Structural design process — load cases, FOS (factor of safety): Visual walkthrough

This deep dive expands the central result of Structural design process — load cases, FOS (factor of safety). If a word here feels heavy, the parent note has the reference table.

Step 1 — A part, and the push on it

WHAT. Picture one real piece of a spacecraft: a metal strut. Something pushes or pulls on it. That push is a force — we will call its size $P$ , measured in newtons (N), the unit of a push (roughly the weight of a small apple is $1 N$ ).

WHY start here. Before any formula, we need one honest physical thing: a load. Everything else is bookkeeping around this single arrow.

PICTURE. The strut is the grey rod. The magenta arrow is the force $P$ trying to stretch it.

Step 2 — We never know $P$ exactly: the limit load

WHAT. In real life we cannot measure the launch push before we fly — we forecast it. Our best forecast of the largest push the part will ever feel is named the limit load, written $P_{limit}$ .

WHY a special name. "Limit" means the edge of the expected world — the worst case among all launch load cases. But nature scatters: the true push might land a little above our forecast. Hold that thought — it is the entire reason a factor of safety exists.

PICTURE. A fuzzy cloud of possible real loads. Our forecast $P_{limit}$ (violet line) sits at the top of what we expect, but the cloud has a tail poking past it.

Step 3 — Fear inflates the demand: multiply by $F O S$

WHAT. To cover the scatter (and our analysis errors, weak welds, temperature effects), we deliberately design against a bigger load than we forecast. We multiply: $load we size for P_{design} = a number > 1 F O S \times our forecast P_{limit}$

$F O S$ (factor of safety) is a pure number with no units, like $1.25$ .
Multiplying stretches the demand arrow. It does not touch the material — it inflates the push we pretend to feel.

WHY multiply the load, not the metal. The uncertainty lives in the load (we guessed the push). So we pad the thing we are unsure about. The material strength is a measured fact; we leave it alone.

PICTURE. The violet limit arrow is stretched by $F O S$ into the longer orange design arrow. Same direction, more length.

Step 4 — What the part can actually take: the allowable

WHAT. Every material has a ceiling — a load (or stress) beyond which it misbehaves. We call that ceiling the allowable, $S_{allow}$ . From Stress and Strain — Yield vs Ultimate Strength there are two ceilings:

Yield ceiling — go past it and the part stays bent forever (permanent deformation).
Ultimate ceiling — go past it and the part rips apart (rupture).

WHY two ceilings. A bent bracket may still hold; a snapped one never does. So we get two separate limits and must clear both.

PICTURE. A vertical strength bar for the material, with two marks: the lower yield line and the higher ultimate line.

Step 5 — The face-off: design load vs allowable

WHAT. Now put the two on the same axis and compare. We demand: $what it can take S_{allow} \geq what we make it face F O S \times S_{limit}$

WHY same axis. Comparing a force to a strength only makes sense if both are the same kind of quantity — both loads, or both stresses. We line up the inflated demand bar against the allowable bar and just look at which is taller.

PICTURE. Two bars side by side: orange = inflated demand $F O S \times S_{limit}$ , violet = allowable $S_{allow}$ . If the strength bar is taller, we survive.

Step 6 — Turn "taller" into a number: the Margin of Safety

WHAT. "Taller by how much?" We divide the strength bar by the inflated demand bar, then subtract $1$ : $M S = \frac{S _{allow}}{F O S \times S _{limit}} - 1$

Read it term by term:

$\frac{S _{allow}}{F O S \times S _{limit}}$ = ratio of strength to inflated demand. If strength is exactly equal, this is $1$ .
$- 1$ shifts the "just barely OK" point to zero, so the number reads as fractional spare.

So $M S = 0.20$ means " $20%$ strength to spare beyond the factored load".

WHY subtract 1. People find " $0$ = exactly enough, positive = spare, negative = fails" far easier to eyeball than " $1$ = exactly enough". The $- 1$ recenters the ruler on the pass/fail line.

PICTURE. The bar comparison of Step 5, now with the extra strip on top labelled $M S$ — the fraction the strength bar sticks up above the demand bar.

Try the parent's Example 2 on the picture: $S_{allow} = 180$ , $S_{limit} = 120$ , $F O S = 1.4$ . Inflated demand $= 1.4 \times 120 = 168$ . Then $M S = 180/168 - 1 = 0.071$ — a thin sliver of spare, and the picture shows exactly how thin.

Step 7 — Edge cases: the three ways the picture degenerates

WHAT. A rule is only trustworthy if we know what it does at the extremes. Three special situations:

$M S = 0$ exactly — strength bar equals demand bar. Passes, but zero spare. One bad weld and it fails.
$M S < 0$ — demand bar is taller than strength bar. Fails. Redesign: thicken, add a rib, or change material (see Finite Element Analysis of Spacecraft Structures to find where it fails).
$F O S = 1$ — no inflation at all. Demand bar = raw limit load. Then $M S$ compares strength to our unpadded forecast — any scatter above the forecast breaks the part. This is why real spacecraft never use $F O S = 1$ ; qualification testing (see Qualification vs Acceptance Testing) uses factors above ultimate.

WHY show all three. So you never meet a bar chart you cannot read. Positive, zero, negative, and the degenerate no-cushion case are the only outcomes — and here they all are.

PICTURE. Three mini bar-pairs side by side: pass with margin (green tick), $M S = 0$ knife-edge (amber), and fail (red cross).

The one-picture summary

WHAT. The entire chain, left to right: forecast the load → inflate it by $F O S$ → stand it against the material allowable → read off $M S$ .

Recall Feynman: tell the whole walkthrough to a friend

I have a rocket part and something pushes on it. I can't measure that push before I fly, so I make my best guess of the worst push — that's the limit load. But my guess could be low, so I make the part face a push that's, say, $1.25 \times$ bigger — that multiplier is the factor of safety; it grows the push, never the metal. Now I ask what the part can actually hold — its allowable — and there are two answers: the "don't bend forever" answer (yield) and the "don't snap" answer (ultimate). I stand the inflated push next to the strength on one ruler. If strength is taller I'm safe; how much taller, as a fraction, is the margin of safety $M S$ . I subtract $1$ so that $0$ means "exactly enough", positive means "spare", negative means "redesign". Zero margin is a knife-edge, and a factor of $1$ means no cushion at all — both are how parts get killed on launch day. That's it: guess the push, pad it, compare to strength, read the margin.

Recall

Which arrow does $F O S$ stretch — the load or the strength? ::: The load (demand). It leaves the material strength untouched. In $M S = S_{allow} / (F O S \cdot S_{limit}) - 1$ , what does the ratio equal when strength exactly meets the inflated demand? ::: Exactly $1$ , so $M S = 0$ — the knife-edge, just barely OK. Why does $F O S = 1$ never appear on real spacecraft? ::: It gives no cushion; any real load scattering above the forecast breaks the part. What are the two allowable ceilings and what does passing each prevent? ::: Yield (no permanent bend) and ultimate (no rupture); the part must clear both.