22-Mec-B5 Product Design and Development · May 2018

Question 2 of 7: The Engineer's Design Process Against the Artist's

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

Notes on this paper

Paper format. National Exams, May 2018, 16-Mec-B5 Product Design and Development — THREE (3) hours, OPEN BOOK, one approved Casio or Sharp calculator permitted. Question 1 is compulsory and carries 40 marks; four of the remaining six questions are chosen, each worth 15 marks, for 100 marks in total. Only the first five questions appearing in the answer book are marked. Most answers are expected in essay form or as tables, figures and charts, and clarity and organisation carry marks in their own right.

Scope of this solution. All seven questions are answered in full, not the five a candidate would attempt, so that the paper works as a study resource. Where the examiner offers a choice of product, one is selected and carried consistently through every part, which is exactly what the question's own guidance note asks for. Numeric illustrations are engineering estimates built from stated, ordinary data; every one of them.

Reference texts for 22-Mec-B5.

Ulrich, K. T. and Eppinger, S. D., Product Design and Development, 6th ed., McGraw-Hill — the framework text for this exam code (development process, concept selection, product architecture, DFM, economics).
Dieter, G. E. and Schmidt, L. C., Engineering Design, 5th ed., McGraw-Hill — problem definition, requirement cascades, material and process selection, design for X.
Pahl, G., Beitz, W., Feldhusen, J. and Grote, K.-H., Engineering Design: A Systematic Approach, 3rd ed., Springer — systematic embodiment design and the function structure.
Boothroyd, G., Dewhurst, P. and Knight, W., Product Design for Manufacture and Assembly, 3rd ed., CRC Press — the DFA index and process-based cost estimating.
Ashby, M. F., Materials Selection in Mechanical Design, 5th ed., Butterworth-Heinemann — material indices, screening and ranking, process-cost models.
Kalpakjian, S. and Schmid, S. R., Manufacturing Engineering and Technology, 7th ed., Pearson — process capability and manufacturing constraints on design.
Cooper, R. G., Winning at New Products, 5th ed., Basic Books — stage-gate governance and gate criteria.
Vaver, D., Intellectual Property Law: Copyright, Patents, Trade-marks, 2nd ed., Irwin Law, together with the Canadian Intellectual Property Office (CIPO) guides — the Canadian statutory frame for Question 4.

Question 2: The Engineer's Design Process Against the Artist's (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 — Comparing the two processes (4 marks)

Both processes begin with a brief, move through divergent exploration, converge on a candidate, and realise it physically, and both are genuinely creative. What separates them is where the goal comes from and who is entitled to declare the work finished.

The engineer designing a functional part receives the goal from outside: a customer need, a system interface, a load case, a regulatory limit. The process is therefore requirement-driven and convergent. Needs are gathered and translated into target specifications with measurable values; concepts are generated against a function structure; one is selected against a datum by an explicit method; the embodiment is analysed, dimensioned and toleranced; and the result is verified against the very specifications that started the work. Physics, cost, manufacturability and the law are external constraints that the engineer may not choose. The deliverable is not the part but a definition of the part complete enough that a stranger in another company can make it correctly. The engineer signs and seals the design and carries professional liability for it, which is why the process is documented at every step.

The artist designing an installation supplies the goal from inside. The process is intent-driven and, in an important sense, divergent all the way to the end: the work is finished when it says what the artist wants it to say, and the artist is the authority on that. Constraints exist — site, budget, safety, the structural capacity of a gallery floor — but most of the important constraints are chosen, adopted because they generate interesting work, and may be abandoned mid-project. The deliverable is the installed work itself and the experience it produces, not a transferable description of it.

Three consequences follow. The engineer's process is reproducible: two competent engineers with the same brief will converge on similar solutions, and that convergence is a feature. The artist's process is deliberately not reproducible; convergence between two artists on the same brief would be a failure. The engineer's process is auditable, because a third party must be able to reconstruct why each decision was made. And the engineer's process has a stopping rule that is external and testable, while the artist's stopping rule is internal and judged.

Part B — Challenges in capturing the design details (4 marks)

For the engineer, the challenge is completeness without over-constraint. The definition must be unambiguous to a manufacturer who cannot ask a question: a 3D master model under configuration control, model-based definition or fully dimensioned drawings, a datum reference frame and geometric tolerances that state function rather than convenience, material and heat-treatment specifications, surface finish, joining and torque specifications, a bill of materials with effectivity dates, and a control plan naming the critical-to-quality characteristics. Two failure modes sit on either side of the target. Under-specify and the supplier makes a legitimate part that does not work. Over-specify and cost rises for nothing: a tolerance tightened out of caution rather than function is paid for on every part ever made.

