Engineering Stress-Strain vs True Stress-Strain

Why This Article Is the Flagship for Material Data Entry

The conversion from engineering stress-strain to true stress-strain to plastic strain is one of the most common and most error-prone steps in preparing a material model for non-linear FEA. The tensile test produces engineering stress and engineering strain — force divided by original area, extension divided by original length. But large-deformation plasticity models in FE solvers work in the current configuration, and they expect true stress and true strain (or true stress and plastic strain). Entering engineering stress-strain data where true stress-plastic strain is expected produces a material model that misrepresents the yield and hardening behaviour — the predicted response is wrong, and any margin computed against it is meaningless. This article walks through the definitions, the conversions, the limitations, and the exact chain of steps that takes raw test data and converts it into the format the FE solver expects.

Engineering Stress and Engineering Strain

Engineering stress and engineering strain are the measures obtained directly from a tensile test using the original specimen dimensions. The engineering stress is the applied force divided by the original cross-sectional area. The engineering strain is the extension divided by the original gauge length. These measures are simple to compute and they are what material data sheets typically report. They are adequate for the elastic region and the early plastic region, where the cross-sectional area and the gauge length have not changed significantly. But as deformation becomes large — particularly after the onset of necking — the engineering measures diverge increasingly from the actual material state, because the original dimensions no longer represent the current geometry. Engineering stress-strain is a response property; it is a valid characterisation of the test, but it is not the measure that large-deformation plasticity models expect.

Engineering stress:

  σeng = F / A0

Engineering strain:

  εeng = ΔL / L0 = (L − L0) / L0

where:
  F      = applied force (N)
  A0     = original cross-sectional area (m²)
  L      = current gauge length (m)
  L0     = original gauge length (m)
  ΔL     = extension = L − L0 (m)

True Stress and True Strain

True stress and true strain account for the changing geometry of the specimen during the test. The true stress is the applied force divided by the instantaneous cross-sectional area — not the original area. The true strain is the natural logarithm of the ratio of the current length to the original length, which accumulates strain incrementally as the specimen elongates. For small strains (the elastic region), the engineering and true measures are nearly identical. For large strains (the plastic region), they diverge: the true stress continues to rise after the ultimate tensile strength because the area at the neck is reducing faster than the material hardens, while the engineering stress (force over original area) falls. The true measures represent the actual material state in the current configuration, which is what large-deformation plasticity models require.

True stress:

  σtrue = F / Ainstantaneous

True strain:

  εtrue = ln(L / L0)

For uniform deformation (constant volume, valid before necking):

  σtrue = σeng (1 + εeng)
  εtrue = ln(1 + εeng)

where:
  F               = applied force (N)
  Ainstantaneous  = current cross-sectional area (m²)
  L               = current gauge length (m)
  L0              = original gauge length (m)

The Specimen Deformation Sequence

The tensile test proceeds through a sequence of deformation stages, and the validity of the engineering-to-true conversion depends on which stage the material is in. The diagram below shows the specimen progression and the corresponding stress-strain curve behaviour.

TENSILE SPECIMEN DEFORMATION SEQUENCE

  Stage 1: INITIAL SPECIMEN (undeformed)
  ┌───────────────────────────────────────┐
  │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│
  └───────────────────────────────────────┘
   ← L0 (original gauge length) →
   ↑ A0 (original cross-section) ↓
  Geometry: uniform. Engineering & true measures identical.

  Stage 2: UNIFORM ELONGATION (after yield, before necking)
  ┌───────────────────────────────────────────────────┐
  │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│
  └───────────────────────────────────────────────────┘
   ← L (increased) →
   ↑ A (uniformly reduced) ↓
  Deformation is uniform along the gauge length.
  Constant-volume conversion σtrue = σeng(1+εeng) IS VALID here.

  Stage 3: NECKING (after ultimate tensile strength)
  ┌───────────────────────────┐
  │░░░░░░░░░░░ ▓▓ ░░░░░░░░░░░│
  └───────────────────────────┘
               ↑ local neck — area drops sharply here
  Deformation is LOCALISED at the neck.
  Cross-section is NON-UNIFORM along the gauge length.
  Constant-volume conversion is NO LONGER VALID.
  True stress must be computed from measured neck area.

