Automation in Construction

Transcription

1 Automation in Construction 35 (2013) Contents lists available at ScienceDirect Automation in Construction journal homepage: Generating construction schedules through automatic data extraction using open BIM (building information modeling) technology Hyunjoo Kim a,, Kyle Anderson b,1, SangHyun Lee b,2, John Hildreth a,3 a Engineering Technology and Construction Management, The University of North Carolina at Charlotte, United States b Civil and Environmental Engineering, University of Michigan, Ann Arbor, United States article info abstract Article history: Accepted 17 May 2013 Available online 4 July 2013 Keywords: Scheduling Building information modeling (BIM) ifcxml Data exchange Critical path method Activity sequence The architecture, engineering, and construction industries have had rapid technological advancements over the last decade, particularly in the area of building information modeling (BIM). BIM stores all the information of a building and can be leveraged for many new and exciting applications including the generation of quantity takeoffs, 4D scheduling, and building simulations. The main objective in this study is to establish a framework for automating the generation of construction schedules by using data (e.g. spatial, geometric, quantity, relationship and material layer set information) stored in BIM. Using the extracted information, the proposed system in this research creates construction tasks, computes activity durations using available activity production rates, applies sequencing rules, and finally outputs a schedule. To demonstrate the functionality of this framework, a prototype system has been developed to import BIM representations with basic building elements such as slabs, walls, doors, windows, roofs, floors, and ceilings in two story buildings Elsevier B.V. All rights reserved. 1. Introduction Construction scheduling has come a long way in the last 20 years, a time when many field supervisors felt that using formal scheduling was irrelevant to day to day operations and a time consuming distraction [9]. Since then, it has become an integral part of most construction projects, but remains a time-consuming, error prone and tedious task done manually [5]. As a result a wealth of research was conducting investigating how the process of schedule generation could be improved by automating activity generation, duration estimation and determining sequence logic [5]. However, many of these efforts require substantial manual input including that of the physical model [1]. Recently with the technological advancement and prevalence of building information modeling (BIM) and 3D modeling in the architectural, engineering, and construction (AEC) industries new opportunities exist for improving scheduling processes. By combining the built-in intelligence of BIM with previous research efforts we can further advance the automation of schedules. Even today, scheduling is still mostly accomplished manually, which can be an extensive and very time consuming process. A reason Corresponding author. Tel.: ; fax: addresses: h.kim@uncc.edu (H. Kim), kyleand@umich.edu (K. Anderson), shdpm@umich.edu (S. Lee), John.Hildreth@uncc.edu (J. Hildreth). 1 Tel.: Tel.: Tel.: ; fax: for this is that this process remains insufficiently supported by software applications [14]. Since it is difficult to interact between scheduling software and BIM, many benefits of the benefits of BIM technology proposed in research papers remain unexploited [15]. Being able to exploit information stored in BIM to assist in generating schedules could help achieve significant time reductions in scheduling compared to traditional manual methods of scheduling. Previous research has demonstrated the feasibility of generating construction schedules for construction process by using the state of the art technologies including BIM, but to date, little work directly focused on this topic and has been able to successfully completely automate this process. Before BIM technology came into being, there were many attempts to automate the scheduling process. Cherneff et al. [4] and Zozaya- Gorostiza et al. [17] developed systems to integrate CAD with construction schedules using knowledge based systems. Although the research showed the potential to improve productivity in the AEC industry through their proposed approaches, data extraction remained a significant issue [11]. Later Fischer and Aalami [10] and Aalami and Fischer [1], built on this work and devised a method to generate activities and their sequences using construction method templates, CMMT (customizable construction model templates) by generalizing activity elaboration and sequencing knowledge. This allowed activities to be represented at different levels of detail as desired by the target group, but the process still required significant manual input. Several more recent efforts have attempted to use information stored in either 3D CAD models or BIM for processes related to /$ see front matter 2013 Elsevier B.V. All rights reserved.

