Experimental Measurements in Wind Tunnel

[Old_Trafford]

Hello everyone,

the forum looks great and I have been reading it for a long time, but I have only registered.

I am writing this post because I have to analyse two flow fields in a wind tunnel.

1) The first flow is generated from a helicopter blade, with Re=2000000, Mach=0.92 and Strouhal=0.01. The wind tunnel can work for 10 seconds every hour. Also, the tunnel has side windows and a liner on the top.
I want to measure the velocity of the flow in the airfoil by using a laser and water seeding. Would you advise that? That else would you advise me to do in order to make experimental measurements?


2) The flow over a car must be analysed and the separation point must be found, using an open wind tunnel. The Re=1600000 and the model length is 250mm. What would you advise on that?

Thank you for your time,

George

 

[bigfooted]

I would advise against using water droplets as seeders. They tend to evaporate, especially at such high velocities. Are you using a PIV system?

For the car, if the only thing you need is to find the separation point, just put a thread on a stick and move it over the car. Or use a pitot tube.

 

[Old_Trafford]

I think that the PIV system is the best option. What seeders would you use?

Would you advise thermometry as an option for the car? Could you please explain the "thread on a stick" option?

Thank you very much

 

[bigfooted]

usually the boundary layer is very small up to the point of separation. After that vortices will form, which cause changes in flow direction. If you put a small thread on a stick and move the stick close to the car, the thread will follow the flow downstream. After the separation point, due to the vortices, the thread will behave more erratic. We used to stick threads on the trailing edge of airfoils to visually see at which angle of attack we would get separation. You can also use a small microphone and hear the flow separation.

For the PIV system: I guess you're at a university if you have access to a PIV system so just ask the lab guys what they have. Usually, polystyrene, aluminium, or certain oils that don't evaporate that fast.

 

[boneh3ad]

Oh goodness, oil PIV is a huge pain. It gets all over the place. TiO2 is a popular choice as well.

 

[RandomGuy88]

There are smoke generators that work really well for PIV. They are really nice because they are very safe and don't leave the residue that oil does.

I don't work with the types of tunnels that can only run for short periods of time, but I would imagine that the short runtime will make it so that you would have to do a lot of runs to get enough good PIV data. The lasers and cameras typically can get image pairs at a rate of only 10Hz. Assuming there will be some startup time before you can actually reach a steady state you will probably end up with less that 100 image pairs per run. This isn't to say this won't work it will just take several runs to get enough data to get good statistics. It would certainly be better if you had high speed lasers and cameras.

As for the separation point on the car I would suggest mini tufts or surface oil flow. Mini tufts are small pieces of string that you place onto the surfaces the pieces of string will point in the direction of flow and at the separation point you will see a change in direction of flow. The disadvantage of this method is that you don't get great resolution because there has to be space between strings, the strings may interfere with the flow and it is time consuming to put them on.

Surface oil flow is very easy and a great qualitative method. When I do it, I mix a UV dye with mineral oil and then airbrush the mixture onto the surface. The airflow shears the oil so you can see the direction of the flow very near the surface. If there is a separation line the oil will typically pool in that area because the shear stress decreases substantially.

 

[Old_Trafford]

I am thinking of using PSP (pressure sensitive paint). I know that this is used mainly in low speed flows (automotive industry) and steady flows but i wonder if i could use it in unsteady transonic conditions. The Mach number is 0.92 and the Strouhal number is 0.01 leading to a frequency of 33.4 Hz. Since a typical time for pressure sensitive paint to function is 1s, would this method give an averaged pressure distribution over the airfoil? (in this 1sec duration). Furthermore, the wind tunnel will work for 10 sec. Will this destroy my pressure distribution (due to the unsteady nature of the flow due to the shocks) or will it give an averaged distribution?

 

[Old_Trafford]

For calculating the velocity field around the car model(case 2) in Re=1.600.000 (U=96m/s)
in an open wind tunnel. Would you recommend LDA or PIV? LDA is surely more difficult to set up and to get the whole 3D results in transverse grids along the 250mm model. That is why i was thinking about 3D PIV method.

 

[boneh3ad]

For one, PSP will absolutely work in transonic or supersonic or even hypersonic flows. As long as there is a pressure differential involved, it will react.

Second, if the response time of the paint is 1 second and you expect your flow to fluctuate at 33.4 Hz, you are not necessarily going to get an average. You will pretty much get nonsense. That does not appear to be a great method for you to use here. I don't see any problem with using PIV as you had originally planned as long as you make an intelligent choice of seed particles such as TiO₂. That would work fine, you would just need to use a fairly large amount since this sounds like it is not a recirculating tunnel.

I assume if you are treating this as a helicopter problem then the airfoil will be "flapping"? You could phase lock the PIV image pairs to specific points during the flapping motion so that you get a whole bunch of images at a whole bunch of identical points in your cycle. Then you could do some phase averaging to get your averages for each point in the flapping motion. Of course if you are limited to 10 Hz, then you just phase lock your system to the flapping for a given portion of the phase and then stop, switch the point you are locked with and re-run the experiment to get another point.

I'd second RandomGuy88's suggestion for surface oil flow on the car if all you want is the separation point. It is super easy, cheap, and gives you essentially infinite spatial resolution.

 

[RandomGuy88]

Actually pressure sensitive paint is generally much better suited for high speeds. I have a lot of experience with this method though it is all at low speeds. The signal strength depends on the pressure changes so at high speeds you get much better signal to noise ratio. I have spent a great deal of time working on improving the signal to noise ratio at low speeds.

A potential problem with the method at high speeds is that PSP is very sensitive to temperature. Meaning that if the temperature changes in time or in space you will see a difference in PSP signal which you may incorrectly interpret as a pressure change. This causes problems at high speeds because of the large temperature changes that can occur in these flows. With that said you can and probably should use binary PSP. The binary paint contains two luminescent molecules. One molecule is pressure sensitive while the other is temperature sensitive. This gives you a reference signal that can be used to correct for temperature effects. It complicates the setup slightly depending on how you go about it.

The time constraint may be an issue. Assuming the flow reaches a steady state including thermal equilibrium you would be fine, just limited in the number of images you acquire. There is unsteady PSP, it is developed specifically to have a fast response time. It can capture frequencies of a few kHz. Might be worth investigating.

I take it this helicopter blade is not rotating correct?

A difficulty with PSP is that you need to acquire an image while the tunnel is on and an image when the tunnel is off. Because the loads on the model usually cause the model to move slightly you have to align the two images before you can get good results. This requires that the surface of your model must have sufficient number of markers. Points on the surface that are easily distinguishable and are in both images. By locating the markers in both images you can apply image alignment algorithms.

 

[Old_Trafford]

Thank you for everything Guys