progress

The following section is a description of what Ultra Flite has completed this semester. We have made adequate progress in all of our stated objectives. The following discussion includes an overview of each objective as stated in section two. Each subsequent section will present data obtained and procedures exercised to date.

5.1 Refilm

Ultra Flite has reviewed the prior semester’s video footage and found it to be inadequate. There exist problems in the orientation of the hummingbird in the top view of the video. Figure 5.1 depicts the setup from last semester. The mirror

Figure 5.1: Fall 1997 Test Setup Model Environment [3]

was not held correctly in place at an angle of 45° . This caused the hummingbird to appear hovering at an increased yaw angle, as seen in Figure 5.2. The mirror was not secured at an angle of 45° every time filming took place. Because of this problem, the orientation of the hummingbird, in the top view of the footage, was not in perfect symmetrical flight. Due to the imprecise orientation of the hummingbird, the video acquired from last semester was not sufficient since it allowed for too much error in the grid analysis of the hummingbird’s wing position. Therefore, a decision was made by Ultra Flite to refilm the hummingbird in hovering flight to attain a more precise record of its flight characteristics.

Figure 5.2: Previous Semester’s Digitized Model of Hummingbird in Hovering Flight with an Increase in Yaw Angle [3, p. 33]

5.1.1 Mirror Assembly

Figure 5.3 is a picture showing the actual test setup at the Zoology Department at The University of Texas at Austin. The mirror from last

Figure 5.3: Fall 1997 Test Setup [3, p. 21] semester can be seen above the flight cage. As described before, it is held in place by a string connected to a light fixture. Because there is no assurance the mirror would stay at the constant 45° incline, it prompted Ultra Flite to investigate a better and more reliable mirror assembly. Our team designed and assembled a very stable and more accurate mirror assembly. The following is a drawing of the new and improved mirror assembly shown in Figure 5.4.

Figure 5.4: Multiple Views of New Mirror Assembly The mirror assembly is made completely of wood, except for the mirror itself. The dimensions of the assembly are also given in Figure 5.4. In addition, Ultra Flite decided to enlarge the mirror to 14 in. x 14 in. This enlarges the total surface area of the new mirror by 52 in², which is approximately a 36% gain. The mirror assembly consists of two outer side panels and two smaller inner side panels. The decrease in size of the inner panels automatically makes a slot for the mirror to fit into place. Wood stoppers were placed on both ends of the inner panels to ensure the mirror wouldn’t fall out of place. Finally, a thin wood panel was placed on top of the mirror assembly to ensure that the mirror does not dislodge from the embedded slot. The most important aspect of building the new mirror assembly was to make accurate measurements and cuts on both the inner and outer wood panels. The main purpose of constructing a new mirror assembly was to ensure its reliability, consistency, and stability. In order to attain this accuracy, we decided to cut a square wood block diagonally which made a perfect triangle at an inclination of 45° . 5.1.2 Test Area Reconfiguration The original test setup was used for refilming with minor adjustments. The setup is seen in Figures 5.1 and 5.3. The clear glass box was used in the filming process to contain the hummingbird without obstructing the camera’s view. A feeder was placed near the top edge of the glass box to ensure the hummingbird image would fit in the viewing area of the camera. Capturing the side and top view of the hummingbird in hovering flight was a difficult task, but with mirror adjustments and camera placement, we were able to capture both views. During filming, additional lights had to be placed near the test setup as shown in Figure 5.3. We decided to use additional lights because the high-speed camera borrowed from Dr. Clemens, at the Department of Aerospace Engineering at The University of Texas at Austin, was especially sensitive. The final choice of camera will be discussed in the next subsection. In addition, we decided to place clear white paper on the bottom of the glass box. The video footage of the hummingbird was not clear due to the debris on the bottom of the box. During filming, the hummingbird, Felice, would not cooperate. Felice did not react to the filming environment as well as last semester’s hummingbird, Bob, did. Therefore, we decided to place live red flowers at the end of the feeder to simulate a real environment. In turn, Felice was more cooperative, and we were able to see the hummingbird in hovering flight for a short duration of time.
5.1.3 Post Processing and Choice of Camera

Using Dr. Clemens’ high-speed camera, we performed the initial filming of the hummingbird. Although the video of the hummingbird displayed adequate quality of hovering flight through the television monitor, the film did not seem to be sufficient for post-processing.

