Learn · Time-fractional operators

When the past still affects the present.

Classical models describe the instantaneous rate of change of a field quantity. Time-fractional models add a fading memory of what happened in the past—helping simulations represent hereditary materials, anomalous heat flow, history-dependent thermal stress, and many other non-classical processes.

Memory kernel
Every prior event contributes
PRESENTOLDER HISTORYNOW
α = 1local-in-time limit
α < 1history-weighted response
1 < α ≤ 2diffusion–wave response
The fractional order as a modeling dial

One order parameter continuously connects different rates of change.

A time-fractional derivative generalizes integer differentiation to noninteger order. Its specific physical meaning comes from the governing equation in which the operator is used.

Fractional operator (Caputo definition)
D0αf(t)=1Γ(1α)0t1(tτ)αdf(τ)dτ
Mathematical meaning

The operator is nonlocal in time: it combines rates from every earlier instant through a power-law kernel. The order α controls how those past rates are weighted. The displayed form applies for 0 < α < 1; for 1 < α < 2, the Caputo definition uses the second derivative and requires a second initial condition.

Physical meaning in applications

The operator alone does not prescribe a physical mechanism. Its interpretation depends on its role in a mathematical model. Its integration in a constitutive law can express material memory, while in a balance equation can describe anomalous transport or a history-dependent coupled response. The order and accompanying coefficients must therefore be tied to data, units, and the physics of that application.

Apply the operator

Select a test function and move the α slider to see effect on the derivative.

f(t)=t2,D0αt2=Γ(3)Γ(3α)t2α
Fractional derivatives of t squaredCurves show Caputo derivatives for orders zero, one half, one, one and one half, and two, with the selected derivative order highlighted.01200.51normalized time, tderivative value
Reference ordersα = 0α = 0.5α = 1α = 1.5α = 2selected α
0 · function1 · first derivative2 · second derivative

For α = 0.75, the curve is approximately 1.765t1.25. Time is nondimensional so the orders can be compared on one plot.

One formulation, a wider modeling range

Fractional operators provide a continuous extension of classical models across broader dynamical regimes.

The integer-order case remains a built-in limit, while the order α opens a continuous path toward memory, anomalous transport, and higher-order transient behavior. A native fractional formulation can therefore move between regimes within one model family, with the coefficients and order tied to the application and its evidence.

Application examples
01 · Viscoelasticity

Creep and relaxation remember the loading path.

Polymers and soft materials do not respond through a single elastic stiffness or viscous time constant. Their present deformation and stress depend on how the system evolved.

Application fit
PolymersAdhesivesCoatingsDampingSoft tissueSeals
What fractional modeling enables

In CAE, a fractional constitutive law can capture broad time-scale behavior with fewer branches and parameters than a large Prony series—simplifying calibration and reducing the cost of long transient analyses.

Fractional modelClassical model · α = 1
Constant stress
history-dependent response
lowhigh

Current normalized strain: 0.55

normalized timestrain
creep comparison between fractional and classical normalized viscoelastic models
0.62
longer memoryclassical limit
t = 0.08

Normalized fractional standard-linear-solid response. Curves are educational and use a guarded Mittag–Leffler approximation, not fitted material data.

02 · Anomalous heat diffusion

Heat does not always spread on the Fourier clock.

Heterogeneous, porous, layered, and microstructured media can trap or delay energy transfer. A history-aware transport law can capture a thermal pulse whose evolution does not follow classical (Fourier) diffusion.

Application fit
BatteriesPorous mediaInsulationBiological mediaCompositesThermal protection
What fractional modeling enables

In CAE, one fractional heat-transfer model can span delayed diffusion, Fourier behavior, and wave-like transport—improving transient temperature and thermal-load predictions when experiments show non-Fourier response.