2 286 H. Kim et al. / Automation in Construction 35 (2013) automated schedule generation. Tulke and Hanff [15] demonstrated the viability of using element quantities stored in BIM to generate durations for scheduling tasks using production rates. Their primary objective in this work was not on schedule creation, but rather on using this technique for expediting the 4D simulation process. Later Kataoka [12], conceived the concept of using simple 3D models to generate quantity takeoffs, schedules, and 4D visualizations by creating a structural planning process using the interpretable templates (SPLIT) system. This system takes simple building geometry and applies known construction methods to it and subsequently generates possible building construction configurations [12]. While innovative, this approach focuses on constructing different possible framing systems and comparing schedules generated by using different methods instead of generating schedules for designed structures from BIM. Tauscher et al. [14] proposed a system to semi-automatically generate schedules based on the data extracted from files using the Industry Foundation Classes (IFC) standard. In their work, they used case based reasoning (CBR) to determined task durations based on information retrieved from similar cases. However, the paper did not attempt to elaborate the details of actual outputs of their proposed system [14]. This previous research shows that there is a need for enhanced interoperability between BIM and scheduling software. The related research efforts in this field have demonstrated advancements in utilizing information in BIM or 3D models for scheduling related applications, but comprehensively leveraging of the intelligence of BIM for automated scheduling generation remains to be done. Therefore, in this paper we consider scheduling within the context of building information modeling (BIM) with an end goal of generating a construction schedule through automatic data extraction from a BIM file, focusing on automating physical model input. The individual goals of this research are as follows: firstly, extracting material, location and quantities for all individual elements from BIM and storing the data according to their unique location in the structure(s); secondly, generating activities and durations based on derived building elements, materials, and quantities; thirdly, developing sequencing for activities generated in the second step; fourthly, generating the output of a preliminary schedule in scheduling software compliant formats, and lastly, refinement process to facilitate the data exchange between BIM authoring and scheduling tools. In order to test the feasibility of the proposed approach, a prototype of the framework has been developed and tested in a case study. The BIM-based scheduling was applied to generate the schedule for two small sized buildings in the range of 6000 to 10,000 ft 2 with the basic building components such as slabs, walls, windows, doors, floors, ceilings, roofs and so on. The application can be extended to scheduling of a more complex building with necessary information and sequencing rules provided. The remainder of the paper details the development of the framework, presents a demonstration of feasibility, and concludes with a summary and limitations of the current research. 2. Developing the framework To address these issues a framework to generate schedules, given a set of sequencing rules, based on BIM has been suggested. The proposed process is deconstructed into five phases: A) construction of the BIM, B) parsing of BIM data, C) transforming of parsed data to activity data, D) generation of schedule, and E) refinement process (Fig. 1). During BIM creation, in addition to designing all the building elements, the user defines construction work zones (if desired) to prioritize construction of one area over another. This is a way to set construction order if multiple buildings exist or to decompose a large building into several smaller sections. The completed BIM is then exported in the XML variant of the open industry standard Industry Foundation Classes (IFC). IFC is a standard established by the BIM community to allow for model exchange between its various commercial software vendors [3]. The XML schema variant of IFC, ifcxml, is used for the demonstration in this research since it allows for access to more processing tools when compared to STEP files [8]. The prototype system, the implementation of the listed five processes written in the Ruby computer language, then reads and extracts relevant spatial, quantity, material, and relational information of all building elements in each designated work zone. Next, durations and resources related to those elements are calculated and assigned by using production rates stored in a database while considering the user defined sequencing rules. Durations are determined by using production rates from RSMeans [13]. Fig. 2 illustrates the previous descriptions of B) parsing of ifcxml data, C) transforming of parsed data to activity data, and D) generation of schedule in greater detail. This provides a detailed list of the informational requirements for this framework. Lastly, the system writes the schedule in the format of a Microsoft Project file, enabling the user to utilize built-in features in MS Project, such as calculations of early start/finish, late start/finish, Fig. 1. Methodology flowchart.

