22-Mec-B5 Product Design and Development · May 2013
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Paper format. National Exams, May 2013 — 07-Mec-B5, Product Design & Development. Three hours; open book; no calculator permitted. Question 1 is compulsory and carries 40 % of the paper; four of the remaining six questions are chosen, each worth 15 %, for 100 % in total, and only the first five questions appearing in the answer book are marked. Note 5 of the paper states that most questions require an essay answer or the use of tables, figures and charts, and that clarity and organisation of the answer carry marks; Note 1 invites the candidate to state any assumption made where a question is open to interpretation, and that licence is used several times below with each use flagged. All seven printed questions are worked here — 130 marks of material against the 100 marks a candidate would actually attempt — so that the set serves as a complete study resource.
Reference texts. Ulrich & Eppinger, Product Design and Development (McGraw-Hill) — the framework text for this exam code, and the source of the generic development process, the needs-to-metrics translation, concept screening and concept scoring used throughout; Dieter & Schmidt, Engineering Design (McGraw-Hill) for the specification, materials and process-selection material; Pahl & Beitz, Engineering Design: A Systematic Approach (Springer) for systematic concept generation and the function structure; Boothroyd, Dewhurst & Knight, Product Design for Manufacture and Assembly (CRC) for the design-for-assembly and design-for-manufacture rules; Ashby, Materials Selection in Mechanical Design (Butterworth-Heinemann) and Kalpakjian & Schmid, Manufacturing Engineering and Technology (Pearson) for the process-selection charts and cost models. Canadian context is taken from the Patent Act, Industrial Design Act, Trademarks Act and Copyright Act (Canadian Intellectual Property Office), from CSA standards (notably CSA B651 Accessible design for the built environment), from the Canada Consumer Product Safety Act, and from Engineers Canada / EGBC guidance on professional practice and on equity, diversity and inclusion in the profession.
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.
1. The controlled digital product definition: CAD models, drawings and the product data management system. This is the authoritative channel, and its defining property is that it is controlled — every item has a part number, every document a revision, and the system distinguishes work in progress from released data, so that a manufacturing engineer can tell whether the geometry in front of them is safe to cut tooling from. The information it carries is the unambiguous what: geometry, dimensions and tolerances with a declared standard such as ASME Y14.5 or ISO GPS, materials and finishes, the bill of materials and the assembly structure. Its strengths are precision, persistence and traceability, and the fact that downstream systems (process planning, purchasing, service documentation) can consume it directly. Its weakness is that it records decisions without recording why they were made, and it is slow: a change reaches the team only when it is released.
2. Synchronous, face-to-face work around a shared object: design reviews, sketch sessions and physical prototypes. This is the high-bandwidth channel, and its value is that misunderstandings surface immediately. A whiteboard sketch session resolves an interface argument in minutes that would take a week of drawing revisions; a formal design review at a gate brings manufacturing, quality, service and purchasing to the same table where their objections can be tested against each other; a daily stand-up keeps the team's model of the project synchronised at low cost. The most effective form is a physical prototype or mock-up used as a boundary object, because it communicates across specialities and across languages without needing translation: everyone can see that the hitch cannot be reached with gloves on. The weakness is that nothing said in the room persists unless someone writes it down, which is why the review minute and the decision record are part of the method rather than an afterthought.
3. Structured written artefacts that carry requirements and rationale. Between the two sits the family of documents whose job is to record intent and reasoning: the product specification with its metrics and target values, interface control documents, design failure mode and effects analyses, concept-selection matrices with their weights and sensitivity results, analysis and test reports, engineering notebooks and the issue tracker. These carry the why, and they are the only channel that survives staff turnover. They are also the channel that regulators, auditors and the professional obligations of the engineer of record depend on. Their weakness is latency and discipline: they are written after the fact by people under schedule pressure, and they decay silently if they are not tied to a gate that requires them.
The three are complementary, not competing, and the practical rule that makes a team work is that decisions made in the fast channel must be landed in a slow one: a whiteboard agreement becomes an interface control document entry, a review objection becomes a tracked action item, a chosen concept becomes a scored matrix in the design record. Teams fail not because they lack channels but because decisions live only in the fastest one.
