arashadrangi-blog

[arashadrangi-blog]

Week 10 Blog Post

This week, my team and I put the final touches on our presentation for our probe mini project, and finished up the code. The remaining values that we had to find were the pitch angle and the velocity, and these values were obtained through analysis of our pressure coefficient data.

The plots used to obtain phi are shown above. The plot to the left allowed us to obtain phi, by using the slope of Cp_alpha vs phi, holding theta constant. Velocity was also obtained by using the stagnation port of the pressure probe at 0 phi, assuming all the inlet pressure to be dynamic pressure. This assumption was valid since the pressure ports measure gauge pressure, and the stagnation port at 0 degrees phi is facing directly to the flow. For future work on the project, someone else could figure out a way to calculate the velocity regardless of what angle of phi you are at, instead of doing and initial angle of 0 degrees to obtain stagnation port data. In other words, modify the code so that velocity calculations are not dependent on the stagnation port at 0 degrees. 

Reflecting on the course as a whole, an important concept I learned about aerodynamics as a result of the course was that small details in the shape of an object can greatly improve its aerodynamics as a whole. For example, extending the span of a wing can improve its lift to drag ratio, or adding a duct extension to the back of a truck can reduce its drag, by allowing the flow to stay attached longer and reducing the turbulent wake behind it. On that note, I had little understanding of what a wake was, nor did I understand how it relates to drag. Now, I understand that a larger size of the wake correlates with increased drag, and its size allows us to calculate the friction drag of the object.

One thing that still confuses me is why a wing loses lift when it stalls. From my understanding, once a wing enters stall, the control surfaces can no longer properly steer the aircraft due to flow being bypassed by the whole front end of the wing. The only thing that the control surfaces deflect is turbulent air, which provides little use for control of an aircraft. However, if the aircraft is still moving, it should still be producing lift. Perhaps I am misunderstanding what stall is; if it was the point that you began losing lift, that would make a whole lot more sense.

My 307 highlight was the flowviz challenge. I am interested in controls and acrobatics, so I was interested in seeing how maneuvers of quads would effect the aerodynamics. It was pretty cool to see how the wake behind a quad can change based on what maneuver you are performing (i.e yaw, pitch, roll), as well as providing my team and I insight on rotorcraft aerodynamics.

My 307 lowlight was probably the first lab; it was a deep dive into material that we hadn’t explored in a whole quarter, and some of it needed to be refreshed. For me, it was a bit of a rocky climb, but the climb was made nonetheless. The tricky thing with fluids and aerodynamics is that not all of it is immediately intuitive; after this course, I can now say that I could look at an object, such as a chair, and tell you why it wouldn’t be so aerodynamic. At the beginning though, not so much. 

As to what awaits for me in the aerospace world, I am still unsure of where my passions lie. My favorite course so far was aircraft dynamics, and I really enjoyed the computational programming side of our supersonics class. I also enjoyed the programming assignments for this class too, so I will probably end up somewhere along working in aircraft controls or doing heavy computational programming for supersonic/transonic aircraft. I am not solely set on doing aircraft as well; I enjoyed fluids as a whole, and would not mind working in other fields unrelated to aerospace, so as long as I get to be the one coding stuff up.

[arashadrangi-blog]

Week 9 Blog Post

This week, the team and I went into heavy analysis for our probe calibration mini project. We analyzed our data graphs for the two different velocities that we ran the probe at (5 and 10 meters a second), and determined pressure coefficients across the probes.

To determine calibration for the 5 hole probe, the plots for the pressure coefficients must be used. The pressure coefficients were determined by using the differential pressure between the high and low pressure ports (ports 2 and 4), and the two side ports (ports 1 and 3). Polyfitting an equation to the pressures would allow us to determine the two angles at which the free stream is acting on the probe. 

To calculate the pressure coefficients, the pressure ports were paired into ports 1,3 and ports 2,4. The pressure differential was determined, and then the value was non-dimensionalized by dividing by the stagnation pressure minus the average pressure. This formula was derived from researching previous methods for 5 hole probe calibration; the denominator term is essentially the deviatoric pressure, and serves mainly to non-dimensionalize the pressure differential while reducing the CP values to reasonable numbers.

