Current practice shows that construction design change management relies heavily on the experience of practitioners to assess the impact of a proposed change. This U.K. government- and industry-funded research (Arup) is concerned with mitigating the risk associated with a practitioner making a judgment disproportionate to the true impact of a design change.
Several design management tools and techniques have been reviewed (including the Analytical Design Planning Technique [ADePT] and the “last planner” methodology). Suggestions have been made on how these can be adapted to apply to design change management. A construction design change management (CDCM) model is proposed as a possible solution, enabling practitioners to make an informed decision regarding the true impact of proposed changes.
The CDCM model concept incorporates a design structure matrix (DSM) and process map generation to create a checklist of rework. It also records the reason for deviation if the true impact is different to the assessed impact. The cost, resource, deviation, and reason for deviation are stored in a database and are available when a similar change is required on a similar project, allowing compensation to be applied to the predicted impact.
Keywords: design change management, design structure matrix (DSM), last planner, impact assessment, risk
All projects can be represented by the project management triangle (Figure 1), where, scope, cost, and time are the project constraints represented on each corner. In all projects, scope, time, and cost are connected. For example, in order to produce equivalent project outcomes (scope) within a shorter duration (time), the project cost will need to be increased (cost). Likewise, when a design change occurs, there is a change to the scope of the project, therefore, it is necessary to change the project cost and/or duration. When referring to the impact of the design change, it is this change to the project cost and/or duration that needs to be considered. In order to calculate the additional project cost, it is necessary to know the additional resource needed to complete the rework/ redesign. This research considers the deviation between the predicted impact of a change and the true impact once a design change has been implemented. This deviation is in terms of cost, resource, and time.
Design changes during the detailed design phase of a built environment project usually have a negative impact on the design program, in terms of additional resource, cost, and project duration. Since design is iterative in nature, the consequences of a change can rattle through various engineering disciplines, making the impact difficult to predict. Current practice, within the project sponsor organization, shows that project teams calculate the effect of a design change on the program using their experience of similar changes during previous projects. There is a risk in relying solely on a practitioner's knowledge, in that they could make an inaccurate assessment or they may leave the business, resulting in a reduction in specialist decision-making expertise.
A number of design management tools and techniques have been reviewed such as the Analytical Design Planning Technique (ADePT) (Austin, Baldwin, Li, & Waskett, 1999a, 1999b, 1999c) and the last planner methodology (Ballard & Howell, 1998; Choo, Tommelein, Ballard, & Zabelle, 1999; Ballard, 2000, Choo, Hammond, Tommelein, Ballard, & Austin, 2003). The authors of this paper suggest what can be done to adapt these tools/techniques in order to address change management. A construction design change management (CDCM) model is proposed as a possible solution. This would support practitioners in making informed decisions with respect to the true impact of proposed changes. Throughout this paper, the term practitioner is used to represent any member of the design team who has a direct effect on the design program, for example, engineers, architects, and project managers.
The CDCM model incorporates a design structure matrix (DSM) and process map generation to create a checklist of rework. It also records the reason for deviation, in terms of cost and resource, if the true impact is different to the assessed impact. The cost, resource, deviation, and reason for deviation are stored in a database and are available when a similar change is required on a similar project, allowing compensation to be applied to the predicted impact. A possible way to allow compensation is to analyze previous projects and assess the percentage of deviation (PD), in terms of cost, resource, and time, for similar changes. Then after initially predicting the impact of the change based upon the process map, the PD can be added if the reason for PD is assumed likely to occur in this change.
This paper reviews some design management models and identifies how they can be adapted to address problems in design change management. The CDCM model is then proposed as an aid to assessing the impact of design changes. Although the focus of this paper is about how the CDCM model has been developed as a concept, the methods for building the model are introduced, verified, and implemented.
Problems in Construction Design Change Management
Many practitioners believe construction design is being managed effectively using protocols documenting change requests. However, after deeper interrogation of such protocols, it is evident that within each protocol, in order to aid the decision on whether to implement a change, an assessment of the impact, in terms of resource and cost, is required. After gathering data from industry specialists in an international design consultancy, Arup, it is evident that various protocols are used depending upon the project managers preference.
Figure 2 shows a process used within Arup (the sponsoring company) that assesses the impact of a change. Within this process, there are two tiers of risk for each design change; each discipline team leader is asked to assess the impact of the change for their discipline. The first tier of risk occurs when the team leader decides which team members to consult. These team members are then asked how they will be impacted by the change. The second tier of risk occurs when the team member assesses the impact based upon their experience. This can be seen in Figure 2. The opinions of the industry specialists interviewed, within Arup during this study, highlight a major discrepancy in whether they believe using a practitioner's experience is acceptable.
