Building information modeling and discrete event simulation: Towards an integrated framework

Automation in Construction 44 (2014) 73–83

Contents lists available at ScienceDirect

Automation in Construction journal homepage: www.elsevier.com/locate/autcon

Building information modeling and discrete event simulation: Towards an integrated framework Weizhuo Lu ⁎, Thomas Olofsson Construction Engineering and Management, Department of Civil, Environmental and Natural Resources Engineering, Luleå University of Technology, 97187 Luleå, Sweden

a r t i c l e

i n f o

Article history: Received 3 July 2013 Received in revised form 30 March 2014 Accepted 5 April 2014 Available online xxxx Keywords: BIM DES Change Intelligence Maintenance

a b s t r a c t The development of a realistic Discrete Event Simulation (DES) model needs the complete specification of the interdependencies between activities and resources. Also, the maintenance of an existing DES model is challenging especially when changes in the logical relationships between activities, resource allocation and design need to be considered. The process of development and maintenance is time-consuming, error-prone and it restricts the application of DES within the construction industry. In this research, a Building Information Modeling (BIM) and DES framework is proposed to enable the implementation and integration of DES in the planning and follow-up of construction activities. The framework consists of: (1) A building information modeling process that exports material quantity take-offs, schedules and required resources to a relational database and (2) an intelligent simulation engine that automatically reads information from the database at the start of each simulation run. This implies that changes in the building information modeling process, such as design modification, different resource allocations and alternative construction methods can be explored without manually checking and re-formalizing the simulation model. A preliminary prototype has been developed by using the proposed BIM– DES framework. The initial results show that the proposed BIM–DES framework reinforces both elements by providing valuable additional information. BIM provides the product and process information to DES, facilitating the building and maintenance of the DES model; the DES model evaluates the construction performances and provides valuable feedback to the BIM process for decision support. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Construction projects are usually delivered within an uncertain environment in which resources and activities interact with each other in a complex manner. The uncertainty that exerts tremendous influence on project performances stems from the inherent variations in activity duration, delivery of material, machine failure and workforce productivity, which are often attributed to the temporary nature of the project organization, on-site production and loosely-coupled supply chains [1,2]. Traditional deterministic methods cannot model and manage the level of uncertainty involved in construction projects. Discrete event simulation (DES), which explicitly incorporates uncertainty, has been used as an effective approach to better capture the complicated interactions and uncertainties found in construction operations [3]. While the benefits of using DES as a decision support tool have been recognized, it has not been widely adopted by the construction industry [4]. One of the reasons for this lack of implementation is the amount of manual work needed to specify and maintain the interdependencies between activities and resources in the construction supply chain [5,6]. In particular, when there are changes to logical relationships, resource allocation or ⁎ Corresponding author. Tel.: +46 920 492866; fax: +46 920 49 28 18. E-mail address: [email protected] (W. Lu).

http://dx.doi.org/10.1016/j.autcon.2014.04.001 0926-5805/© 2014 Elsevier B.V. All rights reserved.

design, the simulation model has to be updated and re-formalized to accommodate the new conditions. Small changes to the input of the simulation model often lead to extensive manual modifications [7]. This process is time-consuming, error-prone and restricts the application of DES to the construction industry. A possible solution for improving the flexibility of a simulation model is to make the necessary input data variable by linking it to a database [8]. On the other hand, as will be discussed later, the data captured by Building Information Modeling (BIM) can be used further by other applications [9]. BIM represents the processes and activities of development and uses a computer-generated model to simulate the planning, design, construction and operation of a facility [10]. The resulting model, a building information model, is a data-rich parametric digital representation of a facility, from which relevant data needed to support construction, fabrication and procurement can be extracted and analyzed [11]. Thus, BIM has the potential to provide a simulation model with design and planning data [7,12]. In this study, a BIM and DES framework is proposed that enables the implementation and integration of DES in the planning and follow-up of construction activities. The purpose of the integrated BIM–DES framework is twofold: BIM provides the product and process information for the DES, which facilitates the building and maintenance of the DES model; the DES model evaluates the construction performance and

74

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83

Fig. 1. The integrated framework of BIM and DES.

provides valuable feedback and decision support to the BIM process (the processes of Building Information Modeling). The remainder of the paper is structured as follows. In Section 2, the authors review previous studies of DES and BIM related to the research work and describe the necessary steps for an integrated framework. In Section 3, the integrated framework between DES and BIM is proposed and modeled with function model methodology (IDEF0) [13]. In Section 4, a preliminary prototype of the proposed BIM–DES framework is developed. A summary of the contributions of the proposed framework and discussions about future research are presented in the final section of the paper. 2. Related research 2.1. DES in construction DES is the modeling of systems in which the state variable changes at a discrete set of points in time [14]. It has been adopted as an effective technique in understanding the behavior of systems and evaluating

