- Email: [email protected]
Information Quality Assessment for Facility Management
Advanced Engineering Informatics 33 (2017) 181–205
Contents lists available at ScienceDirect
Advanced Engineering Informatics journal homepage: www.elsevier.com/locate/aei
Information Quality Assessment for Facility Management Puyan A. Zadeh a,⇑, Guan Wang a, Hasan B. Cavka a, Sheryl Staub-French a, Rachel Pottinger b a b
Department of Civil Engineering, University of British Columbia (UBC), Vancouver, Canada Department of Computer Science, University of British Columbia (UBC), Vancouver, Canada
a r t i c l e
i n f o
Article history: Received 14 December 2016 Received in revised form 13 June 2017 Accepted 14 June 2017
Keywords: Information Quality Building Information Modeling (BIM) Facility Management (FM) BIM for FM Model-based Project Delivery
a b s t r a c t Assessing the quality of building information models (BIMs) is an important yet challenging task within the construction industry as projects are increasingly being delivered with BIM. This is particularly essential for facility management (FM) users as downstream information consumers that depend on the quality of models developed in the previous project phases. The research presented in this paper addresses this challenge by introducing a framework for information quality assessment (IQA) of BIMs for FM uses. The IQA framework is the outcome of an extensive study of two large owner organizations involving numerous BIM projects. The framework is structured based on the essential FM subjects: assets, spaces, and systems, and the model characteristics: objects, attributes, relationships, and spatial information. The framework is then operationalized through the development and evaluation of information quality (IQ) tests using BIM model checking tools across three projects with different levels of detail and complexity. The proposed IQA framework and associated tests advance the state of knowledge about BIM quality in terms of methods to represent and evaluate conformance to owner requirements. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction With the growing adoption of BIM (building information modeling) within the AECOO (architecture, engineering, construction, owner and operator) industry, owner organizations are increasingly requiring BIM as part of the project delivery process and exploring how BIM can be leveraged for facility management (FM) purposes [1]. Research shows that a high number of private and public owners believe in the importance of developing capabilities in their organizations to leverage BIM for the operation phase [1]. Owners believe that a key benefit in using BIM for operation and maintenance comes from the complete and accurate information provided by the delivered models [1]. However, several studies have identified the lack of information quality (IQ) as a major barrier for this aim [2–5]. Specifically, researchers confirm that poor IQ of delivered information causes significant costs and rework for the operations phase [6,7]. Therefore, it is critical for stakeholders in the AECOO industry to be able to assess the quality of BIMs at different stages throughout project delivery and at handover to ensure the usefulness of building information for
⇑ Corresponding author. E-mail address: [email protected] (P.A. Zadeh). http://dx.doi.org/10.1016/j.aei.2017.06.003 1474-0346/Ó 2017 Elsevier Ltd. All rights reserved.
operation and maintenance purposes. This requires clear, structured, and flexible methods for describing and assessing the quality of delivered models in terms of conformance to owner requirements. IQ is described and interpreted in different ways by researchers and owner organizations. The proposed approaches in related literature mainly focus on assuring the quality of BIMs during the modeling phase. For instance important organizations such as BSI [8] GSA [9] LACCD BIMS [10] SBCA [11] provide measures for modelers to avoid quality related issues in their modeling process without proposing specific quality assessment methods [8–11]. Other research works, such as Tribelsky and Sacks [5], have their focus on the data exchange between different models and propose approaches to assess the quality loss in such exchanges [5]. Furthermore, another research stream aims to develop and improve evaluation methods focusing on the quality of model conformance to industry standards such as conformance of Industry Foundation Class (IFC) outputs [4] and Model View Definition (MVD) [12]. Although these approaches provide an important step forward, these works are limited to generic checks offered by common BIM authoring tools that help modelers avoid different IQ issues. Thus, additional research is needed to better understand how to
182
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
characterize the quality of BIMs and evaluate their conformance to owner-specific requirements. The main objective of this research is to address this research gap by providing a structured framework for information quality assessment (IQA) of BIMs for facility management purposes. This framework was developed based on an extensive study of two large owner organizations involving a series of BIM-based projects in which we were able to interview the stakeholders and observe their operation and maintenance processes. The specific research questions pursued include the following: 1. What are the information needs of owner organizations for creating intelligent FM systems? 2. What are the relevant IQ dimensions and related characteristics required to systematically understand and assess the models? 3. How can IQ tests be operationalized to evaluate the conformance of a given BIM for owner-specific information requirements? In response to these questions, we developed an IQA framework based on the identified owner information needs. The framework allows users to systematically characterize the information quality dimensions that are relevant for a particular owner and assess the IQ of BIMs at different project stages with respect to the owner’s FM requirements. The structure of this framework is organized based on four different model characteristics: entities, entity attributes, the relationships between entities, and the spatial information (location and shape) of each entity. The structure of the framework also considers the three essential FM terms: assets (equipment), spaces, and MEPF (mechanical, electrical, plumbing, and fire safety) systems. The model characteristics and FM terms describe the subject of each required IQA test in the framework. Moreover, the framework indicates for each IQA test, the required proxy indicators and benchmarks, and it proposes relevant methods to perform the IQA tests. Using this framework, we operationalized the specific IQA tests for three different projects with different size, complexity and level of detail to show the feasibility and adaptability of the introduced framework in practice. Using the introduced framework in this research is grounded in firsthand observations in actual projects and provides the owners and stakeholders the awareness about the IQ issues and aims to encourage them to support the overall goal of modelbased project delivery. The implementation of IQA tests on exam-
ples from the practice is a proof of feasibility of establishing structured quality control strategies in construction projects. Furthermore, the variety of the practical examples introduced in this research aims to showcase the comprehensibility of IQA tests to cover different quality issue types in a BIM. The framework’s feasibility and comprehensibility in this quality research follow the interpretive and theoretical validity concept introduced in [13,14]. In the next section, we provide examples of representative quality issues in delivered BIMs based on the BIM projects we analyzed. In Section 3, we discuss the research background and related works that includes studies from computer sciences (CS) and the AECOO domain. Then in Section 4, we introduce our case studies and the different steps in our methodology to develop the proposed IQA framework. In Section 5, we provide a detailed explanation of our IQA framework. Then in Section 6, we describe how to operationalize IQ tests from the framework based on selected examples from our case study projects. Finally, Section 7 provides some concluding remarks. 2. Practical motivation regarding current quality issues of BIMs for FM The motivation of this research has its roots in studying the deliverables of several BIM projects and interviewing numerous FM personnel within two different owner organizations. The provided examples in this section are drawn from what has been observed in those projects and cover all typical quality issues of BIMs for FM. In this regard, we especially focused on the identification of obstacles in establishing methods for model-based analysis and challenges in utilizing delivered BIMs in the operations phase of a building. Analyzing a diverse range of BIM projects through the lens of building operations has highlighted that the quality of BIMs often does not satisfy the expected level of quality for FM purposes, which in turn causes issues for BIMs being useful for operation and maintenance purposes. These IQ issues could be observed across different project phases up to and including project handover. For a better understanding of the various types of issues, the following figures provide typical examples of IQ issues from our case studies and highlight the specific information quality dimensions that are exemplified in each example. 2.1. Example 1
? (a) As-is
(b) BIM
Example 1. This example from the Project #3 shows that the large white expansion tank in the as-is photo (left) is missing in the mechanical BIM (right). Therefore, the model has an incomplete representation of the as-is.
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
183
2.2. Example 2
(a) Assets as a part of a system
(b) Assets without a system assignment
Example 2. These two snapshots demonstrate the assets which are assigned to a building system (left) and assets without any system assignment (right) from Project #1, i.e., the relevant MEPF system attribute for these assets is empty. However, in the reality, these assets are in fact parts of different systems. The right snapshot includes important assets such as AHUs and some heat pumps that are not assigned to a any MEPF system. Thus, the model contains incompleteness regarding to the system definitions.
2.3. Example 3
Example 3. In this example from Project #3, the attribute ‘‘Manufacturer” has a different value in BIM (right) in comparison with the actual work sheet on the actual heat pump (left). Therefore, the model represents the as-is situation inaccurately.
2.4. Example 4
Example 4. This example from Project #2 shows the redundancy in naming spaces within a project which contradicts the owner’s uniqueness policy for naming spaces. We discuss this example in Section 6 in more detail.
184
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
2.5. Example 5
Sign Standards and Guidelines 3.1
Sign Classification for Interior Signs
Signs are given a sign code based on the following classification system. Each installed sign will have a unique number attached to its sign code. For Interior signs, the sign code includes the building code. Lower case letters (e.g., N2a) refer to sub-types. This number will be associated with specification drawings and can also be attached to the sign itself to aid in identification. Example 5. This is an example of the space naming requirements for buildings from Owner #2’s projects. According to this requirement, the space names must follow a specific format and must be unique within the entire Owner #2’s buildings. The challenge here is to find a systematic way to evaluate the well-formedness and redundancy of these names.
2.6. Example 6
(a) As-is
(b) BIM
Example 6. This example shows a blue box with the object type ‘‘Generic” in the BIM model (right) which is per se not understandable what it represents. After a walkthrough in the mechanical room of Project #3, we could identify that the blue box is a placeholder for the MCC (Motor Control Center) (left). This example, highlights the lack of understandability of some model parts. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
According to our observations, the examples above are representative IQ issues that often occur in common BIM projects. They show that some IQ issues might be related to deficiencies in single information pieces (Examples 1 to 3), some are related to the entire model and its compliance to specific rules (Example 4 and 5), and some others are about how the information is represented by the model to the user (Example 6). We discuss these three different levels (entity, model, and user) in a separate work in more detail [15]. The examples above cover the different IQ issue types that we have been observing in different projects from the FM perspectives. These IQ issue types can be generally divided into following five categories: 1. 2. 3. 4. 5.
Incompleteness (Example 1 and 2) Inaccuracy (Example 3) Redundancy (Example 4) Well-formedness (Example 5) Understandability (Example 6)
In addition to the examples above, we also identified issues related to the ‘‘level of detail” (LOD) of the model or the ‘‘level of representation” of the reality. However, LOD issues can be understood as model issues in representing the reality in a sufficient (complete) and accurate way based on the user’s requirements. For instance, when the model only includes a generic geometry, like a box as a placeholder, to represent an air handling unit
(AHU), this might satisfy the architect’s needs but is likely not sufficient to meet the needs of the FM users. In this example, on the one hand the AHU representation is ‘‘incomplete” and does not include important parts, such as filters, heating/cooling components, blower, etc., but on the other hand, the AHU representation is also ‘‘not accurate” enough in showing its orientation, connections to other equipment, and the available free space for maintenance activities. The white box in Example 6 is an example for poor LOD representation of an AHU, whereas the turquoise expansion tank provides a sufficient representation. Hence, the LOD issues can be broken down into two fundamental issues that are covered in our examples above, which are the issues related to the incomplete representation of the required information and the issues related to the inaccurate representation of the required information. Furthermore, reviewing these examples shows that such errors can be caused at different project stages from the design up to project delivery. Generally speaking, any interactions with the model might be a source of IQ problems, including activities related to modeling, data exchange between different project participants, and the data consumption as discussed in [4]. However, the modeling procedure has the most impact on assuring the IQ of models, which is why many organizations such as BSI (British Standards Institution), GSA (General Services Administration), LACCD (Los Angeles Community College District), and SBCA (Singapore Building and Construction Authority) provide guidelines for modelers
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
to assure the quality of BIMs [8–11]. Particularly from the FM perspective, developing and maintaining the models are important. FM users, as the downstream information consumers, are very dependent on the quality of modeling in the previous phases. As shown in the examples above, it is critical for owners to be able to assess the representation of FM information needs during the modeling process to ensure that these significant IQ issues are avoided. The following sections provides the necessary background on the topic of IQ from the perspectives of the AECOO industry and relevant works in computer science (CS). 3. Research background The structured IQA approach which is introduced in this research is developed based on the related research works in the areas of building information modeling and computer science. In this section, we introduce the research background which is accordingly divided into the role of BIM and its quality for FM as well as the fundamental points in the IQ discussion from the computer science perspective. 3.1. BIM quality BIM can be described as an environment for collecting and maintaining all building-related information which is critical for collaboration and information exchange between different disciplines and across different project phases [16]. Detailed explanations about the BIM principles are discussed in [17]. As mentioned in the introduction, the role and significance of BIM for different disciplines is growing [1]. In particular, BIM has a great potential for being used as the source of information for setting up CMMS (Computerized Maintenance Management System) and as a tool to track and store changes happening to the as-is situation of facilities [18]. Therefore, BIM has the potential to fundamentally change the project delivery process [19]. Nevertheless, research shows that BIM has not been successfully used in a broader spectrum for the facility maintenance and operation purposes [1]. One explanation for this outcome is because of the difficulties that the owners have in assessing the quality of the delivered models. As a consequence, in many projects, owners do not include FM information needs in the contracts and project requirements. In other words, one can’t ask for something that one can’t assess. There are a few research efforts that specifically focus on assessing the IQ of models in the AECOO domain. Berard in [2] presents a framework for assessing design information quality from the builder’s perspective. Although this is different from the FM perspective, it provides a useful approach to develop a suitable framework for FM, since the presented framework in [2] corresponds with the related works in CS and is structured through identification of relevant IQ dimensions. Kasprzak et al. discussed in [20] a method for developing a standard procedure for the verification of the completeness and accuracy of modeled facility information. This work is an important point of departure for developing owner requirements tailored for facility operation and maintenance purposes which can be used as a benchmark for IQA tests. Similarly, Du et al. in [3] introduce benchmarking metrics for the IQA of BIM cloud performances that also includes IQ related metrics, such as model quality, accuracy, and usefulness. In [4] Solihin et al. introduce a testing methodology for validation of IFC files. Although this work does not have a specific FM focus, it does provide a useful model for developing structured quality tests. Furthermore, many researchers and organizations have been working on the measures and control mechanisms to assure the quality of models during the modeling process [8–10,21,11]. These research efforts provide generic guidelines (usually based on
185
checklists) without detailed and specific assessment approaches for the facility operation and maintenance. A popular approach for IQ control of BIMs for FM is to export the mechanical models to the COBie (Construction Operations Building information exchange) sheets and investigate the errors and outcomes as discussed in [22,23,20,24–26]. In this regard, there are also quality assurance approaches with MVD (Model View Definition) and IDM (Information Delivery Manual) from buildingSMART [27], which similarly focus on the information exchange between different information sources. Some assessors use such approaches to extract information from a model and check whether that model is in compliance with a specific required data structure, which is for example defined through MVD (similar approaches are observed with COBie). An overview about the different literatures from the AECOO domain on BIM-IQ is introduced in [15]. All these approaches are designed to be used by the modelers, while the owners and operators as the actual information consumers after the project delivery are still lacking a comprehensive and structured approach to assess the quality of delivered BIMs [16]. There are a variety of tools that can be used for IQ assessment of BIMs. BIM authoring tools like Autodesk Revit and BIM-based construction management tools such as Navisworks include basic features that allow users to run simple queries. The ‘‘Schedules” feature in Autodesk Revit is a great example of such tools. In addition, there are software applications such as Solibri Model Checker (SMC) or iTWO that are capable to create complicated queries and extract required information from a BIM. 3.2. IQ in CS In contrast to the AECOO domain, the subject of IQ has been discussed in computer science research for more than two decades. Generally speaking, there are two major fundamental research streams in the CS domain; one has the main focus on the usefulness of information for the information consumers, and the other one is based on how the information is modeled in an information system, which are both introduced below. The first stream is based on the original work of Wang and Strong [28], which proposed a framework for different IQ dimensions from the information consumer’s perspective [28]. According to Wang and Strong [28], the ‘‘fitness for use” (or the information usefulness) is the major criterion for IQ; a piece of information is of high quality when it is ‘‘useful” for its consumer [28]. Therefore, the identification of consumer’s information needs is an essential step in assessing the quality of information systems like BIMs. In later research [29], the same research group implemented the proposed IQ framework in [28], and applied it on different case studies from healthcare, finance, and manufacturing. Describing IQ from this perspective is also suitable for applications in domains such as the AECOO. For instance, the author in [30] states that in an information system the information and the functionality for managing the information should supply ‘‘value” to the information consumer. ‘‘Value” can be understood as a function of information quality, which is defined in terms of its usefulness in making decisions [31]. Nevertheless, there is still a great need for studies which have a specific focus on IQ in the AECOO domain and transfer the available knowledge from CS into this domain. Wang and Strong and later Lee et al. identified a large number of IQ dimensions from the academics’ and practitioners’ perspective that they categorized in four essential categories [28,29]: 1. Intrinsic with dimensions such as accuracy (including precision, correctness), consistency, unambiguity, reliability, and believability. 2. Contextual with dimensions such as completeness, quantity, timeliness, and level of detail.
186
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
3. Representational with dimensions such as understandability, readability, interpretability, and meaningfulness. 4. Accessibility with dimensions such as accessibility, and availability. These IQ dimensions might overlap and it is up to the assessors to define a clear distinction between them. This categorization is an important point of departure for structuring framework for new research domains as for AECOO. Depending on the domain, the significance of the dimensions above might change. In our studies, we have observed that researchers in the AECOO domain frequently describe their dimensions differently. For instance, while Solihin et al. and Kasprzak et al. refer to IQ issues with the missing objects as information ‘‘Correctness” [4,20] Berard uses the term ‘‘Correction” [2], Jylhä and Suvanto see it as a part of information ‘‘Availability” [32] and Tang et al. use the term ‘‘Com prehensiveness” [33]. Thus, developing similar categorization as in [28,29] is an important contribution in terms of formalizing the IQ discussions within the AECOO domain. The second research stream for describing IQ dimensions has its roots in the essential work of Redman [34]. According to [34], a model-based information system can be described through three basic characteristics: entity, attributes of the entities, and the values of the attributes. In recent years, the theory of models is integrated (separately from the IQ discussion) into the AECOO domain, and several researchers have been working on additional elements that are specific for this domain. A precedent setting work is presented in [35] where the authors describe five basic characteristics for describing a BIM: 1. ‘‘Functional type” is the equivalent to the entity which is not only covering the building objects but also views, schedules, plans, etc. 2. ‘‘Geometry” is basically the shape of the object. 3. ‘‘Attributes”. 4. ‘‘Relations” between objects. This was originally covered in [34] through special attributes. 5. ‘‘Behavioral rules” such as: a window must be always attached to a wall. One noticeable point that is missing in this categorization of model characteristics of a BIM is the object’s location which is an essential term in three-dimensional models (as information systems). Therefore, we adjust the ‘‘Geometry” with the ‘‘Spatial information” to represent both the object’s geometry and the location. Thus, the basic model characteristics for a BIM can be categorized as follows: 1. 2. 3. 4. 5.
Entity types Spatial information Entity attributes Relations Behavioral rules
The current studies in CS domain about the IQ are mainly based on the two main streams above, i.e. the works of Wang and Strong [28] and Redman [34]. However, there are still research works with interesting perspectives that could be useful for applications in the AECOO domain as well. One of these works is presented in [36] where the authors propose an IQ framework for the entire information system rather than focusing only on single information sources. Considering the heterogeneous structure of information systems in construction projects, having a BIM as the only information source is unrealistic. Therefore, such viewpoint is helpful when the IQ within the entire project should be assessed.