3 H. Kim et al. / Automation in Construction 35 (2013) A) Building Data Loading Part IFCXML Geometry Data Parsing Wall Geometry Property Parsing Get Wall Polygon Points Coordinates Get Wall Depths Get Wall Origin Coordinates Get Wall Story Coordinates Get Wall Direction Ratios Door/Window Property Parsing Get Door/Window Origin Coordinates Get Door/Window heights Get Door/Window Widths Foundation Property Parsing Slab\Ceiling\Floor Property Parsing Get Polygon Points Coordinates Get Depths Get Origin Coordinates Get Story Coordinates Get Direction Ratios IFCXML Material Data Parsing Door/Window Material Property Parsing Wall Material Property Parsing Slab/Ceiling/Floor Material Property Parsing B) Geometry /Material Information Geometry Data Wall Geometry Data Door/Window Geometry Data Foundation Geometry Data Slab Geometry Data Ceiling Geometry Data Floor Geometry Data Material Data Wall Material Data Door/Window Material Data Foundation Material Data Slab Material Data Ceiling Material Data Floor Material Data C) Generating a Schedule Activity Duration Sequencing RSMeans Daily Output Calculate / store schedling information Sequencing rules in Figure 9 & Table 1 Fig. 2. Overall scheduling process of data parsing, geometry/material information and generating a schedule. total/free floats, automatically identifying the critical path and so on. Each process will be explained in the following five sections. A. BIM model preparation The BIM model preparation process has a limited number of steps (Fig. 3). For this framework, the first step in the process is to obtain an architectural model built in a 3D CAD authoring program which has built-in exportation to ifcxml. Built in 3D CAD authoring software, the model is assumed to have basic building elements to the construction of the project such as slabs, walls, doors, windows, roofing, flooring, and ceilings. After designing these elements, created built-in materials and/or material layers will be assigned to each element. Assigning the materials to each element is critical. The names used for each element are later used as filters to identify proper production rates for activities. This makes it imperative to use names similar to, but not necessarily identical to the names used in the productivity database (how this works will be explained in greater detail in section C transforming of parsed data to activity data). Once all building elements have been prepared and materials assigned to each element, the user defines work zones for the ordering of construction. The building work zones are used to define the Fig. 3. Procedure during BIM creation. Fig. 4. ifcxml parsing flowchart.

4 288 H. Kim et al. / Automation in Construction 35 (2013) Fig. 5. Abbreviated view of IfcRelContainedInSpatialStructure in ifcxml. sequencing of construction areas for multiple structures or areas within larger structures (e.g. construct walls in different work zones). There are many possible ways to represent different zones in BIM. One simple method, and the one employed in this prototype, is to create a slab element representing the construction zone. All elements above and below this slab are considered to be part of this construction zone. However, if all building elements were drawn on the same slab, it would be necessary to separate the slab into multiple work zones in this process. Even though it might be an additional step in other applications, it worked well in the case study of the research. Next, the model is exported in ifcxml file format from the BIM authoring software. B. Parsing ifcxml data Information from the BIM now must be extracted from the ifcxml data and the proposed prototype takes the ifcxml representation of the BIM and reads it (Fig. 4). For the purposes of constructing construction schedules, a logical way to sort through all the project data and building elements is by story, and is done so through IfcRelContainedInSpatialStructure. IfcRelContainedInSpatialStructure relates elements to a particular location within the spatial structure of the project including the site, building, story or space and then provides a reference to where information about each element in that location can be found [2]. How building element information is related to the location being described by IfcRelContainedInSpatialStructure in XML is illustrated in Fig. 5. Elements are identified by type under RelatedElements in IfcRelContainedInSpatialStructure. Theprototype parses the references (the string listed in quotations after ref = (B) in Fig. 5) for each like element (e.g. wall, slab, window). The window references, for example, would be identified as i45046, i45173, i45427, and i45672 (C in Fig. 5). Using the reference ids the prototype systematically parses attribute information Fig. 6. Entity relationship diagram.

