22-Mec-B5 Product Design and Development · December 2017
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Paper format. National Exams, December 2017. Three (3) hours. OPEN BOOK; an approved Casio or Sharp calculator is permitted. Question 1 is compulsory and carries 40 marks; four (4) of the remaining six (6) questions are chosen, each worth 15 marks, for 100 marks attempted out of 130 printed. Only the first five questions appearing in the answer book are marked. The marking scheme is printed on page 4 of the paper and is reproduced against each question below. Most answers are expected in essay form, supported by tables, figures and charts.
How to use this document. Every one of the seven printed questions is answered in full, not just the five a candidate would attempt, so that the set works as a study resource. This is a descriptive design-methodology paper: the marks are for method, structure and judgement rather than for arithmetic. Where a number genuinely sharpens an argument — a DFA index, a process break-even, a capability index, a material index — it is computed explicitly and framed with Given. and Find. so the reasoning can be checked. All monetary figures are Canadian dollars.
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.
Both are engineers solving a constrained problem, but they are solving problems with different structures, and the differences are systematic rather than a matter of personality.
The product designer works forward from an incompletely stated human need toward an artefact that does not yet exist. The problem is underdetermined: many artefacts would satisfy the need, and the requirements are necessary but never sufficient, so the designer must widen the solution space before narrowing it and must supply judgement the requirements do not contain. The dominant uncertainty is about the requirement itself — whether the need has been correctly understood — and the dominant risk is building the wrong thing well. The unit of work is the concept; the time constant is the product programme; and success is discovered late, in the market.
The manufacturing process engineer works backward from a fully specified artefact toward a repeatable transformation that will produce it at rate, at cost and within tolerance. The problem is overdetermined: the part geometry, material and tolerances are fixed inputs, and the engineer is searching a much narrower space of process parameters that simultaneously satisfy all of them. The dominant uncertainty is about variation — not whether the process can make one good part, but whether it will make a million — and the dominant risk is a capable-looking process that drifts. The unit of work is the process step; the time constant is the shift; and success is measured continuously and immediately.
Three contrasts follow from that structure and are worth stating explicitly. First, divergence against convergence: the designer is professionally obliged to generate alternatives, whereas alternatives are a cost to the process engineer, who wants the smallest feasible parameter window. Second, the single case against the distribution: the designer reasons about the nominal design and checks the extremes, while the process engineer reasons about the distribution from the start — a mean is nearly useless to them without a standard deviation. Third, irreversibility of commitment: the designer’s early decisions are cheap and their late ones expensive, while the process engineer inherits all of those decisions as fixed and can only optimise within them, which is exactly why DFMA insists the two be in the room together during embodiment rather than at handover.
The similarity is worth a sentence too, because the question says compare as well as contrast: both iterate against evidence, both must trade competing objectives without a single figure of merit, both are constrained by the same physics and the same statutory obligations, and both are professionally accountable for a judgement that the requirements alone do not determine.
| Product designer | Manufacturing process engineer | |
|---|---|---|
| Primary artefact | 3-D CAD model with a defined design intent and a parametric feature tree; the model is the master | Process flow diagram and station layout tied to a routing; the routing is the master |
| Requirements | Requirements matrix or house of quality, tracing customer needs to measurable metrics with target and marginal values | Control plan and process FMEA, tracing each product characteristic to the process parameter that produces it and the control that holds it |
| Geometry and variation | GD&T on the drawing, datum reference frames, tolerance stack-up analysis, worst-case and statistical | Capability studies and control charts on the same characteristics: histograms, X-bar and R charts, Cp and Cpk reports |
| Decisions | Pugh screening matrix, then a weighted decision matrix with the weights fixed before the ratings, plus a sensitivity check | Designed experiments and response-surface plots; main-effect and interaction plots that locate the robust operating window |
| Performance evidence | Simulation results (FEA, CFD, tolerance Monte Carlo), prototype test reports, validation against the requirement matrix | Run-at-rate data, first-article inspection, gauge R&R studies, scrap and rework Pareto charts, OEE dashboards |
| Communication style | Visual and narrative: renderings, exploded views, animations and physical models, aimed at a mixed audience | Numeric and procedural: work instructions, one-point lessons, standard work sheets and andon boards, aimed at an operator on shift |
The single most important difference in the table is the last-but-one row. The designer’s evidence is largely predictive — simulations and prototypes standing in for a product that does not yet exist in quantity — whereas the process engineer’s evidence is observational and arrives continuously. Handover failures are usually failures to convert the first kind into the second: a critical characteristic that was verified once on three prototypes and never became a controlled characteristic with a capability requirement and a measurement system behind it.
The designer assesses success against the need and against the business case, on a timescale of months to years. The concrete measures are: requirement verification coverage, that is, the proportion of requirements with a closed verification record — on the hair-dryer programme of Question 1 this stood at 61 of 68 requirements, or 89.7 per cent, at the design-freeze gate, and the seven open items are the honest statement of programme risk; achieved unit cost against target; customer acceptance measures from user trials and, after launch, from returns, warranty claims and satisfaction surveys; time to market against plan; and the eventual commercial measures of share and margin. The characteristic difficulty is that the decisive feedback arrives long after the decisions that caused it, which is why the designer leans on leading indicators such as prototype test results and user-trial outcomes.
The process engineer assesses success against rate, cost and variation, continuously. The standard composite measure is overall equipment effectiveness,
$$\text{OEE}=A\times P\times Q$$
the product of availability, performance and quality rates. On the redesigned hair-dryer line, with $A=0.88$, $P=0.94$ and $Q=0.985$, $$\text{OEE}=0.88\times 0.94\times 0.985=0.815$$ so $\boxed{\text{OEE}=81.5\ \text{per cent}}$, which is a good result — 85 per cent is the conventional world-class benchmark — and, more usefully, the decomposition says immediately that the 12 per cent availability loss is the largest of the three and should be attacked first.
Alongside OEE sit rolled throughput yield, capability indices on every critical characteristic, and cost per unit against the should-cost model. Rolled throughput yield is the one that repays the redesign of Question 1 twice over: at a 98 per cent first-pass yield per station, a three-station line returns $0.98^{3}=0.941$ while the original six-station line returned $0.98^{6}=0.886$. Halving the number of stations therefore lifted the end-to-end yield by 5.5 percentage points on its own, before any station was improved — a benefit the DFA index never claimed and the assembly-time calculation never showed.
The final point to make is that the two definitions of success can conflict, and managing the conflict is an engineering responsibility rather than a management one. A process engineer optimising OEE will resist product variety, engineering changes and short runs; a designer optimising customer value will want all three. The resolution is not to pick a winner but to agree the measures jointly at the start of the programme, so that the designer carries a manufacturability target such as the DFA index and the assembly-time budget, and the process engineer carries a flexibility target such as changeover time. Shared measures are what make concurrent engineering work in practice rather than only on the organisation chart.
| Result | Value |
|---|---|
| Requirement verification coverage at design freeze | 61 of 68, i.e. 89.7 per cent |
| Overall equipment effectiveness | 0.88 × 0.94 × 0.985 = 0.815 |
| Largest OEE loss | Availability, 12 per cent |
| Rolled throughput yield, 3 stations at 98 per cent | 0.941 |
| Rolled throughput yield, 6 stations at 98 per cent | 0.886 |
| Yield gained from halving the line | 5.5 percentage points |
