Question 1 of 7: The Design Process Applied to a Universal-Design Product

Question 1: The Design Process Applied to a Universal-Design Product (40 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.

Product selected: (i) the bottle opener, for a user whose hand grip strength is limited by arthritis. It is chosen deliberately: it is the smallest and cheapest of the three products, which means the design process must be shown in full without hiding behind system complexity, and it is the one whose failure mode is purely a mismatch between the force the product demands and the force the user can supply. That makes it an almost ideal vehicle for universal design, because the same change that makes the product usable by a person with rheumatoid arthritis — less force, a larger grip, tolerance of one-handed operation — makes it better for everybody, which is precisely Ronald Mace's original argument for the term. The design direction adopted and carried through parts A to F is: a mainstream kitchen product, sold through ordinary retail at an ordinary price, that happens to demand roughly one third of the hand force a conventional opener demands and that carries no visual signal of being a medical aid.

Part A — Three ways of establishing design specifications (6 marks)

A specification is not a wish list; it is a set of measurable metrics, each with a marginal value that must be met and an ideal value that the team is trying to reach. Three genuinely different routes to that set are available, and a real programme normally uses all three with one of them dominant.

(1) User-driven specification — voice of the customer translated into metrics. This is the Ulrich & Eppinger route: gather raw data from users by interview, by observation in the place the product is actually used, and by focus group; interpret the raw statements as need statements in the user's own terms; organise the needs into a hierarchy and weight them; then, for every need, ask what physical quantity would tell you whether the need is met, and set marginal and ideal values on that quantity. The strength of the method is that nothing enters the specification that cannot be traced to a stated need; its weakness is that users describe symptoms and not solutions, and a poorly run study collects preferences for existing products rather than underlying needs.

(2) Benchmark-driven specification — competitive benchmarking and reverse engineering. Buy every opener on the Canadian market, including the assistive-device catalogue models, measure them on a common test rig, and tabulate the results as a competitive benchmarking chart against the needs. The specification then becomes "at least as good as the best competitor on every metric, and decisively better on the two or three metrics the user weighting says matter most." The strength is objectivity and speed — the numbers are real, measured, and defensible to management; the weakness is that it anchors the programme to the existing solution space and will never by itself produce a specification for a feature nobody currently sells.

(3) Standards- and requirements-driven specification. Derive the specification top-down from regulation, standards and published anthropometric and biomechanical data: the Canada Consumer Product Safety Act for hazard requirements, CSA B651 and CSA/ISO 21542 for accessible-design force and clearance limits, the Accessible Canada Act and provincial statutes for the policy frame, and published grip- and pinch-strength distributions for rheumatoid and osteoarthritic populations for the force targets. In this route the specification is a compliance matrix, and the design must be shown by test to satisfy it. The strength is that it produces hard, non-negotiable numbers and an audit trail; the weakness is that it guarantees only adequacy, never desirability, and standards lag practice by years.

A fourth technique, quality function deployment (the House of Quality), is best understood not as a fourth source of data but as the matrix that reconciles the three above by mapping weighted needs onto engineering metrics and exposing where two metrics conflict.

Part B — Nature of the information collected (9 marks)

Method (1), user-driven specification, is carried forward, because the whole premise of the question is a mismatch between product demand and user capability and only the user study measures that capability directly. The study would be a contextual inquiry: roughly twenty-five participants with physician-confirmed rheumatoid or osteoarthritis of the hand, plus a control group of ten users without hand impairment, observed opening bottles in their own kitchens, followed by a semi-structured interview. Four families of information come out of it.

Biomechanical capability data. Maximum voluntary power-grip force by hand and by grip span, measured on a Jamar-type dynamometer; lateral (key) pinch and tip-pinch force; wrist pronation and supination torque; wrist and finger range of motion; and, critically, the pain-limited rather than the strength-limited value of each, because an arthritic user stops at the onset of pain and not at the onset of muscular failure. This is the data that fixes the force targets: a typical adult power grip is of the order of 300 N, while the target population commonly presents around 70 N, that is roughly one quarter of it, and sustains far less than that without pain.

