2.1   A sample RDF graph ................................................................................................. 20 Figure 2.2   An example of OWL ................................................................................................. 20 Figure 2.3   A generic workflow for Semantic Web deployment ................................................. 22 Figure 3.1   An overview of research steps ................................................................................... 25 Figure 3.2   The PICS project action plan ..................................................................................... 27 Figure 3.3   An overview of the PICS energy design workflow ................................................... 28 Figure 4.1   Research steps for developing and implementing the approach ............................... 31 Figure 4.2   A workflow of UMI................................................................................................... 32 Figure 4.3   The TLF structure ...................................................................................................... 34 Figure 4.4   The class diagram of the TLF structure ..................................................................... 35 Figure 4.5   An overview of data integration approach ................................................................ 36 Figure 4.6   A map of the target UBC neighborhood.................................................................... 38 Figure 4.7   The IFC model of Engineering Student Center (ESC) .............................................. 38 Figure 4.8   Sample UBC ventilation policy [40] ......................................................................... 40 Figure 4.9   An example of ventilation settings ............................................................................ 41 Figure 4.10   Sample UBC design material palette [44] ............................................................... 42 Figure 4.11   An example of collected opaque materials .............................................................. 43 Figure 4.12   A RDF graph of the Campus-Ventilation ontology developed for Design Scenario #1................................................................................................................................................... 45 Figure 4.13   An example of the IFC structure [44]...................................................................... 47 Figure 4.14   An example of redefined building structure ............................................................ 47 Figure 4.15   Classes of building elements, their common properties and spatial structure in the new building ontology .................................................................................................................. 48 Figure 4.16   The building ontology representing building materials for Design Scenario #2..... 49 Figure 4.17   The implementation of GeoSPARQL ..................................................................... 50 Figure 4.18   An example of linking two ontologies .................................................................... 51 Figure 4.19   The designed functionalities of the FME plug-in .................................................... 52 viii  Figure 4.20   Examples of Geodatabase transformation in FME .................................................. 54 Figure 4.21   The structure of material information in IFC .......................................................... 57 Figure 4.22   Processes of customized transformers for material data manipulation ................... 58 Figure 4.23   An example of RDF triples in Oracle database ....................................................... 59 Figure 4.24   An example of the final TLF ................................................................................... 61 Figure 4.25   An example of the energy simulation results .......................................................... 62 Figure 4.26   ESC building parts for energy simulation ............................................................... 62 Figure 4.27   A map of annual total operational energy use ......................................................... 64 Figure 4.28   A map of annual energy use per square meter ........................................................ 65   ix   Acknowledgements  I would like to express sincere gratitude to my supervisors Dr. Sheryl Staub-French and Dr. Rachel Pottinger, who have inspired me throughout my Master’s studies. I also owe particular thanks to Dr. Puyan A. Zadeh, whose patience to my endless questions in this research.   I thank tremendous help to my research from Dr. Jon Salter, Juchan Kim from the School of Architecture and Landscape Architecture. Staff from Safe Software Inc. also provided superb support to my work, including Dave Campanas, Steve MacCabe, Paul Nalos and many others who spent their time.   Special thanks to my family for their love and support throughout my years of education.     1   Chapter 1: Introduction  1.1 Background and Problem Statement  Increasing urbanization has caused a corresponding increase in energy consumption from the design, construction and operation of the built environment. According to the Natural Resources Canada’s reports, every year, 31% of energy use and 28% of Greenhouse Gas (GHG) emissions come from commercial, institutional, and residential construction [1, 2]. This highlights the importance of energy policies in the construction industry and the need to optimize building energy performance for achieving sustainability goals within urban communities. Alternative energy policies for decreasing energy consumption and the related emissions are usually combined with financial factors that need to be considered during the design phase through measuring and predicting energy costs. Given such consideration, urban energy modeling, a computerized process to simulate the energy consumption related to buildings, is an essential tool for planners and decision makers in urban planning and design.  At the community (neighborhood to municipal) scale, designers use urban energy modeling to understand building energy consumption and to develop and manage sustainable energy strategies. In this way, community planners are able to review the energy costs of existing built environment factors and evaluate various energy strategies that can increase energy efficiency in communities. Unfortunately, it is difficult to collect and maintain detailed building data from a neighborhood with a large number of buildings. Therefore, current urban modeling processes usually work with relatively low-level and abstract building data [3], and as a consequence, sometimes planners have to make assumptions about unknown building data, such as the number of floors and rooms, or thermal properties of walls. This can cause many uncertainties and errors in the final energy modeling results.   An additional problem is the separation between the energy design processes at the community scale and the building scale. The individual building energy consumption directly affects the energy performance of the whole community, and the external environment (e.g., climate 2  conditions and community policies) also dictates the design of building systems. Thus, the two design processes at the different scales are interrelated and should not be treated as independent from each other. The deficiencies described above demonstrates the need for new approaches and technologies that integrate both community-scale and building-scale data to support the overall energy design of a community.   At the urban scale, Geographic Information System (GIS) technologies provide powerful scalable visualization and spatial analysis. The adoption of GIS has become part of the mainstream in energy modeling and large-scale mapping [4], from building blocks to geographic regions. However, a building is often treated ‘as a whole’ in traditional GIS platforms, and the geometries of building components and their related properties are highly simplified. For instance, walls do not have detailed geometric and material property information in traditional GIS platforms, which is a weakness in representing building details, particularly given the importance of enclosures in assessing building performance. Hence, there is a need to represent and provide building data at a high level of detail (LOD) to be integrated into the urban energy modeling process.   At the building scale, building information at a high LOD can be captured using Building Information Modeling (BIM) technologies. BIM is “a digital representation of physical and functional characteristics of a facility” [5], representing detailed building component geometry with embedded attributes, such as dimensions, materials and energy use data, combined with three-dimensional (3D) representations of a facility. In the Architecture, Engineering Construction and Owner-operated (AECO) domain, BIMs are increasingly being used to support building design (e.g., design coordination and conflict detection), construction (e.g., simulation of construction schedules), and operation (e.g., asset management). The increasing use of BIM in the delivery of building projects provides a foundation for the potential use of rich and highly detailed building information in urban design and community planning. In comparison to community-scale design, energy design at the building scale focuses on the whole building design together with its structure, systems and site. All these elements are critical to optimize energy design, and they are required to function together to meet both occupants and sustainability requirements. Therefore, BIM is a potential technology that can be integrated in 3  energy design processes as it can provide high LOD information about building systems and components.  The technologies at the community and building scales each provide important characteristics for community energy planning. GIS provides macro-level descriptions and exact geographic coordinates of an object, such as cities, land, and building landscapes; whereas BIM focuses on the micro level representation of building components, such as walls, windows, and mechanical, electrical and plumbing (MEP) systems. Given their relative strengths, integrating GIS and BIM data can be very beneficial to support community energy design. For example, when a municipality aims to evaluate the energy performance of a community with rudimentary building data from GIS, it is very challenging to calculate an accurate result. Thus, using BIM as a platform to provide detailed building information can offer a better estimation of building performance through a more detailed energy simulation. Therefore, the integration of community-scale data and building-scale data for urban energy modeling purposes can be achieved through the integration of GIS and BIM data. However, current technologies lack the ability to integrate multidimensional contexts, i.e., building components and systems, individual buildings, and the entire neighborhood [6]. More specifically, current technologies and approaches do not support interoperability between GIS models and BIM, which is critical for accurate and reliable urban energy modeling.  Several research and industry efforts have attempted to integrate GIS and BIM models or translate models [7-9], specifically based on Industry Foundation Class (IFC), the major data exchange format in the AECO industry. Most of these attempts focus on the translation between GIS and BIM data from one data format to another to support spatial visualization. This means that they focused on creating correspondences between GIS and BIM models based on their schemas. As GIS and BIM are from two different domains and their LODs are different, some syntax of the models may never be successfully translated [10]. For instance, some representations in BIM do not have corresponding representations in GIS, which leads to information loss when converting one type of BIM format to another GIS data format. Such an approach also requires a comprehensive understanding of each data structure and identification of similarities and differences to create correspondences among those data formats. Hence, it is practically impossible to represent all of them in one model in either GIS or BIM platforms. 4  Therefore, an approach for addressing the semantic heterogeneities between GIS and BIM is needed to achieve better data integration, and such an approach should focus on the meaning and the logic that the data represents rather than how the data is structured.   Another limitation of existing approaches for GIS and BIM integration is that they do not consider information needs during the integration. Since both GIS and BIM can contain rich data and only part of this data might be used for energy design, it is critical to understand user’s information requirements first and then create project-related solutions rather than integrating all the data together to produce large and complex hybrid models. For example, there are over 800 classes defined in the IFC standard, and it is practically impossible to integrate all of them with another GIS model that also contains rich data contents. Therefore, project-related solutions to the integration of GIS and BIM are needed, and such solutions should allow queries on the integrated data to satisfy the user’s information requirements, create energy scenarios, and reduce unnecessary integration work.   This research investigates the feasibility of using ontologies to achieve semantic integration between GIS and BIM data to support the domain of community energy design. An ontological approach is used because ontologies are able to represent the logic structure of the data and are particularly useful for representing domain-specific knowledge. This research involves three different domains: GIS, BIM and energy design. The proposed solution needs to convey meaning among these domains and make the integrated information understandable to the community of designers who often lack knowledge about GIS and BIM. An advantage of the ontology is that it enables “analysis and reuse of the domain knowledge, and share of common understanding of the information structure among people and software” [11]. For example, with an ontology, we can create conceptual representations of GIS and BIM data so that designers do not have to study and examine the content and structure of the original datasets. Furthermore, such ontologies can be shared with another group of people who want to perform similar tasks. They can easily reuse the ontology or extend it to avoid duplicate work.    5  1.2 Research Objectives  The ultimate goal of this research is to develop an approach to integrate GIS and BIM data that supports energy design at the community scale. This is based on the observed data challenges in urban energy modeling and the strengths of GIS and BIM technologies. Moreover, considering the limitations of current studies on integrating GIS and BIM, the research has three specific objectives: 1. Demonstrate the usage of both community-scale and building-scale data on energy modeling at the community scale. 