AERO 307: Wind Tunnel Lab

By Eric Koch

[ekochaero-blog]

...And Rounding Out the Quarter

So with ten fun- and work-filled weeks behind us, it is time to reflect back for on the wind tunnel journey for the last time. I’d first like to start by saying a few words about the lab hack project. I think our project with the octagon inserts were a lot of work for just two short weeks, especially when combined with the work load from other classes. That being said, my group was able to come together, plan things out, and then execute the plan in just enough time. At times it was hard and time consuming work, but in the end we were able to get everything done that we planned to, and it turned out really well on top of that. 

I think this applies to the various teams I worked with at each point in the quarter, too. We all worked together well, divided up tasks as best we could, and got stuff done. What was even more impressive was that everyone seemed motivated to work, and even be extra ambitious with our projects. The flow vis project, for instance, could have been a fairly simple undertaking. But, instead of just going with the flow (pun intended), we chose to do something creative, despite the extra hours in the lab that it entailed, and the results of that effort speak for themselves. Because I was able to be a part of such a unique and interesting project, I got exposed to some really cool aerodynamics testing techniques, and I think I am more intrigued by the subject as a result. 

An additional result of the flow-vis project was the cementing of the most memorable moment of the lab for me: the dye explosion. Really, no matter what happened during any other part of the quarter, this one had it beat. But I think my second most memorable moment would have to be when we got the dye system working, and witnessed some impressive flow visualization around our favorite wing section. The collective oohs and ahs from everyone standing around the tunnel (like the aerodynamics nerds that we know we are) was almost as entertaining as what was occurring in the water channel itself.

Once again, I think this course really boosted team work skills across the board. Going into senior design in the fall, I think I will be better prepared to coordinate tasks and keep everyone on the same page, as well as be less afraid to think outside the box. Therefore, I am actually really excited to see what we come up with in senior design.

Last, a word about the wind tunnel lab itself. As I have said several times, having the opportunity to work in this setting has been a singular and rewarding experience. Having access to such high-tech equipment was a privilege, and I will fortunately have the opportunity to spend more time in the lab over the summer, through various pursuits. At the end of this course, I certainly feel like I’ve gained not only more knowledge about aerodynamics, but also a renewed interest in the subject.

It’s been real.

[ekochaero-blog]

Rounding out the Edges...

With a baseline collected, we set to work on manufacturing and installing the inserts. Once we obtained the dimensions for the triangles from the CAD model, we traced out the lines onto our sheets of polycarbonate, went to the machine shop, and cut them out on the vertical band saw. Upon returning to the lab, we immediately got to work on installing our first test insert. Unfortunately, this process was slowed upon learning that we would need to take extra precautions due to the (now painted over) layer of lead-based paint on the tunnel walls.

Eventually, though, we got the first piece installed:

We wanted to make sure our method was going to work before installing the remaining inserts, so we fired up the tunnel. Lo and behold, it worked beautifully: the plastic was held firmly in place by the screws, nothing vibrated as air traveled over it, and it looked good. Satisfied, we set to work on the remaining pieces. This process turned out to be rather uncomfortable, as we had to contort our bodies around the cone in the diffuser to install the upper inserts. In the end, though, it looked just as we had imagined it would:

Photo Courtesy of Harrison King

We then proceeded to do the same with the other diffuser section. Unfortunately, I did not obtain a photo of the end result for that section.

The process for the foam inserts ended up being relatively straightforward. We just had to convert a drawing of the CAD model to a specific file type (.dxf, in case you were wondering), and the CNC hot wire did the rest. To install them in the tunnel, we had to take the entire section out of line with the rest of the tunnel, epoxy the inserts into place, and put everything back.

Photo Courtesy of Harrison King

Once again, we ran the tunnel (once everything was back in place, of course) to ensure that everything would hold, and it did! Somehow, this project went better than one might have expected. We did run another test with collecting data once everything was installed, but unfortunately the results from analyzing the data is not available at this time (thanks, finals week!). 

