Planning the use of spatial resources on projects
the case of construction
Dr. Graham M. Winch, Professor of Construction Project Management, Manchester Centre for Civil and Construction Engineering, UMIST
Planning lies at the heart of project management. The critical path method and its variants have developed into a suite of sophisticated tools for planning the sequence of tasks in the work breakdown structure through time. More recent developments such as critical chain have turned attention to the resources required for task execution. However, we still lack tools for the effective planning of an equally important aspect of task execution—the space in which the task is to be executed. This conference paper—based on current research financed by the United Kingdom’s (UK) Engineering and Physical Sciences Research Council—will introduce the concept of critical space and its relevance for project planning.
Critical space analysis is relevant to that subclass of projects where task execution not only creates the completed facility, but also creates, temporally, the spaces in which tasks are to be executed. This includes construction, shipbuilding, oil and gas rigs, and process plants. The fundamental planning problem is that task execution space availability is dynamic, not only because different trades parade through the same space and may clash spatially, but because the spaces themselves change as, for instance, floors are laid creating work spaces, and walls are built closing off work spaces. Of course, experienced project planners do take into account the availability of task execution spaces, but this is based on experience, rarely formalized beyond the most basic rules of thumb. Moreover, most of the recent work on this topic has focused on the layout of static site facilities, not the weekly dynamic of spatial availability.
The VIRCON (Virtual Construction) project is a collaborative research project between four UK universities (Teesside, UCL, UMIST, and Wolverhampton) and six construction firms, including AMEC, Balfour Beatty, Carillion, and Skanska. Its aim is to develop CAD-linked analytic tools for the strategic project planning of construction projects, and this paper will report on initial conceptual work in developing our vision of the future possibilities of construction project planning, and also report on the requirements capture phase of the project which involved interviews with a number of experienced construction project planners.
The paper will start by reviewing some of the main recent developments in construction project planning. It will then go on to present the findings from the requirements capture phase of the project, which has identified the renewed importance of project planning with the development of new forms of procurement such as partnering, and public/private partnerships. The need for “quick and dirty” tools, rather than analytic elegance was also identified. The paper then goes on to discuss the VIRCON vision in more detail, showing the relationships between critical space analysis, critical path analysis, and how they might come together in a space-time broker for project planning. Finally, the recent progress of the VIRCON project will be reported.
The Development of Construction Project Planning
The theory and practice of construction project planning has been under attack for some time, particularly from the work of Laufer and his colleagues (e.g., Laufer and Tucker 1987). They and others argue that it is control rather than action orientated, deterministic, and is over-reliant on critical path analysis. This critique chimes with Golratt’s (1997) claim that there has being nothing new in project planning for forty years. However, the 1990s has seen considerable innovation in project planning techniques, and construction project planning in particular, all of which attempt to move away from (futile?) attempts to refine the predictive capabilities of critical path analysis through sophisticated stochastic methods, when the problem is one of lack of data about the future.
There are a number of such developments that will be discussed seriatim:
• Goldratt’s own critical chain approach, building on his theory of constraints
• The development of lean construction concepts, and, most relevantly, last planner
Exhibit 1: Critical Chain and Last Planner Combined (Winch 2002)
• The development of 4D planning, where the 3D product model is given a time dimension
• The development of the analysis of the spatial aspects of project execution.
The critical chain method (CCM) was discussed extensively at the previous Project Management Institute (PMI®) Research Conference in Paris. It addresses two of the key problems of CPM—the inherent uncertainty of task durations and the associated opportunistic behavior in establishing the true duration of tasks, and the resourcing of tasks. In CCM, the critical chain is the longest resource constrained path through the network, theorized as a constraint to be elevated. Thus a critical chain looks like a critical path, but it includes resourcing in the dependencies.
The conceptual shift from critical path to critical chain by including resourcing issues in the latter might be considered simply a technical development, moving on from the resource leveling approaches that are well established. However, the elevation of the constraint introduces a much more radical aspect. This elevation starts from two observations:
1. That actual duration for any planned task is unknown, but can be assumed to be normally distributed around expected task distribution.