Reviewing the related literature highlights that IQ is a relatively new subject in the AECOO domain and there is still a need for further research in order to establish common ground for mutual understanding and knowledge exchange within the domain. Furthermore, it is also important to recognize that there are fundamental research works in CS that can be transferred into the AECOO domain to shape the IQ discussion in a structured way that is in alliance with the similar discussions in other research domains. This review also shows that each discipline – including FM – has different information requirements and as such, the IQ dimensions need to be described for each discipline accordingly. 4. Methodology The methodology in this research includes several data collection and analytical steps based on our case study projects to investigate the following research questions: 1. What are the information needs of owner organizations for creating intelligent FM systems? 2. What are the relevant IQ dimensions and related characteristics required to systematically understand and assess the models? 3. How can IQ tests be operationalized to evaluate the conformance of a given BIM for owner-specific requirements? In this research, we followed a specific roadmap in order to seek answers to the above questions. Fig. 1 illustrates the major steps of our roadmap, which is a generic approach that can be applied for different FM related purposes. This roadmap was previously introduced and applied for space management in one of our former research projects [16]. 4.1. Case study projects As mentioned earlier, this research is the outcome of studying several institutional and commercial projects for two different large public owners. However, for demonstration purposes and to provide a better overview, in this paper we explicitly focus on three projects that have different levels of complexity, occupancy, scale, and detail, demonstrating sufficient diversity in our IQA examples. The two owners we investigated are: 1. Owner #1 is a large public owner organization that is responsible for the planning, building, and managing many publicly owned infrastructure in one of the Canadian provinces. Its facilities include schools, hospitals, cultural facilities, health facilities, etc. This owner is investing C$34.8 billion in the public infrastructure over the next five years. 2. Owner #2 is a large Canadian public owner organization in the education sector that delivers and maintains educational, housing, and health care facilities. This owner is investing about C $528 million in its construction projects over the next two years. The case study projects investigated include: 1. Project #1 is a museum owned by Owner #1. This C$375.5 million, 39,000 m2, project is still ongoing and it is scheduled to be finished in 2017. Over 29 different stakeholder organizations have been involved in this large public project. As a consequence, the information exchange and collaboration have been a key point. The mechanical and architectural BIMs in this project have been developed in Autodesk Revit and attempt to provide a level of detail for fabrication (LOD 400). Since BIM has had a significant role in this project, and because our research
187
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
Operaonalize IQA Tests (sec. 6)
Formalize Informaon Needs (sec. 4.3.) Study Current Pracce (sec. 4.2.) - FM roles - Informaon systems - Informaon sources - Guidelines and standards
- FM uses - Informaon type - Informaon characteriscs - Level of detail
Determine IQ Dimensions (sec. 4.4.) - Completeness - Accuracy - Redundancy - Well-formedness - etc.
Develop a Framework for IQA (sec. 5) - Model parts to be checked - IQA indicators - Benchmarks
Fig. 1. Roadmap for development of a structured framework to design and perform IQA tests.
team has been involved in several studies related to this large project, this has been a great source for the majority of our case examples for IQA. 2. Project #2 is an educational and housing complex which was finished in fall 2016 for Owner #2. This project consists of two 18 story towers for student housing and a 4 story academic building. It is 45,108 m2 and had a construction budget of C$90 million. The architectural BIM in this project is developed in Autodesk Revit and is used for design, coordination, and fabrication purposes (LOD 400). However, the mechanical model is developed in CATIA V5 which is not a BIM native environment. Therefore, in this research work we focus on the space related BIM IQ issues from this project which are based on the architectural model. 3. Project #3 is an educational and office building for Owner #2. This 4 story building with a gross area of 5675 m2, and a total project cost of C$35 million was completed in 2011. In this project, 18 different stakeholders were involved. However, the use of BIM was limited to design purposes. Both mechanical and architectural models are developed in Autodesk Revit (LOD 300). 4.2. Study current practice The first step in our research was to study the current practices related to the facility management at the owner organizations of our case studies. For this aim, one necessary step was to perform interviews with the engaged personnel and observe the work with the currently employed information systems for FM. The interviews were performed in a semi-structured way and the interviewees included personnel who have been responsible for creating, extracting, maintaining, and consuming the FM information from the project deliverables. In this way, we were able to understand the big picture of the FM information lifecycle at the owner organizations. The interviewees included 18 different FM roles at Owner #1 and 13 stakeholders at Owner #2. A comprehensive list of the interviewees is provided in Appendix A and a detailed analysis of the outcomes is published in [37]. One of the outcomes of these interviews was the identification of the employed information systems and FM applications at the partner organizations. These include: Building and Land Information Management System (BLIMS), Facilities Capital Planning and Management (VFA Facility), Facility Maintenance System (FMS), Work Order Request Tracking System (WORTS), RAP (planning program) at Owner #1
PeopleSoft, Archibus, VFA Facility, and an inhouse developed document management system at Owner #2 A further outcome was the analysis of the relevant information sources, i.e., the type and format in which the required information is available at each organization. The identified information sources in our case studies can be categorized in following categories: 1. Mechanical and architectural BIMs (at the design, construction, and handover stage). 2. Digital documents such as 2D drawings, asset manuals, and further scanned contractual documents. 3. Paper-based documents such as part of the project’s handover drawing sets, manuals, and specifications. This investigation provides an overview about the diversity and extension of the produced information for FM in the owners’ construction projects. Based on the performed interviews and observations, it was also possible to identify the relevant guidelines and standards that the owners develop to indicate the required information for FM use in their projects. In our case studies the organization specific guidelines and standards include: 1. 2. 3. 4.
Owner #1’s Technical Guidelines Owner #1’s Basic Master Specifications (BMS) Owner #1’s Owner Statement of Requirements (SOR) Owner #2’s Technical Guidelines including guidelines for: (a) Commissioning (b) Drawing and Specifications (c) Closeout and Turnover Procedures (d) Demonstration and Training (e) Operating and Maintenance Manuals 5. Owner #2’s Sign Standards and Guidelines These guidelines and standards have been useful in setting up benchmarks to investigate whether the necessary information for FM is captured during the previous project phases. Furthermore, this analysis supported the formalization of the owner organizations’ information needs. 4.3. Formalize information needs The outcome of the previous step helped to understand how the owner organizations in their current practices determine what
188
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
information is needed to be defined or captured in the models (or other deliverables) in order to be useful for FM purposes. Based on these findings, we were able to formalize the ‘‘information needs” of the owner organizations in our case studies. At this point, we would like to emphasize that the owner’s information requirement is a very project related subject matter and this can vary from case to case. However, with formalizing the owner’s information needs at this step, we cover all typical issues that we identified in our observations and introduced in Section 2. In addition, we discuss and analyze the owner’s information requirement more extensively in a separate publication [37]. Such a formalization provides clarity on essential points relevant to the development of an IQA framework, including: What FM activities need to be supported by model-based information (information that is extracted from a BIM) delivery? These activities could include: reactive and preventive maintenance, task management, space management, service history management for assets, document management, etc. What type of assets and systems should be captured in an intelligent FM system? Obviously in most facilities, not all single equipment needs to be tracked in a computerized system and so this step will help to prioritize the different assets and systems. Which characteristics of the required assets and systems are relevant for FM? How detailed and how current (up-to-date) does the required information need to be in order to be considered useful for FM? Studying the current practices at the partner organizations in our case studies resulted that it is essential for the owners to receive models which can be used as data sources for their FM systems. It is also important for the owners that the information transition from models to FM system is combined with minimum effort. In [37], we provide more detail about the owner requirements for FM. Table 1 provides examples of required information by the owners of our case studies from the IQ perspective: Examples above and similar phrases from the identified guidelines that are followed by the owners in our case studies are indications of certain levels of IQ that the owners expect from the deliverables and so we can formalize them in more generic statements such as:
Delivered information pieces must be complete. Delivered information has to follow a specific format. Delivered information must accurately represent the as-built. Delivered information has to completely include all necessary attributes of important assets.
In addition to the owners’ specific statements about their information needs, we also studied the relevant IQ requirements in different general standards and guidelines such as in [8–11]. This additional investigation confirmed the formalized general statements above from our case studies. Examples from these standards are given in Appendix B. A detailed list of required information in our case studies is published in [38] in which the information requirements are categorized according to their FM function and specific use. 4.4. Determine IQ dimensions After studying the current FM practices at Owners #1 and #2, and identification of the information sources, relevant guidelines, and finally formalizing the information needs, we came to the conclusion that the required IQ dimensions for FM purposes are often buried in different requirement clauses which need additional interpretation for a structured and computerized assessment. According to our analysis, in our case studies the owners expect that the delivered information must be: 1. 2. 3. 4. 5.
Complete Accurate (correct and precise) Understandable Unambiguous Well-formed (in compliance with requirements)
As one might expect, these dimensions and the statement before could be relevant for many other construction projects and are not dependent on the case studies. However, through our direct observations in our case studies, we were able to analyze how each of these dimensions correspond with the identified IQ issues in the delivered models (relevant examples are given in Section 2). These IQ dimensions are in accordance with the introduced works from computer science [28,29] and our previous related works [15,16]. However, the identified IQ dimensions have differ-
Table 1 Examples of owner information requirements for FM. Source
Detail
Owner #1’s – basic master specifications, Section 15720: Mechanical identification
Required attributes for AHUs: Manufacturer, Model, Fan, Capacity L/s, S.P. kPa, Motor kW, Heating Coil, Fin Series, Type, Height mm, Width mm, Rows, Capacity W, Air Entering °C, Air Leaving °C, Air S.P. Drop kPa, Water, L/s, Water entering °C, Water Leaving °C, Water Pressure Drop kPa, Steam Cap. kg/s, Steam Pressure kPa, Cooling Coil, Fin Series, Type, Height mm, Width mm, Rows, Capacity W, Air Entering (°C DB, °C WB), Air Leaving (°C DB, °C WB), Water, L/s, Water Entering °C, Water Leaving °C, Water Pressure Drop kPa, Saturated Suction °C
Owner #2’s technical guidelines, commissioning
Verify that O&M documentation left on site is complete.
Owner #2’s closeout and turnover, Sections 01781 and 01782
Required information format/contents” as specific sections.
Owner #2’s closeout and turnover, Section 4
Required as-built information Contract Drawings Specifications Addenda Change Orders and other modifications to the Contract Reviewed shop drawings, product data, and samples Field test records Inspection certificates Manufacturer’s certificates
189
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
ent level of necessity and value for different information consumers. Hence in each project, this step should be based on the current practices at the related owner organization. 5. Framework for creating and performing BIM-IQA tests for facility management The structure of this framework is organized based on the required tests for identified relevant IQ dimensions and is broken down into following parts that we discuss in more detail in this section: 1. 2. 3. 4.
Subjects of the IQA Proxy Indicators Test Benchmarks Performance Types
In this way, the framework provides specific structure for developing and performing each required IQA test. It should be noted that the introduced IQA framework in this work is developed in accordance with the previous works of [28,29,15,16]. Particularly, the categorization of the IQ dimensions is based on the research works of Wang and Strong [28] and Lee et al. [29] that we translated for AECOO domain according to the FM needs. In the next two sections, we discuss the development of this framework in more detail and will explain the operationalization of the IQA test based on the specific examples from out case studies. Table 2 shows a summarized version of the developed framework in order to provide a better overview about the main framework elements. With this framework, the assessors are able to structure their way of thinking about required IQ of delivered models, and to systematically create and perform IQA tests. A detailed version of this framework can be found at the end of this section (Table 3). In the following, we introduce the details of the main framework elements. 5.1. IQ dimensions 5.1.1. Completeness Completeness is one of the most mentioned IQ dimensions in the literature from both CS and AECOO. Ballou and Pazer in [39] considered an information system as complete if ‘‘all values for a certain variable are recorded” in that specific information system. Wand and Wang enhance this description and define completeness as the ‘‘ability of an information system to represent every meaningful state of the represented real world system” [40]. In this definition the term ‘‘every meaningful state” has a significant role which emphasizes that not ‘‘all values” must be recorded in order to have a complete information system. Similarly, Redman describes completeness as publication of all relevant information for the information consumer [34]. Finally, Assaf and Senart define completeness as all needed information to represent all information related to a real world entity [36]. Thus, from the CS perspec-
tive a complete information system shouldn’t necessarily represent all possible information about a real entity and this system would be still complete if it includes the relevant and meaningful information which is useful for the information consumer. Hence, the information consumer and their needs have a central role in determining the completeness of an information system. In the AECOO domain, the researchers agree about the necessity of information completeness, however, they do not provide an unambiguious interpretation. Completeness has been in this domain especially discussed in connection with the information exchange between different sources or models. For instance, the authors in [4] interpret the completeness as ‘‘the minimum requirement for object count [. . .] during export or import process.” This very practical interpretation can be also understood so that a model will be complete if the number of modeled elements is equal to the number of required elements. In this way, completeness check can also be seen as an existence or quantity check for required information. Based on the descriptions above we can describe completeness for BIM as following: ‘‘A BIM is complete when all necessary model characteristics are existing in it and corresponding to relevant objects or states in the real world or requirements.” Although this IQ dimension might be easy to understand, one should be careful that the main point here is about the existence of necessary information and not their verification or correctness. In other words, assessing the information completeness is about investigating what information is in the model and what is missing with respect to the requirements.
5.1.2. Value accuracy Information accuracy is another widely required IQ dimension by the standards and guidelines in different fields and therefore, it plays a central role in the works of information technology researchers. Redman et al. in [41] describe the accuracy of a piece of information in a model as the degree of closeness of the modeled value to the corresponding correct value. The correct values (or the benchmarked values) can be determined by the requirements or the reality. Similarly, Marriott in [42] defines the information accuracy as the ‘‘closeness of computations or estimates to the exact or true values that the statistics were intended to measure.” Sometimes it would be helpful to have definitions for the opposite meaning of a complex term to separate it from the similar terms. So as for ‘‘inaccuracy”, Wand and Wang explain that inaccuracy appears when an information system represents a real-world state different from the one that should have been represented [40]. This specific point of view to the accuracy is also called ‘‘information correctness” or ‘‘information integrity.” Hence, assessing the information correctness is about checking whether the values are correct which could be seen as a true/false question. However, following the interpretation of Redman and Marriott, the information accuracy can be understood as ‘‘information precision” which is a different aspect of accuracy. Whereas for the information correctness, a value can only be either correct or incorrect (true or false), the information precision is a measurable quantity.
Table 2 Framework for creating and performing BIM-IQA tests for asset and space management purposes (summarized version). 5.1 IQ dimensions
– – – – –
Completeness Value accuracy Consistency Well-formedness Understandability
5.2 Subjects of IQA 5.2.1 Model
5.2.2 FM
– – – –
– Assets – Spaces – Systems
Entities Attributes Relations Spatial information
5.3 Proxy indicators
5.4 Benchmarks
5.5 IQA performing types
– Existence – Filter and present – Filter and compare
– Related Docs – Reality – Model
– Visual – Fully automated – Semi-automated
190
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205 Table 3 Detailed framework for creating and performing BIM-IQA tests for asset and space management purposes.
(continued on next page)
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
191
Table 3 (continued)
In general, the information accuracy has a wide meaning and can basically cover several other IQ dimensions or be a result of other dimensions. Therefore, the definitions above are representing a specific interpretation of information accuracy. A different IQ aspect that can be considered under the umbrella of accuracy is the ‘‘level of representation” which implies how detailed the information is modeled. In other words, it describes if the modeled information has the required ‘‘level of detail” (LOD) (or level of representation) to accurately represent the reality in a model [34]. If we describe information accuracy as the necessity of accurate representation of the reality, then inaccuracy could be directly related to issues with many other IQ dimensions. For instance, if a model is incomplete it is consequently not representing accurately the real word. The same interpretation applies when we discuss the information well-formedness, consistency, or understandability. More concrete examples in this regard from the AECOO domain are given in [2] which highlights the relation of accuracy to completeness by defining incorrectness as ‘‘missing objects, geometries and properties”, in [3] which connects accuracy with LOD through defining it as the ‘‘degree to which a set of inherent characteristics of BIMs fulfils desired requirements,” and in [4] where the authors emphasize the relation between information
correctness and well-formedness. Hence, to avoid confusion and in order to have a consistent discussion, we consider the information correctness as the accuracy and describe it for the context of BIM as follows: ‘‘A BIM is accurate when all required model characteristics represent correctly the relevant objects or states in the real world.” Based on this definition, assessing the information accuracy of a BIM would be about verifying values, geometries, relations, and spatial information of the objects in the model with respect to the requirements. 5.1.3. Consistency (redundancy) Information consistency ensures that each information consumer has a consistent representation of the information. Inconsistencies appear when datasets overlap and represent the same or similar concepts in a different manner, or when their specific content does not correspond [34]. According to [36], information consistency in a model deals with the question that ‘‘does the [modeled] information contradict itself?” In other words, in a consistent model there shouldn’t be any contextual conflict between model characteristics. Information inconsistency occurs when different modeled information pieces refer to the same state or com-
192
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
ponent in the reality. According to [40], this includes the physical representation and representation of values in the information system. An example for inconsistent physical representation would be duplicate objects in the model which represent the same object in the reality that also can be understood as information redundancy. This is also in accordance with the interpretation of [2] about inconsistency and with [4] when they describe the correctness. An example for inconsistent values would be attribute values which refer to the same subject such as values for space use that might be defined as ‘‘classroom,” ‘‘washroom,” or ‘‘office.” Thus, we can describe the information consistency for BIMs as follows: ‘‘A BIM is consistent in representing the real world when all required model characteristics can be followed back to the reality without any semantic or actual conflicts.” Information consistency is also known as ‘‘information unambiguity” or ‘‘information uniqueness” [40,43]. Based on our definition, assessing the information consistency of a BIM would be about identifying the redundancies and contradictions in representation of values. This ambiguity can lead to inaccurate representation of the reality or the design intent. This means ambiguity can cause information inaccuracy, which is an example for one IQ issue creates another IQ issue. Recent BIM authoring tools, such as Autodesk Revit and ArchiCAD, are software that have an object-oriented data structure and provide many intelligent features to capture issues such as redundancy of object identifiers, duplications in values, placing an object multiple times at the same spot, etc. These software characteristics can guarantee the modelers the assurance of this IQ dimension. Therefore, we wouldn’t elaborate more the redundancy dimension in the IQA discussion. However, the developers of BIM tools still need to be aware of this IQ dimension’s significance and it is highly recommended to perform suitable redundancy check. 5.1.4. Well-formedness (compliance) This IQ dimension is required by different regulations and standards such as BSI [8], GSA [9], LACCD [10], and SBCA [11]. In [8], well-formedness is defined as the information format and it requires that the owner organizations must describe exactly what information types they expect at the project delivery in the OIR (organizational information requirement) of each project. Also [11] requires to have a data exchange protocol which includes the requirements for the file formats and the naming conventions. Similar requirements are formalized by [9] as well. Different researchers refer to this IQ dimension through different terms such as ‘‘Information Typing” [36], ‘‘Data Formats” [44,29], ‘‘Well-formatted Information” [28], ‘‘Compliance” [2], and ‘‘Well-formedness” [4,16]. Nevertheless, these researchers have a common interpretation of well-formedness. So the information well-formedness in BIMs can be described as follows: ‘‘A BIM contains well-formed information when all required model characteristics are defined in compliance with relevant standards or regulations.” The relevant standards and regulations, such as IFC or OmniClass, can determine the well-formedness of the model structure and semantics, or they can determine the required format of values such as owner requirements for naming conventions of entities. If the modeled information is not in compliance with the required regulations, the model is not representing the reality or the design intent precisely. This means that the lack of information wellformedness can cause information inaccuracy too. 5.1.5. Understandability Information understandability is an abstract IQ dimension which is dependent on the information consumer’s knowledge and experience, as well as on the representation technology. Wang
and Strong describe the information understandability as the ease of understanding and interpretability of information [28]. Assaf and Senart refer to this IQ dimension as the ‘‘information comprehensibility” and require that the data concepts must be understandable to humans and convey logical meaning of the described entities and allow the easy consumption [36]. Hence, the information understandability in BIMs can be described as follows: ‘‘A BIM contains understandable information when the typical information consumer is able to connect all required model characteristics to a real object or state in the constructed building.” Issues related to the incomprehensive information can often happen in a BIM as a result of modeling objects with generic type or with unspecified shapes like boxes and spheres as placeholders. Similarly to above, the lack of information understandability in a model can lead to inaccurate representation of the reality or the design intent. Thus, incomprehensible information can be another reason for information inaccuracy. 5.2. Subjects of IQA An essential part of the framework is to determine what parts of the model should be exactly the subject of IQA tests. In other words, it is necessary to indicate to what model parts the tests should be applied. In this regard, the framework covers two different perspectives. One is model theory perspective and the other one is from the viewpoint of FM users. 5.2.1. Model theory As described in the research background section, the basic characteristics of a BIM can be categorized based on the former work of Pratt [45] and Eastman et al. [35] in five categories as shown in Fig. 2. Based on the interpretation in Fig. 2, one should imagine an ‘‘Entity” as a profile in a BIM for different terms, like building elements, systems, families, views, etc. which is describing a real object or state. This entity is the place where other object characteristics will be defined and collected. In addition, all of these characteristics can have a value which is a digital amount or a certain textual content. All characteristics have significant value for building projects. However, the entity rules are implemented in the most of the BIM authoring tools and so they can be captured through a warning analysis in such tools. For example, doors as entities have the rule to always be attached to a wall. So if the designer ignores this rule while modeling, the majority of BIM authoring tools will show a corresponding error message. Furthermore such rules mainly support the design process and can be excluded from IQA for FM.
