Opportunities abound, but questions remain.
By Terry Wohlers
"Viewpoint" is a monthly
column authored by Terry Wohlers for Time-Compression
This column was published in the November/December 2004 issue.
Rapid prototyping (RP) technology has become indispensable at countless design and manufacturing organizations worldwide. The opportunity now exists for RP to penetrate other industries, including architectural/engineering/construction (A/E/C). My father was a building contractor, so I grew up around construction sites. Long ago, I developed a deep appreciation for what goes into a building, and more recently, envisioned the benefits that quick and inexpensive models might provide to this large and developed industry.
CAD tools for A/E/C professionals developed rapidly in the early 1980s, roughly parallel with CAD software for electrical-mechanical design. The introduction of AutoCAD and the personal computer made using it affordable and productive for even the smallest firms. As the years passed, more powerful software and hardware became available, allowing designers and drafters to illustrate proposed designs in three dimensions. To date, many construction projects, ranging from small to spectacular, have been described in 3D using CAD.
The prospect exists to apply RP to A/E/C with the hope of reaping many of the same benefits derived from its use in product design. RP is a powerful communications tool that promises to ease the labor-intensive effort of making architectural models. For many years, this industry has crafted elaborate models and site designs using a wide range of conventional shop tools and materials such as paper, wood, foam and plastic. RP could dramatically speed the process and reduce cost.
At the COFES (the Congress of the Future of Engineering Software) conference in May 2003, I attended a session of about 35 people that debated the pros and cons of designing and communicating in 3D over 2D in the A/E/C industry. Before attending this session, I had assumed that this industry had made good progress in its transition to 3D over the last several years. However, it was brought to my attention that most continue to resist its use. I asked the group, “If scaled, physical models were available to you for free, would that motivate you to design in 3D?” The question received a lot of reaction from this vocal group and I was hoping for some consensus. Instead, I heard a range of comments that made me wonder whether this industry would ever be ripe for RP. Also, I learned that this very educated and well-connected group, including many industry pioneers and luminaries, was not up-to-date on the advances in RP. At that moment, it became crystal clear to me that this industry had not advanced much in its transition to 3D, compared to the world of manufacturing.
About a year later, I became affiliated with a residential construction project that involved a CAD software product called SoftPlan version 12. After communicating with the designer and receiving files from him, I discovered that the software semi-automatically creates a 3D model of the design when producing the floor plan and elevation drawings. With an interest in this project and RP, I wanted to see how easy or difficult it would be to produce an RP model from the data. I was aware that most 3D CAD architectural models were not closed volumes as required by RP, but I was hoping that SoftPlan might make it easier. Architectural designers that produce 3D models usually produce them for visualization and presentation purposes, so if the model looks good, it’s good enough. Those familiar with the requirements of RP understand that this thinking is flawed if you’re hoping to produce an RP model without a lot of hassle and expense.
Initially, I was fairly impressed by the data sent to me by the designer. I received DXF files of the main floor plan, basement plan and elevations. The elevation drawings included front and rear perspectives of the house. However, I did not receive a single 3D model of the house in any file format. After requesting it repeatedly from the designer, I discovered that no one had ever requested this data from him, so he was unfamiliar with how to export it. This required a phone call to a technical support person at SoftPlan Systems to learn how to do it. The procedure was simple and straightforward.
The 3D model data looked okay overall, but upon closer examination, one could easily see that it was anything but a closed volume. For example, some small sections of the roof were missing. Also, there were no floors in the house, yet it contained furnishings and cars in the garage. Knowing that it was going to require some special work, I contacted Charles Overy of LGM (Minturn, CO), a company that produces visualization products for architecture, development and land planning.
Overy has created software that shrink wraps the exterior of a CAD model and creates a new, closed volume so that the model can be built on an RP machine. The software was designed specifically for problematic CAD files produced by A/E/C professionals and it hopes to bridge the gap from architectural CAD to RP.
