AECbytes "Building the Future"
Article (August 12, 2009)
Model Based Estimating to Inform Target Value Design
Saurabh Tiwari, Josh Odelson, Alan Watt, Atul Khanzode
The use of Target Value Design (TVD) or Target Costing is one of the focus areas of the application of Lean Construction methods to large healthcare projects. The Lean Construction Institute defines Target Costing as a practice which incorporates cost as a factor in design to minimize waste and create value. The cardinal rule is that the Target Cost for a project should never be exceeded. In most traditional project delivery approaches, cost follows design, but on projects where the TVD approach is used, cost should dictate what gets designed to ensure that the target cost is not exceeded. As a result, rapid cost feedback to the design team is paramount in this process. One mechanism for providing this rapid cost feedback is extracting quantities from the virtual model and model-based estimates. In this article, we discuss the lessons currently being learned in applying Building Information Modeling (BIM) tools, such as model-based estimating, for TVD on a large healthcare project in Northern California.
Sutter Medical Center Castro Valley (SMCCV) is a $320 million, six-story, 130-patient bed replacement hospital in Castro Valley, CA, for Sutter Health. Sutter Health adopted Lean Project Delivery for all its projects in early 2003 and has been promoting the use of the Integrated Form of Agreement (IFOA) as the contract method for project delivery. The IFOA approach is similar to the Integrated Project Delivery (IPD) framework promoted by the AIA. DPR Construction is the general contractor and one of 11 members of this IFOA/IPD team for the SMCCV project. DPR was selected as the builder and part of the IPD team for this project having successfully completed a variety of Sutter Health projects, including the Camino Medical Office Building in Mountain View, CA. The Camino Mountain View project has served as an early, and very successful, example of the implementation of Lean Construction methods and the use of Building Information Modeling (BIM) for healthcare projects.
The SMCCV project has just broken ground and foundation work on the project is about to begin. The IPD team on the project has worked collaboratively over the last year to develop a highly integrated design using Lean Construction Methods, such as Value Stream Mapping, Work Structuring and Production Control, and TVD. (These are described in more detail in the AECbytes article, Sutter Medical Center Castro Valley: Case Study of an IPD Project.) The IPD team has also used BIM tools for MEP coordination, quantity extraction and estimating to support the TVD process.
Why Model-Based Cost Estimating?
In the Target Costing process, cost should be an input to design. There has been a lot of research on how to make this happen. Teams have used organizational approaches, like bringing a group of cross-functional team members together in an IPD team, to be able to provide this rapid cost feedback. Cost feedback based on a BIM has also been suggested as a potential option to rapidly iterate through the design options (see Glenn Ballard’s “Innovation in Design” Presentation available at the Lean Construction website). Stanford University’s Center for Integrated Facility Engineering pioneered the concept of estimating based on a product model in early 2000.
Model-based cost estimating is the process of integrating the object attributes from the 3D model of the designer with the cost information from database of the estimator. Model-based estimating has proved to be a leaner approach compared to traditional 2D drawing-based estimating. Using the 3D model to estimate rather than the 2D drawings utilizes the object attribute of the 3D model rather than “assuming” the same, based on flattened 2D drawings. The process is not only quicker but also eliminates scope for errors and omissions. Figure 1 compares the two processes and shows how the cycle time is reduced from 8 weeks to 3 weeks.
Figure 1. The manual estimating process compared to the model-based estimating process.
To date, there have not been many implementations of cost estimating on a large project to accomplish Target Costing goals, despite the recent hype in the industry about what BIM can do for cost estimating. The SMCCV project provided the opportunity to implement model-based cost estimating since one of the requirements during the preconstruction phase on the project was to give the team real-time input on constructability and cost feasibility and provide timely cost feedback on design alternatives and changes on a regular basis. As a result, the SMCCV project team developed a rapid, repetitive and consistent mechanism for evaluating design against the budget and the target values of the client. By adopting the leaner model-based estimating approach, the cycle time of cost feedback—beginning at incorporating design changes in the 3D model to updating cost status of design due to the changes—was dramatically reduced, as shown in Figure 1.
