1-2-3 Revit: Not All BIM is Parametric15 Feb, 2005 By: AIA ,Rick Rundell Cadalyst
Learn to recognize this modeling engine so you can ensure your software choice is effective for building design
One of my first articles for this column explored the pros and cons of the various types of design technologies currently in use for BIM (building information modeling) solutions: CAD, object CAD and parametric building modeling. This month's article will focus on the most advanced of these technologies, parametric building modeling. Next month I'll discuss more extensively why parametric building modeling is so critical to BIM and is the de facto underpinning of a purpose-built BIM solution like Autodesk Revit. This month, I'll start by offering a layman's definition of parametric modeling as it is used for building applications, then I'll provide several tests you can use to determine if a particular BIM solution is using a true parametric building modeler as characterized below.
What is Parametric Modeling?
Volumes have been written about parametric modeling. In this short article, I can't possibly detail all its intricacies. But for those who want further information, I'll provide some relevant links at the end of this article.
Original CAD engines used explicit, coordinate-based geometry to create graphic entities. Editing these "dumb graphics" was cumbersome and extremely error-prone. Documentation was created by extracting coordinates from the model and generating standalone 2D drawings. As graphics engines matured, graphical entities were combined to represent a design element (a wall or a hole, for example). Depending on the software, the models became "smarter" and were a bit easier to edit. Surface and solid modelers added more intelligence to the elements and enabled the creation of complex forms.
But the result was still an explicit (coordinate-based) geometric model, which was inherently difficult to edit and had a tenuous relationship to extracted drawings that easily fell out of synch with the model.
Then came parametric modeling engines that used parameters (numbers or characteristics) to determine the behavior of a graphical entity and define relationships between model components -- for example, "the diameter of this hole is 1 inch" or "the center of this hole is midway between these edges." This meant that the design criteria or intent could be captured during the modeling process. Editing the model became much easier and preserved the original design intent.
This was the breakthrough that gave credibility to the concept of a digital design model. The mechanical design world (at the forefront of parametric modeling) made MCAD parametric modeling the status quo for mechanical design.
What About Buildings?
Unfortunately, MCAD parametric modelers don't scale to a building project. They usually rely on one of two basic technologies to propagate change: history-based, which plays back the design steps for the model each time a change is made; or variational, which attempts to simultaneously solve all conditions with each change. Using these change engines to resolve even a small building is prohibitively slow.
MCAD modelers also generally require the user to embed a lot of constraints -- that is, relationships -- so the change technologies described above can recalculate the result. These "fully constrained" models are unworkable for designing buildings -- and unnecessary, as there are relatively few constraints that really matter to a building designer.
The technology that made parametric modeling work for building design, and therefore enabled parametric building modeling, is the context-driven change engine used in Autodesk Revit.
Purpose-Built for Building Design
Autodesk Revit uses a context-driven change engine to update a partially constrained model, creating a network of building element relationships (inferred by the software and/or set by the user). It then uses this network to help resolve changes later. When you sketch or place components, Revit retains inter-element relationships, but there is no particular order to these relationships. Subsequently, as you modify one element, the parametric change engine determines which other elements need to be updated and how to make the change. The approach is scalable to building applications because it never starts with the entire building model; it always starts with a few elements explicitly touched by the user and continues with selective propagation of changes -- minimizing the number of elements that must be updated.
Is it Really a Parametric Building Modeler?
The following are some examples of how a BIM solution built on a parametric building modeler performs compared with other technologies:
1) Does your software use you to coordinate and manage change?
With a geometry-based product, the user is normally expected to identify all the geometry that is affected by the change, selecting it with a stretch box or similar command. Geometry that's not visible or is turned off cannot be selected and will have to be found and corrected manually.
In a parametric building modeler like Revit, simply selecting and moving a wall in the first floor plan will cause all the related elements to adjust automatically (figure 1). The roof will move with the wall, preserving any overhang relationship, the other exterior walls will extend to remain connected to the moved wall, and so on. This associativity is a defining feature of a true building information modeler.
Figure 1. With state-of-the-art parametric building modeling, BIM software can coordinate a change made anywhere, including on sheets ready to plot, everywhere it matters: in 3D views and drawing sheets, schedules and elevations, sections and plans.
2) Are the terms "extracted" or "generated" used to describe the creation of drawings?
If so, this is a clear indication that you're talking about a geometric modeler. Some software includes libraries of commands or utilities that regenerate or update drawings and schedules based on changes to a building model. But this process is strictly one-way, relying on CAD operators to make sure all the updates have been made. This operation is analogous to running another set of reports on a database after the data has been updated; the reports are just dead artifacts reflecting the state of the data when the report was run.
The test for a state-of-the-art parametric building modeler is whether it can coordinate changes and maintain consistency at all times. It's like working in a spreadsheet. Update the model in one place, and all views, drawings and schedules are instantly synchronized.
3) If you slide a section key across a plan view, does the section update immediately?
Conventional geometry-based products typically do not integrate graphic annotations into the building model. The full integration of drafting annotations into the building model itself is an important component to maintaining the connection between the graphics deliverables and the model. In a geometry engine-based product, a dimension string will simply be text, or at best it might update if you change the underlying geometry. In a change engine-based product, editing the dimension text will change the underlying geometry in a corresponding way.
Similarly, in a geometry engine-based product, the section view and the section key line are usually separate and distinct. The section key line is simply a dumb annotation. In a parametric building modeler, a section key often serves to define the section cut itself. Move the section key line or flip the section key and the section view will immediately update.
4) Does the BIM solution rely on "smart" or "intelligent" objects?
Object-based modelers are common today. They include at the simplest level symbolic drafting aids, much like plumbing or furniture templates that provide tracing guides for manual drafting. As the industry began to associate data such as a key number or name with these symbols, they were dubbed "intelligent" or "smart." In some cases this data, such as a height dimension, could affect the geometry of the symbol, making that data a "parameter" and the symbol "parametric." Other basic relationships such as "hosting" were introduced between symbols, allowing a window to remain attached to a wall if the wall was moved.
However, the missing piece is the network of relationships among and between all the pieces of the building. This is the strength of a parametric building modeler: recording, presenting and managing relationships no matter where they occur in the building.
An effective parametric building modeler manages object data at the component level, but more importantly allows information about relationships between all the components, views and annotations in the model. A door to a stairwell can be locked in place a specific distance from the riser of the stair to ensure egress clearances; a door can be locked a specific distance from a wall to ensure furnishing clearance or pull-side clearance for accessibility. The entire model contains information, not just the objects in it.
The essence of the architectural design of a building is in the relationships that can be embedded in the building model. The creation and manipulation of these relationships is quite literally the act of designing. Parametrics gives designers direct access to these relationships and are natural and intuitive ways of thinking about buildings using a computer, just as a spreadsheet is a tool for thinking about numbers or a word processor is a tool for thinking about words.
But not all BIM solutions are effective parametric building modelers as described above. Use the tests above to see if yours is.