Overy began by importing the 3D model data into Form-Z to redraw missing parts and make corrections before creating a buildable STL file with his software. He said that Form-Z and Rhino work well for this kind of modeling and repair work. Some of the missing elements included the main and lower floor slabs, sections of the roof and an area above the front door. Also, he had to move one wall plane because it was misplaced. He replaced the back deck, which consisted of planks and trusses, with a single extruded shape so that the scaled down model would survive the RP build and post processing. He then exported the file as a 3-DS file and ran it through the online version of his software to create a watertight surface model.
It became apparent that the 3D model data that I received and forwarded to Overy was not in the best of shape for RP. Based on e-mail correspondence with the designer and SoftPlan technical support, it was unclear to me whether the problem was with the software, the user, or both. In this case, it probably didn’t matter. I learned that the problem is with the process.
Marty Doscher of Morphosis (Santa Monica, CA)—a company founded in 1972 that produces unconventional architecture—agrees. “Neither the designer nor the software cares whether the model is a closed volume,” he says. And because RP has never been an objective, there is a general lack of understanding of the requirements among users and producers of 3D CAD software for A/E/C applications. Most A/E/C software producers and users don’t see the need or opportunity for RP.
Cost also has been an obstacle to the acceptance of RP for A/E/C. While the price of RP systems and models has declined dramatically, it’s still relatively expensive. This is especially true when you consider the amount of machine time and material that is consumed when building one of these scaled models. It’s not unusual for one to consume the entire build volume of the RP machine.
Construction drawings are the standard for communicating design intent among builders. However, the digital representation of these drawings does not carry the information needed to construct a physical prototype using an RP machine. Some A/E/C professionals create 3D CAD models, but they are an exception. When they are produced, they are usually not fully closed, watertight (solid) models required by RP systems.
As RP prices decline further and CAD software and repair tools improve, the A/E/C industry will warm up to what RP has to offer. In the future, documentation for a construction project will consist of drawings and a scaled RP model that point to one another. According to Doscher, companies also will use the 3D model data to drive manufacturing processes, such as milling machines and RP systems, to fabricate odd shapes in full scale for the actual building. In fact, his company is already beginning to do it. “It’s a very small but growing piece of the business,” he says.
It will take years—possibly a decade—before A/E/C fully embraces the idea of using CAD data and RP to routinely produce scaled models. To the makers of RP machines, this may not matter. Even a small fraction of this large industry could translate into interesting business opportunities for them. When it becomes mainstream, brace yourself for an impressive surge in RP system sales. However, be patient because it will take time.
Industry consultant, analyst and speaker Terry Wohlers is principal consultant and president of Wohlers Associates, Inc. (Fort Collins, CO). For more information visit http://wohlersassociates.com.
By Charles Overy, LGM
Some architectural CAD programs already export files that look like they might be suitable for RP. However, there are four issues that frequently make this data difficult to handle or completely unusable.
Unlike most traditional RP models, architectural models are small representations of very large objects. Common model scales in the U.S. are 1 inch = 8 feet, 1 inch = 16 feet, and 1 inch = 50 feet. These scales represent reductions of 10 to more than 100 times. Common building elements, such as glass or a steel beam, become paper-thin when they are scaled down and attempted on an RP machine.
Most building designs are communicated on paper or in 3D virtual renderings, so there is very little benefit to using the kind of solid modeling software that is now fairly common in other design professions. A 3D plane may rightfully convey the design intent for architecture when a 3D solid, by itself, would be used for mechanical design. As a result, even when 3D is used, CAD programs usually do not generate the kind of closed volumes that are required by the RP process.
A building is not a single solid. Most A/E/C projects are more similar to complex mechanical assemblies than they are to an individual part. Architectural CAD projects make use of object libraries and components, particularly when the project is drawn in 3D. These objects are likely to have the scale and solids problems described above. When multiple instances of these objects are used in a building design, the problems of exporting an STL model are compounded.
Unlike many other RP processes, where the engineer is looking for a model that closely matches the design in scale, and often in material and function, a physical architectural model is almost always produced for appearance only. However, what one sees in an architectural model depends on the scale and other factors such as the construction material and the maturity of the design. An example is the small cast souvenirs that one can buy of structures such as the Empire State Building and the Eiffel Tower. These curios intuitively convey the shape and design, but in reality, the geometric features of the model are nothing like the actual structure.