High-order immersed boundary method for subsonic aeroacoustics and compressible Navier-Stokes equations

summary

The gist

A new high-order immersed boundary method for subsonic aeroacoustics and compressible Navier-Stokes equations is proposed…

In short

A new high-order immersed boundary method was developed for subsonic aeroacoustics and compressible Navier-Stokes equations. It uses a stencil of points along the boundary normal and 4th-order Lagrange polynomial interpolation to reconstruct flow variables at ghost points. This approach successfully handles cut-cell issues and has been validated for both acoustic wave scattering and viscous compressible flows.

Key concepts

Immersed Boundary Method (IBM)
A numerical technique used to solve fluid dynamics problems where the computational grid does not align perfectly with the physical boundary of an object. It involves placing a discrete set of points on or near the boundary to impose boundary conditions, allowing the simulation to proceed in a fixed Cartesian grid.
Ghost Point Method
A two-step process used within IBMs to determine fluid flow values at points that lie outside the main computational domain (ghost points). First, a least-squares interpolation reconstructs the flow on a stencil near the boundary, and then a high-order Lagrangian extrapolation is used to impose boundary conditions.
Fourth-Order Accuracy
This refers to the spatial accuracy of the numerical scheme. The method is designed to achieve fourth-order accuracy, meaning that errors in approximating variables across space decrease proportionally to the fourth power of the grid spacing, leading to highly precise results for wave scattering and flow simulations.
Cut-Cell Problem
A geometric challenge in IBMs where the immersed boundary intersects the Cartesian grid such that image points fall into cells that are only partially filled with fluid or are too close to the boundary. The proposed method addresses this by using a specific stencil of points along the normal to better handle these intersections.

Terminology used across episodes

This episode discusses

The paper

High-order immersed boundary method for subsonic aeroacoustics and compressible Navier-Stokes equations · Read on arXiv

Institut PPRIME, CNRS-University of Poitiers-ENSMA

Transcript

Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.

Kai: We're looking at the paper: "High-order immersed boundary method for subsonic aeroacoustics and compressible Navier-Stokes equations".

Mira: Here's the gist of the paper. A new high-order immersed boundary method for subsonic aeroacoustics and compressible Navier-Stokes equations is proposed, which has been validated for both acoustic wave scattering and compressible flow past a cylinder The proposed approach relies on a stencil of points aligned along the normal to the immersed boundary, on which a high-order (here fourthorder) Lagrange polynomial interpolation is applied to reconstruct the variables at ghost points The method is based on a discrete ghost point method that determines flow field values at ghost points in two steps A least-squares interpolation is first used to reconstruct the flow on a stencil of points located on the normal to the immersed boundary Then, a 4th order Lagrangian extrapolation based on the normal stencil points adjacent to the boundary point is used to impose the boundary conditions at the immersed boundary by reconstructing the field values at the ghost points This methodology preserves the global spatial accuracy of the underlying finite-difference scheme (a fourthorder Dispersion-Relation-Preserving, DRP, scheme) while effectively handling cut-cell issues Handling Geometric Challenges The classical image point method faces the cut-cell problem when the immersed boundary intersects the Cartesian grid in such a way that the image point falls within a cell that is only partially filled with fluid or is located too close to the boundary To address this, existing approaches often resort to geometric corrections, such as shifting the image point farther into the fluid by introducing a length correction or including the boundary point itself in the interpolation stencil The proposed high-order IBM utilizes a stencil of k + 1 points P0, …, Pk distributed along the normal probe to obtain variables at ghost points using a k-th order Lagrangian interpolation For k = 4, this corresponds to 5 points, matching the desired order of the interpolation of the DRP scheme used in this work Validation and Accuracy The method was first validated on an academic problem of acoustic wave scattering by a circular cylinder using the linearized Euler equations A convergence study demonstrates its expected spatial order of accuracy Second, its robustness is assessed for viscous compressible flows governed by the Navier-Stokes equations around a cylinder at various Reynolds numbers, in both steady and unsteady regimes The results show excellent agreement with reference data from the literature For acoustic wave scattering, a convergence study confirms that the present method is 4th order accurate in space Application to Compressible Flow The method was applied to simulate a subsonic compressible flow past a circular cylinder for Reynolds numbers Re = 40 and Re = 150 For the steady flow at Re = 40, the results show good agreement with literature data, except for the drag coefficient For the compressible flow at Ma = 0.5, a clear Mach number dependence is observed in the results The pressure coefficient Cp increases upstream with increasing Mach number and decreases downstream as the Mach number increases Aeroacoustic Noise Generation The method was also applied to aeroacoustic noise generation by a separated flow, highlighting the method’s ability to capture the radiated acoustic field, including its characteristic dipole directivity The time-averaged root mean square (rms) of the fluctuating pressure, prms, is evaluated over a set of 180 points distributed on a circle of radius r = 75D to determine the acoustic radiation pattern The directivity curve is in close agreement with that reported by Inoue and Hatakeyama The characteristic profile of a dipole, oriented perpendicular to the flow axis and aligned along the y = 0 axis, is clearly observed in the directivity plot A new high-order immersed boundary approach for computational acoustic and low Mach compressible has been proposed The method has been employed for the scattering of an acoustic pulse by a circular cylinder and a convergence study on the reflected wave has demonstrated the 4th order accuracy in space of this method The flow at low Mach numbers past a circular cylinder for Reynolds numbers Re = 40 and Re = 150 has also been investigated, and a good agreement is found with the literature, demonstrating the validity of the numerical method The Strouhal number, as well as the fluctuating lift and drag coefficients, are in good agreement with the values reported by Inoue and Hatakeyama The acoustic pressure field decay along the axis y > 0 is characterized by a reduction in amplitude of pressure fluctuations occurring in two distinct regimes The decay trend following 1/r1/2 is clearly visible in the far field, compared with data from Inoue and Hatakeyama The directivity of the cylinder’s acoustic radiation is shown in Fig.