various strategies for their operation. Since the development of CYCLic Operations Network (CYCLONE) [15], DES has been used to develop computer-based simulation models of construction projects in order to analyze and optimize their behaviors [16]. After the introduction of CYCLONE, a number of construction simulation systems were developed, such as STROBOSCOPE [17], Simphony.net [3], RiSim [18], and SDESA [19]. These simulation systems provide useful tools for project managers to replicate the dynamic interactions between resources and activities in order to evaluate overall performances and to produce a more reliable prediction of a construction system, taking into account the presence of inherent uncertainty and unforeseen conditions [20]. Despite its advantages, the practical applications of DES in construction are limited. The large number of input data required to build a simulation model has been one of the reasons restricting the use of DES in the construction industry [21]. Manual data entry is not only timeconsuming but also an error-prone process [22]. A study from the National Institute of Standards and Technology (NIST) reported [23] that the lack of interoperability between individual systems accounted for a substantial increase in construction costs. Some approaches have

Fig. 2. The relational database diagrams.

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83

been proposed to use the available data to facilitate the building of a DES model. Yilin Huang et al. [6] argued that computer-aided design (CAD) data can provide a relevant source of data to enable data-driven simulations. Robertson and Perera [24] used a corporate business system as the data source for simulation and demonstrated that the approach reduced the need for manual data entry. Randall and Bolmsjo [25] presented a database-driven simulation approach in which an ERP (Enterprise Resource Planning) database was used to reduce the development time of DES models for the construction of large factories. Also, data-driven simulation means that the DES model can be parameterized by providing data through a set of, for example, data forms, tables and spreadsheets [26]. Jeong et al. [27] argued that development of a databasedriven simulation model is feasible and useful to automatically provide different parameter values for a simulation model. Xu et al. [12] extracted product information from a CAD design into a simulation model in order to calculate earthwork quantities and enable the user to feed the simulation model with input data. From previous studies, it has been shown that data-driven simulation provides a possible solution to speed up the definition of the DES model. However, there is no systematic and formal framework to investigate how to utilize the information contained in a BIM to facilitate the building of a DES model. This is surprising, considering the increasing development and use of BIM in construction projects where more and more life cycle information is used. This information can also be used in the development of a DES model of the project. Changes caused by unanticipated site conditions, material unavailability, or new client requirements are common in the construction industry [28]. Such changes include alterations to the logical relationships between activities, changing resource allocation and the modification of the design of the building either at the design stage, the planning stage of construction, or even when construction is already underway [29,30]. Changes are a major cause of delays, cost overruns and performance deviations from expected outcomes. When a change is proposed, the expected impact of the change on the project has to be assessed. Such an assessment has to take into account: (a) the direct impacts of the changes and (b) the indirect impacts on other subsystems or other project elements, such as the construction tasks, resource availability and project duration [31]. However, the large number of elements and the complexity of the dependent relationships that exist between these elements make it difficult to identify the impact that a change will have [31,32]. This is also true from a DES practitioner's view, where one of the most significant challenges is how to coordinate and adapt the simulation model to changes occurring in a construction project. A conventionally constructed DES model needs to be manually updated when changes occur. Hence, the maintenance of the model can be inefficient and prone to error. Therefore, the simulation model needs to be adaptive to changes in data, process interdependence and construction methods. If synchronization support cannot be provided when input conditions change, or if different scenarios cannot be easily explored without manually re-formalizing the simulation model, DES will not be adopted by the construction industry [4]. For the assessment of changes in the design and construction process, the simulation model should be easily adaptable and provide evaluation of the impact of the changes [8]. Previous studies have not considered the impact of changes on the maintenance of simulation models. Motivated by previous research on database-driven simulations, the authors have identified BIM as a possible source of data to facilitate the building of a DES. Also, it is proposed to solve the challenge of coordinating the DES model with the frequent changes that normally occur in construction projects by storing relevant BIM information in a continuously updated database.

Fig. 3. CARS model (Adapted from [54,56]).

or Virtual Design and Construction (VDC) [35]. Fischer and Kunz [35] defined the use of VDC as “the use of integrated multi-disciplinary performance models of design-construction projects, including the product (a building), organization of the design–construction–operation team, and work processes, to support explicit and public business objectives”. Li et al. [36] developed Construction Virtual Prototyping (CVP), which integrates product, process and resource models of construction projects to support the construction planning. A Building Construction Information Model (BCIM) has also been proposed as an extension of BIM to create the project schedule [37]. To create the project schedule, components in BIM contain timing data that indicate the start and finish time and specify the process information for constructing a particular component [38]. The information can be attached to the objects in BIM, including geometry details, take-off, construction methods specification, schedule and cost estimates, as well as providing the advantages of being able to insert, extract, update and modify digital data for analyses [39,40]. BIM-based quantity take-offs are used to generate input data in order to calculate the execution times for construction processes [41]. In order to use BIM as the source of information for DES, BIM must contain both product (building) and process information (planned construction activities required to realize the product).

2.2. BIM at the construction stage The process of using BIM models to improve the planning, design and construction process [33] is also referred to as nD modeling [34],

75

Fig. 4. The product and process information carried by task attributes.

76

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83

Fig. 5. Intelligent task in DES.