Current Model-Based Cost Estimating Solutions
Due to the fragmented nature of the construction industry, and consequently the BIM software industry, there is no one software solution that caters to all aspects of the cost estimating process. Every trade uses the modeling tool that best suits their needs, leading to interoperability issues. Adding to the challenge is the variety of cost databases (e.g., Excel, Timberline, Quickbid, Accubid, etc.) being used by different companies. Identifying one modeling tool and one estimating tool for a project seems impractical in today’s work setup.
There are a variety of organizations working on implementing interoperability solutions between the applications used in the construction industry, such as the buildingSMART alliance and the National Institute of Building studies, which have published the Industry Foundation Class (IFC) and National BIM Standard respectively. We hope that one day the interoperability challenge is truly tackled; however, in today’s world, we have not been able to rely on the exchanges to accomplish results on large projects like SMCCV.
Currently, there are two main options available for implementing model-based cost estimating in the market.
OPTION #1: The first scenario includes a cost estimating solution that works exclusively with just one specific 3D modeling tool and one specific cost database program, as shown in Figure 2. This type of estimating software is usually developed by the same company that develops the 3D modeling tool. For example, CADEst is the estimating tool for CADDuct/CADPipe, all of which are MAPSolid products. Similarly, VICO Estimator is the tool for VICO Constructor, and the upcoming Autodesk QTO is for Autodesk-based modeling products. The issue in this case is that if a company’s cost database is stored in different software and they use a wide variety of BIM tools, they will have to transition their whole modeling and cost estimating platform to this one-stop-shop package of 3D modeling and cost estimating. This type of tool has a greater barrier for entry in the market as if effects not only the cost estimating practices but also the 3D modeling practices.
Figure 2. Cost estimating solutions that combine a specific 3D modeling tool with a specific cost database program.
OPTION #2: The second scenario includes a neutral platform in which you can bring in models from different sources (e.g., Innovaya Visual Estimating, Tocoman, etc.). In this case, the cost estimating software developer either creates an exporter plug-in for each 3D modeling software or it uses a neutral platform like IFC to bring in the data. This type of cost-estimating software does not bind the designers to any specific 3D modeling software. Although, in this case as well, the existing cost database will have to be transitioned to the software that is supported by these model-based cost estimating tools.
On the SMCCV project, implementation of model-based cost estimating has been successful at different levels for different trades. The extent of usage of 3D modeling for cost estimating and quantification for different trades at SMCCV is described below and illustrated in Figure 3:
- Architectural and Structural: Modeled in Revit and successfully estimated using Innovaya.
- Fire Protection: No compatible model-based estimating tool available in the market to integrate with AutoSprink, which is the software used for modeling.
- Electrical: Even though the model is created in AutoCAD MEP, Innovaya cannot be used for estimating because the cost database is not in Timberline or MC2. In this case, cost quantities are being driven out of the model.
- Structural Steel: Tocoman is being explored as an option for linking the Tekla model to the Timberline cost database.
- Mechanical and Plumbing: CADEst is the preferred model-based estimating tool for CADDuct/CADPipe, since they are developed by the same company. However, the Mechanical/Plumbing subcontractors have not yet replaced their existing estimating software (QuickPen, Accubid) with CADEst. (Some of the reasons are mentioned in the next section.) As a result, cost quantities from the model are being done manually.
Figure 3. The software being used for 3D modeling and model-based cost estimating on the SMCCV project.
Challenges in Implementing Model-Based Cost Estimating
The challenges of model-based estimating go beyond finding appropriate software solutions. To transition from manual estimating processes to a model-based estimating process takes substantial effort, time and cost. In our experience, the easier part is the purchase of new programs and transferring the estimating database from one source to another. The more difficult part is the cultural shift and training required. Estimators must be thoroughly trained in the software and run test cases to make sure that the information coming out of the model is accurate and can be trusted. At first, the model-based estimating process may also take more time than their traditional way of estimating. However, after time and greater proficiency using the software, the new method should take less time than the older method, achieving results like the SMCCV project.