Real Object
Composition
Model
Entity has
Attributes
Spatial Info.
Relations
Rules
have
Values Fig. 2. Characteristics for representation of a real object (or state) in a 3D Model, UML notation.
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
193
Therefore, from the model theory perspective, it is necessary to consider the following four model characteristics as the subjects of IQA tests:
cators here correspond with the studied projects in this research and the identified typical IQ issues. We have divided the proxy indicators into three generic categories:
1. Entities: Assets as the most important mechanical components 2. Attributes: Asset specs (properties) which include (but not limited to): (a) Mechanical (performance) (b) Fabrication (model, make, SN) (c) Dimension (geometry) 3. Relations: Systems as asset-asset relationships 4. Spatial Information: Location and shape of assets
1. Proxy indicators which query the ‘‘existence” of the IQA subject, e.g., checking whether an asset exists in the model 2. Proxy indicators which ‘‘query and present” the IQA subject to the assessors and they compare them manually to the benchmarks, e.g., querying all assets in the mechanical room and providing a list of them that the assessors need to use it later as a checklist during their walkthroughs 3. Proxy indicators which ‘‘query and compare” the IQA subject with the benchmarks (no manual process), e.g., checking whether any spaces have redundant names
5.2.2. FM perspective The structure of the framework also considers the three essential FM terms, namely: 1. Assets (equipment), such as air handling units (AHUs), heat pumps, boilers, etc. 2. Spaces, such as rooms, corridors, etc. 3. MEPF Systems, which are mechanical, electrical, plumbing, and fire safety systems While assets and spaces are actual objects in BIM and represent geometric objects in the reality, MEPF systems represent the nongeometric (semantic) aspect of the reality. In other words, MEPF systems describe the semantic between the assets and determine which assets are working together as a system for a specific purpose. According to the object oriented data structure of IFC, the geometric information is already defined at the object level. Therefore, adding assets to the MEPF systems will provide indirectly the geometric information. For example, the definition of an HVAC system doesn’t need explicitly the geometry information of the pumps, AHUs, etc. because the considered assets already bring that information to the system and so with defining an HVAC system, we only need to determine the semantics, i.e., ‘‘which” assets are ‘‘how” interacting with each other. Thus, storing information about the geometry of assets in the MEPF systems will be redundant and geometric information is not necessary for the definition of an MEPF system. Hence, MEPF systems are a non-geometric representation of the reality in a model. MEPF systems can be generally divided into two following categories: 1. ‘‘System Classifications” that correspond to the standard definitions for MEPF systems, such as OmniClass, MasterFormat, and UniFomat (it is also possible that owners have their own standards in referring to these system classes). Examples for MEPF system classifications are: domestic hot water, domestic cold water, exhaust air, supply air, etc. 2. ‘‘Subsystems” are specific sets of assets that are usually connected to each other to fulfil a specific FM task. For instance, a branch of the air system that serves a certain section of a building falls into this category. The definition of subsystems is very flexible and it is up to the modelers to determine which assets need to be part of a subsystem. Having subsystems in a model will support troubleshooting and commissioning processes in terms of giving the practitioners a structured workflow. 5.3. Proxy indicators The practical part of the IQA happens through checking or measuring the relevant proxy indicators. Proxy indicators are generic queries formalized in the human language and are independent of assessment tools and IQA performance methods. The details of each proxy indicator depend on the required IQ dimension and the subjects of the assessment. Therefore, the discussed proxy indi-
5.4. Benchmarks In assessing the proxy indicators, the assessor needs to compare the outcomes against appropriate benchmarks. We have categorized these benchmarks into three different types: 1. Related documents: a variety of different types of documents, such as the owner’s requirements, project deliverables, building codes, municipal regulations, etc. Related documents are benchmarks for the ‘‘required information” in contrast to the built reality which is – if available – the benchmark for the ‘‘real information.” Therefore, this benchmark is suitable to investigate if a BIM represents what is required to be constructed. 2. Built reality: the actual constructed environment with as-built components. Since in a construction project there are always deviations between what is designed and what is built, this benchmark has a significant role in the IQA of models and is suitable to investigate whether a BIM represents what is actually constructed. 3. Model(s): used to identify contradictions among different information pieces within the model(s). Generally, all different models in a project can be considered as benchmarks. However in this work, we make the assumption that a single mechanical BIM contains all required information. Determining which benchmark is suitable for an IQA depends on its purpose. For instance, the related documents would be suitable benchmarks if the assessors are interested in assessing the IQ of a design BIM, while the built reality would be more relevant during the commissioning process. 5.5. IQA performing types The operationalization of the designed IQA tests can be generally divided into the three following categories: 1. Visual tests: In this performance method, the assessors compare visually the available information with benchmarks. For this purpose, they either go to the built object on-site and check the required information (we refer to these checks as ‘‘walkthroughs”), or they do the visual checks using images and scans (we refer to these checks ‘‘imaging/screen tests”). Examples for imaging tests are photos taken from the construction field, laser scans, webcams, etc. 2. Fully automated tests: Some tests have merely to do with the information within a model and so the tests can be performed using available methods in BIM authoring tools or BIM viewers. A typical example for fully automated tests is the information consistency check that identifies whether there is duplicate information for a specific term in the model (e.g. the serial numbers of different assets).
194
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
3. Semi-automated tests: This is the most common type of IQA tests. For these tests, we first filter the required information from the model using available methods in BIM authoring tools or BIM viewers. Then we check the results with the benchmark manually. For example, in order to check the specs for the entire installed pumps in a building, we filter and group them properly from the model first, and then compare them with the related documents as benchmark. 5.6. Detailed framework for creating and performing BIM-IQA tests for FM Using the descriptions above, we developed a framework for assessing the IQ of BIMs for facility management purposes (Table 3. With this framework, we provide a generic instruction for systematically performing BIM-IQA tests which is easy to understand and to perform. The starred tests in this table are discussed in more detail in the next section. 6. Operationalizing the IQA framework using case study projects In this section, we provide specific examples from our case studies to demonstrate the application of our framework for IQA. The operationalization of these tests follows a generic structure which corresponds with the structure of the framework. For each selected IQ dimension and subject of assessment, the assessor needs to identify the relevant proxy indicators and benchmarks and perform the IQA as indicated in the framework. The examples below will give a more detailed description about this operationalization. Since these instructions include generic steps, they are applicable with the majority of BIM tools. The only difference between different BIM tools is that some applications provide specific functions for querying models and so significantly reduce the amount of manual work in some IQA tests. To demonstrate the tool independency, the selected examples in this section are partly performed with Revit Schedules, which is a very limited tool within the Autodesk Revit, and partly performed with Solibri Model Checker (SMC), which is a special tool for querying BIMs. 6.1. Completeness of asset attributes The modeled assets in a mechanical BIM must include the necessary attributes for FM use with a certain value. Generally, these attributes can be categorized into three categories: 1. Attributes related to the fabrication of the assets: these are essential for finding additional information about the assets such as manuals, further specs, etc. 2. Attributes related to the performance of the assets: these depend usually on the asset’s type such as pump head, flow rate, voltage for a pump. 3. Attributes related to the assets’ geometry: these include the dimensional attributes that are usually computable from the object geometry such as height, volume, etc. Asset geometries are relevant to spatial issues like accessibility and transportability of the assets. However, using 3D BIM authoring tools implies that all assets would inevitably have geometry attributes. Thus, we exclude checking this category from our assessment. From the categories above, the existence of the first category is necessary because based on information about the fabrication, it would be possible to find the related information for the other two categories. Therefore, BIM guidelines such as GSA and LACCD BIMS address this point by requiring the following three fabrica-
tion attributes for all assets in a mechanical BIM regardless of their type [9,10]: Make or Manufacturer Model or Mark Serial Number (can’t be determined at the design or fabrication stage) For this assessment, we should filter assets through their families (or types) and the three attributes above, and then identify the incomplete asset attributes. Therefore, the proxy indicator here is the existence check of required attributes. Once the completeness of fabrication attributes is completed we can continue with assessing the type related attributes. Alternatively, these two steps can be combined to perform a type by type check as in our example below. Determining which performance attributes are necessary for a specific asset type depends on owner’s requirements and the followed guidelines by the owner. Therefore, this semiautomated assessment type needs to employ them as benchmarks. 6.1.1. IQA Test 1: Completeness of pump attributes In this example from Project #1, we combine the two attribute categories for pumps and assume that following attributes must be represented in an FM system for operation and maintenance purposes:
Make Model Pump HP (Pressure) Pump Head Flow Rate Fluid Type Voltage
In many projects, such as our case studies, the owner requirements did not specify the required information at this level of detail. However, this example shows that requiring specific asset attributes would be a feasible requirement. Furthermore, the attribute ‘‘Serial Number” is excluded because the final purchases aren’t finalized yet and so the model does not include any suitable attribute for the assets’ serial number. As shown in Fig. 3, we select the required attributes above together with the Pump Type, Family, Type and their quantity in Autodesk Revit to set up the required query for creating a suitable schedule. The schedule is filtered by the ‘‘Pump Type” with the condition of ‘‘parameter exists” and the option for ‘‘ Itemize every instance” is unchecked. These results show that there are no values set for the performance attributes: Pump HP, Pump Head, and the Flow Rate. We consider the zeros as not set and so incomplete. As for the attribute Voltage, even when the values are spread between two different attributes with the same name, we consider them as complete because the information exists. This issue is related to another IQ dimension, namely ‘‘Redundancy.” According to our investigations, the reason for having these two separate columns in this example is related to using families created by different manufacturers. In our case study project, one manufacturer defined ‘‘Voltage” as a ‘‘type property” in their families while the other manufacturer defined it as an ‘‘instance property.” 6.2. Completeness of subsystems As explained earlier, subsystems are specific sets of assets that are usually (physically) connected to each other for a specific FM purpose (Fig. 4). Subsystems are the breakdown of general systems in terms of dividing the MEPF systems into branches in different zones. However, the components of a subsystem are not necessar-
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
195
Fig. 3. Completeness test of pump attributes in the Project #1’s mechanical BIM. This test shows the incompleteness of values for Pump HP, Pump Head, and the Flow Rate.
Fig. 4. An example of a completely defined subsystem in the Project #1’s mechanical BIM. All 13 pieces are assigned to the subsystem ‘‘Mechanical Exhaust Air 39”.
196
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
ily from the same MEPF system type. For this assessment, we need to check if all assets are assigned to at least one specific subsystem. In other words, this assessment is about checking the existence of a relationship and so the proxy indicator type is checking an existence. The completeness assessment of subsystem can be performed fully automated since the benchmark is the model itself. In Revit, subsystems are defined for each asset through a specific attribute called ‘‘System Name.” Thus, to perform this assessment, we should create a query which investigates if all assets have a value for this attribute. 6.2.1. IQA Test 2: Incompleteness of subsystem In this example, we can investigate the incomplete subsystem definitions to identify the lacking information pieces. For this purpose, we set up a query for creating a Revit Schedule which contains assets’ Family, Type, and System Name. In addition, we define a filter for System Name when it ‘‘equals” the ‘‘ ” (blank). In this way, all assets with incomplete subsystem definition will be filtered and listed. Fig. 5 shows an example from our case study. These incomplete subsystem definitions need to be accordingly addressed by the modelers. Such incompleteness lead to disconnections within the semantics of the subsystems and so affect their usefulness for FM purposes. Fig. 6 visualizes the results of the same example on the same project with the Solibri Model Checker to give a better understanding about the range of this IQ issue within a complex project. 6.3. Accuracy of asset attributes The accuracy of the asset attribute values is required explicitly by several regulations and standards such as GSA, LACCD, COBIM,
and BSI [9,10,24,46]. Through this assessment, we can investigate whether the values of necessary attributes of an asset are accurately defined in the model. Similar to the completeness assessment of asset attributes, the accuracy assessment should include the consideration of two different attribute categories: 1. Attributes related to the fabrication: Make, Model, and (if exist) Serial Number 2. Attributes related to the performance. These attributes could be for pumps as an example: pressure, flow rate, fluid type, voltage, etc. Obviously, this assessment is a highly knowledge-based and time-consuming procedure. Hence, such assessment cannot be performed automatically. The general approach for this assessment is to filter and sort the assets according to their types and families and then browse them individually and check if their attribute values are accurately defined. In fact, it is reasonable to combine this assessment with the asset accuracy assessment, since the IQA steps are similar. That means finding an asset in a model, checking if it is the same asset type as it is installed on site, and then assess the necessary attributes. The benchmark for the accuracy check of attributes can include the related documents or the identification plate on the actual installed equipment. 6.3.1. IQA Test 3: Accuracy of pump attributes Fig. 7 shows the list of necessary attributes and their values for the centrifugal pumps from the mechanical BIM of Project #1. Fig. 8 is an example for a pump’s submittal data sheet that can be used as benchmark in such assessments. In creating the required Revit Schedule, a filter condition is defined to choose only
Fig. 5. Incompleteness of subsystems in the Project #1’s mechanical BIM.
197
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
(a) Assets as a part of a system
(b) Assets without a system assignment
Fig. 6. Completeness and incompleteness of subsystems in the Project #1’s mechanical BIM.
Fig. 7. Inaccuracy in the values for mark and manufacturer of centrifugal pump attributes in the Project #1’s mechanical BIM.
Submittal Data Information FI Series Pumps EFFECTIVE: MARCH 1, 2010
SUPERSEDES: OCTOBER 1, 2009
Museum Downtown __________________________________________
JOB CONTRACTOR
ENGINEER
Priority Mechanical Ltd. ___________________________
REP.
ITEM NO.
MODEL NO.
IMPELLER DIA.
G.P.M.
P-7 & P-8
FI8013
11.6" (294.6 mm)
2369.6 (149.5 l/s)
SERVICE
301-1489
5.3°C CHWS (30% P.G.)
1760 RPM MODEL 8013
Dialog ______________________________
Kehoe Equipment Ltd. _____________________________________ HEAD/FT.
134.0 (400 kPa) WEIGHT
H.P.
ELEC. CHAR.
150 (112.5 kW)
575/3/60
Lbs (kg)
PUMP/MOTOR
2619 (1188)
Fig. 8. Related document for a centrifugal pump type in the Project #1.
the ‘‘Pump Type” which ‘‘contains” the keyword ‘‘Centrifugal” and ‘‘Itemize every instance” is checked. In this example, the submittal data information shown in Fig. 8 represents the information related to two first centrifugal pumps listed in the schedule (Fig. 7). In this document the ‘‘ITEM NO.” represents the pumps’ ‘‘Mark” in the Revit Schedule which is inaccurate. Instead, the attribute ‘‘Mark” should correspond with the ‘‘MODEL NO” in the submittal document with value of ‘‘FI8013.” The main reason for this inaccuracy is that the stakeholders decided to install centrifugal pumps from the manufacturer ‘‘Taco” instead of following the suggestion of the designers which was using ‘‘Bell & Gossett” pumps. As a result, all relevant attributes are set inaccurately except the ‘‘Fluid Type.” An explanation for this case could be that the designers had used the centrifugal pumps from ‘‘Bell & Gossett” as a placeholder. However, the selected Revit Family for these pumps were still incomplete. Looking at Fig. 7 shows that the values for ‘‘Pump HP” are missing, and the values for ‘‘Pump Head” and ‘‘Flow Rate” are set to 0.
For correcting such problems, Autodesk provides several predefined families for different assets that the designers can employ in their models. For instance, a suitable Revit Family for FI8013 centrifugal pumps manufactured by Taco can be found here: http:// seek.autodesk.com/product/latest/agg/taco/Taco/FI8013. Furthermore, additional details and manuals about this type of centrifugal pumps is provided on the manufacturer’s website under: http://www.taco-hvac.com/uploads/FileLibrary/301-1489. pdf. Collecting such information should happen before the project handover and must be determined in the owner’s requirements. 6.4. Accuracy of subsystems In this assessment, we check for each subsystem whether it contains the correct components and if these components are connected to each other correctly. Similar to the other accuracy assessments, this procedure is also dependent on the assessor’s
198
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
knowledge and experience. Furthermore, it requires individual checks for each single subsystem. In assessing the accuracy of subsystems, there are two different type of checks required. Since many BIM authoring tools such as Revit automatically assign assets to a new subsystem, if they stay alone in a subsystem (a subsystem with one single component),
they cause inaccuracies. As explained earlier, subsystems represent a group of assets with a specific task and having a subsystem with only one member contradicts the concept of subsystems. In addition, assets in a subsystem need to be connected in an accurate and meaningful way. Thus, the two type of checks for assessing the accuracy of subsystems are investigating the subsystems with
Fig. 9. Subsystem ‘‘Mechanical Exhaust Air 1” with a single component ‘‘Terminals-SDE-Price_Industries Return: Size 12” is an example inaccurately defined subsystems in the Project #1’s mechanical BIM.