Task and context data. What is actually being opened — the 26 mm crown cap of a beer or soda bottle, the twist-off crown, the 28 mm plastic screw closure of a juice or water bottle, ring-pull and vacuum-lug lids; where the task happens; whether the bottle can be braced against a counter or must be held in the air; whether the second hand is available at all, since many users have bilateral involvement; how many bottles are opened at a sitting; and what the user currently does when the opener fails, which is usually to ask somebody else, and which the interviews reveal to be the single most resented outcome.

Product-interaction and failure data. Where the current product slips, where it hurts, where it is dropped, how it is stored and retrieved, how it is cleaned, and whether it is left on the counter or hidden in a drawer. Observation matters more than interview here, because users habitually under-report workarounds they have stopped noticing.

Affective and commercial data. Whether the user would be willing to have the product visible in a kitchen shared with guests — the stigma question, which repeatedly determines whether assistive products are actually used; what the user expects to pay and where the user expects to buy it; and how the product is talked about, which supplies the language for the eventual marketing claim.

Part C — Target specifications (9 marks)

The needs from part B are now converted into measurable metrics. Two of them — the crown-cap and twist-off force targets — are set by a short piece of statics, which is worth doing explicitly because it is what converts a need ("must not hurt to use") into a dimension (a handle length).

Given. A standard 26 mm crown cap requires a prying force at the cap skirt of about $F_{cap}=100\ \text{N}$, applied at a lever arm equal to the cap radius, $r=13\ \text{mm}$, measured from the fulcrum that the opener rests on. A twist-off crown or plastic screw closure requires a removal torque of about $T=1.5\ \text{N}\cdot\text{m}$. The target user can apply roughly $70\ \text{N}$ of power grip against $300\ \text{N}$ for an unimpaired adult, and the design target adopted from the interviews is that no operation shall demand more than $F_{allow}=30\ \text{N}$ of one-handed effort.

Find. The handle length and side-lever arm that bring both opening operations below the 30 N target, and the resulting set of target specifications.

Approach. Treat the opener as a simple lever in each mode, equate the moments about the fulcrum to size the arm, then read the remaining metrics off the four information families of part B.

1. Size the crown-cap lever from moment equilibrium. Taking moments about the fulcrum that bears on the top of the cap, $$F_{user}\,L = F_{cap}\,r$$ so the shortest handle that still meets the target is $$L_{min}=\frac{F_{cap}\,r}{F_{allow}}=\frac{(100)(0.013)}{30}=0.0433\ \text{m}=43.3\ \text{mm}.$$ Anything longer than about 43 mm satisfies the target, so the length is free to be set by the hand rather than by the physics.
2. Choose the handle length from the hand, not from the minimum. A handle that fills the palm of a 95th-percentile male hand and still fits a 5th-percentile female hand is about 120 mm long, so take $L=120\ \text{mm}$. The mechanical advantage is then $$MA=\frac{L}{r}=\frac{120}{13}=9.23,$$ and the force the user must actually supply falls to $$F_{user}=\frac{F_{cap}\,r}{L}=\frac{(100)(0.013)}{0.120}=\boxed{10.8\ \text{N}}$$ which is a factor of 2.77 below the 30 N target and roughly one sixth of the user's available grip. That margin is deliberate: it is what allows the product to be used repeatedly, and on a bad day, rather than once under maximum effort.
3. Size the twist-off jaw for torque. The twist-off mode is a pure torque problem. With the jaw gripping the cap and the handle acting as a side lever of arm $a$, $$F_{user}=\frac{T}{a}=\frac{1.5}{0.090}=\boxed{16.7\ \text{N}}$$ for $a=90\ \text{mm}$, again inside the 30 N target. The twist-off mode is therefore the governing case, and it is the one to test first on a prototype.
4. Fix the grip geometry from the hand data. Grip force peaks at a handle diameter near 35 mm for the general adult population and the optimum shifts larger for arthritic hands, because a larger diameter reduces the finger flexion angle and therefore the joint loading. Specify a 35 mm nominal diameter with an elliptical rather than circular section, so that the handle cannot roll in the hand and so that the user does not need to squeeze to hold it.

The full target specification list follows. Each metric carries a marginal value, which must be met for the product to ship, and an ideal value, which is the target the team is aiming at — the two-column form is what makes a specification testable rather than aspirational.