Ultra Flite had many complications using this camera. The first problem we encountered was that the camera shutter speed could be set at a maximum of ten-thousandth of a second, but no image appeared on the monitor. Although we thought the shutter speed was adequate, we were not able to film at the maximum shutter speed. We discovered that a shutter speed of ten-thousandths of a second did not allow for much light to pass through the camera to amplify the object in the picture. Another possibility for filming was that we could adjust the gain of the camera so more light could pass through. As a result, the object in the picture could be seen, but not with the clarity required. The image on the monitor began to lose resolution due to the increase in gain. Another possibility for increasing the resolution of the image was to change the shutter speed to a lower setting. We tested the filming until we found an adequate setting. Finally, we set the shutter speed to four-thousandths of a second and with a little gain added, the image was satisfactory. While we continued to film at the specified settings, we were not aware of the difficulties we would encounter. When the post-processing software was used to capture each frame of the bird in hovering flight, the image of the bird, especially the wings, were distorted. After looking at the still images of the film on the computer and comparing them to last semester’s footage, Ultra Flite decided to discontinue using Dr. Clemens’ camera for filming. We found that the resolution and quality of the still images were lower than prior semester’s images. In result, we will resort to using the high-speed video camera utilized last semester, which is made available by Dr. Ken Ball at the Department of Mechanical Engineering (ME’s) at The University of Texas at Austin.

Figure 5.5: Fall 1997 Still Image of Hovering Flight [3, p. 25] Figure 5.5 is an example of what last semester’s still image of the hummingbird in hovering flight resembles. With the ME’s high speed video camera, we will be able to capture up to 2000 frames per second. Although we already have the video from last semester, we cannot use their footage due to the uncertain angle between the mirror and the base of the platform. Therefore, we will have to refilm the hummingbird in hovering flight utilizing the new mirror assembly. 5.1.4 New Video Footage After examining the video footage taken with the high-speed camera provided by Dr. Clemens, we decided to use the high-speed video camera that was made available by The Mechanical Engineering Department. The coordination of the Mechanical Engineering, Zoology, and Aerospace Engineering Department was a great effort. Due to incompatible schedules among the three departments, a delay in the schedule allowed Ultra Flite only one chance to conduct the refilming of the hummingbird in hovering flight. The test setup of the cage and mirror assembly was very similar to the initial refilming setup. A few adjustments were made with the lights, mirror assembly, and feeder. For the second run of refilming, Ultra Flite engineers decided to place a white poster board on the bottom and back wall of the cage for better results. The addition of the white poster board created a better contrast between the hummingbird’s wings and the background. With the white poster boards, a back-lit illusion provided the necessary contrast needed to distinguish the hummingbird wing from the background of the test setup environment. The lights in the room were insufficient for the camera to capture a clear image, therefore, additional lights were placed near the test setup. The head lamps were placed inside the cage and directed towards the white poster boards. The two lamps used were sufficient in producing the amount of light needed for the necessary image quality. Precaution was also taken to make sure the temperature inside the cage did not rise above 32 ° C (~90 F). A thermometer and a small electric powered fan were placed inside the cage to warn us of the inside temperature and to cool the inside air, respectively. The mirror assembly was moved accordingly to capture the most surface area of the hummingbird wing motion. The feeder inside the cage was sometimes moved to another location to get better results of the hummingbird wing motion. After some experience, we learned that it was best to cover the feeder while shifting or repositioning the mirror assembly to allow the hummingbird to work up an appetite. Although refilming of the hummingbird took us approximately 3 hours, we were able to capture the video footage necessary for analysis. Before refilming took place, Ultra Flite engineers decided on a plan for refilming. Our concern was capturing all three views (side, top, rear) at once. After some discussion, we decided it was not necessary to capture all three views. The most important aspect of refilming was verification of the wing motion of the hummingbird in hovering flight. Therefore, Ultra Flite decided to capture each view individually and to discard any multiple-view footage. This change allowed us to concentrate on only one view at a time. This means more surface area of the hummingbird can be seen on the footage. The one-view-at-one-time approach will give us at least two times more area coverage which allows us to have a more clear and distinct picture of each frame captured. 5.1.4.1 Post-Processing of New Video Footage This semester, we decided to use super VHS while transferring the captured footage onto tape from the high-speed video camera. This option was not used in the previous semester so we thought it might have better. Each segment of the footage was downloaded individually from the high-speed video camera onto the tape. Two segments of the side view, one segment of the top view, and one segment of the rear view were downloaded onto the tape. Each segment of the video was approximately 4 minutes long with one frame per second which gave us multiple cycles of the wing motion. In order to obtain a computerized movie, we used Apple Video software to capture each segment. Connections from the VCR to the computer allowed for information transfer. The save option of the Apple Video allows for capture of the segment being played on the VCR. After creating each movie of the different views, individual pictures were obtained from each frame for analysis. Adobe Premiere 4.2.1 was used to create the individual pictures of each frame from the computerized movie. The following list describes the procedure necessary to convert the movie into individual pictures:

1. Pull down New Project Menu, Choose Presentation 160x120

2. Pull down File Menu, Choose Import

3. In New Project Window, Drag movie into the Construction Window

4. Remove Audio Stream

5. Pull down File Menu, Choose Export, Select Frame as PICT

After splitting up the movie into individual pictures and saving them as a series of pictures, there will be approximately 50 pictures made in PICT format. 5.1.4.2 Resulting Pictures and Movies of Post-Processing The follow figures are the resulting pictures created in Adobe Premiere, in PICT form. Figure 5.6 is a side-view picture of the hummingbird in hovering flight. In the picture, it shows the frame number, the number of frames per second during recording, and the number of frames per second during playback. With all this information, we were able to keep track of each image and its order in the wing motion cycle. In Figure 5.6, the camera is positioned on its side, therefore, the hummingbird is actually flying horizontally.

Figure 5.6: Side-View of Hummingbird in Hovering Flight The next figure is a picture of the top-view of the hummingbird in hovering flight. In Figure 5.7, the hummingbird is shown flying at a roll and yaw angle. The picture doesn’t depict a symmetrical hummingbird. This same

Figure 5.7: Top-View of Hummingbird in Hovering Flight problem was encountered last semester, and it was determined that the previous mirror assembly was the cause. After building a new and more reliable mirror assembly, the same problem was encountered. Therefore, it was determined that the hummingbird flies at a roll and yaw angle. Ultra Flite engineers believe that the hummingbird flies at a roll and yaw angle for various reasons. One theory is that the feeder has a large entrance which does not require the hummingbird to fly straight into the feeder. Knowing the feeder can be approached from any angle, the hummingbird does not have to fly directly perpendicular to feed. As a result, the hummingbird in the video is not positioned symmetrically. Another theory proposed by Dr. Chai is that the hummingbird is aware of our presence in the room. As the hummingbird feeds, her situational awareness keeps her from flying in a normal state. Figure 5.8 on the next page is a picture of the rear-view of the hummingbird in hovering flight. The rear view was not captured last semester, therefore, we decided to get the additional third view. In our research, no references contain a wing motion pattern for the rear-view of the hummingbird during hovering flight. With this new data of the rear-view, we were able to see the pattern the hummingbird wing-tip traces. Figure 5.8 also shows the hummingbird flying at a roll angle. In this particular picture, the hummingbird is slightly rotated clockwise.

Figure 5.8: Rear-View of Hummingbird in Hovering Flight 5.1.5 Results of Refilming The procedure to analyze the wing motion of each view consisted of using Adobe Photoshop 4.0.1 and Microsoft Excel 97. After creating the images from each movie, open Adobe Photoshop must be opened. The following list of commands was the procedure necessary to obtain the coordinates of the wing-tip:

1. Open Picture

2. Pull down Image Menu, Choose Rotate Canvas 90°

3. Keep a Record of Each Frame

4. Pull down View Menu, Choose Zoom In

5. Pull down Window Menu, Choose Show Info

6. Point to Wing-tip and Record Value

7. Repeat Procedure Until Cycle Completes

8. Repeat Procedure for All Views

After obtaining the coordinates of the wing-tip for one complete cycle in each view, open Microsoft Excel. The following list of commands is the procedure necessary to obtain a figure of the wing motion pattern:

1. Open New Workbook

2. Type in All Corresponding Coordinates

3. Select All Coordinates

4. Click on Chart Wizard or Insert, Chart

5. Choose XY Scatter

6. Choose Smoothed Lines w/o Markers

7. Click on Finish

8. Repeat Procedure for All Views

Figure 5.9: Wing Motion Pattern for Side-View of Hovering Hummingbird Figure 5.9 and Figures 5.10-11 on the following page illustrate plots of data points collected from the video footage. Figure 5.9 is a graph of the data points collected from the images of the side-view of the hovering hummingbird. The figure traces the wing-tip pattern as it cycles through once. The pattern in Figure 5.9 is not typical of accepted theory. Our hummingbird wing-tip traced out an airfoil shape pattern instead of the accepted figure-eight pattern. Figure 5.10 is a graph of the data points collected from the image of the top view of the hovering hummingbird. Figure 5.11 displays a half-oval shape. This is comparable to theory.

Figure 5.10: Wing Motion Pattern for Top-View of Hovering Hummingbird

Figure 5.11: Wing Motion Pattern for Rear-View of Hovering Hummingbird Figure 5.11 is a rear-view picture of the wing motion pattern for a hovering hummingbird. The pattern is not complete due to incomplete data recovered from the movie of the hummingbird during flight. With extrapolation techniques, one can safely say that the cycle will continue to connect with the beginning point. Using the rear-view of the hummingbird, it is confirmed that the side-view pattern is correct relative to the rear-view. The rear-view also confirms that the side-view does not trace a figure-eight pattern. Ultra Flite engineers examined the side-view multiple times. Each test came from different time cycles of the hummingbird during hovering flight. After reviewing the data from three trials of the side view, the same airfoil pattern appeared. There is no doubt our hummingbird wing-tip traced an airfoil shape pattern instead of the traditional figure-eight outline. 5.1.6 Theory versus Experiment In the previous section, our experimental findings were discussed. There exists a discrepancy on the side-view wing pattern motion. Our results show that the side-view wing pattern motion is shaped like an airfoil, as shown in Figure 5.9. After extensive analysis on additional footage of the side-view, we concluded that for our hummingbird, the side-view wing pattern traced an airfoil shape. Theory says that the side-view wing pattern motion should trace a figure-8 motif. The following is an excerpt from Weis-Fogh [12]: According to Stolpe & Zimmer's accurate tracings, the stroke plane is not horizontal but tilted by about 11° so that the wing-tip describes a horizontal figure-of-eight when seen from the side. Apparently Hertel had access to the slow-motion films from which the original analysis was made and confirmed that the movement is sinusoidal. In addition, he found that the wing is twisted like a propeller blade during both upstroke and downstroke so that each section along the wing-axis has the same instantaneous geometrical AOA and, also, that the average effective AOA is attained almost instantaneously at either end of the stroke and remains almost constant until the wing reaches the other extreme. Theoretical analysis has proven to obtain the figure-8 motif, as read in the above paragraph. The author of the article was not the original researchers to discover the figure-8 pattern. Instead, it was Stolpe & Zimmer who discovered the figure-8 motif back in 1930's in Germany. The original paper submitted by Stolpe & Zimmer was written in German, and we were not able to find the article explain their research procedure.

Although we found a different wing pattern from accepted theory, we cannot yet conclude that all hummingbirds side-view wing pattern motion is an airfoil shape. Therefore, we cannot yet conclude that theory is wrong. But our findings show that there is more research to be conducted so that our results can be confirmed or disproved. Ultra Flite engineer's believe that due to the fact that the original discovery of the figure-8 motif was back in the 1930's, there is a possibility that accepted theory is not accurate. To confirm our results, the use of new sophisticated technology to film the hummingbird is necessary.

