Something I talk about a lot is the virtuous relationship between design modeling and the process of design itself. I use the term “design modeling” to indicate the broad practice of creating virtual models that approach a true representation of the designed object or system. My contention is that the act of design modeling improves the design itself; creating visual and informational representations of a design as a part of the design process (and not merely a communication of the final design) is a feedback mechanism that improves the quality of design.
This is not a new idea, certainly. Lots of people get it, know it. Stating it makes it seem trivially obvious.
However, I think that there are large chunks of people who do design that don’t get it. If you were to state the above they’d nod their heads and say “sure, of course”, but on a day to day basis they don’t practice it. Design modeling is something that is intentional and contemplative.
So with that in mind, I’m going to talk about design modeling for a little bit.
Design Modeling in the Built Environment
In the AEC (architectural, engineering, and construction) industry, the term for design modeling is BIM (Building Information Modeling). The idea with BIM is that instead of 2D drawings of pieces of a building that have no informational dimension to them, you create 3D parametric objects that can be assigned informational properties. A 3D window object, for example, can be assigned properties like visible light transmittance, insulation value, cost, etc. A pump can be assigned values for flow, energy consumption, cost, suction head required, etc.
The true value of BIM comes when the properties of all of the elements talk to each other. The chilled beam element is assigned a flowrate. The piping system reads the flowrate from the chilled beam, and all the other elements in that piping system; the total flowrate is passed on to the pump, which can be read off and compared to the equipment selection.
The closer the virtual model is to reality, the better the feedback is for the designer. There are two critical parts to design feedback: speed and accuracy. Speed is simply the reaction time of the simulation to varying input. If you have to sit around and wait for your computer to crunch away through the night if you vary one input, the connection in your mind to how everything fits together is going to be very tenuous.
Garbage In, Garbage Out
Last week I modeled three different mechanical distribution systems for a building. It was short notice and high priority, so I threw the first two models together quickly. They weren’t very pretty but they for the most part visually communicated what we intended to communicate – this chilled beam goes here, those ducts go there, pipe routes through here, etc. Since I did it quickly, it was only a 3D model; I didn’t take the time to build any useful information into the model elements, even though the software gives me that capability.
Consequently, as our team dug into the model and started vetting it, actually figuring out what was going on in the design model proved difficult. We wanted to figure out how much air the model was actually calling for and if the duct sizes were appropriate, which involved counting up the airflows of all the terminal units and summing them together. Doing so by hand (which we had to do, since I hadn’t built the model to be able to even support going back in and inserting intelligent information) proved time-consuming, complex, and error-prone — especially since we were under a crunch (meaning sleep-deprived, over-caffeinated, and irritable).
Doing it Right(er)
Two days later I built another model of a similar but distinct system type. This time, I took the time to make sure my model elements could have the airflows built into them, and made sure that they talked to the ductwork system accurately. It had been a while since I’d done that so it took me a little bit of extra time to clear the cobwebs, but once I had it figured out I was able to finish the model nearly as fast as the previous models I had built. And since this time the entire air-side system actually knew how much air flow I had designed into it — since the design model was a closer approximation to reality than my previous models — the design vetting and coordination process was much faster, simpler, and less prone to error.
I used a view that color-coded the ductwork according to friction loss, sized the ductwork in the software on the fly, and at the end had a visual indication that all of the duct branches and mains were sized within our design guidelines. I was able to create an automated equipment schedule that read off the airflow of each element, summed everything, and provided immediate feedback on the actual amount of air I had designed in versus our targeted values.
My little bit of extra effort to make the model intelligent and informational, rather than just three dimensional, saved us hours and provided a better sense that our design was in line with our actual intent.
If you take the time to do it right, design modeling supports better design, engineering, and understanding of the design itself. There’s a mental feedback the closer your virtual model approaches physical reality, and that cycle is virtuous.