Yield stress, ultimate stress — material behavior: Question bank

Definition Allowable stress, required FoS, and the margin formula (defined here, used below)

$FoS_{req}$ = the required factor of safety — the safety multiplier the standard demands (spacecraft: ~1.25 against yield, ~1.5 against ultimate).
Allowable stress $σ_{allow}$ = the largest stress we permit, obtained by dividing the material limit by the required FoS: $\sigma_\text{allow}=\frac{\sigma_u}{\text{FoS}_\text{req}}\quad\text{(ultimate)}.$$ The required FoS is *already inside* $\sigma_\text{allow}$ — do not divide by it a second time.$
Margin of safety compares the allowable to the applied stress $σ_{app}$ : $MoS = \frac{σ _{allow}}{σ _{app}} - 1 \geq 0$

Why this is the formula: "safe" means the applied stress must not exceed the allowable, i.e. $σ_{app} \leq σ_{allow}$ . Dividing gives $σ_{allow} / σ_{app} \geq 1$ ; subtracting $1$ re-centres "just safe" at $0$ , so a positive MoS is spare capacity and a negative MoS is a fail. Because the required FoS was already baked into $σ_{allow}$ , putting another $FoS_{req}$ in the denominator would double-count the safety factor — that is a common and costly mistake.

True or false — justify

A material with a higher Young's modulus

E

(the elastic slope, in

Pa

) is always stronger.

False.

E

is stiffness (resistance to elastic stretch), not strength. Two steels can share

E

yet have wildly different

σ_{y}

; strength lives in

σ_{y}, σ_{u}

, not the elastic slope.

Yield stress

σ_{y}

changes if you make the rod thicker.

False.

σ_{y}

is a material property (force per unit area). A thicker rod carries more force

F

(in

N

) at yield (

F = σ_{y} A

), but the stress limit itself is unchanged.

Once a part yields it has failed and separated.

False. Yield is the onset of permanent (plastic) deformation — the part is bent but still in one piece. Separation happens later, at fracture, well past

σ_{u}

Engineering stress equals true stress right up to fracture.

False. They agree only while the area barely changes. After necking the real area

A_{inst}

shrinks fast, so true stress (

σ_{t} = F / A_{inst}

) climbs while engineering stress (

σ = F / A_{0}

) falls.

The ultimate stress

σ_{u}

is the stress at the exact moment the specimen breaks.

False.

σ_{u}

is the peak of the engineering curve, reached before fracture. After the peak the specimen necks and the engineering stress at fracture is actually lower than

σ_{u}

Doubling the applied force always doubles the risk of yielding.

True in the sense that it doubles the stress (

σ = F / A_{0}

, area fixed), which is what yield compares against — so you go from

σ

2 σ

relative to the same fixed

σ_{y}

A margin of safety of exactly

0

means the design has failed.

False.

MoS = 0

means the applied stress exactly equals the allowable (which already includes the required FoS) — it passes, but with zero spare. Negative MoS is the failure case.

Spacecraft use the same factor of safety for yield and ultimate.

False. Yield typically uses

FoS_{req} \approx 1.25

and ultimate

\approx 1.5

. The larger ultimate FoS reflects that fracture is catastrophic and non-recoverable, whereas yield only ruins fit/spec.

Yield and ultimate stresses are the same numbers whether you pull or push the material.

Generally false. Ductile metals have roughly symmetric yield in tension and compression, but compression suppresses cracking so brittle materials (glass, ceramics, concrete) are far stronger in compression than tension — the two limits can differ by an order of magnitude.

Spot the error

"Stress is just how hard you push, so a bigger force means a more stressed part."

The error: it ignores area. Stress is ==

F / A_{0}

==, force per area. The same force on a thinner rod gives higher stress; thickening

A

lowers stress without changing the load.

MoS = σ_{allow} / (FoS_{req} \cdot σ_{app}) - 1

, dividing again by the required FoS to be extra safe."

The error: double-counting the safety factor.

σ_{allow}

already equals

σ_{y} / FoS_{req}

, so the FoS is baked in. The correct form is

MoS = σ_{allow} / σ_{app} - 1

; adding another

FoS_{req}

makes the design look far more marginal than it is.

"The 0.2% offset line starts at the origin, parallel to nothing in particular."

Wrong on both counts. The offset line starts at

ε = 0.002

(not the origin) and is drawn ==parallel to the elastic slope

E

==. Its crossing with the curve defines

σ_{y}

(see the figure under "Why questions").

σ_{y}

is where the curve first becomes non-linear, so it equals the proportional limit."

Not the same point. The proportional limit is where Hooke's law

σ = E ε

first breaks. Yield is where permanent strain begins — it sits slightly higher, and for gradual-yielding alloys we pin it with the 0.2% offset.

"To be safe we should use true stress in the design allowables, since it's more physical."

Handbooks and allowables are almost always engineering stress (

F / A_{0}

). Feeding a true-stress value into an engineering-stress formula mis-reads

σ_{u}

, because the two diverge after necking.

"A part loaded to

0.9 σ_{y}

is fine because it hasn't yielded."

It hasn't yielded, but for yield the check is against the allowable

σ_{allow} = σ_{y} / FoS_{req}

. With

FoS_{req} = 1.25

the allowable is

0.8 σ_{y}

, so

0.9 σ_{y}

fails the margin even though no permanent set has occurred yet. See Factor of safety and margins.

"If MoS is positive by yield, the part is automatically safe against fracture too."

