Untitled

[ademaree28-blog]

Last Post!

Well, here we are. Last post of the quarter. So the first piece we printed came out ok. The layers split a little, but I think with some putty it’ll be usable. The second piece wasn’t so successful. The printer is having a problem with the heating element, so it stops about 20 minutes in every time. The other printers aren’t working so well either, so we won’t be able to have the fairing manufactured, but we ‘ll make sure to include documentation for doing that later. Here’s some pictures of the prints.

The most memorable moment in 307 for me was making the wing tips for the winglet challenge. I had never used a 3D printer before this class, so it was excited to take a design, prototype it using the printer, then get to test it in the tunnel. I really enjoyed this because it felt like for the first time in a lab, every group wasn’t doing the exact same experiment.

The most challenging aspect of the quarter was correcting for error coming from the equipment, determining what could be causing the error, and then finding a way to still get something meaningful out of that data. It was a pain, but I think it’s not the last time I’ll be doing it.

I think the team dynamics were pretty good overall. I didn’t have any issues with team members doing little to nothing and leaching a grade off of the rest of the group. I do have a couple of general takeaways though. One is that in the future, it would be useful to have some understanding of different strengths in the group. Often, we would divide up tasks for efficiency, but one team member would sometimes spend a very long time on a single task, which could have possibly been done much quicker by another. In the future, I’ll try to get a sense of what other team members are good at/like to do, and attempt to divide tasks accordingly. If someone ends up with a task that isn’t a strength of there’s, it’s not necessarily a problem, but it could be good to check in with them to be sure they’re not stuck and wasting their time. Another thing would be to make sure we check over each other’s work to ensure we all feel things are being done correctly and up to our standards.

[ademaree28-blog]

Final touches

Our test piece printed pretty well, and it looks like it should fit the traverse just fine. A couple of problems we ran into were that it was a bit tight at the back of the traverse plate, and we also didn’t accommodate a couple of bolts running along the front of the traverse plate. I’m glad we decided to print the test piece instead of wasting time and filament on a full scale one.

Next, we went to work updating the model to deal with the problems. The next thing to consider was mounting. We used the engineering drawings from the traverse documentation to find mounting points, but on further thought decided not to print the fairing with the mounting holes. Since our model was going to be printed in multiple pieces we were concerned that the holes wouldn’t be in the right spots when we were done. Instead, we decided to make some circular cuts where we could seat the bolts and washers without interfering with airflow. Once the pieces were assembled, we could drill the required holes.

Here’s the final model with end pieces and the guide holes. The model we be glued into 3 pieces: two on the side where it mounts to the x axis and one on the opposite side. With the model finished, we went ahead and started printing the first piece, which should take about 13 hours.

[ademaree28-blog]

First crack at the fairing

After some further discussion, we decided that 3D printing would be the best option for manufacturing our design. This would allow for the design to be easily be changed or remade later and also ensures that the geometry is known and can therefore be modeled in CFD.

With that decision made, we went to work on obtaining detailed dimensions of the traverse and drawing a top view of it.

Here’s what we came up with. Rather than trying to fit the geometry of the traverse exactly, we decided that the best way to design the fairing would be to generate an airfoil geometry that can accommodate the worst case top profile of the traverse, then extrude this profile through the solid fairing.

Here’s the basic model we came up with. The actual fairing will be much longer, but we decided to create a thin strip like the one above as a test piece. The plan is to print the model above and test it out on the traverse to make sure it isn’t going to interfere with anything.

[ademaree28-blog]

Starting work on the Traverse Shell

After receiving the original design report for the traverse and a list of project requirements, we were ready to begin working out a design. A summary of the requirements provided by Reiss were that the fairing should cover the entire z-axis assembly without impeding motion or requiring disassembly of the traverse. It should also be usable at any operating conditions and removable.

Here’s a couple of pictures of the traverse z-axis:

Note the large black track in the above pictures which is being used to cover pressure tubes. This is the thickest part of the traverse and will largely affect the size of our design. After some discussion, it was decided that our design could assume a track about half this size.

In order to ensure the fairing could actually be manufactured using 3D printing, we decided to make a simple preliminary design and load it into Cura. The design is essentially an airfoil with dimensions large enough to encompass the entire top profile of the z-axis and a large rectangular cut for the z-axis to reside. Print time in Cura came out to about 40 hours, minimum. I’m a bit concerned about that because with everyone else using the 3D printer, I wonder if we’ll be able to get that much time on it before next Thursday. That’s also assuming that everything goes right with all of the prints. The design did actually fit on the print bed though, which we were concerned about.