5.2 Space-Frame Model of Hummingbird Ultra Flite is working to create a NASTRAN space-frame model of the hummingbird. In order to complete this task, Ultra Flite has been working on several different procedures. With a space-frame model of any object, one must have extensive knowledge of the internal structure of that specimen. There are, however, complications specific to a hummingbird. A Ruby-Throated Hummingbird, such as Felice, has an average body size of approximately three inches from head to tail [13]. Because of the petite size of the hummingbird, the bone structure is also very small. Therefore, it is very difficult to get exact measurements of the size and placements of these bones. 5.2.1 Hummingbird Bone Structure At this time, we have not yet reached a particular solution for attaining specifications on a hummingbird bone structure. We have pursued research at different venues such as the internet, university libraries, and direct contacts in order to locate existing information about the hummingbird skeletal structure, but our leads have proven dead-ends. Although some researchers have studied the muscles, anatomy, and flight characteristics of the hummingbird, we have not been able to find related data on the bone structure. 5.2.1.1 Dissection Other possibilities for obtaining information concerning the bone structure of the hummingbird exist. For example, we were given the chance to dissect a deceased hummingbird to be provided by Dr. Chai of the Zoology Department. Some of the problems raised in the dissection of a hummingbird include lack of experience in dissection, imprecise incisions, and a fragile body structure. The main concern in dissection was that the body structure of the hummingbird is known to fall apart if feathers are removed. Therefore, Ultra Flite decided to find an alternative to dissection. 5.2.1.2 Direct Image Correlation An alternative to dissection is direct image correlation. Direct image correlation is an approximate modeling of the bone structure. Attaining correlation results in obtaining an actual image of the bone structure of the hummingbird. For example, an image of the hummingbird wing could be obtained by x-ray technology. With an x-ray image of the wing, bones in the wing would be displayed. An approximate measurement of a real hummingbird wing bone-structure, such as the bone extending from the wrist to the end of the wing, could be conducted. With this measurement, one could use the x-ray image of the wing and make direct correlations.

Figure 5.12: A Close-Up Picture of Dimensioned Wing [14, p. 38] For instance, if the x-ray image of the wing, containing the same bone extending from the wrist, was one unit in length then it would relate to the approximate measurement of the actual bone extending from the wrist. In this experiment, one unit length equals 5 mm as shown in Figure 5.12. For example, if another bone was measured to be 2/3 the unit length on the x-ray image, then the real bone length would be 3.33 mm. Although this is a possibility in obtaining dimensions on the bone structure, it is not the most accurate or reliable. 5.2.1.3 CT Scanner An alternative to dissection and direct image correlation is Computed X-Ray Tomography, also known as CT Scan. As mentioned before in section four of this report, a CT scan is a completely nondestructive technique for visualizing features in the interior of opaque solid objects and for obtaining digital information on their 3-D geometries and properties. Although medical CAT-scanning is similar, the high-resolution X-Ray CT differs due to its ability to resolve details as small as a few tens of microns in size and objects made of high-density materials.

One of the first industrial CT scanners was purchased by The University of Texas at Austin by members from the W.M. Keck Foundation, the National Science Foundation, the Geology Foundation of The University of Texas at Austin, and the College's of Natural Sciences and Liberal Arts at The University of Texas at Austin [9]. Reuben Reyes, from the Department of Aerospace Engineering at The University of Texas at Austin, has been most helpful in explaining the possibilities of the CT scanner. He has mentioned that a CT scan of the hummingbird will offer data on the physiology of the hummingbird, especially the much needed bone structure. A picture of the CT Lab is shown in Figure 5.13 below.

Figure 5.13: CT Lab at the Geology Department [9] After contacting Dr. John Kappelman from the Department of Anthropology, we received confirmation on time availability for a scan this semester. We obtained permission to proceed with the initial test scan from Dr. Stearman. Dr. Chai will assist in the preparation of the preserved hummingbird for the scans.

The hummingbird is approximately three inches in height; therefore, a scan at 500 microns will result in ~250 slices. At one minute per slice and $104.00 per hour, approximately $260.00 will be spent paying for the scanning time. Although the scan is only 2 1/2 hours, the post processing of the information collected could take much longer. An example of an "emu egg" is shown in Figure 5.14 on the following page.