2. Existing estimates of expected task durations are padded because they include safety time, greatly extending the length of the critical path.
In a situation where actual task duration is uncertain, and managers are held accountable for meeting stated durations (deadlines), then task duration estimates are going to be at the worst case end of the distribution, otherwise managers risk overrunning their deadlines. CCM instead proposes that the average estimated duration should be used—there is a technical debate regarding whether the mean or the median of the distribution is more appropriate. An inevitable result of this is that managers will overrun their planned durations half of the time—this problem is solved by removing from managers the absolute responsibility to meet task-level duration deadline, and, instead providing incentives for early completion. This approach was tried on the A13 project in East London with great success (Barber et al 1999).
One of the main problems with managing by deadlines—or more specifically, latest start dates—is that there is a strong tendency to start work even if not all the resources required for the completion of the task are available. This tendency— known as multitasking by Goldratt—has the inevitable result of increasing the variability of task durations by introducing greater uncertainty, and extending average task durations due to switching costs between tasks. These problems have been addressed in more depth by the advocates of “shielding production” through the last planner technique. See Ballard and Howell (1998), and, more generally, the work of the Lean Construction Institute for detail.
Their research starts from the observation that construction project managers do not manage task execution but manage contracts—the work of managing processes on site is not seen as the responsibility of managers in construction, but is delegated to self-organizing gangs. They argue that it is the responsibility of management to ensure that tasks are executed as efficiently as possible. Efficient task execution will reduce variability in task execution, and because the distribution of durations is skewed towards the pessimistic end, will also reduce average task execution times. The key to efficiency is shielding task execution so that tasks only start when precedent tasks have been completed, and all the resources are available. Such ready-to-start tasks are known as quality assignments.By making only quality assignments, project managers can both reduce costs through increased efficiency and reduce durations by eliminating uncertainty. In order that resource utilization is not reduced due to delays in task commencement due to shielding, project managers are expected to build up buffer stocks of quality assignments off the critical path to which underloaded resources can be allocated.
The approach is called last planner because making quality assignments is the last stage in the project planning process. The planning horizon is typically one week, and the decision-making process is delegated down to the level of first-line supervision. It is a tactical tool, not a strategic one. Although last planner and critical chain were apparently developed independently, they would appear to be highly complementary, with critical chain solving the strategic problem inherent in last planner—if the project ran out of quality assignments, progress would grind to a halt. The weekly decision-making cycle needs to be in the context of “look-ahead” planning on a monthly or quarterly cycle using critical chain. The difference between this approach combining critical chain and last planner, and CPM-driven task allocation can, perhaps, be illustrated in the manner shown in Exhibit 1.
Developments in CAD over the last few years have led to the development of the capability to create sophisticated 3D product models that are invaluable for both visualization and analysis purposes. This has inspired the development of 4D construction planning where the 3D model is related to the program so that the visualization of the “assembly” week-by-week can be achieved. This vision is well-captured by the Stanford/Walt Disney Imagineering team (Fischer et al 2001):
Planners, designers, and engineers will use 4D technologies to analyze and visualize many aspects of a construction project, from the 3D design of a project to the sequence of construction to the relationships between schedule, cost and resource availability data. Planners will create, update and maintain, and deliver a 4D object model throughout a design/construction project. The 4D technologies will enable planners to generate various views of this 4D object model to clearly communicate the spatial and temporal aspects of construction schedules to all project participants. 4D technologies will no longer be used to simply animate the sequence of construction but will be used to communicate a wide range of project data much more clearly and efficiently than possible today. Planners, designers, and engineers will use 4D environments to visually relate data much like the way engineers use gradated color 3D models to visualize the stresses on structures.
This is an exciting area of research, and many challenges remain to be solved (Fischer et al 2001). However, benefits have already been identified in areas such as refining the program (Koo and Fischer 2000), in addition to the important improvements in communications with the client about the plans for the construction of its facility.