Kai: (14b) and is in close agreement with that reported by Inoue and Hatakeyama The influence of reflections from the domain boundaries, which contaminate the calculation, is visible on the directivity curve References 1.

Mira: C.

Kai: S.

Paper summary: Kai: So, to recap, this paper introduces the "High-order immersed boundary method for subsonic aeroacoustics and compressible Navier-Stokes equations." The core thesis is that they've created a new discrete ghost point method that uses high-order schemes to solve problems involving both acoustics and compressible flow when using Cartesian grids.

Mira: They claim this method works by first using a least-squares interpolation on points along the normal to the immersed boundary, then employing a fourth order Lagrangian extrapolation based on adjacent normal stencil points to impose those boundary conditions at ghost points.

Lev: The reason they put that fourth order Lagrangian extrapolation in there is to ensure they keep the global spatial accuracy of their underlying finite difference scheme, which they state is a fourth order dispersion-relation-preserving, or DRP, scheme.

Kai: And this addresses the cut-cell problem you hear about with image point methods by using a stencil of k plus one points along the normal probe to reconstruct variables at ghost points using a k minus order Lagrangian interpolation. For them, they set k equal to four, meaning five points for the interpolation, which matches the fourth order accuracy they need.

Mira: So it’s not just about being high order; it's specifically about structuring that stencil in a way that solves those geometric reconstruction challenges inherent in immersed boundary methods when the grid intersects the boundary.

Lev: It means they are trying to achieve high-order accuracy without sacrificing stability or introducing numerical dispersion issues, which is usually the trade-off you have to make when you move beyond second order ghost point approaches.

Kai: And this has been validated on acoustic wave scattering with linearized Euler equations, showing a convergence study that confirms it's four order accurate in space.

Mira: But the real test was applying it to viscous compressible flows governed by the Navier-Stokes equations around a cylinder at various Reynolds numbers, and the results showed excellent agreement with reference data from the literature.

Lev: They tested this for both steady and unsteady regimes, which shows it’s not just a toy problem solution but something that actually holds up under more realistic flow conditions.

Kai: And when they simulate subsonic compressible flow past a circular cylinder at re equal forty and re equal one hundred fifty, the results show good agreement with literature data, except for the drag coefficient at re equal forty.

Mira: And for that compressible flow at mach number zero point five, there's a clear Mach number dependence in the results. The pressure coefficient up stream increases with increasing mach number and then decreases down stream as the mach number increases.

Lev: That Mach number dependence is important because it shows how much you need to model that compressibility correctly for accurate predictions. And I gotta ask, what about the limitations? What's not covered by this method?

Kai: The authors mentioned a couple of things: they noted that reconstructions can degrade stability and accuracy, which is why they suggested completing the method with WENO schemes or another solution, or reducing the order of interpolation or using a hybrid approach.

Mira: So they aren't claiming perfection; they are suggesting that if you want to push beyond this method's performance even further, you might need to use something like a weighted essentially non-oscillatory scheme.

Lev: And from an engineering standpoint, if we were building real hardware for this, the overhead of implementing these high order reconstructions would land somewhere in the computational cost and memory usage.

Kai: But for now, it seems like this method is a very solid tool for subsonic aeroacoustics and low Mach compressible flows that needs to be used.

Conclusion: Mira: So looking at the paper "High-order immersed boundary method for subsonic aeroacoustics and compressible Navier-Stokes equations," the title itself really sums up what they did—a high order approach for both subsonic aeroacoustics and compressible Navier-Stokes equations.

Kai: And I think what's important to remember is that they’ve successfully developed a discrete ghost point method that handles complex geometries without sacrificing accuracy when you need it.

Lev: It seems like the main implication is that for engineers, this could mean reliable simulation of flows around cylinders where you have those tricky grids, especially if you need high fidelity data on noise generation.

Mira: Exactly. We're talking about being able to get a good handle on how noise radiates from those objects in ways that match experimental data, like the directivity patterns they got for the cylinder’s acoustic radiation.

Kai: So it’s a method that has shown itself to be reliable when pushed into viscous flows, and it gives us better accuracy than some existing methods.

Lev: It demonstrates that when you are dealing with these complex geometries in CFD, you can achieve higher order accuracy without introducing massive instability right away, which is what we need for real simulations.

More episodes

← Home