2. Develop an ontology-based approach to integrate GIS and BIM data that represents the logic structure of the data and addresses the semantic heterogeneities between the data. 3. Develop a query-based process to extract information from the integrated GIS-BIM data to satisfy user data requirements for energy analysis.  Regarding the first objective, we develop relevant design scenarios for energy consumption of buildings in the University of British Columbia (UBC) campus. We use these scenarios in a case study to illustrate how to leverage the community-scale (GIS) and the building-scale (BIM) data to benefit urban energy modeling. To satisfy the second objective, an ontological approach for data integration is proposed as a potential solution to the domain problems of data inoperability between GIS and BIM. It includes analyzing the required information for the energy simulation and creating a suitable ontology to integrate GIS and BIM data. This ontology involves the knowledge of our case study as well as GIS and BIM, which allows sharing the semantics among the GIS, BIM and the energy design domains and further improving the interoperability between GIS and BIM by representing their data logics. Lastly, we conduct relevant queries on the integrated data to provide the required information for the energy simulation of a UBC campus neighborhood with high-level building information.   The main contributions of the research are: 1. Through a case study, this research demonstrated how the integration of GIS and BIM can facilitate building energy design at the community scale.  6  2. This research developed an ontological approach to achieve semantic integration between GIS and BIM data in the domain of community energy design. 3. The research developed queries on integrated GIS-BIM data to extract designers’ data requirements for urban energy modeling purpose.  1.3 Research Methodology  This research is a proof-of-concept (POC) study that demonstrates the feasibility of proposed concepts or theories through a pilot case [12]. The idea of POC has been widely used in engineering prototype design and software development [13-15]. Although there might be distinct processes to realize POC in different industries, a proof-of-concept project generally includes the following five major tasks:  1. Describe problems to provide the base of the theory or prototype. 2. Explore solutions to the problems, including reviewing existing studies and proposing a new solution. 3. Develop details about the solution. 4. Develop a pilot study to test the solution, including the scenarios and assumptions. 5. Perform an implementation based on the pilot study (usually incomplete) and report the results. According to the content of these tasks, we designed five research steps for this research. An overview of these steps is shown in Figure 1.1 and descriptions of each step are described next.7  Figure 1.1   An overview of research steps Investigate urban energy modeling •Understand current energy modeling approaches •Indentify challenges and data requirements •Identify the need of technology integration  Review current technologies  •Study GIS and BIM in urban energy modeling •Explore GIS and BIM integration •Porpose a solution to GIS and BIM integration for energy design Review PICS project •Understand the PICS methodology •Identify the connection to PICS Develop an approach for GIS and BIM integration •Examine Urban Modeling Interface •Design an approach for GIS and BIM integration to support energy modeling in the UMI platform Implement the approach •Develop a case study and collect data •Implement the approach for data integration •Visualize results Semantic Integration of BIM-GIS data  8  1. Investigate urban energy modeling In this step, the problems of current urban energy modeling and relative background were identified, which was summarized in Chapter 1.1.  2. Review current technologies This step is to review the technologies that can help to address the domain problems, and explore potential solutions. The review including urban energy modeling approaches, GIS, BIM and the integration between them. 3. Review the PICS project The research presented in this dissertation is part of another broader Pacific Institute for Climate Solution (PICS) funded project, which is investigating alternative methods to simulate the energy and emissions in British Columbia’s (BC) built environments. We reviewed the PICS methodology and technologies to inform our research approach for integrating BIM and GIS data. 4. Develop an approach for GIS and BIM integration Based on the PICS’s energy modeling methodology, we investigated the Urban Modeling Interface (UMI) approach as the testbed for our proposed approach for integrating GIS and BIM in the energy modeling process. In this step, a review of the UMI mechanism was conducted to develop the ontology-based approach for integrating GIS and BIM data as a potential solution to the research problem identified.  5. Implement the approach  A case study of the University of British Columbia (UBC) campus was developed to demonstrate the workability of the integration approach, and to perform a proof of concept implementation of the proposed integration approach. The implementation includes integrating GIS and BIM data and visualizing the results of the energy simulation.   The PICS project plays an important role in the design of the research presented in this dissertation. As the research is one piece of puzzle in the whole PICS project, the scope of the research is also bounded by the action plan of the PICS project. The aim of the PICS project is to investigate the simulation of the energy and emissions impact of policy, technology and behavior options in built environments, and our research is particularly focused on the technologies. More specifically, the research explores the technologies for GIS and BIM data integration and the simulation results visualization to support the PICS energy simulations. Since the PICS project 9  uses a specific modeling tool, UMI, our exploration is also based on this tool rather than a generic energy modeling approach.   1.4 Overview of the Dissertation  The thesis is structured as follows: Chapter 2 introduces the research background, including a literature review on urban energy modeling, an overview of GIS and BIM for energy design, GIS and BIM integration, and semantic data integration. Chapter 3 is the research methodology conducted for this research. It first presents the research steps and then a background of the PICS project and its connection to this research. Chapter 4 describes the proposed integration approach, its implementations and the results. It includes an examination of the UMI platform, the design of the integration approach, the development of the UBC case study, and the detailed implementation together with the results. Finally, there is a conclusion section summarizing the major outcomes and findings of the research.   10   Chapter 2: Related Background  2.1 Urban Energy Modeling  To accommodate increasing urban growth and energy consumption, communities should be designed to be as energy-efficient as possible. Energy efficiency measures the difference in how much energy is used to provide the same level of performance by the same type of buildings [16]. Urban energy modeling and spatial modeling are effective tools to examine and predict such measurements of building energy. Regarding the basic approaches for urban energy modeling, there are two kinds of models: Top-down and bottom-up [17]. The top-down model typically requires highly aggregated data and little technological details. It is often used for examining long-term energy use trends, usually beyond 10 years. In contrast, the bottom-up model is produced based on individual building data, which is more suitable for energy modeling at the community scale [18]. The bottom-up model often adopts statistical methods to process individual building data to cope with the variety of data availability. Perez and Robinson [19] identified difficulties in the data collection process for bottom-up models and argued that there is a separation between desired models and their resource requirements.   There are many studies on developing bottom-up models to simulate urban energy costs and demands. Jones et al. [3] developed an Energy and Environmental Prediction (EEP) model to quantify the energy use of the built environment to support urban designers in planning energy strategies. This research conducted cluster analysis to group the buildings with similar energy performance and emissions together. There are five types of information identified as key factors that influence the energy performance. The details are listed in Table 2.1. These data are either related to the geo-location, e.g., building address, or the building design, e.g. building dimensions, which means that the modeling process required both geographic data and detailed building data. This research described the required data level and highlighted the data needs on the building stock, which is useful in assessing how to incorporate such data technologies at the building level, including technologies like BIM.     11  Table 2.1   Key factors of building energy performance [3] Information type Description Location Geographic coordinates or detailed address Building dimensions Heated ground floor area, exposed end area, storey area, façade area Age Building built age Built form Number of storeys, window area, storey height, window-to-wall ratio Assumptions  Assumptions made for data unavailable, examples including thermal properties, number of rooms and floors  Reinhart and Davila proposed an implementation workflow for Urban Building Energy Models (UBEM) by merging detailed individual Building Energy Models (BEM) and regional level building models [20]. The UBEM required data including climate information, building geometry, construction standard and usage schedules. UBEM produced good results on predicting operational building energy use across neighborhoods [21]. However, the required modeling work for collecting data at the individual building level was extensive, and the largest uncertainty was from insufficient details about buildings, such as thermal properties. Such limitations denote the needs for integrating technologies at the building scale in the urban energy modeling.  Taylor et al. [22] proposed a new spatial approach that enables modeling non-domestic urban building energy.  The study tried to address the issue on determining suitable modeling unit at the urban scale using both geometric and non-geometric information about individual buildings. The information requirements are described in Table 2.2. However, there is a significant issue of data incomparability that the linkage between datasets is usually non-existent, and it requires huge efforts on linking the data sets manually. This indicates the need of improving the interoperability between the different datasets in the urban energy modeling process.      12  Table 2.2   Summary of data requirements for SCU [22] Information type Description Data Source Built structure Wall, roof and ground areas, orientation, heights Digital maps and LiDAR data Business taxation Taxable area, building functionality, breakdown space use UK Valuation Office Agency records  Through examining the information demands of these urban energy models, we found that all of them require both urban-level data and detailed building design data. Although part of the information can be obtained from traditional GIS databases, such as building locations and shapes, many data are related to the detailed building design, e.g., window-to-wall ratio, which are often inaccessible and require additional surveys. Furthermore, even though all the data sets may be obtained with extensive effort, the disparate data sources and their incompatibility are still big challenges in the implementation of these urban energy models.    In conclusion, the availability and interoperability of urban data and detailed building data is a significant issue in the energy modeling process. Technologies, such as GIS and BIM, to collect, manage and analyze data needs to be further explored.   2.2 Overview of GIS and BIM  Geographic Information System (GIS) is a system that represents and analyzes real-world objects and their relationships [23]. The strengths of GIS are its highly scalable visualization and analytic power. Traditional GIS can represent geographical data from the global to the building scale, and it can conduct extensive spatial analysis, such as finding the correlation between locations and characteristics. At the community scale, similar energy strategies are usually based on spatial-adjacent buildings (buildings that are adjacent or close together), because they have many common natural and human conditions, such as climate conditions and development policies. In this context, GIS has matured in supporting planners to manage, analyze and visualize data. There are many industrial examples that demonstrate the effectiveness of GIS in energy planning. Fabbri et al. [24] completed a study to analyze the distribution of energy performance for a historical center district in Ferrara, Italy, and demonstrated that GIS offers a 13  sufficient data model to evaluate energy indicators and characteristics related to a district, town, or city. Ramachandra and Shruthi [25] used GIS to integrate different layers of information (climatic zones, wind velocities, etc.) to identify wind energy potentials in the city. The research also confirmed that GIS is well suited for managing and analyzing multidisciplinary spatial data in the decision-making process.  