[ekochaero-blog]

Geometry is Everything

The next phase of our effort would be to plan how to actually go about creating the inserts. The initial concept involved cutting high density insulation foam to the desired shape with a CNC hot wire. After doing a bit more research on this approach, we decided that it would be too time consuming and potentially a waste of rather expensive foam to try to cut out such large and complex geometry sixteen times. So we set about brainstorming an alternative. Eventually, Harrison had the idea to use sheets of plastic, cut out in triangles, that would be bent into the desired circular shape. The triangles would be held in place by wood screws drilled into the diffuser walls. After looking into it, we discovered this method would not be any more expensive than the foam. Fortunately, Dr. Doig had some 1/8 in polycarbonate lying around, so we were good to go without spending more than a few dollars on the hardware.

With our method decided on, I whipped up some CAD models to visualize our approach, as well as obtain the geometry for the triangles. While taking measurements of the tunnel, we also discovered a gap between the diffuser and the tunnel with held the debris fence in place. We later decided to make foam inserts for this section, since the geometry was relatively straight-forward compared to the others.

Next, we needed to do some preliminary tests to get a baseline to compare to later. In order to do this, we made a sheet of tufts and attached it to one of the corners in the downstream section of the diffuser.

Though not shown in this image, running the tunnel with the tufts proved that there was some separation going on here. What is not entirely certain is how much of the turbulence is due to the lip, and how much is an effect of the fan blades themselves, since they do not exactly produce smooth flow.

 We also obtained open-tunnel data from another group that was adjusting the fan blades, which would tell us the velocity achieved at certain RPMs.  After installing the inserts, we would be able to compare the results to this baseline data in order to get an estimate of the impact the inserts have.

[ekochaero-blog]

Enter the Octagon...

Following the fantastic conclusion to the flow vis project, we moved on to the final endeavor for the course: the Lab Hack Challenge. For this project, we have been asked to choose from a list of projects that either improve the wind tunnel lab in some way, or involve assisting in an ongoing student project. I chose the octagon lab hack, which entails fitting the octagonal diffuser with inserts that smooth the transition to and from the circular fan housing.

Here is what the diffuser looks like downstream of the test section:

Looking back upstream from the end of the tunnel:

From these images, you can clearly see the rather abrupt change in geometry. The problem with this may not be clear, however. While somewhat small, the corners of the octagon effectively create a blockage in the flow headed towards the fan housing, which causes some loss in efficiency. In the section downstream of the fan housing, the corners create a lip for the air to separate past, causing turbulence and, ultimately, more loss in efficiency. Therefore, by placing inserts in those corners, we hope to mitigate those issues and positively influence the flow through the fan. It is difficult to say how much we expect to gain from this, but we shall see.

[ekochaero-blog]

Water Channel Flow-Vis: Restoring Dignity One Drop of Dye at a Time

With the disastrous first test behind us, we pushed on to achieve what we set out to do in the first place: conduct some jaw-dropping flow visualization with the infamous NACA 4412 airfoil. We decided our objective would be to observe flow separation and stall at several angles of attack, as well as wing tip flow characteristics for several wing configurations. For the latter, we hoped to be able to capture the formation of wing tip vortices, and the effect winglets have on this behavior. I designed and modeled the winglets, while Jake modeled the wing(specifically designing in a rod that could be used with the existing mounting system on the channel), 3D printed it all, and then lightly sanded it to get a nice surface finish.

Our straight wing, nicely rendered in Solidworks:

And the two winglet configurations:

It is worth noting that both winglets carry the NACA 4412 profile all the way through. This means that, theoretically, we could observe very comparable coefficients of lift and drag across the winglets as the straight wing. However, several variables, such as print quality and resolution, and changes in dimensions after sanding, might eliminate any effects this fact might have. Regardless, let’s move on to the experimental setup and results!