The hardest single element to capture is the accumulation of tolerances, because each dimension looks reasonable on its own drawing.

Given. Four features stack to close an assembly gap; each is drawn at $\pm 0.10$ mm, each interpreted as a $\pm 3\sigma$ production limit. The gap specification is $\pm 0.25$ mm.

Find. Whether the stack meets the gap specification on a worst-case basis and on a statistical basis, and the resulting process capability.

Compare the worst-case and statistical stacks. The arithmetic worst case adds the tolerances directly, whereas independent random contributors add in quadrature: $$\begin{aligned}T_{\text{WC}}&=\sum_{i=1}^{4}t_i=4(0.10)=\pm 0.40\ \text{mm} \\ T_{\text{RSS}}&=\sqrt{\sum_{i=1}^{4}t_i^{2}}=0.10\sqrt{4}=\pm 0.20\ \text{mm}\end{aligned}$$ The worst-case stack violates the $\pm 0.25$ mm gap; the statistical stack clears it comfortably.
Convert the statistical stack into a capability the plant can be held to. With $\sigma_i = 0.10/3 = 0.0333$ mm per contributor, the stack standard deviation is $\sigma_{\text{stack}}=0.0333\sqrt{4}=0.0667$ mm and $$C_p=\frac{\text{tolerance width}}{6\sigma_{\text{stack}}}=\frac{0.50}{6(0.0667)}=\boxed{1.25}$$ which is acceptable but below the usual 1.33 target, so the honest answer to give manufacturing is that the design passes statistically and has no margin for a process shift.

That distinction is the detail most often lost in transmission: the drawing carries the same numbers in both cases, and only the accompanying stack analysis says which interpretation the design relies on. Communicating design intent, not just design values, is what makes the difference.

For the artist, the challenge is the opposite one: the essential content is tacit and resists specification. Materiality, scale relative to the body, the quality of light at a particular hour, the sequence in which a visitor encounters the work — these are the design, and none of them reduces cleanly to a dimension. Fabrication drawings therefore tend to be produced after the design is settled, to procure and install rather than to design, and much of the intent survives only in the artist's presence during installation. The consequences appear later: re-installation at a second venue, conservation of degrading materials, and authenticity disputes over what may legitimately be replaced are all failures of capture, and the profession's answer — the artist's installation manual, interviews recorded with the artist, detailed photographic documentation — is a deliberate attempt to borrow the engineer's discipline for the parts that can bear it.

Part C — The importance of iteration in both processes (3 marks)

Iteration is essential to both, but it does different work in each, and the difference is measurable in one case and not in the other.

For the engineer, iteration is convergent: each pass measures the remaining gap between the current design and the specification, and removes a fixed fraction of it. That behaviour is geometric.

Given. An initial requirement gap normalised to 1.0, a per-iteration gap-closure ratio $\rho = 0.45$ typical of a well-instrumented analysis-and-test loop, and an acceptance target of 3 per cent residual gap.

Find. The number of iterations required to reach the target.

Model the residual gap and solve for the pass count. After $k$ iterations the residual gap is $g_k=\rho^{k}$, so reaching $g_k \le 0.03$ requires $$k \ge \frac{\ln(0.03)}{\ln(0.45)}=\frac{-3.5066}{-0.7985}=4.39 \Longrightarrow \boxed{k=5\ \text{iterations}}$$

Convergence of the requirement gap. Because the decay is geometric, the first two passes remove four-fifths of the gap and the last two remove almost nothing — which is why front-loading iterations is worth more than adding them at the end.

Two things follow that matter more than the number itself. First, the return on iteration falls away sharply: passes one and two close 79.8 per cent of the gap and passes four and five close only 7.3 per cent between them, so an iteration deferred to the end is worth a fraction of one done early. Second, the cost of an iteration rises by roughly an order of magnitude per phase — a change during concept costs a drawing revision, the same change after tooling costs a die, and after launch it costs a recall. Multiplying a falling benefit by a rising cost gives the rule that engineering practice actually follows: iterate hard and cheaply while the design is still soft, and treat a late iteration as a failure of the early ones.