However, we found that there were more creative ways to calculate values for our angles and velocity. The primary angles that we were concerned with were the roll and pitch angles of our probe. After finding these, the stagnation port at the tip of the probe could be used to find the velocity. To determine the roll angle, we observed the trends in our two pressure coefficient plots. We found that the CP values for both pairs of ports represented trigonometric functions, only differing by a phase shift. Dividing the two plots by each other yielded a -tangent graph, which was completely independent from phi. The graph is shown above. This gave us the information required to solve for the roll angle. 

In the coming week, we hope to devise a way to calculate the roll angle and velocity. Currently, we are theorizing that the slope of the pressure differentials could give us information about the roll angle, and that the velocity could be obtained from assuming that the stagnation port is entirely composed of dynamic pressure. However, both of these theories are not set in stone, and more analysis of data will have to be performed to derive formulas for these values. We also are preparing for our presentation in the week to come

[arashadrangi-blog]

Week 8 Blog Post

This week, my two teams and I focused on writing our force balance lab report, and conducting more tests for our mini project, which involves calibrating the 5 hole probe for the traverse. The probe project has become more involved, since gathering usable data has proven to be a difficult task.

For my teams for balance report, I conducted our data analysis and plots using matlab. The L/D plots for both wing configurations are shown above. Based on our analysis, we found that the high aspect ratio wing was more efficient at the reynolds number that we tested at (~150,000), which matched our initial predictions. Failure to factor in the sting interference would have skewed the data, as we can see the overlap on the graph. With our test matrix, we would not have been able to characterize the full range of performance of both wings if we did not properly account for the extra lift and drag caused by the sting. 

For our probe calibration project, the first week of testing (last week) ended up yielding poor results, due to the offset in our probe’s positions. This offset was caused by the probe being slightly bent, and the four holes on the outside of the probe not lining up properly with the angle that the probe was being adjusted about. This caused our pressure data to be skewed, and start at values that we were not quite ready for. We performed a second test this week, with better positioning for the probe (accounting for the bend and lining up the probe holes) and a higher test velocity of 10 m/s. Plotting the pressure data showed reasonable trends in terms of ports that were about the axis of rotation (ports 1 and 4) being the high and low pressure ports, and these two switched roles (port 1 going from high to low and port 4 going from low to high) at 45 degrees. However, ports 2 and 3 did not show these same characteristics, and remained close together in value for much of the test. We believe that this could be due to the bend in the probe causing the two ports to be closer to the outer region of the flow field, causing a lack of influence from the flowstream. This was determined after analyzing our test setup, after producing our pressure plots.

In the week to come, my team and I hope to perform more testing on our probe, and come up with a good calibration algorithm for determining the 3 angles at which the velocity is approaching the probe. Research online has proven useful at this point, with some algorithms already produced by graduate students and NASA researchers for calibration of 5 and 7 hole probes. We hope to base our algorithm off of theirs.

[arashadrangi-blog]

Week 7 Blog Post

This week, my team and I performed our force balance experiment, as well as starting setup for our mini project, which involves creating a matlab script for the traverse probe so that we can calculate velocities over each pressure port on the probe. 

For the force balance, we utilized the sting to measure a series of angle of attacks. We aim to calculate the lift and drag of the wing using the load data gathered from the new 3 axis load cell in the wind tunnel. Our team chose one reynolds number to test all our configurations at, and adjusted the speeds of the two different aspect ratio wings so that we could meet this requirement. One wing had a high aspect ratio, utilizing a long wing span with a shorter chord, while the other wing had a longer chord but shorter wing span. Based of current jet designs, I believe that the high aspect ratio wing will perform better in terms of lift and drag. My prediction stems from larger wingspan aircraft having better lift to drag ratio, due to the volume of air that they produce lift from. To factor in the sting interference, we ran the sting at 3 different speeds, corresponding to wing off and the 2 speeds for the wings. The sting lift and drag would be subtracted from the lift and drag of each wing. 