In addition to the cultural shift and “buy-in” challenge, there is also the question of who will pay for the transition from one software to another. Should it be the owner of the project interested in adopting model-based estimating or should it be the design-assist subcontractor, who will benefit from the set up repeatedly on other projects?
DPR’s Model-Based Cost Estimating Solution for Self-Perform Work
DPR has tried and tested Innovaya Visual Estimating (described at length in this AECbytes article), which has proven to work especially well for the self-perform scope of work (Drywall, Concrete, Doors/Frames/Hardware). Innovaya currently works in the context of AutoCAD and Revit as the modeling software and Timberline or MC2 as the estimating software. DPR uses both Revit and Timberline, and therefore, Innovaya was used as the model-based cost estimating solution for DPR’s self-perform work at SMCCV (see Figure 4).
Figure 4. The process of model-based cost estimating using Innovaya at DPR.
At present, 86% of the cost estimate comes from the model for DPR’s self-perform work (see Figure 5), which is 15% of the total project cost.
Figure 5. The extent of model-based estimating in DPR’s self-perform work in the SMCCV project.
The remaining 14% of the DPR self-perform work estimate cannot be done from the model for the following reasons:
- The element is not in the 3D model (e.g., temporary shoring).
- The element is part of the cost assembly related to a modeled component that cannot be determined by examining physical attributes. For example, the quantities of construction joints cannot be calculated from any property of slab on grade (i.e., perimeter, area, etc.). Its quantification depends on how the slab on grade is broken down into different pours.
- The model is not intelligent enough to give a desired quantity. For example, the length of a concrete wall against slab on grade will provide the length of the expansion joint, but currently this information cannot be quantified from the model, because the model does not know there is a wall adjacent to the slab on grade.
- Model-based estimating does not work when the cost is a function of time and not the 3D element. For example, construction trailers, temporary power, equipment, etc., are dependent on the duration of multiple construction activities and the project as a whole.
Model-Based Cost Estimating in the IPD Setting
In the SMCCV project, the model-based cost estimating process involved early and intense collaboration between the following:
- Architects and structural engineers, who were developing the model.
- DPR self-perform work estimators, who have the estimating knowledge and database.
- DPR Virtual Building Group BIM engineers, who are experienced in both areas and know how to integrate the two.
The big incentive for the architects and structural engineers to be actively involved in this process was the IFOA contract, where all—in this case 11—signatories share the common pool of profit on the SMCCV project. Since the goal of this process was to accomplish the Target Cost and, therefore, ensure that the common pool of profit pool is preserved for all parties, the IPD team understood it was essential to work together to secure their share of the profit, as illustrated in Figure 6.
Figure 6. The incentive to work collaboratively on the SMCCV project to reduce the over-budget amount to secure fully funded IFOA profit.
Comparing multiple projects, we believe that the key to collaboration is bringing in a cross-functional team of designers, general contractor and subcontractors early with an incentive to collaborate and optimize on the whole project. When this does not happen, accomplishing the same results becomes challenging for a variety of reasons, both technical and organizational. On some other projects, it has been challenging for DPR to persuade architects to make changes to their models for estimating purposes, because the contractual agreement between the owner and the architect does not stipulate that one of the intents for the 3D model is cost estimating. On other projects, we have seen issues with the interoperability between the BIM authoring system and estimating database because of the fact that team members were invited to join the design process after Detailed Design was completed and this issue was not identified until late in the process. To benefit from the time savings of model-based cost estimating, it is important for the owner to clearly state the intent of 3D modeling in a contractual format when awarding the project to different participants.
It took approximately 3 months of setup time for the cross-functional team of the architects, engineers, self-perform work estimators, and BIM engineers to automate the cost-estimating process on the SMCCV project and generate fortnightly cost estimate updates on design changes. This is outlined in Figure 7. This early collaboration in the IPD approach has resulted in significant time savings, taking one estimator in each self-perform work group just two days to generate an updated model-based cost to meet the “once every two weeks” cost estimating cycle.
Figure 7. The upfront work required to automate the cost estimating process.
