CIFE CENTER FOR INTEGRATED FACILITY ENGINEERING A Study of Handling Design Information in Construction Companies By Ole Berard & Martin Fischer CIFE Working Paper #WP128A November 2011 STANFORD UNIVERSITY COPYRIGHT © 2011 BY Center for Integrated Facility Engineering If you would like to contact the authors, please write to: c/o CIFE, Civil and Environmental Engineering Dept., Stanford University The Jerry Yang & Akiko Yamazaki Environment & Energy Building 473 Via Ortega, Room 292, Mail Code: 4020 Stanford, CA 94305-4020 A Study of Handling Design Information in Construction Companies Ole Berard, PhD Student Martin Fischer, Professor Keywords: design information, BIM, VDC, quality, information management Abstract Building Information Modeling (BIM) is gaining a strong foothold in the construction industry but design intent is often communicated through documents, even when a design and fabrication model is available. BIM can lead to consistent and more coordinated design entailing fewer errors and improve the communication of design information to fabrication and construction. However, most construction professionals are not yet using BIM for their work. Through qualitative research interviews, practices related to finding, using, and sharing building information were studied on eight U.S. construction projects in contrast to practices in Denmark, found through a detailed case study, to understand the practices in use and determine the discrepancies between traditional, current, and advanced practices. This research shows that design information is not structured for the planning and construction tasks of builders, which makes retrieving information error prone and labor intensive. In the interviews, nine types of design information issues related to planning and construction were identified. Other than “well coordinated,” the interviewees had no specific quality criteria for design information and the case studies showed little evidence of a holistic perspective regarding the value of design information. Detailed knowledge of handling design information can inform development of information systems and methodologies and improve design information. 1 1 Introduction In the architectural, engineering, and construction (AEC) industry, design information and processing pf design information is the basis of many work tasks of builders, such as estimation, scheduling, and procurement. In prevalent practice, information about the design of a building is transferred from the designer to the builder through documents either on paper or increasingly in digital form. The emergence of Building Information Modeling (BIM; Eastman and Siabiris, 1995) and Virtual Design and Construction (VDC; Eastman and Siabiris, 1995) implicates replacing documents with digital integrated information. Research has established that BIM and VDC methodologies can improve the design process (Flager et al., 2009; Shea et al., 2005) and design review (Hartmann and Fischer, 2007), the builder’s estimation (Staub-French et al., 2003), planning (Heesom and Mahdjoubi, 2004), and work coordination (Khanzode et al., 2008), and they facilitate prefabrication (Sacks et al., 2010). However, according to an industry study (McGraw-Hill, 2009), the use of BIM and VDC methodologies by builders is mostly limited to reducing conflicts (83 percent of the responding contractors), communication (82 percent), spatial coordination (78 percent), and prefabrication (78 percent). Many studies show that the quality of the design and design information is related to problems in construction (Johansson and Granath, 2010; Josephson and Hammarlund, 1999; Love and Li, 2000) and operation (Chong and Low, 2006). The perceived value of design information is not well aligned between designers and builders (Andi and Minato, 2003; Tilley and McFallan, 2000). In the United Kingdom the perceived quality of traditional tender documents in terms of missing information, incomplete drawings, conflicting information, inadequate specifications, software difficulties, and errors and mistakes has been declining the last 10 to 15 years (Laryea, 2011). It has also been established that management of design information can be time consuming (Flager and Haymaker, 2007). However, there is only a limited amount of detailed empirical understanding of how builders retrieve necessary information for their tasks from the extensive pool of design information. The relation between informational deficiencies and construction and operation problems has been established. The discrepancy between the alleged values of design information is well known in 2 research and practice. Still, there is little insight into the different types of informational problems that builders face in prevalent document-based, as well as BIM-based, design information. The objective of this study is to shed light on builders’ handling of design information in three ways. First, we establish the common practice of how builders retrieve design information for their work tasks. Second, we identify the types of problems that professionals encounter when retrieving information. Third, we describe how the identified issues are influenced by the use of BIM and VDC methodologies. In further research, in-depth understanding of the specific informational problems can lead to the use of a holistic framework to measure the quality of information, methodologies, and organizational settings. For AEC professionals this framework eventually enables the receiver (i.e., the builder) to require quality information from the sender (i.e., the designer). 2 2.1 Review Bridging the Gap between Designers and Builders Since the European renaissance (early 15th to early 17th century; (Ackerman, 1997) the role of the builder and the designer have been separate; until then the builder was also the designer. As a result, the design specifics needed to be mediated to the builder. Typically, this was and still is done through drawings and specifications. The construction of the Palazzo Sansedoni (1340) in Siena, Italy, is one of the earliest examples of the use of orthogonal and to-scale drawings and specifications (Toker, 1985), as known in present practice. Information systems (IS) are defined by Alter (2008, p. 451) as “system(s) in which human participants and/or machines perform work (processes and activities) using information, technology, and other resources to produce informational products and/or services for internal or external customers.” The AEC industry is reliant on IS consisting of human participants (in this case, the builder), rather than machines (i.e., computers) to perform work on information from other actors (in this case, the designer) using information technology (IT) to view, not for automation, to create informational products (e.g., estimates, schedules, logistic plans). 