F F ε̄ = 0.44 DIC FIELD
Aerospace & Defence · Material Characterisation · VFM

Full-Field Material Calibration of Inconel 625 — Without Classic Coupon Methods

5 Johnson–Cook parameters. 3 strain rate regimes. 3 specimens instead of 12. Simultaneous identification from full DIC fields — with validated uncertainty quantification and independent test confirmation.

Virtual Fields Method Johnson–Cook Inconel 625 Digital Image Correlation SHPB 10⁻³ – 10⁴ s⁻¹ Sensitivity-Based VF Uncertainty Quantification
Scope a calibration study All Case Studies

Project Snapshot

MaterialInconel 625 (NiCrMo)
SectorAerospace & Defence
MethodVirtual Fields Method + DIC
ModelJohnson–Cook (5 params)
Strain rate range10⁻³ – 10⁴ s⁻¹
Specimens used3–4 vs 9–12 (classic)
Validation error4.2% peak force
UQ methodMonte Carlo (DIC noise)
EquipmentServo-hydraulic · SHPB · >100k fps
DeliverableJC material card (LS-DYNA / Abaqus)

5 parameters. 3 regimes. Validated against an independent impact test.

Independent validation · 4.2% peak force error · 3.8% energy error · Inconel 625 batch
Specimen reduction
3–4 specimens vs 9–12
for classical coupon suite
Same alloy batch
4.2%
Validation force error
Independent Charpy-type
impact specimen at 5 m/s
Classical method: 11%
50k
Field points per image
Full DIC strain field
vs 1 global force-disp.
curve (classic method)
7×10⁷
Strain rate span
10⁻³ to 10⁴ s⁻¹
Quasi-static · Intermediate
SHPB dynamic

A nickel superalloy. Five coupled parameters. Seven decades of strain rate.

Inconel 625 is widely used in aerospace, nuclear, and defence applications — turbine casings, pressure vessels, high-velocity impact shields — where the material must be proven safe across a wide range of temperatures, strain rates, and stress states. Reliable simulation of these components depends entirely on the quality of the constitutive material card.

Classical calibration builds that card from a suite of uniaxial specimens: smooth tensile for A, B, n — then separate high-rate compression tests for C — then heated specimens for m. Each test is post-processed independently, the stress–strain curve is curve-fitted, and parameters are handed off sequentially. Each step propagates the errors of the previous one.

The core problem: every classical test delivers a single global scalar — one force–displacement curve. The stress and strain fields inside the specimen are assumed uniform and never measured. For a material like Inconel 625 where the hardening, rate sensitivity, and failure mode interact non-trivially, this assumption is a source of systematic bias that no amount of additional specimens resolves.

The classical approach doesn't lack data — it lacks spatial information. VFM doesn't run more tests; it extracts ten thousand times more mechanical information from each one.

Inverse identification from full strain fields — one optimisation, all parameters.

SIMFORGE applied the Virtual Fields Method (VFM) — a full-field inverse technique grounded in the Principle of Virtual Work — to identify all five Johnson–Cook parameters simultaneously from DIC strain fields acquired during a small number of purpose-designed heterogeneous tests.

Rather than assuming a uniform stress state, the method exploits the spatial gradient of the strain field as a rich mechanical signal. Virtual fields are constructed analytically to be maximally sensitive to each target parameter, directly conditioning the identification problem.

Specimen geometry was not arbitrary. A sensitivity-driven FE pre-study was used to select notch geometry such that the condition number of the identification system was minimised — guaranteeing parameter decoupling and numerical robustness before a single test was run.

ParameterPhysical meaningRegime
AInitial yield stressQuasi-static
B, nStrain hardening coefficient & exponentQuasi-static
CStrain rate sensitivityIntermediate + SHPB
mThermal softening exponentHigh-temp campaign
ε̇₀Reference strain rate (fixed)10⁻³ s⁻¹ baseline

The Virtual Fields Method — core mechanics.

VFM is built on the weak form of the equilibrium equation — the Principle of Virtual Work. In standard FEA, this principle is used to solve for displacements given a constitutive model. VFM inverts the problem: given measured displacements and strains (from DIC), it solves for the constitutive parameters.

