An innovative LDV data processing method for statistical error corrections. Application to homogeneous and non-homogeneous seeding

In this experimental study, statistical errors in the processing of raw Laser Doppler Velocimetry (LDV) data are discussed. Acquisitions have been made in an open-circuit wind tunnel in the wake flow of a simplified car model, for validation rates up to hundreds of Hertz. Both homogeneous and non-homogenous seeding conditions are considered. For homogeneous spatial concentration seeding, the velocity systematic error is presented regarding previous works. Correction methods are then discussed according to their assumptions. Then, this paper aims at characterizing the burst random error related to non-homogeneous seeding. The experimental assessment of the available methods (interarrival time weighting and transit time weighting) according to the seeding conditions allows a deeper understanding of their ability to correct both velocity systematic error and burst random error. Finally, a new method is proposed mixing the previous one, leading to reliable (velocity systematic error correction) and repeatable (burst random error correction) results whatever the seeding conditions in our range of validation rates.