3 In other industries IS has increased the effectiveness and efficiency of the organization (Hevner et al., 2004) by improving the handling of information and knowledge. This is typically achieved through IS by automating business processes (Broadbent et al., 1999) and providing necessary information for human intervention. In other engineering industries, such as automotive and aerospace, object-oriented product models and the integration of production knowledge for the benefit of design and production is an established topic in research (Hvam, 1999; Yang et al., 2008) and industry- and companywide design IS are common. More current IS research and development seeks to provide the user with relevant information through task (Holz et al., 2006), context (Ye and Fischer, 2002), and location (Cheverst et al., 2002) awareness. Mithas et al. (2011) show that providing access to precise, timely, and reliable data and information to relevant entities and stakeholders (i.e., information management) enhances organizational performance in terms of process, customer, and performance management. It might seem implicit in IS research that digital information is preferable over paper. However, paper can be a key artifact within an organization that is not easily replaced digitally and might be incorporated in IS, as described by Mackay (1999) for air traffic control and Nomura et al. (2006) for commercial airline pilots. 2.2 From Omniscient Databases to Task-Specific Information In the AEC industry the emergence of micro computers (i.e., PC) and Computer Aided Design (CAD) in the 1980s have fostered the vision of using computers, databases, and applications for design, planning, and construction instead of drawings and specifications. Eastman (1981) describes “integrated design databases” that connect geometry and product properties to enhance productivity and enable design and shop drawings, as well as cost estimates. As CAD became widely accepted in AEC during the 1980s the geometry was stored in CAD databases. The CAD geometry could then be linked to object data in other databases and extracted by logical queries to perform “other [than drawing] functions” (Teicholz, 1989). Because the information needs of AEC professionals change as the design, planning, or construction of the building evolves, Zamanian and Pittman (1999) envision flexible schemata that define information exchange based on objectoriented models with information and functionality through simple heuristics that adapt throughout the project’s lifetime and connect manufacturer data into the design. Previous research focused on 4 an omniscient database; current research strives to understand information in the task’s context (Eastman et al., 2010). Boddy et al. (2007) envision developing construction project and taskspecific (e.g., estimation, energy analysis, scheduling) scenarios that describe information in the context of the work performed—integrating both unstructured data from documents and structured information from databases through semantic analysis in the information repository. Traditionally, it was believed that the connection of databases would provide the infrastructure for an AEC IS. Today, information has to be seen in the context of the task that needs to be performed. Since work routines in the AEC industry change (Hartmann et al., 2009) and construction knowledge has not yet been formalized sufficiently (Fischer, 2006), the task-specific information is presently unique for every project. 3 3.1 Method Definition and Limitation Design is “a specification of an object, manifested by an agent, intended to accomplish goals, in a particular environment, using a set of primitive components, satisfying a set of requirements, subject to constraints” (Ralph and Wand, 2009, p. 106). The primary agents in the design of a building are the designers, architects, and engineers, but product information from building product manufacturers and detailing and engineering from builders also contributes to the design. Information is essentially “news and facts about something” (Losee, 1997, p. 257). For this research design information are the building specifics needed by the design team (architect, engineer, manufacturer, builder, etc.) for communicating with relevant recipients (such as owner, user, municipality, builder). In this research we focus on the subset of design information needed by the builder to estimate, plan, and construct the building. This subset currently consists of drawings (cross sections, details, floor plans, elevations), specifications for the different disciplines, and schematics and schedules. As the design evolves, changes are covered by addendums, answers to requests for information (RFI), and change orders. The traditional medium of design information is documents. Recently, design information is supplemented with, but not replaced by, BIMs. For this research the design information itself is independent of the medium. 5 3.2 Data Collection and Coding The focus of this research—to establish the common practice of builders to retrieve design information, identify the types of problems that builders encounter, and describe how the problems are influenced by the use of BIM and VDC methodologies—is highly explorative. Explorative studies require a qualitative and inductive research methodology that allows the theory to derive from empirical studies. The research methodology was based on Grounded Theory (Glaser and Strauss, 1967), which derives theory on an inductive basis with no preconceptions. The data was collected by direct observations in a case study (Yin, 2009) and additional qualitative research interviews (Kvale, 1996) with builders to achieve in-depth understanding of the research focus. The analysis was based on the Grounded Theory methodologies of constant comparison (to be able to adjust the interview guide continuously) and open coding (to derive the types of problems inductively). The current practice of finding design information was established by direct observations and interviews of builders in a commercial construction project in Denmark, referred to as the Danish case. The qualitative research interviews were performed with builders in the U.S. Pacific West (Seattle, WA; Los Angeles, CA; and the San Francisco Bay area, CA), referred to as the U.S. interviews. The sampling strategy for the U.S. interviews was extreme cases, that is, unusual cases that are especially good or problematic (Flyvbjerg, 2006). To sample projects that are especially good in BIM and VDC methodologies the U.S. Pacific West was chosen out of the researchers’ evaluation, as this is one of the most advanced regions with respect to BIM and VDC methodologies. Participants were initially sampled based on the researchers’ knowledge about their