A fourth area of development is related to the spatial configuration of execution on site. Construction processes, together with many other engineering projects, have an inherent spatial dimension. Execution of the design is physically constrained, and, usually, site-specific. Not only is the site a defined—and usually confined—space, but also the configuration of that space changes continually as the project progresses on site. There have been two approaches to this problem in recent research. The first is to address the site-layout problem—in other words to try to improve the layout of the site, particularly temporary installations and materials storage facilities, with the aim of ensuring that subsequent activity is as efficient and effective as possible. The work of Tommelein at Berkeley, coming from an operations research perspective, is perhaps the best-known contribution here (e.g., Tommelein et al 1990; Zouein and Tommelein 1999). Her work on SightPlan its derivatives represents a sustained attempt to apply the latest modeling techniques to the problem. Others such as Cheng and O’Connor have used Geographical Information Systems (GIS), while Retik and Shapira (1999) have applied Virtual Reality (VR) techniques to this problem.
The second—and complementary—approach is to focus more directly on task execution. Thabet and Beliveau (1994) developed basic algorithms for modeling workspace demand for repetitive tasks in multistory buildings. Riley (Riley and Sanvido 1996) conducted extensive empirical work on site in multistory buildings to develop a detailed typology of construction space use during task execution. Further work (e.g., Riley and Sanvido 1997) has explored ways of analyzing the space requirements for task execution so as plan for the most efficient and effective task execution, and explored the benefits of a 4D environment.
As Exhibit 1 suggests, there are important complementarities between all of these strands of work, and many of the leaders in each of the strands have collaborated in developing this new perspective on project planning—see for instance Tommelein, Riley, and Howell (1999) which applies theory of constraints ideas to construction scheduling. From the perspective of the spatial configuration of construction task execution, the principal common threads through this body of current research would appear to be:
• Space is a resource, and therefore can be conceived of as a constraint to be elevated in critical chain theory.
• 4D construction planning allows the visualization of available space at sequential points in the program.
• The availability of task execution space is an element in the determination of quality assignments in a last planner environment.
• The site layout, or external spatial availability and the task execution, or internal, spatial availability are essential aspects of construction planning, and need to be combined in one analytic model.
Exhibit 2. Space Classifications
Others are currently addressing these themes—what we wish to do in this paper is to present a particular approach that, we believe, complements this ongoing work. This is the VIRCON concept, and, in particular the concept of critical space analysis.
The Concept of Critical Space Analysis (CSA)
One distinctive feature of construction projects that distinguishes them from those in other sectors is that the spaces available for task execution change as the project progresses through the program. While the spatial configuration of the production process in a factory remains static during production, the spatial configuration of the construction process is continually changing as, for instance, product elements are fixed, or temporary works removed. In manufacturing, spatial configuration is usually given considerable attention as the layout of production equipment is planned, yet in construction, relatively little attention is paid formally to the allocation of tasks to spaces. This is not to suggest that construction planners do not consider these matters—they are too important to be ignored—but to argue that spatial task allocation decisions are made on the basis of experience and intuition, without the support of tools such as those available for the sequencing of tasks such as critical path analysis and its derivatives.
Others have also noticed this gap in the toolkit available to construction planners. Researchers such as Tommelein and her colleagues (1991, 1992), Retik and Shapira (1999), and Chen and O’Connor (1996) have made important advances in providing such tools, but they focus on the factory anal-ogy—layout planning is essentially a one-off preprocess activity. A more “dynamic” element has been introduced by Zouein and Tommelein (1999), but they still focus on the layout and movement of plant and site installations evolving over time between stable states rather than the dynamic spatial configuration of task execution itself. Only a few have turned their attention to task execution spaces—most notably Thabet and Beliveau (1994) and Riley and Sanvido (1995, 1997). In particular, the earlier paper by Riley and Sanvido is seminal in providing an empirically-based classification of the different types of spaces required for, and associated with, task execution. It is this work that has formed one of the main influences on the development of the concept of critical space analysis.