Figure 5.14: 3D reconstruction of 274 0.5 mm Slices of an Emu Egg. A Cutaway Rendering Displaying the Emu Embryo Skeleton In Situ, Side View [9]. Dr. Peng Chai informed us of a hummingbird that could be used in the scan. Unfortunately, the hummingbird was immersed in a fluid causing difficulty in keeping the specimen still for the duration of the scan. Reuben Reyes was our liaison to the CT-Scan Laboratory. As a previous employee with the College of Natural Sciences, he had many contacts in the Geology Department and was fully aware of the capabilities of the lab. While trying to coordinate a time to scan with the CT-Scan Laboratory, we were notified that Dr. Tim Rowe from the Department of Geological Sciences had also expressed his wishes to conduct a scan on a hummingbird. Seeing this as an opportunity to cut costs, Ultra Flite contacted Dr. Stearman for a possible joint scan.

Dr. Tim Rowe wanted higher resolution, therefore, a smaller slice thickness. He also wanted to use his own bird rather than one that Dr. Chai would provide. Since all these changes would alter the financial calculations, Dr. Stearman discussed the matter with Dr.Rowe. They decided a slice thickness of 0.1 mm to be used on Dr. Rowe’s bird. Ultra Flite was informed the specimen, although not an adult, was considered almost fully grown. The first scan was initiated, and the density spectrum of the hummingbird was gathered. The image data was downloaded to the LRC after the CT-Scan Lab finished the processing.

Ultra Flite then downloaded 577 images in the TIFF format. One could divide these images into two sets. The CT-Scan takes pictures at every other location then returns to take pictures of the skipped locations. Therefore, the first set of pictures (1-288) were the images taken from bottom to top. The second set of images were taken from top to bottom, this time acquiring images of the skipped parts. On the following page, Figure 5.15 illustrates one of these images, 512 x 512 pixels with 8 bitmaps (256 colors). Since the images include everything from bone to muscle, it was necessary to separate the bones from the rest of the bird in order to create a 3-D model of the bone structure. An image processing program, Adobe Photoshop v 4.0, was used to accomplish this task.

Figure 5.15: "Humm100.TIFF", the 100^th Slice from the CT-Scan. The Shades of Gray Show Different Densities, Where White is Most Dense (Bone.) Ultra Flite used the following commands in Photoshop to isolate the bones:

1. Pull down the Filter menu, select Noise, choose Median

2. Pull down the Image menu, select Mode, choose Indexed Color

3. Pull down the Image menu, select Adjust, choose Levels, find correct level settings that show only the bone

4. Pull down the Image menu, select Mode, choose RGB Color

5. Pull down the Image menu, select Mode, choose Indexed Color (adaptive, 4 colors)

6. Pull down the Image menu, select Mode, choose Color Table and change every color to black, except for white

Table 5.1 details the threshold levels for each frame range. One can enter the three values under Adjusted Levels in the window that opens on procedure number 3 above to get an image that shows mostly bone.
Table 5.1: Threshold Levels for Respective Frame Ranges

Frame No. Range	Adjusted Levels
009-040	62-1.00-171
041-069	171-0.19-199
070-140	146-0.17-192
141-144	150-1.00-152
145-159	150-0.32-152
160-169	174-0.32-176
170-179	190-1.00-192
180-210	157-1.00-159
211-239	175-1.00-177
240-259	196-1.00-199
259-279	149-1.00-151
280-288	129-1.00-131

On the following page, Figure 5.16 illustrates how the previous figure would look like after the procedure explained above.

