The Aesthetic of Design For Manufacturing & Assembly

Every project has its challenges, those parts of the project that leave you mentally exhausted, but I think most of us also have those things that truly charge us up. For me, one of those has always been DfMA (Design for Manufacturing and Assembly) and within that, it has been creating Skids.

Redesigning a job to be more modular or breaking down a mechanical room into prefab skids and assemblies is probably one of the most engaging parts of being in VDC and I, like many others, do it almost compulsively when looking over a design model or drawings. Figuring out in virtual models or in the fab shop how to adjust the designs to fit the fabrication techniques available has been the ultimate mix of Legos and video games. I have stayed up late into the night chasing a better design layout or trying to solve a challenging system in an overly small area. I love it.

When I moved to CAISSON, I assumed DfMA would still be a part of my job, and I looked forward to it. I figured I would just be helping others with it instead of doing it myself. So I was surprised to realize when it started coming up that I had never written down anything about that process, and worse than that, I really had little idea how to teach or even talk about that process to others. In practice, I had always known who was good at it on my team and just picked those people to hand off to. I viewed it as an almost innate talent. There were people who were good at it, and people that were not.

Since taking this role at CAISSON, I have gotten to spend time with a wider variety of designers and installers, most of them more than happy to talk skids and DfMA. In doing that, I realized that the people who seemed to have a talent for it talk about it more in terms of an aesthetic than a process. They do not say, “First I do A then I do B.” They talk about symmetry, space, and the feel they want. I am talking about pipes, valves, and steel. These are not architectural spaces, but the comments they were making mirrored those of architects talking about much more refined spaces. I heard similar sentiments from a Foreman in the field. The Foreman wants to balance minimal fittings against the idea of clean lines and efficient install. They would never cut diagonally across space to get pipes to a sink just to save a few fittings, regardless of how much the estimator might beg them to do so.

So, in thinking about how to teach DfMA, I believe that beyond the processes for manufacturing we need to help people build that aesthetic understanding. This means that exposure is extremely valuable. If we want people to produce great skids, we need to expose them to as many great skids as we can. We also should show them bad skids. I think we can identify key properties that make up a great skid or assembly, and we can even define a process of a sort, but having a good understanding of the aesthetic may be far more helpful.

I’m not saying that DfMA is an art. The aesthetic I’m talking about is more to do with comparing values. Symmetry would have a value that is in tension with the amount of material and valve access and free space. People are good at making these comparisons and judgments, and often do so almost intuitively, but generally only if they visualize the options. In training a person to make great skids, we are building a mental library to make comparisons with.

This is similar to what is being done by computers when machine learning is applied to design challenges. The computers produce millions of design iterations, then evaluate and optimize those designs to fit criteria provided by the programmer or designer. The criteria define what a good design is, and acts like an aesthetic for the designs to be measured against.

In structural components, this process often produces forms that look unlike anything a human designer would create. Some components can be visually fascinating, looking like bones or skeletal structures. For the structural designers, these represent a new aesthetic for design efficiency. In terms of mechanical systems, the results are not as dramatic. Mechanical designs are generated using components that limit the design variations, but as technology and manufacturing advances, we are likely to see even mechanical designs take on a new appearance.

For now, I believe the limitations of machine learning solutions mean that the vast majority of DfMA design work will still fall to humans to perform. So, understanding how to communicate DfMA practices and train people to create skids and assemblies is likely to remain a major part of off-site construction for the foreseeable future. This means that if we as an industry want to have better DfMA designers, we need to find ways to encourage design exposure. Making an effort to send designers to places that allow them to build those mental libraries should be a critical part of their training. Places like prefabrication events and conventions, or even just more time in the fab shop or on site, can be hugely beneficial in building an understanding of the aesthetic of DfMA and enhancing the quality of the skids and assemblies they produce.