L.C. Berselli,D. Wells,X. Xie,T. Iliescu
Abstract:Spatial filtering has been central in the development of large eddy simulation reduced order models (LES-ROMs) and regularized reduced order models (Reg-ROMs), In this paper, we perform a numerical investigation of spatial filtering. To this end, we consider one of the simplest Reg-ROMs, the Leray ROM (L-ROM), which uses ROM spatial filtering to smooth the flow variables and decreases the amount of energy aliased to the lower index ROM basis functions. We also propose a new form of ROM differential filter and use it as a spatial filter for the L-ROM. We investigate the performance of this new form of ROM differential filter in the numerical simulation of a flow past a circular cylinder at a Reynolds number $Re=760$.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: How to improve reduced - order models (ROMs), especially regularized reduced - order models (Reg - ROMs), through spatial filtering techniques in order to carry out numerical simulations of convection - dominated fluid flows more efficiently and relatively accurately.
Specifically, the authors focus on:
1. **Application of spatial filtering in large - eddy simulation (LES) and regularized reduced - order models**: Spatial filtering plays a crucial role in these models. It can help smooth flow - field variables, reduce energy aliasing phenomena, and improve the accuracy of numerical simulations.
2. **Proposing a new ROM differential filter**: This new form of filter is used for the Leray reduced - order model (L - ROM), and its performance in the problem of flow around a cylinder is tested.
3. **Verifying the performance of the new filter**: Evaluate the performance of the newly proposed ROM differential filter under different parameter settings through numerical experiments, especially its application in the problem of flow around a cylinder at a Reynolds number of 760.
### Main research contents
1. **Construction of reduced - order models (ROM)**:
- Select modes {⃗ ϕ₁,..., ⃗ ϕᵈ} to represent the repetitive spatial structures of a given flow field.
- Select the dominant modes {⃗ ϕ₁,..., ⃗ ϕʳ} as the basis functions of the ROM.
- Use Galerkin truncation ⃗ uʳ = ∑ⱼᵃⱼ⃗ ϕⱼ.
- Replace ⃗ u with ⃗ uʳ and project it onto the ROM space Xʳ := span{⃗ ϕ₁,..., ⃗ ϕʳ}, obtaining a low - dimensional dynamic system:
\[
\dot{\vec{a}} = A\vec{a} + \vec{a}^\top B \vec{a}
\]
where \(\vec{a}\) is the unknown ROM coefficient vector, and \(A\) and \(B\) are ROM operators.
2. **ROM differential filter**:
- Based on the classical Helmholtz filter, define the filtered flow field \(F(\vec{u}_r)\) to satisfy:
\[
(I - \delta^2 \Delta) F(\vec{u}_r), \vec{v}) = (\vec{u}_r, \vec{v}), \quad \forall \vec{v} \in X_f
\]
- Propose two different versions of the ROM differential filter: the finite - element version (FE Version) and the ROM version (ROM Version).
3. **Leray reduced - order model (L - ROM)**:
- The Leray model stabilizes the numerical solution by introducing a filtering operation to handle nonlinear terms. The L - ROM equation is:
\[
(\vec{w}_r)_t = \frac{1}{Re} \Delta \vec{w}_r - F(\vec{w}_r) \cdot \nabla \vec{w}_r - \nabla p
\]
- Approximate the convolution operation through the differential filter:
\[
F(\vec{w}) = (\delta^2 \Delta + 1)^{-1} \vec{w}
\]
4. **Numerical results**:
- The newly proposed ROM differential filter was tested in the problem of flow around a cylinder at a Reynolds number of 760. The results show that the new filter can effectively smooth the flow field, reduce kinetic - energy aliasing phenomena, and has a low computational cost.
### Conclusion
This paper proposes a new ROM differential filter and applies it to the L - ROM, achieving encouraging numerical results. The new filter significantly improves the accuracy of numerical simulations while maintaining computational efficiency. Future research will further explore the application of the new filter in more complex flow problems and make a detailed comparison with standard filters.