After years of practical use in different industries, GIS data have been generated across diverse data formats with heterogeneous syntax and structure. Some examples of popular data formats are Geodatabase (GDB), Shapefile, TIGER, and GEOTIFF, and all these data formats are developed for different GIS platforms, which causes many complexities and difficulties in data exchange. To improve data interoperability, Open Geospatial Consortium (OGC) has developed Geography Markup Language (GML) that can express geographical features. GML serves as an open interchange format for geographic information exchange on the internet [26]. Recently, there is a growing interest in using 3D city models in GIS software for urban facility management and urban planning. An extension to GML, CityGML, has been established to fulfill this kind of demand to represent a city’s geographic features in three-dimensions. Currently, the coupling of urban-scale energy modeling and 3D city models has attracted considerable attention. Bahu et al. [27] connected a 3D city model with urban energy system models to conduct an analysis for heat energy demand and solar potential. With enhanced 3D analysis, this study showed significant improvements of modeling energy systems in the city’s built environment. Another study using CityGML to calculate district heating demand was completed by Nouvel et al. [28]. The authors evaluated the simulation results with 3D city models, finding that the simulation results are closer to real demand, and the higher the LOD of the model, the more accurate the results obtained. Since CityGML also can model 3D buildings, many studies on GIS and BIM integration rely on integrating CityGML and IFC, which are described in detail in Section 2.3.   2D and 3D GIS have become dominant tools in large-scale energy modeling because of the powerful analysis and visualization, whereas the unavailability of individual building scale data is a common issue in most situations according to our review on different urban energy modeling approaches. Furthermore, GIS is very limited at the building scale because many building 14  components do not have detailed representations in traditional GIS platforms (e.g., walls). Accurate data collection and management are mature at the city scale through ground surveys or aerial photography by governments and large organizations while they are haphazard at the building scale. For example, many buildings in the city still only have paper-based design documentation and such data are very difficult to use for energy modeling. Recently, there is a growing interest in using Building Information Modeling (BIM) to address such issues. Unlike GIS, the scope of BIM is usually at the individual building scale. The National BIM Standard-United States (NBIMS-US) defines Building Information Modeling (BIM) as “a digital representation of physical and functional characteristics of a facility” [5]. BIM is able to capture a high LOD of the information about building objects, such as geometries and dimensions, with a three-dimensional representation. Among various BIM standards, Industry Foundation Class (IFC), developed by buildingSMART, is the major data exchange format to facilitate building data interoperability in the AECO industry. IFC is platform neutral, and it conveys all information needed to construct a facility, such as dimensions, materials and costs [29]. In comparison to community-scale design, energy design at the building scale focuses on the whole building design together with its structure, systems and site. All these elements are critical to optimize energy design, and they are required to function together to meet both occupants’ and sustainability requirements. Therefore, BIM is an advantageous technology in such processes as it can provide high LOD information about a building object through its lifecycle.   In recent years, designers have shown great interest in utilizing BIM for low-energy and sustainable building design. Lee et al. [30] investigated the impact of different architectural components on building energy performance using a BIM-based energy modeling system, aiming at minimizing changes during the design process. The authors found that through BIM-based analysis, architects can infer the energy performance of many alternative designs. Martino Di Giuda et al. [31] conducted a project on applying BIM technology to assist the energy efficient retrofitting in a school building. The project proposed a new working philosophy of using BIM at different work stages, which leads to optimized building maintenance or upgrading strategies. Although BIM has proved to be successful in gaining accurate energy simulation results, the computing cost is a big challenge when it is applied to the urban scale, especially 15  using high LOD data. Thus, it is crucial to establish a balance between data LOD and expected accuracy of the results.   2.3 GIS and BIM Integration  Despite GIS and BIM having different operational scopes, they both describe real-world objects. GIS provides macro-level descriptions and exact geographic coordinates of an object, such as cities, land, and building landscape; whereas BIM focuses on micro-level of building components, like walls, windows and MEP systems [32]. Integrating information from different sources and different LOD is advantageous in the urban energy design process. The integration of GIS and BIM allows designers to obtain information from both macro and micro level built environment data, and furthermore, enables the strengths of these two technologies to be exploited. Researchers have thus begun to investigate different approaches for combining these data sets to achieve a better energy design process. Niu et al. [8] developed a web-based BIM-GIS visualization system to support both urban planners and building energy designers. The study used Green Building XML (gbXML) as the exchange format to convert Autodesk Revit and Keyhole Markup Language (KML) files into the developed system. However, this study focuses on visualizing a building’s performance data in GIS systems instead of using both GIS and BIM data as the inputs for energy modeling. In Europe, a collaborative research STREAMER [6] is underway to achieve a semantic BIM+GIS approach to resolve challenges in energy-efficient healthcare buildings and the district design. The research only has strategic plans without a detailed approach for the integration. Therefore, neither of them have formalized an approach to integrating data to support energy modeling from a technical perspective.  Outside of the energy domain, there are also many researchers exploring the possibility of integrating GIS and BIM data. Liu and Issa [7] investigated the use of ArcGIS, Revit, and AutoCAD Civil 3D to connect and visualize building MEP (mechanical, electrical and plumbing) systems and outdoor sub-surface pipeline networks. They converted the Revit model and Shapefile GIS data into AutoCAD Civil 3D to visualize the elevation and the position of the pipeline network using both building interior data and building site surface data. Laat et al. [9] described a workflow for converting IFC models to CityGML by implementing a CityGML extension called GeoBIM, aimed at integrating IFC information into GIS platforms. The 16  proposed schema matching results are listed in Table 2.3. The table provides a good basis for the author to identify the useful IFC classes and their corresponding representations in the GIS domain. Similarly, El-Mekawy et al. [33] also did a study on a unidirectional translation between CityGML and IFC. The study proposed a geospatial interoperable model, called Unified Building Model (UBM), and adopted it in a real case. The authors investigated the insights of the IFC schema, identifying thirteen important classes and entities that can represent building and architectural elements in GIS. A translation between IFC and CityGML through UBM is described in Table 2.3. The subsequent study [34] mainly worked on evaluating issues of schema mismatching, heterogeneity of different data sources, missing geometric information and complexity of geometrical information representation, which highlights the difficulty of a one-to-one mapping between the two data formats and the need for an intermediate model for translation. Therefore, the approach proposed in this dissertation adopts the intermediate model approach instead of translating BIM or GIS directly.   Table 2.3   Mapping between IFC and CityGML classes [9] IFC Class CityGML Class IfcBuilding Building BuildingAddress Address IfcSpace Room IfcWindow Window IfcFlowTerminal FlowTerminal IfcFurnishingElement BuildingFurniture IfcDoor Door IfcRoof RoofSurface IfcSlab RoofSurface or FloorSurface (Depending on IfcSlabTypeEnum) IfcColumn Column IfcBeam Beam IfcRailing Railing IfcStair Stair 17  IFC Class CityGML Class IfcWall InteriorWallSurface or WallSurface (Depending on boundaryType)   Table 2.4   IFC-UBM-CItyGML mapping [33] IFC UBM CityGML IfcBuilding UBMBuilding _AbstractBuilding IfcBuildingStorey UBMStorey BoundarySurface             RoofSurface             WallSurface             GroundSurface              Other building elements IfcSpace UBMSpace             UBMOpenedSpace             UBMClosedSpace Room IfcSlab UBMLevel          (Ground Slab) UBMGround GroundSurface         (Floor Slab) UBMFloor FloorSurface         (Ceiling Slab) UBMCovering            UBMCeiling CeilingSurface IfcRoof UBMCovering RoofSurface           UBMRoof  IfcWall UBMWall          (Exterior Wall) UBMExteriorWall WallSurface         (Interior Wall) UBMInteriorWall InteriorWallSurface IfcCurtainWall UBMWall   UBMCurtainWall WallSurface IfcOpeningElement UBMOpening Opening          IfcDoor          IfcDoor             Door          IfcWindow          IfcWindow            Window 18  IFC UBM CityGML IfcBeam UBMBuildingInstallation BuildingInstallation IfcColumn UBMBuildingInstallation BuildingInstallation IfcCovering UBMBuildingInstallation BuildingInstallation IfcStair UBMBuildingInstallation BuildingInstallation IfcRailing UBMBuildingInstallation BuildingInstallation IfcRamp UBMBuildingInstallation BuildingInstallation  The results of these data mappings provide a good basis to represent the correspondences between GIS and BIM features, which can be a reference to further integration and save efforts for schema examinations. However, most of these studies still try to understand the data schema and conduct mapping on geometric classes. Since GIS and BIM are from two different domains, they use different terminologies and data representation approaches to describe the same object, which results in large differences in data schema structure. This also causes complexities and problems in the mapping process. The more dissimilarities between data structure, the harder the model integration is. Therefore, solving heterogeneities of GIS and BIM data structures is not the ultimate way to achieve the integration.  Semantics can convey the meaning of a language and reflect elements using domain knowledge. Therefore, solving semantic heterogeneities between GIS and BIM in the building domain can facilitate information integration. Mignard et al. [35] developed a semantic model extension to BIM called UIM (Urban Information Modeling), which is allowed to populated with information from IFC and CityGML. Karan et al. [36] applied semantic web technology to identify the interoperability between BIM and general GML model. They developed and tested an ontological framework for data exchange between BIM standard and GML and then alleviated the knowledge requirement of the IFC schema in GIS to retrieve data.   Although these studies achieve better data integration using semantic approaches than direct data mapping, none of them is dedicated to a specific domain application. There are over 800 classes defined in the IFC schema, and it is practically impossible to integrate all of them into one model. Furthermore, only part of the information in BIM is useful for a given domain problem. Taking 19  urban energy simulation as an example, in addition to those classes identified for the building element representations (i.e., walls), special attention should also be paid to non-geometric information like building element’s dimensions and thermal properties. Therefore, in our approach, the domain problem and the data requirements for energy modeling are built first to avoid unnecessary integration work.  2.4 Semantic Integration and Ontology-based Approach  To achieve information integration between different data sources, ontology-based approaches are commonly used in solving semantic heterogeneities in structured data to realize seamless connectivity between information sources [11, 37]. Since the research presented in this dissertation aims to achieve GIS and BIM integration based on their data meaning and logic, an ontological approach will be used to achieve GIS and BIM integration for community energy design.   Among different ontology-related technologies, Semantic Web is a Web 3.0 technology developed by World Wide Web Consortium (W3C), and it is a way of linking data between systems or entities that allows for rich, self-describing interrelations of data [38]. It supports the use of ontologies and allows the computer to process and interpret such semantic information over the internet.   