Setting the Groundwork for Model-Based Cost Estimating
For the SMCCV project, the first step was a sanity check to identify components that were modeled incorrectly for estimating purposes (either the quantities were inaccurate or the elements were not broken down they way they are constructed). Next, the list of identified components was provided to the architects and structural engineers, who then incorporated those changes incrementally in the model over the span of two months. The list also included some parameters that needed to be added to the objects to automate the mapping process with the cost assemblies due to the limitations of the modeling software (Revit). For example, ceiling height information was added as a shared parameter to a wall, because it was needed to quantify the wall surface area that was required for finish taping.
Parallel to the model modification and information addition effort, existing cost assemblies in Timberline were modified so that they could be mapped with the 3D model objects. Figure 8 shows the 3D model object with the DPR object parameters fed in on the left side and the cost assembly created in Timberline on the right side.
Figure 8. Mapping a 1 Hr Wall from the 3D model to its Cost Assembly in Timberline through Innovaya.
Once the cost assemblies and 3D model objects were able to be mapped, then the cost was generated using the mapping process. The next step compared the cost with the traditional estimate to see if the costs matched. Transitioning a traditional estimate to a model-based estimate was a cumbersome process, because it was hard to resolve the quantity variations if the quantity takeoffs using the traditional estimate were not recorded.
The third part completed, as a pre-requisite, was to train the estimators in the model-based estimating tool. The time it took for an estimator to get comfortable with the model-based estimating software varied from one to two months. Once training was complete, estimates could be generated once every two weeks.
Other Benefits of Model-Based Cost Estimating
Along with being able to communicate the cost for the design every 2 weeks and with considerably less time involvement (nearly 1/5th of the time of traditional estimating due to savings in quantity takeoff and automated mapping of objects with cost assemblies), the model-based cost estimating process also helps evaluate the cost of design and construction alternatives with quantity trends for model objects, as shown in Figures 9 and 10 respectively.
Figure 9. An example of checking the behavior of the model over time—the Quantity Trend chart for Pile Caps.
Figure 10. Cost Comparison of design/construction alternatives.
SMCCV is truly a ground-breaking project, as DPR has taken model-based cost estimating to the maximum extent possible. As a result, there have been a number of lessons learned, as listed below. Most of the lessons are process and organization related; some of them are also software related.
- Senior Company Management buy-in of model-based cost estimating: If the senior company management sees the value in the model-based cost estimating process and endorses it, it is much easier to implement within the company. This is one of the major reasons why some of the trades are still generating traditional estimates as there is still resistance to move away from traditional estimating practices.
- Contractual language of the project to support collaborative work environment: Compared to non-IPD projects, it has been easier to work with designers and get requests of model modifications entertained because of the IFOA contractual setting. The IFOA leverages a collaborative work environment by providing incentives, such as a common pool of profit.
- Not all cost estimates can be model-based: Some of the items in the estimate cannot be quantified or formulated from the existing 3D elements in the model. Items like construction joints in slabs are means and method items, which need to be manually quantified. Also, there are time-based cost elements (e.g., man lifts, temporary power, trailers, etc.), which are estimated by how long they are on the jobsite and cannot be easily quantified from the 3D model.
- Transitioning traditional estimates to model-based estimates: A visual record in the form of marked up drawings of what was a part of the hand takeoff is important to have, so that quantities can be compared easily with the model quantities.
- A new software tool does not always perform the way you expect it to: Implementation of new technology is not always successful the first time. A lot of collaboration with the software developer is required to make it work to give you the desired result.
- Always check the quantities from the model at least once: Some of the elements might have been modeled using a tool that does not give you the right quantities. In case of SMCCV project, there were irregular shared pile caps whose quantities were not read correctly. Taking another example, Revit gives you the flexibility of modeling certain elements in different ways but quantification does not work with all of them. For example, openings can be modeled using an “edit profile” tool or “opening tool” or an “opening family” or a “void extrusion.” The only way openings get quantified is if they are modeled using opening tool or by using the opening family.