technology maturity. The sample was extended by recommendations of the initial sample, asking the question, “Who else would be good to talk to?” The U.S. interviews consisted of three parts: (1) a semistructured section to explore both current-use BIM and VDC methodologies, and current praxis and problems in information findings; (2) a presentation of the findings from the Danish case to enhance the interviewees’ understanding of the research and confirm the findings; and (3) a validation of the study and an in-depth discussion of design information. At every U.S. interview between one and three interviewees were present, preferably from different levels within an organization to get a holistic picture of the 6 problem. The interviews were documented by notes taken during and after the interview, and pictures and documents from the interviewees such as drawings, slideshows, and handouts— following the Grounded Theory approach of “everything is data.” Notes were preferred over audio recordings to encourage interviewees to provide open and honest answers. The coding of the data (open coding) led to the identification of 129 information issues. The coding led to three levels of problems: (1) 77 unique issues, (2) 16 high-level issues, and (3) nine types. The identified high-level (HL) issues and types are described below. For the purpose of this research the use of BIM and VDC methodologies is categorized into three profiles (see Figure 1). The purpose is to communicate the BIM status of the interviewees’ projects. More complex classifications do exist in practice and research; however, this study needed a very simple model, based on three profiles observed during the study that relate to the industry study’s findings (McGraw-Hill, 2009). The profiles are used to cross reference the nine types of problems identified in this research through the 8 US interviews and the Danish case study in order to inform qualitatively whether the BIM and VDC profile influences the identified types of problems. Figure 1. BIM and VDC profiles based on observations in the study. 4 4.1 The Danish Case The Sample The Danish case is a commercial construction project including 35.000 m2 of office space and 18.000 m2 of parking space outside of Copenhagen, Denmark, with a design-build contract. Most of the building was designed in traditional 2D, and the contractor had access to the 2D drawings, both digital and paper. At the time of the research the construction work was almost finished. The 7 building was chosen to study the process of finding information for windows in a sloped curtain wall that supports the elevated roof of the atrium connecting several office wings (see Figure 2). The curtain wall measures 142 meters long and 3.5 meters high and consists of 108 windows, some of which can be opened. The design was in 2D documents, but the researcher built a simple 3D model of the curtain wall. This model enabled providing the facts above almost instantly and more reliably compared to the 10 to 15 minutes the traditional method takes to find such information. Figure 2. Drawing of the Danish case curtain wall. The design information on the Danish case is filed as paper, as well as digitally by the builder. For this study information finding was done on the digital version. The digital version was available in a folder structure (see Figure 3), first by discipline, that is, architectural, structural, and mechanical, electrical, and plumbing (MEP), and then by revision date. New revisions of the drawing or specification, received by email, are archived as PDF documents and a few as CAD files in a folder for each revision date. During the 3½ years of design and construction a total of 577 revisions of architectural, structural, and MEP drawings and specifications were sent out and 477 RFIs were issued by the builders. Three of these revisions were directly related to object of the study (i.e. curtain wall), whereas eight revisions addressed drawings also showing the curtain wall, but no RFI was issued. This low number of revisions is due informal communication between designer and builder. 8 Figure 3. Drawings and specifications folder structure for the Danish case. 4.2 The Practice of Finding Information The practice of finding information about the curtain wall as it was identified by direct observation is depicted in Business Process Mapping Notation (BPMN; White and Miers, 2008) and shown in Figure 4. First the current documents are identified, and then the architectural specification is opened, as it also contains references to drawings. After having accessed the drawings referenced in the specification, the builder goes through the list of drawings to find more relevant drawings, as not all are referenced in the specification or on the drawings. If no more relevant drawings are available and the identified information still is not sufficient or is ambiguous, it is necessary to ask someone: either the designer or the building product manufacturer. To clarify a product with the manufacturer it is often sufficient to use the website or to speak with a sales representative. Requesting information from the designer is a formal process called RFI that potentially results in an update of the design information. Getting an RFI answer from a designer often takes a week or more depending on the complexity of the question. In the Danish case 20 documents had to be opened to study the available information about the curtain wall. Only half of these documents showed relevant information; the whole process took one hour. The process has to be repeated at least partially for every revision, eight in this case. 9 Figure 4. BPMN flowchart of finding design information about a curtain wall for the Danish case. 4.3 Design Information Problems in the Danish Case The high level (HL) issues identified during the Danish case are described below. They are marked “HL” and are consecutively numbered. An overview including the related types can be seen in Table 1. When the design information requires clarification the timing (HL01) of the designer’s or manufacturer’s answer can interfere with the builder’s work flow. The process of finding (HL02) the desired information took net one hour, because the structure (HL03) of the design information was not appropriate for the task. Design information is structured around areas (i.e., plans) and views (e.g., elevations) rather than around products; hence, information about one product is in many documents. This process was repeated frequently during design, planning, and construction by several roles (e.g., project manager, design manager, estimator, foremen) and for several tasks (e.g., estimation, procurement, scheduling, work planning). Therefore, a significant amount of time 10 is spent on information finding. Estimate: 8–12 professionals