2.4.1 Overview of Semantic Web  Resource Description Framework (RDF) and Ontology Web Language (OWL) are two important components of Semantic Web to represent ontologies. OWL is a Semantic Web language designed to represent the knowledge about things and the relations between things, and RDF is a standard model for exchanging such data on the internet [39]. OWL supports customized rule descriptions that allow machines to conduct logic reasoning, for instance, finding all individual elements that are instances of a concept, or checking whether an instance follows the defined rule. W3C also developed SPARQL protocol for querying RDF models, which provides a good tool for this research to query the integrated GIS and BIM data.   20  RDF consists of a set of statements called triples comprising a subject, a predicate (property) and an object. A RDF model can be visualized in graph using nodes and arcs, which represents triples (subject-predicate-object) and their relationships (Figure 2.1). RDF expresses information about resources, i.e., subject, predicate and object in RDF models. Anything in the world can be a resource (e.g. concepts, physical things, texts and numbers). These resources are denoted using literal or Uniform Resource Identifier (URI), a string of characters providing the means to identify a resource. Literals are plain values with data types, such as numbers and dates. URIs often begin with a common substring called namespace to avoid name conflicts among different RDF models. As a namespace is often long and repeated frequently in the model, a namespace prefix is often defined to represent the full text of the namespace. For example, a namespace URI ‘http://www.w3.org/1999/02/22-rdf-syntax-ns#’ can be denoted with prefix ‘rdf’, so all the URIs of resources under this namespace can begin with ‘rdf:’ instead of the full string.    In OWL language, there are three main components: class, individual and property. A class is a concept that describes a group of things. An individual is an instance of a class. A property is the relationship between individuals. A Property can either be an object property that links two individuals, or a data type property that links an individual to a data value. An example of representing OWL ontology using RDF model is illustrated in Figure 2.2. The graph shows two triples: ‘Person-hasName-Name’ and ‘PersonA-isA-Person’. Here ‘Person’ and ‘Name’ are classes, ‘hasName’ and ‘isA’ are properties, and ‘PersonA’ is an instance of ‘Person’.   Person Name hasName isA PersonA  Figure 2.2   An example of OWL Subject Object Predicate Figure 2.1   A sample RDF graph 21  2.4.2 Applications of Semantic Web  There have been many applications of Semantic Web technology in both GIS or BIM domains. Malgundkar et al. [40] created an ontology for urban traffic analysis and unified different sources of traffic data into a GIS-based system. Rich semantics were added, resulting in a formal expression of traffic knowledge and quick data retrieval through ontology-based reasoning and inference rules. Pauwels et al. [41] did an investigation on using the Semantic Web technology to resolve 3-D information conversion issues, demonstrating the value of this technology to improve the interoperability among different 3-D information representations. Niknam and Karshenas [42] showed the potential of using the ontology and Semantic Web to combine information from various sources, including BIM, estimating knowledge, and construction material cost data for construction cost estimation. Costa and Madrazo [43] connected BIM with a data catalog of precast concrete components to improve the assembly and dimensioning process in construction.   Many resources about Semantic Web have been developed by some GIS and BIM organizations. Open Geospatial Consortium (OGC) has developed the GeoSPARQL standard to allow representing and querying geospatial data using Semantic Web data [44]. Using this standard, the geometric information can be expressed in a RDF model and connected with other ontologies. In the BIM world, IFC provides rich building semantics. buildingSMART [45] has published an ifcOWL ontology for the IFC standard, making it available for Semantic Web usage. IfcOWL is a direct translation of the IFC EXPRESS schema to OWL.   These studies demonstrate the effectiveness of Semantic Web in integrating different sources of GIS or BIM data, and the findings of existing studies and available resources, indicate the possibility of integrating GIS and BIM together.   Through reviewing how these studies applied Semantic Web, a generic workflow to deploy Semantic Web was developed. Figure 2.3 shows an overview of the process. It includes three main tasks:  1. Develop ontology 22  2. Transform data 3. Link data The first task includes analyzing data, capturing their characteristics and then developing an ontology. In this task, a schema of the data is obtained, and domain concepts and data relationships are identified. Based on project-related information, users often re-name or build new classes and properties to create a formal ontology. The second task is to transform data into a RDF format. At this stage, the original data are converted to a set of RDF triples based on the developed ontology. The third task is to link different data sets. The goal of this task is to identify the similar data classes and relationships among different data sources and then create links between datasets, so that the information can be shared among datasets.   Although there are some tools that can help to match and align schemas or ontologies in the linking data process, manual work is still necessary to verify the automatic mapping results and to provide additional domain knowledge to identify data relationships. Once data are linked together, they are available to be stored in a single RDF model file or any other databases, and users can access to these data through different application platforms. Users can perform data queries with SPARQL standards, and the results are in line with the original data. It usually Figure 2.3   A generic workflow for Semantic Web deployment 23  requires a large amount of work to develop an ontology that covers all the concepts in the original data source; therefore, it is significant to define data requirements and only capture useful concepts.   This workflow provides a basis for the implementation of Semantic Web on GIS and BIM integration in our research. The three main tasks are indispensable for the application of Semantic Web; therefore, there are corresponding steps or sub-steps to these task in the designed approach for the integration in this research.    24   Chapter 3: Overview of Research Methodology  This research was designed using proof-of-concept and scenario-based methodology[13-15], which demonstrates the feasibility of proposed concepts or theories through a pilot case [12]. The idea of POC has been widely used in engineering prototype design and software development [13-15]. Although there might be distinct processes to realize POC in different industries, a proof-of-concept project generally includes the following five major tasks:   1. Describe research problems to provide the base of the theory 2. Explore the solutions to the problems, including reviewing existing studies and propose a new solution 3. Develop details about the solution 4. Develop a pilot study to test the solution, including the scenarios and assumptions 5. Perform an implementation based on the pilot study (usually incomplete) and report the results  According to the content of these tasks, we designed five research steps to conduct the study on integrating GIS and BIM for community energy design. Figure 3.1 shows an overview of these steps, which are then described in detail. 25   Figure 3.1   An overview of research steps Investigate urban energy modeling •Understand current energy modeling approaches •Indentify challenges and data requirements •Identify the need of technology integration  Review current technologies  •Study GIS and BIM in urban energy modeling •Explore GIS and BIM integration •Porpose a solution to GIS and BIM integration for energy design Review the PICS project •Understand the PICS methodology •Identify the connection to PICS Develop an approach for GIS and BIM integration •Examine Urban Modeling Interface •Design an approach for GIS and BIM integration to support energy modeling in the UMI platform Implement the approach •Develop a case study and collect data •Implement the approach for data integration •Visualize results 26  1. Investigate urban energy modeling Current urban energy modeling approaches at the community scale were reviewed first. Their main problem is the usage of low-level and abstract building data, which causes large uncertainties in the final modeling results. Therefore, the need of technology integration (GIS and BIM) was identified.  2. Review current technologies At this step, current technologies were reviewed to provide a theoretical basis of the solution, including urban energy modeling related techniques, their data requirements, and GIS and BIM for energy modeling. Through examining existing research about data integration, Semantic Web was recognized as the technology to conduct the integration in this research. A review of Semantic Web related project provides a fundamental for the design of implementations.   3. Review the PICS project This study is relied on a broader community-scale research about investigating the simulation of the energy and emissions in different built environments, which is a Pacific Institutional Climate Solutions (PICS) funded project. Therefore, there needs an understanding of the PICS project methodology. Furthermore, the role of this research in the PICS project as well as the connection between the two studies needs to be clearly identified. 4. Develop an approach for GIS and BIM integration In this step, the mechanism of the UMI platform was examined to inform the specific domain problem and data requirements for the integration. After that, a potential data integration approach was produced as the solution to the research problems.  5. Implement the approach  A case study of the University of British Columbia (UBC) campus was developed to demonstrate the workability of the approach, and a proof-of-concept implementation was performed. The implementation includes the integration of GIS and BIM data and visualization of the results.   In the next sections of this chapter, some related background about the PICS project together with our research scope is described in detail.   27  3.1 Background on the PICS Research Project  This study is part of a broader research project funded by PICS, which is a two-year community-scale research project on increasing energy efficiency in British Columbia’s (BC) built environment. The overall objective of the PICS project is to investigate the simulation of the  energy and emissions impact of policy, technology and behavior options in different built environments. The PICS project action plan has a design of four sections, shown in Figure 3.2.    During the early investigation stage of the PICS project, the PICS team tested and identified a promising energy modeling platform, UMI, as the main energy modeling tool. UMI uses a neighborhood model, often a GIS model, and a standardized Template Library File (TLF) to perform energy simulation. A neighborhood model stores individual building massing information (e.g. size and volume). This model consists of a geometric layer that represents spatial relationships among buildings in the neighborhood, such as façade and shading effects. A TLF contains building templates which stores non-geometric data for energy simulation purpose, such as building materials, heating schedules, thermal loads and spaces. Details about the UMI platform is described in more detail in Section  4.1. Policies and scenarios •Inventory built environment related policies •Develop a framework for generating policy scenarios Geo-spatial modeling •Model built environment for six BC communities •Develop archetypes representitive of communities Energy modeling •Model energy and emission impact in archetypes •Visualize performance Recommend-ed policy  •  Review scenarios with stakeholders •  Recommended policy change Figure 3.2   The PICS project action plan 28   To incorporate UMI into the PICS action plan, the PICS research team developed a workflow that demonstrates the iterative energy design process at the community level. Figure 3.3 shows the main processes, demonstrating a full design cycle. The total work of one cycle includes four processes, namely:  1. Data integration 2. Energy simulation 3. Visualization 4. Design scenario review   In the PICS energy design workflow, design scenarios are linked with original IFC and Geodatabase data and serve as important inputs for the energy simulation in the UMI platform. These scenarios are simulated, visualized, evaluated and modified until designers are satisfied with the results. IFC and ESRI Geodatabase are the data formats for BIM and GIS model respectively. IFC provides the information about individual buildings, and Geodatabase provides neighborhood information and serves as the neighborhood model in UMI. Designers can integrate these data and obtain a TLF through the data integration process. Then an energy Figure 3.3   An overview of the PICS energy design workflow 29  simulation can be performed using this TLF and the neighborhood model. After that, designers can visualize the results on maps, review the results, propose new designs and evaluate them again.  Regarding the major processes in this design workflow, data integration involves integrating GIS and BIM model and extracting building properties to create a TLF for UMI. This process adopts Web Semantic technology introduced in Section 2.4, and is implemented in the Feature Manipulation Engine (FME) developed by Safe Software [46]. Energy simulation is conducted in the UMI environment, and then operational energy performance is obtained. For data visualization, the results are exported from UMI to the ArcGIS software for further energy mapping and visualization. Then designers and other decision makers can evaluate the energy performance, propose changes, modify the design, and conduct energy modeling again until a satisfaction is achieved. More details about the data integration in the workflow are given in Chapter 4.   The investigation of the PICS project on energy simulation are from three perspectives: policies, technologies and behaviors. The research presented in this dissertation is focused on the technologies, particular related to data for community energy design. Our role in the PICS project is to investigate the technologies for combining both community-scale and building-scale data to support the exploration of the energy impact on the built environment. The scope of our research in the PICS design workflow is also shown in Figure 4.1. The tasks are to develop approach for GIS and BIM data integration process and perform the simulation results visualization. We conduct a study on integrating GIS and BIM data, and provide the PICS team with the required parameters for the energy simulation. Other members in the PICS project help to run the energy simulation and provide the results, and following that, we perform the results visualization.   