Question 1 of 7: Improving the functionality of a consumer product (40 marks)

Question 1 — Improving the functionality of a consumer 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.

Given. The product selected is the household blender, taken as a conventional 1.5 L jar machine driven by a single-speed universal motor, with the assumed parameters listed in the table below; the design objective set by the question is to improve ease of use and range of applications.

Find. A defended design direction for the blender, carried consistently through parts A to E: how customer input is gathered, three functional improvements traceable to that input, the societal consequences and how to improve them, the translation of those improvements into engineering specifications, and the priority rule used when the specifications cannot all be met.

Approach. Follow the Ulrich and Eppinger front end in order — gather needs, turn needs into measurable target specifications, generate and screen concepts, then prioritise by weighted trade-off — and anchor the whole argument on three quantities that can actually be measured, so that each design claim is falsifiable rather than rhetorical.

A. Three general ways to obtain end-customer input, and one developed into an activity

Customer input at the start of a design programme is not one activity but a family of them, and they differ in what they can and cannot see. The three that cover the useful ground are the following.

(i) Direct interaction in the use environment — contextual inquiry, in-home observation and shop-along visits, in which the engineer watches the product being used where it is actually used. This is the only method that reliably surfaces latent needs, the ones the customer does not report because the workaround has become invisible to them: the jar left unwashed because the base cannot be immersed, the towel folded over the lid because it leaks, the appliance kept on the counter because lifting it is awkward.

(ii) Structured elicitation from a sample — interviews, focus groups and questionnaires. Interviews and focus groups are exploratory and generate the vocabulary of the need; questionnaires are confirmatory and generate the weighting of needs across a defined population. Ulrich and Eppinger's rule of thumb is that roughly thirty well-run interviews with a single market segment surface about ninety per cent of that segment's needs, after which new interviews mostly repeat what is already on the wall.

(iii) Mining evidence the customer has already produced — warranty and returns data, service-call transcripts, retailer reviews, and lead-user communities. This is the cheapest input available and the only one that is longitudinal: it reports what failed after two years of ownership, which no laboratory session can show. Its weakness is that it is a survivor's account, biased toward the articulate and the aggrieved.

Taking method (i) forward, a contextual-inquiry programme becomes a useful engineering activity only when it is instrumented and sampled deliberately rather than run as a set of friendly visits. It is developed as follows. Recruit fifteen households stratified deliberately across the variable that matters for this product: hand strength and dexterity. Five households with at least one member over 70 or reporting arthritis, five with young children, five general. Each session runs ninety minutes in the participant's own kitchen and follows a fixed task list — make a smoothie with frozen fruit, make a hot soup, grind coffee, clean everything — with the participant narrating aloud and the session video-recorded from two angles. Two measurements are taken on the spot rather than left to impression: grip and pinch strength with a hand dynamometer, and the torque required to release that household's own blender collar with a digital torque gauge. The raw observations are then written as need statements in the customer's language, affinity-diagrammed into a hierarchy, and finally importance-rated by a confirmatory survey. Sizing the survey is arithmetic, not judgement: to estimate the proportion of owners who report a given need to within ±8 % at 95 % confidence,

$$n = \frac{z^{2}\,p(1-p)}{E^{2}} = \frac{1.96^{2}\times 0.25}{0.08^{2}} = 150.1 \;\Rightarrow\; \boxed{n = 151 \text{ respondents}}$$

using the conservative $p = 0.5$. The deliverable of the activity is therefore a ranked, weighted needs list with a stated confidence interval — an input a specification can be written from — and not a set of anecdotes.

B. Three ways customer input improves the functionality of the blender

The three needs that the programme in part A would be expected to surface, and the design responses that follow from them, are set out below. Each is stated as a measured quantity, because a functional improvement that cannot be measured cannot be verified later.

(1) Range of applications: give the machine torque, not just speed. The recurring complaint about a domestic blender is that it liquefies well and does everything else badly — it stalls on nut butter, climbs out of stiff dough, and merely rattles ice. That is not a power problem, it is a speed problem, because a universal motor delivers its rated power at very high speed and therefore very little torque. With the assumed data the shaft power is

$$P_{\text{sh}} = \eta P_{\text{in}} = 0.62 \times 1200 = 744\ \text{W}$$

and the available torque follows directly from $T = P_{\text{sh}}/\omega$ with $\omega = 2\pi n/60$. At the direct-drive speed,

$$\omega_{\text{hi}} = \frac{2\pi \times 18\,000}{60} = 1885\ \text{rad/s}, \qquad T_{\text{hi}} = \frac{744}{1885} = 0.395\ \text{N}\cdot\text{m}$$

whereas a 4.5:1 reduction to a second, low-speed setting gives, from the same motor and the same 744 W,

$$\omega_{\text{lo}} = \frac{2\pi \times 4\,000}{60} = 418.9\ \text{rad/s}, \qquad \boxed{T_{\text{lo}} = \frac{744}{418.9} = 1.776\ \text{N}\cdot\text{m}}$$

a 4.5-fold increase in torque for no increase in installed power, cost of copper, or breaker load. This is the single change that most widens the range of applications, and it is invisible to a customer, who will ask for “more power” when what they need is a different point on the same power hyperbola.