For the artist, iteration is generative rather than convergent. Successive studies, maquettes and site tests are not reducing a measured error; they are discovering what the work is about, and a late iteration that changes the meaning of the piece is a success rather than a defect. There is no residual gap to plot because there is no fixed target to subtract from. The stopping rule is satisficing — the artist stops when further work stops adding — and the practical constraints on it are the opening date and the budget rather than an acceptance criterion. The shared lesson is that both processes fail in the same way when iteration is compressed away: the engineer ships an unconverged design, and the artist ships an unresolved one.

Part D — Assessing the success of each process (4 marks)

The engineer's process is assessed against evidence, and the assessment is designed at the same time as the product. Four instruments do the work. A requirement verification matrix demands that every requirement be closed by test, analysis, inspection or demonstration, and completeness of that matrix — not the opinion of the design team — is what permits release. Process capability on the critical characteristics shows the design can actually be made: $C_{pk}\ge 1.33$ is the usual gate. Reliability demonstration converts a claim into a sample size, and the arithmetic is worth stating.

Given. A reliability claim of $R = 0.97$ at the warranty milestone, to be demonstrated at 90 per cent confidence by a zero-failure test.

Find. The number of units that must survive with no failures.

Size the zero-failure demonstration. With no failures permitted, the confidence that reliability exceeds $R$ after $n$ trials is $C = 1-R^{n}$, so $$n=\frac{\ln(1-C)}{\ln R}=\frac{\ln(0.10)}{\ln(0.97)}=\frac{-2.3026}{-0.03046}=75.6 \Longrightarrow \boxed{n=76\ \text{units}}$$ Raising the claim to $R=0.99$ at the same confidence would take 230 units, which is exactly why accelerated and physics-of-failure methods exist: brute-force sampling becomes unaffordable long before the reliability targets become ambitious.

The fourth instrument is field performance: warranty claims per thousand units at a fixed month-in-service, returns, and cost of poor quality. It is the only one of the four that measures the outcome rather than the process, and it arrives too late to change the design, which is why the first three exist.

The artist's process cannot be assessed with any of these, and importing them produces nonsense. Its instruments are the judgement of informed viewers, critical review, curatorial and peer recognition, the depth and duration of audience engagement, acquisition into a collection, and — the only genuinely decisive one — whether the work still holds attention years later. These are ordinal, contested and lagging, and the honest position is that they are the right measures for the thing being measured. The instructive comparison for an engineer is that the two sets are not different in rigour but different in who is entitled to judge: the engineer's success is decided by an external standard the engineer did not write, and the artist's by a discourse the artist is a participant in.

Topic. Requirement-driven convergent design against intent-driven generative design; capture of design definition; iteration economics; verification and validation.

Key relations. Tolerance stacks, $T_{\text{WC}}=\sum t_i$ and $T_{\text{RSS}}=\sqrt{\sum t_i^{2}}$; capability, $C_p = T/6\sigma$ and $C_{pk}=\min[(USL-\mu),(\mu-LSL)]/3\sigma$; geometric iteration, $g_k=\rho^{k}$ with $k=\ln g_{\text{target}}/\ln\rho$; zero-failure demonstration, $n = \ln(1-C)/\ln R$.

Why this works. The root-sum-square stack is legitimate only because independent random variables add in variance, not in range, so four independent $\pm 0.10$ mm contributors give a $\pm 0.20$ mm stack rather than $\pm 0.40$ mm; the price of that halving is that it is a probabilistic statement, so it must be backed by a control plan that keeps the contributors independent and centred. The geometric iteration model follows from each pass attacking the remaining gap rather than the original one, which is what a feedback loop with a fixed gain does. The zero-failure sample size follows from the probability that $n$ independent units all survive being $R^{n}$; setting that equal to the residual risk $1-C$ inverts to the logarithmic form.

Common pitfall. Using the statistical stack while managing the plant as though the worst-case stack applied, or the reverse. RSS is not a licence to loosen tolerances; it is a commitment to keep the contributing processes centred and capable, and it collapses as soon as one contributor shifts its mean. The parallel pitfall in Part D is judging a creative process by engineering instruments, or an engineering design by the enthusiasm of its designers.

Source. Ulrich and Eppinger, Product Design and Development, 6th ed., Ch. 3–7 (needs, specifications, concept selection) and Ch. 13; Kalpakjian and Schmid, Manufacturing Engineering and Technology, 7th ed., Ch. 36 (process capability); Dieter and Schmidt, Engineering Design, 5th ed., Ch. 1 and Ch. 14 (tolerance analysis and reliability).

← Question 1 Next: Question 3 →