Green's Function Integral method for Pressure Reconstruction from Measured Pressure Gradient and the Interpretation of Omnidirectional Integration

Qi Wang, Xiaofeng LiuarXiv 2023 · DOI · arXiv

In the world of fluid mechanics, pressure is a "hidden" variable. While we can easily film particles moving in a flow using Particle Image Velocimetry (PIV) to get velocity, measuring the instantaneous pressure field is notoriously difficult. Researchers usually calculate the pressure gradient from velocity data and then try to "reconstruct" the pressure field.

A new paper by Qi Wang and Xiaofeng Liu from San Diego State University introduces the Green’s Function Integral (GFI) method, a mathematical shortcut that makes this reconstruction significantly faster and more robust, especially for complex 3D flows.

TL;DR

The GFI method replaces computationally expensive line-integration techniques (like ODI) with a single convolution based on Green’s functions. It provides a mathematically equivalent but much faster way to reconstruct pressure fields from noisy data, achieving a 14x speedup in 2D while handling complex 3D geometries with ease.

What Problem Does This Paper Address?

To get pressure from PIV, researchers use the Navier-Stokes equations to determine the pressure gradient ($\nabla p$). However, PIV data is inherently noisy. If you simply integrate the gradient along a single path, the errors accumulate.

The current gold standard is Omnidirectional Integration (ODI), which averages thousands of "zigzag" integration paths from all directions to cancel out noise. While accurate, ODI is a computational nightmare in 3D, often requiring high-end GPUs to be practical. The challenge was to find a method that retains ODI's noise-canceling "magic" without the heavy lifting of calculating infinite line integrals.

Methods

The authors propose the Green’s Function Integral (GFI) method. Instead of drawing lines through the data, they treat the pressure reconstruction as a boundary value problem solved via Green’s identities.

  1. The Convolution Kernel: They use the Green’s function of the Laplacian operator as a convolution kernel. This relates the pressure at any point directly to the pressure gradient across the entire domain.
  2. Boundary Compatibility: A major hurdle in pressure reconstruction is the boundary. GFI introduces a compatibility condition—a set of linear equations—that solves for the pressure on the boundaries first, which then informs the rest of the field.
  3. Mathematical Equivalence: The authors prove that GFI is the mathematical limit of ODI as the number of integration paths goes to infinity. Essentially, GFI is what ODI is trying to be, but expressed as a clean, continuous integral.

Datasets

The researchers validated GFI using two primary sources:

  • Johns Hopkins Turbulence Database (JHTDB): They used data from forced isotropic turbulence to test the method in a 2D plane (254 x 254 grid) with 40% added noise.
  • Synthetic 3D Multi-connected Domain: To test complex geometry, they created a 3D cube with a spherical "void" in the center (simulating a bubble or a solid object) and reconstructed the pressure from turbulent flow data mapped onto an unstructured tetrahedral mesh.

Key Insights and Ideas

  • The "Aha" Moment on Denoising: Through spectral analysis (SVD), the authors showed that the GFI operator acts as a natural low-pass filter. It attenuates high-frequency noise in the pressure gradient, which is why it performs so well on "dirty" experimental data.
  • Geometry Independence: Because GFI doesn't rely on "rays" or "paths," it doesn't care if your domain is a simple square or a complex 3D space with holes (multi-connected domains).
  • Power Law of Error: They discovered a non-dimensionalized power law ($\kappa \approx 0.89$) that describes how the error in reconstructed pressure decreases as the grid resolution increases.

Key Results

  • Speed: In a 2D test case, GFI was 14 times faster than the state-of-the-art ODI method (4.3 seconds vs. 59.6 seconds).
  • Accuracy: Despite the speedup, GFI’s accuracy was nearly identical to ODI. In 2D isotropic turbulence with heavy noise, the relative difference between the two was within 10% of the pressure fluctuation's standard deviation.
  • 3D Capability: The method successfully reconstructed pressure in a 3D domain with a central void, showing only a 2% relative difference from the ground truth.

Limitations and Future Work

The authors noted that when using unstructured tetrahedral meshes in 3D, the current version of the algorithm produced some "spurious behavior" because it lacks higher-order interpolation.

Future directions include: * Implementing higher-order interpolation schemes to clean up results on unstructured meshes. * Exploring multi-grid formulations to further boost computational efficiency for massive 3D datasets. * Applying the method to real-world experimental PIV data, such as a rising bubble in a liquid, where the multi-connected domain capability would be a game-changer.

Summary updated

Abstract

Accurately and efficiently measuring the pressure field is of paramount importance in many fluid mechanics applications. The pressure gradient field of a fluid flow can be determined from the balance of the momentum equation based on the particle image velocimetry (PIV) measurement of the flow kinematics, which renders the experimental evaluation of the material acceleration and the viscous stress terms possible. We present a novel method of reconstructing the pressure fields from error-embedded pressure gradient measurement data. This method utilized Green's function of the Laplacian operator as the convolution kernel that relates pressure to pressure gradient. A compatibility condition on the boundary offers equations to solve for the boundary pressure. This Green's function integral (GFI) method has a deep mathematical connection with the state-of-the-art omnidirectional integration (ODI). As mathematically equivalent to ODI in the limit of an infinite number of line integral paths, GFI spares the necessity of line integration along zigzag integral paths, rendering generalized implementation schemes for both two and three-dimensional problems with arbitrary geometry while bringing in improved computational efficiency. In the current work, we apply GFI to pressure reconstruction of simple canonical and isotropic turbulence flows embedded with error in two-dimensional and three-dimensional domains, respectively. Uncertainty quantification is performed by eigenanalysis of the GFI operator in a square domain. The accuracy and computational efficiency of GFI are evaluated and compared with ODI.
BibTeX
@misc{wang2023green,
  author = {Qi Wang and Xiaofeng Liu},
  title = {{Green's Function Integral method for Pressure Reconstruction from Measured Pressure Gradient and the Interpretation of Omnidirectional Integration}},
  year = {2023},
  eprint = {2304.05726},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2304.05726},
}

Loading related content...

Figure