Fig. 10. Subsystem ‘‘Mechanical Exhaust Air 76” is the correct host for the component ‘‘Terminals-SDE-Price_Industries Return: Size 12” which is the only member of the subsystem ‘‘Mechanical Exhaust Air 1” (see Fig. 9).
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
199
Fig. 11. Disconnection in defining the subsystem ‘‘Mechanical Exhaust Air 76”. In this 3D view the components that are not a member of the subsystem ‘‘Mechanical Exhaust Air 76” are hidden.
single components, as well as checking the connectivity of its components. The first assessment has an existence proxy indicator, has the model as benchmark, and can be performed fully automated. The second assessment needs suitable queries that present the results to the assessor to visually evaluate the accuracy of subsystems based on the related documents.
6.4.1. IQA Test 4: Accuracy of subsystem In this example, the built-in System Browser in Revit can be used while the specific column for ‘‘Number of Elements” is checked in the browser setting. Now, we can go through the subsystems and evaluate their accuracy. Fig. 9 shows a return element as a ‘‘single component subsystem” while this return element has to be part of the rest of the branch (‘‘Mechanical Exhaust Air 76”) shown in Fig. 10. We highly recommend that the assessors go through all subsystems and visually controls the connectivity and accuracy of the assigned components. Fig. 11 gives an example from a disconnected subsystem which requires completion.
objects which has to be unique as well. This might cause redundancy issues. 6.5.1. IQA Test 5: Redundancy of object identifiers In the Project #1, the modelers used the attribute ‘‘Mark” as an equivalent to the asset identifier. Therefore, we created a Revit Schedule which includes this attribute together with the number of assets which have the same value ‘‘Count.” Identifiers with a number greater than 1 are redundant and need to be corrected (Fig. 12). However, as one sees, the modelers were consistent in describing the values of the attribute Mark in terms of following a specific pattern.
6.5. Redundancy of assets Many BIM standards and regulations require unique identification numbers for assets such as [9,24]. Common BIM authoring tools, like Revit, address this requirement by following an objectoriented approach in programming and in their data models. In this way, each modeled object (or entity) gets a unique ID when it is created. So any conflict with its uniqueness will produce an error by the modeling tool. Nevertheless, working merely with IDs would be confusing for non-programmers. Therefore, some modelers prefer to create a naming system for referring to their
Fig. 12. Redundancy of object identifiers.
200
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
6.6. Well-formedness of space attributes In this assessment, we investigate whether the information pieces in a BIM are defined in compliance with relevant standards or regulations. This can be performed in a fully automated way in
which we create and run queries based on the rules indicated in the related documents as benchmarks. Performing wellformedness tests is very dependent on the assessment tool. In case the assessment tool doesn’t provide rule-based queries, performing well-formedness tests becomes a semi-automated approach which
Sign Standards and Guidelines 3.1
Sign Classification for Interior Signs
Signs are given a sign code based on the following classification system. Each installed sign will have a unique number attached to its sign code. For Interior signs, the sign code includes the building code. Lower case letters (e.g., N2a) refer to sub-types. This number will be associated with specification drawings and can also be attached to the sign itself to aid in identification. Fig. 13. Owner #2’s Sign Standard and Guidelines.
Fig. 14. Query settings for checking the naming convention of space numbers in Project #2. This query basically is a translation of the Owner #2’s requirement about the space signs. In this way, the assessors are able to check the well-formedness of the spaces numbers in a model.
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
201
is again based on querying simpler queries and evaluating them visually.
with confusing abbreviations) by running specific queries within the model as the benchmark in a fully automated process.
6.6.1. IQA Test 6: Well-formedness of space attributes in project #2 Many owner organizations have specific rules for their space attributes. These are usually indicated in extra guidelines such as – in this example – Owner #2’s Sign Standard and Guidelines. In this example, we assess whether the space numbers in the architectural model of the Project #2 follow the fairly complicated signing rule of Owner #2 (Fig. 13). For performing this assessment, we utilized the Solibri Model Checker to create complex queries. Fig. 14 shows the translation of the Owner #2’s rules into the Solibri queries and Fig. 15 shows an example of one of the spaces found in the Project #2 architectural model which do not match with the Owner #2’s rules for space names.
6.7.1. IQA Test 7: Understandability of assets in project #3 mechanical room Fig. 16 shows the results of querying generic objects in the mechanical BIM of Project #3 in the Solibri Model Checker. A closer look at some of these objects which are actually located in the mechanical room of Project #3, shows that all of them have the generic shape of boxes and meant to be placeholders (Fig. 17). In a further investigation, we can see in Fig. 18(b) the same blue box at the top left of in Fig. 17. Reviewing its properties gives the user a hint about the identity of this equipment under the attribute ‘‘type” with the value ‘‘MCC.” A user who is not familiar with this expression wouldn’t be able to understand what this blue box is representing. However, it is possible to go to the already built mechanical room and find out what is represented in the model. After a walkthrough in the mechanical room, we could identify that the blue box is a placeholder for the MCC (motor control center) in the Project #3 (Fig. 18(a)). The examples in this section show how we can systematically assess BIMs for different IQ issues related to the FM using the introduced framework in Section 5. The used BIMs in the examples above are from projects with different size, complexity and level of detail which demonstrate the feasibility and adaptability of our framework in practice. Moreover, the tests are described in a way that they are adaptable to be performed with different tools for querying BIMs. So the framework is project and tool independent.
6.7. Understandability of assets As mentioned in Section 5, in order to assess the understandability of model components, we need to investigate generic objects or objects with unspecified shapes like boxes and spheres as placeholders. Understandability is an IQ dimension on the user level as explained in [15]. Therefore, it is highly dependent on the user’s domain knowledge. For instance, not all model users are familiar with abbreviations such as MCC (Motor Control Center), CHWS (Chilled Water Supply), etc. For this type of assessment, we need to check the existence of the generic objects (or objects
Fig. 15. An example for checking the well-formedness of space numbers in the Project #2 which shows that the space number 10100 does not follow the owner’s requirements stated in 19.
202
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
Fig. 16. Understandability issue in the mechanical BIM of Project #3.
Fig. 17. Blue box as a placeholder in the mechanical BIM of Project #3.
7. Conclusion The rapid increase of using BIM in the AECO industry creates many new opportunities for practitioners to exchange and use data. High quality BIMs can especially create great potentials for downstream information users, such as FM practitioners. However, assessing the quality and usefulness of delivered BIMs to the owners is a significant challenge and requires further research. In order to address this gap and support the ultimate goal of establishing
approaches for model-based project delivery, a new framework for understanding and assessing the information quality of BIMs from the FM perspective is presented this work. This framework is developed based on the information needs of the owners in our case studies and provides actual examples about IQ issues and their assessment. The framework is organized according to the main model characteristics: objects, attributes, relationships, and spatial information. It also considers the major FM terms for operation and maintenance, namely: assets, spaces, and MEPF systems. Furthermore, we identified five relevant IQ dimensions for FM and discussed the assessment of these IQ dimensions with specific examples from our case studies. As demonstrated in Section 3, IQ dimensions, such as accuracy or completeness, can be interpreted in different ways by the researcher. This may lead to ambiguities and difficulties in mutual understanding among researchers and practitioners. To avoid such issues, we discussed each relevant IQ dimesion and provided clear definitions for each of them. We also emphasized that the relevant IQ dimensions might vary from project to project and it is up to the assessors to define and identify them in each case. However, the presented approach for developing the IQA framework remains the same. In our proposed approach, the assessor needs to complete three steps for each IQ dimension and each subject of assessment: 1. Identify the relevant proxy indicators 2. Determine the benchmarks 3. Choose how to perform the IQA tests
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
(a) As-is
203
(b) BIM Fig. 18. Investigating the model placeholders in the actual mechanical room of Project #3.
Finally, we provided a series of actual tests from our case study projects to show how to operationalize the IQA framework in practice and to demonstrate the feasibility and practicality. The examples also show that performing IQA tests is an application independent approach and the tests can be performed with any BIM authoring tool that provides basic filtering features. However, using specific software applications for querying BIMs can provide more flexibility and save time in performing the IQA tests. Using the IQA framework, the owners will not only be able to perform IQ controls through the course of a project, but based on the structure of this framework they are able now to formalize better their information needs for the operation and maintenance phase. In other words, by using this framework the owners can develop ‘‘quality assessment strategies” for their projects. They can develop their information requirements based on the framework’s structure and perform IQA tests throughout different project phases which ultimately can help to have higher model quality in delivered BIMs at the time of project handover. In this way, our framework contributes to establishing model-based project delivery approaches. As a future work we envision to work on a series of tests to demonstrate the efficiency of IQA tests and their impact on the projects. This can be done through for example performing systematic IQA tests in parts of a project and compare its efficiency with a control group. For instance in a large project, it is common to have several design teams. In such a case study project, it is conceivable to establish IQA tests for a specific design team and select another design team with similar tasks as the control group. Then, through observing the amount of model corrections and growth, and through observing the volume of digital communications related to the models within the groups and between the designers and builders, we will be able to get a better understanding about the impact of establishing IQA tests on a project. The findings of this investigation will be published in a separate work. Furthermore, we aim to perform a cross-project analysis on our case study BIMs and provide a comparison between different projects considering each IQ dimension. Appendix A. List of interviewees List of the interviewees at the partner organizations for studying the current FM practices and identification of organizational information needs from FM perspective. More details are provided in [37]. Developing Owner Requirements for BIM-enabled Project Delivery and Asset Lifecycle Management). Owner Organization #1: 1. 2. 3. 4. 5.
Assistant Deputy Minister of Project Services Branch Executive Director of Project Services Branch Head, Procurement Standards at Project Services Branch Executive Director of Project Delivery Branch Project Director of Project Delivery Branch
6. Senior Project Manager at Project Delivery Branch 7. Director of Project Portfolio Management 8. Director of Facility Evaluation at Technical Services Branch 9. Facility Design Coordinator at Technical Services Branch 10. Supervisor-CAD/GIS and Records at Technical Services Branch 11. CAD/GIS Technologist at Technical Services Branch 12. Building Security Systems Engineer at Technical Services Branch 13. Assistant Deputy Minister at Regional Operation Branch 14. Director of Regional Operation Branch 15. Facilities Manager at Regional Operation Branch 16. Operations Supervisor at Regional Operation Branch 17. Director of Capital Development 18. Logistics Coordinator Owner Organization #2: 1. Facility Information Systems (FIS) manager, who is responsible for data collection and maintenance of the facility space information; 2. Manager of Space Inventory, who is also responsible for the information maintenance; 3. Facility manager of the Brown Zone as an information consumer who manages the operation and maintenance activities and building managers in that zone; 4. Building administrator of one of the case examples (Project #3) as an information consumer; 5. Technical specialist for the Building Management Services (BMS), who is explicitly assigned to work at Project #3 and is another information consumer; 6. Records Retrieval System Administrator 7. Manager of Technical Services 8. Main7tenance & Renewal Senior Analyst 9. Maintenance Technical Specialist 10. Head Maintenance Engineer 11. Millwright 12. Building Operation Program Manager 13. Head Maintenance Engineer (BMS Center) Appendix B. Examples of IQ requirements in general standards and guidelines 1. ‘‘BIM models are validated against the standards established in the BIM Guide Series to ensure accuracy and completeness of the Model.” (GSA BIM Guide Series 008: 2.2.7 Monitoring Compliance and Submittals [9]) 2. ‘‘Each BIM equipment object in the As-Built BIM shall contain geometric data and a minimum set of attributes:. . .” (GSA BIM Guide Series 008: 1.3.1 Accurate As-Built Geometry and Spatial Program BIM [9])
204
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
3. ‘‘Example asset information requirements:” (BSI - PAS 1192-3: A.3 Specific asset information requirements [8]) (a) Legal information (b) Commercial information (c) Financial information (d) Technical information (e) Managerial information 4. ‘‘Model objects shall contain IFC parameters and associated data applicable to building system requirements.” (LACCD Building Information Modeling Standards: 2.4 Modeling Requirements [10]) 5. ‘‘Each space shall include the following spatial information:” (LACCD Building Information Modeling Standards: 2.4.6 Program Spatial Requirements [10]) (a) Space type – Omniclass (Table 13) (b) Space number – Omniclass (Table 14) (c) Space name (d) Space description (e) Department (f) Program 6. ‘‘The format of each information exchange shall be defined by the organization that has defined the OIR (organizational information requirements) and the AIR (asset information requirements).” (BSI, PAS 1192-3: 2014, p.10 [8]) 7. ‘‘Models should include all appropriate dimensioning as needed for design intent, analysis, and construction. Level of detail and included model elements are provided in the Information Exchange Worksheet.” (BIM project execution planning guide v2.0: SECTION J: QUALITY CONTROL [47]) 8. ‘‘ensure that the Project Facility Data set has no undefined, incorrectly defined, incorrectly named or duplicated elements and the reporting process on non-compliant elements and corrective action plans.‘‘ (USC BIM Guidelines: SECTION F: BIM AND DATA QUALITY CONTROL [26])
References [1] SmartMarket Report, SmartMarket Report: The Business Value of BIM for Owners, McGraw Hill Construction, 2014.. [2] O. Berard, A framework for assessing design information quality from the builder’s perspective, in: Building Information Modeling for Managing Design and Construction: Assessing Design Information Quality, Technical University of Denmark, 2012. [3] J. Du, R. Liu, R. Issa, BIM cloud score: benchmarking BIM performance, J. Construct. Eng. Manage. 140 (11) (2014) 04014054. http://dx.doi.org/10.1061/ (ASCE)CO.1943-7862.0000891. [4] W. Solihin, C. Eastman, Y.-C. Lee, Toward robust and quantifiable automated IFC quality validation, Adv. Eng. Inform. 29 (3) (2015) 739–756, ISSN: 14740346 http://linkinghub.elsevier.com/retrieve/pii/S1474034615000725. [5] E. Tribelsky, R. Sacks, Measuring information flow in the detailed design of construction projects, Res. Eng. Des. 21 (3) (2010) 189–206. [6] B. Foster, BIM for facility management: design for maintenance strategy, J. Build. Inform. Model. (2011) 18–19. https://www.brikbase.org/sites/ default/files/Pages/%20from%20jbim_spring11-2.foster.pdf. [7] Michael P. Gallaher, Alan C. O’Connor, John L. Dettbarn, Linda T. Gilday, Cost Analysis of Inadequate Interoperability in the U.S. Capital Facilities Industry, 2004. . [8] BSI, Specification for information management for the operational phase of assets using building information modelling, PAS 1192-3:2014, The British Standards Institution, ISBN 978 0 580 83910 8, 2014. . [9] GSA, GSA Building Information Modeling Guide Series: 08 – GSA BIM Guide for Facility Management, 2011. . [10] LACCD BIMS, LACCD Building Information Modeling Standards, 2010. . [11] SBCA, Singapore BIM Guide Version 2, 2013. . [12] G. Lee, Y.H. Park, S. Ham, Extended process to product modeling (xPPM) for integrated and seamless IDM and MVD development, Adv. Eng. Inform. 27 (4) (2013) 636–651. [13] J. Maxwell, Understanding and validity in qualitative research, Harvard Educ. Rev. 62 (3) (1992) 279–301. [14] S. Thomson, Qualitative research: validity, Joaag 6 (1) (2011) 77–82.
[15] P. Zadeh, S. Staub-French, R. Pottinger, Review of BIM quality assessment approaches for facility management, in: ICSC15: The Canadian Society for Civil Engineering 5th International/11th Construction Specialty Conference, University of British Columbia, Vancouver, Canada, 2015, pp. 1887–1896. https://circle.ubc.ca/bitstream/handle/2429/53726/Zadeh_P_et_al_ICSC15_ 342_Review_Of_Bim.pdf?sequence=1. [16] P. Zadeh, H. Burak Cavka, S. Staub-French, BIM Information Quality Analysis for Space Management, Osaka University, Osaka, Japan, 2016. [17] C.M. Eastman, P. Teicholz, R. Sacks, K. Liston, BIM Handbook: A Guide to Building Information Modeling for Owners, Managers, Designers, Engineers and Contractors, second ed., John Wiley and Sons, 2011, ISBN: 9781118021699 https://books.google.ca/books?id=aCi7Ozwkoj0C. [18] A. Akcamete, B. Akinci, J.H. Garrett, Potential utilization of building information models for planning maintenance activities, in: Proceedings of the International Conference on Computing in Civil and Building Engineering, 2010, pp. 151–158. [19] L. Sabol, Building information modeling & facility management, in: IFMA World Workplace, Design + Construction Strategies, LLC\, 2008. [20] C. Kasprzak, A. Ramesh, C. Dubler, Developing standards to assess the quality of BIM criteria for facilities management, in: AEI 2013 – sBuilding Solutions for Architectural Engineering, ASCE, 2013, pp. 680–690. http://ascelibrary. org/doi/abs/10.1061/9780784412909.067. [21] National Institute of Building Sciences, National Building Information Modeling Standard: Transforming the Building Supply Chain Through Open and Interoperable Information Exchanges, 2007.. [22] B. Becerik-Gerber, F. Jazizadeh, N. Li, G. Calis, Application areas and data requirements for BIM-enabled facilities management, J. Construct. Eng. Manage. 138 (3) (2011) 431–442. [23] BIM Task Group, COBie Data Drops, 2012. . [24] H. Kulusjärvi, COBIM: Common BIM Requirements – Series 6 Quality assurance, 2012. . [25] P. Teicholz, BIM for Facility Managers, Wiley, ISBN 9781118417621, 2013. . [26] USC, Building Information Modeling (BIM) Guidelines, 2012. . [27] BuildingSMART, Specifications Welcome to buildingSMART-Tech.org, 2017. . [28] R.Y. Wang, D.M. Strong, Beyond accuracy: what data quality means to data consumers, J. Manage. Inform. Syst. 12 (4) (1996) 5–33, ISSN: 07421222 http:// www.jstor.org/stable/40398176. [29] Y.W. Lee, D.M. Strong, B.K. Kahn, R.Y. Wang, AIMQ: a methodology for information quality assessment, Inform. Manage. 40 (2) (2002) 133–146, ISSN: 0378-7206 http://www.sciencedirect.com/science/article/pii/ S0378720602000435. [30] B. Hicks, Lean information management: understanding and eliminating waste, Int. J. Inform. Manage. 27 (4) (2007) 233–249. [31] M. Guttman, The information content of BIM: an information theory analysis of building information model (BIM) content, Res. J. 03(02) (2011) 29. [32] T. Jylhä, M.E. Suvanto, Impacts of poor quality of information in the facility management field, Facilities 33 (5/6) (2015) 302–319. http://dx.doi.org/10. 1108/F-07-2013-0057. [33] P. Tang, E. Anil, B. Akinci, D. Huber, Efficient and effective quality assessment of As-Is building information models and 3D laser-scanned data, in: Proceedings of ASCE International Workshop on Computing in Civil Engineering, McGrawHill, Miami, FL, 2011. [34] T.C. Redman, Data Quality for the Information Age, Artech House, Boston, 1996, ISBN: 978-0-89006-883-0. [35] C. Eastman, Y. Jeong, R. Sacks, I. Kaner, Exchange model and exchange object concepts for implementation of national BIM standards, J. Comput. Civil Eng. 24 (1) (2010) 25–34, ISSN: 0887-3801/2010/1-25-34/$25.00 http://dx.doi.org/ 10.1061/(ASCE)0887-3801(2010)24:1(25). [36] A. Assaf, A. Senart, Data quality principles in the semantic web, in: 2012 IEEE Sixth International Conference on Semantic Computing (ICSC), IEEE, 2012, pp. 226–229. [37] H.B. Cavka, S. Staub-French, E.A. Poirier, Developing Owner Information Requirements for BIM-enabled Project Delivery and Asset Lifecycle Management, 2017, in press. [38] H.B. Cavka, S. Staub-French, R. Pottinger, Evaluating the alignment of organizational and project contexts for BIM adoption: a case study of a large owner organization, Buildings 5 (4) (2015) 1265–1300. [39] D.P. Ballou, H.L. Pazer, Modeling data and process quality in multi-input, multi-output information systems, Manage. Sci. 31 (2) (1985) 150–162, ISSN: 0025-1909 1526-5501 http://pubsonline.informs.org/doi/abs/10.1287/mnsc. 31.2.150. [40] Y. Wand, R.Y. Wang, Anchoring data quality dimensions in ontological foundations, Commun. ACM 39 (11) (1996) 86–95. http://dl.acm.org/ citation.cfm?id=240479. [41] T.C. Redman, C. Fox, A. Levitin, Data and data quality, in: Understanding Information Retrieval Systems: Management, Types, and Standards, Taylor & Francis, 2011, p. 269, ISBN: 978-1-4398-9196-4. [42] F.H.C. Marriott, A Dictionary of Statistical Terms, Longman [u.a.], Harlow, 1991, 5th ed., reprint ed. ISBN: 978-0-582-01905-8 978-0-47021349-0.
