The Gold Standard of Rheology | Stress Relaxation After a Finite Loading Ramp

In an ideal strain-step test, strain jumps instantly from zero to γ₀ and then stays fixed. A real controlled ramp takes a finite time tᵣ to reach γ₀. A ramp-and-hold test therefore contains two distinct periods: deformation increases, then deformation is maintained.

Set t = 0 at the beginning of loading. During the linear ramp, γ(t) = γ₀t/tᵣ. At t = tᵣ, the hold begins. Define a second clock u = t − tᵣ for elapsed hold time, so u = 0 marks the end of the ramp.

Keep both clocks labelled. A result reported ten seconds after ramp start is not the same observation as ten seconds into the hold. Here γ is shear strain, σ is shear stress and τ is the material’s relaxation time, not a stress symbol.

Contents · Next chapter

A Maxwell element places an elastic spring and a viscous dashpot in series. They carry the same stress, while their strains add. Let the spring shear modulus be G and dashpot viscosity be η. The resulting linear equation is dσ/dt + σ/τ = G dγ/dt, with τ = η/G.

When strain is held, dγ/dt = 0 and stress decays exponentially. For an ideal instantaneous step, σ_step(u) = Gγ₀ exp(−u/τ), where u is then time since the step. That expression assumes the step itself took no time.

MIT’s Engineering Viscoelasticity notes develop spring–dashpot models and their limitations. We use the shear version of that linear framework to derive a finite-ramp teaching case. One relaxation time is not a universal description of real polymers, gels or suspensions.

Contents · Previous chapter · Next chapter

During the ramp the shear rate is constant, dγ/dt = γ₀/tᵣ. Starting from a relaxed state with σ(0) = 0, solve the linear equation to obtain σ(t) = Gγ₀(τ/tᵣ)[1 − exp(−t/τ)] for 0 ≤ t ≤ tᵣ.

At the end, σ_end = Gγ₀A, where A = (τ/tᵣ)[1 − exp(−tᵣ/τ)]. The spring does not carry the entire imposed strain: the dashpot has already deformed while loading was underway.

You can obtain the same result by adding the surviving contributions from small strain increments. An increment imposed early in the ramp has longer to relax than one imposed near the end. Integrating those differently aged contributions gives the factor A.

Contents · Previous chapter · Next chapter

Once γ stays fixed, the forcing term becomes zero. Using the end-of-ramp stress as the initial condition gives σ_hold(u) = Gγ₀A exp(−u/τ), for u ≥ 0.

For positive ramp duration, 0 < A < 1. At identical elapsed hold time, this stress is smaller than an ideal step to the same γ₀ at the start of that hold. That is a comparison between two loading histories, not evidence by itself of damage or nonlinearity.

There is also a useful interpretation for a general linear material. Stress divided by final strain after the ramp averages its relaxation modulus over elapsed ages from u to u + tᵣ. It therefore need not equal the ideal-step relaxation modulus at u. The exponential model lets us calculate that average explicitly.

Contents · Previous chapter · Next chapter

Take G = 1,000 Pa, τ = 10 s, γ₀ = 0.010 and tᵣ = 10 s. The dashpot viscosity is η = Gτ = 10,000 Pa·s. The ramp rate is 0.0010 s⁻¹.

An ideal step would start at Gγ₀ = 10.0 Pa. Our ramp has A = 1 − exp(−1) ≈ 0.63212, giving σ_end ≈ 6.3212 Pa. Ten seconds into the hold, stress is 6.3212 exp(−1) ≈ 2.3254 Pa.

For the ramp case, the hold begins at total time t = 10 s. The last row is therefore at t = 30 s after ramp start. Labelling that distinction prevents an apparent disagreement caused only by different time origins.

Contents · Previous chapter · Next chapter

Write x = tᵣ/τ, so A = (1 − exp(−x))/x. When x is small, A approaches 1, with leading approximation A ≈ 1 − x/2. When x is large, A approaches 1/x.

All rows reach strain 0.010 in the same model material. A slower ramp leaves less stress at the start of the hold because more relaxation has occurred during loading. The final strain alone does not specify the mechanical state.

Making a commanded ramp faster is not an unlimited remedy. Instrument dynamics, inertia and actual strain tracking can become relevant. The intended waveform and the waveform delivered to the specimen should be distinguished.

Contents · Previous chapter · Next chapter

If σ_end/γ₀ is mistaken for the instantaneous spring modulus, the ten-second ramp gives 632.12 Pa rather than 1,000 Pa. The apparent value is lower by about 36.8%. It is a ramp-dependent response ratio, not the model’s G.

