Most of a product’s cost is committed in the earliest stages, when hardly anything has been spent. By the time the spending really starts, the decisions that fix it have already been made. DFMA is what you do inside that window — which makes it as much a discipline of timing as of design.
There is an asymmetry that explains almost every argument between the technical office and production: most of a product’s cost is committed in the earliest stages, when hardly anything has been spent. By the time the spending really starts — tooling, pilot runs, capacity — the decisions that fix that cost have already been made.
This is not an intuition: it is the central argument of Boothroyd and Dewhurst’s work on design for manufacture and assembly. And it has an uncomfortable practical consequence: a cost reduction programme that starts once the product is already in series arrives too late for what weighs most. It can squeeze purchasing and fine-tune cycles, but it cannot remove parts that already have a mould.
Until there is a tooling purchase order, almost anything can change and almost nothing costs. Afterwards, every change has a budget, a lead time and a conversation with the supplier. DFMA is what you do inside that window, which makes it as much a discipline of timing as of design.
DFMA puts together two things with different objectives, and they are worth separating because sometimes they pull in opposite directions.
And the classic conflict: combining four parts into one improves DFA and may worsen DFM, because that single part needs a more complex mould, with slides and a higher risk of sink marks. The decision is not “always combine”, but comparing what you save in assembly against what you pay in tooling and process risk — and that comparison is made with numbers, not with principles.
DFA analysis starts from a simple and fairly brutal idea: for each part there are three questions. If the answer to all three is no, that part is a candidate to disappear by merging into another. The result is called the theoretical minimum part count, and it almost never matches what is on the drawing.
| Part | Does it move relative to the others? |
Must it be of another material? |
Must it be separable? |
Verdict |
|---|
The theoretical minimum is not a target to hit: it is a question. A twelve-part assembly with a minimum of four is not necessarily badly designed — there may be mould cost, serviceability or manufacturing reasons that justify every one of them. What the number tells you is how many of those reasons you have to be able to explain, and in how many of them nobody has ever asked.
These criteria are for thermoplastic injection-moulded parts, which is the most common case. The specific values depend on the material and the mould, and they are set by the material datasheet and by whoever is going to mould it, not by a general table. What does not change is the question you have to ask at each point.
The material cools at different rates where the thickness changes, and that is where sink marks and warping appear. When more stiffness is needed, you get it with ribs, not by thickening the wall.
A rib as thick as the wall creates a mass of material right at the junction: that is where the sink mark appears, and it shows on the visible face of the part.
Without draft the part drags on its way out and scuffs, or sticks. Textured surfaces need more draft than smooth ones, and that is decided before the mould is etched.
A sharp internal corner concentrates stress and is where the part breaks. It also hinders filling. A small radius solves both and costs nothing in design.
A clip or a recess that does not release in the opening direction forces a slide: more mould, more maintenance, more cycle time. Sometimes it is worth it; the point is knowing that you are paying for it.
A tight tolerance in plastic is paid for in inspection, in rejects and sometimes in a different process. The question is whether that dimension needs that tolerance or inherited it from a machined version.
All of the above translates into one thing: fewer parts and fewer operations mean fewer part numbers to buy, store, inspect and restock. The assembly saving is usually the smallest of the three, and it is the only one visible in the cost breakdown. The other two sit in purchasing and in quality.
The text that formalises DFMA. This is where the three theoretical-minimum questions come from, and the idea of measuring assembly efficiency instead of arguing about it.
It places DFMA inside the complete development process and treats the conflict between combining and separating parts as a product architecture decision, not just a cost one.
The specific values for wall thickness, ribs, radii and draft come from the datasheet of the material you are going to use. They are the right source for any figure that goes on a drawing.
The international standard on general tolerances in plastics moulding. Useful for arguing with data about whether a dimension is achievable with the intended process.
DFMA without a target cost is a list of good practices. With one it becomes a criterion: every part consolidation is judged by how much it moves unit cost.
The most profitable review of a drawing costs one meeting. Whoever is going to cut the mould sees in ten minutes the undercuts and wall thicknesses that will cost three months of corrections.