5 H. Kim et al. / Automation in Construction 35 (2013) Fig. 7. ifcxml mapping of element quantities through IfcRelDefinesByProperty. (location, shape, quantity information, and material information) about each element by following the paths from the reference id to the attributes (Fig. 6). For a typical wall, IfcWallStandardCase, several relevant pieces of information are required. Initially the prototype checks to see in what building area the item is located. Using the IfcWallStandardCase reference from IfcRelContainedInSpatialStructure, i44880, a series of iterations are processed to determine the location coordinates of the wall relative to the center point of the model, (0, 0, 0). In ifcxml, the coordinates for each item are written relative to itself, with the first coordinate usually identified by their two dimensional coordinates (x, y). Two sets of coordinates can be obtained from the wall element, the coordinates of each corner of the wall and the offset of these points from the center of the project. Since the element vertex coordinates are relative to the element, it becomes necessary to determine the element's actual distance to the center point of the project by summing the distance to the center of the project and the relative coordinates. The offset coordinates (x, y) illustrate the distance from the relative coordinates to the center point of the project. With this information the building area that the wall is located in can be determined by combining the relative and absolute coordinates of the wall and comparing them with the boundaries of the various building areas. For example, if the first vertex in the wall has a relative coordinate of ( 9, 0), combining this with its offset coordinates of ( 688.5, 31.0) means that this vertex is located at ( 697.5, 31.0) relative to the center point of the project. This pairing of data would be checked against the boundaries of each of the defined work zones until it is determined in which zone this building element is located. Next, the prototype parses the quantities of the referenced wall as shown in Fig. 7. Quantities relate to building elements through the relationship IfcRelDefinesByProperties. Here the prototype uses the wall reference to find the IfcRelDefinesByProperties that relates to it through its related objects reference id. The quantities of the wall are found using the path: IfcRelDefinesByProperties => IfcElementQuantity. Under ifcelementquanitity, individual element attributes include element width, height, length, gross side area, net side area, gross volume, net volume, and gross footprint area (Fig. 8). Lastly, material information must be gathered. Materials relate to the elements through the relationship IfcRelAssociatesMaterial. The method for gathering this information is similar to the previous, in that the prototype searches for IfcRelAssociatesMaterial with a related object with the reference id of the wall. Once it is found, the following path locates the layer set name of the materials associated with the wall: IfcRelAssociatesMaterial => IfcMaterialLayerSetUsage => IfcMaterialLayerSet => IfcMaterialLayer. For example, for wall i44880 the layer set is named Wall2 4+1/2 Gyp.Board. As a layer set may have only one material or more, as in this case, the prototype reads for each material layer and parses each individual material name and thickness that composes the layer set. At this point, data that can be compiled about the wall includes its composition, the building area it is located in, and all of its quantities and geometric configuration. C. Transforming of parsed data to activity data Each element's quantity and material information are stored along with its location (building area location and story). Based on the type of elements extracted, the prototype establishes the various activities. The next step is to identify what activities these elements are part and then to determine the durations associated with each activity in each building area on a per story basis related to these activities. Since in this phase of the research our objective is to demonstrate how information stored in BIM can be extracted and applied for use in automated schedule generation, we use a set of predefined activities for each building element, dependent upon the element location and material composition, and user defined resource availability quantities, which means that the proposed system requires for all the names of the building components to be pre-determined in generating the construction activities. Future extension of this work will focus on how to determine which activities should be generated for specific building elements based on a combination of information stored in BIM and user input. To generate the resulting activity durations, each individual element and its associating material data reference a well-established productivity data [13] for their construction. Fig. 9 Fig. 8. IFC elements diagram.

