Engineering MakingPareto Tradeoffs: When You Cannot Maximize Everything
Atlas stands at a large drafting table covered with bicycle frame prototypes, a laptop showing a scatter plot with a curved frontier line, and sticky notes marking wall-thickness and stiffness values on each frame. He holds a caliper in one hand and points at the frontier curve with the other, explaining to a student why reducing wall thickness to save weight always reduces stiffness within the same frame design.
- Explain why maximizing two competing objectives simultaneously is mathematically impossible in most real engineering systems.
- Identify which design candidates lie on a Pareto frontier and which are dominated by better alternatives.
- Compare designs on a Pareto plot to justify a selection based on stakeholder priorities rather than a single metric.
- Predict how shifting a design along the frontier changes the balance between two competing objectives.
- Construct a two-objective tradeoff plot and label the frontier, dominated region, and infeasible region.
Key terms
- Objective
- A performance or cost metric an engineer wants to maximize or minimize in a design.
- Dominated design
- A candidate that another design beats on every objective at once, so it is never worth choosing.
- Pareto frontier
- The set of non-dominated designs where improving one objective forces sacrificing another.
- Infeasible region
- Combinations of objectives unachievable with current materials and processes, beyond the physical limit today.
- Scalarization
- Combining multiple objectives into one weighted score to select a single design point.
Why You Cannot Maximize Everything
Within a fixed design family and material, competing objectives are physically coupled, so pushing one degrades another. Thinning a frame's walls cuts mass but lowers stiffness because bending resistance falls with the wall thickness; thickening them restores stiffness but adds mass. No setting maximizes both, which is why the honest output of optimization is not a single winner but a curve of balanced compromises. Recognizing this coupling early prevents the impossible promise of a design that is best on every metric simultaneously.
Domination and the Frontier
On a two-objective plot, design X dominates design Y if X is at least as good on both objectives and strictly better on one. Dominated designs are pure waste: a frontier alternative beats them with no downside, so they are discarded. The non-dominated survivors form the Pareto frontier, where every point is a legitimate choice and moving along the curve trades one objective for the other. Crucially, two frontier designs can both be correct for different users, while a dominated design is always wrong.
Frontier Shifts Versus Movement Along It
Two very different events look similar on a plot. Moving along the frontier trades one objective for another within today's technology and reflects a change in stakeholder priorities, not capability. Shifting the entire frontier outward, so that more performance is achievable at every cost level, signals a genuine technological improvement such as a better motor or material that expands what is feasible. Distinguishing these is essential: the first redistributes a fixed pie, while the second grows it.
Worked examples
Given frame designs by (mass kg, stiffness N/mm) — A(0.9, 180), B(1.1, 210), D(1.2, 190) — where lower mass and higher stiffness are preferred, determine which design is dominated.
- State the domination rule: a design is dominated if another is no worse on both objectives and strictly better on at least one.
- Compare D(1.2, 190) with B(1.1, 210): B has lower mass (1.1 < 1.2) and higher stiffness (210 > 190), so B beats D on both objectives.
- Conclude D is dominated by B and should be discarded.
- Check A(0.9, 180) and B(1.1, 210) against each other: A is lighter but less stiff, B is stiffer but heavier, so neither dominates the other.
- Identify A and B as non-dominated frontier points representing legitimate trade-offs.
Answer: Design D is dominated by B (B is both lighter and stiffer); A and B are non-dominated frontier points and both remain valid choices.
Activity
Six bicycle frame designs are plotted by mass (kg) and stiffness (N/mm). Drag each design card to the correct zone — Pareto Frontier, Dominated, or Infeasible — on the design space map.
Practice
Given four laptop designs by weight and battery life, identify which are dominated and which lie on the Pareto frontier.
Explain the difference between moving along a Pareto frontier and the entire frontier shifting outward.
Common mistakes to avoid
- Averaging two normalized objectives objectively scores designs.A simple average hides a 50/50 weight that belongs to stakeholders, and poor weights can even select a dominated design.
- Frontier designs that are superseded were dominated all along.Designs optimal under old technology are not dominated by each other; an outward frontier shift supersedes them by expanding what is feasible.
Check your understanding
A structural engineer plots tensile strength versus cost for eight steel alloy candidates. After removing all dominated alloys, three remain. What is true about these three remaining alloys?
A student argues: 'We should just normalize both objectives to a 0–1 scale and average them — that gives an objective overall score for every design.' What is the most accurate critique of this reasoning?
On a Pareto frontier plot of drone battery life (maximize) versus drone mass (minimize), a new motor design shifts the entire frontier outward — meaning more battery life at every given mass level. What does this frontier shift represent?
Recap
Within a fixed design family competing objectives are coupled, so optimization yields a Pareto frontier of non-dominated trade-offs rather than one winner. A dominated design, beaten on every objective, is always discarded, while frontier designs are all legitimate for different stakeholders; the engineer characterizes the frontier and lets priorities, not a hidden averaging weight, select the operating point.
Reflect
Think of a purchase where you traded one quality for another, and consider where your choice sat on its Pareto frontier.