In addition, BIM provides the ideal solution to the synchronization of design modification. Traditionally, changes are managed manually. Manual methods of communicating changes among affected parties have proved time-consuming and often ineffective. In BIM, late changes in the design of building components need to be updated by using the parametric intelligence and rules of the BIM authoring tool. Crosssystem updates can be checked and coordinated in the BIM, allowing any change to be accurately reflected and coordinated [40]. By further linking the updated design with the estimation and scheduling process, a BIM can provide rapid feedback on the implications of changes of quantity take-off information [40]. The ability of BIM to coordinate and update changes in design and quantity information is utilized in the proposed integration framework. In this way, the proposed framework uses BIM to consider design changes and, as a result of those alterations, any changes of quantity take-off can be propagated from the BIM

to the DES model. Other unanticipated changes, such as late delivery of materials and breakdowns, are incorporated into the DES model. BIM has not been fully integrated with discrete-event construction simulation. The available information from BIM is not used in such simulations, which makes the generation of a simulation model inefficient [42]. Most current integration efforts using BIM have been related to the analysis of energy use, lighting, cost estimation and scheduling. However, previous research has identified the potential for utilizing the quantitative data contained in the BIM model to a greater extent [9]. BIM has the potential to facilitate the building of simulation models by providing essential input data from the design, bill of quantities and planning information, such as schedules, delivery dates and available resources [7,12]. With the increasing use of BIM from the design to the construction phase, there is a new opportunity to integrate BIM and DES for use in daily planning and control by contractors and subcontractors.

Fig. 6. The process of extracting information from BIM.

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83

77

Fig. 7. Snapshot of the quantity take-off.

3. An integrated BIM–DES framework Fig. 1 shows the proposed integrated BIM–DES framework, in which the process of generating and managing information is described by using IDEF0. There are five elements in the IDEF0 functional model. The activity (or function) is represented by the boxes; inputs are represented by the arrows pointing into the left-hand side of an activity box; outputs are represented by arrows pointing out the right-hand side of an activity box; the arrows pointing into the top portion of the box represent constraints or controls on the activities; finally, the arrows pointing into the bottom of the activity box are the mechanisms for carrying out the activity. A detailed explanation of IDEF0 can be found in [13]. As shown in Fig. 1, the process of integrating BIM and DES starts with the BIM model produced by architects and design engineers from which the product information can be extracted to generate constructionspecific information, such as the construction-level quantity take-off and scheduling. 3.1. BIM design The framework starts with the development of a parametric 3D computer-aided design model for the product according to the owner's

requirements and in accordance with the local regulation and standards. By using BIM authoring tools, the output of the BIM design is the explicit representation of the product to build. The building parts or components of the model need to be classified to support the transfer of information to the construction stage. In Sweden, two classification systems are used by the construction industry to sort and identify building components in the planning and cost estimation software SBEF and BSAB. SBEF is the old Swedish classification system, and BSAB is the new system which will, ultimately, replace the SBEF system in the Swedish construction industry [43]. However, both systems are still in use by Swedish planning and cost estimation software. 3.2. Construction quantity take-off Usually, architects and design engineers do not consider the process information in the BIM, such as the means and methods to realize the project, which means that the BIM model is mostly used in the design stage. However, by extending the use of BIM from design into the construction stage by linking the quantity take-off of building components with construction recipes, BIM then has the ability to estimate the resources needed in the construction supply chains [44]. The construction recipe maps components to the required representation of

Fig. 8. Snapshot of the schedule and resource information.

78

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83

Fig. 9. Relational database.

the construction steps and resources [45], as well as defining what operations and resources are needed for construction of the component on-site or for the installation of prefabricated components [40]. The recipe holds information about how the components are built and consists of the separate work tasks and resources to construct a specific component. By linking recipes with components, the extended BIM encompasses not only the product information (design) but also the cost estimate and planned construction information in order to realize the product (tasks and associated resource). Normally, temporary elements are not defined in the BIM (e.g. formwork). However, the quantities of such elements can usually be calculated based on the component properties [46]. The database characteristics of a BIM provide the ability to extract quantities in different forms. For example, a concrete column could be linked to a recipe consisting of three tasks: formwork, rebar and concreting [47]. The column type and its gross surface area can be used to compute the type and quantity of formwork and the associated labor and equipment requirements in the construction process. A specific type of building object can have different recipes i.e. the tasks and resources needed to construct the object. For example, some columns have a recipe for in-situ concrete pouring with rebar inside of a wood form while others have a recipe using a precast concrete column, fabricated off-site [48].

Fig. 10. BIM–DES task.

3.3. Scheduling As recipes define the required construction steps and resources, they can be used to generate a task list [40,49,50]. In addition, by adapting a location breakdown structure (LBS) [51], the location where the task happens and quantity per location can be specified. The schedule can then be generated based on the chosen recipes and location-based quantity take-off, related to the construction site and resource constraints [51,52].

