A reduced simulation applied to viscoelastic fatigue of polymers using a time multi-scale approach based on Partition of Unity method

Sebastian Rodriguez,Angelo Pasquale,Jad Mounayer,Diego Canales,Marianne Beringhier,Chady Ghnatios,Amine Ammar,Francisco Chinesta

2024-08-09

Abstract:The simulation of viscoelastic time-evolution problems described by a large number of internal variables and with a large spectrum of relaxation times requires high computational resources for their resolution. Furthermore, the internal variables evolution is described by a set of linear differential equations which involves many time scales. In this context, the use of a space-time PGD approximation is proposed here to boost their resolution, where the temporal functions are constructed following a multi-scale strategy along with the Partition of Unity method, in order to catch each dynamic efficiently. The feasibility and the robustness of the method are discussed in the case of a polymer in a non-equilibrium state under cyclic loading.

Computational Engineering, Finance, and Science

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: when simulating the time - evolution problem of viscoelastic materials (such as polymers) with a large number of internal variables and a wide range of relaxation time spectra, how to reduce the demand for computational resources. Specifically, the authors propose a time multi - scale approximation method based on the Partition of Unity method to improve the simulation efficiency of the viscoelastic fatigue behavior of polymers under cyclic loading in non - equilibrium states.

### Specific description of the problem

1. **High computational cost**: Traditional simulation methods require a large amount of computational resources to solve viscoelastic problems involving multiple time scales, especially when the number of internal variables is large and the range of relaxation times is wide.
2. **Multi - time - scale problem**: The evolution of internal variables is described by a set of linear differential equations, which involve many different time scales. This makes it very difficult to accurately capture all dynamic changes in a single time step.
3. **Viscoelastic behavior under cyclic loading**: For polymer materials, understanding their behavior under cyclic loading conditions is more challenging than for metal materials, because their creep effect under the mean stress will cause the cyclic behavior to evolve slowly and be affected by temperature coupling, especially under high - frequency excitation.

### Solutions

To address the above challenges, the authors propose the following solutions:

- **Space - time PGD (Proper Generalized Decomposition) method**: By separating time and space and using low - rank approximation to construct a simplified model of the problem, the computational complexity is reduced.
- **Time multi - scale approximation**: Introduce a time multi - scale strategy, combined with the Partition of Unity method, to efficiently capture the dynamic behaviors at different time scales. This method allows the use of coarse grids at the macro - scale and fine grids at the micro - scale, thereby significantly reducing the computational cost.
- **Handling transient behavior**: For the transient behavior that may occur in the function, combine the classical single - scale approximation and the multi - scale approximation to ensure that accurate results can also be obtained in the initial transient stage.

### Application examples

The paper verifies the effectiveness of the proposed method through numerical experiments, including:
- The case of a single internal variable
- The case of 50 internal variables

The results show that although in some cases the multi - scale approximation may require more iterations to converge, it can generally correctly approximate the reference solution and can significantly reduce the computational cost in a large time interval.

### Conclusions and prospects

This research shows that it is feasible to use PGD combined with the time multi - scale approximation method to predict the behavior of viscoelastic materials under high - cycle fatigue loads. Future work will further extend to three - dimensional problems and test more complex nonlinear viscoelastic behaviors. In addition, more abundant shape functions will be explored to reduce the number of required time multi - scale modes.

### Formula summary

- Evolution equation of internal variables:
\[
\frac{dz[j]}{dt}+\frac{1}{\tau_j}(z[j] - z[j]_{\infty}) = 0,\quad1\leq j\leq N
\]
where \(z[j]_{\infty}=E_{rj}^{\infty}\frac{\partial u}{\partial x}\), \(E_{rj}^{\infty}=p_jE_r\).

- Time multi - scale approximation:
\[
\lambda(t)\approx\sum_{k = 1}^{ms}\sum_{i = 1}^{n}N_i(t)q_k^iG_T(\tau(t - t_i))g_k
\]

Through these methods, the paper provides a new way to efficiently simulate the dynamic behavior of complex viscoelastic materials.