Efficient Time-Resolved Pressure Estimation by Sparse Sensor Optimization and Non-Time-Resolved PIV

TL;DR

This paper compares two methods for estimating time-resolved pressure fields from non-time-resolved PIV data, demonstrating that the second approach is significantly faster and more efficient, especially at higher temporal resolutions.

Contribution

The study introduces and compares two approaches for pressure estimation from non-time-resolved PIV data, highlighting the computational efficiency of directly estimating pressure fields first.

Findings

Approach two is approximately thirty times faster than approach one at higher time resolutions.

Estimating pressure fields directly from non-time-resolved data reduces computational load.

Both approaches are validated using flow over a cylinder with down-sampled PIV data.

Abstract

Pressure field estimation from PIV data has been a well-established technique. However, time-resolved pressure estimation strongly depends on the temporal resolution of the PIV measurements. Generally, PIV data has limited time resolution creating challenges to understand high Reynolds number flows. To overcome this challenge, sensor data measured at few optimized locations with higher time resolution is combined with PIV data using data driven methods to reconstruct time resolved velocity fields. In this context, if we wish to estimate time resolved pressure fields from non-time resolved PIV data, there are two possible approaches. Approach 1: reconstruct time-resolved velocity field first from non-time resolved PIV data using sensor data, and then time-resolved pressure fields are estimated from time-resolved pressure fields by applying pressure Poisson equation. Approach 2: first estimate non-time resolved pressure fields from non-time resolved velocity field measurements using pressure Poisson equation and then reconstruct time resolved pressure fields directly from non-time resolved pressure fields. These two approaches are compared in this study. These approaches are demonstrated for actual PIV data of flow over a cylinder. Time-resolved PIV measurements are down-sampled to mimic non-time-resolved velocity data. It was found that the approach two is approximately thirty times faster than approach one when time resolution is improved from 1 Hz to 50 Hz. This is expected because in the second approach the pressure Poisson equation needs to be solved only with non-time resolved velocity fields which reduces the computational load.