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205 [43] P. Cykana, A. Paul, M. Stern, DoD guidelines on data quality management, in: Proceedings of the Conference on Information Quality, Cambridge, 1996, pp. 154–171. [44] T. Redman, Data Quality: The Field Guide, Data management series, Digital Press, 2001, ISBN: 978-1-55558-251-7 https://books.google.ca/books?id= iE2qsxDyEEYC. [45] M.J. Pratt, Extension of ISO 10303, the STEP standard, for the exchange of procedural shape models, in: Proceedings of Shape Modeling Applications, 2004, 2004, pp. 317–326.
205
[46] BSI, New BIM standard reduces asset management costs even further j BSI Group, 2014.. [47] J. Messner, C. Anumba, C. Dubler, S. Goddman, C. Kasprzak, R. Kreider, R. Leicht, C. Saluja, N. Zikic, BIM Project Execution Planning Guide, Tech. Rep., 2010.
Contents lists available at ScienceDirect
Advanced Engineering Informatics journal homepage: www.elsevier.com/locate/aei
Information Quality Assessment for Facility Management Puyan A. Zadeh a,⇑, Guan Wang a, Hasan B. Cavka a, Sheryl Staub-French a, Rachel Pottinger b a b
Department of Civil Engineering, University of British Columbia (UBC), Vancouver, Canada Department of Computer Science, University of British Columbia (UBC), Vancouver, Canada
a r t i c l e
i n f o
Article history: Received 14 December 2016 Received in revised form 13 June 2017 Accepted 14 June 2017
Keywords: Information Quality Building Information Modeling (BIM) Facility Management (FM) BIM for FM Model-based Project Delivery
a b s t r a c t Assessing the quality of building information models (BIMs) is an important yet challenging task within the construction industry as projects are increasingly being delivered with BIM. This is particularly essential for facility management (FM) users as downstream information consumers that depend on the quality of models developed in the previous project phases. The research presented in this paper addresses this challenge by introducing a framework for information quality assessment (IQA) of BIMs for FM uses. The IQA framework is the outcome of an extensive study of two large owner organizations involving numerous BIM projects. The framework is structured based on the essential FM subjects: assets, spaces, and systems, and the model characteristics: objects, attributes, relationships, and spatial information. The framework is then operationalized through the development and evaluation of information quality (IQ) tests using BIM model checking tools across three projects with different levels of detail and complexity. The proposed IQA framework and associated tests advance the state of knowledge about BIM quality in terms of methods to represent and evaluate conformance to owner requirements. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction With the growing adoption of BIM (building information modeling) within the AECOO (architecture, engineering, construction, owner and operator) industry, owner organizations are increasingly requiring BIM as part of the project delivery process and exploring how BIM can be leveraged for facility management (FM) purposes [1]. Research shows that a high number of private and public owners believe in the importance of developing capabilities in their organizations to leverage BIM for the operation phase [1]. Owners believe that a key benefit in using BIM for operation and maintenance comes from the complete and accurate information provided by the delivered models [1]. However, several studies have identified the lack of information quality (IQ) as a major barrier for this aim [2–5]. Specifically, researchers confirm that poor IQ of delivered information causes significant costs and rework for the operations phase [6,7]. Therefore, it is critical for stakeholders in the AECOO industry to be able to assess the quality of BIMs at different stages throughout project delivery and at handover to ensure the usefulness of building information for
⇑ Corresponding author. E-mail address: [email protected] (P.A. Zadeh). http://dx.doi.org/10.1016/j.aei.2017.06.003 1474-0346/Ó 2017 Elsevier Ltd. All rights reserved.
operation and maintenance purposes. This requires clear, structured, and flexible methods for describing and assessing the quality of delivered models in terms of conformance to owner requirements. IQ is described and interpreted in different ways by researchers and owner organizations. The proposed approaches in related literature mainly focus on assuring the quality of BIMs during the modeling phase. For instance important organizations such as BSI [8] GSA [9] LACCD BIMS [10] SBCA [11] provide measures for modelers to avoid quality related issues in their modeling process without proposing specific quality assessment methods [8–11]. Other research works, such as Tribelsky and Sacks [5], have their focus on the data exchange between different models and propose approaches to assess the quality loss in such exchanges [5]. Furthermore, another research stream aims to develop and improve evaluation methods focusing on the quality of model conformance to industry standards such as conformance of Industry Foundation Class (IFC) outputs [4] and Model View Definition (MVD) [12]. Although these approaches provide an important step forward, these works are limited to generic checks offered by common BIM authoring tools that help modelers avoid different IQ issues. Thus, additional research is needed to better understand how to
182
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
characterize the quality of BIMs and evaluate their conformance to owner-specific requirements. The main objective of this research is to address this research gap by providing a structured framework for information quality assessment (IQA) of BIMs for facility management purposes. This framework was developed based on an extensive study of two large owner organizations involving a series of BIM-based projects in which we were able to interview the stakeholders and observe their operation and maintenance processes. The specific research questions pursued include the following: 1. What are the information needs of owner organizations for creating intelligent FM systems? 2. What are the relevant IQ dimensions and related characteristics required to systematically understand and assess the models? 3. How can IQ tests be operationalized to evaluate the conformance of a given BIM for owner-specific information requirements? In response to these questions, we developed an IQA framework based on the identified owner information needs. The framework allows users to systematically characterize the information quality dimensions that are relevant for a particular owner and assess the IQ of BIMs at different project stages with respect to the owner’s FM requirements. The structure of this framework is organized based on four different model characteristics: entities, entity attributes, the relationships between entities, and the spatial information (location and shape) of each entity. The structure of the framework also considers the three essential FM terms: assets (equipment), spaces, and MEPF (mechanical, electrical, plumbing, and fire safety) systems. The model characteristics and FM terms describe the subject of each required IQA test in the framework. Moreover, the framework indicates for each IQA test, the required proxy indicators and benchmarks, and it proposes relevant methods to perform the IQA tests. Using this framework, we operationalized the specific IQA tests for three different projects with different size, complexity and level of detail to show the feasibility and adaptability of the introduced framework in practice. Using the introduced framework in this research is grounded in firsthand observations in actual projects and provides the owners and stakeholders the awareness about the IQ issues and aims to encourage them to support the overall goal of modelbased project delivery. The implementation of IQA tests on exam-
ples from the practice is a proof of feasibility of establishing structured quality control strategies in construction projects. Furthermore, the variety of the practical examples introduced in this research aims to showcase the comprehensibility of IQA tests to cover different quality issue types in a BIM. The framework’s feasibility and comprehensibility in this quality research follow the interpretive and theoretical validity concept introduced in [13,14]. In the next section, we provide examples of representative quality issues in delivered BIMs based on the BIM projects we analyzed. In Section 3, we discuss the research background and related works that includes studies from computer sciences (CS) and the AECOO domain. Then in Section 4, we introduce our case studies and the different steps in our methodology to develop the proposed IQA framework. In Section 5, we provide a detailed explanation of our IQA framework. Then in Section 6, we describe how to operationalize IQ tests from the framework based on selected examples from our case study projects. Finally, Section 7 provides some concluding remarks. 2. Practical motivation regarding current quality issues of BIMs for FM The motivation of this research has its roots in studying the deliverables of several BIM projects and interviewing numerous FM personnel within two different owner organizations. The provided examples in this section are drawn from what has been observed in those projects and cover all typical quality issues of BIMs for FM. In this regard, we especially focused on the identification of obstacles in establishing methods for model-based analysis and challenges in utilizing delivered BIMs in the operations phase of a building. Analyzing a diverse range of BIM projects through the lens of building operations has highlighted that the quality of BIMs often does not satisfy the expected level of quality for FM purposes, which in turn causes issues for BIMs being useful for operation and maintenance purposes. These IQ issues could be observed across different project phases up to and including project handover. For a better understanding of the various types of issues, the following figures provide typical examples of IQ issues from our case studies and highlight the specific information quality dimensions that are exemplified in each example. 2.1. Example 1
? (a) As-is
(b) BIM
Example 1. This example from the Project #3 shows that the large white expansion tank in the as-is photo (left) is missing in the mechanical BIM (right). Therefore, the model has an incomplete representation of the as-is.
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
183
2.2. Example 2
(a) Assets as a part of a system
(b) Assets without a system assignment
Example 2. These two snapshots demonstrate the assets which are assigned to a building system (left) and assets without any system assignment (right) from Project #1, i.e., the relevant MEPF system attribute for these assets is empty. However, in the reality, these assets are in fact parts of different systems. The right snapshot includes important assets such as AHUs and some heat pumps that are not assigned to a any MEPF system. Thus, the model contains incompleteness regarding to the system definitions.
2.3. Example 3
Example 3. In this example from Project #3, the attribute ‘‘Manufacturer” has a different value in BIM (right) in comparison with the actual work sheet on the actual heat pump (left). Therefore, the model represents the as-is situation inaccurately.
2.4. Example 4
Example 4. This example from Project #2 shows the redundancy in naming spaces within a project which contradicts the owner’s uniqueness policy for naming spaces. We discuss this example in Section 6 in more detail.
184
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
2.5. Example 5
Sign Standards and Guidelines 3.1
Sign Classification for Interior Signs
Signs are given a sign code based on the following classification system. Each installed sign will have a unique number attached to its sign code. For Interior signs, the sign code includes the building code. Lower case letters (e.g., N2a) refer to sub-types. This number will be associated with specification drawings and can also be attached to the sign itself to aid in identification. Example 5. This is an example of the space naming requirements for buildings from Owner #2’s projects. According to this requirement, the space names must follow a specific format and must be unique within the entire Owner #2’s buildings. The challenge here is to find a systematic way to evaluate the well-formedness and redundancy of these names.
2.6. Example 6
(a) As-is
(b) BIM
Example 6. This example shows a blue box with the object type ‘‘Generic” in the BIM model (right) which is per se not understandable what it represents. After a walkthrough in the mechanical room of Project #3, we could identify that the blue box is a placeholder for the MCC (Motor Control Center) (left). This example, highlights the lack of understandability of some model parts. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
According to our observations, the examples above are representative IQ issues that often occur in common BIM projects. They show that some IQ issues might be related to deficiencies in single information pieces (Examples 1 to 3), some are related to the entire model and its compliance to specific rules (Example 4 and 5), and some others are about how the information is represented by the model to the user (Example 6). We discuss these three different levels (entity, model, and user) in a separate work in more detail [15]. The examples above cover the different IQ issue types that we have been observing in different projects from the FM perspectives. These IQ issue types can be generally divided into following five categories: 1. 2. 3. 4. 5.
Incompleteness (Example 1 and 2) Inaccuracy (Example 3) Redundancy (Example 4) Well-formedness (Example 5) Understandability (Example 6)
In addition to the examples above, we also identified issues related to the ‘‘level of detail” (LOD) of the model or the ‘‘level of representation” of the reality. However, LOD issues can be understood as model issues in representing the reality in a sufficient (complete) and accurate way based on the user’s requirements. For instance, when the model only includes a generic geometry, like a box as a placeholder, to represent an air handling unit
(AHU), this might satisfy the architect’s needs but is likely not sufficient to meet the needs of the FM users. In this example, on the one hand the AHU representation is ‘‘incomplete” and does not include important parts, such as filters, heating/cooling components, blower, etc., but on the other hand, the AHU representation is also ‘‘not accurate” enough in showing its orientation, connections to other equipment, and the available free space for maintenance activities. The white box in Example 6 is an example for poor LOD representation of an AHU, whereas the turquoise expansion tank provides a sufficient representation. Hence, the LOD issues can be broken down into two fundamental issues that are covered in our examples above, which are the issues related to the incomplete representation of the required information and the issues related to the inaccurate representation of the required information. Furthermore, reviewing these examples shows that such errors can be caused at different project stages from the design up to project delivery. Generally speaking, any interactions with the model might be a source of IQ problems, including activities related to modeling, data exchange between different project participants, and the data consumption as discussed in [4]. However, the modeling procedure has the most impact on assuring the IQ of models, which is why many organizations such as BSI (British Standards Institution), GSA (General Services Administration), LACCD (Los Angeles Community College District), and SBCA (Singapore Building and Construction Authority) provide guidelines for modelers
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
to assure the quality of BIMs [8–11]. Particularly from the FM perspective, developing and maintaining the models are important. FM users, as the downstream information consumers, are very dependent on the quality of modeling in the previous phases. As shown in the examples above, it is critical for owners to be able to assess the representation of FM information needs during the modeling process to ensure that these significant IQ issues are avoided. The following sections provides the necessary background on the topic of IQ from the perspectives of the AECOO industry and relevant works in computer science (CS). 3. Research background The structured IQA approach which is introduced in this research is developed based on the related research works in the areas of building information modeling and computer science. In this section, we introduce the research background which is accordingly divided into the role of BIM and its quality for FM as well as the fundamental points in the IQ discussion from the computer science perspective. 3.1. BIM quality BIM can be described as an environment for collecting and maintaining all building-related information which is critical for collaboration and information exchange between different disciplines and across different project phases [16]. Detailed explanations about the BIM principles are discussed in [17]. As mentioned in the introduction, the role and significance of BIM for different disciplines is growing [1]. In particular, BIM has a great potential for being used as the source of information for setting up CMMS (Computerized Maintenance Management System) and as a tool to track and store changes happening to the as-is situation of facilities [18]. Therefore, BIM has the potential to fundamentally change the project delivery process [19]. Nevertheless, research shows that BIM has not been successfully used in a broader spectrum for the facility maintenance and operation purposes [1]. One explanation for this outcome is because of the difficulties that the owners have in assessing the quality of the delivered models. As a consequence, in many projects, owners do not include FM information needs in the contracts and project requirements. In other words, one can’t ask for something that one can’t assess. There are a few research efforts that specifically focus on assessing the IQ of models in the AECOO domain. Berard in [2] presents a framework for assessing design information quality from the builder’s perspective. Although this is different from the FM perspective, it provides a useful approach to develop a suitable framework for FM, since the presented framework in [2] corresponds with the related works in CS and is structured through identification of relevant IQ dimensions. Kasprzak et al. discussed in [20] a method for developing a standard procedure for the verification of the completeness and accuracy of modeled facility information. This work is an important point of departure for developing owner requirements tailored for facility operation and maintenance purposes which can be used as a benchmark for IQA tests. Similarly, Du et al. in [3] introduce benchmarking metrics for the IQA of BIM cloud performances that also includes IQ related metrics, such as model quality, accuracy, and usefulness. In [4] Solihin et al. introduce a testing methodology for validation of IFC files. Although this work does not have a specific FM focus, it does provide a useful model for developing structured quality tests. Furthermore, many researchers and organizations have been working on the measures and control mechanisms to assure the quality of models during the modeling process [8–10,21,11]. These research efforts provide generic guidelines (usually based on
185
checklists) without detailed and specific assessment approaches for the facility operation and maintenance. A popular approach for IQ control of BIMs for FM is to export the mechanical models to the COBie (Construction Operations Building information exchange) sheets and investigate the errors and outcomes as discussed in [22,23,20,24–26]. In this regard, there are also quality assurance approaches with MVD (Model View Definition) and IDM (Information Delivery Manual) from buildingSMART [27], which similarly focus on the information exchange between different information sources. Some assessors use such approaches to extract information from a model and check whether that model is in compliance with a specific required data structure, which is for example defined through MVD (similar approaches are observed with COBie). An overview about the different literatures from the AECOO domain on BIM-IQ is introduced in [15]. All these approaches are designed to be used by the modelers, while the owners and operators as the actual information consumers after the project delivery are still lacking a comprehensive and structured approach to assess the quality of delivered BIMs [16]. There are a variety of tools that can be used for IQ assessment of BIMs. BIM authoring tools like Autodesk Revit and BIM-based construction management tools such as Navisworks include basic features that allow users to run simple queries. The ‘‘Schedules” feature in Autodesk Revit is a great example of such tools. In addition, there are software applications such as Solibri Model Checker (SMC) or iTWO that are capable to create complicated queries and extract required information from a BIM. 3.2. IQ in CS In contrast to the AECOO domain, the subject of IQ has been discussed in computer science research for more than two decades. Generally speaking, there are two major fundamental research streams in the CS domain; one has the main focus on the usefulness of information for the information consumers, and the other one is based on how the information is modeled in an information system, which are both introduced below. The first stream is based on the original work of Wang and Strong [28], which proposed a framework for different IQ dimensions from the information consumer’s perspective [28]. According to Wang and Strong [28], the ‘‘fitness for use” (or the information usefulness) is the major criterion for IQ; a piece of information is of high quality when it is ‘‘useful” for its consumer [28]. Therefore, the identification of consumer’s information needs is an essential step in assessing the quality of information systems like BIMs. In later research [29], the same research group implemented the proposed IQ framework in [28], and applied it on different case studies from healthcare, finance, and manufacturing. Describing IQ from this perspective is also suitable for applications in domains such as the AECOO. For instance, the author in [30] states that in an information system the information and the functionality for managing the information should supply ‘‘value” to the information consumer. ‘‘Value” can be understood as a function of information quality, which is defined in terms of its usefulness in making decisions [31]. Nevertheless, there is still a great need for studies which have a specific focus on IQ in the AECOO domain and transfer the available knowledge from CS into this domain. Wang and Strong and later Lee et al. identified a large number of IQ dimensions from the academics’ and practitioners’ perspective that they categorized in four essential categories [28,29]: 1. Intrinsic with dimensions such as accuracy (including precision, correctness), consistency, unambiguity, reliability, and believability. 2. Contextual with dimensions such as completeness, quantity, timeliness, and level of detail.