However, for this exact single exponential, σ_hold(u)/σ_end = exp(−u/τ). Two positive tail values therefore give τ = (u₂ − u₁)/ln[σ(u₁)/σ(u₂)]. The common ramp factor cancels, so the ideal noiseless tail still identifies the correct τ.

A constant shift of time origin does not change an exponential slope when amplitude is freely fitted. Problems arise when the fit constrains the amplitude incorrectly or mixes clocks inconsistently. For example, forcing a 10 Pa step amplitude onto the 2.3254 Pa observation at u = 10 s gives an apparent τ of about 6.86 s instead of 10 s.

Contents · Previous chapter · Next chapter

For a generalised Maxwell model with an equilibrium spring G∞ and arms Gⱼ, τⱼ, each relaxing arm receives its own ramp factor Aⱼ = (τⱼ/tᵣ)[1 − exp(−tᵣ/τⱼ)]. During the hold, σ/γ₀ = G∞ + Σ GⱼAⱼ exp(−u/τⱼ).

The equilibrium spring’s contribution is G∞ because its stress depends on the final strain without decaying. Fast relaxing arms can be strongly suppressed during a slow ramp, while slower arms retain more of their initial contribution.

Normalising by the end stress no longer removes one universal factor. The ramp has changed the relative weights of the modes. A fitted single time constant may depend on the available time window and ramp, even though the underlying multi-mode material model is unchanged.

Contents · Previous chapter · Next chapter

Inspect measured strain and stress together. Record ramp shape, achieved duration, final strain, hold stability, sampling, temperature and sample preparation. A rounded corner or overshoot is a different input from the ideal piecewise linear history used here.

Verify an appropriate linear regime, measurement sensitivity, boundary contact and instrument response. Slip, evaporation, ageing or changing sample structure can invalidate the simple interpretation. A good exponential-looking tail does not by itself establish that the entire experiment met the model assumptions.

Use the rotational-rheometry guide for the measurement chain. The primary Abramowitch and Woo finite-ramp study addresses ramp effects in a specific ligament-analysis framework. It supports taking loading history seriously; its biological model and results should not be substituted for our single Maxwell calculation.

Contents · Previous chapter · Next chapter

Does a lower end stress mean the material softened?

Not necessarily. A longer ramp predicts a lower end stress in the same unchanged Maxwell element. Compare loading histories before attributing the difference to a material change.

Is stress divided by strain always a relaxation modulus?

For an ideal strain step in linear response, that ratio defines the relaxation modulus. After a finite ramp, it reflects the accumulated response to the whole ramp. Name the protocol rather than silently applying the step definition.

Can the earliest modulus be recovered from the tail alone?

Only within a justified model and with enough information about the ramp and relevant timescales. Fast modes may have largely relaxed before the hold. Dividing by a very small ramp factor can magnify measurement and parameter uncertainty rather than restore lost information reliably.

Contents · Previous chapter · Next chapter

Keep G = 1,000 Pa, γ₀ = 0.010 and tᵣ = 10 s, but change τ to 20 s. Now x = 0.5 and A = 2[1 − exp(−0.5)] ≈ 0.78694. The end stress is 7.8694 Pa; after ten seconds of hold it is about 4.7730 Pa.

The material with the longer relaxation time retains more stress through the same ramp. Predict that direction before calculating. If γ₀ then doubles within a verified linear model, every stress doubles while τ and A remain unchanged. A failure of that proportionality invites investigation of the model or measurement.

Finally, two hold stresses are 4.0 Pa and 2.0 Pa separated by ten seconds in a stipulated single-exponential case. The inferred relaxation time is 10/ln 2 ≈ 14.43 s. Those two values alone do not identify G without the final strain, ramp history and adequate support for the model.

Contents · Previous chapter · Next chapter

Quantities and experimental evidence

Return to the Rheology root for creep, flow and oscillation. Use rotational rheometry for torque-to-response interpretation, and rheology terminology and loading history for preserving protocol meaning.

Compare a different rheological problem

Yield-stress pipe flow and the moving plug studies spatially varying steady flow. Its yield model answers a different question from linear stress relaxation. Polymer chemistry supplies molecular context without making one Maxwell arm a complete molecular explanation.

Sources and learning transfer

The main mathematical reference is Roylance’s MIT Engineering Viscoelasticity notes. The finite-ramp ligament study offers a primary application with its own constitutive assumptions. Our numerical material, ramp comparisons and transfer exercises are constructed educational cases.

Continue through How X Works and the Gold Standard of Education. The useful habit is to carry the actual input history into the interpretation, rather than treating the final setting as the whole experiment.

Contents · Previous chapter · Return to Rheology