Based on the print time concerns, we started to brainstorm some other manufacturing ideas. One would be to use aluminum sheeting and have it bent into an airfoil shape. We could then 3D print brackets to mount with. A concern with this is that the shape of the fairing may not be uniform, making it difficult to model in CFD software. Also, we have to worry about obtaining the material, and a damaged or imperfect design couldn’t be fixed by just pulling up the old CAD model and reprinting. We also thought about using foam or wood. The wood seemed much too heavy for practical use, and the foam would likely have issues with strength and mounting.

[ademaree28-blog]

Miniprojects

Flow viz projects were presented today, so I started work with my new team on our miniproject, streamlining the traverse. There was some initial confusion, since most of us were picturing the streamlining being the horizontal portion of the traverse that slides in and out of the tunnel. The portion we’ll acutally be working with is the z axis. With all of the moving parts and complex geometries, this is gonna be a lot more difficult, but we'll manage.

We spent the lab time measuring the traverse and getting a sense of the shapes and movements we’ll need to accommodate. Considering the complexity of the shape, we’re likely going to 3d print the design. We noticed that since it needs to slide up and down on a track, we’ll likely need to have a large gap in the front of the shell that we design. It seemed to us that the air flowing in through that gap would be a bit of an issue, and we’re not sure how to deal with it yet. One idea is to add weather stripping to close the gap, but we’re not sure if that’ll be enough. More next week

[ademaree28-blog]

Finishing flow viz

We decided to go with the roof rack for our flow viz and finished up the last of our videos. Here’s a quick peak.

It looks like the flow is still separating with the roof rack, and the wake actually appears to be quite a bit larger, so there’s probably more drag. Also, we’re seeing the same Von Karmon vortices from the cylinder lab! This shouldn’t be a surprise since the racks are cylinders, but I imagine this will likely mean the racks are noisy.

Also, after discussing it with instructors/TA’s, it seems like the reason that there is so much more separation than we expected is that the flow is extremely low Reynolds number. At real world speeds and dimensions, separation would be much further back. Based on this, I would guess the same is true for the Von Karmon vortices.

[ademaree28-blog]

On to flow viz

With our report turned in, we began playing around with some ideas for flow viz. my group was interested in the water channel, because we’ve seen some really cool videos taken with the dye in the water channel, and they look a lot better than some of the videos we’ve taken of smoke in the tunnel.

We decided to use the Tesla model and spent most of the period trying our different amounts of dye and figuring out the optimal distance to put the model at, etc. Here’s one of the videos we took.

It definitely looks pretty cool if nothing else. It’s got separated flow behind the car, which we expected, considering the blunt rear of the car, but it looks like separation occurs at the top of the hood, which we didn’t expect.

We’re not really sure what we’re going to do for the final flow viz project as of yet. We tried putting something in front of it to simulate drafting another vehicle, but the result was just dye everywhere due to the turbulent wake of the leading object. Some other ideas we have are to attempt some drag calculations usig Matlab’s image processing, attaching a spoiler or something to it to see the effect, or adding a roof rack to the top of it. I imagine the Matlab one is going to be more time than we have, and it might still not yield meaningful results, so probably not that one.

[ademaree28-blog]

Finishing up the report

Halfway through week 7, and we’re still hard at work on the report. Most of the results and preliminary stuff is complete, so now we’re just working on touching up after some of the comments on our report and we’re still dealing with the force balance data. Accounting for the significant drift in the load cell from the first couple of labs has been challenging, and some of our stuff still doesn’t quite make sense afterwards.

I mentioned in an earlier post that I wasn’t sure about some of the pressure data either. After looking in to it some more, it seems like some of the pressure tubes were leaking a bit, which caused them to read higher pressures than they should have been. In the end, we decided to ignore those ports and interpolate the pressures from the nearby ports, which gave us much more reasonable Cp plots.

For the new plot, the matches xfoil pretty much perfectly. The minor offset seen is expected except for the low pressure at the trailing edge. I’m not actually sure what’s going on there, but I have a couple of theories. The first is simply that assuming the pressure at the trailing edge is just the average of the two nearby ports may just be wrong. The second is that this is caused by viscous effects which our xfoil analysis didn’t account for.

Also, after processing our winglet data, it looks like we did see some improvement!

The winglet yields a maximum lift coefficient of about 2.1, compared to the control with 1.78. The efficiency in terms of L/D is the same for both of them, but if the same efficiency can be achieved with a higher magnitude of lift coefficient, that still seems to be an improvement. It seems to me this could allow for lower take off speeds/shorted takeoff distances for the same size aircraft if nothing else.

I really enjoyed this experiment, since this is the first time full design process I’ve been through at Cal Poly, from concept to testing. If I could go back and do it again, I’d like to have another control case which would be just increasing the span of the wing. The design we went with resulted in an increase span of about 120mm, and we know that that should always improve efficiency. A good winglet design should be better than an equivalent spanned conventional wing, so it would have been helpful to be able to compare to this.