186
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
3. Representational with dimensions such as understandability, readability, interpretability, and meaningfulness. 4. Accessibility with dimensions such as accessibility, and availability. These IQ dimensions might overlap and it is up to the assessors to define a clear distinction between them. This categorization is an important point of departure for structuring framework for new research domains as for AECOO. Depending on the domain, the significance of the dimensions above might change. In our studies, we have observed that researchers in the AECOO domain frequently describe their dimensions differently. For instance, while Solihin et al. and Kasprzak et al. refer to IQ issues with the missing objects as information ‘‘Correctness” [4,20] Berard uses the term ‘‘Correction” [2], Jylhä and Suvanto see it as a part of information ‘‘Availability” [32] and Tang et al. use the term ‘‘Com prehensiveness” [33]. Thus, developing similar categorization as in [28,29] is an important contribution in terms of formalizing the IQ discussions within the AECOO domain. The second research stream for describing IQ dimensions has its roots in the essential work of Redman [34]. According to [34], a model-based information system can be described through three basic characteristics: entity, attributes of the entities, and the values of the attributes. In recent years, the theory of models is integrated (separately from the IQ discussion) into the AECOO domain, and several researchers have been working on additional elements that are specific for this domain. A precedent setting work is presented in [35] where the authors describe five basic characteristics for describing a BIM: 1. ‘‘Functional type” is the equivalent to the entity which is not only covering the building objects but also views, schedules, plans, etc. 2. ‘‘Geometry” is basically the shape of the object. 3. ‘‘Attributes”. 4. ‘‘Relations” between objects. This was originally covered in [34] through special attributes. 5. ‘‘Behavioral rules” such as: a window must be always attached to a wall. One noticeable point that is missing in this categorization of model characteristics of a BIM is the object’s location which is an essential term in three-dimensional models (as information systems). Therefore, we adjust the ‘‘Geometry” with the ‘‘Spatial information” to represent both the object’s geometry and the location. Thus, the basic model characteristics for a BIM can be categorized as follows: 1. 2. 3. 4. 5.
Entity types Spatial information Entity attributes Relations Behavioral rules
The current studies in CS domain about the IQ are mainly based on the two main streams above, i.e. the works of Wang and Strong [28] and Redman [34]. However, there are still research works with interesting perspectives that could be useful for applications in the AECOO domain as well. One of these works is presented in [36] where the authors propose an IQ framework for the entire information system rather than focusing only on single information sources. Considering the heterogeneous structure of information systems in construction projects, having a BIM as the only information source is unrealistic. Therefore, such viewpoint is helpful when the IQ within the entire project should be assessed.
Reviewing the related literature highlights that IQ is a relatively new subject in the AECOO domain and there is still a need for further research in order to establish common ground for mutual understanding and knowledge exchange within the domain. Furthermore, it is also important to recognize that there are fundamental research works in CS that can be transferred into the AECOO domain to shape the IQ discussion in a structured way that is in alliance with the similar discussions in other research domains. This review also shows that each discipline – including FM – has different information requirements and as such, the IQ dimensions need to be described for each discipline accordingly. 4. Methodology The methodology in this research includes several data collection and analytical steps based on our case study projects to investigate the following research questions: 1. What are the information needs of owner organizations for creating intelligent FM systems? 2. What are the relevant IQ dimensions and related characteristics required to systematically understand and assess the models? 3. How can IQ tests be operationalized to evaluate the conformance of a given BIM for owner-specific requirements? In this research, we followed a specific roadmap in order to seek answers to the above questions. Fig. 1 illustrates the major steps of our roadmap, which is a generic approach that can be applied for different FM related purposes. This roadmap was previously introduced and applied for space management in one of our former research projects [16]. 4.1. Case study projects As mentioned earlier, this research is the outcome of studying several institutional and commercial projects for two different large public owners. However, for demonstration purposes and to provide a better overview, in this paper we explicitly focus on three projects that have different levels of complexity, occupancy, scale, and detail, demonstrating sufficient diversity in our IQA examples. The two owners we investigated are: 1. Owner #1 is a large public owner organization that is responsible for the planning, building, and managing many publicly owned infrastructure in one of the Canadian provinces. Its facilities include schools, hospitals, cultural facilities, health facilities, etc. This owner is investing C$34.8 billion in the public infrastructure over the next five years. 2. Owner #2 is a large Canadian public owner organization in the education sector that delivers and maintains educational, housing, and health care facilities. This owner is investing about C $528 million in its construction projects over the next two years. The case study projects investigated include: 1. Project #1 is a museum owned by Owner #1. This C$375.5 million, 39,000 m2, project is still ongoing and it is scheduled to be finished in 2017. Over 29 different stakeholder organizations have been involved in this large public project. As a consequence, the information exchange and collaboration have been a key point. The mechanical and architectural BIMs in this project have been developed in Autodesk Revit and attempt to provide a level of detail for fabrication (LOD 400). Since BIM has had a significant role in this project, and because our research
187
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
Operaonalize IQA Tests (sec. 6)
Formalize Informaon Needs (sec. 4.3.) Study Current Pracce (sec. 4.2.) - FM roles - Informaon systems - Informaon sources - Guidelines and standards
- FM uses - Informaon type - Informaon characteriscs - Level of detail
Determine IQ Dimensions (sec. 4.4.) - Completeness - Accuracy - Redundancy - Well-formedness - etc.
Develop a Framework for IQA (sec. 5) - Model parts to be checked - IQA indicators - Benchmarks
Fig. 1. Roadmap for development of a structured framework to design and perform IQA tests.
team has been involved in several studies related to this large project, this has been a great source for the majority of our case examples for IQA. 2. Project #2 is an educational and housing complex which was finished in fall 2016 for Owner #2. This project consists of two 18 story towers for student housing and a 4 story academic building. It is 45,108 m2 and had a construction budget of C$90 million. The architectural BIM in this project is developed in Autodesk Revit and is used for design, coordination, and fabrication purposes (LOD 400). However, the mechanical model is developed in CATIA V5 which is not a BIM native environment. Therefore, in this research work we focus on the space related BIM IQ issues from this project which are based on the architectural model. 3. Project #3 is an educational and office building for Owner #2. This 4 story building with a gross area of 5675 m2, and a total project cost of C$35 million was completed in 2011. In this project, 18 different stakeholders were involved. However, the use of BIM was limited to design purposes. Both mechanical and architectural models are developed in Autodesk Revit (LOD 300). 4.2. Study current practice The first step in our research was to study the current practices related to the facility management at the owner organizations of our case studies. For this aim, one necessary step was to perform interviews with the engaged personnel and observe the work with the currently employed information systems for FM. The interviews were performed in a semi-structured way and the interviewees included personnel who have been responsible for creating, extracting, maintaining, and consuming the FM information from the project deliverables. In this way, we were able to understand the big picture of the FM information lifecycle at the owner organizations. The interviewees included 18 different FM roles at Owner #1 and 13 stakeholders at Owner #2. A comprehensive list of the interviewees is provided in Appendix A and a detailed analysis of the outcomes is published in [37]. One of the outcomes of these interviews was the identification of the employed information systems and FM applications at the partner organizations. These include: Building and Land Information Management System (BLIMS), Facilities Capital Planning and Management (VFA Facility), Facility Maintenance System (FMS), Work Order Request Tracking System (WORTS), RAP (planning program) at Owner #1
PeopleSoft, Archibus, VFA Facility, and an inhouse developed document management system at Owner #2 A further outcome was the analysis of the relevant information sources, i.e., the type and format in which the required information is available at each organization. The identified information sources in our case studies can be categorized in following categories: 1. Mechanical and architectural BIMs (at the design, construction, and handover stage). 2. Digital documents such as 2D drawings, asset manuals, and further scanned contractual documents. 3. Paper-based documents such as part of the project’s handover drawing sets, manuals, and specifications. This investigation provides an overview about the diversity and extension of the produced information for FM in the owners’ construction projects. Based on the performed interviews and observations, it was also possible to identify the relevant guidelines and standards that the owners develop to indicate the required information for FM use in their projects. In our case studies the organization specific guidelines and standards include: 1. 2. 3. 4.
Owner #1’s Technical Guidelines Owner #1’s Basic Master Specifications (BMS) Owner #1’s Owner Statement of Requirements (SOR) Owner #2’s Technical Guidelines including guidelines for: (a) Commissioning (b) Drawing and Specifications (c) Closeout and Turnover Procedures (d) Demonstration and Training (e) Operating and Maintenance Manuals 5. Owner #2’s Sign Standards and Guidelines These guidelines and standards have been useful in setting up benchmarks to investigate whether the necessary information for FM is captured during the previous project phases. Furthermore, this analysis supported the formalization of the owner organizations’ information needs. 4.3. Formalize information needs The outcome of the previous step helped to understand how the owner organizations in their current practices determine what
188
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
information is needed to be defined or captured in the models (or other deliverables) in order to be useful for FM purposes. Based on these findings, we were able to formalize the ‘‘information needs” of the owner organizations in our case studies. At this point, we would like to emphasize that the owner’s information requirement is a very project related subject matter and this can vary from case to case. However, with formalizing the owner’s information needs at this step, we cover all typical issues that we identified in our observations and introduced in Section 2. In addition, we discuss and analyze the owner’s information requirement more extensively in a separate publication [37]. Such a formalization provides clarity on essential points relevant to the development of an IQA framework, including: What FM activities need to be supported by model-based information (information that is extracted from a BIM) delivery? These activities could include: reactive and preventive maintenance, task management, space management, service history management for assets, document management, etc. What type of assets and systems should be captured in an intelligent FM system? Obviously in most facilities, not all single equipment needs to be tracked in a computerized system and so this step will help to prioritize the different assets and systems. Which characteristics of the required assets and systems are relevant for FM? How detailed and how current (up-to-date) does the required information need to be in order to be considered useful for FM? Studying the current practices at the partner organizations in our case studies resulted that it is essential for the owners to receive models which can be used as data sources for their FM systems. It is also important for the owners that the information transition from models to FM system is combined with minimum effort. In [37], we provide more detail about the owner requirements for FM. Table 1 provides examples of required information by the owners of our case studies from the IQ perspective: Examples above and similar phrases from the identified guidelines that are followed by the owners in our case studies are indications of certain levels of IQ that the owners expect from the deliverables and so we can formalize them in more generic statements such as:
Delivered information pieces must be complete. Delivered information has to follow a specific format. Delivered information must accurately represent the as-built. Delivered information has to completely include all necessary attributes of important assets.
In addition to the owners’ specific statements about their information needs, we also studied the relevant IQ requirements in different general standards and guidelines such as in [8–11]. This additional investigation confirmed the formalized general statements above from our case studies. Examples from these standards are given in Appendix B. A detailed list of required information in our case studies is published in [38] in which the information requirements are categorized according to their FM function and specific use. 4.4. Determine IQ dimensions After studying the current FM practices at Owners #1 and #2, and identification of the information sources, relevant guidelines, and finally formalizing the information needs, we came to the conclusion that the required IQ dimensions for FM purposes are often buried in different requirement clauses which need additional interpretation for a structured and computerized assessment. According to our analysis, in our case studies the owners expect that the delivered information must be: 1. 2. 3. 4. 5.
Complete Accurate (correct and precise) Understandable Unambiguous Well-formed (in compliance with requirements)
As one might expect, these dimensions and the statement before could be relevant for many other construction projects and are not dependent on the case studies. However, through our direct observations in our case studies, we were able to analyze how each of these dimensions correspond with the identified IQ issues in the delivered models (relevant examples are given in Section 2). These IQ dimensions are in accordance with the introduced works from computer science [28,29] and our previous related works [15,16]. However, the identified IQ dimensions have differ-
Table 1 Examples of owner information requirements for FM. Source
Detail
Owner #1’s – basic master specifications, Section 15720: Mechanical identification
Required attributes for AHUs: Manufacturer, Model, Fan, Capacity L/s, S.P. kPa, Motor kW, Heating Coil, Fin Series, Type, Height mm, Width mm, Rows, Capacity W, Air Entering °C, Air Leaving °C, Air S.P. Drop kPa, Water, L/s, Water entering °C, Water Leaving °C, Water Pressure Drop kPa, Steam Cap. kg/s, Steam Pressure kPa, Cooling Coil, Fin Series, Type, Height mm, Width mm, Rows, Capacity W, Air Entering (°C DB, °C WB), Air Leaving (°C DB, °C WB), Water, L/s, Water Entering °C, Water Leaving °C, Water Pressure Drop kPa, Saturated Suction °C
Owner #2’s technical guidelines, commissioning
Verify that O&M documentation left on site is complete.
Owner #2’s closeout and turnover, Sections 01781 and 01782
Required information format/contents” as specific sections.
Owner #2’s closeout and turnover, Section 4
Required as-built information Contract Drawings Specifications Addenda Change Orders and other modifications to the Contract Reviewed shop drawings, product data, and samples Field test records Inspection certificates Manufacturer’s certificates
189
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
ent level of necessity and value for different information consumers. Hence in each project, this step should be based on the current practices at the related owner organization. 5. Framework for creating and performing BIM-IQA tests for facility management The structure of this framework is organized based on the required tests for identified relevant IQ dimensions and is broken down into following parts that we discuss in more detail in this section: 1. 2. 3. 4.
Subjects of the IQA Proxy Indicators Test Benchmarks Performance Types
In this way, the framework provides specific structure for developing and performing each required IQA test. It should be noted that the introduced IQA framework in this work is developed in accordance with the previous works of [28,29,15,16]. Particularly, the categorization of the IQ dimensions is based on the research works of Wang and Strong [28] and Lee et al. [29] that we translated for AECOO domain according to the FM needs. In the next two sections, we discuss the development of this framework in more detail and will explain the operationalization of the IQA test based on the specific examples from out case studies. Table 2 shows a summarized version of the developed framework in order to provide a better overview about the main framework elements. With this framework, the assessors are able to structure their way of thinking about required IQ of delivered models, and to systematically create and perform IQA tests. A detailed version of this framework can be found at the end of this section (Table 3). In the following, we introduce the details of the main framework elements. 5.1. IQ dimensions 5.1.1. Completeness Completeness is one of the most mentioned IQ dimensions in the literature from both CS and AECOO. Ballou and Pazer in [39] considered an information system as complete if ‘‘all values for a certain variable are recorded” in that specific information system. Wand and Wang enhance this description and define completeness as the ‘‘ability of an information system to represent every meaningful state of the represented real world system” [40]. In this definition the term ‘‘every meaningful state” has a significant role which emphasizes that not ‘‘all values” must be recorded in order to have a complete information system. Similarly, Redman describes completeness as publication of all relevant information for the information consumer [34]. Finally, Assaf and Senart define completeness as all needed information to represent all information related to a real world entity [36]. Thus, from the CS perspec-
tive a complete information system shouldn’t necessarily represent all possible information about a real entity and this system would be still complete if it includes the relevant and meaningful information which is useful for the information consumer. Hence, the information consumer and their needs have a central role in determining the completeness of an information system. In the AECOO domain, the researchers agree about the necessity of information completeness, however, they do not provide an unambiguious interpretation. Completeness has been in this domain especially discussed in connection with the information exchange between different sources or models. For instance, the authors in [4] interpret the completeness as ‘‘the minimum requirement for object count [. . .] during export or import process.” This very practical interpretation can be also understood so that a model will be complete if the number of modeled elements is equal to the number of required elements. In this way, completeness check can also be seen as an existence or quantity check for required information. Based on the descriptions above we can describe completeness for BIM as following: ‘‘A BIM is complete when all necessary model characteristics are existing in it and corresponding to relevant objects or states in the real world or requirements.” Although this IQ dimension might be easy to understand, one should be careful that the main point here is about the existence of necessary information and not their verification or correctness. In other words, assessing the information completeness is about investigating what information is in the model and what is missing with respect to the requirements.
5.1.2. Value accuracy Information accuracy is another widely required IQ dimension by the standards and guidelines in different fields and therefore, it plays a central role in the works of information technology researchers. Redman et al. in [41] describe the accuracy of a piece of information in a model as the degree of closeness of the modeled value to the corresponding correct value. The correct values (or the benchmarked values) can be determined by the requirements or the reality. Similarly, Marriott in [42] defines the information accuracy as the ‘‘closeness of computations or estimates to the exact or true values that the statistics were intended to measure.” Sometimes it would be helpful to have definitions for the opposite meaning of a complex term to separate it from the similar terms. So as for ‘‘inaccuracy”, Wand and Wang explain that inaccuracy appears when an information system represents a real-world state different from the one that should have been represented [40]. This specific point of view to the accuracy is also called ‘‘information correctness” or ‘‘information integrity.” Hence, assessing the information correctness is about checking whether the values are correct which could be seen as a true/false question. However, following the interpretation of Redman and Marriott, the information accuracy can be understood as ‘‘information precision” which is a different aspect of accuracy. Whereas for the information correctness, a value can only be either correct or incorrect (true or false), the information precision is a measurable quantity.
Table 2 Framework for creating and performing BIM-IQA tests for asset and space management purposes (summarized version). 5.1 IQ dimensions
– – – – –
Completeness Value accuracy Consistency Well-formedness Understandability
5.2 Subjects of IQA 5.2.1 Model
5.2.2 FM
– – – –
– Assets – Spaces – Systems
Entities Attributes Relations Spatial information
5.3 Proxy indicators
5.4 Benchmarks
5.5 IQA performing types
– Existence – Filter and present – Filter and compare
– Related Docs – Reality – Model
– Visual – Fully automated – Semi-automated
190
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205 Table 3 Detailed framework for creating and performing BIM-IQA tests for asset and space management purposes.