3.4. DES As shown in Fig. 1, after the generation of the schedule, the tasks in that schedule and associated product and process information are exported to a relational database in which a primary key and a foreign key are used to represent the logical link between tables. Relational database schemas are very flexible. Individual relationships and tables can be added, modified and removed without disturbing the rest of the schema [53]. In the relational database, tasks' names are set to be the primary key. Hence, the task is the key to retrieving the corresponding information from the relational database. As seen from the relational database diagrams (see Fig. 2), 5 tables have been created to hold product and process information about the construction project. The table “Sequence of

Fig. 11. BIM–DES specification.

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83

79

Fig. 12. The information read by the DES model.

locations” stores the information about the construction sequence of tasks. The sequence of construction activities can be changed if the order does not violate the logical constraint. “Material supply” contains the material name, arrival time and quantity of the delivery. “Resource” stores the required resource for the task including machine and workforce requirements. “Bill of materials” contains the location-based quantity take-off data from BIM for the required materials for the execution of the task. “Summary” summarizes the product and process information for the construction, including the scheduled start and finish time. After the information has been exported from the BIM to the database, the DES intelligently reads the task information in the dataset. By intelligently, we mean that the DES task knows the required resource, materials and preceding activities for its execution. The task contains an information packet carrying the required product and process information for building components/elements when these tasks enter the DES model. The generated information descriptions are based on Object-Action-Resource (OAR) ontology proposed by Darwiche et al. [54]. OAR ontology effectively presents the design of a product and the construction information needed to define an activity's work scope for that design [55]. It was extended to the Component, Action, Resource and Sequencing (CARS) model to describe who (resource) is doing what (action) when (time) and where [56], see Fig. 3. The CARS model has been used to integrate process and product information (see Fig. 4): 1) Product information such as 3D geometric information, quantity take-off and location 2) Process information such as resources, construction task, schedules and construction sequence The quantity take-offs are physical quantities of design components, such as length, area and weight of rebar, as well as the volume of concrete beams. The required resources in the process information define the materials, the workforce, the equipment, as well as the process quantity take-off for temporary elements, such as areas of formwork [57]. Activity durations are typically calculated as the quantity take-off from the BIM multiplied by the productivity. Fig. 5 shows the work flow of an intelligent task when the activity enters the DES model. First, the task broadcasts the resources (machinery, workforce and materials) required to perform its operations and competes with other tasks in the schedule of access for available resources, in order to finish their individual operations. The task also consumes materials in order to produce components necessary for the subsequent activities. Each resource receives this broadcast message and checks its own state to decide whether it can service the request. If the resource has the capacity to perform the operation, it sends out a message to the task. The current task is on hold until all required requests are satisfied, including the finishing of predecessor tasks. After the task had been finished, it will be pulled from the schedule and marked as completed. In the BIM–DES framework, when a task has been completed, simultaneously, a “virtual material” is produced to

indicate the status of the task. Whenever a task is completed, all tasks that have not yet begun are checked to determine whether the prerequisites for starting (status of preceding tasks and the availability of the resource) are fulfilled. This is repeated until all the tasks in the schedule have been accomplished. The following statuses for the resource are used: (1) idle: the machine/workforce is idle and waiting for the next operational request, (2) operation: the machine/workforce is processing a task and (3) breakdown: the machine has broken down and cannot process any tasks. The proposed DES model takes into account the available resources, the interdependencies between individual tasks and simulates the dynamic and interactive relationships between tasks, workforce, machinery and materials in the material supply chain. Messages between tasks, resources and the surrounding DES environment are passed through events, properties and states. States dynamically define values which change at discrete points in time throughout the execution of the simulation model, such as the state of workers and machines: idle, busy and breakdown. An event is a notification that something important has happened, such as the event that a machine has accepted the seize request sent out by a task, or a task has been completed. Properties are the input parameters associated with the object that can be specified by the user interface; they are static values (e.g. processing time, bill of materials) which do not change during the running of the simulation model. The BIM provides information about what to build (product) and how to build it (process) in order to deliver the product. The DES model incorporates the specifics of the construction site environment, such as the variation in the materials supply chain, activity duration and probability of machine failure. The data read by DES are imported from BIM to the DES model via the relational database, in which different tasks share resources, consume materials, produce components and transport materials. 4. Preliminary prototype The preliminary prototype of the proposed integrated framework has been implemented and developed by using Autodesk Revit Structure as the BIM authoring tool, Tocoman iLink as the quantity take-off tool, Consultecs Bidcon and Plancon as the construction cost estimation and planning software and Simio as the DES tool. Fig. 6 shows the Table 1 Description of three construction alternatives. Alternative

The cast-in-place concrete slab is divided into 4 locations and three independent crews – carpenter, ironworker and concreter – move through these locations sequentially.