For the mini project, my team and I this week had to figure out a way to set up and test our probe system. We utilized a probe testing machine in the wind tunnel, and figured out which port corresponded to each of its tubing by taping all the ports over and running air over them. We then devised a mounting mechanism using large binder clips. After figuring out how we would test the probe, we then devised a test matrix. We plan only on testing a 45 degree range, since we can mimic the other 3 ports through symmetry. The figure shows our testing configuration. 

Overall, I feel good about this weeks testing, since we gained valuable results which matched our predictions (in terms of the force balance experiment). The high aspect ratio wing ended up producing more lift, and although it produced slightly more drag, it was not that much more than the low aspect ratio wing. For next week, we plan on writing the report for the force balance lab, and testing our mini project.

[arashadrangi-blog]

Week 6 Blog Post

This week, the team and I performed our flow viz presentations, and I was reassigned with a new team for our force balance experiment. For the force balance experiment, we took a variety of data using the three axis load cell, mounted on the sting in the wind tunnel.

The goal of the force balance experiment was to determine the lift and drag on two different NACA 4412 wings of different aspect ratios. Our team tested without the wing mounted, running the tunnel at 0, 15, and 19 meters per second. This test without the wing was ran to gain interference data from the sting, so that it could be subtracted off from the actual wing data. This was done because the sting will produce some lift and drag when the actual wing is ran. Data gathered from the test was simple enough to interpret. Unlike the previous load cell, the new one used greatly simplified things since you no longer need to break the components of each axis for the forces to get lift and drag. Instead, lift and drag are entirely reliant on Z and Y components respectively. This makes it far easier to observe wing characteristics during the test.

This weeks testing and analysis went well. Trends for lift graphs based on different angles of attack were sensible, since the lift curves increased as we increased the angles of attack. We chose to run both wings at the same reynolds number, in order to provide the most comparable data, so that only the aspect ratios would change the lift curves. To do this, the smaller wing ran at 19 m/s and the larger wing ran at 15 m/s. Although we have not yet plotted our data, we feel that the wing with the higher aspect ratio will produce more lift, due to more area/chord covering the test section of the tunnel.

During the coming week, we hope to analyze our data, produce lift and drag curves, and write our report. This experiment is a good opportunity to observe wind tunnel testing through load cells rather than pressure ports. Load cells remove some elements of calculation from the test, but can have issues with calibration, more so than pressure ports.

[arashadrangi-blog]

Week 5 Blog Post

This week, the team and I conducted one final test on our quads in the wind tunnel, and began working on our presentation. We hoped to gain more interesting information about quads and how the fog machine shows flow at higher angles of attack

At a higher angle of attack, the quad can be visualized to be “moving” backwards, since multirotor aircraft usually fly backwards by having their front props spin faster than their back, causing a moment that pitches the multirotor upwards. When this effect was visualized with the tuft grid, the team witnessed air that was passing over the quad being sucked backwards, creating a very interesting effect. Visualizing this with the tuft grid also yielded cool results, and we witnessed different flow characteristics for different maneuvers. This was the first round of tests where there was a more distinct difference in flow visualized when the multirotors yaw/pitch/rolled. Shown below is a snippet of the multirotor at a high alpha, performing a full throttle maneuver.

The team felt good about this weeks testing. The only issue was that the testing for this week was conducted in sunlight, making it harder to gain valuable data from the flow visualization. Despite this, testing high angles of attack for quad copters using laser flow visualization allowed us to see the recirculation towards the back end of the quad more clearly, providing interesting data to compare against the tuft grid.

For the week ahead, we plan on further expanding our report, and prepping for our presentation.

[arashadrangi-blog]

Week 4 Blog Post

This week, our team performed tons of quad copter wind tunnel testing, and encountered interesting phenomenon as well as various problems. Both made the wind tunnel testing interesting and amusing for the team.