(2) Ease of use: bring the interlock collar under the accessible operating-force ceiling. The dynamometer measurement in part A is the one that turns a soft complaint into a specification. The jar is locked to the base by twisting a collar against a gasket, and the user applies that torque as a couple with both hands on the jar body. For a couple of two equal forces $F$ separated by the grip diameter $D$, $T_{c} = F D$, so

$$F = \frac{T_{c}}{D} = \frac{2.6}{0.110} = 23.6\ \text{N}$$

against the CSA/ANSI B651 ceiling of 22.2 N for an operable control. The baseline therefore fails the accessibility criterion outright, which is precisely what the arthritic-household stratum in the sample would have reported as “I can't get the jar off.” Two independent levers close the gap. Moulding two grip lugs onto the collar enlarges the effective grip circle to 150 mm:

$$F = \frac{2.6}{0.150} = 17.3\ \text{N} \;<\; 22.2\ \text{N}$$

and specifying a lubricated silicone lip seal in place of the rubber compression gasket lowers the breakaway torque to 2.0 N·m, giving

$$\boxed{F = \frac{2.0}{0.150} = 13.3\ \text{N}}$$

which retains a 40 % margin against the ceiling for gasket ageing and for the swelling that follows repeated dishwasher cycles. Equivalently, the specification can be written as a torque limit directly: at $D = 150$ mm the collar torque must not exceed $22.2 \times 0.150 = 3.33$ N·m, and 2.0 N·m sits comfortably inside it.

(3) Range of applications again: let the machine finish a hot soup in one jar. The observation that participants blend soup and then reheat it on the stove points to a capability the machine almost has already. Viscous dissipation at high tip speed is a heater; the only question is whether it is fast enough to be useful. Assuming 85 % of the shaft power reaches the fluid, $P_{f} = 0.85 \times 744 = 632.4$ W, and heating 1.5 kg of soup from 20 °C to a 75 °C serving temperature needs

$$Q = mc\,\Delta T = 1.5 \times 4186 \times 55 = 345\,345\ \text{J}, \qquad \boxed{t = \frac{345\,345}{632.4} = 546\ \text{s} = 9.1\ \text{min}}$$

Nine minutes is a usable claim, and it is a claim that can be specified, measured and advertised. It also forces two safety requirements that the baseline does not have: a vented lid cap, because a sealed jar of 75 °C liquid will lift its own lid, and a jar material with a service temperature above 100 °C.

C. Societal impact of the change, and three improvements to that impact

The redesign has effects beyond the buyer. Positively, a blender that is operable at 13 N and that crushes, kneads and heats replaces two or three other appliances and admits a much wider range of users — older people, people with reduced grip, people cooking one-handed — which is a real gain in independent living. Negatively, a machine with a gearbox, a second speed and a heating claim has more parts, more failure modes and a stronger incentive to be replaced rather than repaired, and every additional appliance function sold as “new” accelerates the disposal of a working old one. The three improvements that would most improve the balance are the following.

(i) Design for repair and for a long service life. The dominant environmental fact about a low-duty appliance is that it is not dominated by its use phase, which is the opposite of the usual intuition. A screening life-cycle inventory makes this concrete. Taking a glass jar at 1.35 kg (15 MJ/kg), an ABS housing at 1.00 kg (95 MJ/kg) and a motor and drive assembly at 1.25 kg (35 MJ/kg average for the steel-and-copper mix):

$$E_{\text{emb}} = 1.35(15) + 1.00(95) + 1.25(35) = 20.3 + 95 + 43.8 = 159\ \text{MJ}$$

Against this, three minutes of daily use at 1.2 kW is $0.06$ kWh/day, so over a five-year life

$$E_{\text{use}} = 0.06 \times 365 \times 5 \times 3.6 = 394.2\ \text{MJ}, \qquad E_{\text{tot}} = 159 + 394.2 = 553.2\ \text{MJ}$$

and the use phase is only $394.2/553.2 = \boxed{71.3\ \%}$ of the total, not the 90 %-plus that an always-on appliance such as a refrigerator would show. Nearly three tenths of the burden is decided in the bill of materials, before the product is ever plugged in. Doubling the service life from five to ten years therefore lowers the annualised burden from

$$\frac{553.2}{5} = 110.6\ \text{MJ/yr} \quad\text{to}\quad \frac{159 + 788.4}{10} = 94.7\ \text{MJ/yr}, \qquad \boxed{-14.4\ \%}$$

The honest reading of that number is that longevity helps but does not dominate, and that the strongest argument for repairability here is not energy at all but material recovery, cost to the owner, and the fact that appliances are discarded for the failure of one CAD 4 part. In the Canadian setting this now has legislative support: the 2024 amendments to the Copyright Act (Bills C-244 and C-59) removed the technological-protection-measure barrier to diagnosis, maintenance and repair, so a design that publishes a service manual and sells the seal, the coupling and the switch as spares is aligned with, not ahead of, the regulatory direction.