1

Pouring concrete with crane and bucket, crane up-time between failure follows an Exponential (100) hours distribution. Pouring concrete with pump, order formworks half an hour before the operation is scheduled to start. As alternative 1 with design modifications in location 3 and location 4

2 3

80

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83

Fig. 13. Exploring alternative construction methods.

process steps: (1) the development of the BIM design with classified building objects, (2) quantity take-off from the BIM design and cost estimation using construction recipes, (3) scheduling of activities based on quantities per work location, construction recipes and resource constraints and (4) performance evaluation using DES. Simio supports the modeling of both discrete and continuous systems, along with largescale applications based on systems dynamics and agent-based modeling [58]. We customized Simio to the specific requirements of the study. Fig. 7 shows a snapshot of the quantity take-off from BIM. Note that the quantity take-off (Mängd) is the total quantity which includes four locations. Fig. 8 shows a snapshot of the schedule and resource allocation of the task. This example involves the construction of a cast-in-place concrete slab for a residential building project. The construction of an in-situ concrete slab normally consists of erecting the formwork, installing the reinforcement and placing the in-situ concrete. Due to the limited storage area, the reinforcing bar and framework are transported to the construction site according to the material delivery plan: what type of material, when and how much should be delivered. The actual arrival of materials onto the construction site occurs according to its scheduled date 25% on time, 1 day early 25% of the time and 1 day late 50% of the time. The probability of the arrival time was based on typical variations as described in interviews with the project manager; this probability can be changed according to the specific project situation. Readymixed concrete is transported to the construction site by a concrete mix-truck, the load capacity of which is approximately 10 m3 of wet concrete per truck. The cast-in-place concrete slab is divided into 4 locations and three independent crews – carpenter (formwork), ironworker (reinforcement) and concreter (pouring concrete) – move through these locations sequentially. As shown in Fig. 9, a relational database is used to store relative product and process information for the task including scheduled start time, scheduled due time, the bill of materials (BOM), resources, material supply and sequence of location. In the relational database, BOM defines the component materials and required quantity for the construction task. For example, for pouring concrete at location 2, the materials include 95 m3 concrete estimated from the BIM quantity take-off. In addition, by including the preceding tasks in the BOM, the logical relationships between tasks are maintained. The pouring of concrete at location 2 begins only when the placement of reinforcement at location 2, the pouring of concrete at location 1 and 95 m3 concrete are seized by task 2. The resource table specifies

the required number of machines and workforce for the task (e.g. crane, concrete truck-mix and concreter). The material supply shows what type of materials, when and what quantity will be delivered to the construction site. The sequence of location assigns the construction sequence to the task. In the BIM–DES framework, the BIM–DES specifications need to be defined. The first is “Reading information from database”, which means that the task reads corresponding information from the relational database and assigns the arrival time of the task (see Fig. 10). The second is the BIM–DES specification (see Fig. 11). This interface defines how the processing time is estimated when the task enters the work locations i.e. the quantity take-off from BIM (workload) is multiplied by the processing time per unit. This implies that if the material quantity take-off has been changed due to a design modification, such changes can be updated and reflected directly into the DES model. Note that the processing time per unit is calculated by using Swedish productivity statistics [59]. The construction method and consumed material quantity also need to be defined in the BIM–DES specification. The information read by the DES model is shown in Fig. 12. Table 1 shows the three construction alternatives. The managers compare the schedule reliability of these different alternatives to mitigate the associated risk. Specific changes, such as alternative construction methods, can be investigated and examined to see how they affect the schedule reliability. How the change affects the completion time of the project is an important issue to be considered before changes can be approved and implemented [60]. The conventional way is to build two simulation models in order to compare two of these construction alternatives. For pouring concrete with crane and bucket, a CYCLONE simulation needs to build the crane cycle, concrete mix-truck cycle, materials order cycle, ironworker cycle, carpenter cycle and concreter cycle in which the full range of possible states associated with each relevant cycle must be identified. All of these simulation elements and their interdependencies must be specified and integrated manually to formalize the simulation model. For alternative 2, a separate model has to be developed to include the pump cycle and incorporate the altered process logical relationship. The adoption of the process interdependence or the construction methods lead to extensive modifications to the original simulation model if a traditional approach is used [7]. In the BIM–DES framework, however, the managers can explore alternative construction methods and process interdependence by simply

Fig. 14. Changing process interdependence.

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83

81

Fig. 15. Animation and visualization of the simulation model.

changing the resource and logical relationships in the database. The task entity can automatically read this updated information and send out the seize requests to the relevant resource. Fig. 13 shows an example of the dialog: the resource for the task can be manipulated by replacing the resource for pouring concrete from crane and bucket to concrete pump. Furthermore, a change in process interdependences is easy to implement by changing the component materials of the specific construction task in the database. As shown in Fig. 14, the sequential interdependence between pouring concrete at location 4 with the other construction tasks can easily be changed if the logical relationships are satisfied. Hence, the BIM–DES framework facilitates the testing of “what if” scenarios in the search of alternative modifications to the construction process. Based on the BIM–DES framework, managers can experiment with, and quickly evaluate, a number of process alternatives. Also, design modifications are common even when the construction process has already commenced. The BIM–DES framework provides a mechanism to propagate design modifications from the BIM to the DES. The impact of the design changes on the progress and construction performance can simply be evaluated by updating the BOM list in the database. Fig. 15 shows the simulation model in which the managers can visualize the progress of the project. Benchmarking uses the target due times shown in Fig. 12; Fig. 16 summarizes the schedule reliability of each construction tasks for three alternatives. For alternative 1, the schedule reliability of the construction task (placing reinforcement on location 4) is 63.65%, while tasks (pouring concrete from location 1 to location 4) have even lower schedule reliabilities when the impacts of variability from the preceding tasks are considered. For alternative 2, compared with alternative 1, the schedule reliabilities of placing reinforcement and pouring concrete have been greatly improved. For alternative 3, the design modifications result in a reduction of schedule