Localized heat pulse · 1D domain
Fractional modelClassical model · α = 1
Subdiffusive regime0 < α < 1fractional frontα = 1 Fourier frontlocalized energy spreads slowly and remains source-centered
Diffusion–wave regime1 < α < 2fractional frontα = 2 wave-limit frontfronts propagate toward the boundaries
Conceptual comparison of a slowly spreading subdiffusive temperature field and propagating diffusion-wave fronts that reflect from the slab boundaries.
Temperature profileDiffusion–wave · α = 1.35
distance through domaintemperature
Shaded area: departure from the Fourier prediction at the selected time.
1.35 · diffusion–wave
subdiffusiveFourier · α = 1thermal-wave limit
t = 0.18

Normalized 1D modal field from an initially localized pulse with fixed-temperature ends. The reduced-order response spans subdiffusion, Fourier diffusion, and the diffusion–wave range; its α → 2 endpoint visualizes the hyperbolic thermal-wave limit associated with Cattaneo-type behavior, not the full mixed-derivative Cattaneo equation. Note that this is a notional schematic and not a thermal design calculation.

03 · Fractional thermoelasticity

Changing the thermal history changes the stress history.

Temperature, expansion, constraint, and stress are coupled. When thermal transport carries memory, the timing and distribution of thermoelastic response can change as well.

Application fit
Thermal shockLayered structuresElectronicsCoatingsCracked componentsThermal cycling
What fractional modeling enables

Fractional thermoelasticity links thermal history directly to mechanical response, improving predictions of stress timing, peak loads, and residual deformation in thermally cycled components.

Thermal shock · coupled response
Boundary condition
Fractional formulation
α = 1.30
memory-aware
TEMPERATURE FIELDTHERMAL STRESS FIELDshocked facefixed end
Classical formulation
α = 1.00
integer limit
TEMPERATURE FIELDTHERMAL STRESS FIELDshocked facefixed end
Peak stress
0.56fractional0.64classical
Time to peak
0.20fractional0.20classical
Late-time stress
0.35fractional0.35classical
Peak thermal-stress history
Fractional modelClassical model · α = 1
normalized timepeak stress
Peak thermal-stress histories for fractional and classical one-sided thermal shock
1.30 · diffusion–wave
subdiffusiveFourier · α = 1thermal-wave limit
t = 0.12

Normalized reduced-order comparison of one-sided heat penetration, linear thermal expansion, and support-dependent stress. Fixed–fixed retains a uniform constraint contribution; fixed–free emphasizes transient gradient stress. Values are illustrative, not design predictions.

What this comparison shows. The applied shock, geometry, and supports are identical; only the time operator changes. Fractional heat transport changes how quickly the thermal front penetrates the component and therefore the magnitude, timing, and persistence of thermal stress. At α = 1, both formulations converge.
04 · Anomalous transport in porous media

Transport inherits the medium’s hidden time scales.

Flow and wave measurements in heterogeneous media often show delayed arrival, broad transitions, attenuation, and long tails. A time-fractional approach can represent the cumulative effect of trapping, tortuosity, and distributed relaxation without resolving every microscopic mechanism.

Application fit
GroundwaterReservoirsFiltrationGeophysicsAcoustic dampingPorous absorbers
What fractional modeling enables

In CAE, a fractional transport law can reproduce delayed breakthrough, dispersion, and persistent attenuation within an effective continuum—reducing the need to resolve every pore or introduce many discrete relaxation scales.

History-dependent transport
Fractional formulationα = 0.65
virtual sensorsourceheterogeneous medium
Integer-order referenceα = 1.00
virtual sensorsourceheterogeneous medium
Signal at the virtual sensor

Pressure arrival and breakthrough broadening

Fractional modelInteger-order reference · α = 1
normalized timepressure
Pressure signal at a sensor in a heterogeneous medium
0.65
stronger retentioninteger limit · 1
t = 0.08

Normalized conceptual comparison with fixed source, medium, and sensor location. The Darcy view represents memory in storage or resistance; the acoustic view represents distributed attenuation and dispersion. Values are illustrative, not design predictions.

What this comparison shows. A time-fractional operator turns unresolved trapping, tortuosity, and distributed relaxation into a measurable shift in arrival time, spreading, and persistence. The integer-order Darcy or wave models remain the recovered limit.