Conceptual Design Process Questions and Answers

Answer: A

Correct answer: Designs where no objective improves freely

A Pareto front is the set of non-dominated designs: for each of them, no objective can be improved without making another worse. A design lighter and cheaper than another dominates it and pushes it off the front. The front is therefore not one design but the whole surface of legitimate compromises - every point on it is optimal, and choosing among them means weighting the objectives against each other.

Answer: B

Correct answer: That judgement belongs to the customer

Collapsing several objectives into a single score requires weights, and those weights encode a value judgement - how much fuel burn is worth how much acquisition cost - that the design team is not entitled to make on the customer’s behalf. Presenting the front keeps that judgement where it belongs and makes the trade explicit. It guarantees nothing about the lightest design, and it certainly requires more analysis rather than less.

Answer: C

Correct answer: 90 per cent

The saving is

Ninety per cent fewer analyses, in exchange for resolving the main effects and the important interactions rather than every combination. That is the bargain design of experiments offers, and it is what makes a five-variable study feasible within a schedule at all.

Answer: D

Correct answer: 3 per cent

The error relative to the true value is

Three per cent is respectable for a surrogate used to guide a search, but it is not good enough to decide on: a three per cent error in take-off weight is three tonnes on a ninety-tonne aircraft. The usual discipline is to search with the surrogate and confirm the shortlisted designs with the full analysis.

Answer: A

Correct answer: Both A and R are true, and R is the correct explanation of A

Both are true and the reason is why the check matters. A surrogate is fitted to a sample of full analyses and interpolates well within the region those samples cover; outside it, or in a corner where the sampling was sparse, its predictions can be badly wrong and it gives no warning. Optimisers are drawn precisely to extremes, so the design a surrogate search recommends is often exactly where its accuracy is least trustworthy.

Answer: B

Correct answer: Both A and R are true, and R is the correct explanation of A

Both are true and the reason accounts for the practice. When objectives compete there is no single optimum, only a front of non-dominated designs, and selecting a point on it means deciding what one objective is worth in terms of another. That weighting is a commercial and operational judgement belonging to the customer, so the design team presents the front and the trade it embodies rather than concealing the decision inside a weighted score.

Answer: C

The first two are correct. The third overstates the role: the surrogate is built from high-fidelity runs and its promising predictions are confirmed by more of them, so it redistributes that analysis rather than replacing it.

Answer: D

Correct answer: 1 and 2 only

Statements 1 and 2 define the Pareto front and its consequence. Statement 3 is the opposite of the truth: combining objectives into one score does not remove the judgement, it merely hides it inside the weights, which are chosen by whoever writes the objective function. Making the weighting explicit, or presenting the front instead, is the honest treatment.

Answer: A

The four combine into the standard modern workflow: choose points wisely, fit a cheap model, search it, then present the trade-offs rather than a single answer. Each step exists because the full analysis is too expensive to run exhaustively.

Answer: B

Correct answer: It requires no high-fidelity analysis to construct

This is not a limitation but a false statement about how surrogates work. A surrogate is built from high-fidelity analyses - typically a designed sample of them - and its whole purpose is to make those expensive runs go further. The genuine limitations are the other three: accuracy depends on sampling density, extrapolation beyond the sampled region is unreliable, and predictions should be confirmed with the full analysis before a decision rests on them.