and 3–6 times per project for net one hour results in 24–72 hours per product. The significant effort involved leads the builder to ask, “When do I stop?” Consequently, the builder will evaluate time consumption against risk of insufficient knowledge. The consequence of a wrong evaluation is construction errors that could be prevented. Type High-Level Issues Problems Related to the: Access HL09 Builder’s effort to access design information. Coordination HL11 Coordination level of different disciplines. Correctness HL06, HL07, HL12 Extent of missing, incorrect, or outdated design information. Distribution HL13, HL14 Handling and distribution of the design information. Format HL02, HL03, HL08 Flexibility and conciseness of the medium. Handling HL04, HL15 Effort to transform or update information for work tasks. Precision HL05, HL10 Representation of actual and existing conditions, which are accurate and unambiguous. Relevance HL01 Timing of the information delivery. Volume HL16 Number of documents and files, and other media. Table 1. Types of design information problems and related HL issues identified in the Danish case (HL01–HL08) and the U.S. interviews (HL09–HL16) through open coding. The design information in the Danish case included three basic types of information: First, explicit information is apparent and unambiguous. Among these are measurements, building products, and color, surface covering, and materials requirements. Second, inferred information has to be calculated, measured, counted or inferred, or derived from the design information. A piece of information is transformed (HL04) and becomes suitable for a purpose. Examples include quantities, elevations, some dimensions, and opening direction. Third, implicit information is using product requirements instead of functional requirements. The actual requirement is than implicit in the product type, for example, instead of requiring sun shading, a product that provides sun shading is 11 required. Implicit information adds ambiguity (HL05), for example, it is uncertain whether sun shading is a requirement or the product was chosen by mistake. Besides the three types of information in the Danish case, necessary information was missing (HL06) or was difficult to derive. The placement of the glass façade according to the module grid and the angle of the sloped façade were not provided; the angle could be derived by the Pythagorean theorem. Product data was outdated (HL07); two products in the specification did not exist anymore or had changed name. Information was scattered (HL08) over multiple documents. A product name that had not been mentioned before appeared on a detail drawing. 5 5.1 The U.S. Interviews The Sample The U.S. interviews (see Table 2) intended to focus on specific construction projects. Unfortunately, this was not possible in all cases, and two interviewees had a broader company perspective. Eight situations (two companies and six unique construction projects) were discussed and four follow-up interviews with software developers and VDC/LEAN consultants were conducted to verify the broader relevance of the identified problems. In total, 23 persons (17 from the eight situations and six from follow-up interviews) participated in 14 interviews (10 and four). The interviewees represented different levels of management: three participants came from top management (COO and CEO) of the company, five were from higher project management (project director, design manager, senior project manager), three were from company/regional BIM management, and six worked on the construction projects (project engineer, MEP specialist, planner). The combination of different levels of management provided a wide-ranging perspective, from the day-to-day problems to the overall strategy of information management within the company. No. Type VDC Value USD Phase Companies Interviewed Interviewees 1 Amusement Level 2 200 m + Finishing 1: CM (Large, Worldwide) 2 – PM & PE 1: GC (Large, U.S. Wide) 1 – BM Park Ride 2 Company Level 2 12 3 Health Care Level 3 300 m + Pre Con 1: GC (Large, U.S. Wide) 2 – BM & PE 4 Health Care Level 2 300 m + Constr. 2: GC (Medium, CA) 3 – TM, PM, & Sub (Medium, North. CA) PE 1 – PM 5 Company Level 3 6 Health Care Level 2 300 m + 7 Bio Tech Level 2 50 m 8 Education Level 1 10 m 1: GC (Large, Worldwide) 1 – BM Pre Con 1: GC (US, Worldwide) 1 – PM Finishing 2: CM (Medium, North. CA) 4 – TM & 3 PE Sub (Small, North. CA) 1 – TM 1: CM (Medium, North. CA) 1 – PM Constr. VDC level = see Error! Reference source not found., Pre Con = Pre Construction, Constr. = Construction, CM = Construction Manager, GC = General Contractor, Sub = Subcontractor, PE = Project Engineer, BM = BIM Management, PM = Project Management, TM = Top Management. Table 2. Case profiles and companies. The companies’ role on the construction projects was primarily the planning, procurement, and operations as construction managers or general contractor. They handle design information and experience many of the issues and challenges; the construction work is performed by subcontractors or self-performing units, which is why two interviews were conducted with subcontractors. In total, 10 construction companies participated in the study: four large (more than 500 employees) with worldwide or U.S.-wide activities, four medium sized (between 100 and 500 employees), and two small (fewer than 100 employees) primarily based in northern California. The BIM and VDC maturity was evaluated according to three levels: traditional (no models used), current (MEP coordination), and advanced (more than MEP coordination) practice (see Figure 1). Health care projects, hospitals as well as medical office buildings, are overrepresented in the study due to the current focus on application of BIM and VDC methodologies on projects. Case profile number eight was a traditional project chosen as a control project. 