22-Mec-B5 Product Design and Development · December 2016

Question 2 of 7: Design for Manufacture and Assembly

Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)

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.

A — Functional definition (3 marks)

Design for Manufacture and Assembly is the practice of choosing the geometry, the material, the joining method and above all the part count of a product so that the total cost of making and putting it together is minimised, while the product still delivers its required function. Stated functionally rather than as a slogan: DFMA is a design activity, carried out by the design team with manufacturing engineering in the room, that treats manufacturability and assemblability as design requirements of equal standing with performance, and that measures the result. Boothroyd, Dewhurst and Knight split it into two halves that must both be done. Design for assembly asks whether each part needs to exist at all, and applies the three-question test: a part is separate only if it moves relative to its neighbour, must be of a different material for a fundamental reason, or must be removable for assembly or service. Design for manufacture then asks whether the surviving parts are shaped so that the chosen process can make them easily — adequate draft, uniform wall, generous radii, tolerances no tighter than the function requires, features accessible to the tool. The measurable output is the assembly efficiency index

which turns a qualitative argument into a number that can be tracked through design reviews.

B — A common strategy consistent with DFMA objectives (6 marks)

The most widely used strategy is part-count reduction by functional integration, combined with self-locating and self-fastening features that eliminate separate fasteners. A sub-assembly of a stamped bracket, two spacers, four screws, four washers and a nut plate is replaced by a single moulded or die-cast part whose bosses locate the mating component and whose moulded snap fingers retain it. Nothing is lost functionally; what disappears is eleven part numbers, eleven purchase orders, eleven line-side bins, eleven chances to fit the wrong part, and roughly nine seconds of screwdriving.

The strategy is powerful because fasteners are the single worst value in most assemblies. A screw costs a few cents but consumes six to ten seconds of handling, alignment and driving, needs torque control, and is the commonest source of both line rework and field warranty. Related moves in the same family are designing parts to be symmetric so they cannot be inserted the wrong way (or grossly asymmetric so the error is obvious), assembling from a single direction along a vertical axis so gravity assists and the part does not need to be reoriented, providing lead-in chamfers so insertion is self-aligning, and using a common platform of modules across a product family so volume concentrates on fewer part numbers.

The roof rack of Question 1 is a worked instance of exactly this strategy: integrating the clamp bodies into the tray extrusion and moulding the buckle into the strap end took the design from 38 parts and 570 s to 22 parts and 330 s, lifting the DFA index from 0.063 to 0.109 and taking $10.00 per unit out of the factory cost. That is the shape of the result a well-executed DFMA study returns: a large, verifiable cost reduction obtained without changing what the product does.

C — When to apply DFMA, and when it is too late (2 marks)

DFMA belongs in concept development and early system-level design, before the architecture is chosen and while the part count is still an open variable. This is where roughly 70 % to 80 % of the eventual manufacturing cost is committed, even though only a small fraction has yet been spent — the classic committed-versus-incurred cost divergence. A DFMA review at concept selection can still delete a whole sub-assembly; the same review at detail design can only shave a boss.

It is too late once production tooling has been cut. From that point the die, mould or fixture is a sunk asset, the change requires new tooling plus requalification, supplier agreements and volume commitments are in place, certification and regulatory testing must be repeated, and service parts for the old configuration must be carried in parallel. Practically, the last genuinely cheap opportunity is the design freeze that releases tooling; after that a DFMA finding becomes a business case rather than a design decision, and most such findings are deferred to the next model.

D — Manual against automated assembly (4 marks)

The two routes reward opposite design choices, so the decision must be taken before the parts are detailed rather than after. Manual assembly tolerates variety, compliance and judgement: a person can flex a harness past an obstruction, feel a snap engage, cope with a part presented at any orientation and switch between variants without changeover. What a person is poor at is sustained precision, high force, and anything requiring more than about two seconds of alignment; so manual designs should avoid parts that tangle or nest, avoid tight-clearance insertions, avoid tools that must be picked up and put down, and above all avoid awkward postures and the need to hold a part while fastening it.

Automated assembly reverses every one of those. A feeder needs parts that are rigid, non-tangling, non-nesting and orientable by a simple track feature; a robot needs a single vertical insertion axis, generous lead-in chamfers, a self-locating datum scheme and consistent presentation; and the cell as a whole needs tight process capability, because a machine has no judgement and an out-of-tolerance part becomes a jam rather than a fitted part. Automation rewards low variety and punishes changeover, so it suits a mature, high-volume, few-variant product.

Beyond design detail there is a straightforward economic test, and it is the one that usually decides.

Given. A manual cell for the Question 1 rack needs $25,000 of fixtures and takes 330 s per unit at a burdened rate of $42/h. An automated cell — robot, feeders, vision and guarding — costs $480,000 and runs at a marginal cost of $0.95 per unit including its share of operator attendance. Planned volume is 20,000 units per year over an eight-year model life.

Find. The break-even volume between the two routes, and hence which one this product should use.

1. Express the manual variable cost. Assembly labour is time at rate, $u_m=t\,r/3600=330\times 42/3600=3.85$ dollars per unit, against $u_a=0.95$ for the automated cell — the machine is four times cheaper per unit once it is running.
2. Equate the two total-cost lines. Setting $T_m+n\,u_m=T_a+n\,u_a$ and solving for the crossover volume,$$n^{*}=\frac{T_a-T_m}{u_m-u_a}=\frac{480{,}000-25{,}000}{3.85-0.95}=\frac{455{,}000}{2.90}$$which evaluates to $$\boxed{n^{*}=156{,}897\ \text{units}}$$or 7.8 years of production at 20,000 units per year.
3. Interpret against the programme. Over the whole eight-year life the plant will build 160,000 units, so the manual route costs $641,000 and the automated route $632,000. The automated cell wins by about 1.4 % of programme cost — which is well inside the uncertainty on volume, on the capital quotation and on the eight-year life itself. The correct engineering answer is therefore to assemble this product manually, design it for manual assembly, and revisit automation only if volume rises materially.