After the last post, you’re probably wondering how we were going to go about pressurizing the dye system this time. Well, fortunately, Cody was able to locate a new hand pump for us, which worked perfectly: It held the right amount of pressure like a champ for the entirety of the experiment. We needed to replace the line for the green dye(which was the only color left after our mishap), but otherwise the dye system behaved for the duration. We set up LED worklights in several different positions to give us the right lighting for various angles. As you’ll see, Jacob worked the camera like a pro, as we made sure the rest of the process went smoothly. The mount we used with the wing (black apparatus shown in the image above) enabled us to easily change angle of attack, and held the wing firmly in place. All that was left was to watch in awe as the water channel, dye, and wing section came together and worked in harmony to produce some pretty stunning images.

First, we tested for separation at stall. At around 10 degrees, the flow is clearly still attached:

But, by around 20 degrees, the wing is clearly stalled. Not only that, it looks stunning:

Next up was comparing wing tip vortex formation across our different configurations.

For the straight wing, tip vortices are very clearly forming, and there is a significant amount of entrainment of the dye along the chord. Now, as we look at the winglets, we hope to see a decrease in the amount of entrainment, as well as a reduction or even total prevention of tip vortex formation.

With the single winglet above, it seems that there is less entrainment of the dye, and no clearly formed vortices. In other words, it seems to be doing its job!

Results are a bit unclear with the “split scimitar” winglet, shown above. However, the addition certainly seems to have had an effect on the degree of entrainment. One thing to keep in mind while we are looking at these is that the winglets are not perfect. They were not thoroughly engineered to produce optimal efficiency, and we were very limited on the amount of sanding we could do on such thin cross sections, which left slightly rough surfaces in places. Therefore, some noisiness in the dye is to be expected. Even in light of that, we feel that we were able to observe some pretty useful (and fun to look at) behaviors.

Closing Comments...

Now that we have collected all of our mesmerizing visuals, I can look back on the last two weeks and be proud of all that we accomplished, despite our troubles. This experiment gave us the opportunity to really contribute to the lab by getting the water channel up and running again (and leave more behind than intended, specifically on the floor), while also learning more about flow behaviors around our favorite wing section. It was, all things considered, a pretty rewarding experience, and I look forward to contributing even more to the lab in the last few weeks of the quarter (preferably without wreaking any more havoc).

[ekochaero-blog]

Victories and (Epic) Failures in the Lab

On Tuesday following our efforts to install the filtration system, we were finally able to fill up the channel. it was not long before water started to leak out of the filter housing. Fortunately, we hadn’t put that much water into the system yet, so we were able to unscrew the housing from the lid, adjust the O-ring seal, and screw it back in (properly seated this time), without making too much of a mess. To our surprise and satisfaction, all of the other plumbing that we had installed held up without a single leaky seal. So we filled up the system the rest of the way(all ~100 gallons of it!) and fired up the pump motor. With everything still holding up, we decided to run a dye test ahead of our actual experiment using an existing aircraft model.

This is where things got interesting. The good news: the dye system with delivery lines and pressurization lines was pretty much ready to go. The bad news: the hand pump used for pressurizing those lines was busted. So we went to find Cody, our favorite Aero lab tech, in search of another way to pressurize the system. Sparing the details, we eventually got everything rigged up to a portable air compressor with a decent amount of pressure left in the tank, and a reliable regulator to keep from over pressurizing the lines. This setup allowed us to conduct the test seen in the video above. Unfortunately, we would need to find a more sustainable way to pressurize the system that would not require us to run a loud compressor every few minutes for our actual experiment. So we set out to find another regulator that could be used with the shop air (the compressed air line that runs throughout the lab). Eventually, we did obtain one, and hooked it up to the system and prepared for what we thought would be one last successful test of everything before calling it a day. Well, sometimes things don’t go as planned. Basically, the regulator couldn’t handle the immense pressure in the shop air line. Immediately after switching on the air, the regulator shot up to 50 psi--not bad when you’re trying to inflate a mountain bike tire, very bad when the output of the regulator connects somewhat loosely to half-full (that’s right, I am an optimist) bottles of dye, and you only need about 10 psi. Being the lucky student who volunteered to operate the regulator, my optimism quickly turned to horror as the pressure dial’s needle refused to turn in any direction but clockwise. Amidst the chaos of yells to switch off the air, I sat, frozen, pondering what was transpiring not two feet from my face, until suddenly one of the lines--the red dye line--popped off the bottle, spraying the (fortunately small) amount of dye in the line at the time onto the floor in front of me. I grabbed the line as fast as I could in an attempt to kink it, but it was in vain. The pressure continued to rise, and just as I stood up and turned away to escape my impending fate, the blue line popped off as well, and, not unlike a great volcano, exploded into a geyser of smurf blood. The results of this disaster were, as you would expect, not pretty:

Photo Courtesy of Ransom Cutshall

We spent the next hour and a half cleaning up the lab (and ourselves) as thoroughly as possible. Luckily, no one was physically injured. But with our wounded prides, we cleaned as best we could as we mulled over the events leading up to that moment: where we went wrong, and what we needed to do to prevent this from happening again. We still came out with a positive outlook on things at the end of it though:

Not shown but equally blue: Jake. Photo Courtesy of Dr. Doig

And even when I thought I had enough of flow-vis for one day, there was still more:

Just not the type I wanted to deal with.

Joking aside, though, I learned an important lesson here about the importance of lab procedures and safety. Fortunately, Dr. Doig was forgiving enough to allow us to test another day, but we would certainly be sure to think twice and think hard before throwing any switches again.

[ekochaero-blog]

Preparing to Visualize some Flow

Writing the report for the NACA 4412 experiments was a lot of work, but my team worked very hard to get it mostly done a bit early (a day early in fact), so it wasn’t last minute. The result of these efforts was that it enabled me to dedicate a large chunk of Sunday to work with my flow-vis group to get the water channel up and running. [AERO 307 has become life.]

Basically, in order to get the water channel up to snuff, we would need to install a filtration system into the existing plumbing so dye that is injected into the flow for visualization could be filtered out, thus maintaining consistently clear water. The plan was simple:

With the materials already supplied by the lab, we set about to do some good old plumbing work.

I calculated the dimensions for the pipe that we would need for our desired placement of the filter, based on optimal placement of the valve lever and the ability to remove the filter tank for replacement and maintenance. Then all that was left was to measure twice, cut once, and glue! Fortunately, Garrett had some experience with proper plumbing techniques, so we left it to him to get the fittings and seals just right.

After 5 short hours of work, the job was done:

And it looks good! Unfortunately, we would need to wait for another day to actually fill up the system and put our work to the test...

[ekochaero-blog]

Balancing Forces...and Time

Well, it’s done. The thorough report that is the culmination of a month’s worth of testing on the NACA 4412 has been turned in, and now we [continue to] look at visualizing flow over our favorite airfoil using proper techniques. But first, I would like to take this post to reflect on the past month in the lab and some of the challenges and successes we faced during that time.

The week leading up to the report deadline was a chaotic one: problems with the force balance system involving electromagnetic interference early in the week left groups unable to test during their scheduled times, the report required attention and strongly discouraged procrastination, and thought needed to be put into looking ahead to flow vis. Fortunately, the problems with the force balance were sorted out just in time for my group to test and collect some sweet data. Unfortunately, the corresponding analysis on this data occupied the better part of the 72 hours between the experiment and the deadline (thus why there is no post about our force balance testing prior to this). 

Correcting Data...and then Correcting Again

Speaking of data analysis, crunching the numbers for the force balance experiment proved to be much more involved--and time consuming--than expected. This was mostly as a result of certain considerations needing to be made with our data. First, we learned that there was a finite amount of “drift” in the load cell (the sensor that measures the applied forces using a healthy combination of physics and black magic) which would need to be accounted for. This basically meant that the load cell would read zero force (assuming it was calibrated correctly) at the start of data collection, but by the end, once no external forces are being applied (i.e. the tunnel is off), the sensor read some non zero force. Fortunately, this was easily accounted for by recording the initial and final readings and subtracting the linearly related values from the data. 

