Material Insights · Understanding the Material World

What engineers do: rules of thumb, the method and the kinds of engineering knowledge: Normal design and radical design

All courses / Understanding Engineering / Module 1: What engineers do: rules of thumb, the method and the kinds of engineering knowledge Part 3 of 4

Normal design and radical design

Most of what engineers do is improve things whose working idea is already settled. A small part of engineering tries something whose working idea is unknown, with no promise that it will work. The two kinds of work need different knowledge, carry different risks and reward different habits. Knowing which kind a technology faces tells you a great deal about how fast it can improve and how likely it is to fail.

3.1 Normal design improves a device whose working idea and shape are known

Normal design starts from firm ground. The working idea is proven, the usual layout is known, and a version built along familiar lines can be expected to work. What remains is to make it better or to make it cope with harder demands: lighter, cheaper, longer-lived, more efficient, easier to repair, or suited to a new kind of buyer. Designing next year's two-wheeler, a bigger tractor, a lower-cost refrigerator or a longer span for a girder bridge are all normal design.

This work is easy to underrate. Shaving 5% off a gearbox's losses or a girder's weight can take years of clever effort, and the gains pile up. But each change is a step inside a tradition, not a break with it. Most engineers spend most of their careers here: in a two-wheeler plant or a road contractor's design office, far more people refine familiar products than attempt new ones.

Normal design is also where rules of thumb, tables of data and standards do most of the work. Because the working idea is stable, a community can store what it has learnt in codes, handbooks, drawings and in the heads of experienced staff. A young engineer inherits all this. That inheritance is why a modern ceiling fan, which seems so simple, is the result of a century of small, well-recorded improvements.

The payoff from small gains is large when volumes are large. Assume a scooter maker sells one million vehicles a year in India (see 'Motorcycle and scooter' in the Products table), and a redesign, from a lighter frame and better-tuned fuel injection, saves each scooter an assumed 10 litres of petrol a year, a few per cent of what a typical rider burns. Units first: 10 L x 1,000,000 = 10 million litres a year, or 10,000 cubic metres. At an assumed round ₹100 a litre this is ₹100 crore a year kept by buyers (calculated), for a change nobody would call a revolution. Pump prices in the big cities were ₹102 to ₹112 a litre in November 2026, so the round figure errs low.

Data Worked estimate: what a small saving is worth at scale
StepWorkingResult
Saving per scooterAssumed 10 litres of petrol a year10 L
Fleet of new scooters in a yearAssumed 1 million sold1,000,000
Total fuel saved a year10 L x 1,000,00010 million litres
Value to buyers10 million L x ₹100 per litre (assumed round figure)₹100 crore a year (calculated)

All inputs are round assumptions for the exercise, not a claim about any maker. Petrol cost ₹102 to ₹112 a litre in four metro cities in November 2026.

Source: HDFC Sky, 'Petrol, diesel prices today, November 11, 2026' (from Goodreturns), petrol Rs 102 to Rs 112 a litre in Delhi, Mumbai, Bengaluru and Chennai, https://hdfcsky.com/news/petrol-diesel-prices-today-november-11-2026-check-latest-rates-in-delhi-mumbai-kolkata-chennai-bengaluru-hyderabad
  • Vincenti, What Engineers Know and How They Know It (Johns Hopkins University Press, 1990) · ch. 1, 'Introduction: Engineering As Knowledge'
  • HDFC Sky, 'Petrol, diesel prices today, November 11, 2026' (from Goodreturns) · petrol Rs 102 to Rs 112 a litre in Delhi, Mumbai, Bengaluru and Chennai, https://hdfcsky.com/news/petrol-diesel-prices-today-november-11-2026-check-latest-rates-in-delhi-mumbai-kolkata-chennai-bengaluru-hyderabad

3.2 Radical design starts with a working idea nobody has proved

Radical design starts on open ground. The layout is undecided and the working idea itself may be untested. There is no expected shape, no code to follow, and no reason yet to believe the thing will perform well enough to be worth pursuing. So the first goal is humble: build something that shows the idea can work at all.

Solid-state lighting is a good case. For a century the household lamp was a thin metal wire heated until it glowed. That is a working idea based on heat. A light-emitting diode uses a different idea: electrons crossing a junction in a semiconductor give off light directly. The first red LEDs were dim curiosities. Blue ones took decades and were so hard that their inventors received a Nobel Prize. No one improving filament bulbs could have arrived at an LED by small steps, because it changed the working idea, not the details.

Work of this kind looks different. Rival layouts compete at once, because no one knows which is best. Old rules of thumb mislead, since they grew up around a different idea or material. Progress arrives in jumps and surprises, most attempts fail, and the survivors harden into the next tradition.