[ademaree28-blog]

4412 Reflection

We’re all done with our testing, so all that’s left now is to write up the report. Processing all of the data from the last 4 labs is a bit of a nightmare. We split up the work throughout my group, but even so, it takes about a day of work to turn all of the data from one of the labs into all of the plots and information we’d like to discuss.

The most interesting aspect of the past couple of weeks has been the winglet testing. I found it a lot more interesting to get to test out something we had designed to see if it has the desired effect than to test something like a NACA airfoil, for which I can find any information I need online. I also found this experiment a bit easier, as I could easily grasp the end goal of the experiment.

Overall, I think the lab has been a good continuation of 304. I haven’t found any of it to be far more difficult than 304, but the workload has certainly increased. Some of the most important skills I think I’ve gained from these labs are communication/working with a team, analysis/quantification of error, and technical writing.

[ademaree28-blog]

Week 6

We got some winglet testing in this week using the force balance. The force balance is definitely giving much more consistent data this time. Turning off the linear drive eliminates a lot of interference, so our results should be very usable. It was a lot more enjoyable performing a lab when we felt like the data we were gonna get out would actually make sense.

Before even going into actual results, we noticed from looking at the labview readouts that our design is definitely increasing the lift by 2-4 newtons, which is exciting. Most of this likely comes from the extra 120mm of total span that it adds to the wing. Without looking at the data, it’s not clear what the effect is on drag.

We also got some cool flow viz using the smoke machine, which was my favorite part of the lab.

We were a little disappointed that we didn’t get any flow viz without our winglet to compare, but the visualization was still interesting. You can see the vortices shedding off the tip of the wing. This is what wing tip devices are intended to prevent, since it results in induced drag. The fact that they’re still present means that our wing tip device could not completely eliminate induced drag, but it’s possible that it did reduce this effect.

[ademaree28-blog]

Pitot Probe

Today we were looking at the infinite wing again by using the pitot probe to get the total drag behind the wing. This would have been a lot simpler with the pressure rakes, but we were having problems with them, so we just used the pitot probe with the traverse. It will work the same and likely get more consistent data, since all of the pressures are from the same pressure probe. Unfortunately, it means there were over a hundred files for us to go through. Below, we’ve got a preliminary plot of our data.

The data looks really good for this experiment and is pretty much doing exactly what we expected. By using the loss in momentum, which is the area between the curves, we can calculate the total drag on the wing.

Once we finished in the lab, we finally set out to get our winglets printing on the 3d printer. Here’s our finished product.

Since we printed both wings at once, we were a little worried they wouldn’t come out right, but they’re pretty much perfect. we’re excited to test tese out next week!

[ademaree28-blog]

Today, we spent most of the lab period developing a model for our winglet and preparing it for 3d printing. We found it pretty difficult to model using standard cad software like Solidworks, so we ended up using Vehicle Sketch Pad and exporting to an stl file. We went with a simple Hoerner wing tip design as discussed in a previous post. Attached below is a shot of our model.

It’s not as fancy as some of the other winglets I’ve seen the other groups doing, but I think that at these low speeds, it’s going to be just as if not more effective than the others. The increased span alone should yield greater efficiency, but I’m hopeful that the Hoerner tip will limit the spanwise flow a bit.

As promised, I’ve also got some plots from the infinite wing data. ere’s one below for the wing at 20 m/s and at 2 degrees AoA.

Looking at it, it doesn’t look completely ridiculous, but it doesn’t quite look like I expected based on Cp plots I’ve seen elsewhere. For one, there’s a weird spike, even after removing a known bad port. Also, at that spike, the pressure is greater than one, which is impossible for the assumed incompressible flow.

[ademaree28-blog]

Infinite Wing Data

Today, we hit the lab to get data for the infinite wing. Data collection was a lot simpler for this experiment, so we were actually able to get all of the runs we needed in one go. I don’t have any plots of the data to display at the moment, since we accidentally uploaded the wrong folder to google drive, so I’ll have some of those in the next post.

In the mean time, here’s some sweet flow viz we took!

Since our focus in our experiments is on max L/D, the flow viz is pretty boring. The smoke pretty much flows over the wing in the same way for all of the angles we tested at, and there’s nothing interesting going on in the wake. In this video, we decided to slowly ramp the angle of attack up to see separation just to get something interesting happening with the flow viz. Surprisingly, flow didn’t appear to separate until about 17 degrees, which was much higher than I expected. In Theory of Wing Sections, separation appears to occur at about 14 or 15 degrees.