The 2 main forms of flow viz that we ran for our experiment were the fog machine and tuft grid. For the fog machine, we used a laser to visualize a cross section of the flow, then used the floodlights to get an overarching picture of what was happening. We ran several iterations of our test matrix with each quad, with yaw, pitch, roll, high climb, and various combinations of the above to witness the best results that we could. A good amount of the maneuvers ended up looking not too different, however, we were able to catch some interesting snippets of flow visualization from our testing. Shown above are two images from testing: a high climb maneuver performed by our smallest quad model, and a high angle of attack configuration with high climb using the tuft grid. The laser shows the flow being pulled down below the quad at very high angles, due to the large amount of down force caused by the quad. The tuft grid allows us to visualize flow being channeled through the center of the grid, since the high alpha configuration quad was pulling air from the tuft grid zone and down through the quad, creating a large area of turbulence where the grid was present.

 The team felt great about this weeks testing overall. However, we would like to do another round of flow viz using the fog machine to get high alpha data. We only got the chance to obtain this data using tuft grids, and we believe that we can get even more interesting footage if we actually use the fog machine. It was harder to visualize the effects of the flow using the tuft grid, since we had limited time to assemble it and didn’t accurately predict the area that we wanted to have tufts on the grid. The fog machine is much more forgiving, since it can be adjusted quickly on the fly.

Looking into the future for our report, it would be awesome to implement some kind of simulation into our presentation (streamlines in matlab, maybe some cfd). It is difficult for us to gather more quantitative data for a flow visualization experiment since no pressure ports were really used. However, it would be interesting to be able to mathematically predict flow characteristics under the quad copter and above for each respective maneuver.

[arashadrangi-blog]

Traverse Blog Post

For the wake measurement lab, the wind tunnel traverse system was used to take measurements of the wake size across the wind tunnel. Setting up the traverse involved a good bit of assembly, a little bit of arduino troubleshooting, and wire manufacturing.

The traverse had not been used in a few months, and this was the first lab that the new VI and setup would be used for. The traverse arm had to be reinstalled, and all three axis were tested. Unfortunately, the Z axis was having issues due to the piece connecting the motor shaft and traverse arm being in poor operation. However, for the purpose of the experiment, only the Y axis was to be used, so the z axis was manually moved into position. The old RS232 cables used for the traverse were constantly breaking, so one had to be professionally manufactured. This involved a bit of wiring and soldering, along with using some store bought rs232 heads and mounts to secure the cable. Finally, the traverse needed a bit of troubleshooting on the arduino side, since a few wires got yanked out. However, using pin values found in the VI, I was able to figure out where the wires went. Finally, we had to test the traverse. Kyle, Cyrus and I ran through a test matrix of 40 points, spanning the borders of a general test matrix that people in the lab would run. The traverse operated well, and was ready to go for the lab!

[arashadrangi-blog]

Week 3 Blog Post

This week, the team and I worked on our tech memo for the NACA 4412 infinite wing experiment and created test plans for the flow viz challenge. For the flow viz challenge, we hope to test different quad copters, and use different methods of flow visualization.

Careful consideration was taken when planning our flow viz challenge tests. Our team decided to use the fog machine with a laser, and tuft grids to visualize flow behind different quad copters. We will also perform different maneuvers with our quads while it is held stationary in a mount, so that we can observe how the propeller movements affect the flow. The maneuvers we picked were 10 seconds of yaw, pitch, and roll, along with a high climb maneuver. As a bonus, we wanted to test our quads at different angles of attack, namely a 0 angle and a high angle (i.e 18 degrees), and observe how these angles might effect flow characteristics. Different models of quad copters were used for testing as well, so that we could observe any possible changes due to the geometry of the quad.

For the week of testing to come, we would like to meet the goals for our test matrix, and obtain good video from the fog and tuft grid testing. We also need to manufacture the tuft grid which we are going to use for wind tunnel testing. For our tuft grid, we will be using a wooden frame coupled with chicken wire to compose the mesh. Proper tuft grid spacing needs to be picked; tufts which are too close together will stick to each other and have difficulty responding to flow disturbances, while tufts too far apart wont show very useful information. We ended up deciding with 2 inch spacing on each tuft, and covering a 1x1.5 ft rectangle, which was shifted closer to the bottom of the grid to provide visual on downforce from the drones. Towards the end of the week, we got the chance to manufacture the final frame and wire for the tuft grid, shown above.