Two material and manufacturing consequences follow directly from that table and are carried into part D. Metric 1 combined with metric 7 rules out a solid metal body, so the structure is a glass-filled polyamide or a die-cast zinc-aluminium core; metrics 9 and 10 force a thermoplastic elastomer over-mould of Shore A 60 to 70, food-contact grade, bonded to the core in a two-shot mould rather than glued on; and metric 11 at 50 000 units per year puts the programme squarely in injection-moulding territory rather than machining.

Part D — Three design concepts (6 marks)

The three concepts below are deliberately different architectures, not three stylings of the same architecture: a hand-held lever, a mounted machine, and a mechanism that adapts to the cap. Each sketch carries the rough dimensions that come out of part C.

Part E — Methodology for comparing the alternatives (4 marks)

The methodology adopted is the two-stage concept selection of Ulrich & Eppinger: concept screening followed by concept scoring, with an explicit sensitivity check on the result. It proceeds in six steps.

1. Fix the selection criteria from the specification, before any concept is looked at. The criteria are drawn from the target specification table of part C so that the comparison is against the stated needs and not against the designers' preferences. Six criteria are used here: user force and torque demand, closure-type coverage, portability, unit manufacturing cost, cleanability and durability, and non-stigmatising appearance.
2. Choose a datum. The reference concept is the conventional flat steel crown-cap opener — the "church key" — because it is what the target users already own, so every judgement is relative to a product the panel has actually handled.
3. Screen with a Pugh matrix. Each concept is rated only as better (+), same (0) or worse (−) than the datum on each criterion, and the net score is used to eliminate the clearly dominated concepts and to expose the specific weaknesses that should be designed out before the next stage. Screening is coarse on purpose: it is fast, it is robust to bad data, and it is the stage at which concepts should be combined and improved rather than merely ranked.
4. Fix the weights, still before any rating. Weights come from the needs hierarchy of part B and must be agreed and recorded before the scoring ratings are entered; otherwise the weights are quietly tuned until the favourite wins, which is the commonest way this method is abused.
5. Score on a 1–5 scale and compute the weighted total. For concept $j$ the total is $$S_j=\sum_{i=1}^{n} w_i\,r_{ij},\qquad \sum_{i=1}^{n} w_i = 1$$ where $w_i$ is the weight of criterion $i$ and $r_{ij}$ is the rating of concept $j$ on it. Ratings are given against the datum, which is defined to score 3 on every criterion.
6. Test the result for sensitivity, then decide. Re-run the totals under a defensibly different weighting — here, an assistive-device sales channel rather than mass retail — and report whether the ranking survives. A ranking that flips under a plausible change of weights is not a decision; it is a signal that the two concepts are equivalent on the evidence available and that a prototype test is needed to separate them.

Part F — Ranking and selection (6 marks)

Given. The three concepts of part D, the six criteria of part E, the mass-retail weighting $w=(0.30,\ 0.15,\ 0.15,\ 0.20,\ 0.10,\ 0.10)$, and the ratings tabulated below. Find. The screened set, the weighted ranking, and the selected concept, with a statement of how robust the selection is.

Screening first. Against the church-key datum:

All three concepts beat the datum on the two criteria that matter most and none is dominated outright, so all three continue; concept B is carried forward despite its negative net score because its single strength — it demands no grip force whatever — is the strongest response in the set to the primary need, and screening exists to preserve exactly that kind of concept. Scoring follows.

Working the winning column explicitly, $S_A = (0.30)(4)+(0.15)(4)+(0.15)(5)+ (0.20)(5)+(0.10)(5)+(0.10)(4) = 1.20+0.60+0.75+1.00+0.50+0.40$, so $$\boxed{S_A = 4.45,\quad S_C = 3.65,\quad S_B = 3.30}$$ and concept A leads concept C by 0.80 of a rating point.

Now the sensitivity check. Suppose the product were sold through an assistive-device channel rather than mainstream retail: closure coverage would matter as much as force, and unit cost would matter far less, giving $w'=(0.30,\ 0.30,\ 0.15,\ 0.05,\ 0.10,\ 0.10)$. Recomputing, $S'_A = 4.30$, $S'_C = 4.10$ and $S'_B = 3.30$. The ranking is unchanged, but the margin between A and C has collapsed from 0.80 to 0.20, and if the cost weight were driven all the way to zero the two concepts would tie exactly at 4.25. The honest conclusion is therefore that A wins the mass-retail case decisively and the clinical case only narrowly.