5.2 Discussing Results from the Danish Case To confirm and substantiate the findings of the Danish case, they were presented during the U.S. interviews. The U.S. interviewees confirmed the process, the time it took (HL02), and problems in timing (HL01). The net time of one hour was recognized but because builders hardly ever work one 13 hour without interruption, the gross time was estimated to be a whole day. An MEP coordinator mentioned that it took one week to figure out a floor plan, including multiple products, before being able to ask questions. However, some U.S. projects use a window schedule that compiles the information in one document and probably speeds up the process specifically for windows. The typical designer answers questions within one to four weeks and longer for more complicated questions. Problems with missing information (HL06), ambiguity (HL05) in functional requirements, outdated (HL07) product data, and scattered information (HL08) were also recognized. Outdated products have been seen, as far as specifying products that are not legal by state law. An explanation offered is that building products change regularly, and designers do not update their catalogues as often. “Drawings are the language construction speaks!” —MEP subcontractor Drawings have been the established method of communication between design and construction for centuries. Some subcontractors address issues with the structure (HL03) of drawings by using enhanced shop drawings, including all the information necessary for the installation in the field, which are, at least partially, extracted from BIMs. Disagreement was on the question “When do I stop?” Although some would look at more documents, such as structural cross sections and addendums and answers on RFIs, others would stop sooner and either ship their risk downstream to the subcontractor or upstream by making assumptions in the product submittals to the designers. 5.3 Quality of Design Information When asked “What is quality of design information?” builders typically answered “That it is well coordinated!”—which usually meant actual dimensions were given with no conflicts between disciplines. Builders struggled with differentiating between good design that fulfills the owner’s requirements and good design information that fulfills the builder’s requirements. Only one 14 interviewee (advanced practice) distinguished quality of design information by time to find information, relevancy for the task, not out of date, latency, and access. The largely monolithic view on information quality—well coordinated—seems to be shared throughout the industry and is evidence that only a few are concerned with receiving documentation that suits their requirements from the designers. “Our biggest problem is (design) changes.” —MEP construction manager The biggest problem pointed out by the interviewees is that the design and the design information is fluid and changes often. Design changes are implemented frequently, often because the client is undecided. Clients and designers do not seem to understand the builder’s need to freeze the design. However, when asked directly to elaborate on the problems and issues they encounter with design information all professionals described numerous issues and a wide range of problems they encounter. 5.4 Information Problems The physical copy of the current set of design information is often located at the site office; because of the physical distance for the subcontractor access (HL09) can be a problem. Also, it can be difficult to identify the current version and retrieve a complete and up-to-date set. Easier and digital access would speed up the process. There is a gap between what is designed and what is built. This gap widens when the design is not based on precise (HL10) and actual dimensions, specific products, and existing conditions. Laser scanning is gaining popularity to capture existing conditions. Precision is also related to designed tolerance in building products that are assembled on site that cannot be absorbed by the construction, but are adding up so that they exceed the overall tolerance. The design is done by many different organizations and persons. Lack of coordination (HL11) among them can result in potentially expensive field conflicts. Traditional design is coordinated on light tables in 2D. With geometrical 3D models coordination can be automated by taking all three 15 dimensions and even installation sequence into account. The result of the coordination is shop drawings that are coordinated between mechanical, electrical and plumbing systems. In addition to information that is missing (HL06) and outdated (HL07), information can be profoundly incorrect (HL12) and overlooks requirements and legislation. Builders spend time managing information; forwarding relevant information to other actors; distributing (HL13) and routing RFIs, RFI answers, addendums, and submittals; and a myriad of emails need to be screened for relevant information. Builders use a variety of IT systems for different tasks, but many cannot transfer information between them. Many disciplines and trades use the architect’s drawings as a foundation for their work. These drawings have to be reduced in detail and distributed every time the architect releases a new version, which is why interoperability (HL14) is an issue. Design changes do not always entail updates (HL15) of drawings and specifications, but are published in addendums and RFI answers. The information is spread over even more documents, and information management becomes more difficult. A solution is to hand draw the information about where a product is installed (i.e., as-built documentation) into paper drawings. Problems arise when this is not put into the digital drawing, which makes it difficult to trace such information at a later stage. Design information is spread over different media, such as emails, drawings, specifications, RFI answers, addendums, and hand sketches, consequently it is difficult to retain an overview of what is being built (see Figure 5), simply because of the physical volume (HL16). For example, the specifications on a studied US project consisted of more than 1,600 pages, which makes them difficult to handle by the construction team. Consequently, builders start memorizing parts that are important to them, which can lead to assumptions. Also, the volume can cause gaps between subcontractors and lack of understanding of the scope of the work for the builder. 16 Figure 5. Example of voluminous design information for a Danish construction project (photo by author). 6 VDC and BIM Methodologies’ Influence on Design Information To evaluate qualitatively whether the use of BIM and VDC methodologies influence the types of issues in design information, the three levels—traditional, current, and advanced practice (see Figure 1)—were cross-referenced with the nine types of information problems (see Table 1 in the original data set. For the three levels all quotes concerning a type of issue were extracted and analyzed, and the issue was then evaluated to discern whether it was a major issue that has impact on productivity, an issue that was time consuming, or was improved, which means that the issue was addressed significantly compared to traditional practice (see Figure 6). 