Some practitioners believe that on projects where teams are co-located (located within the same geographical office) it may be acceptable to use experience to assess the impact of a change. However, in projects which are complex in nature, where multidisciplinary expertise is drawn from dispersed project teams, the impact of a change is more difficult to assess since it is necessary to determine how other disciplines will be affected. It is suggested that the risk associated with making an incorrect judgment of the impact of a change is increased when a project team are both multidisciplinary and not co-located.
The authors suggest that relying solely on an individual's experience is not acceptable for two specific reasons:
- There is a risk that the individual could make an incorrect judgment. This risk is highlighted in Figure 2.
- Strategically, relying solely on an individual's experience will not add any value to the company, since when that individual is no longer employed by the firm, his or her expertise will be lost.
An accurate impact assessment of a proposed design change is essential to enable an informed decision of whether it is worth implementing the change to maximize efficiency in the design process and to prevent overruns. This research is concerned with mitigating the risk associated with a practitioner making a judgment disproportionate to the true impact of a design change. A CDCM tool is being developed to aid design practitioners in assessing the impact of a change and to record the impact of changes for future reference.
The next section of this paper is a literature review of current design management tools, which can be adapted and developed to address the problem of solely using a practitioner's experience in assessing design changes. The design management tools being reviewed address slightly different construction design management problems.
A Review of Some Design Management Models
In 1965, Steward (1981) developed the design structure matrix (DSM), a matrix representation of a process. The matrix can be reordered and interrogated to find the optimum order of carrying out tasks, through eliminating the amount of rework.
The DSM can be used as an alternative method to bar charts for scheduling design work. The advantage of using a DSM is that it accounts for the iterative nature of design work. The DSM represents the relationships between activities. Figure 3 shows a basic process map of a sample project and a corresponding DSM. The design tasks are listed both down the left hand side and along the top. Tasks are always to be carried out in the order they are listed in the matrix. Each marker (• or ×) in the matrix determines the relationship between two tasks. The task on the left of the marker is dependent upon information from the task above the marker. Since the order of tasks in the DSM is the order in which the tasks will be carried out, any marker under the diagonal is reliant only on tasks, which have already been completed. Whereas, a marker above the diagonal represents design iterations, where the information/data required is initially estimated.
The need to estimate and carry out design iterations can be reduced through reordering the matrix, hence changing the order in which tasks are carried out. Figure 4 shows the reordered matrix. When reordering the matrix, the aim is to eliminate the markers above the diagonal. If this cannot be done, it is optimum to cluster the markers into groups as close to the diagonal as possible, this is called partitioning.
The use of a DSM is essential in assessing the optimum order to carry out tasks. The use of the matrix has advantages in terms of analytically interrogating the data.
In the early 1990s, Steven Eppinger (1991) identified the need for a design process model to manage complex concurrent design. Eppinger stated, “Concurrent engineering does not simplify the design process, rather it adds a tremendous amount of intertask coupling which makes the overall job considerably more difficult”.
The relationship between any two tasks can be described as dependent (sequential), independent (parallel), or interdependent (coupled), as shown in Figure 5. Eppinger (1991) used a DSM to order tasks in a process. Once the matrix has been partitioned, the dependency can be seen between tasks. Sequential (series) and parallel tasks are positioned only below the diagonal, since each task being carried out is only reliant on data from tasks which have previously occurred. Coupled tasks are grouped around the diagonal, since iteration is required when data is required from task, which has not previously been completed. This is shown in Figure 6. If there are any markers above the diagonal, some coupling of tasks is required. The design manager can manipulate the DSM to manage the design program in various ways. Through removing coupling, design time can be reduced and through adding additional coupling, product quality can be improved. The design manager can explore various project programs and choose the most appropriate for their company or product.
The DSM was incorporated into the Analytical Design Planning Technique (ADePT) during the late 1990s. Austin et al. (1999a, 1999b, 1999c) identified that design is an iterative process requiring assumptions and rework until a suitable solution has been developed. Previous network analysis planning techniques (e.g., traditional programs, made up of bar charts) do not account for the iterative nature of design and monitor progress based upon the completion of design deliverables. ADePT uses a DSM to identify the information required to carry out a task. The availability of this information is then monitored to facilitate more effective planning and management of building design.
ADePT is a software tool made up of three main stages (Figure 7):
- A generic design process model (DPM) of the detailed design phase,
- Design structure matrix analysis, and
- Project and discipline design programs.