  STRESS-STRAIN CURVES (schematic):

  Stress
   ↑
   │        True stress  ╱─────── keeps rising
   │                   ╱
   │                 ╱
   │    ────────────●  ← Ultimate tensile strength (engineering peak)
   │   ╱          ╲       Engineering stress  ╲──── falls
   │  ╱             ╲
   │ ╱                ╲
   │╱                   ╲──── Fracture
   └──────────────────────────────→ Strain
         ↑ elastic   ↑ uniform   ↑ necking
         region      plastic     region

  KEY: Engineering references ORIGINAL geometry (A0, L0).
       True references INSTANTANEOUS geometry (Ainst, L).
       The two curves coincide at small strain and diverge at large strain.

Limitations After Necking

The constant-volume conversions — σtrue = σeng(1+εeng) and εtrue = ln(1+εeng) — are valid only during uniform deformation, before the onset of necking. Once necking begins, the deformation is no longer uniform along the gauge length: the cross-section reduces rapidly and locally at the neck while the rest of the gauge length deforms little. The engineering strain, computed from the average extension over the original gauge length, no longer represents the strain at the neck. The conversion formulas, which assume uniform deformation and constant volume distributed evenly, break down. After necking, the true stress can only be determined by measuring the actual cross-sectional area at the neck — which requires additional instrumentation (such as a diametral extensometer or optical measurement) during the test, or post-test measurement of the fracture surface. This is a critical limitation: the simple conversion of published engineering stress-strain data to true stress-strain is valid only up to the ultimate tensile strength. Beyond it, the converted data is not reliable unless the neck area was measured. For FE models that need the full curve to fracture, this must be handled carefully — either by truncating the curve at the ultimate, by using a Bridgman correction if neck geometry is available, or by using a hardening law extrapolated from the uniform region.

The conversions σtrue = σeng(1+εeng) and εtrue = ln(1+εeng) are valid ONLY during uniform deformation — before necking. After the ultimate tensile strength, deformation localises at the neck, the cross-section is non-uniform, and these formulas no longer apply. Converting published engineering data beyond the ultimate without measured neck geometry produces unreliable true stress-strain data.

Why Engineering Data Cannot Simply Be Entered into Plasticity Models

Large-deformation plasticity models in FE solvers operate in the current configuration. The solver computes the deformation incrementally, updating the geometry at each step, and it evaluates the constitutive law using true stress and true strain measures. If the engineer enters engineering stress-strain data and the solver interprets it as true stress-strain, the yield and hardening behaviour is misrepresented: the material appears to be softer than it actually is (because engineering stress is lower than true stress at the same strain after yield), and the hardening is understated (because the engineering curve falls after the ultimate while the true curve continues to rise). The predicted plastic response is then wrong — the structure appears to yield earlier and redistribute load differently than the real material would. This is not a minor numerical discrepancy; it is a misrepresentation of the material behaviour that can change the predicted failure mode and the computed margin. The engineer must convert the data to the form the solver expects before entering it.

Conversion to Plastic Strain

Most FE plasticity models require true stress versus plastic strain, not true stress versus total strain. The plastic strain is the permanent component of the total strain — the total true strain minus the elastic strain. The elastic strain at any point on the curve is the true stress divided by the Young's modulus. So the plastic strain is computed by subtracting the elastic strain from the total true strain at each point on the curve. This step removes the elastic component and leaves only the permanent deformation, which is what the plasticity model tracks internally. The conversion must be done point by point along the curve, and it should be verified: the plastic strain should be zero at the yield point (where the total strain is entirely elastic up to yield), and it should increase monotonically thereafter.

Plastic strain (from true stress and true total strain):

  εplastic = εtotal,true − σtrue / E

where:
  εplastic      = true plastic strain (dimensionless)
  εtotal,true   = true total strain = ln(1 + εeng)
  σtrue         = true stress = σeng (1 + εeng)
  E             = Young's modulus (Pa)

  The term σtrue / E is the elastic strain recovered on unloading.
  At the yield point: εplastic = 0 (all strain is elastic up to yield).
  After yield: εplastic increases as the curve progresses.
  The plastic strain must be monotonically increasing for the solver.

The Conversion Chain — From Test Data to FE Input

The full conversion chain takes the raw measured engineering stress-strain curve and produces the true stress versus plastic strain table that the FE plasticity model expects. Each step has a specific purpose and a specific assumption. The engineer must perform each step, verify the result, and document the traceability of the final data back to the original test. The diagram below shows the chain.

