Every organization is perfectly designed to produce precisely the results that it does.
        —Variant of a common systems-thinking saying

Unlike other large and complex human undertakings, a plant is a very tightly coupled system of subsystems. Large airplanes can lose big pieces of their insides without affecting the flight behavior of the plane; ships can be largely dismantled, yet stay afloat as long as the integrity of the hull is not breached; big buildings can survive as habitable structures even if great chunks of them are destroyed.

But plants are different. They are much more like organisms than large inanimate structures; they have a dynamic stability, a homeostatic integrity, which is a function of all their parts interconnecting and working properly together.

I maintain that the end of all computer-aided design is simulation; that's where we no longer create only geometry, but the full range of behaviors of the thing we want to build. We model the environment in which it will be built, and how it will interact with the plant.

Because of its homeostatic integrity—the tendency of the system to maintain a particular state, including "awareness" of all its components—a plant can be monitored for metaphoric "leaks." Materials, pressure, heat, or information cannot disappear unnoticed. That makes plants easier to simulate than systems that lack such integrity, because it is much easier to symbolically represent closed systems than open ones.

Yet simulation is almost unknown among plant design and construction firms. The technology is there to do it, but the cultural and methodological infrastructure is wanting.

I've seen demos of Dassault Systemes' CATIA CCPLANT and Bentley's PlantSpace, and they provide great simulation capabilities—and I'm sure there are others. The concept of "4D design," incorporating time as a modeled dimension, has been propagated by Stanford's Center for Interactive Facility Engineering (CIFE), and has resulted in a variety of startups with 4D features in their products. (Recent interesting addition: VirtualSTEP. See Web site of that name.) But what percentage of the products' users exercise the simulation capabilities? Not a very large one.

I grant you that 4D systems require a conceptual leap and a great deal of sophistication. But how about something simple, like "sitecams"? These are inexpensive video cameras mounted in such a way as to provide a view of a construction site. They are equipped with a url and a wired or wireless connection to the Internet. Simple fixed-mount units cost only a few hundred dollars; fancy ones have servo motors that allow the viewer to "drive" the camera around the site. Why are they not on every single construction site?