How to Do Flow Visualization Through Background Oriented Schlieren (BOS)

Background Oriented Schlieren (BOS) is an established method to visualize the flow, which is invisible to our eyes. Different from the classical schlieren visualization technique, BOS does not require the expensive concave focusing mirror. Perhaps your interests in BOS was ignited by NASA's BOS video showing the shock wave on the full scale supersonic fighter jet. That is impressive, isn't it! So I am planning to introduce you the steps to carry out BOS experiment at home or in any laboratory. Two BOS experiments will be demonstrated: the hot jet flow from a hot air gun and the candle flame. Just to assure you, BOS is not as difficult as you imaged.

Smartphone.

The smartphone certainly needs to have a camera. I believe most of the smartphones nowadays should be suitable.

I actually also tried to use a DSLR camera, however, due to the flipping motion of the mirror, there was vibration, which affected the image quality and the image post-processing procedure. I believe a mirror-less camera is also suitable. It is worth trying if you have one.

Tripod

A stable tripod is very important, it is also used to prevent any vibration and movement of the camera. In my opinion, a tripod is necessary in BOS experiment.

Bluetooth Shutter

The Bluetooth shutter is used to avoid controlling the camera by touching the screen, so that vibration or movement can be avoided.

Background Pattern

A suitable background is critical in BOS. The pattern I used is randomly generated particles. The particle image generator is provided by imagico (http://www.imagico.de/map/jsdotpattern.php). The particle generator interface is shown in the picture. Basically, there are two parameters to be controlled: the particle distance and image size. Please feel free to explore the particle generation in the website. Once the particle image is saved into your computer, you can print it out. Regarding the particle density in printed image, the rule-of-thumb is that we should be able to distinguish each individual particle in the recorded image. You might need to adjust the particle distance to achieve a suitable particle density.

The experimental setup is shown in the figure. You probably realised that the experiment was performed on my kitchen worktop. It is the most stable surface in my house. The hot air gun also needs to be mounted to a stable base, because the fan within the air gun would cause vibration, and the hot air gun is likely to move due to the reacting force from the hot jet. In my setup, a base built by Lego bricks were used to hold the air gun. A heavy metal piece was also added to hold the base in place firmly.

There is another trick. The switch of the air gun was set to always on, and the on/off was controlled by using the switch in the wall power plug. This was implemented to avoid touching the hot air gun.

Now we can start the experiment by taking images of the background pattern while the airgun is on. Please also remember to take a photo when the air gun is off.

The background images before and after the air gun was switched on are compared here. The overall images does not seem to vary. But if we zoom in towards the gun outlet region, as shown, particle displacement can be seen. The particle displacement is due to the refractive index change in the hot air. I also measured the hot air temperature through a thermal couple, it reaches nearly 300 degree C.

Great! We need to calculate the particle displacement now!

The aim of the BOS image processing is to calculate the particle displacement. Several algorithms do the job. I chose to use the cross-correlation, which is also the algorithm used in PIV (Particle Image Velocimetry). If you are interested in the cross-correlation method, you can watch the video clip in YouTube（https://youtu.be/E1IyvwVWf6A）.

I used PIVLab in MATLAB to calculate the particle displacement, as it has a user-friendly interface and easy to start. More importantly, it is FREE! There are also some other programmes can do the job. It is really up to your choice.

Once the images are uploaded into PIVLab, we should set up the parameters for the cross-correlation:

• PIV algorithm： FFT window deformation
• Window size: Decreasing window size method is adopted in PIVLab. Basically, the final resolution is determined by the final smallest window. However, in the present BOS measurement, the particle displacement is quite small, it is not very critical to use the decreasing the windows. In the attached image, I chose to use two sets of windows, both have 50% window overlap.
• Sub-pixel estimator: default setting is recommend.

The background knowledges of cross-correlation are not elaborated in detail. If you want to have in-depth introduction, we can also communicate personally.

Now, feel free to click the 'ANALYZE' button. The particle displacement should be calculated for you.

The cross-correlation results will provide two displacement components, one for displacement in x-direction, the other in y-direction. The question now is to decide which component should be used to visualize the hot jet flow. Please follow the rules below:

X component = vertical knife edge in classical schlieren

Y component = horizontal knife edge in classical schlieren

The hot jet is along the horizontal direction, if we want to visualize this flow in classical schlieren, a horizontal knife edge is preferred. Therefore, the y-component is used for flow visualization in this experiment. The result is shown in the attached figure. The hot air flow from the outlet is clear.

Now I want to do a little bit discussion on the result. If we have a closer look at the color bar beside the BOS image, we can find that all the particle displacements are below 1 pixel, which is very small. In my opinion, the small displacement should affect the uncertainty of the displacement calculation. Perhaps we can achieve a larger displacement by optimizing the background pattern.

Luckily, I previously visualized the jet flow of the same hot air gun using classical schlieren with a Z-shaped light path and two concave mirrors. The result is also attached. You can find the classical schlieren images reveals more details.

Using the same procedure, the air heated by the candle flame was also visualized through BOS. This experiment is slightly easier than the hot air gun, as there is no moving part of the test rig. The BOS images are also compared with classical schlieren images.

Perhaps you already realized the Pro and Cons of BOS in comparison to classical schlieren.

Pro: the visualization area is not restricted by the size of concave mirror, potentially you can achieve a very large area.

Cons: The resolution and sensitivity is weaker than classical schlieren.

Now you can start your own BOS experiment!

You can enjoy more schlieren experiments by visiting Luftvis website: www.schlierenvisualization.co.uk