17 Figure 6. Qualitative evaluation of the extent to which identified issues were addressed. The Danish case study identified eight HL issues (HL01–HL08) concerning five types: precision, correctness, format, relevance, and handling. The U.S. interviews identified eight additional HL issues (HL09–HL16) and touched all nine types. This discrepancy is not due to the additional HL issues and types not relevant on the Danish case but because the study did not reveal them explicitly. Many of the issues had been resolved because the study took place late in the project. Other issues, such as access and volume, seem to be accepted as general conditions of the AEC industry. Access only becomes explicit in the context of discussing the situation of multiple stakeholders; it grows as an issue moving downstream. Also, distribution and relevance are issues highly related to the collaboration on a project. The issue types are related to the product’s design information (format, handling, volume) as it impacts the project organization of creating and receiving it (access, distribution, relevance) as well as to the process of building it (coordination, correctness, precision). Traditional practice does not address any of these issues. Current practice tries to improve the product by utilizing BIMs and organization by easing access to information by many actors, as well as providing coordinated design. Advanced practice is characterized by changing all three perspectives. 6.1 Traditional Practice Traditional practice, without any models, was identified in the control project (Table 2). Lack of coordination of design information was a problem in multiple situations. For example, the roof supports were misplaced according to the design, which caused repeated problems with piping. The MEP subcontractors build where they found space, rather than according to the design. In total, over the construction period of two years there were 315 RFIs—a substantial amount for the project size. As-built information was drawn in by hand (format) and addendums and RFI answers did not entail updates of drawings and specifications (correctness); therefore, they were pasted onto the existing drawings by the project manager (handling). Physical drawings were in big piles in the construction office (voluminous), and subcontractors had to access the most current information (access) there. Waiting times on RFIs were long (relevance), the answers afterward were 18 distributed to the relevant subcontractors manually, and answers in the form of hand sketches did not take all conditions into account (precision). Therefore, all types were significant issues on the project. 6.2 Current Practice Five of the eight projects used geometrical 3D models for coordination (current practice). For the five, coordination was significantly improved and field problems were reduced. Design and production models were created to coordinate the geometry and models were exchanged through 3D CAD files (interoperability). Accessibility was often improved by sharing information online, for example, on an ftp site. Distribution was still a major issue because updates were notified manually through email and the ability to track the changes was limited. Also, much information was in the form of digital drawings and specifications (format) and models were used for other work tasks only to a limited degree (handling). A subcontractor suggested freezing the agreed-upon design by burning it onto a CD; a pragmatic solution, but it shows a lack of an effective information management system. Precision was a big problem: “Designers have it in their paper but does it fit in reality?” Laser scans were a cost-effective way to address existing conditions in the design but builders missed realistic conditions in design information and experienced incorrect, missing, or outdated information (correctness). The gap between design intent and construction was often bridged by time-consuming RFIs and submittals (relevance). 6.3 Advanced Practice Two projects with advanced practice were sampled. These projects solved some of the issues by the use of technology and more importantly by closer collaboration; these projects build on contracts that encourage collaboration, such as Integrated Project Delivery (AIA, 2009). Extensive 3D modeling by subcontractors of their work’s product (e.g. MEP systems and dry wall) improves coordination and information management systems improve accessibility of the design information. More importantly, the collocation between designers and builders results in precise and correct information and relevance is improved by coordinating the design sequence and information output 19 with the builders, through so-called pull-scheduling sessions; as major issues are addressed, new problems arise. Close collaboration, coordination, and collocation reduce the time spent on redesign through regular review and feedback. However, current pull-scheduling sessions require many actors to attend day-long meetings and each actor contributes to only a small part of each meeting. Subcontractors might not be willing to provide all the needed information; they are typically paid for design services but their revenue is generated by production and not design consulting. They have to ensure getting the job and not give competitors an advantage by providing too many insights. They keep product information private until a construction contract is signed, as product information and knowledge is crucial for their bids. Not all designers participated in the collocation; especially architects, who prefer their downtown offices. Collocation and design-assist is expensive and the owner needs to be willing to make the extra investment. As projects get more advanced, interoperability becomes a bigger issue because it is desirable to use the BIMs for additional purposes, for example, to get the design models into the facility management system. Information is not necessarily reused. Designers build one model and then subcontractors build another one for their own scope of work. Even though much information was in the models, a document for a very comprehensive specification was needed. The advanced practice projects were in the design phase, and here models dominated, but as the projects move into planning and into the field many of the work tasks might still be based on drawings, entailing labor-intensive transformation of information, is assumed by the interviewees. 