[ademaree28-blog]

Infinite Wing Prep

This week we’re all done with the force balance, and we’re starting on the infinite wing with pressure probes. Since the other groups were in the lab, we focused on prepping our matlab code for the lab. After about an hour and a half of struggling with indexing and other issues, our code is now plotting airfoil shape, the pressure coefficient over the chord, and calculating the drag and lift coefficients. As far as the drag and lift coefficients go, there’s really no way to verify that our code is right as of yet, since the sample data is wind off. Here’s a screenshot of the pressure plot.

It’s evident from this that port 8 is giving false readings, so unless it gets fixed, we’ll likely be ignoring this pressure in our calcs.

I also did some more research on the wing grids that I discussed last week. According to an experiment performed by some MIT students, the wing grid shows no benefit until a critical Reynold’s number is reached, and the critical Reynold’s number is much higher than any we are testing at. Additionally, based on some discussion on pilotfriend.com, it seems that the most effective tip design might be just an increased span with a sharpened tip. According to the site, most other parameters such as twist, dihedral, sweep, thickness, and chord length have more significant effect on stall than on max L/D.

[ademaree28-blog]

Wrapping Up the Force Balance

No time in the lab today, but one of the other groups is going to supply us with the final data that we need. Since we weren’t in the lab, we went to work on processing the data for the blue wing. Once that was done, we also tried to quantify some of the error in our data collection. Below are the plots we obtained.

CL vs alpha for all runs

CL/CD vs alpha for all runs

For the CL vs alpha plots, the data looks reasonable. CL varies almost linearly with alpha, which is what’s expected for small angle of attack. Error also appears to be pretty small except for the second run at 10 m/s and 4 degrees. An unexpected feature of the graph, however, is that CL decreases between -2 degrees and 0 degrees, which doesn’t make sense. Also, the lift coefficients seem unusually high, since xflr5 was predicting CL of less than one for all AoA tested. Other groups seem to have similar scaling discrepancies in their results.

As for the CL/CD plots, these really don’t look right at all. xflr5 predicted max CL around 11, which is much higher than any of our experimental results. Additionally, the propagated errors here seem way too large. There are only a couple of operations that went into these calculations, so I don’t think they should be this big. Next week, we’ll need to do some analysis to determine if we made a mistake somewhere in the error propagation.

For our winglet design we found something really interesting called a wing grid. Here’s a picture of one:

I believe the concept is similar to the split scimitar, but it appears to be quite a bit easier to machine. According to some research done by Dr U. La Roche, this configuration could produce similar lift and efficiency of a wing with almost twice the span and dramatically reduces induced drag. More on this next week!

[ademaree28-blog]

Wing 2

Today we got back into the lab to test the blue wing. As with the red wing, we didn’t have time to get our last runs at -4 degrees angle of attack, but hopefully we’ll have some extra time later to knock those out.

We did our testing at the end of the class period, so there wasn’t time to look at any plots, but my expectation is that we’re going to see a lower L/D for this wing as a result of the lower aspect ratio. As shown below, that’s also what the xflr5 polars predict.

Analysis of red wing @ 10 m/s

Analysis of blue wing @ 10 m/s

The data we actually obtained from the red wing is not matching xflr5. Our lift coefficients are significantly higher than what is predicted by xflr5. Most other groups seem to have this issue as well, so I’m guessing there’s an issue with the forces being read by the force probe. The L/D seems within the realm of possibility, but definitely less than predicted by xflr5.

Cl vs alpha for the red wing from testing

L/D vs alpha for the red wing from testing

On Thursday, we’ll hopefully get our final data, get plots for the blue wing, and possibly do some more winglet design.

[ademaree28-blog]

Testing!!

We finally got the chance to get in the tunnel and start testing. Yay! It was a slow start due to some accuracy issues with the stepper motor for adjusting angle of attack. We spent about 20 minutes at the beginning of the lab changing the angle and checking manually. We eventually figured out that the issue was backlash in the gears, causing an error of about .8 degrees. We were able to avoid this by always moving the motor in the same direction.

Once that issue was taken care of, testing went pretty smoothly and we were able to get all but one set of data that we needed. We’ll have to do the last set next week. We spent the rest of the lab trying to sort out the data and work out bugs in our matlab script to get some nice plots. Here’s one of the plots we’re generating.

This plot is definitely nonsense, since in this AoA range, the slope of the CL vs alpha plot should be linear and the lift should be increasing in magnitude. We’re gonna have to keep working on our code to see if we made a mistake somewhere because I don’t think the data should be this bad.

For next week, we plan to get our last set of data from this wing, then test the blue wing to see how they compare. We’re gonna work more on our code over the weekend to fix the bugs, then hopefully we won’t need to do anything else we the new data but add it to the folder and run the code.