A second strand of work that has strongly influenced the development of the CSA concept is space syntax (Hillier and Hanson 1984; Hillier 1996). This work investigates the influence of spatial configuration on human movement. This work is behavioral, focusing on how individuals choose to move through spaces—essentially following the longest lines of sight. Such analysis—supported by analytic tools such as AxeMan—allows the prediction to a high degree of probability of likely human movement patterns through any given configuration of spaces from the 2D plan of the space. To date, work in space syntax has been on existing spaces within and between buildings, and proposals for new buildings and urban spaces. The VIRCON project—together with that on customer movement during refurbishment in the associated RaCMIT project—is, we believe, innovative in applying these techniques to buildings during the process of their construction.
Any analysis of the spatial configuration of the construction process needs a set of definitions of space-types. Researchers at Virginia Tech pioneered here, while the work of Riley and Sanvido is authoritative. However, these authors focused on the spaces constrained by the fixing of product elements, and did not explicitly consider the total available space, as defined by the site boundary. This is, possibly, due to their choice of repetitive multistory building construction as an exemplar. In this, the site layout literature has an analytic advantage. Our proposals here were agreed at a VIRCON team meeting in January 2002. Exhibit 2 illustrates our proposed classification. The main elements are:
• Total Space (t): the total area bounded by the site boundary
• Product Space (p): the area of (t) occupied by fixed product components
• Installation Space (i): the area of (t) occupied by temporary installations
• Available Space (a): the balance of space left when (i) and (p) are deducted from (t)
• Required Space (r): the space required by a task for execution, storage, safety, and movement.
At this stage of CSA development, only analysis of required space for task execution is required. Available spaces can also be categorized as areas and/or paths (Riley and Sanvido 1995). There are several area subcategories. In addition to work areas, layout, unloading, staging, storage, prefabrication, tools and equipment, and hazard and protected spaces can be identified. Likewise, there are several path subcategories: material, personnel, and debris.
CSA is proposed as the simulation, visualization and optimization of the spatial loads placed on a construction project by its scheduled tasks and resources. Spatial loading is the ratio of required space to available space—Thabet and Beliveau (1994) call this the Space Capacity Factor. A ratio greater than unity means congestion. Critical Spaces are defined as those where loading is at or greater than unity, i.e., there is no spatial slack. This approach is analogous to Critical Path Analysis (CPA), which is applied to time rather than space.
It is proposed that only spaces of the type “work area” (which for VIRCON purposes are referred to as “execution spaces”) are permanently linked to scheduled tasks. That is to say, a task with a product output that “makes the building grow” can only execute in its intended space. However, if support tasks or supplementary areas and paths are required by a task, these can generally be located in a variety of available spaces. A space may have many roles during its existence. For example: execution space, material path, hazard area, and so on. In fact it may be easier to categorize a space by roles that it cannot undertake. By way of example, a space that in no part of its life span contains more than one entrance can never be a path.
Resources require space. For example, it has been stated that an individual construction worker has a minimum spatial requirement of 28m2—see Kelsey Winch and Penn (2001) for a review. Suppliers of construction plant normally supply data on the safe working envelope of their plant—see Heesom and Mahdjoubi (2002c) for a review. Resources are allocated to tasks, which then sum their spatial requirements. Tasks are always allocated to one or more execution spaces, but may also require other combinations of area and path spaces, depending on the nature of the trades involved. The load on a space will vary during its life span, in response to the changing requirements of its allocated tasks.
The VIRCON Concept
Critical Space Analysis is one input into the overall VIRCON concept, which is presented in Exhibit 3. The core of the VIRCON concept is the project database developed in MS Access (Dawood et al 2001). This contains both graphical data originating from the designers in AutoCAD, and program data originating from the constructors in Microsoft® Project®.It has been developed using Uniclass (Crawford et al 1997; see also Kang and Paulson 2000), which is the UK implementation of the principles contained in ISO Technical Report 14177 Classification of Information in the Construction Industry of 1994. It is also compliant with BS 1192-5 Construction Drawing Practice: Guide for the Structuring and Exchange of CAD Data (1998), which is the UK implementation of ISO 13567. These are used to develop both the product breakdown structure, derived from the CAD product information, and the work breakdown structure for project programming. Drawing on this database, both critical path and critical space analysis can be undertaken and then analyzed jointly in the space-time broker. In parallel with this activity, a 4D model of the progress of the project is generated, visualizing the spatial availability sequentially over the life of the project. Outputs from these analyzes can take various forms:
Exhibit 3. VIRCON System Diagram (Kelsey et al 2001)
Exhibit 4.The VIRCON Collaborators
• 4D visualization of project progress
• Clash detection of overlapping activities—both temporally and physically
• Analyzes of spatial constraints
• Space syntax analysis of movement paths.