reliability of pouring concrete on location 4 from 80.18% to 19.92%. To deliver the project on time, the managers need to reconsider the implications of design modifications and adopt alternatives to enhance the schedule reliability.

5. Conclusions Support for efficient integration between simulation models, other software and tools is believed to be the future of DES. In this paper, a BIM–DES integrated framework is proposed and developed, in which BIM provides the product and process information to DES which facilitates the building and maintenance of the DES model; the DES model provides a valuable addition to the BIM as it can simulate the dynamic interdependencies and interactions between planned construction tasks. The proposed integrated framework between BIM and DES reinforces both approaches by providing additional benefits which are otherwise impossible when they are used separately. The reinforcement provided by the framework can be highlighted as follows: ➢ The integrated BIM and DES framework provides a more efficient platform than the traditional simulation approach, as the manual specification and maintenance of the simulation model is substituted with a dynamic database read by the DES model. ➢ The integrated BIM and DES framework can evaluate the construction performances of alternative designs, different resource allocations and alternative use of construction methods without manually checking and re-formalizing the simulation model, meaning that alternatives in the building information modeling process can be explored.

Fig. 16. The comparisons of schedule reliability with a confidence level of 95% and 1000 replications.

82

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83

The estimation of the “quality” of a developed schedule (good or bad) is normally based on experience. The proposed DES–BIM framework can make a more detailed estimation, taking into account the variation in productivity of the workers, the capacity of the supply chain and the uncertainty on the construction site. The framework can also be of use in virtual design and construction classes to support students exploring the relationship between products and processes in scheduling exercises [45]. It must be easy to change and re-evaluate a product or process which is why the strategy of using BIM's capacity for coordinating changes and intelligent DES modeling is adopted in this study. Despite the fact that the integration of BIM and DES provides an effective tool for the integration of product and process models, there are some limitations of the preliminary prototype. There is still some information that needs to be entered manually in the BIM process e.g. productivity statistics such as labor and equipment rates [59], the total number of laborers and the machine failure probability. Databases, such as Sage Timberline RSMeans Commercial Knowledgebase, can be connected to the proposed integrated framework to semi-automate this process. Being a pilot study, the prototype was designed to demonstrate the potential of this integration framework between BIM and DES, which will be further tested in a future study of a relocation project in the city of Kiruna in Sweden. We are aware of several research applications which might be helpful in our future studies, such as the demonstration of how to use templates to extract product and process information from BIMs by Konig et al. [7]. Marcel and Avi [61] represented the available construction methods (materials, labor, equipment, production rates) for a project in a matrix and embedded them in the BIM. For industrial applications, the prototype needs to be developed further to simplify the exchange of information between the BIM and the DES application. The integrated BIM–DES framework deals with changes of product and process, and how these changes affect the project performance. Since changes in managerial strategies and actions can significantly impact on the project performance, system dynamics (SD) [62], a method which analyzes and examines the effectiveness of managerial strategies and actions within a changing project environment, may also be incorporated into the proposed framework in future studies. Acknowledgments The project was financially supported by the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS). References [1] R. Vrijhoef, L. Koskela, The four roles of supply chain management in construction, Eur. J. Purch. Supply Manag. 6 (2000) 169–178. [2] A. Segerstedt, T. Olofsson, Supply chains in the construction industry, Supply Chain Manag. Int. J. 15 (2010) 347–353. [3] D. Hajjar, S.M. AbouRizk, Unified modeling methodology for construction simulation, J. Constr. Eng. Manag. 128 (2002) 174–185. [4] S. AbouRizk, Role of simulation in construction engineering and management, J. Constr. Eng. Manag. 136 (2010) 1140–1153. [5] A. Skoogh, A methodology for input data management in discrete event simulation projects, Proceedings of the 2008 Winter Simulation Conference, 2008, pp. 1727–1735. [6] Yilin Huang, M.D. Seck, A. Verbraeck, From data to simulation models: componentbased model generation with a data-driven approach, Proceedings of the 2011 Winter Simulation Conference, 2011, pp. 3719–3729. [7] M. Konig, C. Koch, I. Habenicht, S. Spieckermann, Intelligent BIM-based construction scheduling using discrete event simulation, Proceedings of the 2012 Winter Simulation Conference, Berlin, 2012. [8] D. Steinhauer, M. Soyka, Development and applications of simulation tools for oneof-a-kind production processes, Proceedings of the 2012 Winter Simulation Conference, Berlin, 2008. [9] R. Jongeling, J. Kim, M. Fischer, C. Mourgues, T. Olofsson, Quantitative analysis of workflow, temporary structure usage, and productivity using 4D models, Autom. Constr. 17 (2008) 780–791.