The generic DPM was compiled using a modified version of the IDEF0 notation (called the IDEF0v notation, see Figure 8), taking into account intra- and cross-disciplinary information (Austin et al., 1999a). The DPM contains 600 design tasks and 4,600 information requirements. At the beginning of each project, the generic DPM is made project specific through eliminating unnecessary tasks depending upon the project constraints, this is done through a series of prompted questions in the ADePT Design Builder software.
The project specific model is transferred into a DSM. Analysis (partitioning and tearing) is then carried out on the matrix to find the optimum order to carry out tasks to reduce the amount of rework. Once the order has been established, a project program is developed. The amount of resource is then added to the program and the DSM is refined until the optimum design program is identified.
ADePT management have recently used ADePT to assess the impact of design changes, through manipulating the DSM, on a commercial project (Paul Waskett, ADePT Management Ltd., personal communication, October 2, 2009). After identifying a task that requires re-evaluation because of an imposed design change, it is necessary to move that task to a future period in the schedule. This means moving the task down the matrix until it is below the task currently being completed. Once the task is in place, it is necessary to identify if any other tasks require rework by checking if the task has any dependencies in the upper diagonal. If other tasks require rework, they must also be moved down the matrix. The process is only complete once all rework is identified and there are no more dependencies in the upper diagonal (or the dependencies have been partitioned). Figure 9 shows an example of tracing rework in a DSM. Figure 9a shows the DSM before any changes are imposed. After completing floor beam design (task 10) a change is implemented; this requires rework to the floor system design (task 8). In Figure 9b, task 8 has been moved down the matrix. It can then be seen that task 10 is dependent upon task 8; therefore, task 10 must also be reworked. Hence, task 10 is moved down the matrix, as shown in Figure 9 c. The final matrix represents the new order required to complete the tasks, including the rework. This method identifies what needs to be done and the optimum order to carry out tasks once a change has been implemented.
At a similar time that Austin et al. (1999a, 1999b, 1999c) were developing ADePT, Choo et al. (1999) developed the WorkPlan database program to aid in developing weekly work plans adopting the last planner methodology. The term last planner refers to the individual or group of people who decide what tasks are to be carried out on a day-to-day basis. Traditionally the last planner will allocate work either based on “project schedule” or “whatever is generating the most heat” (Ballard, 2000.
Choo et al. (1999) explained that traditionally program schedules are produced by a project manager who may not have a clear understanding of the work to be performed. For example, they may be unaware of what the constraints are on a task and whether the resources required to carry out the task are available at the required time. This traditional schedule identifies what task SHOULD be done at a given time, in order to satisfy the project objectives. The last planner system (LPS) proposes that the last planner also executes a schedule. Their schedule should take into account both what the management believe SHOULD be done and combine this with what physically CAN be done. The last planners schedule represents what WILL be done. Ballard and Howell (1998) suggested that the last planner should carry out this schedule on a weekly basis using a weekly work plan.
The WorkPlan software, developed by Choo et al. (1999) is a database tool, which can be used to guide the last planner through the process of preparing weekly work packages, identifying constraints, releasing work packages, allocating resources, and assessing reasons for divergence from the schedule. WorkPlan is an alternative to other scheduling software in that it takes into account the availability of resource and equipment required to carry out a task.
A further advantage of WorkPlan is its requirement to record the reason for any divergence from the weekly work plan. WorkPlan uses these data to produce a report representing the reliability of the current planning system. This report can be used by management to identify where improvement is required in the planning system.
During the early 2000s, researchers from the Lean Construction Institute, ADePT research, and WorkPlan research worked together to combine ADePT and WorkPlan Last Planner to develop the DePlan tool. DePlan is a tool for integrated design management, incorporating the ADePT method for optimizing tasks and the last planner philosophy for scheduling work which not only SHOULD be done but CAN be done (Choo et al., 2003).
How Do These Design Management Models Need to Be Adapted to Address the Problems in Construction Design Change Management?
Current practice within the sponsoring company shows the most common method used for design planning to be the traditional project program, usually using Microsoft Project or Primavera software packages. Applying sections of ADePT in the reverse order will allow a traditional project program to be converted into a DSM. Once a design change is proposed, this DSM can be manipulated as suggested by ADePT management (Paul Waskett, personal communication, October 2, 2009) to reorder the matrix to determine a checklist of design tasks, which require rework. This checklist could then be converted into an IDEF0v process map, which will allow the practitioners assessing the impact of a change to visualize the required rework.
Eppinger (1991) described how the DSM can be manipulated. Through coupling and decoupling, design quality can be improved or design time reduced. This can apply to redesign, for example, through uncoupling tasks and deciding the results of the previous design revision will suffice and design time can be reduced.