7 Discussion It is time consuming to find and compile information in a structure that does not support this task. For this reason builders assess when to stop searching for specific information, based on the time it takes and the risk of detecting incomplete information. As shown in recent literature, the view on information is shifting from the idea of an omniscient database to considering information in the context of the work task. An IS that is aware of the task and the context could provide the desired information for the builder’s work. This would save time and reduce the risk of finding incomplete 20 information. The challenge is to provide the builder with relevant information for the task at hand. But how can the system know what information is relevant in the AEC industry where processes and information requirements are not well defined? Advanced practice uses pull scheduling for design information to clarify what information is required; only relevant information is produced. Current pull-scheduling sessions as observed in advanced practice require many actors to attend day-long meetings and each actor contributes to only a small part of each meeting. These meetings address a couple of the types of issues by providing timely information and potentially increasing precision, correctness, and coordination. The output of the pull-scheduling meetings is not generalized; it contributes only to the participants’ personal experience and basically, every project starts over. A common and shared way to document information requirements for actors and tasks on a project-to-project basis can help collect these experiences. Potentially the documentation of information requirements can be generalized by analyzing multiple sets and reduce the time used on pull scheduling. However, it is important to take into account the unique composition of construction projects. This reduces not only time spent on information finding and creation but it becomes explicit which information is missing to perform work tasks. Paper drawings are an important part of the AEC industry of today and there is an important lesson in current literature concerning air traffic control and pilots. Paper drawings need not necessarily be replaced but IS could provide better paper drawings as an output, providing more relevant information on a piece of paper as well as on a computer screen. 8 Further Research The identified problem types are guidelines for areas that need improvement. Further research can transform the types into a framework of design information quality, consisting of metrics and a methodology. The metrics should be based on the identified types of issues as well as results from the existing literature on information quality in other fields. Measuring design information quality is necessary to improve the information provided to the builder. Holistic metrics provide guidance to 21 areas that need improvement and follow-up on initiatives. A proactive measurement system can ensure that issues are taken care of during each project by pinpointing problems and enabling the receiver to require information from the creator. The metrics need to be supplemented by a methodology to define the information requirements. The information needed to perform a task varies depending on the task, the individual actor and the context of the construction project. To improve Design Information Quality a methodology that can identify task, actor and context relevant information is needed to require information from the creator. Pull scheduling, as used by advanced projects, can serve as an inspiration for a methodology to identify such information. The methodology needs to document the information requirements with a clear-cut documentation that captures the process and the information, as well as the organization, to communicate information requirements unambiguously to the creator. 9 Conclusion The current practice of finding task-specific product information in the design information was established by detailed observation on a Danish case. The current practice is time consuming and error prone, the structure of the design information does not support it, and information for a work task is spread over multiple documents. Most builders describe design information quality as monolithic but “well coordinated.” This research identified 16 HL issues that builders have with design information. Eight were identified by studying a Danish case and eight more through 14 U.S. interviews. Together they provide a nuanced representation of the various issues builders have with design information. Design information is understood in the wider set and is provided by architects and engineers, as well as by subcontractors and building-product manufacturers. Of the 16 HL issues nine types were identified that are related to design information: product (format, handling, and volume), project organization or form of collaboration (access, distribution, and relevance), and building process (coordination, precision, and correctness). Together, all nine issues provide a holistic view of the problems faced by builders under the design, planning, and construction of a building; focus on product, organization, and process are the means of addressing them. 22 The comparison of eight U.S. projects reveals qualitatively that advanced practice in BIM and VDC methodologies significantly improves the identified types of issues. This is achieved by utilizing BIM and changing the organization to a closer collaboration while influencing the process of building with precise, correct and task relevant information. However, this closer collaboration comes at a cost. The design and construction teams both need to invest time in design coordination, planning, pull scheduling, and collocation, and the owner’s upfront investment will increase. Further research is needed to define a framework to improve design information quality that should consist of metrics and methodology. The metrics would measure the effect of improvements in design information, and proactive measurements on construction projects would continuously improve the design information. A methodology would develop information requirements based on the task, context, and actor performing it. 10 Acknowledgments The authors would like to express their gratitude to all of the people and companies involved in this study. We would especially like to thank the CIFE members and MT Højgaard for their support; the Ejnar and Meta Thorsens Foundation for funding this research; and Rolf Büchmann-Slorup, Flemming Vestergaard and Jan Karlshøj for their valuable feedback. 11 References Ackerman, J. S. (1997) Villard de honnecourt's drawings of reims cathedral: A study in architectural representation. Artibus Et Historiae, 18(35), pp. 41-9. AIA (2009) C191™-2009, standard form multi-party agreement for integrated project delivery. American Institute of Architects, Washington, D.C, USA. Alter, S. (2008) Defining information systems as work systems: Implications for the IS field. European Journal of Information Systems, 17(5), pp. 448–469. Andi, and Minato, T. (2003) Design documents quality in the japanese construction industry: Factors influencing and impacts on construction process. International Journal of Project Management, 21(7), pp. 537-46. Boddy, S., Wetherill, M., Rezgui, Y., Cooper, G. (2007) Computer integrated construction: A review and proposals for future direction. Advances in Engineering Software, 38(10), pp. 677-87. 