(continued on next page)
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
191
Table 3 (continued)
In general, the information accuracy has a wide meaning and can basically cover several other IQ dimensions or be a result of other dimensions. Therefore, the definitions above are representing a specific interpretation of information accuracy. A different IQ aspect that can be considered under the umbrella of accuracy is the ‘‘level of representation” which implies how detailed the information is modeled. In other words, it describes if the modeled information has the required ‘‘level of detail” (LOD) (or level of representation) to accurately represent the reality in a model [34]. If we describe information accuracy as the necessity of accurate representation of the reality, then inaccuracy could be directly related to issues with many other IQ dimensions. For instance, if a model is incomplete it is consequently not representing accurately the real word. The same interpretation applies when we discuss the information well-formedness, consistency, or understandability. More concrete examples in this regard from the AECOO domain are given in [2] which highlights the relation of accuracy to completeness by defining incorrectness as ‘‘missing objects, geometries and properties”, in [3] which connects accuracy with LOD through defining it as the ‘‘degree to which a set of inherent characteristics of BIMs fulfils desired requirements,” and in [4] where the authors emphasize the relation between information
correctness and well-formedness. Hence, to avoid confusion and in order to have a consistent discussion, we consider the information correctness as the accuracy and describe it for the context of BIM as follows: ‘‘A BIM is accurate when all required model characteristics represent correctly the relevant objects or states in the real world.” Based on this definition, assessing the information accuracy of a BIM would be about verifying values, geometries, relations, and spatial information of the objects in the model with respect to the requirements. 5.1.3. Consistency (redundancy) Information consistency ensures that each information consumer has a consistent representation of the information. Inconsistencies appear when datasets overlap and represent the same or similar concepts in a different manner, or when their specific content does not correspond [34]. According to [36], information consistency in a model deals with the question that ‘‘does the [modeled] information contradict itself?” In other words, in a consistent model there shouldn’t be any contextual conflict between model characteristics. Information inconsistency occurs when different modeled information pieces refer to the same state or com-
192
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
ponent in the reality. According to [40], this includes the physical representation and representation of values in the information system. An example for inconsistent physical representation would be duplicate objects in the model which represent the same object in the reality that also can be understood as information redundancy. This is also in accordance with the interpretation of [2] about inconsistency and with [4] when they describe the correctness. An example for inconsistent values would be attribute values which refer to the same subject such as values for space use that might be defined as ‘‘classroom,” ‘‘washroom,” or ‘‘office.” Thus, we can describe the information consistency for BIMs as follows: ‘‘A BIM is consistent in representing the real world when all required model characteristics can be followed back to the reality without any semantic or actual conflicts.” Information consistency is also known as ‘‘information unambiguity” or ‘‘information uniqueness” [40,43]. Based on our definition, assessing the information consistency of a BIM would be about identifying the redundancies and contradictions in representation of values. This ambiguity can lead to inaccurate representation of the reality or the design intent. This means ambiguity can cause information inaccuracy, which is an example for one IQ issue creates another IQ issue. Recent BIM authoring tools, such as Autodesk Revit and ArchiCAD, are software that have an object-oriented data structure and provide many intelligent features to capture issues such as redundancy of object identifiers, duplications in values, placing an object multiple times at the same spot, etc. These software characteristics can guarantee the modelers the assurance of this IQ dimension. Therefore, we wouldn’t elaborate more the redundancy dimension in the IQA discussion. However, the developers of BIM tools still need to be aware of this IQ dimension’s significance and it is highly recommended to perform suitable redundancy check. 5.1.4. Well-formedness (compliance) This IQ dimension is required by different regulations and standards such as BSI [8], GSA [9], LACCD [10], and SBCA [11]. In [8], well-formedness is defined as the information format and it requires that the owner organizations must describe exactly what information types they expect at the project delivery in the OIR (organizational information requirement) of each project. Also [11] requires to have a data exchange protocol which includes the requirements for the file formats and the naming conventions. Similar requirements are formalized by [9] as well. Different researchers refer to this IQ dimension through different terms such as ‘‘Information Typing” [36], ‘‘Data Formats” [44,29], ‘‘Well-formatted Information” [28], ‘‘Compliance” [2], and ‘‘Well-formedness” [4,16]. Nevertheless, these researchers have a common interpretation of well-formedness. So the information well-formedness in BIMs can be described as follows: ‘‘A BIM contains well-formed information when all required model characteristics are defined in compliance with relevant standards or regulations.” The relevant standards and regulations, such as IFC or OmniClass, can determine the well-formedness of the model structure and semantics, or they can determine the required format of values such as owner requirements for naming conventions of entities. If the modeled information is not in compliance with the required regulations, the model is not representing the reality or the design intent precisely. This means that the lack of information wellformedness can cause information inaccuracy too. 5.1.5. Understandability Information understandability is an abstract IQ dimension which is dependent on the information consumer’s knowledge and experience, as well as on the representation technology. Wang
and Strong describe the information understandability as the ease of understanding and interpretability of information [28]. Assaf and Senart refer to this IQ dimension as the ‘‘information comprehensibility” and require that the data concepts must be understandable to humans and convey logical meaning of the described entities and allow the easy consumption [36]. Hence, the information understandability in BIMs can be described as follows: ‘‘A BIM contains understandable information when the typical information consumer is able to connect all required model characteristics to a real object or state in the constructed building.” Issues related to the incomprehensive information can often happen in a BIM as a result of modeling objects with generic type or with unspecified shapes like boxes and spheres as placeholders. Similarly to above, the lack of information understandability in a model can lead to inaccurate representation of the reality or the design intent. Thus, incomprehensible information can be another reason for information inaccuracy. 5.2. Subjects of IQA An essential part of the framework is to determine what parts of the model should be exactly the subject of IQA tests. In other words, it is necessary to indicate to what model parts the tests should be applied. In this regard, the framework covers two different perspectives. One is model theory perspective and the other one is from the viewpoint of FM users. 5.2.1. Model theory As described in the research background section, the basic characteristics of a BIM can be categorized based on the former work of Pratt [45] and Eastman et al. [35] in five categories as shown in Fig. 2. Based on the interpretation in Fig. 2, one should imagine an ‘‘Entity” as a profile in a BIM for different terms, like building elements, systems, families, views, etc. which is describing a real object or state. This entity is the place where other object characteristics will be defined and collected. In addition, all of these characteristics can have a value which is a digital amount or a certain textual content. All characteristics have significant value for building projects. However, the entity rules are implemented in the most of the BIM authoring tools and so they can be captured through a warning analysis in such tools. For example, doors as entities have the rule to always be attached to a wall. So if the designer ignores this rule while modeling, the majority of BIM authoring tools will show a corresponding error message. Furthermore such rules mainly support the design process and can be excluded from IQA for FM.
Real Object
Composition
Model
Entity has
Attributes
Spatial Info.
Relations
Rules
have
Values Fig. 2. Characteristics for representation of a real object (or state) in a 3D Model, UML notation.
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
193
Therefore, from the model theory perspective, it is necessary to consider the following four model characteristics as the subjects of IQA tests:
cators here correspond with the studied projects in this research and the identified typical IQ issues. We have divided the proxy indicators into three generic categories:
1. Entities: Assets as the most important mechanical components 2. Attributes: Asset specs (properties) which include (but not limited to): (a) Mechanical (performance) (b) Fabrication (model, make, SN) (c) Dimension (geometry) 3. Relations: Systems as asset-asset relationships 4. Spatial Information: Location and shape of assets
1. Proxy indicators which query the ‘‘existence” of the IQA subject, e.g., checking whether an asset exists in the model 2. Proxy indicators which ‘‘query and present” the IQA subject to the assessors and they compare them manually to the benchmarks, e.g., querying all assets in the mechanical room and providing a list of them that the assessors need to use it later as a checklist during their walkthroughs 3. Proxy indicators which ‘‘query and compare” the IQA subject with the benchmarks (no manual process), e.g., checking whether any spaces have redundant names
5.2.2. FM perspective The structure of the framework also considers the three essential FM terms, namely: 1. Assets (equipment), such as air handling units (AHUs), heat pumps, boilers, etc. 2. Spaces, such as rooms, corridors, etc. 3. MEPF Systems, which are mechanical, electrical, plumbing, and fire safety systems While assets and spaces are actual objects in BIM and represent geometric objects in the reality, MEPF systems represent the nongeometric (semantic) aspect of the reality. In other words, MEPF systems describe the semantic between the assets and determine which assets are working together as a system for a specific purpose. According to the object oriented data structure of IFC, the geometric information is already defined at the object level. Therefore, adding assets to the MEPF systems will provide indirectly the geometric information. For example, the definition of an HVAC system doesn’t need explicitly the geometry information of the pumps, AHUs, etc. because the considered assets already bring that information to the system and so with defining an HVAC system, we only need to determine the semantics, i.e., ‘‘which” assets are ‘‘how” interacting with each other. Thus, storing information about the geometry of assets in the MEPF systems will be redundant and geometric information is not necessary for the definition of an MEPF system. Hence, MEPF systems are a non-geometric representation of the reality in a model. MEPF systems can be generally divided into two following categories: 1. ‘‘System Classifications” that correspond to the standard definitions for MEPF systems, such as OmniClass, MasterFormat, and UniFomat (it is also possible that owners have their own standards in referring to these system classes). Examples for MEPF system classifications are: domestic hot water, domestic cold water, exhaust air, supply air, etc. 2. ‘‘Subsystems” are specific sets of assets that are usually connected to each other to fulfil a specific FM task. For instance, a branch of the air system that serves a certain section of a building falls into this category. The definition of subsystems is very flexible and it is up to the modelers to determine which assets need to be part of a subsystem. Having subsystems in a model will support troubleshooting and commissioning processes in terms of giving the practitioners a structured workflow. 5.3. Proxy indicators The practical part of the IQA happens through checking or measuring the relevant proxy indicators. Proxy indicators are generic queries formalized in the human language and are independent of assessment tools and IQA performance methods. The details of each proxy indicator depend on the required IQ dimension and the subjects of the assessment. Therefore, the discussed proxy indi-
5.4. Benchmarks In assessing the proxy indicators, the assessor needs to compare the outcomes against appropriate benchmarks. We have categorized these benchmarks into three different types: 1. Related documents: a variety of different types of documents, such as the owner’s requirements, project deliverables, building codes, municipal regulations, etc. Related documents are benchmarks for the ‘‘required information” in contrast to the built reality which is – if available – the benchmark for the ‘‘real information.” Therefore, this benchmark is suitable to investigate if a BIM represents what is required to be constructed. 2. Built reality: the actual constructed environment with as-built components. Since in a construction project there are always deviations between what is designed and what is built, this benchmark has a significant role in the IQA of models and is suitable to investigate whether a BIM represents what is actually constructed. 3. Model(s): used to identify contradictions among different information pieces within the model(s). Generally, all different models in a project can be considered as benchmarks. However in this work, we make the assumption that a single mechanical BIM contains all required information. Determining which benchmark is suitable for an IQA depends on its purpose. For instance, the related documents would be suitable benchmarks if the assessors are interested in assessing the IQ of a design BIM, while the built reality would be more relevant during the commissioning process. 5.5. IQA performing types The operationalization of the designed IQA tests can be generally divided into the three following categories: 1. Visual tests: In this performance method, the assessors compare visually the available information with benchmarks. For this purpose, they either go to the built object on-site and check the required information (we refer to these checks as ‘‘walkthroughs”), or they do the visual checks using images and scans (we refer to these checks ‘‘imaging/screen tests”). Examples for imaging tests are photos taken from the construction field, laser scans, webcams, etc. 2. Fully automated tests: Some tests have merely to do with the information within a model and so the tests can be performed using available methods in BIM authoring tools or BIM viewers. A typical example for fully automated tests is the information consistency check that identifies whether there is duplicate information for a specific term in the model (e.g. the serial numbers of different assets).
194
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
3. Semi-automated tests: This is the most common type of IQA tests. For these tests, we first filter the required information from the model using available methods in BIM authoring tools or BIM viewers. Then we check the results with the benchmark manually. For example, in order to check the specs for the entire installed pumps in a building, we filter and group them properly from the model first, and then compare them with the related documents as benchmark. 5.6. Detailed framework for creating and performing BIM-IQA tests for FM Using the descriptions above, we developed a framework for assessing the IQ of BIMs for facility management purposes (Table 3. With this framework, we provide a generic instruction for systematically performing BIM-IQA tests which is easy to understand and to perform. The starred tests in this table are discussed in more detail in the next section. 6. Operationalizing the IQA framework using case study projects In this section, we provide specific examples from our case studies to demonstrate the application of our framework for IQA. The operationalization of these tests follows a generic structure which corresponds with the structure of the framework. For each selected IQ dimension and subject of assessment, the assessor needs to identify the relevant proxy indicators and benchmarks and perform the IQA as indicated in the framework. The examples below will give a more detailed description about this operationalization. Since these instructions include generic steps, they are applicable with the majority of BIM tools. The only difference between different BIM tools is that some applications provide specific functions for querying models and so significantly reduce the amount of manual work in some IQA tests. To demonstrate the tool independency, the selected examples in this section are partly performed with Revit Schedules, which is a very limited tool within the Autodesk Revit, and partly performed with Solibri Model Checker (SMC), which is a special tool for querying BIMs. 6.1. Completeness of asset attributes The modeled assets in a mechanical BIM must include the necessary attributes for FM use with a certain value. Generally, these attributes can be categorized into three categories: 1. Attributes related to the fabrication of the assets: these are essential for finding additional information about the assets such as manuals, further specs, etc. 2. Attributes related to the performance of the assets: these depend usually on the asset’s type such as pump head, flow rate, voltage for a pump. 3. Attributes related to the assets’ geometry: these include the dimensional attributes that are usually computable from the object geometry such as height, volume, etc. Asset geometries are relevant to spatial issues like accessibility and transportability of the assets. However, using 3D BIM authoring tools implies that all assets would inevitably have geometry attributes. Thus, we exclude checking this category from our assessment. From the categories above, the existence of the first category is necessary because based on information about the fabrication, it would be possible to find the related information for the other two categories. Therefore, BIM guidelines such as GSA and LACCD BIMS address this point by requiring the following three fabrica-
tion attributes for all assets in a mechanical BIM regardless of their type [9,10]: Make or Manufacturer Model or Mark Serial Number (can’t be determined at the design or fabrication stage) For this assessment, we should filter assets through their families (or types) and the three attributes above, and then identify the incomplete asset attributes. Therefore, the proxy indicator here is the existence check of required attributes. Once the completeness of fabrication attributes is completed we can continue with assessing the type related attributes. Alternatively, these two steps can be combined to perform a type by type check as in our example below. Determining which performance attributes are necessary for a specific asset type depends on owner’s requirements and the followed guidelines by the owner. Therefore, this semiautomated assessment type needs to employ them as benchmarks. 6.1.1. IQA Test 1: Completeness of pump attributes In this example from Project #1, we combine the two attribute categories for pumps and assume that following attributes must be represented in an FM system for operation and maintenance purposes:
Make Model Pump HP (Pressure) Pump Head Flow Rate Fluid Type Voltage
In many projects, such as our case studies, the owner requirements did not specify the required information at this level of detail. However, this example shows that requiring specific asset attributes would be a feasible requirement. Furthermore, the attribute ‘‘Serial Number” is excluded because the final purchases aren’t finalized yet and so the model does not include any suitable attribute for the assets’ serial number. As shown in Fig. 3, we select the required attributes above together with the Pump Type, Family, Type and their quantity in Autodesk Revit to set up the required query for creating a suitable schedule. The schedule is filtered by the ‘‘Pump Type” with the condition of ‘‘parameter exists” and the option for ‘‘ Itemize every instance” is unchecked. These results show that there are no values set for the performance attributes: Pump HP, Pump Head, and the Flow Rate. We consider the zeros as not set and so incomplete. As for the attribute Voltage, even when the values are spread between two different attributes with the same name, we consider them as complete because the information exists. This issue is related to another IQ dimension, namely ‘‘Redundancy.” According to our investigations, the reason for having these two separate columns in this example is related to using families created by different manufacturers. In our case study project, one manufacturer defined ‘‘Voltage” as a ‘‘type property” in their families while the other manufacturer defined it as an ‘‘instance property.” 6.2. Completeness of subsystems As explained earlier, subsystems are specific sets of assets that are usually (physically) connected to each other for a specific FM purpose (Fig. 4). Subsystems are the breakdown of general systems in terms of dividing the MEPF systems into branches in different zones. However, the components of a subsystem are not necessar-
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
195
Fig. 3. Completeness test of pump attributes in the Project #1’s mechanical BIM. This test shows the incompleteness of values for Pump HP, Pump Head, and the Flow Rate.
Fig. 4. An example of a completely defined subsystem in the Project #1’s mechanical BIM. All 13 pieces are assigned to the subsystem ‘‘Mechanical Exhaust Air 39”.
196
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
ily from the same MEPF system type. For this assessment, we need to check if all assets are assigned to at least one specific subsystem. In other words, this assessment is about checking the existence of a relationship and so the proxy indicator type is checking an existence. The completeness assessment of subsystem can be performed fully automated since the benchmark is the model itself. In Revit, subsystems are defined for each asset through a specific attribute called ‘‘System Name.” Thus, to perform this assessment, we should create a query which investigates if all assets have a value for this attribute. 6.2.1. IQA Test 2: Incompleteness of subsystem In this example, we can investigate the incomplete subsystem definitions to identify the lacking information pieces. For this purpose, we set up a query for creating a Revit Schedule which contains assets’ Family, Type, and System Name. In addition, we define a filter for System Name when it ‘‘equals” the ‘‘ ” (blank). In this way, all assets with incomplete subsystem definition will be filtered and listed. Fig. 5 shows an example from our case study. These incomplete subsystem definitions need to be accordingly addressed by the modelers. Such incompleteness lead to disconnections within the semantics of the subsystems and so affect their usefulness for FM purposes. Fig. 6 visualizes the results of the same example on the same project with the Solibri Model Checker to give a better understanding about the range of this IQ issue within a complex project. 6.3. Accuracy of asset attributes The accuracy of the asset attribute values is required explicitly by several regulations and standards such as GSA, LACCD, COBIM,
and BSI [9,10,24,46]. Through this assessment, we can investigate whether the values of necessary attributes of an asset are accurately defined in the model. Similar to the completeness assessment of asset attributes, the accuracy assessment should include the consideration of two different attribute categories: 1. Attributes related to the fabrication: Make, Model, and (if exist) Serial Number 2. Attributes related to the performance. These attributes could be for pumps as an example: pressure, flow rate, fluid type, voltage, etc. Obviously, this assessment is a highly knowledge-based and time-consuming procedure. Hence, such assessment cannot be performed automatically. The general approach for this assessment is to filter and sort the assets according to their types and families and then browse them individually and check if their attribute values are accurately defined. In fact, it is reasonable to combine this assessment with the asset accuracy assessment, since the IQA steps are similar. That means finding an asset in a model, checking if it is the same asset type as it is installed on site, and then assess the necessary attributes. The benchmark for the accuracy check of attributes can include the related documents or the identification plate on the actual installed equipment. 6.3.1. IQA Test 3: Accuracy of pump attributes Fig. 7 shows the list of necessary attributes and their values for the centrifugal pumps from the mechanical BIM of Project #1. Fig. 8 is an example for a pump’s submittal data sheet that can be used as benchmark in such assessments. In creating the required Revit Schedule, a filter condition is defined to choose only
Fig. 5. Incompleteness of subsystems in the Project #1’s mechanical BIM.
197
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
(a) Assets as a part of a system
(b) Assets without a system assignment
Fig. 6. Completeness and incompleteness of subsystems in the Project #1’s mechanical BIM.
Fig. 7. Inaccuracy in the values for mark and manufacturer of centrifugal pump attributes in the Project #1’s mechanical BIM.