But wait, there’s more! Since the wing is mounted to the force balance system via a strut (as seen above), we also needed to account for the forces this generates by running a test on the strut alone. Using a calibration curve, we accounted for this in the data. However, there are some problems with this. Since the strut occupies some of the lifting surface area of the wing, completely removing the forces generated by the strut would be, in effect, also removing some of the force generated by the wing. Unfortunately, this is probably the best method we have for accounting for the strut.

Final Thoughts, and Looking Forward...

All in all, the past month has been a fun-filled learning experience. It was very interesting working with all kinds of measurement methods, from the embedded pressure ports on the infinite wing, to the wake rake, and finally the force balance. The force balance was especially impressive, given the amount of development, time, and money that went into getting it up and running. Though it may still have its idiosyncrasies, it was very exciting to work with such complex equipment. And this, really, sums up my thoughts on the lab so far as a whole: we have been given the resources and (for the most part) left to our own devices to conduct realistic experiments to produce potentially high quality results and findings. While this level of freedom has proven challenging at times (making sure to record all necessary measurements like ambient conditions and apparatus dimensions, accounting for all sources of error, drawing meaningful conclusions...), the result is that we get to truly “learn by doing”, and that’s pretty cool if you ask me.

Stay tuned for a look at what we’re doing to prepare for some awesome flow visualization....in the water channel!

[ekochaero-blog]

Assessing Sources of Error with Flow-Vis

In order to fully investigate and interpret our data, it is important to consider all factors that might influence the results. Therefore, it would be prudent to discuss the effects of additional aspects of the experimental apparatus and their influence on the flow being measured. Specifically, there is the potential to see influences from wall effects from the tunnel walls, interference from the wake rake apparatus itself, as well as discontinuities in the tunnel floor height. While I won’t go into numerical analysis of these sources of error, I will investigate them using a much prettier-to-look-at method: FLOW-VIS!

After collecting the necessary data for the offset wake rake experiment, we had a little bit of time left over--just enough time to do a spur of the moment exercise in injecting smoke into the flow.

Our goal was to get a rough idea of how much disturbance various things created on the flow through the tunnel. Of course, we can’t really quantify these disturbances using such a crude method of flow visualization, but we can at least get an idea of how much it visibly affects the flow, which might gives us a better idea of how much we should expect to see when we look at the numbers. Plus it looks pretty cool. We’ll save high quality flow visualization for when we’re actually meant to later on. For now, I think our efforts here produced some pretty interesting imagery.

First up is the rake. With our smoke aligned near perfectly with the rake by Ryan, you can really see the flow being diffused as it passes through the pressure probes and the body of the rake itself. It doesn’t really seem to have much of an effect upstream, though you can make out an increase in the width of the smoke stream just before reaching the probes.

The effect of the lip in the tunnel floor proved to be difficult to observe. Further down the tunnel, the smoke definitely started to move further from the floor, which may have suggested a boundary layer characterisation. Of course, it could have also just been the smoke diffusing.

Finally, we have wall effects:

At first glance, we didn’t think that the smoke really showed much in the way of wall effects. However, in this short gif, you can see what appears to be 3D effects, as vortices roll off the wing in the span-wise direction. Since we have an “infinite” wing here, we should not be seeing tip vortices that would be normally seen on a finite wing. Therefore, we can only assume that we are in fact seeing some wall effects. Good job smoke vis! Once again, we can’t really use this to quantify the effects, but it is pretty cool to actually see them in action! Side note: watch the smoke follow the contour of the intake upstream of the wing is mesmerizing...

[ekochaero-blog]

On Perturbations...

Continuing on with our current theme of examining sources of error in the lab, one significant subject of interest is that of flow through the tunnel itself. This is important because we make one crucial assumption in order to conduct any of our data analysis: that the flow is steady. Is it really though? Let’s find out.