[arashadrangi-blog]

Week 1 Blog Post

 For our first week of testing, we observed pressure data across a NACA 4412 airfoil in an “infinite wing” configuration across several Reynolds numbers and angles of attack. From our data gathered, we hope to extract the Cl and Cd plots, and compare them with existing data for the airfoil at different Reynolds numbers.        

Picking our test matrix was the biggest challenge. We chose our angles of attack based on already known ranges for stall angle and zero lift angle as well as common Reynolds numbers that the airfoil has been tested at (velocities of 20, 25, 30, 35 m/s for the tunnel). We ended up testing for stall characteristics at angles between 12 and 18 degrees, and testing for zero lift between angles of -3 to -7 degrees. We picked these angles since the NACA 4412 is predicted to produce zero lift around -4 degrees, and is predicted to stall from anywhere between 15-18 degrees depending on the Reynolds number (based of experimental data online and XFOIL simulations).

This weeks testing was interesting. Compared to our XFOIL simulations, our stall occurred at angles 1 or 2 degrees more during the real tests. We predict that the XFOIL simulation may not have predicted accurately since XFOIL makes assumptions that disregard viscous drag, which will prevent the wing from stalling at lower angles. We also had time to check for the reattachment angle at a tunnel speed of 35 m/s, starting at 23 degrees, and the flow reattached at 14 degrees. At a lower Reynolds number, we predict that the flow would reattach at a higher angle, due to its lower kinetic energy, which will allow the wing to grip the air better.

Our team is looking forward to our wake refinement testing, which will allow us to factor in viscous drag in our testing. Testing purely from pressure ports on the wing will only tell us the pressure drag that the wing is experiencing. To determine the viscous drag, we will have to look at the wake size behind the wing, which is determined by how well the flow remains attached to the wing.

[arashadrangi-blog]

Week 2 Blog Post

This week, my team and I worked on observing the size of the wake behind the NACA 4412 airfoil in an infinite wing configuration, picking a constant tunnel velocity and testing a handful of angle of attacks. We hoped to verify our predictions for the size and location of the wake for each angle of attack. Observing the properties of a wing’s wake is important for determining the viscous drag of the wing at a respective Reynolds number and angle of attack.

The test matrix chosen for the experiment was based off our test matrix for week 1. We wanted to test at angle of attacks for two criteria: they had to be before the point of stall and match a dataset that we had taken during week 1. These criteria were chosen so that we could compare data from a previous run, and so that the flow would not separate. If the flow separates, we won’t be able to accurately observe a wake forming behind the wing due to the constant recirculation of flow and poor pressure data. Our team chose a tunnel velocity of 25 m/s, which produced a Reynolds number of 404974. The decision made in choosing this Reynolds number was due to its place in our test matrix; we gathered pressure data during week one across the wing at velocities of 20, 25, 30, and 35.

Our team felt good about this week’s testing overall; there were a few odd things that we did not expect. I was surprised by the sizes of the wakes. To accurately capture them, our lab TA, Kyle, instructed us to measure in 1 mm increments. In the end, each wake ended up being around 10 mm wide, which is rather small. We also noticed that the average Cp of the plots was 1.2, which is higher than it should be for free steam subsonic flow (about 1.0). We predicted that this was due to errors from the scanivalve readings.

For the week ahead, we plan to finish analyzing our data and writing our report. The final pieces of data that we need are the Cl plots and Cd plots.

[arashadrangi-blog]

A non convergent solution at an alpha of -4 degrees and the same reynolds number. We predict that based on our XFOIL results, the zero lift angle of attack will be somewhere between -4 and -5 degrees, due to the change in CL  from positive to negative between these angles.

[arashadrangi-blog]

NACA 4412 airfoil at an Re of 404E3. This is the alpha right before our stall angle of attack. XFOIL predicts a stall alpha between 16-17 degrees at our current flow conditions.