CONVERSION CHAIN: TEST DATA → FE PLASTICITY INPUT

  ┌─────────────────────────┐
  │ 1. MEASURED ENGINEERING │   From tensile test:
  │    STRESS-STRAIN CURVE  │   σeng = F/A0,  εeng = ΔL/L0
  │  (σeng vs εeng)         │   Raw test output.
  └───────────┬─────────────┘
              │
              │  Apply (valid before necking):
              │    σtrue = σeng (1 + εeng)
              │    εtrue = ln(1 + εeng)
              ▼
  ┌─────────────────────────┐
  │ 2. TRUE STRESS /        │   Now in current-configuration measures.
  │    TRUE STRAIN          │   Elastic + plastic combined.
  │  (σtrue vs εtrue)       │
  └───────────┬─────────────┘
              │
              │  Remove elastic strain:
              │    εplastic = εtrue − σtrue / E
              ▼
  ┌─────────────────────────┐
  │ 3. TRUE STRESS /        │   Permanent strain only.
  │    PLASTIC STRAIN       │   This is what the plasticity model tracks.
  │  (σtrue vs εplastic)    │
  └───────────┬─────────────┘
              │
              │  Verify: monotonic, εplastic = 0 at yield,
              │  consistent with E, within valid range.
              ▼
  ┌─────────────────────────┐
  │ 4. FE PLASTICITY INPUT  │   Entered into the solver material card
  │  (σtrue vs εplastic)    │   per the solver's required format.
  └─────────────────────────┘

  NOTE: The exact required format depends on the solver and the
  material formulation. Check the solver documentation for whether
  it expects true stress vs plastic strain, true stress vs total
  strain, or another measure. Do not assume the format.

Check the Solver's Expected Format

Different FE solvers and different material formulations expect the stress-strain data in different forms. Some expect true stress versus plastic strain. Some expect true stress versus total true strain and compute the plastic strain internally. Some accept engineering stress-strain for small-strain formulations and require true for large-strain. Some have specific keywords or cards that specify the measure. The engineer must establish exactly what the solver expects before preparing the data — and the expectation depends on the solver, the element formulation, the material model type, and whether the analysis is small-strain or large-deformation. The documentation is the authority; assumptions are not. Entering data in the wrong format is one of the most common and most consequential material model errors, and it often produces no warning from the solver — the solver simply runs with the wrong data and produces a wrong answer.

BEFORE ENTERING A STRESS-STRAIN CURVE INTO AN FE SOLVER, ESTABLISH EXACTLY WHAT STRESS AND STRAIN MEASURES THE MATERIAL MODEL EXPECTS. Is it true stress vs plastic strain? True stress vs total strain? Engineering stress-strain? The answer depends on the solver, the formulation, and the element type. Entering the wrong form produces incorrect material behaviour — often with no solver warning.

Engineering vs True Stress-Strain Compared

The following table summarises the key differences between engineering and true stress-strain, including their validity and their typical use in FE analysis.

Verification of the Converted Data

After converting the data, the engineer should verify the result before entering it into the solver. The plastic strain should be zero at the yield point and increase monotonically. The true stress should be greater than or equal to the engineering stress at the same nominal point (since σtrue = σeng(1+εeng) and εeng is positive in tension). The initial slope of the true stress-total strain curve should equal the Young's modulus (since the elastic region is the same in both measures). The curve should be smooth and physically reasonable — discontinuities or non-monotonic segments suggest an error in the conversion or in the raw data. For data beyond the ultimate, if the neck area was not measured, the engineer should flag the limitation and consider truncating the curve or using a validated extrapolation. Verification is a response-property check — it confirms that the data entering the material model is correct, which is a precondition for a meaningful margin.

VERIFY THE CONVERTED CURVE BEFORE ENTERING IT: plastic strain zero at yield and monotonic thereafter; true stress ≥ engineering stress; initial slope equals E; curve smooth and physically reasonable. Data beyond the ultimate without measured neck geometry should be flagged as a limitation.

Key Takeaways

• Engineering stress-strain references original geometry (A0, L0); true stress-strain references instantaneous geometry (Ainst, L)
• For uniform deformation: σtrue = σeng(1+εeng), εtrue = ln(1+εeng) — valid before necking only
• After necking, deformation is localised and the conversions break down — true stress needs measured neck area
• FE large-deformation plasticity models expect true stress and plastic strain, not engineering stress-strain
• Plastic strain = εtotal,true − σtrue/E; it is zero at yield and increases monotonically
• Always check the solver documentation for the exact stress and strain measures the material model expects
• The conversion chain is: engineering curve → true stress/true strain → remove elastic strain → true stress/plastic strain → FE input