The trouble with a wind tunnel is that it uses a practically invisible substance we call “air”, and without some form of flow visualization (i.e. smoke), it is difficult to determine the state of the flow through the tunnel. Therefore, the best we can hope for is to minimize the chances of disrupting the flow so that we can make our assumption of steady flow. While there may be some disturbances to the flow that are naturally occurring in the lab, I think it is useful to discuss those that we can actually account for with careful consideration. Keep in mind that it would be very difficult to correct our data for any significant disruptions to the assumed steady nature of the flow. Rather, we want to ensure that our data can be accepted as valid.

Upon preparing to start collecting data, we always allow for a “warm-up” period of a few minutes, in which the tunnel gets up to speed and steadies out. Now here is the important step: after we set the angle of attack for the next round of data collection, we wait approximately 10 seconds from the time the airfoil is clamped in place to the time we hit “collect” in LabView. The reason for this is that when we rotate the airfoil when the tunnel is on, the motion “perturbs” the flow (thus the title of this post). Basically, the flow over the wing is “pushed” in a direction other than downstream, causing a sort of wave in the wake of the wing. This can affect the flow properties such as pressure both downstream and upstream. Therefore, we wait. Ideally, the longer the time in between this perturbation and data collection, the better. However, our time is limited in the lab, and we want to make sure to get all of the data we need. Regardless of this ideal, 10 seconds has proven to be enough time, since we collect data for approximately 15 seconds, and no significant anomalies or differences are observed from when we start collecting to when we stop. 

With these particular situations accounted for, it seems we might just be able to rely on our assumption for near steady flow, unless of course doors to the outside world opening and closing during testing are to be thought of as significant...

[ekochaero-blog]

The Art of Clamping (and Other Sources of Error)

After several weeks of solid testing on our NACA 4412 airfoil, one thing I have failed to adequately discuss is the presence of many variables in lab procedures that may contribute to overall error in our results. The thing about error is that it is always present in experimentation, lurking in some form behind every bit of data and calculation. The only thing one can do is to alleviate the effects of this error as much as possible by taking great care to consider as many potential sources of error in the lab as possible. 

So, onto discussing the actual sources of error that might have existed in our experiments. First, as the title of this post suggests, clamping actually played a part in potentially messing with the reliability of measurements. 

 This was certainly one of the more mundane tasks in the lab(and somehow became the butt of jokes as the title of “professional clamper” makes its way onto our resumes), but it actually ended up being a pretty critical element in ensuring that we were set to record data. At the lower speeds, the forces generated by the airfoil were not strong enough to really move the airfoil in its place. However, at higher speeds, such as the 20 m/s we tested at, the airfoil readily rotated when not clamped sufficiently, as we discovered. Fortunately, this realization came before any data was affected.

The clamps did have another effect, however. It seemed that at higher speeds, upon clamping the airfoil in place, the angle of attack would occasionally shift slightly (~.5 degrees max). Of course, it is debatable how much of an effect this would really have on the data, but there it is. Although, since error has a tendency to propagate, it is possible that this would have an effect down the line. On the topic of angle of attack, though, I might mention (again) the method we used to check and set the AoA:

As you can (sort of) see from the blue arrow-shaped thing, which was used to pinpoint the angle, there is some inherent amount of uncertainty there that is somewhat difficult to quantify. Nevertheless, if we’re talking about propagation of error, if there is some error between the angle we were aiming for and the angle we actually set it to, and then it shifts up to half a degree when the clamps are set, we could be more than half a degree off from where we meant to be. Of course, when compared to errors that are the results of other variations in testing conditions, such a small difference in angle may not actually have that significant of an impact on the end result.

Speaking of other variations, we also observed fluctuations in tunnel flow velocity. This however did not end up being the result of human error. Upon setting the motor RPM to give us the desired speed, and then proceeding to give it time to stabilize, we still observed deviation from the desired velocity of about +/-.05 m/s. It may be safe to assume this to have negligible effect on the accuracy of results.