The last planner philosophy applies to project planning. A similar philosophy can be used to analyze the impact of a change during the design phase of a project. Throughout this paper, the authors will refer to this as the last practitioner philosophy. The last practitioner is the person who assesses the impact of design changes during the design phases of a project; they are different to the last planner because they may not be involved in planning the original design program. The last practitioner is aware of the physical constraints on carrying out the tasks. They can assess what physically CAN be done in addition to what the change dictates SHOULD be done. The last practitioner can make a more informed impact assessment than someone higher in the hierarchal tree (e.g., a project manager) since the last practitioner is aware of the physical constraints of a task. Current practice within the sponsoring company uses this last practitioner philosophy, as can be seen in Figure 2, where the “team member assesses impact based upon experience.” The proposed CDCM model is concerned with supplying the last practitioner with as much information as possible to allow him or her to make an informed decision, and hence, mitigating the risk associated with the practitioner making a judgment disproportionate to the true impact of the design change.
The last planner philosophy uses percentage plan complete (PPC) and reasons for any deviance in competition to track the percentage of assignments completed in each weekly work plan (Ballard, 2000). This can be adapted for tracking the deviance in assumed impact in terms of cost/time/resource compared to the actual impact of a change. The reasons for any deviance can be recorded and feedback given to the last practitioners. A possible way to allow compensation is to analyze previous projects and assess the percentage of deviation (PD), in terms of cost, resource, and time, for similar changes. Then after initially predicting the impact of the change based upon the process map, the PD can be added if the reason for PD is assumed likely to occur while implementing this change.
Proposed Construction Design Change Management Model to Aid in Assessing the Impact of a Design Change
The important factors in proposing a CDCM model is that it is of benefit to the practitioner who will use it. It is not possible to eliminate the risk of incorrect impact assessments being made. However, it is possible to supply those practitioners with as much data as possible to enable them to make a more guided impact assessment, using both a checklist created from a DSM analysis of the current program and a database of previous changes on similar jobs.
Figure 10 represents the proposed CDCM in visual form. This section summarizes the model by describing each element of the model in turn. Although Austin et al. (1999a, 1999b, 1999c) suggested that traditional project schedules (bar charts, etc.) are unsuitable for managing design, this is still the normal method used for planning in many construction design consultancies. The first stage of the proposed CDCM is to take the traditional project schedule and convert this into a DSM, keeping all existing dependencies intact.
A request for a change can occur for any reason at any time and can be instigated from various sources, for example, a client change, a request from the contractor, or any discipline of the design team. Once a design change has been proposed, analysis can be carried out on the DSM to identify the required rework (through moving effected tasks down the matrix until they are positioned ahead of current time). Since the original project schedule contains details of tasks, people, and disciplines, the revised DSM forms a checklist of tasks requiring rework. A generic model can then be used to add detail to the checklist, providing the project manager and the team leader in each design discipline with a detailed list of rework.
A process map of identified required rework can be produced from the data stored in the revised DSM to help the relevant practitioners to visualize the impact of the change. The appropriate project manager or discipline team leaders can identify from the checklist and process map which team members will be affected by the change. These team members will be referred to as last practitioners since they are the people involved in carrying out the task, and appreciate the constraints on the job, for example required resources.
The last practitioners will be asked to assess the impact of the change for their tasks. The last practitioner is supported in making the impact assessment through both the DSM analysis of expected rework and historic data of changes made on similar projects. When making the impact assessment, the last practitioner needs to take into account the physical constraints on resource to identify what CAN be done in addition to the revised DSM account of what SHOULD be done.
Once each last practitioner has assessed the impact for their tasks, the discipline team leader, project manager, and possibly the client (if client authority is required) can make an informed decision on whether it is worth implementing the change.
Once the change has been accepted, each last practitioner can be informed to implement the change and start rework. Once each last practitioner has completed the rework, each discipline leader is to be notified. The discipline leader or project manager can then assess whether there was a deviance between the true impact and the assessed impact of the change. Any deviance can be reported and the reason for deviance checked, the last practitioner could be informed of the deviance to help them in future assessments.
The information collated through carrying out the change is recorded in the form of a database. The impact of the change in terms of duration, cost, and resources is recorded, in addition to any deviance from the expected impact and reason for deviance. When a change is not implemented this is also inputted into the database for future reference. This database forms an integral part of the CDCM model, because when a change is requested, the database can be scanned for previous projects with similar changes; the impact of these similar changes are then available to the last practitioner when assessing the change. For example, the last practitioner may identify that on similar changes on previous projects, the impact has been consistently underestimated, and therefore, factor in additional resource to account for this deviance.