(ii) Design the noise down, because this appliance is used in shared walls. A blender at 92 dBA is one of the loudest objects in a home. Adding a compliantly mounted motor, a stiffer base and a jar shroud to reach 82 dBA is a factor of ten in acoustic intensity, $10^{(92-82)/10} = 10$, and roughly a halving of perceived loudness. In occupational terms, using the Canadian 3 dB exchange rate about an 85 dBA criterion, the permissible daily exposure is

$$T = \frac{8}{2^{(L-85)/3}}: \qquad L = 92 \Rightarrow T = 1.59\ \text{h}; \qquad L = 82 \Rightarrow T = 16\ \text{h}$$

No domestic user runs a blender for 1.6 hours, so this is not a domestic hearing-damage argument; it is an argument about a commercial smoothie counter where the same machine runs all shift, and about apartment living where the neighbour has no choice in the matter.

(iii) Design for the end of life, and for the users the market usually skips. Concretely: a mono-material housing (one polymer family, marked to ISO 11469) so that the shredded stream is worth recovering; screwed rather than ultrasonically welded joints so the motor and the printed-circuit assembly can be removed in under two minutes; and the accessibility target of part B applied not as a marketing feature but as a floor, since a control anyone can operate is the least expensive form of social inclusion a manufacturer can buy.

D. Converting design improvement ideas into realistic engineering specifications

An improvement idea becomes an engineering specification through four disciplined moves, and it is the fourth that most candidates omit. First, name the metric — the physical quantity that will be measured, with its units. “Easy to open” is not a metric; “peak two-handed couple force to release the collar, in newtons” is. Second, name the test — the apparatus, the sample size, the conditioning and the acceptance rule, because a metric without a test method is unenforceable in a supply contract. Third, set two values, not one: a marginal value, below which the product is not viable and the programme should stop, and an ideal value, which is the target the design is aimed at. Setting a single number either over-constrains the designer or is quietly abandoned. Fourth, trace each specification back to the need it serves and forward to the subsystem that owns it, so that when a specification later has to be relaxed the argument is about a customer need and a responsible engineer, not about a number in a spreadsheet.

The specification values themselves come from three sources in descending order of authority: a mandatory external requirement (a standard, a regulation, a retailer's condition of listing), a benchmark of competitor products measured on the same rig, and, only where neither exists, an engineering estimate that is recorded as an estimate. Applying all of this to the blender gives the target specification table below.

E. Establishing priorities when the specifications cannot all be met

Specifications conflict, and on this product they conflict sharply: metric 5 wants a high-temperature jar, metric 10 wants a cheap one; metric 7 wants mass and damping, metric 3 wants a gearbox that adds a second noise source. The priority rule has to be established before the conflict is visible, or the loudest voice in the room decides it. The procedure is in four steps.

1. Separate the constraints from the objectives. A specification that comes from a regulation, a safety standard or a legal requirement is a must: it is not tradable, and a concept that fails it is deleted rather than penalised. On this product metrics 1 and 6 are musts (an operating force above the accessible ceiling and a jar that softens below the boiling point are both non-negotiable), and the remaining eight are wants that can be scored. Screening before scoring is what stops a high total from hiding a fatal failure.
2. Fix the weights, from the part-A survey, before any concept is rated. The weights come from the confirmatory survey's importance ratings, normalised to unity, and they are recorded and signed off while the concepts are still abstract. Reversing this order — rating first, then adjusting weights until the preferred concept wins — is the commonest failure of decision matrices in practice, and it is undetectable after the fact.
3. Screen, then score. A Pugh matrix against the existing product as datum removes the concepts that are worse on the musts. The survivors go into a weighted matrix, rated 1 (poor) to 5 (excellent) on each want, and the weighted total is $S = \sum w_i r_i$. Here three architectures survive screening: A, the baseline with lugs added only; B, a two-speed geared universal motor with the lever collar and a Tritan jar; C, a brushless motor with electronic torque control, the lever collar, a Tritan jar and an acoustic enclosure.
4. Run a sensitivity pass and let it change the answer. Re-score with the weights moved to the edge of their plausible range. If the ranking survives, the decision is robust and can be defended; if it inverts, the honest conclusion is that the decision is a judgement about weights, not about engineering, and it must go to the person who owns the market positioning.

Under the base weights concept C wins at 4.15 against B at 3.75. Under the cost-dominant weighting — unit cost raised from 0.15 to 0.35, taken out of range of applications and noise — the ranking inverts: B scores 3.60 against C at 3.35. That inversion is the most useful output of the whole exercise, because it says precisely what the decision depends on. If the product is aimed at a premium counter-top segment where cost weighs 0.15, build C. If it is aimed at the mass channel where cost weighs 0.35, build B. What must not happen is that the matrix is presented as though it settled the question by itself.

Where a want simply cannot be met, three moves are available in order of preference: relax the specification with the customer's agreement (drop metric 7 from 82 to 88 dBA, which the survey says costs little), re-partition the architecture so the conflict disappears (a brushless motor delivers metric 3 without a gearbox, which removes the gearbox's contribution to metric 7 at the same time), or de-scope to a later release and record the deferral in the specification document rather than deleting the row. Deleting the row is what makes the same conflict reappear, undiscussed, in the next programme.