23 Broadbent, M., Weill, P., St.Clair, D., Kearney, A. T. (1999) The implications of information technology infrastructure for business process redesign. MIS Quarterly, 23(2), pp. 159-82. Cheverst, K., Mitchell, K., Davies, N. (2002) Exploring context-aware information push. Personal and Ubiquitous Computing, 6(4), pp. 276-81. Chong, W., and Low, S. (2006) Latent building defects: Causes and design strategies to prevent them. Journal of Performance of Constructed Facilities, 20(3), pp. 213-22. Eastman, C. M., Jeong, Y. S., Sacks, R., Kaner, I. (2010) Exchange model and exchange object concepts for implementation of national BIM standards. Journal of Computing in Civil Engineering, 24(1), pp. 25-34. Eastman, C. M., and Siabiris, A. (1995) Generic building product model incorporating building type information. Automation in Construction, 3(4), pp. 283-304. Eastman, C. M. (1981) Database facilities for engineering design. Proceedings of the IEEE, 69(10), pp. 1249-63. Fischer, M. (2006) Formalizing construction knowledge for concurrent performance-based design. Intelligent Computing in Engineering and Architecture, 4200, pp. 186-205. Flager, F., and Haymaker, J. (2007) A comparison of multidisciplinary design, analysis and optimization processes in the building construction and aerospace industries. CIB 24th W078 Conference, Maribor, Slovenia, pp.625-30. Flager, F., Welle, B., Bansal, P., Soremekun, G., Haymaker, J. (2009) Multidisciplinary process integration and design optimization of a classroom building. Electronic Journal of Information Technology in Construction, 14, pp. 595-612. Flyvbjerg, B. (2006) Five misunderstandings about case-study research. Qualitative Inquiry, 12(2), pp. 219-45. Glaser, B. G., and Strauss, A. L. (1967) The Discovery of Grounded Theory: Strategies for Qualitative Research, Aldine Publishing Co., New York, NY, USA. Hartmann, T., Fischer, M., Haymaker, J. (2009) Implementing information systems with project teams using ethnographic–action research. Advanced Engineering Informatics, 23(1), pp. 57-67. Hartmann, T., and Fischer, M. (2007) Supporting the constructability review with 3D/4D models. Building Research and Information, 35(1), pp. 70-80. Heesom, D., and Mahdjoubi, L. (2004) Trends of 4D CAD applications for construction planning. Construction Management and Economics, 22(2), pp. 171-82. Hevner, A. R., March, S. T., Park, J., Ram, S. (2004) Design science in information systems research. MIS Quarterly, 28(1), pp. 75-105. Holz, H., Rostanin, O., Dengel, A., Suzuki, T., Maeda, K., Kanasaki, K. (2006) Task-based process know-how reuse and proactive information delivery in TaskNavigator. Proceedings of the 15th ACM International Conference on Information and Knowledge Management, ACM, Arlington, VA, USA, pp.522-31. 24 Hvam, L. (1999) Procedure for building product models. Robotics and Computer-Integrated Manufacturing, 15(3), pp. 77-87. Johansson, P., and Granath, K. (2010) Using construction deficiency reports and product models as systematic feedback to avoid design errors caused by lack of knowledge. Proceedings of the CIB W78 2010: 27th International Conference, CIB, Cairo, Egypt, pp.paper 47. Josephson, P. E., and Hammarlund, Y. (1999) The causes and costs of defects in construction - A study of seven building projects. Automation in Construction, 8(6), pp. 681-7. Khanzode, A., Fischer, M., Reed, D. (2008) Benefits and lessons learned of implementing building virtual design and construction (VDC) technologies for coordination of mechanical, electrical, and plumbing (MEP) systems on a large healthcare project. Electronic Journal of Information Technology in Construction, 13, pp. 324-42. Kvale, S. (1996) Interviews: An Introduction to Qualitative Research Interviewing, 3rd printing Edn. Sage Publications, Thousand Oaks, CA, USA. Laryea, S. (2011) Quality of tender documents: Case studies from the UK. Construction Management and Economics, 29(3), pp. 275-86. Losee, R. M. (1997) A discipline independent definition of information. Journal of the American Society for Information Science, 48(3), pp. 254-69. Love, P. E. D., and Li, H. (2000) Quantifying the causes and costs of rework in construction. Construction Management and Economics, 18(4), pp. 490. MacKay, W. E. (1999) Is paper safer? the role of paper flight strips in air traffic control. ACM Transactions on Computer-Human Interaction, 6(4), pp. 311-40. McGraw-Hill (2009) The business value of BIM - getting building information modeling to the bottom line. 1st Edn. McGraw-Hill Construction, New York, NY, USA. Mithas, S., Ramasubbu, N., Sambamurthy, V. (2011) How information management capability influences firm performance. MIS Quarterly, 35(1), pp. 237-56. Nomura, S., Hutchins, E., Holder, B. E. (2006) The uses of paper in commercial airline flight operations. Proceedings of the 2006 20th Anniversary Conference on Computer Supported Cooperative Work, ACM, Banff, Alberta, Canada, pp.249-58. Ralph, P., and Wand, Y. (2009) A proposal for a formal definition of the design concept. Design Requirements Engineering, 14, pp. 103-36. Sacks, R., Kaner, I., Eastman, C. M., Jeong, Y. (2010) The rosewood experiment — building information modeling and interoperability for architectural precast facades. Automation in Construction, 19(4), pp. 419-32. Shea, K., Aish, R., Gourtovaia, M. (2005) Towards integrated performance-driven generative design tools. Automation in Construction, 14(2), pp. 253-64. Staub-French, S., Fischer, M., Paulson, B., Fischer, M., Kunz, J., Ishii, K. (2003) A feature ontology to support construction cost estimating. Artificial Intelligence for Engineering Design, Analysis and Manufacturing, 17(2), pp. 133-54. 25 Teicholz, P. (1989) Integration of microcomputer applications. current and future approaches. Proceedings of Construction Congress I - Excellence in the Constructed Project, ASCE, San Francisco, CA, USA, pp.147-57. Tilley, P. A., and McFallan, S. L. (2000) Design and documentation quality survey. 1st Edn. CSIRO, Highett, Victoria, Australia. Toker, F. (1985) Gothic architecture by remote control: An illustrated building contract of 1340. Art Bulletin, 67(1), pp. 67-95. White, S. A., and Miers, D. (2008) BPMN Modeling and Reference Guide, 1st Edn. Future Strategies Inc., Lighthouse Point, FL, USA. Yang, W. Z., Xie, S. Q., Ai, Q. S., Zhou, Z. D. (2008) Recent development on product modelling: A review. International Journal of Production Research, 46(21), pp. 6055-85. Ye, Y., and Fischer, G. (2002) Supporting reuse by delivering task-relevant and personalized information. Proceedings of the 24th International Conference on Software Engineering, ACM, Orlando, FL, USA, pp.513-23. Yin, R. K. (2009) Case Study Research: Design and Methods, 5th Edn. Sage, Thousand Oaks, CA, USA. Zamanian, M. K., and Pittman, J. H. (1999) A software industry perspective on AEC information models for distributed collaboration. Automation in Construction, 8(3), pp. 237-48. 26