Submittal Data Information FI Series Pumps EFFECTIVE: MARCH 1, 2010
SUPERSEDES: OCTOBER 1, 2009
Museum Downtown __________________________________________
JOB CONTRACTOR
ENGINEER
Priority Mechanical Ltd. ___________________________
REP.
ITEM NO.
MODEL NO.
IMPELLER DIA.
G.P.M.
P-7 & P-8
FI8013
11.6" (294.6 mm)
2369.6 (149.5 l/s)
SERVICE
301-1489
5.3°C CHWS (30% P.G.)
1760 RPM MODEL 8013
Dialog ______________________________
Kehoe Equipment Ltd. _____________________________________ HEAD/FT.
134.0 (400 kPa) WEIGHT
H.P.
ELEC. CHAR.
150 (112.5 kW)
575/3/60
Lbs (kg)
PUMP/MOTOR
2619 (1188)
Fig. 8. Related document for a centrifugal pump type in the Project #1.
the ‘‘Pump Type” which ‘‘contains” the keyword ‘‘Centrifugal” and ‘‘Itemize every instance” is checked. In this example, the submittal data information shown in Fig. 8 represents the information related to two first centrifugal pumps listed in the schedule (Fig. 7). In this document the ‘‘ITEM NO.” represents the pumps’ ‘‘Mark” in the Revit Schedule which is inaccurate. Instead, the attribute ‘‘Mark” should correspond with the ‘‘MODEL NO” in the submittal document with value of ‘‘FI8013.” The main reason for this inaccuracy is that the stakeholders decided to install centrifugal pumps from the manufacturer ‘‘Taco” instead of following the suggestion of the designers which was using ‘‘Bell & Gossett” pumps. As a result, all relevant attributes are set inaccurately except the ‘‘Fluid Type.” An explanation for this case could be that the designers had used the centrifugal pumps from ‘‘Bell & Gossett” as a placeholder. However, the selected Revit Family for these pumps were still incomplete. Looking at Fig. 7 shows that the values for ‘‘Pump HP” are missing, and the values for ‘‘Pump Head” and ‘‘Flow Rate” are set to 0.
For correcting such problems, Autodesk provides several predefined families for different assets that the designers can employ in their models. For instance, a suitable Revit Family for FI8013 centrifugal pumps manufactured by Taco can be found here: http:// seek.autodesk.com/product/latest/agg/taco/Taco/FI8013. Furthermore, additional details and manuals about this type of centrifugal pumps is provided on the manufacturer’s website under: http://www.taco-hvac.com/uploads/FileLibrary/301-1489. pdf. Collecting such information should happen before the project handover and must be determined in the owner’s requirements. 6.4. Accuracy of subsystems In this assessment, we check for each subsystem whether it contains the correct components and if these components are connected to each other correctly. Similar to the other accuracy assessments, this procedure is also dependent on the assessor’s
198
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
knowledge and experience. Furthermore, it requires individual checks for each single subsystem. In assessing the accuracy of subsystems, there are two different type of checks required. Since many BIM authoring tools such as Revit automatically assign assets to a new subsystem, if they stay alone in a subsystem (a subsystem with one single component),
they cause inaccuracies. As explained earlier, subsystems represent a group of assets with a specific task and having a subsystem with only one member contradicts the concept of subsystems. In addition, assets in a subsystem need to be connected in an accurate and meaningful way. Thus, the two type of checks for assessing the accuracy of subsystems are investigating the subsystems with
Fig. 9. Subsystem ‘‘Mechanical Exhaust Air 1” with a single component ‘‘Terminals-SDE-Price_Industries Return: Size 12” is an example inaccurately defined subsystems in the Project #1’s mechanical BIM.
Fig. 10. Subsystem ‘‘Mechanical Exhaust Air 76” is the correct host for the component ‘‘Terminals-SDE-Price_Industries Return: Size 12” which is the only member of the subsystem ‘‘Mechanical Exhaust Air 1” (see Fig. 9).
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
199
Fig. 11. Disconnection in defining the subsystem ‘‘Mechanical Exhaust Air 76”. In this 3D view the components that are not a member of the subsystem ‘‘Mechanical Exhaust Air 76” are hidden.
single components, as well as checking the connectivity of its components. The first assessment has an existence proxy indicator, has the model as benchmark, and can be performed fully automated. The second assessment needs suitable queries that present the results to the assessor to visually evaluate the accuracy of subsystems based on the related documents.
6.4.1. IQA Test 4: Accuracy of subsystem In this example, the built-in System Browser in Revit can be used while the specific column for ‘‘Number of Elements” is checked in the browser setting. Now, we can go through the subsystems and evaluate their accuracy. Fig. 9 shows a return element as a ‘‘single component subsystem” while this return element has to be part of the rest of the branch (‘‘Mechanical Exhaust Air 76”) shown in Fig. 10. We highly recommend that the assessors go through all subsystems and visually controls the connectivity and accuracy of the assigned components. Fig. 11 gives an example from a disconnected subsystem which requires completion.
objects which has to be unique as well. This might cause redundancy issues. 6.5.1. IQA Test 5: Redundancy of object identifiers In the Project #1, the modelers used the attribute ‘‘Mark” as an equivalent to the asset identifier. Therefore, we created a Revit Schedule which includes this attribute together with the number of assets which have the same value ‘‘Count.” Identifiers with a number greater than 1 are redundant and need to be corrected (Fig. 12). However, as one sees, the modelers were consistent in describing the values of the attribute Mark in terms of following a specific pattern.
6.5. Redundancy of assets Many BIM standards and regulations require unique identification numbers for assets such as [9,24]. Common BIM authoring tools, like Revit, address this requirement by following an objectoriented approach in programming and in their data models. In this way, each modeled object (or entity) gets a unique ID when it is created. So any conflict with its uniqueness will produce an error by the modeling tool. Nevertheless, working merely with IDs would be confusing for non-programmers. Therefore, some modelers prefer to create a naming system for referring to their
Fig. 12. Redundancy of object identifiers.
200
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
6.6. Well-formedness of space attributes In this assessment, we investigate whether the information pieces in a BIM are defined in compliance with relevant standards or regulations. This can be performed in a fully automated way in
which we create and run queries based on the rules indicated in the related documents as benchmarks. Performing wellformedness tests is very dependent on the assessment tool. In case the assessment tool doesn’t provide rule-based queries, performing well-formedness tests becomes a semi-automated approach which
Sign Standards and Guidelines 3.1
Sign Classification for Interior Signs
Signs are given a sign code based on the following classification system. Each installed sign will have a unique number attached to its sign code. For Interior signs, the sign code includes the building code. Lower case letters (e.g., N2a) refer to sub-types. This number will be associated with specification drawings and can also be attached to the sign itself to aid in identification. Fig. 13. Owner #2’s Sign Standard and Guidelines.
Fig. 14. Query settings for checking the naming convention of space numbers in Project #2. This query basically is a translation of the Owner #2’s requirement about the space signs. In this way, the assessors are able to check the well-formedness of the spaces numbers in a model.
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
201
is again based on querying simpler queries and evaluating them visually.
with confusing abbreviations) by running specific queries within the model as the benchmark in a fully automated process.
6.6.1. IQA Test 6: Well-formedness of space attributes in project #2 Many owner organizations have specific rules for their space attributes. These are usually indicated in extra guidelines such as – in this example – Owner #2’s Sign Standard and Guidelines. In this example, we assess whether the space numbers in the architectural model of the Project #2 follow the fairly complicated signing rule of Owner #2 (Fig. 13). For performing this assessment, we utilized the Solibri Model Checker to create complex queries. Fig. 14 shows the translation of the Owner #2’s rules into the Solibri queries and Fig. 15 shows an example of one of the spaces found in the Project #2 architectural model which do not match with the Owner #2’s rules for space names.
6.7.1. IQA Test 7: Understandability of assets in project #3 mechanical room Fig. 16 shows the results of querying generic objects in the mechanical BIM of Project #3 in the Solibri Model Checker. A closer look at some of these objects which are actually located in the mechanical room of Project #3, shows that all of them have the generic shape of boxes and meant to be placeholders (Fig. 17). In a further investigation, we can see in Fig. 18(b) the same blue box at the top left of in Fig. 17. Reviewing its properties gives the user a hint about the identity of this equipment under the attribute ‘‘type” with the value ‘‘MCC.” A user who is not familiar with this expression wouldn’t be able to understand what this blue box is representing. However, it is possible to go to the already built mechanical room and find out what is represented in the model. After a walkthrough in the mechanical room, we could identify that the blue box is a placeholder for the MCC (motor control center) in the Project #3 (Fig. 18(a)). The examples in this section show how we can systematically assess BIMs for different IQ issues related to the FM using the introduced framework in Section 5. The used BIMs in the examples above are from projects with different size, complexity and level of detail which demonstrate the feasibility and adaptability of our framework in practice. Moreover, the tests are described in a way that they are adaptable to be performed with different tools for querying BIMs. So the framework is project and tool independent.
6.7. Understandability of assets As mentioned in Section 5, in order to assess the understandability of model components, we need to investigate generic objects or objects with unspecified shapes like boxes and spheres as placeholders. Understandability is an IQ dimension on the user level as explained in [15]. Therefore, it is highly dependent on the user’s domain knowledge. For instance, not all model users are familiar with abbreviations such as MCC (Motor Control Center), CHWS (Chilled Water Supply), etc. For this type of assessment, we need to check the existence of the generic objects (or objects
Fig. 15. An example for checking the well-formedness of space numbers in the Project #2 which shows that the space number 10100 does not follow the owner’s requirements stated in 19.
202
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
Fig. 16. Understandability issue in the mechanical BIM of Project #3.
Fig. 17. Blue box as a placeholder in the mechanical BIM of Project #3.
7. Conclusion The rapid increase of using BIM in the AECO industry creates many new opportunities for practitioners to exchange and use data. High quality BIMs can especially create great potentials for downstream information users, such as FM practitioners. However, assessing the quality and usefulness of delivered BIMs to the owners is a significant challenge and requires further research. In order to address this gap and support the ultimate goal of establishing
approaches for model-based project delivery, a new framework for understanding and assessing the information quality of BIMs from the FM perspective is presented this work. This framework is developed based on the information needs of the owners in our case studies and provides actual examples about IQ issues and their assessment. The framework is organized according to the main model characteristics: objects, attributes, relationships, and spatial information. It also considers the major FM terms for operation and maintenance, namely: assets, spaces, and MEPF systems. Furthermore, we identified five relevant IQ dimensions for FM and discussed the assessment of these IQ dimensions with specific examples from our case studies. As demonstrated in Section 3, IQ dimensions, such as accuracy or completeness, can be interpreted in different ways by the researcher. This may lead to ambiguities and difficulties in mutual understanding among researchers and practitioners. To avoid such issues, we discussed each relevant IQ dimesion and provided clear definitions for each of them. We also emphasized that the relevant IQ dimensions might vary from project to project and it is up to the assessors to define and identify them in each case. However, the presented approach for developing the IQA framework remains the same. In our proposed approach, the assessor needs to complete three steps for each IQ dimension and each subject of assessment: 1. Identify the relevant proxy indicators 2. Determine the benchmarks 3. Choose how to perform the IQA tests
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
(a) As-is
203
(b) BIM Fig. 18. Investigating the model placeholders in the actual mechanical room of Project #3.
Finally, we provided a series of actual tests from our case study projects to show how to operationalize the IQA framework in practice and to demonstrate the feasibility and practicality. The examples also show that performing IQA tests is an application independent approach and the tests can be performed with any BIM authoring tool that provides basic filtering features. However, using specific software applications for querying BIMs can provide more flexibility and save time in performing the IQA tests. Using the IQA framework, the owners will not only be able to perform IQ controls through the course of a project, but based on the structure of this framework they are able now to formalize better their information needs for the operation and maintenance phase. In other words, by using this framework the owners can develop ‘‘quality assessment strategies” for their projects. They can develop their information requirements based on the framework’s structure and perform IQA tests throughout different project phases which ultimately can help to have higher model quality in delivered BIMs at the time of project handover. In this way, our framework contributes to establishing model-based project delivery approaches. As a future work we envision to work on a series of tests to demonstrate the efficiency of IQA tests and their impact on the projects. This can be done through for example performing systematic IQA tests in parts of a project and compare its efficiency with a control group. For instance in a large project, it is common to have several design teams. In such a case study project, it is conceivable to establish IQA tests for a specific design team and select another design team with similar tasks as the control group. Then, through observing the amount of model corrections and growth, and through observing the volume of digital communications related to the models within the groups and between the designers and builders, we will be able to get a better understanding about the impact of establishing IQA tests on a project. The findings of this investigation will be published in a separate work. Furthermore, we aim to perform a cross-project analysis on our case study BIMs and provide a comparison between different projects considering each IQ dimension. Appendix A. List of interviewees List of the interviewees at the partner organizations for studying the current FM practices and identification of organizational information needs from FM perspective. More details are provided in [37]. Developing Owner Requirements for BIM-enabled Project Delivery and Asset Lifecycle Management). Owner Organization #1: 1. 2. 3. 4. 5.
Assistant Deputy Minister of Project Services Branch Executive Director of Project Services Branch Head, Procurement Standards at Project Services Branch Executive Director of Project Delivery Branch Project Director of Project Delivery Branch
6. Senior Project Manager at Project Delivery Branch 7. Director of Project Portfolio Management 8. Director of Facility Evaluation at Technical Services Branch 9. Facility Design Coordinator at Technical Services Branch 10. Supervisor-CAD/GIS and Records at Technical Services Branch 11. CAD/GIS Technologist at Technical Services Branch 12. Building Security Systems Engineer at Technical Services Branch 13. Assistant Deputy Minister at Regional Operation Branch 14. Director of Regional Operation Branch 15. Facilities Manager at Regional Operation Branch 16. Operations Supervisor at Regional Operation Branch 17. Director of Capital Development 18. Logistics Coordinator Owner Organization #2: 1. Facility Information Systems (FIS) manager, who is responsible for data collection and maintenance of the facility space information; 2. Manager of Space Inventory, who is also responsible for the information maintenance; 3. Facility manager of the Brown Zone as an information consumer who manages the operation and maintenance activities and building managers in that zone; 4. Building administrator of one of the case examples (Project #3) as an information consumer; 5. Technical specialist for the Building Management Services (BMS), who is explicitly assigned to work at Project #3 and is another information consumer; 6. Records Retrieval System Administrator 7. Manager of Technical Services 8. Main7tenance & Renewal Senior Analyst 9. Maintenance Technical Specialist 10. Head Maintenance Engineer 11. Millwright 12. Building Operation Program Manager 13. Head Maintenance Engineer (BMS Center) Appendix B. Examples of IQ requirements in general standards and guidelines 1. ‘‘BIM models are validated against the standards established in the BIM Guide Series to ensure accuracy and completeness of the Model.” (GSA BIM Guide Series 008: 2.2.7 Monitoring Compliance and Submittals [9]) 2. ‘‘Each BIM equipment object in the As-Built BIM shall contain geometric data and a minimum set of attributes:. . .” (GSA BIM Guide Series 008: 1.3.1 Accurate As-Built Geometry and Spatial Program BIM [9])
204
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205
3. ‘‘Example asset information requirements:” (BSI - PAS 1192-3: A.3 Specific asset information requirements [8]) (a) Legal information (b) Commercial information (c) Financial information (d) Technical information (e) Managerial information 4. ‘‘Model objects shall contain IFC parameters and associated data applicable to building system requirements.” (LACCD Building Information Modeling Standards: 2.4 Modeling Requirements [10]) 5. ‘‘Each space shall include the following spatial information:” (LACCD Building Information Modeling Standards: 2.4.6 Program Spatial Requirements [10]) (a) Space type – Omniclass (Table 13) (b) Space number – Omniclass (Table 14) (c) Space name (d) Space description (e) Department (f) Program 6. ‘‘The format of each information exchange shall be defined by the organization that has defined the OIR (organizational information requirements) and the AIR (asset information requirements).” (BSI, PAS 1192-3: 2014, p.10 [8]) 7. ‘‘Models should include all appropriate dimensioning as needed for design intent, analysis, and construction. Level of detail and included model elements are provided in the Information Exchange Worksheet.” (BIM project execution planning guide v2.0: SECTION J: QUALITY CONTROL [47]) 8. ‘‘ensure that the Project Facility Data set has no undefined, incorrectly defined, incorrectly named or duplicated elements and the reporting process on non-compliant elements and corrective action plans.‘‘ (USC BIM Guidelines: SECTION F: BIM AND DATA QUALITY CONTROL [26])
References [1] SmartMarket Report, SmartMarket Report: The Business Value of BIM for Owners, McGraw Hill Construction, 2014.
[15] P. Zadeh, S. Staub-French, R. Pottinger, Review of BIM quality assessment approaches for facility management, in: ICSC15: The Canadian Society for Civil Engineering 5th International/11th Construction Specialty Conference, University of British Columbia, Vancouver, Canada, 2015, pp. 1887–1896. https://circle.ubc.ca/bitstream/handle/2429/53726/Zadeh_P_et_al_ICSC15_ 342_Review_Of_Bim.pdf?sequence=1. [16] P. Zadeh, H. Burak Cavka, S. Staub-French, BIM Information Quality Analysis for Space Management, Osaka University, Osaka, Japan, 2016. [17] C.M. Eastman, P. Teicholz, R. Sacks, K. Liston, BIM Handbook: A Guide to Building Information Modeling for Owners, Managers, Designers, Engineers and Contractors, second ed., John Wiley and Sons, 2011, ISBN: 9781118021699 https://books.google.ca/books?id=aCi7Ozwkoj0C. [18] A. Akcamete, B. Akinci, J.H. Garrett, Potential utilization of building information models for planning maintenance activities, in: Proceedings of the International Conference on Computing in Civil and Building Engineering, 2010, pp. 151–158. [19] L. Sabol, Building information modeling & facility management, in: IFMA World Workplace, Design + Construction Strategies, LLC\, 2008. [20] C. Kasprzak, A. Ramesh, C. Dubler, Developing standards to assess the quality of BIM criteria for facilities management, in: AEI 2013 – sBuilding Solutions for Architectural Engineering, ASCE, 2013, pp. 680–690. http://ascelibrary. org/doi/abs/10.1061/9780784412909.067. [21] National Institute of Building Sciences, National Building Information Modeling Standard: Transforming the Building Supply Chain Through Open and Interoperable Information Exchanges, 2007.
P.A. Zadeh et al. / Advanced Engineering Informatics 33 (2017) 181–205 [43] P. Cykana, A. Paul, M. Stern, DoD guidelines on data quality management, in: Proceedings of the Conference on Information Quality, Cambridge, 1996, pp. 154–171. [44] T. Redman, Data Quality: The Field Guide, Data management series, Digital Press, 2001, ISBN: 978-1-55558-251-7 https://books.google.ca/books?id= iE2qsxDyEEYC. [45] M.J. Pratt, Extension of ISO 10303, the STEP standard, for the exchange of procedural shape models, in: Proceedings of Shape Modeling Applications, 2004, 2004, pp. 317–326.
205
[46] BSI, New BIM standard reduces asset management costs even further j BSI Group, 2014.