A viscoelastic Rivlin-Ericksen material model applicable to glacier ice

P. Riesen, Kolumban Hutter, Mathias FunkNonlinear 2010 · DOI

Abstract

We present a viscoelastic constitutive relation which describes transient creep of a modified second grade fluid enhanced with elastic properties of a solid. The material law describes a Rivlin-Ericksen material and is a generalization of existing material laws applied to study the viscoelastic properties of ice. The intention is to provide a formulation tailored to reproduce the viscoelastic behaviour of ice ranging from the instantaneous elastic response, to recoverable deformation, to viscous, stationary flow at the characteristic minimum creep rate associated with the deformation of polycrystalline ice. We numerically solve the problem of a slab of material shearing down a uniformly inclined plate. The equations are made dimensionless in a form in which elastic effects and/or the influence of higher order terms (i.e., strain accelerations) can be compared with viscous creep at the minimum creep rate by means of two dimensionless parameters. We discuss the resulting material behaviour and the features exhibited at different parameter combinations. Also, a viable range of the non-dimensional parameters is estimated in the scale analysis.

Introduction

For decades, the "gold standard" for modeling glacier flow has been Glen’s flow law. While Glen’s law is excellent at predicting how ice flows over long periods (stationary creep), it treats ice purely as a non-Newtonian fluid. However, glacier ice isn't just a simple fluid. Laboratory experiments and recent field observations—such as the sudden reversal of flow direction observed at the Gornergletscher in Switzerland—suggest that ice has "memory" and elastic properties. It can store energy like a spring and release it when loads change. Standard models cannot explain these "bounces" or the initial rapid deformation (primary creep) seen when a load is first applied. This paper aims to bridge the gap between the fluid-like behavior of ice on geological timescales and its solid-like behavior on hourly or daily timescales.

Methods

The authors propose a new constitutive relation called the Elastic Modified Second Order Isotropic Material (EMSOIM) model. This model is built within the framework of Rivlin-Ericksen materials, which relate stress not just to the current rate of deformation, but also to the acceleration of that deformation and the total strain.

Key components of the method include:

  • MSOFM Integration: It incorporates the Modified Second Order Fluid Model, which accounts for "normal stress effects" (stresses acting perpendicular to the flow) and strain accelerations.
  • Fading Elasticity: A novel addition is a variable shear modulus that depends on the accumulated strain. This "fading elasticity factor" allows the material to act like a solid initially but "forget" its elastic state as it deforms, eventually transitioning into a purely viscous fluid.
  • Dimensionless Analysis: The complex equations are reduced into two primary dimensionless parameters: $\mathbb{H}$ (representing the significance of strain accelerations) and $\mathbb{K}$ (representing initial stiffness/rigidity).
Schematic of the slab-on-a-slope benchmark problem used to test the EMSOIM model.

The model was tested using a benchmark "slab-on-a-slope" problem (illustrated above), solved via the finite element method using established physical constants for ice, such as Glen’s Law parameters and shear modulus values from existing literature.

Results and Discussion

The EMSOIM model successfully reproduced three distinct phases of behavior in a single framework: an instantaneous elastic response, primary/transient creep, and stationary secondary creep. Unlike Glen’s law, the model shows that when a load is removed, the material undergoes partial or total recovery of its deformation, mirroring the flow reversals seen in the field.

Numerical results showing the influence of dimensionless parameters H and K on velocity/creep rate and displacement over time.

As shown in the figure above, the parameters $\mathbb{H}$ and $\mathbb{K}$ significantly dictate the material's response:

  • The Role of Acceleration ($\mathbb{H}$): Strain acceleration terms are essential to capture "primary creep"—the initial phase where ice deforms rapidly before settling into a steady flow (Panels a and b).
  • The Role of Stiffness ($\mathbb{K}$): Higher rigidity leads to lower overall displacement and can influence the transition to steady state (Panels c and d).
  • Oscillatory Behavior: If the material is very rigid (high $\mathbb{K}$) and acceleration effects are significant (high $\mathbb{H}$), the creep rate can actually oscillate before reaching a steady state.

Through scale analysis, the authors estimated viable ranges for real-world glacier conditions: $\mathbb{H}$ typically falls between 1.5 and 115, while $\mathbb{K}$ ranges from $2.9 \times 10^3$ to $2 \times 10^4$.

Conclusion and Future Work

The EMSOIM model provides a robust mathematical framework for capturing the complex, short-term viscoelastic behaviors of glacier ice. While successful in 1D simulations, future work is required to implement this model in 3D finite element codes, which will be computationally demanding. Additionally, the current model does not yet account for "tertiary creep," where ice flow accelerates again due to damage or recrystallization.

Summary updated

Abstract

We present a viscoelastic constitutive relation which describes transient creep of a modified second grade fluid enhanced with elastic properties of a solid. The material law describes a Rivlin-Ericksen material and is a generalization of existing material laws applied to study the viscoelastic properties of ice. The intention is to provide a formulation tailored to reproduce the viscoelastic behaviour of ice ranging from the instantaneous elastic response, to recoverable deformation, to viscous, stationary flow at the characteristic minimum creep rate associated with the deformation of polycrystalline ice. We numerically solve the problem of a slab of material shearing down a uniformly inclined plate. The equations are made dimensionless in a form in which elastic effects and/or the influence of higher order terms (i.e., strain accelerations) can be compared with viscous creep at the minimum creep rate by means of two dimensionless parameters. We discuss the resulting material behaviour and the features exhibited at different parameter combinations. Also, a viable range of the non-dimensional parameters is estimated in the scale analysis.
BibTeX
@inproceedings{riesen2010viscoelastic,
  author = {P. Riesen and Kolumban Hutter and Mathias Funk},
  title = {{A viscoelastic Rivlin-Ericksen material model applicable to glacier ice}},
  booktitle = {Nonlinear},
  year = {2010},
  doi = {10.5194/npg-17-673-2010},
}
Figure