# Cost, trade-offs and resources: why the best design is rarely the strongest: Optimising, Pareto fronts and the price of margin

Source page: https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/p4/

[All courses](https://www.materialinsights.org/learn/) / [Understanding Engineering](https://www.materialinsights.org/learn/engineering/) / [Module 13: Cost, trade-offs and resources: why the best design is rarely the strongest](https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/) Part 4 of 5

## Optimising, Pareto fronts and the price of margin

Optimising is the habit of looking for the best compromise when two costs pull in opposite directions. With one goal there is often a single answer. With several, such as cost, weight and carbon, there is a set of answers none of which beats another everywhere. This part shows both and how to tell a robust choice from a fragile one.

### 4.1 A single optimum exists where one cost falls as another rises, and it is usually flat

A water pipe for a village supply is a classic pair of opposites. A wider pipe costs more to buy and lay, but the water flows through it with less friction, so the pump needs less power. A narrower pipe is cheap but the pump runs hard every day. Add the two yearly costs and there is a diameter where the sum is lowest.

The physics comes from the module on Fluids. For a fixed flow, the speed of water in a pipe goes as one over the diameter squared, and the pressure lost to friction grows with the square of the speed and falls with the diameter, so pressure loss goes as one over diameter to the fifth power if the friction factor is held constant. Pumping power then also goes as D to the minus 5. The pipe cost, at first approximation, grows in proportion to D.

Put in rounded numbers for 1 km: at 100 mm, the yearly capital charge on the pipe is Rs 40,000 and the yearly cost of pumping energy is Rs 1.2 lakh. Writing x for the diameter divided by 100 mm, the total is 40,000x + 120,000 / x^5. The minimum is where x^6 = 15, so x = 1.57 and the diameter is about 157 mm. At that point the pipe cost is exactly five times the energy cost, a neat check that the balance is right: the exponent of 5 sets the ratio.

The table shows the sum for four standard sizes. The 150 mm pipe is within half a per cent of the true minimum, and even 125 mm is only 18 per cent worse, because the minimum is flat. This is the usual shape of an optimum: costs near it change little, so a standard size, a rounded dimension or a slightly generous margin loses almost nothing.

Data Worked estimate 9: yearly cost of 1 km of village water pipe for four standard sizes (assumed prices)

| Pipe diameter | Capital charge a year (Rs 40,000 x D/100) | Pumping energy a year (Rs 120,000 x (100/D)^5) | Total a year (calculated) |
|---|---|---|---|
| 100 mm | Rs 40,000 | Rs 120,000 | Rs 160,000 |
| 125 mm | Rs 50,000 | Rs 39,300 | Rs 89,300 |
| 150 mm | Rs 60,000 | Rs 15,800 | Rs 75,800 |
| 200 mm | Rs 80,000 | Rs 3,750 | Rs 83,750 |

Prices are illustrative round numbers for the method, not tender rates. The exact optimum is 157 mm at about Rs 75,400 (calculated), so 150 mm is within 0.5 per cent. Pipe cost is taken as proportional to diameter, a simplification.
 Source: Method: standard pipe economic-diameter argument; physics as in the module on Fluids, assumed prices

- Kosky et al., Exploring Engineering (Academic Press, 2013) · ch. 5, 'Engineering Economics' (annual cost of borrowed capital, the basis of the yearly pipe charge)

### 4.2 With several goals there is no single best, only a front of options that nobody beats on every count

Return to the three panels. If cost, mass and carbon all matter, there is no way to call one the winner without choosing how much of each to give up. What can be done is to remove the options that are plainly beaten. Option A dominates option B if A is at least as good on every count and strictly better on one. The options that nothing dominates form the Pareto front, named after the Italian economist Vilfredo Pareto. Choosing among them is then a matter of values, not of calculation.

Using the numbers from Part 3, primary aluminium (4.9 kg, US$15, 74 kg CO2e) is dominated by recycled aluminium (4.9 kg, US$15, 2.5 kg CO2e): same mass and price, much less carbon. Steel stays on the front only because it is the cheapest, by less than US$2 a panel, and recycled aluminium looks almost as good on paper. That near tie should make you suspicious, not pleased. The table leaves out the things that decide it: there is only so much scrap, forming and welding aluminium costs more, and aluminium fatigues in a way steel does not. A Pareto table is only as honest as its axes, and the most important axis is often the one not in it.

Engineers use two practical habits. The first is to set limits first (the panel must not exceed a mass, a cost, a carbon figure) and optimise one remaining quantity within them. The second is a weighted score, where each count gets a weight, but this hides the trade-off in the weights. If a score changes winner when the weights move by a quarter, the design has not really been decided. Test the ranking against a range of weights, as Part 1 tested the pump against a range of discount rates: a choice that survives is robust.