Principle of Virtual Work — dynamic form
−∫V σ : ε* dV + ∫∂V T · u* dS + ∫V ρa · u* dV = 0
σ: Cauchy stress · ε*: virtual strain · T: boundary traction · ρa: inertia term (non-negligible in dynamic regimes)

For the Johnson–Cook model, the VFM cost function assembles a linear system in the parameter space by choosing virtual fields u* that are orthogonal to each parameter's sensitivity field. This yields a closed-form identification equation — no iterative FE model updating required.

Johnson–Cook flow stress
σ̄ = (A + Bε̄ⁿ)(1 + C ln ε̇*)(1 − T*ᵐ)
Multiplicative form: each term identifiable in a dedicated strain rate / temperature regime
VFM identification system
H · p = b  →  p = H⁻¹ b
H: sensitivity matrix (4×4 for isothermal case) · p: parameter vector · b: virtual work from boundary conditions

Geometry optimised for identifiability — not convenience.

The sensitivity-based virtual field approach requires that the strain field span the full parameter sensitivity space simultaneously. A smooth dog-bone specimen fails this requirement: its gauge section is nearly uniform, meaning H is near-singular and parameter cross-talk is severe.

For Inconel 625, a notched flat specimen with a 4 mm notch radius and 3 mm minimum width was selected after iterating on the FE sensitivity analysis. The objective function was the condition number κ(H) — minimised across candidate geometries:

Identifiability criterion
κ(H) = ‖H‖ · ‖H⁻¹‖ → minimise over specimen geometry
Final geometry: κ(H) reduced by 3.1× vs. smooth dog-bone under identical boundary conditions

The optimised geometry produces local equivalent strains up to ε̄ = 0.44 at the notch root before fracture — sufficient to activate the full hardening range — while maintaining quasi-uniform strain in the far-field gauge section for reference state comparison.

Key geometry parameters
Minimum width3 mm
Notch radius4 mm
Gauge length20 mm
Peak ε̄ (notch root)0.44
κ(H) reduction vs smooth3.1×

Three regimes. One unified identification framework.

The full JC identification is structured across three strain rate regimes, each requiring a different experimental platform. The VFM cost function and parameter extraction engine are identical across all three — only the inertia treatment changes.

Regime 01
10⁻³–10⁻¹ s⁻¹
Quasi-static
Screw-driven or servo-hydraulic load frame. Standard 2D DIC at 1–5 fps. Inertia term negligible — classic static VFM balance. Single notched flat specimen per rate.
→ Identifies: A, B, n
Regime 02
1–100 s⁻¹
Intermediate
Servo-hydraulic high-speed machine. 20–100k fps camera. Inertia term ρa enters the balance — must be evaluated from measured acceleration field. Scientifically the richest regime for C identification.
→ Identifies: A, B, n, C
Regime 03
10³–10⁴ s⁻¹
SHPB Dynamic
Split-Hopkinson Pressure Bar. Ultra-high-speed camera (>100k fps). DIC on bar and specimen. VFM enables relaxation of classical SHPB equilibrium assumptions — heterogeneity is exploited, not avoided.
→ Identifies: A, B, n, C (validation)
Why the intermediate regime matters

The 1–100 s⁻¹ window is systematically under-characterised in the literature: too fast for quasi-static machines, too slow to be treated as SHPB wave propagation. Classical approaches either ignore it or interpolate between two extremes. VFM treats it natively — the dynamic balance equation includes the inertia term explicitly from measured acceleration fields, making this the highest-fidelity identification window for the rate sensitivity parameter C.

Five JC parameters. Confidence intervals from DIC noise propagation.

The following parameter set was identified via VFM across the three strain rate regimes. Confidence intervals are derived from a Monte Carlo uncertainty propagation on the DIC displacement noise floor (σ_u ≈ 0.02 px RMS) — providing traceable bounds, not just point estimates.

ParameterSymbolVFM ResultLiteratureRegime
Yield stress A 485 ± 11 MPa 450–520 QS
Hardening coeff. B 612 ± 22 MPa 580–650 QS
Hardening exp. n 0.41 ± 0.02 0.38–0.45 QS
Rate sensitivity C 0.013 ± 0.002 0.010–0.017 Int. + SHPB
Thermal softening m 1.24 ± 0.08 1.10–1.40 High-T
Reference rate ε̇₀ 10⁻³ s⁻¹ Fixed
Validation against independent test

Charpy-type notched impact specimen at 5 m/s impact velocity. FE simulation using VFM-identified JC parameters reproduced peak force within 4.2% and energy within 3.8%. Classical calibration on the same batch: 11% peak force error on the same validation geometry.

More information per test. Fewer tests total.