This study concentrates on GIS and BIM integration for community-scale energy modeling by leveraging the existing PICS project design and the UMI modeling platform. As discussed earlier, the data integration requires an understanding of project-related data requirements. In this case, the integrated GIS and BIM data are used in the UMI platform. Therefore, the exploration of the 30  data integration is also based on the UMI platform. In order to develop the integration approach, it is necessary to examine the UMI working mechanism and its data requirements for the simulation.    Up to this point, Chapter 1-3 has introduced the domain problems, the review of technologies and the PICS project, which are the first three research steps. In next chapter, Steps 4 and 5, the development and implementation of the data integration approach will be described.      31   Chapter 4: Approach, Implementations and Results  Based on the research steps from Figure 3.1 in Section 3.1, the steps related to developing and implementing the approach are extracted and graphically represented in Figure 4.1. Corresponding to these steps, this chapter introduces the UMI platform, the design approach, the case study, the implementation of the approach and the results visualization.  Figure 4.1   Research steps for developing and implementing the approach   4.1 UMI: An Environment for Urban Energy Modeling  UMI is a Rhino-based energy modeling environment for designers and planners interested in neighborhood and city energy performance, developed by the Sustainable Design Lab at the Massachusetts Institute of Technology [28]. The use of UMI has been demonstrated in a recent study in the City of Boston [47], presenting a complete workflow of applying UMI to model the energy consumption of the whole city. UMI uses a neighborhood model, often a GIS model, and a standardized Template Library File (TLF) to perform energy simulation. A neighborhood model stores individual building massing information (e.g. size and volume). This model consists of a geometric layer that represents spatial relationships among buildings in the neighborhood, such as façade and shading effects. A TLF is an eXtensible Markup Language (XML) format and contains building templates which stores non-geometric data for energy Implement the approach 3. Develop a case study and collect data 4. Implement the approach for data integration 5. Visualize results Develop an approach for GIS-BIM integration 1. Examine Urban Modeling Interface 2. Design an approach for GIS and BIM integration to support energy modeling in the UMI platform 32  simulation purpose, such as building materials, heating schedules, thermal loads and spaces. On the urban scale, each building template usually represents one type of buildings, e.g., commercial and residential, and a TLF can contain multiple building templates for different building types. With a specified neighborhood model and a TLF, UMI can perform an energy simulation on the buildings included in the neighborhood model.   4.1.1 UMI Workflow  Figure 4.2 describes four steps to perform urban energy modeling in the UMI environment. The first step is to create a neighborhood model. Users can choose to create a model in the UMI environment or import a model from an external file. The next step is to configure a TLF, which contains major input parameters of the simulation. The details about the TLF are explained in Section next section. In this step, each building in the model should be assigned to a particular building template in the TLF. There are also some additional properties that need to be set outside of the TLF, including the window-to-wall ratio and the floor height of the building. Once all the inputs are ready, users are able to run the simulation. The simulation results can either be visualized directly in UMI and their cloud environment or be exported to other platforms.  4.1.2 TLF: Standardized Building Property Template  A Template Library File (TLF) contains most of the required parameters for energy simulation in UMI. That is to say, it defines the information requirements for this simulation. Since this research uses UMI as the modeling platform, understanding TLF contents and structure is a necessary process to understand the project-related data requirements, which is a prerequisite of GIS and BIM integration in our research.   The main contents of a TLF are summarized in Table 4.1. There are four main categories: general data, constructions, thermal loads and conditionings. A tree structure is used to Create neighborhood model Configure TLF Run simulation Visualize result Figure 4.2   A workflow of UMI  33  systematize these contents in an XML file. The data contents are further categorized into seventeen classes, and each class also contains many simple data properties or nested subclasses.  Table 4.1   Data contents of the TLF [48]  Figure 4.3 shows the overall data classes (1), and an example of nested properties of opaque construction (2,3) in a TLF. This example shows that there are twelve types of opaque materials in the file, and each opaque material class contains many detailed properties (e.g., id, conductivity, density and density).   Category Attributes General data Use, code, description Constructions Exterior wall, roof, ground floor, internal floor, external floor, basement wall, glazing and window to wall ratio, glazing and window to wall ratio, partition, thermal mass type and ratio Thermal loads Occupation density and schedule, equipment density and schedule, lighting density and schedule, infiltration rate Conditioning systems Heating and cooling set points and schedules, mechanical ventilation rates and schedules, natural ventilation rates and schedules. 34    To better understand the simulation parameters and data requirements, we analyzed a sample TLF installed with UMI and generated a class diagram (Figure 4.4) to represent the main data classes and their relationships. Construction, Material, and Schedule are virtual classes here, only for better organizing the data view. To give an example of this diagram, the BuildingTemplate class is associated with the Zone class that represents its core thermal zone information. The Zone class further relates to the ZoneConstructionSet class, linking to its construction component details. The ZoneConstructionSet is aggregated with multiple subclasses of the Construction class, which relates to its façade, roof, partition, ground floor and internal floor information. During the investigation, it was discovered that the BuildingTemplate class is the connection between the geometric layer and the TLF. A TLF can contain multiple instances of the BuildingTemplate class, and each building in the model needs to be assigned to an instance of the BuildingTemplate. Moreover, depending on the desired modeling scale unit, a BuildingTemplate instance can be applied to part of a building (i.e., a room and a storey), the whole building or multiple buildings that belong to the same building type. This shows the flexibility of the modeling scale in the UMI platform. The developers of UMI provides an Figure 4.3   The TLF structure 12 335  independent Template Library Editor allowing to create and manage TLF data, which means that the TLF not only can be used for UMI but also other energy modeling tools when there is an interface between the TLF and software environment. However, even with a graphic-based tool, the user still needs to configure at least several hundred attributes based on a building’s design information. Such work requires large manual efforts, and it is difficult to collect the related data, which are the two biggest challenges of using UMI. These challenges indicate the potential power of integrating BIM into the modeling process to provide detailed building design data for the simulation.     Through a thorough exploration of UMI, the author found that it is an energy modeling tool that makes use of both community-scale data (e.g., neighborhood model) and individual building-scale data (e.g., materials) as input parameters for energy simulation, which makes UMI an advantageous platform to investigate GIS and BIM data integration. Moreover, the challenges we identified in the UMI environment, the data limitations and additional manual work, are typical in other reviewed urban energy models introduced in Chapter 2.1. Therefore, we believe that if Figure 4.4   The class diagram of the TLF structure 36  our proposed approach can solve the challenges of UMI, it is also applicable to other urban modeling methods.   4.2 Design of the Data Integration Approach  Following the design workflow of the PICS project (Figure 3.3 on Page 27) and the UMI mechanism, we propose an approach for the data integration process. This approach includes three main steps (Figure 4.5): 1. Ontology development  2. Data extraction and transformation 3. Data query and loading   This approach is developed based on the generic workflow for Semantic Web deployment, introduced in Chapter 2.4.2. First, ontologies for IFC data and Geodatabase data for the energy design purpose are created. They are then linked based on their similar semantics, e.g., buildings. The ontology-related work is done in Protégé, an open-source ontology software developed by Stanford University [49]. Based on this linked ontology, related data are extracted from the Geodatabase and the IFC model and then are transformed into RDF triples. The last step was to conduct semantic queries over RDF model to retrieve required properties using SPARQL protocol. The results are organized in a XML file following the TLF structure. The details about these three steps are given in Chapter 4.4. Figure 4.5   An overview of data integration approach 37   In order to test the feasibility of the proposed approach, this research is guided by a conceptual case study of the campus community in the University of British Columbia (UBC). Two design scenarios were developed for the implementations of the approach. The scenarios are representative of two spatial scales, which are the community scale and the building scale respectively.    4.3 Case Study: the UBC Vancouver Campus  4.3.1 Data: Geodatabase and IFC  The author collected campus data from the UBC Campus and Community Planning department, which maintains a Geodatabase that stores all GIS-related data about the campus. UBC Building Operations divides the campus into eight service zones (Orange, Green, Red, Teal, Yellow, Brown, Blue and Grey) and organizes the campus in such way for service management. Figure 4.6 is a map derived from the Geodatabase data, showing the target neighborhood of this research. The buildings in this area are color coded by their service zone color. A magnified map on the top-right of the picture shows the surroundings of the Engineering Student Center (ESC) building, which is selected as the individual building case in this study. In addition to the campus geometric data (e.g. building footprint, streets and parcels), this Geodatabase maintains non-geometric data including service zone information and simple building profiles, such as the built year. The external appearance design of the ESC is shown in Figure 4.7 using an IFC model. The IFC model contains the architectural and structural design of the building. To demonstrate our proposed approach, this study integrates these two data sources to support urban energy modeling in the selected UBC neighborhood.    38   Figure 4.7   The IFC model of Engineering Student Center (ESC)  Figure 4.6   A map of the target UBC neighborhood 39  4.3.2 Design Scenarios  The design process is iterative and designers usually produce multiple design scenarios. These scenarios are simulated, visualized, evaluated and modified until a satisfying design is achieved. In the PICS energy design workflow (Figure 3.3), design scenarios are linked with the original IFC and Geodatabase data and serve as important inputs for the energy simulation in the UMI platform. Then the energy consumption under different scenarios can be reviewed, and designers can modify the design and conduct another simulation until they are satisfied with the results.   Regarding data integration, design scenarios inform the data that needs to be integrated for energy modeling. However, at the community level, design scenarios are often conceptual and are text-based descriptions. Therefore, in order to integrate them in the energy modeling process, it is important to convert them from text to a format that a computer can process. In this section, the author developed two scenarios to represent energy strategies related to the UBC ventilation policy and the performance of building construction materials. The main tasks for creating scenarios involved: (1) Reviewing the existing energy strategies for the ventilation and building materials on the UBC campus, and (2) Making reasonable assumptions to convert the strategies and integrate them into Geodatabase and IFC datasets for energy modeling.  Scenario 1: Ventilation Policy at the UBC Community scale  Ventilation is the process to replace indoor air with outdoor air to provide higher indoor air quality. Ventilation policies affect both the cooling and heating performance of the building. Too much ventilation causes high frequency of indoor and outdoor air exchange, resulting in large energy waste. To maximize the energy efficiency and avoid unnecessary costs, UBC has made official energy policies for buildings on campus. A few examples are shown below in Figure 4.8.   The policy in the documentation is just a guideline for the ventilation settings of buildings, and the implementation of the policy is performed by UBC Building Operations. UBC uses zone-based management and each zone has a manager as the zone leader. Hence, in this case, an assumption was made that each service zone ensures its own implementation of the UBC ventilation policy. This means that the actual ventilation setting is related to service zones rather 40  than individual buildings directly. As a result, the ventilation policy data needs to be converted to a table and integrated into the Geodatabase model that stores the campus-wide data.     