Units: where engineering answers go wrong

Ask any engineering lecturer where the marks are lost and the answer is never the physics. It is millimetres entered where the formula wanted metres, and an answer wrong by a factor of a million that still looks like a number.

Seven base units, everything else derived

The SI system defines seven base units — the metre, kilogram, second, ampere, kelvin, mole and candela — and builds every other unit from them by multiplication and division. A newton is a kilogram-metre per second squared. A pascal is a newton per square metre. A joule is a newton-metre, and a watt is a joule per second.

This is not trivia. It means every formula you use is dimensionally consistent by construction, and it gives you a free error check: if the units on both sides of an equation do not match, the equation is wrong regardless of how the arithmetic came out. Working out what units your answer should carry before you calculate is the fastest bug-catcher in engineering.

The kilogram is the odd one out

Every other base unit takes prefixes cleanly — millimetre, kilometre, microsecond. The kilogram already contains a prefix, which makes it the one unit where the pattern breaks. A thousand kilograms is a megagram, conventionally called a tonne; a thousandth of a kilogram is a gram, not a millikilogram.

Because mass appears in nearly every mechanical formula, this is a persistent source of error. Enter grams where a formula expects kilograms and your force, energy and power results are all a thousand times too large, in a way that looks entirely plausible if you are not checking magnitudes.

The square and cube trap

Converting a length is one multiplication. Converting an area or a volume is not, and this catches people constantly. There are ten millimetres in a centimetre — but a hundred square millimetres in a square centimetre, and a thousand cubic millimetres in a cubic centimetre. The factor is raised to the same power as the dimension.

Stress calculations are where this bites hardest, because cross-sectional areas are usually measured in square millimetres and the formula wants square metres. A 500 mm² section is 0.0005 m², not 0.5. Getting that conversion wrong by the obvious factor of a thousand rather than the correct million produces a stress figure a thousand times out — and both the right and wrong answers look like reasonable numbers on a page.

Same units, different quantity

Dimensional consistency is necessary but not sufficient. Torque and work both come out in newton-metres because both are a force times a distance, yet they are not the same physical quantity: work multiplies force by distance moved along its own line, while torque multiplies force by the perpendicular distance to a pivot. By convention torque stays in newton-metres and is never written as joules, precisely so the two do not get confused on a drawing.

Strain is the opposite case — it has no units at all, being a length divided by a length. That is why Young's modulus, which is stress divided by strain, carries the same units as stress. A modulus reported without units, or in units that are not pressure, is a red flag.

Order-of-magnitude checking

Before accepting any calculated figure, ask whether its size is believable. Structural steel yields around 250 to 355 megapascals, so a stress result of 50 MPa is comfortable and 50,000 MPa is a unit error. Atmospheric pressure is about 101 kilopascals. A person weighs roughly 700 newtons. A kettle draws around 3 kilowatts.

Keeping a handful of these anchors in mind catches the great majority of unit errors instantly, because unit errors do not produce slightly wrong answers — they produce answers wrong by factors of a thousand or a million, which are obvious the moment you have something to compare against.

A working habit

Convert everything to SI base units before you start, not partway through. Write the units alongside every intermediate value rather than tracking them mentally. Check that the units of your result match what the quantity should be. Then sanity-check the magnitude against something you know.

Four steps, none of them clever, and between them they eliminate almost every error that is not a genuine misunderstanding of the physics.