Not guaranteed. Yield and ultimate are separate checks with different limits (

σ_{y}

σ_{u}

) and different required FoS. You must clear both independently.

Why questions

Why do engineers care about two stress numbers instead of one "failure" stress?

Because the two failures are qualitatively different: yield ruins the part's shape/spec permanently (recoverable structure gone), while ultimate is separation. A precision antenna can be "failed" at yield long before it would ever fracture.

Figure — Yield stress, ultimate stress — material behavior

Why is the 0.2% offset line drawn parallel to the elastic slope, and not from the origin?

Look at the figure. Unloading from any point in the plastic region traces a line parallel to $E$ back down, and where it lands is the permanent strain left behind. So a parallel line at

ε = 0.002

marks exactly "

0.2%

permanent set," and its crossing with the curve is the repeatable

σ_{y}

Why does the stress–strain curve start out straight?

Atomic bonds act like tiny springs; a small stretch gives a restoring force linear in displacement. Summing over all bonds in the cross-section makes the macroscopic response linear:

σ = E ε

. See Young's modulus and elasticity.

Why does the curve peel away from the straight line at yield?

Dislocations begin to glide — planes of atoms slip irreversibly. That slip is not recovered on unloading, so the strain no longer returns proportionally and the curve bends over.

Why does stress keep rising after yield instead of staying flat (strain hardening)?

Dislocations pile up and tangle, obstructing each other, so more stress is needed to keep them moving. The material literally gets harder to deform until it peaks at

σ_{u}

Why do we need the 0.2% offset at all — why not read yield off directly?

Many alloys have no sharp yield point; the bend is gradual, so "where it yields" is ambiguous. The offset pins yield to a fixed 0.2% permanent strain, giving a repeatable number across labs.

Why does the engineering stress drop after the ultimate point?

A local neck forms; the true cross-section

A_{inst}

shrinks rapidly there. Since engineering stress divides by the fixed original area

A_{0}

, a smaller load over that unchanged

A_{0}

shows up as falling stress even though the material at the neck is more stressed than ever.

Why is a larger factor of safety chosen for ultimate than for yield?

Because fracture is unrecoverable and catastrophic, while yield leaves an intact (if deformed) part. More consequence ⇒ more margin demanded. Links to Structural load cases and launch loads where worst-case loads must never reach fracture.

Why doesn't stiffness (

E

) appear in the yield/ultimate margin checks?

Margins compare stress limits to applied stress.

E

governs how much a part stretches, not the stress at which it permanently deforms or breaks — stiffness and strength are independent design levers. Relevant in Material selection for spacecraft.

Edge cases

What happens to strain in the elastic region if the applied stress is exactly zero?

Strain is exactly zero and fully recoverable — the "spring" is at rest. This is the degenerate baseline:

ε = σ / E = 0

, no stored energy, no permanent set.

A perfectly brittle material (like glass) — where is its yield point on the curve?

There essentially isn't one. It stays linear-elastic right up to fracture, so

σ_{y} \approx σ_{u}

and the plastic/strain-hardening region vanishes. The "Every Young Ultimately Fractures" story collapses to Elastic→Fracture.

Does a material yield and break at the same stress in compression as in tension?

Not necessarily. Compression tends to close cracks, so brittle materials carry far more compressive than tensile stress; ductile metals are more symmetric but their design allowables still list tension and compression separately. This matters for launch-load struts that see both — see Structural load cases and launch loads.

How do temperature and how fast you load change

σ_{y}

and

σ_{u}

Heating a metal generally lowers both

σ_{y}

and

σ_{u}

(bonds and dislocations move more easily), while loading faster (high strain rate) or colder tends to raise them and make the material more brittle. Handbook allowables are therefore quoted at a stated temperature and rate — never assume room-temperature values on a hot thruster bracket or a cryogenic tank.

If a load is applied and fully removed while stress stayed below

σ_{y}

, what remains?

Nothing permanent — the part springs back to its original length (zero residual strain). That's the operational meaning of the elastic region: reversible.

If a part is loaded just past yield and then unloaded, what does the unloading path look like?

It comes down along a line ==parallel to the original elastic slope

E

==, ending at a nonzero (permanent) strain. That leftover strain is exactly the plastic deformation the offset method measures.

What is the margin of safety if the applied stress equals the allowable stress exactly?

MoS = σ_{allow} / σ_{app} - 1

. With

σ_{app} = σ_{allow}

this is ==

0

== — a pass with no spare capacity; any extra load pushes it negative. (No second FoS in the denominator — it is already inside

σ_{allow}

A cyclic load stays below

σ_{y}

forever — is the part safe indefinitely?

Not necessarily. Staying below yield defeats static failure, but repeated cycles can still cause fatigue cracking below

σ_{y}

. That's a different limit — see Fatigue and fracture mechanics.

At the instant of necking (peak

σ_{u}

), what is happening to the true area versus

A_{0}

The local true area

A_{inst}

starts shrinking below

A_{0}

rapidly. So true stress (

F / A_{inst}

) shoots up while engineering stress (

F / A_{0}

) turns down — the two curves split right at

σ_{u}

Recall Quick self-test

Name the three confusions the mnemonic collapses to. ::: Stiffness vs strength, yield vs fracture, engineering vs true stress. Does $σ_{y}$ depend on the rod's area? ::: No — material property; only the failing force scales with area. Why is there no extra $FoS_{req}$ in the MoS denominator? ::: Because $σ_{allow} = σ_{y} / FoS_{req}$ already contains it; dividing again would double-count the safety factor.