joshmoncino-blog

[joshmoncino-blog]

Post 10 (6/10)

This week, we gained access to the laser cutter and wind tunnel in order to make and test our disc cover. 

Kyle made the CAD design for the cover, which could then be uploaded into the computer to start cutting (the two images below are the model made in Creo). Unfortunately, there was an issue with the file format, so we had to revert to using an illustrator file. While this allowed us to cut out the sector and the two circles, it prevented us from adding a key. We were hoping to add some stability to our design by having the two ends of the sector lock into each other, thereby holding everything together even further and helping to reduce the size of the gaps in the disc cover. Also, it seems that the material we used was not ideal for laser cutting, as the speed required to cut through the material led to a significant amount of burning on the edges and surface. This is more for aesthetic reasons, though, so it had no effect on the performance of the cover.

Once we cut out the discs, we had to spend some time enlarging one of small center circles in order to fit it over the axle. In order to do this, we simply attached a filing attachment to the hand drill and filed away until the disc fit. As can be seen in the image below, the sector of plastic that was cut out to allow for the folding of the circle into a cone worked excellently. We were worried that the angle we calculated would result in a gap that was too large, leading to airflow inside the disc cover, but once the whole attachment was clamped together, the gap closed nicely to create minimal exposure to the spokes. 

After filling the inner circle to size, we drilled the holes for the screws. To provide maximum stability with minimum warping, we decided to use 4 sets of two screws at even distributions along the perimeter of the cover. The image below shows how the screw went from one side through the other, with a nut at the end. The tension in the plastic, which resulted from the transformation from circle to cone, in opposition to the force of the screws clamping the two covers together allowed for a very secure design, with next to no rattling. Furthermore, since the tension in the plastic is directed outwards, we did not have to worry about any undesired force on the spokes, as the screws take the entire load. 

With the disc cover successfully attached to the wheel, we now had to test the effect it had on the airflow around the wheel. We ran the wind tunnel at two different RPM’s: 100 and 200. These values correspond approximately to 7m/s and 15 m/s. Furthermore, four tests were done: stationary and spinning wheel with the cover on, and stationary and spinning wheel with the cover off. 

Below are images from the test with the cover off. The first image is a stationary wheel. Smoke was used to qualitatively measure the turbulence of the flow, while tufts were attempted for the dics-on tests, but turned out to give little useful information. 

The smoke stream contacted the tire and moved around along the disc cover. After the smoke stream contacted the tire, the flow moved around and through the spokes, coming out the other end in a highly disrupted state. This disruption indicates a large effect of the spokes and axle on the smoothness of the flow. 

With the wheel now spinning, it is further evident that the spokes and axle greatly disrupt the flow, resulting in an increase in pressure drag that would reduce the speed of the solar car.

With the cover now on, it can be seen for the stationary wheel that the smoke stream remains much less disturbed. There is a small decrease in smoothness, but the effect that the wheel has on the flow is greatly reduced.

It seems that, for a spinning wheel, the flow once again becomes significantly disturbed, but the qualitative measurement techniques employed give little indication as to how much less drag is produced with a disc cover attached. The best indication of the decrease in drag is for the stationary wheel shown above. 

My group and I would consider this project a success. We were able to make this disc cover out of cheap, light, and strong materials, and the wind tunnel testing validates our original belief that this cover would reduce the amount of drag produced by the wheel. We hope that this design, or some version thereof, can be used to help Prove Lab break the world record for the solar car land speed record.

[joshmoncino-blog]

Post 9 (6/4)

With the flow visualization project completed, I was finally able to dedicate all my 307 time to the designing of a disc cover for the Prove Lab solar car. To recap from the last post, the objective was to create a method of reducing the turbulence caused by the spokes of the front wheel as it spins in the fairing. While the fairing covers the vast majority of the wheel, leaving only a few inches of tire and rim exposed to the air, some air will still enter the fairing. This minimal, yet not negligible air movement created the need to minimize the drag caused by the movement of the spokes. 

Our group thought the most effective method would be a disc cover. We thought of this idea based on the disc covers seen on many professional racing bikes, as shown in the image below. 

