01
Forces, and how structures stand up
Everything structural reduces to a small number of ways material is loaded, and recognising which one you're looking at tells you what will fail.
Tension pulls apart, and materials like rope, cable and steel are excellent in it. Compression pushes together, and stone, masonry, earth and concrete are strong there and weak in tension, which is why they're used in arches and thick walls and not as beams. Bending combines both, putting one face in tension and the other in compression, which is why a beam fails at its surfaces and not through the middle. Shear slides one part past another, and torsion twists.
That's why traditional materials are used the way they are. An arch turns a span into pure compression so stone can bridge it. A truss turns a span into a set of members each in pure tension or compression, which is far more efficient than bending.
Triangles are the reason. A triangle can't change shape without changing the length of a side, while a rectangle folds sideways freely. Every brace, every truss and every diagonal is that fact applied.
And load has to reach the ground continuously. Tracing the path from where a load is applied, through every member, to the foundation, and asking whether each part can carry it, is the single most useful structural habit there is.
02
The simple machines, and what they trade
Every mechanism is a combination of a handful of devices, and all of them trade force against distance.
A lever, a pulley, an inclined plane, a wedge, a screw and a wheel and axle all do the same thing: they let a smaller force move through a larger distance to produce a larger force through a smaller one. Nothing is created, and the arrangement is what changes.
That trade is the whole of mechanical advantage, and knowing it lets you size things. If a system multiplies force by a factor, it divides speed by the same factor and everything downstream has to carry the larger load.
Friction is the constant tax. It reduces the advantage, it produces heat, and it causes wear. Bearings, lubrication and rolling instead of sliding are all attempts to reduce it, and in some places it's deliberately increased instead, which is what a brake and a wedge rely on.
Gears and belts change speed and torque in the same trade, with different tolerance of misalignment and overload. A belt slipping under load is a protection; a gear train has none unless one is built in.
03
Materials, and why they fail
Choosing material is choosing a set of properties, and failures nearly always trace to a property that was ignored.
Strength is not one number. A material can be strong in compression and weak in tension, strong but brittle, or weak but tough. Toughness, the ability to absorb energy without breaking, is frequently more important than raw strength in anything that takes a shock.
Stiffness is separate from strength and is confused with it constantly. A stiff material deflects little under load; a strong one resists breaking. A structure can be strong enough and still unusable because it flexes too much.
Fatigue is the failure people don't anticipate. Repeated loading well below the breaking point will eventually crack a material, and the crack grows invisibly until it fails suddenly. Anything that vibrates, flexes or cycles is subject to it, and it's behind a large share of unexpected mechanical failures.
Then the environment: corrosion, rot, ultraviolet degradation, and the fact that dissimilar metals in contact and in the presence of moisture corrode faster than either alone. And temperature, which changes dimensions, and can make a tough material brittle when cold.
04
Measuring, and knowing how accurate you need to be
Getting dimensions right is half of practical work, and knowing how right they need to be is the other half.
Accuracy is closeness to the true value; precision is repeatability. They're different, and a measurement can be precisely wrong, which happens whenever an instrument is consistently misreading.
Errors accumulate. Measuring from each mark to the next carries every error forward, while measuring everything from one datum does not. That's why setting out from a single reference is the standard practice and why cumulative measurement produces work that's out at the far end.
Better still, transfer directly and not measuring at all. Marking a part from the thing it must fit, or using a story stick, removes the numbers and the errors that come with them. Craft traditions do this constantly and it's more accurate than a tape.
And match the tolerance to the job. Not everything needs to be exact, and pursuing precision where it isn't needed wastes time that the parts that do need it should have had.
05
Designing so that failure is safe
This is the part that separates engineering from making, and it's the most transferable idea in the subject.
Everything fails eventually. The question a designer asks is not whether but how, and whether that failure is tolerable. A rope that stretches and frays gives warning; a brittle component that snaps does not. Given the choice, choose the one that warns.
A factor of safety is deliberate excess capacity, and it exists to cover what you don't know: variation in materials, loads larger than expected, deterioration, and errors in the calculation itself. It's not padding, it's the acknowledgement that the inputs are uncertain. Anything carrying people gets more of it.
Design the failure point deliberately. A shear pin, a fusible link, a relief valve and a sacrificial component all exist so that when something is overloaded, the cheap and safe part goes first. That's why the wooden teeth in a mill gear were wooden.
And prefer arrangements that fail into a safe state. A brake held off by pressure fails on when the pressure goes; a valve held open by power closes when the power stops. Deciding which state a thing rests in when everything else fails is frequently the most important decision in a design.