The performance advantage of VFM over classical calibration is not marginal — it is structural. Classical calibration extracts one scalar per test. VFM extracts 10,000–50,000 spatial data points per image. This fundamentally changes the information density of the identification problem.

Specimens required (full JC identification)
Classic
9–12 specimens
VFM
3–4 specimens
Test campaign duration
Classic
~6 weeks
VFM
~2 weeks
Data points per test event
Classic
~10
VFM
10k – 50k / image
Independent validation error (Inconel 625)
Classic
11% peak force
VFM
4.2% peak force

For rare or expensive nickel alloys — single-crystal, ODS, AM-deposited — the 3× reduction in specimen count alone can justify the VFM approach on material cost. For high-rate aerospace qualification programmes, the combined time and accuracy improvement is the differentiator.

Four findings that change how you specify a material card.

01

Parameter C is systematically overestimated by classical methods

Classical identification of C relies on comparing two independently obtained stress–strain curves at different rates. Any systematic error in either test (friction, wave dispersion, alignment) transfers directly into C. VFM identifies C from the spatial strain rate gradient within a single specimen — the systematic errors cancel. For Inconel 625, the VFM-derived C was 18% lower than the classical estimate from the same batch.

02

The intermediate strain rate regime is not a gap — it's the most informative window

The 1–100 s⁻¹ range is where inertia effects become measurable but wave mechanics do not dominate. VFM's explicit inertia term makes this regime natively tractable. The dynamic VFM balance provides richer constraint on both B/n and C simultaneously — in a single test — than either a quasi-static suite or a SHPB set provides individually.

03

Uncertainty quantification is built in — not an afterthought

The VFM system H·p = b is linear in the parameters (for the isothermal JC case). This means a Monte Carlo propagation of DIC noise through the identification system yields full confidence intervals on every parameter simultaneously — in minutes, not days. Classical methods typically report no UQ, or derive it from repeat specimen scatter, which conflates identification error with material variability.

04

Specimen geometry is an engineering design decision — not a standard

Classical standards (ISO 6892, ASTM E8) prescribe smooth geometry specifically to achieve uniform stress — which is exactly what VFM does not want. The notched geometry used here is not a compromise; it is a precision instrument. The sensitivity-based FE pre-study that selects the notch radius is as important as the test itself — and it takes one morning, not a separate test campaign.

What was delivered.

Specimen designFE sensitivity analysis to minimise κ(H) — notch geometry selected for parameter decoupling across all five JC parameters
Test programme3–4 specimens across quasi-static, intermediate (servo-hydraulic), and SHPB regimes — DIC acquisition at each
VFM identificationFull-field PVW cost function · sensitivity-based virtual fields · simultaneous 4-parameter isothermal identification + thermal softening m from separate high-T campaign
Uncertainty quantificationMonte Carlo propagation of DIC displacement noise (σ_u = 0.02 px) through H·p = b — 95% CI on all parameters
ValidationIndependent Charpy-type impact test at 5 m/s · FE simulation vs experimental force-time history · 4.2% peak force error
DeliverableJC material card for LS-DYNA / Abaqus / Nastran SOL 700 · parameter report with sensitivity maps · full identification traceability
Similar calibration studies we scope
  • Titanium alloys (Ti-6Al-4V) — aerostructure impact qualification
  • CFRP orthotropic identification via sensitivity-based VFM
  • High-entropy alloys — simultaneous multi-phase identification
  • Additive manufactured Inconel 718 — anisotropy quantification
  • Rate-dependent polymers — viscoelastic VFM identification
Why SIMFORGE for material calibration

VFM is not a bolt-on service — it requires simultaneous expertise in inverse methods, DIC acquisition, high-rate mechanics, and uncertainty quantification. SIMFORGE combines all four in a single engagement. The deliverable is not just a number — it is a parameter set with traceable confidence bounds, validated against an independent test, ready to drop into your simulation solver.

Scope a project

Material to characterise. Solver to feed.
Tell us the alloy and the regime.

Whether it's a nickel superalloy for bird-strike simulation, a titanium structure under ballistic loading, or a CFRP layup for crash analysis — if it needs a high-fidelity constitutive card with traceable uncertainty, we scope it in a 30-minute call and deliver a fixed-fee proposal within 48 hours.

Scope a calibration study Aerospace & Defence capabilities

Initial response · 24h  ·  NDA available  ·  Fixed-fee proposal within 48h