We needed a material that was light, relatively cheap, easily cut, and flexible. On the first day of the project, we went to Home Depot and found a shatterproof plastic sheet that was big enough to allow for the circular cut with a diameter greater than that of the wheel, while thin and light enough to be practical for the solar car. The measurement of this cut occupied the next few days of work. With the limited tools available to use for measuring, we had to figure out ways to accurately measure the wheel’s dimensions. 

Below is a sketch of the dimensions we were able to determine. One issue was the change in diameter of the axle from one side of the wheel to the other. The first major issue was the measurement of the hypotenuse of the triangle shown below. This is essentially the path that lies flat along the spokes and is the place where our disc cover would rest. However, since the disc cover had to be bent from a circle into a cone, the plastic cut could not simply be the diameter of the wheel, as the distance covered by the spokes is larger than the wheel diameter. Furthermore, in order to be able to bend the disc at all, a circular cut-out had to be made around the axle, allowing both for room the axle to protrude and for the material to bend into the desired shape. 

The above image shows some of the early calculations that were made in order to determine the true circle diameter that would have to be cut out in order to create the shape. These calculations were not the ones we used in the end, as we went back to Prove Lab multiple times in order to remeasure in more accurate ways. However, the general process is about the same. By using the dimensions in the diagram, we were able to calculate the added length (x1), that would be required for the cone to be cut out. This x1 length was then added to the measured length along the spoke, and therefore gave the actual diameter of the circle. 

Furthermore, in order to be able to bend the cone at all, some material had to be removed from the circle. To do this we would have to cut a sliver of some angle from the center of the circle to the diameter. We were able to calculate the required angle by using the surface area change from circle to cone. The area of the circle is π*r^2, while the surface area of the cone is π*r*l. Since “r” for the circle would equal “l” for the cone, the surface area of the circle would always be greater than that of the cone. This loss in surface area would be equal to the area of the sector that would have to be cut out in order to create the cone. By using the area of a sector, A = (theta/2)*r^2, the angle of the sector could be calculated. 

And finally, the small circle that would have to be cut out about the center of the plastic sheet was approximated fairly roughly based on the diameter of the axle. Since the circle was not very large, the angle change would not be significant, and we knew that any errors could be corrected via sanding or filing later on. 

With all the dimensions worked out and the material ready, we now just have to wait for our scheduled time on the laser cutter in order to start the hands-on aspect of this project. 

[joshmoncino-blog]

Post 8 (5/27)

On Thursday we gave our presentation and I think it went pretty well. I was impressed by the depth of analysis that some other groups went into. One of the major difficulties for our group was analyzing in such a way as to be able to draw meaningful conclusions. Without many numbers to work with, the qualitative analysis was difficult to present in a useful way, without making the entire presentation sound like pure speculation. However, it was a great opportunity to learn about the Common Research Model. There were many designs built into the model that I had never heard about or even considered as important, such as intentionally causing stall at the root of the wing before the wingtip. Seeing how much more complicated a wing can get than a simple 4412 design was a nice change.

I thought that one of the more interesting presentations was the active flow control (I think that’s what it was called). I liked how it was something that is just now coming into industry, which made it seem very relevant. I’d actually seen the wing they used in their testing and presentation a few days before and remember wondering what the grey discs were for, so it was great to see my question unexpectedly answered.

Now we have our new assignment: to minimize the turbulent effect of the spokes on the wheel of Prove Lab’s solar powered car. Although the fairings will cover the vast majority of the wheels, enough of the rim and tire will be exposed to allow for some airflow to enter the fairing, resulting in turbulence caused by the rotation spokes. To minimize this effect, my group will attempt to build a disc cover, similar to the one shown below. This cover will be made of fairly rigid plastic, but will be flexible enough to be bent into the cone shape required to cover the spokes. 

In order to ensure that the disc stays on the spokes, small bolts will be used to clamp together the discs on each side of the wheel. We have not yet decided how best to cover the bolt heads in order to further minimize aerodynamic effects, as our previous idea of countersinking them will not work well with the thin plastic that we bought.

I think that this design will be fairly easy to make, as we already have the plastic and only need to create the cone shape based on the dimensions of the wheel. Due to the simple design, it should be finished within the time frame given, and we really hope it can be used by Porve lab to create a slightly faster vehicle. 

