Question 2 of 7: Design for Manufacture and Assembly

Question 2: Design for Manufacture and Assembly (15 marks)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Part A — Objectives of DFMA (3 marks)

Design for manufacture and assembly is the deliberate use of production knowledge as a design input rather than as a downstream constraint. Its objectives are, in order of the leverage they carry: to reduce the part count by asking of every part whether it must move relative to its neighbour, whether it must be of a different material, and whether it must be separable for assembly or service — a part that fails all three tests is a candidate for elimination or integration; to reduce the assembly time and therefore the assembly labour cost, by making the remaining parts easy to grasp, orient and insert; to reduce the manufacturing cost of each remaining part, by choosing processes and features the process does well, holding only the tolerances that function requires, and standardising materials and fasteners across the product family; to improve quality and reliability, because eliminating a part eliminates every defect mode that part could have contributed and every tolerance it contributed to the stack; and to shorten time to market, because a design that is producible on the first attempt does not consume the schedule in re-tooling. The unifying idea is that assembly cost is largely determined by decisions taken long before any drawing is released, so DFMA is best understood as cost avoidance rather than cost reduction.

Part B — Three design features consistent with DFMA (6 marks)

(1) Part integration and self-fastening features. Replacing a bracket, two screws, a washer and a nut with a single moulded snap-fit boss on an existing wall removes four parts and their handling, and removes a threaded fastener whose torque must be controlled and audited. The classic Boothroyd illustration is the redesign of an electrical assembly in which nineteen parts and 168 s of assembly time became seven parts and 62 s.

(2) Self-locating and self-aligning geometry. Chamfers and lead-ins on every mating feature, generous first-engagement clearances that tighten only at the end of insertion, and a locating pin plus a slot (never two pins in two round holes) so that the assembly is constrained exactly and not over-constrained. This converts a positioning task requiring vision and dexterity into a task that can be done by feel, or by a robot with a cheap gripper.

(3) Symmetry, or deliberate gross asymmetry. A part that is fully symmetric about its insertion axis cannot be inserted the wrong way and needs no orientation step at all; where symmetry is impossible, the part should be made so obviously asymmetric — a large flat, a colour-coded end, a keyed profile — that the wrong orientation cannot be entered. The worst case is the nearly symmetric part, which is slow for a human to orient and expensive for a feeder to sort. Related features in the same family are: unidirectional assembly, so that the product is built by stacking downward along one axis with gravity holding each part; a base part that acts as the fixture; and no flexible or tangling parts such as loose wires, springs or gaskets that must be handled individually.

Part C — When DFMA should be applied (3 marks)

Given. The Boothroyd–Dewhurst design-for-assembly index $\eta = N_{min}t_{ideal}/t_{asm}$, with $t_{ideal}=3\ \text{s}$; an original assembly of $N=19$ parts taking $t_{asm}=168\ \text{s}$, and its redesign at $N=7$ parts and $t_{asm}=62\ \text{s}$, both with a theoretical minimum part count $N_{min}=5$. Find. The improvement the index records, as evidence for when the method must be applied.

The original design scores $\eta_{1} = (5)(3)/168 = 0.0893$, that is 8.93 per cent, while the redesign scores $\eta_{2} = (5)(3)/62 = 0.2419$, or 24.19 per cent — a $$\boxed{2.71\times \text{ improvement in the DFA index}}$$ achieved by a 63.2 per cent reduction in part count and a 63.1 per cent reduction in assembly time. Nothing in that redesign is available once the tooling has been cut.

That is the argument for timing. DFMA belongs in the concept development and system-level design phases — in Ulrich & Eppinger's terms, phases 1 and 2 of the generic process, before the product architecture is frozen — and it must be revisited in detail design as each part is drawn. Part-count reduction is an architectural decision: whether a housing is one moulding or four sheet-metal panels is settled when the architecture is chosen, and by detail design the only moves left are chamfers and fastener standardisation, which are worth having but are worth perhaps a tenth of what the architectural moves were worth. This is the familiar committed-cost argument: something like seventy to eighty per cent of a product's lifecycle cost is committed by the end of concept design, while only a few per cent has actually been spent. Applying DFMA after design release inverts the economics, because every change then costs re-tooling, re-qualification and schedule. In practice this means a manufacturing engineer and a supplier representative sit in the concept reviews as full members of the team, not as reviewers of a finished drawing package — concurrent engineering is the organisational form that makes DFMA timing possible.

Part D — Manual versus automated assembly (3 marks)

The two cases share the objective of part-count reduction but diverge sharply in what they reward. Manual assembly tolerates variety, judgement and compliance: a human can orient a nearly symmetric part, can feel that a snap has seated, can handle a limp cable, and can build several product variants down one line with only a work-instruction change. What a human does badly is repetition under load and anything demanding sustained precision, so the design considerations are ergonomic and cognitive — part weight and reach within the working envelope, no assembly forces above a comfortable one-handed level, no overhead or wrist-twisting operations, unambiguous part appearance to prevent the wrong-part error, poka-yoke features so that a wrong assembly physically cannot be completed, and clear line of sight and access to the joint. Manual assembly has low fixed cost and high variable cost, so it suits low and medium volume, high variant mix, and early production while the design is still settling.

Automated assembly reverses every one of those trade-offs. Feeders cannot exercise judgement, so parts must be rigid, non-tangling, non-nesting, free of burrs and flash, and either fully symmetric or so asymmetric that a vibratory bowl or vision system can orient them reliably; tolerance stacks must be tightened because a robot has no compliance unless compliance is designed in; assembly should proceed along a single vertical axis onto a base part that doubles as the pallet fixture; fasteners should be reduced to one type or eliminated in favour of snaps and press fits, because each fastener type is a separate station; and in-line inspection must be designed in, since an automated line will faithfully reproduce a defect thousands of times before anybody notices. Automation carries high fixed cost in machinery and tooling and low variable cost, so it needs high, stable volume and a frozen design to pay back; the break-even against manual assembly is a straightforward comparison of annualised equipment cost against labour saved. A common and sensible middle course is hybrid assembly: automate the high-volume, high-repetition, low-variety subassembly and leave final assembly and variant configuration to people.