Data Pareto check on the equal-stiffness panels (smaller is better on every column)

| Option | Mass (kg) | Material cost (US$) | Embodied carbon (kg CO2e) | On the front? |
|---|---|---|---|---|
| Steel, hot-rolled | 10.0 | 13.5 | 22 | Yes: cheapest, by a small margin |
| Aluminium, primary | 4.9 | 15 | 74 | No: recycled aluminium beats it on carbon alone |
| Aluminium, recycled | 4.9 | 15 | 2.5 (understated) | Yes, on paper, but scrap supply limits it |
| Carbon-fibre laminate | 3.0 | 105 | 94 | Yes: lightest, so nothing beats it on mass |

The front depends on which axes are counted. Adding a fourth, availability at scale, would push recycled aluminium off it.
 Source: Comparison built from worked estimates 6 and 7 of this module, Part 3  Common misreading

#### "Best" always hides a list of things given up

A design is best only for stated goals, under stated limits, at a stated time. The best engineers state what the design gives up and what would change the answer. A design presented as optimal with no list of what was traded off has hidden its assumptions, not removed them.

- Hammack, The Things We Make (Sourcebooks, 2023) · ch. 2, 'Best: The Vitruvian Man, Kodachrome, and the Constant Search for Better'
- Kosky et al., Exploring Engineering (Academic Press, 2013) · ch. 11, 'Materials Engineering' (optimising one property subject to requirements)

### 4.3 A safety margin is a purchase, and the right amount depends on what failure would cost

In the module on Design, failure and safety, a factor of safety bought a small chance of failure. Here is the money side of that bargain. A margin costs material and weight, which means rupees. A failure costs repairs, lost output, injury and trust. For each extra step of margin, an engineer can ask whether it saves more in expected loss than it costs.

Expected loss is the chance of the failure multiplied by what it costs. A Rs 2 lakh machine part, if it fails, stops a line and brings a Rs 50 lakh loss. If the chance of that failure in the year is 1 in 100, the expected loss is Rs 50,000 a year. A stronger design that halves the chance to 1 in 200 saves Rs 25,000 a year, so it is worth up to Rs 25,000 a year more (about Rs 1.5 lakh over ten years at 10 per cent, calculated). Past that, the margin costs more than it saves.

Two cautions apply. First, the chance of a rare event is the number we know least, so it is a poor basis for shaving a margin that protects lives. Where harm could be severe, codes set a floor and no cost sum is allowed to go under it, as the module on Design, failure and safety describes with ALARP. Second, the cost of a failure falls on people who did not choose the saving, so the one who benefits from the cut and the one who bears the risk are often different. The price of margin has to be judged by someone who answers for both.

- Kosky et al., Exploring Engineering (Academic Press, 2013) · ch. 5, 'Engineering Economics'
- Hammack, The Things We Make (Sourcebooks, 2023) · ch. 2, 'Best: The Vitruvian Man, Kodachrome, and the Constant Search for Better'

### See it in the table
[H₂O Water](https://www.materialinsights.org/commodities/water/)[Al Aluminium](https://www.materialinsights.org/commodities/aluminum/)[St Steel](https://www.materialinsights.org/commodities/steel/)

### Check your understanding

In the pipe example, the diameter 150 mm is chosen instead of the exact 157 mm. What is the cost of that choice?
 At 150 mm the total is Rs 75,800 against Rs 75,400 at 157 mm, a difference of about half a per cent. The optimum is flat, so a standard size costs almost nothing.

Option A is lighter and cheaper than option B but emits more carbon. What follows?
 A dominates only if it is at least as good on every count. B is better on carbon, so neither beats the other and the choice depends on how much each count matters.

A failure costing Rs 50 lakh has a 1 in 100 yearly chance. A change that cuts the chance to 1 in 200 is worth paying for up to:
 Expected loss falls from Rs 50,000 to Rs 25,000 a year, a saving of Rs 25,000. For lethal hazards, codes also set a minimum margin no matter what this sum says.

[← Part 3: Choosing a material under cost, weight and carbon](https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/p3/)[Mark read · Part 5: Resources, recycling and designing for the conditions you have →](https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/p5/)

## Other sections

- [Module overview](https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/index.md)
- [Whole-life cost and the price of money](https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/p1/index.md)
- [Why things get cheaper: learning, scale and standards](https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/p2/index.md)
- [Choosing a material under cost, weight and carbon](https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/p3/index.md)
- [Resources, recycling and designing for the conditions you have](https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/p5/index.md)
- [Key terms, quiz and sources](https://www.materialinsights.org/learn/engineering/cost-tradeoffs-resources/review/index.md)