6 290 H. Kim et al. / Automation in Construction 35 (2013) Start Parse building element information and project data Identify Project Scope Categorize Activity Groups Identify Construction Phases Distribute Building Element Data Into Activity Groups Analyze Individual Element Information Quantity Information Material Composition Spatial Information Reference Productivity Rates Compute Activity Duration Apply Sequencing Rules Calculate and Store CPM Scheduling Information Early Start (ES) Total Float (TF) Early Finish (EF) Free Float (FF) Late Start (LF) Dependent Float (DF) Late Finish (LF) Critical Path (CP) Adjust (extend or shorten) Acceptable Result? (Verified by User) Yes Compile/Activity List No Export Activity List Schedule Critical Path: Network Diagram End Fig. 9. Generating a construction schedule. illustrates the overall process of generating a construction schedule. The prototype takes each building element and its information and references a created productivity database drawn from selected RSMeans values. The prototype checks for keywords in the material name and attributes from the element with the naming of items in the productivity database. If the name alone is not specific enough to obtain a rate, additional checks, such as for item dimensions, are run to obtain a production rate. This productivity is applied to the quantity, which may require unit conversions or count of the element, and the duration for its construction is calculated. This technique is applied to all of the like-items on the story and a total duration for the activity is computed. Table 1 Sequencing rules for task ordering [6]. Criteria in deciding the sequence of an activity Supported by Covered by Embedded in Distance to support 1. Install support first 2. Remove supported first 3. Install hidden first 4. Remove cover first 5. Embedded component first 6. Embedded component before, concurrently or after 7. Closer to support first 8. Least flexible first 9. One providing service first To illustrate the process of referencing the element database for a door installation, the name of the door and the dimensions are required. This information was previously captured along with the story location in their building area. To calculate the duration required for installing a door, a check is run against the name Wood Doors. This prompts the prototype to access a certain section of the productivity database, Wood Doors (in this case, a specific IDof from RSMeans unit cost index). Since the true dimensions of (80 36 ) is identified and the name does not include further description, the system defaults to standard 6-panel wood doors in (cost index from RSMeans). At this point, the productivity rate of 16 doors per day (daily output) is selected, yielding (1 door required/16 doors per day) = days and storing the value as the duration for a single door installation on this story within the defined building area. This process is automatically repeated for each door on the same building story in the same building area and a total time for the activity door installation on this story is recorded. This prototype uses productivity rates for combined activities to reduce the complexity of construction processes and the total number of story activities while maintaining the integrity of the produced schedules. Activity sequencing is based off a set of sequencing rules developed for the prototype. The rules govern the complete order of the construction processes. Task generation considers how the user defined the building

7 H. Kim et al. / Automation in Construction 35 (2013) and assumptions without considering a specific situation given in a project. A construction schedule generated through automatic data extraction can be modified and refined by experienced construction personnel who can take into account the specific situations that exist in a complicated real-world project. 3. Demonstration areas during the creation of the BIM. The building areas that the user defined during the model production are decomposed by story. The sequencing rules used in the algorithm are then applied to determine the sequencing per building area and for the total project. The rules were considered based on its structural consideration and have four different categories such as supported by, covered by, embedded in and distance to support as shown in Table 1 [18]. These rules are founded on the logic that activities such as slab construction require the completion of its vertical supports first. Also, in regard to embedded components, a wall must be constructed before the window or door may be installed in it. In addition, the rules consider that the amount of labor is limited in a given area at a set time. Once the crew has finished their activity in a given area, they can begin the same activity for the next building area upon completion of the required supporting activities. These rules, those listed in Table 1, are an example of a set of rules that have been applied and