Image based velocity profile estimation for Pulsed Ultrasound Velocimetry

A well-established, non-invasive, in-line technique for monitoring the rheological properties of a fluid in a pipe in real time is Pulsed Ultrasound Velocimetry (PUV) in combination with pressure difference measurement (PD). PUV makes it possible to determine the velocity profile along the measuring axis using Doppler echography. One problem that is left to solve is to develop an automatic determination of the position of the pipe wall and to estimate the correct velocity close to the liquid-wall interface. Due to temperature changes, different fluids in pipe, and pipe material it is not possible to set the wall position to a fixed value. The profile is distorted close to the wall, because the received signal is a convolution of sampling volumes with the velocity distribution of particles within the pipe. That is especially a problem when the sampling volume overlaps with both, the wall and the fluid. It has been shown that the physical characterization of the pulse and deconvolution of the signal improve the velocity estimation. However, this method is complex and an easier and especially faster method has to be found. In this work an image based velocity profile estimator is developed and tested for its industrial applicability. The code for the velocity profile estimator was written in MATLAB. The image is generated from the Doppler spectrum and is then processed with the help of functions of MATLAB’s Image Processing Toolbox. It was shown that the determination of the wall position with the newly developed velocity profile estimator works reliable according to the image. The user does not have to select the wall position manually anymore. Furthermore, the shape of the velocity profile that was estimated from the automatic detected wall looks reasonable and is smoother than velocity profiles obtained with the current FFTmethod, especially, when a small number of pulse repetitions was used. A really important improvement was the enhancement of the detected velocity gradient at the pipe wall. It is now realistic and thus the correct velocity can be obtained The correct rheological properties of ketchup could be found in half of the time with the image based velocity profile estimator compared to the FFT-method because the number of pulse repetitions could be decreased.