[10] S. Azhar, M. Hein, B. Sketo, Building information modeling (BIM): benefits, risks and challenges, Proceedings of the 44th ASC Annual Conference, Auburn, Alabama, 2–5 Apr 2008, 2008, pp. 627–634. [11] B. Ernstrom, et al., The Contractors' Guide to BIM, Associated General Contractors of America, 2006. [12] J. Xu, S.M. AbouRizk, C. Fraser, Integrated three-dimensional computer-aided design and discrete-event simulation models, Can. J. Civ. Eng. 30 (2003) 449–459. [13] R.J. Mayer, IDEF0 Function Modeling, Knowledge Based Systems, College Station, TX, 1992. [14] J. Banks, J.S. Carson, B.L. Nelson, D.M. Nicol, Discrete Event Systems Simulation, Prentice Hall, Englewood Cliffs, NJ, 2000. [15] D.W. Halpin, L. Riggs, Planning and Analysis of Construction Operations, Wiley Inter Science, New York, N.Y., 1992. [16] S. AbouRizk, D. Halpin, Y. Mohamed, U. Hermann, Research in modeling and simulation for improving construction engineering operations, J. Constr. Eng. Manag. 137 (2011) 843–852. [17] J. Martinez, P.G. Ioannou, General purpose simulation with Stroboscope, Proceedings of the 1994 Winter Simulation Conference, Orlando, FL, United States, 1994, pp. 1159–1166. [18] D.K.H. Chua, G.M. Li, RISim: resource-interacted simulation modeling in construction, J. Constr. Eng. Manag. 128 (2002) 195–202. [19] M. Lu, Simplified discrete-event simulation approach for construction simulation, J. Constr. Eng. Manag. 129 (2003) 537–546. [20] K.J. Kim, G.E. Gibson, Interactive simulation modeling for heavy construction operations, Autom. Constr. 12 (2003) 97–109. [21] A.A. Oloufa, M. Ikeda, T. Nguyen, Resource-based simulation libraries for construction, Autom. Constr. 7 (1998) 315–326. [22] T. Perera, K. Liyanage, Methodology for rapid identification and collection of input data in the simulation of manufacturing systems, Simul. Pract. Theory 7 (2000) 645–656. [23] M.P. Gallaher, R.E. Chapman, Cost analysis of inadequate interoperability in the US capital facilities industry, US Department of Commerce, Technology Administration, National Institute of Standards and Technology, 2004. [24] N. Robertson, T. Perera, Automated data collection for simulation? Simul. Pract. Theory 9 (2002) 349–364. [25] L.G. Randell, G.S. Bolmsjo, Database driven factory simulation: a proof-of-concept demonstrator, Proceedings of the 2001 Winter Simulation Conference, 2011, pp. 977–983. [26] C. McLean, A. Jones, T. Lee, F. Riddick, An architecture for a generic data-driven machine shop simulator, Proceedings of the 2002 Winter Simulation Conference, 2002, pp. 1108–1116. [27] K. Jeong, L. Wu, J. Hong, IDEF method-based simulation model design and development framework, J. Ind. Eng. Manag. 2 (2009) 337–359. [28] A. Hanna, R. Camlic, P. Peterson, E. Nordheim, Quantitative definition of projects impacted by change orders, J. Constr. Eng. Manag. 128 (2002) 57–64. [29] F. Pena-Mora, M. Park, Dynamic planning for fast-tracking building construction projects, J. Constr. Eng. Manag. 127 (2001) 445–456. [30] P.E.D. Love, G.D. Holt, L.Y. Shen, H. Li, Z. Irani, Using systems dynamics to better understand change and rework in construction project management systems, Int. J. Proj. Manag. 20 (2002) 425–436. [31] S. Isaac, R. Navon, A graph-based model for the identification of the impact of design changes, Autom. Constr. 31 (2013) 31–40. [32] M. Lee, B.Y. Ryoo, K.T. Sullivan, A.S. Hanna, Development of automated change order impact detection and quantification system, Autom. Constr. 18 (2009) 525–535. [33] T. Olofsson, G. Lee, C. Eastman, Editorial—case studies of BIM in use, J. Inf. Technol. Constr. (2008) 244–245. [34] R. Jongeling, M. Emborg, T. Olofsson, nD modelling in the development of cast in place concrete structures, J. Inf. Technol. Constr. (2005) 27–41. [35] J. Kunz, J. Fischer, Virtual design and construction: themes, case studies and implementation suggestions, CIFE Working Paper 097, CIFE, Stanford University, 2009. [36] H. Li, T. Huang, C.W. Kong, H.L. Guo, A. Baldwin, N. Chan, J. Wong, Integrating design and construction through virtual prototyping, Autom. Constr. 17 (2008) 915–922. [37] C.E. Fırat, J. Kiiras, P. Huovinen, A building construction information model for managing projects virtually, Innovation in Architecture, Engineering and Construction (AEC), June 23–25, 2008 Antalya Turkey, 2008. [38] A. Mahalingam, R. Kashyap, C. Mahajan, An evaluation of the applicability of 4D CAD on construction projects, Autom. Constr. 19 (2010) 148–159. [39] J. Goedert, P. Meadati, Integrating construction process documentation into building information modeling, J. Constr. Eng. Manag. 134 (2008) 509–516. [40] C. Eastman, P. Teicholz, R. Sacks, K. Liston, BIM Handbook: A Guide to Building Information Modeling for Owners, Managers, Designers, Engineers, and Contractors, John Wiley & Sons, Inc., 2008. [41] Z. Hu, J. Zhang, BIM and 4D-based integrated solution of analysis and management for conflicts and structural safety problems during construction: 2. Development and site trials, Autom. Constr. 20 (2011) 167–180. [42] Z. Mallasi, N. Dawood, Registering space requirements of construction operations using site-PECASO model, Proceedings of CIB w78 2002 Conference on Distributing Knowledge in Buildings, Aarhus School of Architecture, Denmark, 2002, pp. 83–90. [43] R. Jongeling, M. Emborg, M. Asp, T. Olofsson, Modelling and simulation of cast in place concrete constructions using N-dimensions, Research Report 2005:18, 14021528Luleå Technical University, 2005. [44] H. Norberg, P. Wallström, T. Olofsson, A. Segerstedt, V. Tarandi, L. Isaksson and N. Outters, Byggsynkronisering, Research Report 2009, ISBN: 9789174390070, Luleå Technical University.