Developing the VIRCON Tool
The development of VIRCON—financed by the UK Engineering and Physical Sciences Research Council’s awards numbered GRN000876/890/906—is a collaborative effort between four UK universities, and six UK construction companies, as shown in Exhibit 4. Other industrial collaborators are associated with particular projects used in the development of the VIRCON tool. Our tool is not linked to the virtual environment for teaching construction management of the same name developed in Australia (Jaafari 2001). Development started with a requirements capture phase, reported in Kelsey et al (2001). This consisted mainly of interviews with construction planners in the collaborating companies. A number of clear messages came from this phase of the project:
• Construction planners would welcome tools that enabled the graphical visualization of the program.
• Most planners preferred simple project planning software tools, rather than more sophisticated ones such as Primavera, which tended to be seen as over-rigorous, and confined to major projects.
• Spatial planning was largely intuitive, and trade contractors—particularly groundworks and services—reported difficulties in convincing planners of their spatial requirements for task execution.
• Planners have very little time in which to plan, and any tool provided must be simple to use, with quick iterative capabilities, and integrate with existing tools.
On this basis, and combined with an extensive review of current research in the domains identified above (Heesom and Mahdjoubi 2001a) and current leading edge industry practice (Heesom and Mahdjoubi 2001c), the VIRCON tool was developed using the following principles:
• Construction space planning is essentially a 2D problem—any activity in the third dimension effectively sterilizes the space below it for safety and access reasons, and so all that needs to be planned in most cases is the 2D “footprint” of the activity. Therefore, the VIRCON tool is, in essence, a 3D tool (x, z + t), rather than a 4D tool (x, y, z + t). This is also convenient, because very few designers in construction are presently preparing 3D models, as opposed to 2D drawings. 4D planning will only, we suggest, offer significant advantages in rare, particularly space-critical, planning situations.
• The VIRCON tool is a strategic planning tool. We would not, therefore, expect it to be used in conjunction with last planner techniques. We believe the actual determination of the availability of spatial resource for quality assignments on a week-by-week will remain a physical one by inspection, rather modeled in a virtual environment. The VIRCON tool will give the planned week-by-week availability of spaces.
• An extensive review of current software available for critical path analysis (Heesom and Mahdjoubi 2001b) concluded that Microsoft Project had by far the largest installed base, and also the best data exchange and application programming facilities. Plug-ins are also available for critical chain applications such as ProChain. For these reasons, the VIRCON tool is written to operate in a Microsoft environment, taking advantage of Open Data-base Connectivity (ODBC). For similar reasons, AutoCAD was adopted as the preferred CAD software (Dawood et al 2001).
• It is a decision-support tool for planners, not a decision-making one. Our approach is to visualize the construction program using simple graphics, rather than to optimize the construction program using sophisticated algorithms. The implementation of the sophisticated optimization tools available for critical path analysis in contemporary construction project planning practice has been limited—very often the use of CPA software is limited to the preparation of visualizations of the program, rather than its analysis. We believe that in order to get critical space analysis accepted by construction planners, its benefits for visualization need to be stressed at this stage.
• Most of the benefits of 4D project planning for the visualization of the program can be achieved by “stretching” 2D drawings into the y dimension, without the need to obtain inputs in the form of a 3D product model. This is much closer to the reality of practice on most construction projects—few architects are geared up to produce full 3D product models.