[joshmoncino-blog]

Post 7

This past week we finally wrapped up the report and did our next test for the flow visualization project. Tuesday was spent thinking of things to test. It was difficult coming up with reasonable project ideas, as most of what we wanted to do was to big of a project to finish within the short amount of time we were given. We were originally going to use a motorcycle helmet, covered with separation paint/oil. This would have been fairly simple and interesting, but we wanted to avoid ruining one of Kevin’s helmets. We were then told that we could use Matt Paul’s 3D printed Common Research Model. We were still a little limited as we did not want to ruin Matt’s model, so we finished Tuesday by submitting our proposal. This included running the wind tunnel at several speeds and angling the model at 4 different angles of attack (including 0). Rather than the separation oil, we settled for tufts and smoke to provide the flow visualization that we would later analyze.

We mounted a Go Pro on the ceiling of the tunnel and were able to capture several minutes of video, as well as individual snapshots. Below is the model at 0˚ angle of attack, with smoke aimed at the wingtip. No significant aerodynamic effects are noticeable at this angle of attack, but it is still a good reference point to which the other angles of attack can be compared. 

The next picture is at 10˚ angle of attack. At this angle of attack, spanwise flow is very noticeable due to the alignment of the tufts in the direction of the span. We theorized that there were several factors causing this kind of flow, but believe that the primary reason is the swept nature of the wing. Since the wing is not perpendicular to the freestream flow, some flow is guided along the length of the wing, causing the tufts to align in this way. 

Furthermore, separation can be seen towards the root of the wing, but has not yet become significant over the rest of the wing. At the root, the recirculation caused by separation can be seen through the curling of the tufts. Due to the washout of the wing, the root has the greatest angle of attack, resulting in separation at the root before anywhere else on the wing. This design is intended to prevent stall at the wingtip, which could result in a spiraling decrease in altitude for the aircraft. By ensuring stall at the root before the wingtip, stalling would cause the plane to drop, but not spin. 

And finally, while the tufts are the primary visualization method from above, the smoke gives excellent flow visualization from the side. At an angle of attack of 15˚, as shown here, separation is very evident. This smoke is being directed towards the root, where separation occurs first. This separation can be seen with a very clear cloud of recirculating smoke. 

There are numerous other observations and explanations that will be included in the presentation, but separation and spanwise flow are the main two concepts. 

[joshmoncino-blog]

Post 6 (5/13)

On Friday and part of Thursday, our group worked on finishing all of our analysis and some error explanation for the report. My job was to investigate the finite wing experiments, which was the blue and red wing testing. The very first day of testing this quarter was for the red wing, and we remembered that on that day, not only were we very new to some of the equipment used, but the strain gage in the load cell had broken, along with the added error of drifting that was a recent problem in the tunnel. As a result, we suspected that our data would be very shaky for the red airfoil, and while doing the analysis, I was able to confirm that the data collection from that day was very poor. Fortunately, however, the analysis showed that the blue wing data was reliable and a good representation of the 4412 airfoil.  

I first analyzed the CL vs. Alpha plot shown here: 

This plot shows relatively similar trends between both airfoils while running at both Reynolds numbers. The first strange thing is that the red airfoil at Re = 166,680 seems to stall at 10deg, something that the Theory of Wing Sections does not confirm, and which also stands out when compared to the other curves. The fact that all of the Cl values are incredibly high is a combination of drifting and our inability to account for the sting interference. The sting data that we received was at different angles of attack and different increments than our experiments, which would have required a lot of effort in extrapolation the data to suit ours. While this plot cannot be used to accurately predict specific values, the trends are unaffected by the sting interference, which is why we decided to spend our time in other areas and leaving this as a last priority.

Next, I used the Cd vs. Alpha plot to further understand what happened with finite wings:

The blue wing data is very much what we expected, as it is a relatively smooth curve throughout and never crosses the x-axis. The red data, on the other hand, is far less reasonable and includes negative Cd values, which makes no sense. As a result of this negative data, we were only able to apply the Oswald efficiency calculations to the blue wing, as the negative values produced imaginary numbers. 

I originally included Cl/Cd vs. Alpha plots, but when I asked Bradon to review my analysis, he said that I should replace these with drag polar plots. Incidentally, I am glad that I did, as the drag polar plots are much easier to interpret and excellently indicate the flaws in the red wing data, while showing the success of the blue wing. 

