2D fluid case

High Reynolds number experimental lid driven cavity bench

Experimental set-up

The Lid-Driven Cavity (LDC) consists of a 3D cubic or parallelepiped fully confined domain, where the flow is generated by the motion of one of the wall's cavity. The LDC is a fundamentally attractive fluid mechanic configuration as several well-known flow phenomena occur: shear-flow, boundary layers, eddies and core vortex, curved streamlines, Taylor-Görtler-like vortices, etc. Such rich physic has conducted the LDC to be an ideal benchmark for direct numerical simulations methods development and validation.

The LDC experimental bench developed at the LML is powered by a moving band, allowing to achieve high Reynolds numbers and turbulent flow regimes. Measurements are performed using Particles Image Velocimetry (PIV) using a pulsed Nd:Yag to create a laser sheet, combined with a CCD camera. The flow is seeded with Iriodin particles of about 10µm in diameter.


Driven Cavity dynamic illustration

Particule Image Velocimetry Methods

Classical method

The images won't be scaled and two different meshes are used. The Intercorrelation metric is used with FFT and parabolic sub-pixel interpolation. Note that the image is deformed between the two sequences.


Speed: 0.81s/images


There is no wrong displacement when a 5x5 median filter is used. If one decreases the size of this filter window, wrong displacement vectors can appear. The vorticity field is quite smooth.

Pyramidal Method

The same mesh is used on the two sequences but the image is scaled.



The results are almost identical as the one using the classical method but there is a 1.7 speed-up which is essential when dealing with thousands of images in a test.

Optical Flow Integrated per Block

Tx and Ty for each ZOI




While the time is multiplied by 5 compared with the classical method, one notes that the vorticity field is quite different. The vorticity field has clear high frequencies that are not caught by the classical method.

Tx,Ty & Rz for each ZOI

One reminds that Rz is the vorticity
46.3s /10 images


Results: Vorticity with finite differences

There are no real differences with the field where only Tx and Tz are computed so adding the vorticity does not interfere with the identification of the displacement fields.

Results: Direct Optical Flow measured vorticity

One can note that this field is slightly different from the previous ones, as significant turbulent fluctuations are observed.



The images used herein, the parameters files and the post-processing python script can be downloaded here
Number of download: 35