• The development of Industry Foundation Classes by the International Alliance for Interoperability was not yet well enough advanced for it to be worthwhile attempting to link our work to that initiative (Heesom and Mahdjoui 2001c). We, therefore, opted for a relational database, rather than an object-orientated one (Dawood et al 2001). However, this is not a rejection of standardization as such, as is shown by the adoption of Uniclass and BS 1192-5.
The first stage in the development of the tool was to capture the product and program information from our “test-bed” project, the new Centuria Building for the School of Health at Teesside University, Middlesbrough, UK. This proved to be time-consuming as the architect’s drawings were in ArchiCAD, and the contractor had not developed a network to support its Gantt Charts, and our commitment to Uniclass provided a further dimension of difficulty. Considerable development effort had, therefore, to be expended in establishing protocols for the population of the database. Using the database, tools for the marking up of available space on the 2D product drawings (AreaMan), and critical space analysis, and space-time broking (SpaceMan) were developed and tested (North and Winch 2002). We identified the need for a tool to manage data on resource inputs (ResourceMan), implemented in Microsoft Access® (Heesom and Mahdjoubi 2002)—again this was based on Uniclass principles. A drag-and-drop facility for placing plant, plant paths, and temporary installations such as site accommodation and scaffolding was also developed (PlantMan) (Heesom and Mahdjoubi 2002).A further tool has been developed to detect potential conflicts between the temporary works objects and elements under construction (ClashMan) (Heesom and Mahdjoubi 2002). In parallel, 4D visualization methodologies were developed at both the whole-building scale in an AutoCAD environment (Dawood et al 2002) with a VR interface (ProVis), and at the scale of the piling and services installation trades in a VRML environment (SpaceVis) (Heesom and Mahdjoubi 2002). Data on spatial requirements for piling plant and task execution were obtained from a hospital site in central London. The VIRCON tool is presently in the phase of systems integration, and an early prototype is being evaluated on a second live project—a new school in Manchester, UK.
The Potential of the VIRCON Tool
A further problem identified in the requirements capture phase is that project managers actually have remarkably little time to plan on most construction projects (Kelsey et al 2001). The overall project program is formed during tender and becomes enshrined in the master program. Short tender periods mean that this master program is very much a guess at what the actual task durations might be. This has meant that subsequent programs developed for actually managing the project tend to be constrained by decisions made in haste during tender. In such an environment, sophisticated construction planning techniques are difficult to implement. However, this is situation is changing, at least in the UK (Winch 2000). The development of partnering on the basis of longer-terms relationships between clients, and concession contracting where finance, design, construction, and operation are rolled up into one contract mean that projects have longer lead times from the point of view of those responsible for planning execution on site, and that, therefore, construction project managers have more time to plan their projects.
Thus the development of the VIRCON tool, and many of the others briefly described in the earlier part of this paper, is timely for construction project planning. We believe that the VIRCON approach is distinctive in that it is led by a requirements capture phase, which encouraged us towards an industry-led, rather than a technology-led approach to development. While this may disappoint some of the more sophisticated theorists of 4D planning and related approaches, we believe that it is closer to where the bulk of the (UK) construction industry presently is, and can help to pave the way for the wider acceptance of some of the more sophisticated approaches. Our approach is a QUAD one (quick and dirty), and we make no apologies for that, but it is also one that embodies the latest research in space and movement both during construction, and in the completed building, in the concept of critical space analysis.
The current research project is planned to end in November 2002. Future work on the development of the VIRCON tool is likely to take various directions:
• Integration of movement analysis into the Man tools
• Relationship to IFCs, particularly in database development
• Relationship to critical chain analysis, with space theorized as a constraint
• Further development of 4D visualization methodologies for the Vis tools.
Ballard, G., and G. Howell. (1998). Shielding production: Essential step in production control. Journal of Construction Engineering and Management 124: 11–17.
Barber, P., C. Tomkins, and A. Graves. (1999). Decentralised site management—A case study. International Journal of Project Management 17: 113–120.
Cheng, M. Y., and J. T. O’Connor. (1996). ArcSite: Enhanced GIS for construction site layout. Journal of Construction Engineering and Management 122: 329–336.