The following plot is for the blue wing only, run at two different Reynolds numbers. The tangent lines drawn indicate the maximum efficiency points, and since Cd and Cl are directly related to angle of attack, these points can be traced back to find at which angle of attack the airfoil experiences this max L/D. This data is all very similar to what was expected from a drag polar plot. 

The red wing drag polar plot, however, shows no reasonable trend. While the majority of the low Re number curve follows a similar trend to the blue wing, the entire curve is translated far away from where it should be, as well as having a clearly bad first data point. Furthermore, the high Re number case is far off from what was expected, with about half of the curve in the negative Cd range. As a result of this poor data, no max L/D point could be determined.

This encompasses the majority of my role in this report. My next task is to elaborate more on how exactly the errors in the lab resulted in the data shown above. 

[joshmoncino-blog]

Post 5

This week consisted of testing the rake and the winglets. With the rake, we intended to use the traverse to move the pitot tubes along the cross section of the wind tunnel. Since the infinite wing was vertically placed in the tunnel, this method of pitot tube placement would allow for measurements from the freestream above the airfoil, through the wake, and to the freestream under the airfoil. In taking these total pressure readings across the wake, we hoped to be able to measure the amount of energy loss due to friction drag. This would then help to distinguish the pressure and friction drag components of the total drag when writing our report. 

At the very beginning of our testing, however, the traverse broke, preventing us from moving across the test section in the way that we wanted. We ended up getting very little data, none of which was validated by a second trial, and we ended up not being able to use it in the report. We were able to use Ashton Demaree’s data instead. As a result of this equipment failure, I cannot say much else about the rake experiment. 

The winglet testing was much more successful. The picture blow shows the winglets attached to the upside-down airfoil in the tunnel. Our winglets were laser cut and attached to plywood extensions, so as to maximize the wing aspect ratio within the parameters of the project. This testing period was very short, as we only measured two angles of attack at two different tunnel speeds. The analysis performed later unfortunately showed that the winglets decreased the efficiency of the 4412. While the lift that was produced did increase for all velocities and angles of attack, the drag was also increased such that the ratio of the two led to a decrease in L/D. It is likely that the amount of induced drag produced by this size of an airfoil is small enough that the added parasitic drag that comes as a result of the winglets makes their use counterproductive. It is possible, therefore, that winglets only become useful for larger airfoils. 

[joshmoncino-blog]

Post 4 (4/28)

This week we successfully tested the infinite wing, completed the winglets, and decided how to move forward regarding the blue and red wing experiments. 

Infinite Wing

Our testing of the infinite wing was the first experiment that went very well, with reasonable results and next to no other issues. Paul Vankeppel’s design for the rotating system, giving the wing varying angles of attack, worked perfectly. After talking to Brandon about how this experiment used to be run, with two lab partners attempting to change the angle of the wing in the wind tunnel by the same amount from opposite ends of the test section, this new method seems like a massive improvement. We tested at angles between 0 and 25 degrees in increments of 5, running each angle twice to ensure repeatability, and running everything overall twice for two different Reynolds Numbers (400,000 and 550,000).

Below are some of the plots that we made representing Coefficient of Pressure vs. Location along the Airfoil:

These two plots are the first angle that was tested (0 deg) and the last angle (25 deg), indicating the transition that the airfoil underwent in terms of coefficient of pressure while moving through this range of angles. It can be seen that the max point of the curved line (the top of the airfoil) for 0 deg plot is at -0.8, while at 25 deg, the Cp is essentially -1 for the entire top of the airfoil. This increase in the negative Cp value indicates that the air is moving more quickly and at a lower pressure, which is to be expected for separated flow. Additionally, at this high angle of attack, the entire length of the airfoil is covered in separated airflow, resulting in a fairly constant Cp value. As to why the final converging value for Cp is -1, we do not yet know. It could just be a coincidence for this airfoil and these conditions, but it seems a little too ideal for that to be the case. 

