How a product gets designed

Design starts before any drawing. Someone recognizes a need, and the need has to become a problem specific enough to solve. Then come a candidate design, analysis and optimization, and evaluation against the original need.

The arrows that matter point backwards. Evaluation can send a design back to analysis, or show that the problem was defined wrong. Each loop leaves the problem better understood than before.

A design has to be functional, safe, reliable, competitive, usable, manufacturable and marketable. Those requirements rarely agree.

Tolerance is the clearest example. A tighter tolerance can make a fit more functional and more reliable. It also changes how the part is made, and what it costs.

In one analysis of a pin-hole fit, ±0.14 mm (IT10) may not need finish machining at all. Tightening it to ±0.03 mm (IT7) roughly doubles the cost; ±0.02 mm (IT6) roughly triples it. The right tolerance is the loosest one the function allows.

The effect does not stop at the part. A tolerance changes tooling, tooling changes variation, variation changes margin, and margin changes what the customer experiences.

Why does a shaft shoulder get a generous radius instead of a sharp corner? Because of how parts actually fail.

A static check uses the stress–strain curve for yield and ultimate strength, Mohr's circle for principal stresses, and a failure theory such as von Mises. Many parts pass it and still crack. They fail in fatigue: a load well below yield, repeated millions of times.

Stage I, initiation. Cyclic slip at the surface opens a microcrack along shear planes, near 45° to the load. It starts where stress concentrates: a sharp corner, a scratch, a void. Below a threshold, a small crack may not grow at all.

Stage II, propagation. The crack turns perpendicular to the load and grows a little with each cycle. Changes in loading leave beach marks on the fracture surface. Its growth rate follows Paris' law, da/dN = C(ΔK)^m, which lets engineers estimate remaining life. In failure analysis, beach marks are strong evidence of fatigue, read alongside striations and the loading history.

Stage III, final fracture. The remaining section can no longer carry the load and breaks suddenly.

A sharp corner concentrates stress where Stage I begins. A radius spreads it out. That is the reason for the radius, and why surface finish and residual stress belong in the same decision.

Why not pick a fan from its datasheet? A fan rated at 43.7 CFM delivers that only with nothing in its way. Inside an enclosure, every wall, grille and component pushes back, and the airflow it actually delivers drops.

The model shows where the air really goes. So a fan is chosen from the system, not the datasheet: estimate how hard the enclosure pushes back first, then pick a fan that still delivers the airflow the design needs.

A design is finished when every feature can answer "why".