used in this prototype (Fig. 10); however, if desired it would be possible to develop and implement alternate sets of rules. D. Generate schedule After the activity quantities, names, durations, and sequences are defined, the output can be produced. The system exports activity list of data to the format of a Microsoft Project file. The activities are defined by building boundaries then by story. Each activity lists the activity name in the format of building area story level description, duration in days (assuming 40-hour workweek), start date, end date and predecessors. This also produces a Gantt chart and network diagram with the critical path indicated. By exporting the output in Microsoft Project format, all of the functions of this popular and powerful scheduling tool may easily be applied to the generated preliminary schedule. E. Refinement process Fig. 10. Implemented sequencing of activities. There can be the need to refine the results from the proposed BIM based scheduling process because the construction schedule generated by this research was created through generalized sequencing rules In order to test the prototype, a BIM was created using Graphisoft's ArchiCAD 15 [7]. The model consists of two buildings each representing a separate building area (Fig. 11). The first building area contains a 5850 square foot two story wood framed building. The second building area has a four story structure totaling roughly 9500 ft 2. Both structures have materials assigned to each element using premade composite structure fills from ArchiCAD. No custom layer sets have been used. Basic elements for construction compose each structure, including: walls, slabs, flooring, soffited ceilings, doors, windows, and roofing elements as shown in Table 2. Accordingly, the prototype parses through the ifcxml file for all the information described previously and produces a Microsoft Project output file. The schedule for the BIM consists of 58 activities. This includes 20 for the first building area and 38 for the second. Some activities listed include: cast-in place concrete for the slab on grade, curing, wall framing, and gypsum board installation. Table 3 shows the quantity take-offs arranged by building element (walls, slabs, doors, etc.). Each building element is listed by its unique id, pertinent metrics (for wall: net area of the desired dimensional surface, net volume, depth) and material layer set name. The schedule estimates that construction would take 129 working days (assuming a 40-h workweek), spanning from the starting date of October 10 in the year 2011 through the end date of April 4 of the following year. The critical path determined for the project was heavily dependent on the initial site work, foundation construction, wall framing and floor construction activities. Microsoft Project displays the projected critical path, visible here in the produced network diagram (Fig. 12). Such results could be expected based on the sequencing rules, due to the parameter to ensure that vertical supports are installed first. Table 4 shows the materials of each building component used in the case study. 4. Verification We verified the automatic data extraction process. Specifically, the verification was conducted to make sure that the data automatically extracted from a BIM using ifcxml technology did not lose/skip any data or material properties. Firstly, a visual inspection was conducted by reproducing the same model in a different software platform, SketchUP and confirmed that the original BIM drawn in ArchiCAD 15 and the one reproduced in SketchUp were identical in its geometry and locations of each building component. Secondly, the material quantity take-off list (as shown in Table 3), generated by the system, showed that the material properties were consistent as those of the BIM created at the beginning. On the other hand, the research considered only the basic building components such as slabs, walls, doors, windows, roofs, floors, and ceilings to propose a systematic approach. Therefore, the comparison between the result in the case study and real construction project results was not made due to the fact that our demonstration did not include all the building components that a real project would do. However, we intend to include more building components in the future research and verify the results in a more realistic way. 5. Conclusions Over the last number of years, BIM has changed the landscape of the AEC industry. With the influx of this new technology, many new

8 292 H. Kim et al. / Automation in Construction 35 (2013) Fig. 11. BIM in case study.