The final stage is to update the Microsoft Project schedule, ensuring that there is always an up-to-date program.
Work in Progress and Future Work
The continuous focus of this research is resolving a problem identified in the construction design industry. This research has been presented to many practitioners (both project managers and engineers) within the sponsoring company, and the perceived value of the tool has been approved by those who would be using the tool. However, the proposed CDCM model is in the planning/building phase only and the following stages are to verify and implement the tool.
Although the authors believe the CDCM tool would bring valuable expertise to design change management with little extra day-to-day work for practitioners, it is necessary to appreciate and understand fully any resistance caused by practitioners believing the model requires them to carry out additional work.
This paper has discussed the concept behind producing a support tool to aid in assessing the impact of design changes. In order to produce a tool, which can be used by practitioners on all built environment projects, there is scope to develop the tool further through benchmarking against other industries. The software then needs to be produced, verified, and implemented.
Benchmarking will be carried out through interviewing practitioners from a variety of industries, including construction, aerospace, and manufacture. There are two purposes to benchmarking: (1) to learn from other industries and to incorporate these lessons, where applicable, into the construction design change management model; and (2) benchmarking will provide a platform for determining whether the CDCM model would be applicable for use in other industries.
The software elements of the CDCM model are currently being produced. The main platform will be an executable program produced using Visual Basic. The benefits of using Visual Basic is that it can be used to produce professional, easy to use programs which are familiar to all Windows users. Visual Basic code can also be used to extract the data held within Microsoft Project. It is expected that all software will be produced by early
2011. Some verification of the code will occur during the software building stage. This will be done through testing each block in the CDCM as it is built. This verification will be carried out using test data, supplied by Arup, from previous projects.
Once the software build is complete, it will be verified and tested comprehensively again, using test data supplied by Arup. The authors will work with practitioners to track changes on a range of projects over a three-month period. From here, the tool will be implemented for use on one project initially; it will be monitored and improved, taking on board practitioner comments, before being ruled out on further projects.
In construction design, no single protocol is used to manage design changes. Very little is known about the consequence of design changes. Currently design practitioners use their experience to assess the impact on their specialist field. A CDCM tool is proposed to help mitigate the risk associated with a practitioner making a judgment disproportionate to the true impact of a design change.
The research carried out so far has clearly identified a need for the CDCM tool. Various design management models have been reviewed and suggestions have been made regarding how they can be adapted in order to help evaluate change. A CDCM model has been proposed, incorporating DSM analysis and process maps to create a checklist and visualization of redesign tasks, in addition to creating an historical record of the impact of a change for future reference. The reason for any deviance between the real impact and the expected impact of a change is recorded in a database and recalled when a similar change on a similar project has been proposed.
The remaining stages of this research are to carryout benchmarking against other industries, build, verify, and implement the tool for use in the sponsoring company.
This work has been undertaken as part of an engineering doctorate research project titled “A Support Tool for Assessing the Impact of a Proposed Structural Design Change.” The research is funded by the EPSRC and industrial sponsor Ove Arup and Partners.
Austin, S., Baldwin, A., Li, B., & Waskett, P. (1999a). Analytical design planning techniques: A model of the detailed building design process. Design Studies, 20(3), 279–296.
Austin, S., Baldwin, A., Li, B., & Waskett, P. (1999b). Analytical design planning techniques (ADePT): Programming the building design process. Proceedings of ICE, Structures and Buildings, 134, 111–118.
Austin, S., Baldwin, A., Li, B., & Waskett, P. (1999c). Analytical design planning techniques (ADePT): A dependency structure matrix tool to schedule the building design process. Construction Management and Economics, 17(2), 155–167.
Ballard, G. (2000). The last planner system of production control (PhD thesis). Birmingham, UK: University of Birmingham.
Ballard, G., & Howell, G. (1998). Shielding production: An essential step in production control. Journal of Construction, Engineering and Management, ASCE, 124(1), 18–24.
Choo, H. J., Tommelein, I.D., Ballard, G., and Zabelle, T.R. (1999). WorkPlan: Constraint-based database for work package scheduling. Journal of Construction, Engineering and Management, 125(3), 151–160.
Choo, H., Hammond, J., Tommelein, I., Ballard, G., & Austin, S. (2003). DePlan: A tool for integrated design management. Journal of Automation in Construction, 13(3), 313–326.
Eppinger, S. (1991). Model-based approaches to managing concurrent engineering. Journal of Engineering Design, 2(4), 283–290.
Steward, D. (1981). The design structure system: A method for managing the design of complex systems. IEEE Transactions on Engineering Management, 28(3), 71–74.
© 2010 Project Management Institute