DFMA

Design for Manufacturing and Assembly (DFMA) is philosophy used during the design phase in order to improve cost-effectiveness, reliability, ease of use, scalability, and simplicity of projects without compromising on performance. It is in the early design phases prior to the ordering of materials, machines, and labor that savings can be made.

There's no single checklist to fulfill and then declare DFMA completed, but there are several principles to consider.

The core tenet of my approach is REDUCE AND SIMPLIFY. You can't have sourcing/manufacturing/assembly issues for parts that don't exist! 😉

To elaborate a bit more on the benefits and why I focus most on this above all else:

• Assembly complexity grows exponentially with part QUANTITY. If connecting two parts uses one screw, there is little opportunity for confusion. Increase to two or three and the gain to mechanical security may justify it, but does it really need eight? More fasteners mean longer assembly and disassembly times, more opportunity to lose parts in the work environment, and more steps to include in a manual or guide.

• Assembly complexity grows exponentially with part VARIETY. Keeping track of eight different types of screws each with different head shapes, thread pitches, diameters, and lengths takes effort that needs to be balanced with equal or more value to the product. The reason these varieties of fasteners were made is to fulfill specific needs, whether that be flush mounting, an attractive appearance, chemical resistance, tool-free removal, etc; if the application doesn't call for that need, rule out that type of fastener. As a rule of thumb, I try to use no more than two "small" (≤ M4) and two "large" (> M4) screw varieties in an assembly.

• Using STANDARD PARTS makes servicing a product cheap, painless, and possible. Unless an application REALLY needs that M4.55 × 6 screw (and it almost never does), try to find a way to use an M6 × 10. Users are more likely to maintain a product that doesn't demand they run errands for it.

• Considering the tools accessible to the end user improves their experience with the product. If the product is for generic consumers unlikely to have much beyond common tools, stick to slotted and Philips fasteners. If they're a savvy technician or maker, hex and Torx drives become reasonable.

• Fewer parts means fewer molds to design, a lighter product, simpler packaging, less cost to ship, less material to order, etc. Every part requires attention, and many of the relationships between parts are significant as well. This means the cost reduction function for removing parts is not directly proportional, but exponentially related. The takeaway? Start weeding out parts! This is where most of the savings potential lies, but it's harder to quantify relative to the actions that are taken.

• Simpler part geometries are easier to manufacture, easier to inspect, and less prone to defects. If the part is machined, stick to one or two standard sized and commonly used bits the shop will have available, and try to lay feature out for maximum access to minimize tool changes. If a hole doesn't need to be tapered, leave it as drilled; if two parts have holes that align, design them so the stock can be clamped together and only one plunge of the bit is required. If it's critical to function that a feature be a certain size, then place it accessible for easy inspection. Think about the design of an inspection jig for the part; if it seems complicated or troubling then the part's geometry could use some simplification. Features of constant thickness and size tend to bear thermal and mechanical loads symmetrically, and are less likely to crack or warp during cooling after forming than an irregularly shaped geometry that dissipates heat unevenly.

Some other DFMA principles I incorporate include:

• Consider the part's production volume and compatible processes. Is this a low fidelity prototype part that I don't need more than five of? I can use FDM printing and have some overhangs in the model, then. Is the precision and tolerance of FDM too coarse? Maybe the pieces can be machined. Raise the volume to tens or hundreds of thousands and injection molding may be suitable, in which case all overhangs need to go. It may be confusing on whether to start with available processes to inform design or start with a functional design to rule out incompatible processes, but the only wrong answer is to neglect process entirely.

• Consider ways to modularize any subassemblies. If I design a suite of products that all use AAA batteries, I would save myself a lot of development time, inventory management, and stock keeping organization if I created a single part to seat them rather than a new one for each product. Modularization and data reuse shares many of the benefits of removing parts altogether, since it is another form of reducing redundancy.

• Consider asymmetry to constrain how parts can fit together. Making incorrect assembly impossible is the cheapest way to avoid it from occurring. A tapered pin can only be inserted into a tapered hole one way, whereas standard symmetric dowel pins can be installed one of two ways; if a pin's angular orientation in a hole is important, add a locating notch or other feature so it can only be installed the proper way. Only a small amount of asymmetry is necessary to add installation control while retaining much of the simplicity benefits of a mostly symmetrical design.

• Consider where precision is nice, versus where it is necessary. While it would be nice to have all features perfectly sized and placed, it simply isn't possible. I like to imagine my part or product has a limited "precision budget" that depletes as I "spend" on features to improve their form or position. The mental construct makes me reflect on whether I'm designing "over budget" and is time to revisit my tolerances and recoup some of my budget. Features like clearance holes don't need as much precision as mounting holes.

• Physically start a design working from a piece of stock. When a model starts as a length of stock, the part specification tree outlining all the features becomes a literal to-do list. If the list is very long, producing the part will take AT LEAST that long, and in nine times out of ten it will take longer.

An excellent resource for further reading is UFL's design lab DFMA tips page, which can be found here.