9 H. Kim et al. / Automation in Construction 35 (2013) Table 2 Abbreviated listing of activities, durations, start, finish, and predecessors. Task name Duration Start Finish Predecessors 1 Pre-structure phase 59 days Mon 10/10/11 Thu 12/29/11 2 Bldg. 1: Story 1: Cast-in-place concrete (slab on grade) 10 days Fri 12/30/11 Thu 1/12/ Bldg. 1: Story 1: Wall construction 3 days Fri 1/13/12 Tue 1/17/ Bldg. 1: Story 1: Gypsum board installation 4 days Wed 1/18/12 Mon 1/23/ Bldg. 1: Story 1: Acoustical tile installation 5 days Thu 1/26/12 Wed2/1/ Bldg. 1: Story 1: Ceiling finishings (paint) 2 days Thu 1/26/12 Fri 1/27/ Bldg. 1: Story 1: Wall finishings (paint) 5 days Thu 2/2/12 Wed 2/8/12 4, 5 8 Bldg. 1: Story 1: Door(s) installation 1 day Thu 2/9/12 Thu 2/9/12 6, 7 9 Bldg. 1: Story 1: Window(s) installation 3 days Thu 2/9/12 Mon 2/13/12 6, 7 10 Bldg. 1: Story 1: Floor finishing (carpeting) 6 days Tue 2/14/12 Tue 2/21/12 8, 9, 2 11 Bldg. 1: Story 2: Floor construction 6 days Wed 1/18/12 Wed 1/25/ Bldg. 1: Story 2: Wall construction 3 days Thu 1/26/12 Mon 1/30/ Bldg. 1: Story 2: Gypsum board installation 4 days Tue 1/31/12 Fri 2/31/12 12, 4 14 Bldg. 1: Story 2: Acoustical tile installation 4 days Tue 2/7/12 Fri 2/10/12 20, 5 15 Bldg. 1: Story 2: Ceiling finishings (paint) 2 days Tue 2/7/12 Wed 2/8/12 20, 6 16 Bldg. 1: Story 2: Wall finishings (paint) 5 days Mon 2/13/12 Fri 2/17/12 13, 14, 7 17 Bldg. 1: Story 2: Door(s) installation 1 day Mon 2/20/12 Mon 2/20/12 15, 16, 8 18 Bldg. 1: Story 2: Window(s) installation 4 days Mon 2/20/12 Thu 2/23/12 15, 16, 9 19 Bldg. 1: Story 2: Floor finishing (carpeting) 6 days Fri 2/24/12 Fri 3/2/12 17, 18, 11, 10 and exciting areas of research are now available. This new technology has been applied to many topics including scheduling. While there has been significant research on scheduling as it relates to 4D simulation in the context of BIM, relatively little work in academia has explored the possibilities of automated schedule generation from BIM. Table 3 Abbreviated produced quantity take-offs spreadsheet. File name Model for QT.ifcXML Building square footage 11,900 1 story Wall ID Net area Net volume Layer set type Material 1 Material 1 thickness Material 2 Material 2 thickness Material 3 Material 3 thickness (ft 2 ) (ft 3 ) (in.) (in.) (in.) i Wall Gypsum Insulation Gypsum 0.5 i Wall Gypsum Insulation Gypsum 0.5 i Wall Gypsum Insulation Gypsum 0.5 i Wall Gypsum Insulation Gypsum 0.5 i Wall Gypsum Insulation Gypsum 0.5 i Wall Gypsum Insulation Gypsum 0.5 i Wall Gypsum Insulation Gypsum 0.5 i Wall Gypsum Insulation Gypsum 0.5 i Wall Gypsum Insulation Gypsum 0.5 i Wall Gypsum Insulation Gypsum 0.5 Slab ID Net area Net volume Layer set type Material 1 Material 1 thickness Material 2 Material 2 thickness (ft 2 ) (ft 3 ) (in.) (in.) i /4\ Plywood_1 06 Plywood 1/8\Plywood Wood Frame Door ID Height Width Type Located in wall (in.) (in.) i D 1 Entrance side2 14 i9130 i D 1 Entrance side2 14 i15787 Total count 2 Height Width Window ID (in.) (in.) Type

10 294 H. Kim et al. / Automation in Construction 35 (2013) Regular activity Critical activity Extracted view of Network Diagram Fig. 12. Network diagram showing the critical path created from the prototype. This paper proposes a framework to be used for automated schedule generation from BIM. This system focuses on data exchange using a well-establish international data standard schema, ifcxml. To test the suggested framework, a prototype has been developed and applies the techniques described in the methodology and parses the pertinent spatial element and material layer data from the constructed BIM. Through the information gathered, the proposed BIM based scheduling approach is used to create construction tasks, calculate activity durations using productivity rates from a database, and apply sequencing rules. Once computed the prototype produces a preliminary schedule. In the demonstration presented in this paper, the prototype was successfully applied to a BIM consisting of two separate structures that were modeled with a basic building envelope and exported as an ifcxml file. Applying the prototype system to the produced file, a preliminary schedule was generated and able to be visible in various formats. Consequently, additional work can be conducted to further expand the robustness of the prototype. In the paper, a limited number of basic building components such as slabs, walls, doors, windows, ceilings, and floors were applied in generating the construction schedule in the case study. However, its possible application can be extended to all the building components with further detailing each building component in the process. Expansion of the set of elements that the system can recognize and parse information from the BIM is the logical next step in the development. Increasing the depth of materials and construction methods in the productivity database also requires further work. Another extension of this research could integrate the techniques used here in automatically linking a generated schedule's tasks to building elements for the creation of 4D building simulations. Further, in construction there are many possible ways to construct buildings and each individual element. In this prototype the activities associated with each element type have been predefined and determined based on specific extracted information for the element (e.g. location, material). More work is needed to identify practical ways to be able to determine which activities should be associated with each element while minimizing the use of default settings and user input. While the proposed methodology is able to quickly generate construction schedules, there are a few limitations observed. This methodology suffers from a limitation inherent to ifcxml, which is scalability and complexity [16]. The models that have been tested in the proposed process are relatively simple BIMs with limited details and are able to generate construction schedules in from under a minute to a few, but with increased complexity in models time required will increase, possibly significantly. In developing the BIM based construction schedule, this research utilized ifcxml technology, which consists of a very complex schema which requires a substantial time investment to learn. However, with Table 4 Materials of building components in case study. Type of component Material Daily output RSMeans reference Slab Concrete in place, including forms (4 uses), reinforcing steel, concrete, placement and finishing unless otherwise C.Y indicated Wall Concrete block, autoclave aerated concrete block, solid, , incl. mortar 600 S.F Wall finishing Sprayed on walls, two coats 1560 S.F Flooring Wood block flooring, end grain block flooring, end grain flooring, coated, 2 thick 295 S.F Ceiling Suspended acoustic ceiling tiles, fiberglass boards, film faced, 2 2 thick 625 S.F Door Wood door paneled, interior, six panel, hollow core, 1 3/8 thick molded hardboard, Ea Window Aluminum windows, incl. frame and glazing, commercial grade, stack units, casement, opening 10 Ea Roof framing Lightweight framing, angle framing, field fabricated, 1/2 1/2 1/8 200 L.F Formwork Forms in place, elevated slabs, flat plate, job-built plywood, to 15 high, 4 use 560 S.F Insulation Rigid insulation, 2-1/2 thick, R S.F Painting Walls and ceilings, concrete, drywall, or plaster, primer or sealer coat, smooth finish, brushwork 1150 S.F

11 H. Kim et al. / Automation in Construction 35 (2013) the introduction of simplified XML schema, released by buildingsmart in the year 2012, the burden of complex data parsing is expected to be reduced. References [1] F. Aalami, M. Fisher, Construction Method Models: The Glue Between Design and Construction, ASCE, Conference proceeding of Computing in Civil Engineering, 1998, pp [2] buildingsmart, IFC2x Edition 3 Technical Corrigendum 1, (October 16, 2012). [3] buildingsmart, IFC Specification Technology, specifications/ifc-overview/ifc-overview-summary2011, (December 16, 2012). [4] J. Cherneff, R. Logcher, D. Sriram, Integrating CAD with construction schedule generation, Journal of Computing in Civil Engineering 5 (1) (1991) [5] N. Chevallier, A.D. Russell, Automated schedule generation, Canadian Journal of Civil Engineering 25 (6) (1998) [6] D. Echeverry, W. Ibbs, S. Kim, Sequencing knowledge for construction scheduling, Journal of Construction Engineering and Management 117 (1) (1991) [7] Graphisoft, Graphisoft ArchiCAD 15, ac15/pdf/archicad15_brochure.pdpdf2011. [8] M. Halfawy, T. Froese, Component-based framework for implementing integrated architectural/engineering/construction project systems, Journal of Computing in Civil Engineering 21 (6) (2007) [9] C. Hendrickson, T. Au, Project Management for Construction: Fundamental Concepts for Owners, Engineers, Architects, and Builders, Prentice Hall, NJ, [10] M. Fischer, F. Aalami, Scheduling with computer-interpretable construction method models, Journal of Construction Engineering and Management 122 (4) (1996) [11] T. Froese, M. Fisher, F. Grobler, J. Ritzenthaler, K. Yu, S. Sutherland, S. Staub, B. Akinci, R. Atbas, B. Koo, A. Barron, J. Kun, Industry foundation classes for project management a trial implementation, ITcon 4 (1999). [12] M. Kataoka, Automated generation of construction plans from primitive geometries, Journal of Construction Engineering and Management 134 (8) (2008) [13] RSMeans, Building Construction Cost Data, Construction Publishers & Consultants, Kingston, MA, [14] E. Tauscher, E. Mikulakova, K. Beucke, M. König, Automated generation of construction schedules based on the IFC object model, ASCE, Computing in Civil Engineering 2009, Austin, TX, [15] J. Tulke, J. Hanff, 4D construction sequence planning new process and data model, CIB-W78 24th International Conference on Information Technology in Construction, Maribor, SLV, [16] J. Zhang, A. Webster, M. Lawrence, M. Nepal, R. Pottinger, S. Staub-French, M. Tory, Improving the usability of standard schemas, Information Systems 36 (2) (2011) [17] C. Zozaya-Gorostiza, C. Hendrickson, D.R. Rehak, P. Lim, Construction Planex: a knowledge intensive planner for construction projects, 5th International Symposium on Robotics in Construction, 1988, pp [18] D. Echeverry, A. Beltrán, Bar-code Control of Construction Field Personnel and Construction Materials, Proceedings of Computing in Civil EWngineering, 1997, pp