Creep and Stress Relaxation in Glass Interlayers
The Two Fundamental Experiments
There are two complementary ways to probe the time-dependent behaviour of a viscoelastic material using static (non-oscillatory) tests:
| Test | What is held constant | What is measured | Result |
|---|---|---|---|
| Stress relaxation | Strain ε0 (sudden step) | Stress σ(t) decaying over time | Relaxation modulus E(t) = σ(t)/ε0 |
| Creep | Stress σ0 (sudden step) | Strain ε(t) increasing over time | Creep compliance J(t) = ε(t)/σ0 |
Both tests contain the same fundamental information about the material — they are mathematically interconvertible. However, the practical considerations differ significantly.
Stress Relaxation
A specimen is rapidly deformed to a fixed strain ε0 and held. The stress is recorded as it decreases over time. The ratio σ(t)/ε0 is the relaxation modulus E(t).
At t = 0, the material responds with its instantaneous modulus E0. Over time, molecular chains rearrange and the stress decreases. The curve approaches an equilibrium modulus E∞ at long times.
For PVB interlayers, E0 can be 1000× larger than E∞. This enormous range is what makes viscoelastic characterisation essential — and why a single “stiffness value” is meaningless for design.
Creep
A constant stress σ0 is suddenly applied, and the strain ε(t) is recorded as it increases. The total creep strain has three components:
- εe: instantaneous elastic strain (recoverable, immediate)
- εv(t): time-dependent viscoelastic strain (partially recoverable)
- εp: permanent plastic strain (non-recoverable)
J(t) ≠ 1/E(t) — a common mistake. The creep compliance and relaxation modulus are related by the convolution integral ∫ G(τ)·J(t−τ) dτ = t, not by simple reciprocal. They approach equality only in the glassy and equilibrium limits.
Why Relaxation Is Preferred
While both tests are valid, stress relaxation is strongly preferred for interlayer characterisation:
| Criterion | Stress relaxation | Creep |
|---|---|---|
| Prony series | Direct — E(t) is the Prony series function | Requires numerical interconversion |
| FEM input | All major codes accept relaxation Prony coefficients | Not directly accepted |
| Numerical stability | Fitting decaying exponentials is stable | Fitting rising exponentials is less stable |
| Standards | EN 16613 references relaxation data | Less commonly referenced |
| Equipment | Requires servo-controlled DMA | Can use simpler dead-weight apparatus |
Experimental Details
Equipment
Centelles et al. (2021) used the RSA3 Dynamic Mechanical Analyser (TA Instruments) with a tension clamp for standalone interlayer film specimens. Galic et al. (2022) tested laminated glass beams (500 × 100 mm, 5+0.76+5 mm) in four-point bending using inductive displacement transducers.
Test protocol
- Determine the LVE range via amplitude sweep (ensure strain is small enough for linear response)
- Mount specimen at target temperature; allow thermal equilibration (10–15 minutes)
- Apply sudden step strain ε0 within LVE range
- Record stress σ(t) over 3–4 decades of log time (typically 0.1 s to 10,000 s)
- Repeat at each temperature: −10, 0, 10, 20, 25, 30, 35, 40, 50°C (extended to 80°C for SentryGlas)
All tests must stay within the LVE range. If the strain is too large, the material structure breaks down, the superposition principle fails, and the Prony series concept becomes invalid. Always perform an amplitude sweep first.
From Isothermal Curves to Master Curve
A single isothermal test covers only 3–4 decades of log time. The structural design range for glass spans from 3-second wind gusts to 50-year permanent loads — about 9 decades. The solution is time-temperature superposition (TTS).
By testing at multiple temperatures and shifting the isothermal curves horizontally on the log(t) axis, they collapse onto a single master curve at the reference temperature T0. The master curve typically spans 15–20+ decades — from nanoseconds to centuries.
This master curve is then fitted with a Prony series to obtain the coefficients used in structural design and FEM analysis.
Key Results for Interlayers
Isothermal behaviour by material type
PVB (standard, BG-R20): Extreme temperature sensitivity. E(t) at 1 s ranges from ~1000 MPa at −10°C to ~0.5 MPa at 50°C — a 2000× drop across 60°C. Strong time dependence at each temperature.
SentryGlas (ionomer): Moderate sensitivity up to 50°C. Still ~100–700 MPa across the full design temperature range. Only at 60–80°C does significant softening occur. Retains E∞ = 80.5 MPa long-term.
EVA (EVALAM, EVASAFE): Very low sensitivity. E(t) ≈ 2–5 MPa at all service temperatures. Already in the rubbery plateau (Tg < −10°C). Minimal time dependence — nearly elastic.
TPU: Similar to EVA but approximately 2× stiffer (~14 MPa). No stiffness drop observed in the tested range.
The time range gap
The gap between laboratory time scales and design time scales is enormous:
| Design scenario | Duration | log(t) [s] |
|---|---|---|
| Wind gust | 3 s | 0.5 |
| Barrier load | 30 s | 1.5 |
| Snow (short-term) | 3 weeks | 6.3 |
| Permanent / dead load | 50 years | 9.2 |
| Span: nearly 9 decades — impossible to cover in a single test | ||
This is why TTS and the master curve approach are essential. Without them, extrapolation from short-term tests to 50-year design loads is unreliable.
Explore the Relaxation Data
See how E(t) varies with temperature and time for all interlayer types in our database. Select a material, choose a temperature, and get the full relaxation curve.
Launch Prony Calculator