To map the texts to data tables, the author first checked the required ventilation-related data structure of TLF and then extracted relevant data from the UBC policy. The identified attributes are listed in Table 4.2. The ventilation policy was divided into two parts: ventilation setting and schedules (year, week and day). A ventilation setting is associated with a yearly operating schedule, and a yearly schedule relates to a weekly schedule. Then the weekly schedule also references daily schedules for seven days in a week. A daily schedule contains hourly system running capacity for 24 hours in a day. The capacity value is between 0 to 1. 0 means the system is completely off, and 1 means that it is fully in operation during this one-hour period. Figure 4.9 shows an example of partial ventilation settings for service zones under the guidance of the UBC ventilation policy.         Figure 4.8   Sample UBC ventilation policy [40] 41  Table 4.2   Ventilation-related Properties Ventilation setting   Year schedule  Week schedule  Day schedule Service zone  FromDay  Day 1   H1  Infiltration  FromMonth  Day 2  H2 Infiltration rate  ToDay  Day 3  … Natural ventilation  ToMonth  Day 4  … Natural vent min outdoor air temp (deg·C)  Week schedule  Day 5  H24 Natural vent max outdoor air temp (deg·C)    Day 6   Natural vent max relative humidity    Day 7   Natural vent schedule       Natural vent zone temp setpoint (deg·C)       Scheduled ventilation       Scheduled ventilation ACH       Scheduled ventilation schedule       Scheduled ventilation setpoint (deg·C)              Figure 4.9   An example of ventilation settings 42  Scenario 2: Construction Materials at the Building Scale  Building materials affect the performance of building systems and are significant to the energy design of a building and affect the long-term energy costs of building maintenance. The UBC Vancouver Campus Plan [50] provides a detailed range of materials for architecture. The plan specifies the material usage of buildings on campus. Figure 4.10 illustrates some examples of design material palettes from the plan.    Regarding the requirements of the TLF, it categorizes materials into two groups: opaque and glazing. The two groups have different attributes, summarized in Table 4.3. Based on the UBC design palette and the required attributes from the TLF, we collected data about building materials and their properties as additional information (Figure 4.11). Although the material requirements are applied to the whole campus, the actual construction details about the material, e.g., material thickness, are related to the individual building design. As a result, the material data should be linked with the IFC model of the building, which is the ESC building model in this case.   Table 4.3   Material attributes in the TLF Opaque material attribute Glazing material attribute Conductivity (W/m·K) Conductivity (W/m·K) Density(Kg/m3) Density (Kg/m3) Roughness Dirt Factor Solar Absorptance Back-side IR Emissivity Figure 4.10   Sample UBC design material palette [44] 43  Opaque material attribute Glazing material attribute Specific Heat(J/Kg·K) Front-side IR Emissivity Thermal Emittance IR Transmittance Visible Absorptance Back-side Solar Reflectance  Front-side Solar Reflectance  Solar Transmittance  Back-side Visible Reflectance  Front-side Visible Reflectance  Visible Transmittance    4.4 Implementation of the Approach  For the implementation of the proposed integration approach, we adopted the Semantic Web technology to integrate data and then conducted semantic queries to retrieve information and load them into a TLF. A software called Feature Manipulation Engine [46] (FME) was exploited to integrate Geodatabase and IFC data. FME can read and write more than 300 data formats, including IFC and Geodatabase. It also contains a set of transformers for data structure and content manipulation, and it finally loads them to another format. With FME, we built a repeatable workflow for Geodatabase and IFC integration and automates this workflow. However, FME currently does not support any Semantic Web data (e.g., RDF), so the author chose to build a plug-in for FME to realize the data transformation to RDF model. The final RDF model was stored in an Oracle database. Regarding the visualization process, we exported the simulation results from UMI to ArcGIS, a suite of GIS software, and linked them with the campus Geodatabase. The rest of this section will describe the implementation details about the Figure 4.11   An example of collected opaque materials 44  specific steps implemented in our approach, which include ontology development, data extraction and transformation, and data query and loading.   4.4.1 Ontology Development  As multiple datasets (Geodatabase and IFC) were used for energy modeling in our research, we developed an integration approach that supports querying across these data instead of merging the data together. There were four steps required for this approach: 1. Create a campus-ventilation ontology 2. Create a building ontology 3. Standardize location representations in the ontologies 4. Link the two ontologies   This section describes how the author developed a project-related ontology that represents the semantics of the design scenarios.  The Campus-Ventilation Ontology  The campus-ventilation ontology was created for Design Scenario #1, which looked at ventilation policy at the UBC community scale. In Section 4.3.2, we described how we mapped ventilation policy data into a few tables (Table 4.2 on Page 40) to represent this knowledge. With additional knowledge about the UBC campus organization and the TLF structure, four major classes were identified: Campus Building, Service Zone, Ventilation Setting and Schedule. These classes and properties (relationships between classes) are described using a RDF graph in Figure 4.12. A building is operated by a service zone, and a service zone has its ventilation setting. A ventilation setting has mechanical and natural ventilation schedules. Although in reality, different rooms in a building may have their own ventilation settings, this research focuses on the community-scale, so we created links from the ventilation setting to the service zone instead of single rooms or buildings. After identifying the main classes, the properties of each class were  detailed. For example, a campus building has BL_ID (building id), short name, full name, gross area and height.   45   The Building Ontology  The building ontology was created for Design Scenario #2, which investigated the semantics of the building materials and components that are useful for energy modeling in UMI. BuildingSMART [45] has developed ifcOWL to represent the IFC schema in OWL format. In the IFC schema, there are two important super classes to describe a building’s components and its spatial contexts. They are IfcBuildingElement and IfcSpatialStructureElement. IfcBuildingElement is the parent class of the major function parts of a building, such as floors, roofs and walls. IfcSpatialStructureElement is the generalized class of building spatial structure components, examples including buildings, building storeys and building space. The IFC schema uses a relationship class called IfcRelAggregates to build the hierarchy among these spatial structure components. Another class, IfcRelContainedInSpatialStructure, is used to connect the instances of IfcSpatialStructureElement to its contained IfcBuildingElement instances. Figure 4.13 gives an example of such relationships. An instance of IfcBuilding (#4) has a property called RelatingObject that references to the IfcRelAggregates (#1). This IfcRelAggregates refers to the IfcBuildingStorey (#6) through RelatedObject propoerty. This means that the building storey (#6) belongs to the building (#4). Similarly, the building storey (#6) contains the wall (#11), and this relation is stored in IfcRelContainedInSpatialStructure class. Through an examination of the IFC schema and ifcOwl, we found that although the current IFC data structure allows strict-controlled and well standardized building data in IFC, the complicated relationships make it extremely difficult to query and extract building information. For example, if we want to find out which storey a wall belongs to, the query has to check relationships like RelatedElements–IfcRelContainedInSpatialStructure-RelatingStrucure, which is quite time consuming and difficult for people who do not understand IFC. In this case, the structure needs to be tailored to meet contextual requirements and allow rapid data look-up. Therefore, we Figure 4.12   A RDF graph of the Campus-Ventilation ontology developed for Design Scenario #1 46  simplified the sophisticated relationships and renamed the terminologies to make it more understandable to people without knowledge about BIM or IFC (e.g., urban planners). The new structure is shown in Figure 4.14. A property isContainedIn directly links a building element and a spatial structure, and hasAggregate is used to create a relation between the two spatial structure elements. [51]  47    Figure 4.13   An example of the IFC structure [44]  Figure 4.14   An example of redefined building structure 48  An additional challenge is that the IFC schema has many terminologies that are ambiguous to urban planners. For instance, in the IFC schema, IfcSlab is a class to represent the concept of slab, which is a vertical construction component that encloses a space. This concept is frequently used in construction, but urban planners usually do not understand it well. In an IFC model, floors, roofs, and even stair landings can be instances of IfcSlab. What urban designers are interested in is the functionality of the slab, i.e., Does the slab serve as a floor, a roof or a foundation? Hence, we conducted a semantic translation in this case, and new classes were added to the new building ontology. A list of the identified classes and some common properties of building elements is shown in Figure 4.15.   Additional semantics for Design Scenario #2 also need to be developed in the new building ontology. In IFC, building materials also have a complicated organization. A class IfcRelAssociatesMaterial is used to link building elements and their related material information. Furthermore, materials have multiple arrangements in the IFC standard. They can be arranged by layers, by parts, or by a single material, and additional classes are used to store the material usage and analytical properties. Therefore, we decided not to use the IFC structure and instead used part of the TLF structure (introduced in Section 4.1.2). Figure 4.16 illustrates how materials are connected to building elements. In this case, a wall is a subclass of BuildingElement, and it can have one or several material layers. Each material layer has two common properties: thickness is used for thermal performance calculations, and order denotes its position when the Figure 4.15   Classes of building elements, their common properties and spatial structure in the new building ontology 49  element has multiple material layers. Material layers are associated with the construction materials which we developed as the design scenario.  Location Standardization for Ontologies  The location of a building is important to model an existing community in the real world because it is associated with climate conditions and solar influence. Both GIS and BIM contain rich spatial data that stores the location information, and thus the ontologies we developed should also represent such semantics. However, GIS and BIM usually use different coordinate systems to map the location of geometric elements. GIS often uses georeferenced coordinates while BIM uses Cartesian coordinates. Hence, consistency needs to be achieved between GIS and BIM data. GeoSPARQL protocol is developed by OGC to represent and query geospatial data (geometric data) in RDF. GeoSPARQL defines any spatial object as a Feature that can have a spatial location, and a Feature can associate with one or multiple Geometry through the hasGeometry property. Geometry, such as a polygon, is the representation of the shape of a spatial object. To represent the Geometry for a Feature on the map, the GeoSPARQL protocol uses the asWKT property to store the geometry serialization in Well Known Text (WKT) format, a text mark-up language representing geometries. An example of WKT is a set of longitudes and latitudes. With GeoSPARQL, the ontology of geometries (e.g., a building’s shape) can be linked to the ontology Figure 4.16   The building ontology representing building materials for Design Scenario #2 50  of non-geometric data (e.g., a building’s name). Figure 4.17 shows how the author implemented GeoSPARQL in our ontologies. In this case, BuildingElement and CampusBuilding were specified as subclasses of Feature, and two new classes, 2D polygon and 3D Geometry, were created as subclasses of Geometry. hasGeometry was used to connect BuildingElement and CampusBuilding to their geometries. In this way, we converted Cartesian coordinates of the geometry elements in the IFC model to georeferenced coordinates. As a result, the locations of geometric data in the Geodatabase and the IFC models can be standardized in one common format, which is the WKT format.     Linked Ontology  The final task of the ontology creation step is to link the campus-ventilation and the building ontologies for further data integration. In the OWL syntax, a predefined property sameAs can be used to indicate that two URI references actually refer to the same thing. The instances connected by the sameAs property are considered the same in a RDF model. This can allow the query on any instance of a class also to be passed to the instance of another one that has similar semantics. Figure 4.18 demonstrates how the Building class and CampusBuilding class are linked and how the campus-ventilation ontology and the new building ontology are linked together. If we conduct a query on a building instance in the Building RDF model, the query also works for the same building in the CampusBuilding RDF model and can directly retrieve its information, which supports querying across two models without merging them together into one file.  