Turbulent Flow Particle Deposition Measurement

Project Overview

I engaged in research at the Fluid Transport Laboratory at Johns Hopkins University, where I studied the effects of electrostatics on multiphase turbulent flow. When particles (glass beads, sand, etc.) collide, triboelectric charges form. As these small particles become charged, they will often clump together and deposit themselves on surfaces. This can happen to jets flying over deserts to spacecraft landing on sandy/dusty surfaces like those of the moon/Mars. In order to understand how particles deposit themselves on surfaces in these environments, a project was begun to adapt another research group's deposition height measurement system to apply to the 10-40 micron scale. The name of this technique is known as Digital Image Projection (DIP). This project was less of a design project and more of an image processing and MATLAB intensive project. I had to hand the rest of the project off to the PhD student (Matt Gorman) I was working with when I completed my time at the lab.

The Theory of Digital Image Projection

The DIP technique utilizes the same theory as stereo imaging to obtain 3-dimension images, but one of the cameras is replaced by a projector. The projector and the camera are mounted in the same plane, but are aimed at a similar target. The projector displays a random, greyscale pattern on the target. Once the system is calibrated for the target (zeroed), any change in the height will distort the projection from the projector. This distortion, the geometric theory of stereo imaging, partnered with the MATLAB correlation therefore allows for the height of the target to be determined. This is especially exciting for when a clump of particles deposit themselves, distorting only that specific portion of the target. As a result, the varying height of the clump can be determined.

Geometric Theory for DIP

Image Credit: "An experimental investigation on the surface water transport process over an airfoil by using a digital image projection technique." Hu, et. al., 2015

DIP Setup Schematic

Image Credit: "An experimental investigation on the surface water transport process over an airfoil by using a digital image projection technique." Hu, et. al., 2015

Calibration Test Images

Above is a set of 23 images, starting at the user defined zero. The random greyscale pattern was projected on an optics linear translation stage with a flat platform. The height of the stage was increased in 0.001 inch increments. This is what is causing the above shift to the left of the projection.

Test 1: Projection on Flat Plate

The projector displayed different test patterns on a flat plate of the linear translation stage. My work was using a MATLAB program to procedurally and randomly generate these patterns. The program allowed for the user to update the pattern live, whether it be changing the resolution or changing the size of the pixel "blocks." Additionally, there a section of the code that randomly generates a maze pattern. The user can modify the thickness of the maze walls and the resolution of the maze. Having adjustable patterns allows for the user to determine what the best pattern for DIP is, specifically for how well (accurately) the correlation program can identify the patterns.

Below, two of the patterns, one random greyscale and one maze can be seen. Again, these are about 23 frames from the zero position to being raised in increments of 0.001 inches up to 0.022 inches.

The Results

One example plot from the correlation MATLAB code can be seen on the left. This plot represented the plate at the 0.003 inch height, with the actual heights displayed in the colored part of the plot. It can be seen that there was fairly substantial noise in this analysis, which showed a point of refinement.

After working on this project for a few months, this is about the stopping point for when I ended my time at the lab. However, there was one more contribution I made to this project which can be seen below.

Future Testing - Calibration Target

Resin 3D Printed Target

The CAD model on the left designed to have four grid sections. Each grid quadrant contains circles of random diameter, extruded to different heights (20, 40, 60 and a mixture in the fourth quadrant) representative of the scale that would be observed in the research.

This part was subsequently printed on a stereolithography (SLA) resin printer with a layer height resolution of 10 microns (0.01 mm). This allowed for very accurate tolerances that could be used as a reliable control value.

After calibrating the DIP software and setup on a flat plate, placing this part on the target and running tests would allow for the user to determine if the calibration was done correctly.

Page updated

Google Sites

Report abuse