Unfortunately, we didn’t get any pictures of the flow visualization, but we were able to see the moment at which point flow separated from the airfoil. Once we saw this, we decided to go back and test for two more angles, 16 and 17.5 degrees, as this was the approximate moment at which the separation occurred. Below are the Cp vs. Location plots the lie as close as possible on opposite sides of the stall point:

At 15 deg at a location around 0.05, the Cp values waver a bit, indicating an impending shift in the flow properties. Then, at 17.5 deg, the trend of the plot has clearly shifted from a smooth curve to a hard drop off and a more linear trend. It can therefore be deduced that the stall point is between 15 and 17.5 deg. 

Winglets

Our winglets are now complete. They were made based on a research paper that Sam found, with a small amount of area below the airfoil, and a much larger section above, both shaped like tip-less triangles. These are attached to plywood extensions of the airfoil, thereby maximizing the wing aspect ratio while staying within the experimental parameters. We are looking forward to testing them and seeing how they match up against the designs of other groups.

Red and Blue Wings

The issues with the Red and Blue wing experiments are still bothering our group. We have decided that, since drifting is known to have a very small effect on the final results, we will simply assume it to be negligible for the red wing. As for the blue wing, we will include drifting, because we have useable data to determine its effects, but we do not anticipate a drastic change to our results. We still have the same issue with the strut data, but we have decided to make it a low priority while the rest of the experiments are still being analyzed. If we have more time before the due date of our report, we will interpolate and extrapolate the given strut data to match our matlab script, and then be able to account for the effect that the strut had on our wings.

[joshmoncino-blog]

Post 3 (4/23/17)

This past week has been a combination of trying to figure out why our plots from the first test were off, doing a new test for the blue wing, and working on the winglets. 

The red wing plots have been our foremost concern because of their very high CL and Cd values. We know that an issue lies with the broken strain gage in the wind tunnel, which could account for some of the strange results. Along with that, when I wrote my last post, we had not yet accounted for drifting or the influences of the sting on the experiment. We learned that drifting occurs when the electronics in the load cell begin to heat up. This heat accumulates throughout the experiment, leading to changes in the voltage readings, which ultimately alter the forces that the computer sees. Due to the increase in heat, a constant load throughout the experiment can be perceived by the computer as an increasing load, thereby skewing the data. 

To account for this, we should have taken windoff data before and after every completed trail at each angle of attack. Doing this would have allowed for the comparison of the data points and, assuming a linear trend, this increase could have been accounted for by subtracting the difference from the data and averaging so as to create un-drifted data. Unfortunately, when doing the red wing experiment, we did not fully understand how to account for drifting, and therefore only took windoff data before beginning the trials at each angle of attack. As a result, we do not have the ending windoff data to compare against so we will have to figure out another way to approximately account for drifting. When doing the experiment for the blue wing, we did the windoff data correctly, so maybe we can assume similar amounts of drifting and apply this trend to the red wing as well. 

We also spent some time thinking about how the sting affects the overall reading on the load cell. We were able to incorporate symbolic sting data into our script for the red and blue wing in order to be ready for when the sting data came in from the TA. However, after receiving the sting force data from Brandon, we realized that our experimental angles of attack do not match up with the angles of attack at which the sting was run. While we ran from -10 deg to 15 deg in steps of 5 deg, the sting data has a range of 0 to 11 deg in steps of 2. Because of the mismatch in the dimensions of these data sets, we will have to interpolate and extrapolate so as to approximate the influence of the sting on the system without having to go into the wind tunnel ourselves and re-run the experiment for our own angles of attack. I am worried that this approximated method will be very poor and time-consuming, but it is the best that we can do at the moment.  

As of right now, I do not have the completed blue wing graphs to display in this week’s blog, but I can say that our experiment went very smoothly. As mentioned previously, we were sure to include the appropriate amount of windoff data in order to account for drifting, and now we need to incorporate the sting data so as to create plots that display the Cl/Cd vs. alpha trends for purely the airfoil. 

The one thing that seems to be going our way this week is progress on the winglets. I am not very involved in this aspect of the course as I am focusing more on the red and blue wing experiments, while Sam is doing the majority of the work on the winglets. 

We were originally planning on simply maximizing the area of the winglet both above and below the airfoil, so I am not entirely sure as to why Sam decided to change the design, as shown in the picture above. However, our general original design seems to be being followed, however, and hopefully the end result is a satisfying increase in wing efficiency.  