W. Lu, T. Olofsson / Automation in Construction 44 (2014) 73–83 [45] F. Peterson, T. Hartmann, R. Fruchter, M. Fischer, Teaching construction project management with BIM support: experience and lessons learned, Autom. Constr. 20 (2011) 115–125. [46] J. Irizarry, E.P. Karan, F. Jalaei, Integrating BIM and GIS to improve the visual monitoring of construction supply chain management, Autom. Constr. 31 (2013) 241–254. [47] T. Kala, O. Seppänen, C. Stein, Case study of using an integrated 5D system in a large hospital construction project, Proceedings of the 18th International Group of Lean Construction, July 14–16, 2010 Haifa, Israel, 2010. [48] S.C. Hee, K. Kyungrai, W.S. Dong, Development of 5D-CAD based work process model for aged-housing revamp projects, CIB W78 International Conference on Information Technology in Construction, Santiago, Chile, 2008. [49] I.S. Panushev, S.N. Pollalis, A framework for delivery of integrated building information modeling, Joint International Conference on Computing and Decision Making in Civil and Building Engineering, Montreal, Canada, 2006, pp. 2814–2822. [50] R. Sacks, M. Radosavljevic, R. Barak, Requirements for building information modeling based lean production management systems for construction, Autom. Constr. 19 (2010) 641–655. [51] R. Jongeling, T. Olofsson, A method for planning of work-flow by combined use of location-based scheduling and 4D CAD, Autom. Constr. 16 (2007) 189–198. [52] A.M. Ospina, D. Castro, Interaction of processes and phases in project scheduling using BIM for A/E/C/FM integration, Construction Research Congress 2010: Innovation for Reshaping Construction Practices, Edmonton, Canada, 2010, pp. 939–948.

83

[53] J.L. Harrington, Relational Database Design and Implementation: Clearly Explained, Academic Press, London, 2002. [54] A. Darwiche, R.E. Levitt, B. Hayes-Roth, OARPLAN: generating project plans by reasoning about objects, actions and resources, AI EDAM 2 (3) (1988) 169–181. [55] C. Mourgues, M. Fischer, J. Kunz, Method to produce field instructions from product and process models for cast-in-place concrete operations, Autom. Constr. 22 (2012) 233–246. [56] M. Fischer, F. Aalami, C. Kuhne, A. Ripberger, Cost-loaded production model for planning and control, Conf. on Durability of Building Materials and Components, CIB W78, Workshop, Vancouver, Canada, 1999, pp. 2813–2824. [57] Z. Shen, R.R.A. Issa, Quantitative evaluation of the BIM-assisted construction detailed cost estimates, J. Inf. Technol. Constr. 15 (2010) 234–257. [58] D.T. Sturrock, C.D. Pegden, Recent innovations in Simio, Proceedings of the 2011 Winter Simulation Conference, 2011, pp. 52–62. [59] Wikells Byggberäkningar, Sektionsfakta-NYB 12/13, 2012. [60] S.M. Levy, Project Management in Construction, McGraw-Hill Professional Engineering, 2006. [61] M. Marcel, W. Avi, Descriptive construction methods through BIM-based collaboration, National Institute of Building Sciences-building SMART alliance Conference, 2013. [62] S. Lee, S. Han, F. Peña-Mora, Integrating construction operation and context in largescale construction using hybrid computer simulation, J. Comput. Civ. Eng. 23 (2009) 75–83.