Radical design can also pay off hugely. The table compares a filament lamp and an LED lamp giving about the same light. At four hours a day the saving is about 75 kWh a year per lamp, and at an assumed round ₹7 per kWh, typical of what households pay for their last units each month, it is about ₹520 a year. With a lamp costing a fraction of that, it explains why households and the state moved quickly once the technology matured: India's UJALA programme, launched in January 2015, had sold 36.87 crore LED bulbs at ₹70 each by January 2025.

Data Worked estimate: filament lamp against LED lamp, four hours a day
QuantityFilament lampLED lamp
Light output (about)800 lumens800 lumens
Electric power (assumed round values)60 W9 W
Energy per year at 4 h a day (1,460 h)60 W x 1,460 h = 88 kWh9 W x 1,460 h = 13 kWh
Cost per year at an assumed ₹7 per kWhabout ₹615about ₹92
Saving per lamp (calculated)about 75 kWh, about ₹520

Values are typical round figures for a household lamp, not a particular product. See 'LED lamp' in the Products table and 'Lighting and displays' among the technologies.

  • Vincenti, What Engineers Know and How They Know It (Johns Hopkins University Press, 1990) · ch. 1, 'Introduction: Engineering As Knowledge'
  • Vincenti, What Engineers Know and How They Know It (Johns Hopkins University Press, 1990) · ch. 7, 'The Anatomy of Engineering Design Knowledge'
  • Press Information Bureau, Government of India, 'UJALA: 10 years' backgrounder (January 2025) · 36.87 crore LED bulbs at Rs 70 each, launched 5 January 2015, https://static.pib.gov.in/WriteReadData/specificdocs/documents/2025/jan/doc202516481801.pdf

3.3 The line is blurred by levels: a radical part can sit inside a normal device

Designs are layered. A passenger car is a system made of a body, a drive train, a chassis and electrics. The drive train holds an engine or motor, a transmission and a cooling circuit, and each of these has its own smaller parts. At each level there is a working idea and a usual shape. So novelty can sit at one level while the levels around it stay familiar.

An example from road transport: a car with an electric motor and a battery pack in the floor replaces the engine and fuel tank, which is a radical change in the drive train. Yet the car still has four wheels, steering, seats and a body that a driver recognises at once. The overall device is a normal car with a radical heart. Over time, as the heart becomes familiar, it becomes normal too, and designers stop thinking about it. The same happened with the brushless electronic motor in a ceiling fan. Customers see the same fan. The designer sees a new working idea inside it.

This layering has a practical consequence. When a radical part is added to a normal device, much of the old knowledge still applies, which speeds work. But the interfaces, where the new part touches the old, are where surprises appear. A battery is heavy, so the floor and the suspension have to be redesigned. It is charged, so the building's electrical supply matters. Engineers learn the new knowledge level by level.

Radical and normal are also relative. A change that looks timid to one industry may be deep for another. What matters for your reading of technologies is whether the working idea changed, whether the usual shape moved, and which levels were affected. Where the idea is new at the top, expect uncertainty and slow growth in knowledge. Where it is new only inside a part, expect faster progress.

Value chain A passenger car as layers, with one radical swap
  • Passenger car see Products table
    • Body and chassis normal
    • Drive train where the radical change sits
      • Internal combustion engine the older working idea
      • Electric motor with Lithium-ion battery the newer working idea
    • Steering, brakes and suspension normal, but redesigned for the extra weight
    • Electrics and software normal in kind, much larger in scope
Source: Vincenti, What Engineers Know and How They Know It (Johns Hopkins University Press, 1990), ch. 1, 'Introduction: Engineering as Knowledge'
  • Vincenti, What Engineers Know and How They Know It (Johns Hopkins University Press, 1990) · ch. 1, 'Introduction: Engineering as Knowledge'
  • Vincenti, What Engineers Know and How They Know It (Johns Hopkins University Press, 1990) · ch. 7, 'The Anatomy of Engineering Design Knowledge'

Check your understanding

Which change is radical design in the sense of this part?

Changing the source of light changes the working idea. The others improve a known device.

Why is normal design often a better bet for a buyer who values reliability?

A stable working idea gives a record of how things fail and standard checks to catch the failures.

An electric car keeps four wheels, seats and steering but replaces the engine. This is best described as:

The change is at the level of the drive train. Much surrounding knowledge still applies, but interfaces need new work.
← Part 2: The kinds of knowledge engineers useMark read · Part 4: The fields of engineering and the road ahead →