Crawford, M., J. Cann, and R. O’Leary. (1997). Uniclass: Unified Classification for the Construction Industry.London: RIBA Publications.
Dawood, N., E. Sriprasert, and Z. Mallasi. (2001). VIRCON Project: Data Capture and Database Development.Middlesborough: University of Teesside.
Dawood, N., E. Sriprasert, C. Liu, and Z. Mallasi. (2002). VIRCON Project: 4D Visualisation Development: Whole Building Visualisation. Middlesborough: University of Teesside.
Fischer, M., K. M. Liston, and Walt Disney Imagineering. (2001). Wish-list for 4D Environments: A WDI R&D Perspective. http://www.stanford.edu/group/4D/issues/wishlist.shtml. Accessed 12/04/02.
Goldratt, E. M. (1997). Critical Chain.Great Barrington: The North River Press.
Heesom, D., and L. Mahdjoubi. (2001a). Technology Opportunities and Potential for the Virtual Construction Site: Emerging Research Initiatives.Wolverhampton: University of Wolverhampton.
———. (2001b). Technology Opportunities and Potential for the Virtual Construction Site: Project Planning Software Review. Wolverhampton: University of Wolverhampton.
———. (2001c). Technology Opportunities and Potential for the Virtual Construction Site: Existing Technology and Potential. Wolverhampton: University of Wolverhampton.
———. (2002). Visualisation Development. Wolverhampton: University of Wolverhampton.
Hillier, B. (1996). Space is the Machine. Cambridge: Cambridge University Press.
Hillier, B., and J. Hanson. (1984). The Social Logic of Space. Cambridge: Cambridge University Press.
Jaafari, A., K. K. Manivong, and M. Chaaya. (2001). VIRCON: Interactive system for teaching construction management. Journal of Construction Engineering and Management 127: 66–75.
Kang, L. S., and B. C. Paulson. (2000). Information classification for civil engineering projects by uniclass. Journal of Construction Engineering and Management 126: 158–167.
Kelsey, J., G. M. Winch, and A. Penn. (2001).Understanding the Project Planning Process: Requirements Capture for the Virtual Construction Site.London: Bartlett Research Paper 15.
Koo, B., and M. Fischer. (2000). Feasibility study of 4D CAD in commercial construction. Journal of Construction Engineering and Management 126: 251–260.
Laufer, A., and R. L. Tucker. (1987). Is construction project planning really doing its job? A critical examination of focus, role and process. Construction Management and Economics 5: 243–266.
North, S., and G. M. Winch. (2002). Analysing Critical Space and Time.Manchester: UMIST.
Retik, A., and A. Shapira. (1999). VR-based planning of construction site activities. Automation in Construction 8: 671–680.
Riley, D. R., and V. E. Sanvido. (1995). Patterns of construction-space use in multistory buildings. Journal of Construction Engineering and Management 121: 464–473
———. (1997). Space planning method for multistory building construction. Journal of Construction Engineering and Management 123: 171–180.
Thabet, W. Y., and Y. J. Beliveau. (1994). Modeling work space to schedule repetitive floors in multistory buildings. Journal of Construction Engineering and Management 120: 96–116.
Tommelein, I. D., R. E. Levitt, B. Hayes-Roth, and T. Confrey. (1991). Sightplan experiments: Alternative strategies for site layout. Design Journal of Computing in Civil Engineering 5: 42–63
Tommelein, I. D., D. R. Riley, and G. A. Howell. (1999). Parade game: Impact of work flow variability on trade performance. Journal of Construction Engineering and Management 125: 304–310.
Winch, G. M. (2000). Institutional reform in British Construction: Partnering and private finance. Building Research and Information 28: 141–155.
———. (2002). Managing Construction Projects: An Information Processing Approach.Oxford:Blackwell.
Zouein, P. P., and I. D. Tommelein. (1999). Dynamic layout planning using a hybrid incremental solution method. Journal of Construction Engineering and Management 125: 400–408.
Proceedings of PMI Research Conference 2002