Figure 4.17   The implementation of GeoSPARQL 51    4.4.2 Data Extraction and Transformation  In this step, useful information needs to be extracted from the Geodatabase and the IFC models and should be transformed to RDF triples in the Oracle database. An automated workflow and a plug-in within FME were developed to complete the tasks. The workflow can manipulate the data from Geodatabase and IFC through a set of transformers based on the developed ontology, and the plug-in receives data and converts them to RDF triples and sends them to the database. The design of functionalities of the plug-in is depicted in Figure 4.19. In FME, an instance of a class is called a Feature. The plugin receives the data stream from the last transformer, and reads Features one by one. It first checks the class type of a Feature, and then creates a correspondent instance of that feature class in the RDF model. For feature’s attributes, the plugin screens out the attributes that are not in the ontology, checks the type of the attributes, and finally creates properties using the attribute values.                Figure 4.18   An example of linking two ontologies 52    Geodatabase Transformation  The transformation of the Geodatabase data to RDF model is relatively straightforward. The Geodatabase data was directly mapped to each designated class in the ontology without too many procedures. The class mapping is shown in Table 4.4. Basically, every table was mapped to one class; however, for BUILDINGS_1TO1, its shape was extracted and was converted to the Geometry class based on the designed implementation of GeoSPARQL.   Figure 4.19   The designed functionalities of the FME plug-in 53  Table 4.4   Translation from Geodatabase to the campus-ventilation ontology  Geodatabase table Campus-ventilation ontology class BUILDINGS_1TO1          Shape CampusBuilding 2DPolygon (Geometry) B_Zone ServiceZone VentSetting VentilationSetting Schedule        Year        Week        Day  YearSchedule WeekSchedule DaySchedule  Figure 4.20 demonstrates how buildings, zones and ventilation settings were manipulated in FME. Yellow boxes are the input and output data streams, and blue boxes are the transformers we used. In this case, FME read data from BUILDINGS_1TO1 table, and gave every building a new id. Then the shape of the buildings was extracted and was converted to Geometry. Finally, the plug-in processed data and create RDF triples in the database. Although for each Feature class, the workflow is different, a common procedure is to count the instances of the class and set up a new unique id in the RDF model.    54   IFC Transformation  Since the structure of IFC is very complicated, the transformation process for the IFC data needs many more steps than that of the Geodatabase data. After investigating several IFC models, the author established the data mapping between the source IFC model and our building ontology for data transformation (Table 4.5). Most of the classes are one-to-one mapping, but there are several special cases.      Figure 4.20   Examples of Geodatabase transformation in FME 55  Table 4.5   Mapping between IFC and the building ontology classes IFC class Building ontology class IfcBuilding Building IfcStrorey BuildingStorey IfcSpace BuildingSpace IfcSlab  (PredefinedType=Floor) Floor (PredefinedType=Roof) Roof IfcRoof IFcWallStandarCase Wall IfcWall IfcCurtainWall CurtainWall IfcPlate IfcWindow Window IfcBuildingElementProxy BuildingEquipment IfcMember Structure IfcBeam IfcColumn IfcCovering Covering IfcMaterialLayer MaterialLayer (LayerThickness) (Thickness) IfcMaterialLayerSetUsage   (LayerSetDirection,DirectionSense) (Order) IfcMaterialLayerSet IfcMaterial Material  The first special case is IfcSlab. An IfcSlab instance could be a roof slab, a floor slab or a stair landing, so it was categorized into different classes in the building ontology. This categorization was based on its predefined type value in the IFC model. The second special case is IfcCurtainWall and IfcPlate. Through our examination, we found that IfcCurtainWall usually 56  does not have much useful information and often serves as a container class and the reference of multiple curtain panels. The actual panel information is stored in IfcPlate. In IFC, an IfcPlate instance could be any planar entity other than walls and slabs. If the model is developed strictly following the standards, IfcPlate should have information about its predefined type. However, we found that this situation is very rare. Therefore, during the transformation, we checked the container class of the IfcPlate instances. If an IfcPlate instance is contained in an IfcCuratinWall instance, then we consider it as a curtain panel of the curtain wall, otherwise we ignore it.   The third special case is related to materials. The IFC standard uses a very complicated structure to organize material-related information (Figure 4.21). Basically, in the transformation process, each IfcMaterialLayer instance should be converted to a MaterialLayer instance in the RDF model, and each IfcMaterial instance should be converted to a Material instance. However, to find the order of a material in the layers, it needs to check several classes and properties in the IFC data. In the IFC schema, IfcMaterialLayerSet is a list that contains multiple IfcMaterialLayer instances, and IfcMaterialLayerUsage has two properties (LayerSetDirection and DirectionSense), describing how a single MaterialLayer is arranged in the layer set. LayerSetDirection specifies the orientation of the layers. Layers could be arranged along X,Y axis (horizontal) or Z axis (vertical). For example, the material layers of a wall should be always along X or Y axis because a wall is always perpendicular to the ground. DirectionSense specifies whether the layer is arranged positive or negative along the LayerSetDirection axis. If a material layer is lined in the direction which the axis points to, then it is positive; otherwise it is negative. Another important property is the index of a material layer in the layer set. This index specifies the position of a layer in the set, which is the original order of a material layer in the source IFC model. In our building ontology, we used positive as the standard case. Therefore, if the direction sense is positive then the index is its order in the ontology. On the contrary, the order of a material is reversed based on its index if a layer’s direction sense is negative. With these three properties, the new order of a material layer can be determined in the RDF model. 57   Similar to the Geodatabase data, we also built a workflow in FME to transform data into RDF triples. Processing material-related data is a common step for all features of building elements; hence, a customized transformer composed of a set of sub-transformers was developed to be used for all classes. Figure 4.22 shows how the transformer manipulated material data in the IFC model.         Figure 4.21   The structure of material information in IFC  58  Through the abovementioned processes, both Geodatabase and IFC data can be transformed into RDF models. Figure 4.23 gives an example of the final outcome in the database. It shows a triple about a service zone in the model, which indicates that under the namespace ‘http://www.semanticweb.org/whitney/ontologies/campus#’, the resource named ‘18’ is a subclass of ServiceZone.  Figure 4.22   Processes of customized transformers for material data manipulation 59  4.4.3 Data Query and Loading  SPARQL was used in to query the RDF model, and the results were populated into an empty TFL file. Below is a SPARQL query example. This query can return all walls in the ESC building RDF model, including information about its ID, name, exposure (external or internal) and associated material layers in order.    select wid, name, layer,material_id,thickness,isExternal from table (sem_match( 'PREFIX my: <http://www.semanticweb.org/administrator/ontologies/myifc#> select ?wid ?name ?layer ?material_id ?thickness ?isExternal where  { ?b a my:Wall; my:UID ?wid; my:hasName ?name; my:isExternal ?isExternal; my:hasMaterialLayer ?l. ?l my:UID ?layer; my:hasOrder ?s. ?m my:UID ?material_id. } order by ?wid ?s ', sem_models('IFCBUILDING'),NULL,NULL,NULL,NULL,' '))  Although we have taken the TLF structure into consideration when designing the ontology, the query results still need to be organized and follow the strict TFL structure for simulation purpose. The mapping in Table 4.6 shows the manipulation of the query results. In this table, the mapping to BuildingTemplate class of TFL is variable. BuildingTemplate represents the modeling scale unit in the simulation. The mapping depends on the requirements of designers. If high-level accuracy is required, then each building space or each building storey can be assigned to a BuildingTemplate instance; otherwise a BuildingTemplate instance is created using an individual building or even multiple buildings when a large scale of data needs to be modeled. In this research, individual buildings are considered as the modeling scale unit for the target neighborhood, but the ESC building was divided into multiple parts during the simulation to demonstrate the feasibility of obtaining higher accuracy using BIM.   Figure 4.23   An example of RDF triples in Oracle database 60  A special case in this process is the curtain wall. Although curtain walls are part of wall surface, UMI does not allow the assignment of glazing materials to façades or partition walls. Therefore, we considered curtain walls as non-operable (fixed) windows in this case.   Table 4.6   Mapping between the linked ontology and the TFL classes Linked ontology class TFL class Building/ BuildingStorey/ BuildingSpace BuildingTemplate MaterialLayer Construction (All types) (Layer)  OpaqueConstruction Floor  (isExternal=Yes) ExteriorFloor (isExternal=No) InteriorFloor (Type=Ground) GroundFloor Roof Roof Wall  (isExternal=Yes) Façade (isExternal=No) Parition CurtainWall WindowConstruction Window Structure StructureDefinition VentilationSetting VentilationSetting WeekSchedule WeekSchedule DaySchedule DaySchedule Material Material OpaqueMaterial OpaqueMaterial GlazingMaterial GlazingMaterial    61   Regarding the results of data loading, part of the final TFL data was shown in Figure 4.24. It shows an instance of exterior wall, named Basic Wall. This wall is associated with two material layers, having 0.064 meters and 0.239 meters in thickness respectively.  4.5 Results Visualization  4.5.1 Energy Simulation Results  The energy modeling process was completed by another PICS team member. The author provided partial building templates to demonstrate the feasibility of the proposed design approach. Additional parameters were specified by the energy modeller. The results are estimated monthly operational energy consumption of cooling, heating, equipment, domestic hot water, lighting and the total operational cost respectively. Each type of the results is stored in a separate Comma Separated Value (CSV) file. Figure 4.25 shows an example of energy simulation results of the heating system of individual buildings in the target neighborhood. During the simulation process, the modeller divided the ESC building into three parts based on the domain knowledge of the energy simulation (Figure 4.26), which are a portion of the ground floor with an exposed yard above, the rest of the ground floor and the whole second floor. The Figure 4.24   An example of the final TLF  62  modeller the named these parts ESC-1, ESC-2 and ESC-3 respectively. In UMI, each part was assigned to a different building template class.     Figure 4.26   ESC building parts for energy simulation  Figure 4.25   An example of the energy simulation results  ESC 3 ESC 2 ESC 1 ESC 3 ESC 2 ESC 1 63  4.5.2 Results Visualization  The simulation results were linked to the Geodatabase data based on building names. Then the results were visualized in ArcGlobe, a GIS software developed by ESRI. Figure 4.27 shows annual total energy usage of the target UBC neighborhood and a closer view of ESC building. The background map is the earth satellite imagery. The buildings were extruded to 3D using their heights and were color coded by the estimated energy consumption. Using each building’s gross floor area, we also calculated the operational energy cost per square meter. Similarly, the buildings were also color coded based on the calculation (Figure 4.28). Through such methods, designers can customize the results visualization to support the energy cost evaluation. In addition, designers can view the information about individual buildings during the ecaluation process. Figure 4.27 and Figure 4.28 only show simple building descriptions. With TLF data linking to the neighborhood model, designers can also view the data used for the simulation, such as wall materials and ventilation settings of each building.   The use of the integrated data on community energy modeling and visualization allows: (1) Integrating the design and the simulation of individul buildings, such as materials, into the community environment; (2) Investigating the impact of community-level policies, such as ventilation policiesm, on individual buildings. Thefeore, designers can perform a comprehensive analysis on the overall community design, faciliteated by data integration.      64    Figure 4.27   A map of annual total operational energy use 65    ESC Building Figure 4.28   A map of annual energy use per square meter 66  In this chapter, the author presents the analysis of the UMI mechanism and its core data component, the TLF. The challenges in using the UMI platform are very typical in current urban energy modeling processes, which makes UMI a suitable platform to test the approach proposed in this research. The proposed data integration approach is an ontology-based method, and it adopts the Semantic Web technology to create a problem-specific ontology for GIS and BIM data integration. This ontology can help to solve the semantic heterogeneities between GIS and BIM data in the energy design domain. In addition, a query process was also developed to allow designers to conduct queries on the integrated data and retrieve the desired information for the simulation. A case study on the UBC campus was developed to demonstrate the feasibility of the approach, including two design scenarios at the community scale and at the building scale respectively. With another PICS team member providing the simulation results, the author helped to visualize the results within a 3D geospatial model using the ArcGIS product.    