[ekochaero-blog]

Looking Downstream...and Upstream

These past two weeks working with the NACA 4412 airfoil were awesome. We collected some really good data (and lots of it), and have already conducted lots of interesting analysis. Our plots for lift coefficent (not shown) match near perfectly with plots found in Theory of Wing Sections. Moving forward, we will continue working on the final report as we add more data to the collection after experimenting (hopefully) with the force balance setup. In preparation for our next round of testing, I aim to code up some error analysis to examine our current data, as well as have it along in lab next week to examine the new data...live. 

Looking even further to the future, I have been coming up with ideas for the flow-viz segment. For inspiration, I looked to motorsport once again for inspiration. Back in the AERO 302 days, when we were asked to come up with a flow-viz technique (albeit using much simpler equipment), I thought of something I had seen in Formula 1 testing...

FLOW-VIZ PAINT

Image Courtesy of motorsport.com

Though it occurs to me that Dr. Doig and the wind tunnel crew might not appreciate the aftermath of this awesome technique. I’ll keep thinking...

[ekochaero-blog]

Looking at results...

Finally, let’s take a look at some plots created using data collected during both of the experiments conducted thus far. For the first experiment,  in which we measured the pressure at ports on the wing itself, the  resulting Cp plots look reasonable: 

For Low Speed (12 m/s)

For High Speed (20 m/s)

Gifs Used with Permission from Jordan Rabold

As previously mentioned, we had some discrepancies between trials, and this will be investigated later via error plots. For now though, we can see that the averaged data complies with expectations. As the wing starts from a downwards-pointing, negative angle of attack, there is positive pressure on the upper surface of the wing, and negative pressure on the lower surface. As the wing rotates up to positive angles of attack, the pressure on top drops off and the pressure on the bottom drops off. This would suggest that lift is being generated at positive angles of attack, while a bit of downforce is generated at negative angles, which makes sense. Comparing the two speeds, you can see that the curves follow the same general shape. However, the exact values are greater for the higher speed. This is believable, since higher flow rate is accompanied by higher pressure.

Now if we take a look at the data from the wake rake experiment, we can see how the flow behaves downstream of the wing:

 For Low Speed(20 m/s)

For High Speed (20 m/s)

Gifs Used with Permission from Jordan Rabold

Clearly, the low speed flow results in a greater loss of dynamic pressure in the space immediately downstream of the wing, when compared to the higher speed results. This can be explained in a similar way to the flow over a cylinder at higher speed compared to lower: the flow stays attached longer at high speed. Therefore, there is less of a pressure drop in the wake at higher speed. Also worth noting is the fact that there is a visible boundary layer at the top and bottom on both plots. This is most likely due to wall effects in the tunnel itself. While wall effects are generally not ideal to have around, the good news is, since the rake data quantified the boundary layer created by wall effects, we now know that it is not having a significant effect on the part of the flow we care about most since it does not extend all the way. However, since these are non-dimensionalized plots, it may be prudent to investigate this further with error quantification.

[ekochaero-blog]

On Wake Rakes...They’re Not Gardening Tools

While the data we have collected thus far is being processed and arranged into beautiful plots, I thought I might present a real life example of the wake rakes used in this experiment. Being a fan of both aerodynamics and motorsport, I immediately thought of how formula 1 teams use wake rakes during testing to get on-track data about the flow over the car. Here are some examples of this application:

Photo Courtesy of planetf1.com

While slightly different in appearance from the rake used in lab, this unsightly fence-like object is commonly used by teams in pre-season testing:

It’s all part of that process of verifying data from CFD with Wind Tunnel testing and, perhaps more importantly, Real-World testing.

[ekochaero-blog]

Just Wingin’ It with a NACA 4412 Airfoil...

...except with quite a bit of planning, organisation, and precise execution. This time we move on to the latest foray in the wind tunnel lab: analysing the flow over a NACA 4412 airfoil (AKA a wing).