- Model-based cost estimating is not a click of a button process: As you may have grasped by now, there is a lot of pre-requisite work in preparing the cost assemblies, preparing the model, training the estimators, etc. All of these steps are required to make this process work successfully.
- Start the process early by the end of conceptual design phase: The earlier the teams start this process in the preconstruction phase, the more in sync the model will be for cost estimating, and the more time design will have in the design development phase to react to the regular cost updates to attain Target Value Design.
Model-based cost estimating is the latest step in the evolution of estimating technologies, starting from the hand takeoff of scaled 2D drawings to the digitizers, followed by computer-based 2D takeoff with tools like On Screen Takeoff/Accubid. Some hesitation in accepting the quantities from the model is natural. Trust in model quantities needs to be built by adopting a standardized process of quality control on the model and communicating to the modelers.
The model-based estimating process calls for the collaboration of multiple teams. The successful implementation of this process is as much influenced by the organizational and contractual setup as is by the availability of software solutions in the market that facilitate it. It is imperative for the whole project team to have an understanding on the intent of the usage of 3D modeling. The owner has an important role to play in this process, making a decision to use model-based cost estimating and to mandate that in a contractual format with the project participants. Architects and structural engineers, who build the design models, will have to work collaboratively with the contractors to understand how they intend to use the models for cost estimating and, thereby, input their feedback appropriately in the 3D model.
Model-based cost estimating is not a push of the button process. There is a lot of upfront work that is required to get to a desired level of automation. Even then, there is always some part of the estimate that does not come from the model, which requires time and effort to align with model updates. Technically, the biggest challenge is to find the right suite of solutions that work seamlessly with the 3D model and tie back to the company's cost database.
In spite of the challenges in available technology and process workflow, the advantages of model-based estimating far outweigh the upfront time and effort required to enable the process. It is leaner and smarter, because it automates the time-consuming tasks of quantity takeoff and allows the team to focus on issues like productivity factors and how location of the construction activity affects it. As a result of the automation, it makes cost adaptation to design change management easier and quicker, helping the project team get rapid cost feedback on design status.
The industry in the near future will continue to see multiple model-based cost estimating tools for different trades within a single project that work towards the same goal of using the modeled quantities for cost estimating. The project will eventually benefit from following the common principle of automating the mapping of 3D model objects with the cost assemblies and reducing the time spent on quantity takeoff. Teams will be able to give rapid cost feedback to design changes to attain Target Value Design and deliver the best possible value to the owner.
We would like to thank Digby Christian of Sutter Health for his support for making model-based cost estimating a reality on the SMCCV project, as well as all members of the IPD Team for their contributions in making this happen and helping to drive the industry forward.
About the Authors
Saurabh Tiwari, Senior BIM Engineer at DPR Construction, has managed the coordination process, quantity extraction and the model based estimating process for the SMCCV project. He is instrumental in producing the quantity updates every two weeks, and assisting the project team members in producing the estimates based on the models every two weeks.
Josh Odelson, Senior Project Engineer at DPR Construction, is currently responsible for managing the TVD process at SMCCV and is intimately involved in all the details of using the Model Based Quantity extracts, the Model Based estimates and using this data to help the IPD team make informed decisions.
Alan Watt, Director of Preconstruction Technologies at DPR Construction, is responsible for Preconstruction technologies at DPR and has helped the SMCCV project team implement Model Based Estimating by configuring the assemblies in Timberline and developing customizations to make it work with Revit Structure and Revit Architecture models.
Atul Khanzode, Director of Virtual Building at DPR Construction, leads the Virtual Building Group and helps project teams implement Lean Construction and Virtual Design and Construction methods on their projects. Atul has assisted in bringing together the various team members in making Model Based Estimating and its use for TVD a reality on the SMCCV project.
DPR Construction, Inc. is a national general contractor and construction management company based in Redwood City, Calif., which specializes in technically complex and sustainable projects for advanced technology, life sciences, healthcare and corporate office markets. DPR is recognized as an industry leader in the use of Lean Construction Methods and Building Information Modeling Technologies to deliver enduring value to customers on complex construction projects.