67   Chapter 5: Conclusion  A method of integrating GIS and BIM data to support urban energy design on the community scale is presented in this thesis. Through reviewing different approaches for urban energy modeling and existing information technologies, the author find that the integration of GIS and BIM can be very beneficial in this context. Therefore, this study attempts to address the data challenges in urban energy modeling by combining GIS and BIM technologies using an ontology-based approach. In this approach, GIS and BIM are integrated based on the developed ontology. The approach also automates the generation of the information required by the energy modeling process through a data query process, and then provides useful visualizations of the simulation results. Such work can alleviate the data incompatibility between GIS and BIM in the domain of urban energy design.   This research was a part of bigger research project funded by PICS, which aims to investigate energy simulation, consumption and emission in the built environment. The study follows the overall objectives of the PICS project but is centered on the technological perspective. A software called UMI was used as the platform to perform the energy simulation. One of its significant components, Template Library File (TLF), dictates the data requirements in this case. The study leveraged UMI and TLF and developed an approach for urban energy design. Regarding the data integration approach, Semantic Web was used in the data integration process. An ontology was developed to achieve semantic integration between the GIS (Geodatabase) and BIM (IFC) models. This ontology covers the semantics of two design scenarios and the core data structure of the Geodatabase and IFC data in the domain of energy modeling. We transform the data into RDF triples, using the developed ontology as a principal. Feature Manipulation Engine (FME) is exploited in this research to establish an automated workflow for data extraction and conversion. We also perform queries on the RDF model and load the results in a TLF. A case study of the UBC campus was presented to demonstrate the feasibility of the proposed approach. Two design scenarios: ventilation policy and construction material are developed to lead the execution details. Although we have not tested the method on other platforms, we believe that the approach is also applicable to other urban modeling methods, because UMI and TLF can 68  demonstrate paradigmatic data challenges of urban energy models based on our review of many other studies about urban energy modeling. With the help of other PICS members to develop and run the energy simulation, the author was able to obtain the energy consumption results and visualize them in a three-dimensional geospatial environment.   The key achievements of this research are summarized as follows: 1. A demonstration of integrating GIS and BIM data in the community energy design domain is presented. In this sense, urban designers can better manage and control different sources of data and their design scenarios.  2. A new ontology for semantic integration between GIS and BIM data is developed in the urban energy design domain. The developed ontology can be reused and extended in the future to support other energy modeling approaches. Furthermore, the ontology addresses the issue of urban planners’ lack of knowledge about AECO domain through a semantic translation between BIM and the developed ontology. We expect that planners who do not know BIM can also use our method to obtain building data and perform the simulation. In addition, the accomplishment of the proposed data integration approach in FME saves energy modellers from enormous manual work on collecting data and creating input parameters for the energy simulation. This helps modellers process original rudimentary data and manipulate them easily. Their data requirements are also satisfied through such a process.  3. The development of the query process on the integrated data to satisfy designers’ data requirements. Through the queries, designers can rapidly retrieve policy and design information for energy simulation from the RDF model.  There are three main limitations of this research: 1. The LOD of BIM is much higher than the modeling scale unit in GIS. Considering that this research uses the entire building as a single simplified object in the energy simulation process, multiple types of exterior walls may be retrieved from the IFC model, but only one of them can be assigned to the building in the template. It requires modeller’s decision on what information to use for the simulation. Hence, it is worthwhile to explore a method to accommodate high-level data to the appropriate spatial scale.  69  2. In this project, queries on RDF models require knowledge about GeoSPARQL. A graphic interface is needed to allow non-IT professionals to easily conduct data queries.  3. The quantity and quality of BIM remains a big challenge. BIM is still not a required process in construction projects, and many buildings do not have a BIM. Furthermore, existing BIMs are often incomplete and the data content often does not strictly follow the standard (i.e., no predefined type for IfcSlab), causing difficulties to retrieve related information. Hence, facilitating BIM usage from an urban planning perspective is also very beneficial to the overall BIM promotion process.   70   Bibliography   1. Natural Resources Canada’s Office of Energy Efficiency, ecoENERGY Efficiency for Buildings, N.R.C.s.O.o.E. Efficiency, Editor. 2012. 2. Natural Resources Canada. Energy-efficient new homes. [Video] 2016  [cited 2016 October 27]; Available from: http://www.nrcan.gc.ca/energy/efficiency/housing/new-homes/5023. 3. Jones, P.J., Lannon, S., and Williams, J. MODELLING BUILDING ENERGY USE AT URBAN SCALE. in Seventh International IBPSA Conference. 2001. Rio de Janeiro, Brazil. 4. Rich, S. and Davis, K.H., Geographic Information Systems (GIS) for Facility Management. 2010, IFMA FOUNDATION. 5. NATIONAL BIM STANDARD-UNITED STATES. WHAT IS A BIM? 2016  [cited 2016 July 27]; Available from: https://www.nationalbimstandard.org/faqs#faq1. 6. Sebastian, R., et al., SEMANTIC BIM AND GIS MODELLING FOR ENERGY-EFFICIENT BUILDINGS INTEGRATED IN A HEALTHCARE DISTRICT. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2013. II-2/W1: p. 255-260. 7. Liu, R. and Issa, R.R.A., 3D Visualization of Sub-Surface Pipelines in Connection with the Building Utilities: Integrating GIS and BIM for Facility Management, in Computing in Civil Engineering. 2012. p. 341-348. 8. Niu, S., Pan, W., and Zhao, Y., A BIM-GIS Integrated Web-based Visualization System for Low Energy Building Design. Procedia Engineering, 2015. 121: p. 2184-2192. 71  9. de Laat, R. and Berlo, L.V., Integration of BIM and GIS: The development of the CityGML GeoBIM extension, in Advances in 3D geo-information sciences. 2011, Springer Berlin Heidelberg. p. 211-225. 10. Karan, E.P. and Irizarry, J., Extending BIM interoperability to preconstruction operations using geospatial analyses and semantic web services. Automation in Construction, 2015. 53: p. 1-12. 11. Noy, N.F., Semantic integration: a survey of ontology-based approaches. ACM SIGMOD Record, 2004. 33(4): p. 65-70. 12. Techopedia. What is a Proof of Concept (POC)? 2016  [cited 2017 January 03]; Available from: https://www.techopedia.com/definition/4066/proof-of-concept-poc. 13. The U.S. National Archives and Records Administration, Recommended Practice: Developing and Implementing an Enterprise-wide Electronic Records Management (ERM) Proof of Concept Pilot, T.U.S.N.A.a.R. Administration, Editor. 2006, Electrocic Government. 14. K-12 BLUEPRINT. Planning a Proof of Concept. 2014; Available from: https://www.k12blueprint.com/sites/default/files/POC-Planning.pdf. 15. Hosseini, H.N. and Welo, T. A framework for integrating reliability and systems engineering: proof‐of‐concept experiences. in INCOSE International Symposium. 2016. Wiley Online Library. 16. Natural Resources Canada. Key concepts: energy efficiency. 2016 2016-07-06 [cited 2016 July 22]; Available from: http://www.nrcan.gc.ca/energy/efficiency/buildings/eeb/key/3967. 72  17. Metz, B. and Davidson, O., Modelling and Cost Assessment, in Climate change 2001: mitigation: contribution of Working Group III to the third assessment report of the Intergovernmental Panel on Climate Change. 2001, Cambridge University Press. p. 489-490. 18. Beeck, N.M.J.P.v., Classification of energy models. FEW Research Memorandum. Vol. 777. 2000, Tilburg. 19. Perez, D. and Robinson, D., Urban Energy Flow Modelling: A Data-Aware Approach, in Digital Urban Modeling and Simulation, G.A. Stefan Müller Arisona, Jan Halatsch, Peter Wonka, Editor. 2012, Springer Berlin Heidelberg. p. 200-220. 20. Reinhart, C.F. and Cerezo Davila, C., Urban building energy modeling – A review of a nascent field. Building and Environment, 2016. 97: p. 196-202. 21. Cerezo, C., et al. THREE METHODS FOR CHARACTERIZING BUILDING ARCHETYPES IN URBAN ENERGY SIMULATION. A CASE STUDY IN KUWAIT CITY. in Building Simulation. 2015. Hyderabad, India. 22. Taylor, S., Fan, D., and Rylatt, M., Enabling urban-scale energy modelling: a new spatial approach. Building Research & Information, 2013. 42(1): p. 4-16. 23. Esri. What is GIS. 2016  [cited 2016 December 10]; Available from: http://www.esri.com/what-is-gis. 24. Fabbri, K., Zuppiroli, M., and Ambrogio, K., Heritage buildings and energy performance: Mapping with GIS tools. Energy and Buildings, 2012. 48: p. 137-145. 25. Ramachandra, T.V. and Shruthi, B.V., Wind energy potential mapping in Karnataka, India, using GIS. Energy Conversion and Management, 2005. 46(9-10): p. 1561-1578. 73  26. Open Geospatial Consortium. Geography Markup Language. 2016  [cited 2016 Aug 10]; Available from: http://www.opengeospatial.org/standards/gml. 27. Bahu, J.-M., et al., TOWARDS A 3D SPATIAL URBAN ENERGY MODELLING APPROACH. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2013. II-2/W1: p. 33-41. 28. Nouvel, R., et al. CityGML-based 3D city model for energy diagnostics and urban energy policy support. in 13th Conference of International Building Performance Simulation Association. 2013. Chambéry, France. 29. buildingSMART. IFC Overview summary. 2016  [cited 2016 June 16th]; Available from: http://www.buildingsmart-tech.org/specifications/ifc-overview. 30. KweonHyoung Lee, I.K., SeungYeon Choo, MODEL STUDY OF DESIGN COMPONENTS FOR ENERGY-PERFORMANCE-BASED ARCHITECTURAL DESIGN USING BIM LOD 100. Journal of Green Building, 2015. 10(2): p. 179-197. 31. Giuda, G.M.D., Villa, V., and Piantanida, P., BIM and Energy Efficient Retrofitting in School Buildings. Energy Procedia, 2015. 78: p. 1045-1050. 32. Wu, B. and Zhang, S., INTEGRATION OF GIS AND BIM FOR INDOOR GEOVISUAL ANALYTICS, in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. 2016, XXIII ISPRS Congress. p. 455-458. 33. El-Mekawy, M., Östman, A., and Shahzad, K., Towards Interoperating CityGML and IFC Building Models: A Unified Model Based Approach, in Advances in 3D Geo-Information Sciences, G.K. Thomas H. Kolbe, Claus Nagel, Editor. 2010, Springer. p. 73 - 93. 74  34. El-Mekawy, M., Östman, A., and Hijazi, I., An Evaluation of IFC-CityGML Unidirectional Conversion. International Journal of Advanced Computer Science and Applications, 2012. 3(5): p. 159-171. 35. Mignard, C. and Nicolle, C., Merging BIM and GIS using ontologies application to urban facility management in ACTIVe3D. Computers in Industry, 2014. 65(9): p. 1276-1290. 36. Karan, E.P., Irizarry, J., and Haymaker, J., BIM and GIS Integration and Interoperability Based on Semantic Web Technology. Journal of Computing in Civil Engineering, 2016. 30(3): p. 04015043. 37. Uschold, M. and Gruninger, M., Ontologies and semantics for seamless connectivity. ACM SIGMOD Record, 2004. 33(4): p. 58-64. 38. World Wide Web Consortium. Semantic Web. 2015  [cited 2016 June 17th]; Available from: https://www.w3.org/standards/semanticweb/. 39. W3C OWL Working Group. Web Ontology Language. 2013  [cited 2016 August 10]; Available from: https://www.w3.org/OWL/. 40. Malgundkar, T., GIS Driven Urban Traffic Analysis Based on Ontology. International Journal of Managing Information Technology, 2012. 4(1): p. 15-23. 41. Pauwels, P., et al., Three-dimensional information exchange over the semantic web for the domain of architecture, engineering, and construction. Artificial Intelligence for Engineering Design, Analysis and Manufacturing, 2011. 25(04): p. 317-332. 42. Niknam, M. and Karshenas, S., Integrating distributed sources of information for construction cost estimating using Semantic Web and Semantic Web Service technologies. Automation in Construction, 2015. 57: p. 222-238. 75  43. Costa, G. and Madrazo, L., Connecting building component catalogues with BIM models using semantic technologies: an application for precast concrete components. Automation in Construction, 2015. 57: p. 239-248. 44. Open Geospatial Consortium, OGC GeoSPARQL - A Geographic Query Language for RDF Data. 2012. 45. buildingSMART. buildingSMART International home of openBIM. 2016  [cited 2016 July 29]; Available from: http://www.buildingsmart-tech.org/. 46. Safe Software Inc. How FME Works. 2016  [cited 2016 November 04]; Available from: https://www.safe.com/how-it-works/. 47. Davila, C.C., Reinhart, C., and Bemis, a.J., Modeling Boston: a workflow for the generation of complete urban building energy demand models from existing geospatial datasets. . Energy, in press. 48. Cerezo, C., Dogan, T., and Reinhart, C. Towards standardized building properties template files for early design energy model generation. in ASHRAE/IBPSA-USA Building Simulation Conference. 2014. Atlanta. 49. Stanford Center for Biomedical Informatics Research. Protégé. 2016  [cited 2016 November 05]; Available from: http://protege.stanford.edu/. 50. The University of British Columbia. Vancouver Campus Plan. 2014  [cited 2016 Nov 04]; Available from: http://planning.ubc.ca/vancouver/planning/policies-plans/land-use-governance-documents/vancouver-campus-plan. 51. buildingSMART. IfcRelContainedInSpatialStructure. 2016  [cited 2016 November 05]; Available from: http://www.buildingsmart-76  tech.org/ifc/IFC2x4/rc2/html/schema/ifcproductextension/lexical/ifcrelcontainedinspatialstructure.htm.   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0340680/manifest

Comment

Related Items