Optimization is a disciplined loop, and its first three steps are where most of its value is won or lost.
Step 1: state the objective function, the constraints and the design variables, in numbers. “Make it better” is not optimizable. The team must name one quantity to be minimised or maximised, the constraints that bound the search, and the variables that may be changed together with the ranges over which they may move. Where there are genuinely several objectives, they are either combined into a weighted objective or, better, one is optimised subject to limits on the others, so that the trade-off is visible rather than buried in a weighting.
Step 2: build a model that predicts the objective from the variables, and validate it. The model may be analytical, numerical (finite element or computational fluid dynamics), or empirical from designed experiments. It is worthless until it has been checked against measurement at a few known points, because an optimizer will exploit a model's errors ruthlessly — it will find the corner of the design space where the model is wrong and report it as the optimum.
Step 3: screen the variables. A designed experiment or a sensitivity study identifies which few variables actually move the objective. This is the step that makes the rest affordable, since optimizing over three influential variables is a different problem from optimizing over fifteen.
Step 4: search. Parametric studies, response-surface methods built on designed experiments, or formal numerical optimization are all legitimate; the choice depends on how expensive one model evaluation is. The output is a candidate optimum plus an understanding of how flat or sharp the optimum is.
Step 5: test robustness. A sharp optimum that only exists at nominal dimensions and nominal operating conditions is a trap. The candidate is re-evaluated across the manufacturing tolerance band and the range of real operating conditions — the robust-design view associated with Taguchi — and a slightly worse but flatter design is normally preferred.
Step 6: confirm by physical test, then freeze and document. The optimized design is built and measured against the objective and every constraint, because the confirmation run is the only evidence that the model was right about the optimum and not merely about the region it was validated in.
Example: the electric kettle on the kitchen counter. Suppose the objective is to minimise the energy consumed per cup of tea actually poured, subject to a constraint that one litre still reaches a full boil in no more than five minutes (a 1 500 W element on a 120 V circuit needs about four and a half minutes to bring a litre of cold tap water to the boil, so a shorter target is not physically reachable), that the external surface stays below a touch-safe limit, that the product remains certifiable to the applicable CSA electrical safety standard, and that manufacturing cost does not rise by more than a set amount. The information the team needs, and where it comes from, is specific:
| Information needed | Why it is needed | Where it comes from |
|---|---|---|
| Element power and its heat-transfer efficiency into the water | Sets boil time and the electrical energy input | Element supplier data plus calorimetric bench test |
| Thermal mass of the jug, base and element assembly | Energy spent heating the kettle rather than the water | Bill of materials, masses and specific heats |
| Wall material, thickness and any air gap; lid sealing | Governs standing heat loss and re-boil energy | CAD model plus a measured cooling curve |
| Distribution of fill volumes and of volume actually poured | The dominant loss is boiling water nobody drinks | Instrumented user study — the single most valuable datum |
| Number of boils per day and the ambient and inlet water temperature | Converts per-boil energy into annual energy | User study and Canadian municipal supply data |
| Thermostat cut-out behaviour and re-boil frequency | Controls how often the cycle repeats unnecessarily | Control system test log |
| Touch temperature of the external surface | A hard safety constraint on insulation strategy | Thermal test to the safety standard's method |
| Cost per unit of added insulation or of a minimum-fill indicator | Turns each option into cost per unit of energy saved | Supplier quotations and cost model |
The instructive part of the example is that the optimization does not end where an engineer's instinct expects. Insulating the jug and reducing its thermal mass both help, but the sensitivity study almost always shows that the largest term is the water that is boiled and never poured, so the highest-value design change is a clearly marked minimum-fill line, a sight glass calibrated in cups rather than litres, and a lid that pours without needing an over-fill. That conclusion is only reachable because step 1 defined the objective as energy per cup poured rather than energy per boil — which is the whole argument for spending time on the objective function before touching the model.
