Why the exact same tensile test, on the exact same specimen, produces two different curves — one that appears to drop before fracture, and one that never does.
Pull a tensile specimen to fracture and plot the data one way, and the stress appears to peak, then fall as the test continues — as if the material were getting weaker right before it breaks. Plot the exact same test data a different way, and the stress climbs continuously, right up to the moment of fracture. Nothing about the physical test changed. What changed is the number sitting in the denominator of the stress calculation — and that single choice is the entire difference between engineering stress and true stress.
Engineering stress is calculated as force divided by the original, pre-test cross-sectional area — σ_eng = F / A₀ — a fixed number, measured once before the test ever starts. True stress instead divides by the actual, instantaneous cross-sectional area at that exact moment — σ_true = F / A_instant — a number that keeps shrinking as the specimen stretches and, especially, as it necks down right before fracture. The same split applies to strain: engineering strain is measured against the original gauge length, true strain against the continuously changing instantaneous length. Since the real cross-section is always shrinking during a tension test, true stress is always the larger of the two numbers — and the gap between them grows dramatically in the necking region right before final fracture.
Up until necking begins, deformation is uniform along the gauge length, and — assuming the material's volume stays constant as it stretches — engineering and true values convert cleanly: σ_true = σ_eng (1 + ε_eng) and ε_true = ln(1 + ε_eng). Once necking starts, deformation localizes entirely into that one narrowing region, the uniform-deformation assumption behind those formulas breaks down, and true stress in the neck can no longer be calculated from the engineering curve — it has to be measured directly, typically from the actual minimum neck diameter. That's also exactly why the two curves are drawn overlapping early on and only meaningfully separate once necking begins: before that point the difference is small and roughly correctable; after it, the physical behavior genuinely diverges.
False. The material is actually still strain-hardening — getting intrinsically stronger at the true, instantaneous level — right up until the moment it fractures. The apparent drop in the engineering curve is purely a bookkeeping artifact of dividing by the fixed original cross-sectional area (A₀) while the specimen's real cross-section is rapidly shrinking during necking. Force actually does start to fall during necking, because there is less real material left to carry it — but the stress the material is genuinely experiencing at the neck, computed against its actual shrinking area, keeps climbing the entire time. The true stress-strain curve shows the real physical behavior: continuous strain hardening, all the way to fracture.The engineering curve isn't wrong, exactly — it's answering a different, still-useful question (how much force can this original part size carry) rather than describing the material's intrinsic behavior at the point of deformation.
Explains why a tensile test produces two different stress-strain curves depending on whether the calculation uses the specimen's fixed original cross-section (engineering stress/strain) or its actual, continuously shrinking cross-section (true stress/strain) — and why the engineering curve appears to drop after the ultimate tensile strength while the true curve keeps rising all the way to fracture.
Engineering stress and strain are simple to calculate and standard for most design work: force divided by the specimen's original, pre-test area, and elongation divided by its original gauge length. Both are fixed reference values measured once, before loading begins. What that convenience hides is that the specimen's actual cross-section is shrinking continuously as it stretches — modestly during uniform elongation, dramatically during necking — so the engineering values increasingly understate what the material is truly experiencing as the test progresses.
True stress is force divided by the actual, instantaneous cross-sectional area at each moment, σ_true = F / A_instant, and true strain is based on the continuously changing instantaneous length rather than the fixed original length. Because the real area is always shrinking under tension, true stress is always equal to or greater than engineering stress at the same point in a test, with the gap widening sharply in the necking region. This is also why the engineering stress-strain curve shows a peak at the ultimate tensile strength (UTS) followed by an apparent decline toward fracture — the fixed A₀ denominator can't account for the shrinking real area — while the true stress-strain curve continues climbing right up to fracture, correctly reflecting that the material keeps strain-hardening the whole time. Before necking begins, the two are related by σ_true = σ_eng(1 + ε_eng) and ε_true = ln(1 + ε_eng), assuming uniform deformation and constant volume; after necking starts, deformation localizes and these conversion formulas no longer apply.
Engineering stress-strain values are what most standard design calculations, material spec sheets, and allowable-stress tables use, and they're perfectly adequate for elastic-range and typical working-stress design. True stress-strain data matters specifically wherever large plastic deformation is being modeled directly — metal forming and forging simulations, finite element analysis of ductile failure and necking behavior, and detailed constitutive material models — because engineering values become progressively less representative of actual material behavior exactly where that behavior matters most.
Because true stress divides the same applied force by the actual, currently-reduced cross-sectional area, while engineering stress divides it by the original, larger, pre-test area. Since the specimen's real cross-section only ever shrinks under tensile load (it never grows back), dividing by a smaller number always produces a larger result — so true stress is always greater than or equal to engineering stress at the same point in a test, and the gap grows substantially during necking.
The apparent drop is a pure artifact of the fixed denominator. Engineering stress is force divided by the original area, A₀, and during necking the force the specimen can carry does start to decrease — because there is genuinely less material left in the neck to carry it. But dividing that shrinking force by a denominator that never changes exaggerates the appearance of weakening. The true stress, calculated against the actual shrinking area, shows the material is still strain-hardening the entire time, right up to fracture.
Yes, but only up to the onset of necking. While deformation is uniform along the gauge length and the material's volume is assumed constant, σ_true = σ_eng(1 + ε_eng) and ε_true = ln(1 + ε_eng) hold. Once necking begins, deformation concentrates entirely in the neck, the uniform-deformation assumption fails, and true stress in the neck must be measured directly (for example, from the actual minimum neck diameter) rather than calculated from the engineering curve.
For most conventional design work — sizing a part for elastic loading, checking against a published yield or ultimate strength — engineering stress and strain are the standard and are what published material properties (from a mill certificate or datasheet) are reported in. True stress and strain become important specifically when modeling large plastic deformation directly, such as in metal forming simulations or detailed finite-element analysis of ductile fracture, where the shrinking real cross-section is exactly the behavior being studied.
Necking is characteristic of ductile materials being pulled in tension — once the material's strain-hardening rate can no longer keep pace with the loss of load-carrying capacity from the shrinking cross-section, deformation localizes into a neck rather than staying uniform along the gauge length. Very brittle materials may fracture with little to no visible necking, in which case the engineering and true curves stay much closer together throughout the test since there's little localized area reduction to diverge over.
Try our Mechanical Studio
More calculators, simulators, and guides for this discipline.