While we will be working with several different experimental setups for the wing, for this first round, the experimental airfoil extended across the entire test section--much like the cylinder. Since it was left up to us to decide on our testing points, we decided to test at two different speeds(12 m/s and 20 m/s), and at numerous angles of attack for each speed ranging from -4 degrees to 18 degrees. In case you were wondering, these points were not just chosen at random out of a hat or a complex random number generator in Matlab. For the speed, it was important to do a bit of calculation before deciding on the numbers that will also be used for the later experiments. Since we will be testing a couple different wings, which are made of different materials with different strengths(one aluminum, one foam), we needed to be sure we could match Reynolds numbers--since our goal is repeatability of results--while also testing at two speeds that were different enough without exceeding the max speed of the foam wing. After all of this careful consideration, we decided on 12 m/s, which would allow for the best Reynolds number match later, and 20 m/s, to get some higher speed data. As for the angles of attack, we chose to start at the zero-lift angle and increment all the way to approximate stall angle, both of which were found in Abbott and Von Doenhoff’s Theory of Wing Sections. 

We will now take a break from this long wall of text with a glorious picture of me setting the angle of attack like a pro(seriously, I’m adding a second page to my resume just for this task):

On the subject of setting the angle, the method for actually setting it was far from exact, and at the higher speed, the assembly tended to shift ever-so-slightly when the clamps were locked down. This might have led to some margin of error in the actual angles being tested at--not so much a problem for a single trial, but when you’re averaging data between multiple trials, this might have some effect on the results. On that note, my task for the second trial of this experiment was once again at the Matlab station. Since we had prepared code ahead of time, though, this task was slightly more involved as it consisted of checking the data and comparing the second to the first trial. Here was where it got interesting. There was some not insignificant error between the first and second trials for some of the pressure ports. This could have been due to problems with the ports themselves, changes in ambient conditions, error in the set angle or speed, or something else. Look for plots in later posts to see whether this affected our end results.

[ekochaero-blog]

Final Comments on the Cylinder Lab

Now that my report for the lab has been submitted, there are a few comments I’d like to offload before moving on to the next great adventure in the wind tunnel lab. First, I should probably present the plots for Cp that we came up with:

There is a lot of analysis that could be done on these plots, and most of that can be found in the report. What I would like to say through this medium, however, is that I was really impressed with the quality of the data we collected from the tunnel, despite some anomalies that might be explained away anyways. This serves as a reminder, to me at least, that we have the opportunity through this lab to work with professional-quality equipment that is truly capable of accurately measuring the physics of flow. And that’s pretty cool if you ask me. I am very excited to move on to more complex experiments and put the tunnel to even bigger tests.

[ekochaero-blog]

Analysing the Flow over a Cylinder: Part 3

As discussed in the previous post, we can predict the flow characteristics downstream from the cylinder by calculating the Reynold’s numbers corresponding to the flow speed and reference our figure from before. A bit of Matlab magic et voila: our Reynold’s numbers are 1.6285e+05 and 3.2569e+05 for velocities of 15 m/s and 30 m/s, respectively. After referring to our figure, we might predict that for the slower speed, the downstream flow is asymmetric, while it is possibly asymmetric for the faster speed. Of course, this prediction assumes that some nice-looking figure from the world-wide web is to be wholly trusted, and that their ranges of Reynold’s numbers corresponding to each characteristic flow are exact. It could easily be that the slower speed also fits into the same range as the faster. What we definitely can glean from our calculated Reynold’s number values is that both wakes are fully turbulent.

We may not be able to bank on predictions made by comparing numbers with pictures, but that is where the glory of this lab shines! We are dealing with a wind tunnel here after all. After collecting the data required for our reports, we were left with enough time to do a bit of flow-vis using a fog machine so we could see what was actually going on past the cylinder.

You can see in the gif above that the flow seems to separate early, and is clearly fully turbulent downstream. While it may be due to how the fog machine hose was being held (i.e. not positioned exactly at leading edge), it does appear to not quite be symmetric. Perhaps the internet told the truth in this case...

In this gif, you can clearly see that the flow is staying attached much longer before separating and transitioning to turbulent flow. Again, it may be down to how the hose was held, but this time the flow much more closely resembles symmetry.

Much more analysis and interpretation must be done before coming to concrete conclusions, and I am excited to see how our data holds up to predictions.