[joshmoncino-blog]

Week 2 Post 2

The lab went fairly well last Tuesday. We were able to get almost all of the data that we wanted, having been slightly under time pressure. The experiment ran smoothly throughout, and while we heard that there were difficulties with the load cell and other instruments in the wind tunnel, there were no drastic failures in the data, as far as I can tell. We did notice that the windoff readings of the linear drive fluctuated too often and led to some suspicion that the data would not be entirely reliable. Either way, we finished the lab without problems and, on Thursday, began analyzing. 

The analysis period was unfortunately mostly spent by trying to figure out how to avoid hard coding in all of the file names to each experiment. We had run the red wing at 6 different angles (from -10 deg to 15 deg) twice, to allow for averaging of data, and at two different Reynolds numbers. This, therefore, led to a significant number of .mat files that we finally were able to call in using a for-loop and create the graphs shown above. 

Unfortunately, we did not have time to discuss the meaning of these plots, only being able to verify with each other that the Cl vs. Angle of Attack plot matches the other three Matlab scripts that had been created in our group. Because of this, I hope my analysis in this post is not completely wrong, as I have not been able to consult with my group about the majority of my findings. 

The Cl vs. AoA graph makes the most sense to me. At negative angles, the airfoil creates negative lift, which then becomes more positive as AoA increases. Also, while this might not be true for all airfoils, it seems reasonable that a higher reynolds number would increase the overall lift produced by the wing. As the density of the experiment, as well as the length of the model, were kept as close to constant as possible, the only variation was velocity, and it is known that an increase in flow velocity generally increases the lift under a wing. 

I do not understand why the drag plots switch from negative to positive. I would have thought that, depending on the load cell’s coordinate system, the resulting drag would be all positive or all negative throughout the experiment. This is something I will have to ask my group about, as I am likely forgetting some crucial detail. However, what I do understand (I think), is why the difference between the first and second reynolds numbers is so large. The first reynolds number is lower than the second, and it is known that parasitic drag increases with velocity, while induced drag decreases with velocity. The low reynolds number case would indicate the dominance of the induced drag component over the parasitic drag. Then, with the increase in wind tunnel RPM and subsequent increase in flow velocity, the induced drag diminishes and parasitic drag becomes more prominent. However, since both experiments were run at fairly low velocities, the parasitic drag did not increase as much as the induced drag decreased, resulting in the very horizontal trend of the second reynolds number case. 

Lastly, the efficiency of an airfoil can be determined through the plotting of Cl/Cd vs. AoA. Each airfoil has a unique reynolds number at which it reaches its peak efficiency so this can be the explanation for the differing trends between the two reynolds number cases. However, I do not believe that the negative values in the graph are correct, and most likely come from the negative drag components mentioned earlier. I will look into this with my group to determine the most reasonable explanation or if the graphs have simply been calculated incorrectly. 

One thing to note is that we have not yet corrected for drifting, so this data is all raw, directly from the experiment. Correcting for drifting, and other factors we might find throughout our analysis, will hopefully lead to more reasonable results. 

[joshmoncino-blog]

Week 1

As there were equipment difficulties in the lab, we were not able to test during our lab time on Thursday. However, we have developed a procedure for the red and blue wing comparison, and have begun to discuss how to approach the testing for the infinite wing. 

The red and blue wings are identical in all but aspect ratio. Therefore, any differences that we notice throughout the testing of any variables can be attributed to the differing aspect ratios. 

Using a range of angles of attack (between -10 deg and 15 deg), we will examine the lift and drag generated by each wing. The measurements will be made using discrete angles of attack, as we are looking at forces instead of pressures, so a continuous sweep across this range of angles will not necessarily give accurate results. 

Each angle’s properties will be measured twice in order to maximize data collection within the given timeframe (if more time is available, we will do 3 trials at each angle). Additionally, the entire experiment will be run twice at two different Reynold’s numbers so as to examine the airfoil’s lift and drag under varying conditions. Due to the differing geometries of the two airfoils, the wind tunnel will be run at different speeds to allow for approximately equal Reynold’s numbers.

The analysis of this experiment will include Lift/Drag vs. AoA plots for each Re Number. Due to the only difference between the wings being their aspect ratios, these plots will allow for the understanding of the effect that aspect ratio has on L/D over a range of AoA’s.