Figure 5.16: Black and White Image of "Humm100.TIFF" after Thresholding. The Image Displays What the End Result of a Single Slice Would Look Like. Reuben Reyes suggested entering these commands into a program that would execute each procedure for its respective frame number saving valuable time. However, these procedures proved too complex so we decided to only do the thresholding, since the final result would change minimally. The software necessary for creating a 3-D model was available at the Department of Geological Sciences and Ultra Flite therefore relied on Reuben Reyes to obtain the 3-D model. The following paragraph explains in his own words how Mr. Reyes obtained the 3-D model.
After we did a median filter and threshold of the slices (stack of parallel images), I ran a program called Omniview on a SGI computer. The program Omniview takes each slice and creates an isosurface or 3D mesh. The method used to create this mesh is call "marching cubes". The 3d mesh is made up of facets (3d triangles). The total number of facets in this 3D model was 285,000. The program Omniview has an export option and the 3d model was saved as a Wavefront Object file. The object file was then converted into a Stereo Lithography file with a program called STL_Util. The Stereo Lithography file was then reduced in size with a program call Visualization Toolkit (VTK). With VTK a smoothing filter was applied to the points in the 3d mesh and then it was decimated. This reduced the number of points from 285,000 to 200,000. The 3D model was then saved as a Stereo Lithography for the 3D printout in Mechanical Engineering Dept. The last step was to make a 3d animation and still images in 3D Studio Max. The 3D model was too large to load into 3D Studio Max, so the3d model was cut up into 13 separate Stereo Lithography files. Using the program STL_Util the 13 files were converted into DXF files. The 13 DXF files were then loaded and put back together in 3D Studio Max for the animations and still images.
A problem with the scan was discovered after the thresholding procedures had been completed. During the scan, a shift in the position of the hummingbird occurred several times. This meant the CT-Scan data was useless. This discovery was forwarded to the CT-Scan Laboratory which decided to redo the scan. Also, the size of the image confirmed that the hummingbird was a baby, rather than an adult. The problem with a baby besides the size difference, is that the bones are soft compared to an adult which causes difficulty in recognizing the bone from the tissue. Nonetheless, Ultra Flite decided to create a 3-D model of the hummingbird’s head and beak since no shifts had occurred in that range of images. On the current page and the following pages, Figures 5.17-22 illustrate 6 images of the rendered 3-D model of the hummingbird viewed from 6 different angles.

Figure 5.17: 3-D Model Rendering, Bottom-View Figure 5.18: 3-D Model Rendering, Top-View

Figure 5.19: 3-D Model Rendering, Back-View. Notice that the Structure Protruding from the Bottom is the Start of the Vertebrae

Figure 5.20: 3-D Model Rendering, Right-Side View Figure 5.21: 3-D Model Rendering, Left-Side View

Figure 5.22: 3-D Model Rendering, Front-View The reason one can see holes in the skull is due to inaccurate thresholding. Some of the bone was not included in parts of the final black and white image that was used to separate the bone from the rest of the body. However, it is not a bad first draft of the 3-D model. Reuben Reyes, who has performed numerous CT-Scans and image post-processing, said the scan performed was of bad quality and the thresholding done was not bad. With the new scan, also 12-bit images (4096 shades of gray) will be available that will allow for a better contrast between the bone structure and the rest of the body.

At the time of writing this report, the new scan had not been initiated yet and Ultra Flite is therefore unable to present any new data or the finished 3-D model of the hummingbird.

5.2.2 SDRC I-DEAS When the CT scan is successful and the data collected is filtered, creation of the space-frame model of the hummingbird can be accomplished with the use of Structural Dynamics Research Corporation's (SDRC) I-DEAS software package. The advantage of using I-DEAS is that the information collected from the CT-Scan can be imported into I-DEAS. Also, it is much more practical to use I-DEAS to generate grid points of the space-frame model; therefore, we may transfer the generated grid points to NASTRAN.

At this time, we have not yet used the I-DEAS package, but we believe I-DEAS will be much more user friendly than NASTRAN. The I-DEAS interface is designed for point and click functions. The user-friendly interface allows for better understanding and efficiency of the software program. Therefore, Ultra Flite recommends the use of I-DEAS after the post processing of the data obtained from the new CT-Scan. If for any reason I-DEAS will not accept the output data from the CT-Scan, manual input of the data into I-DEAS to create a space-frame model of the hummingbird will be necessary.

5.2.3 MSC/NASTRAN The use of NASTRAN in creating the space-frame model of the hummingbird needs to be utilized if all else has failed with the I-DEAS software package. NASTRAN is known to be very user unfriendly because there exists no visual interface except for the code being written. Although the original objective was to create a space-frame model in NASTRAN, Ultra Flite has decided to use the I-DEAS software before resorting to NASTRAN. 5.3 Aeroelastic Analysis

Our initial objective three included analyzing the unsteady flight loads on the hummingbird by using an aeroelastic analysis that employs doublet-lattice aerodynamic modeling. In order to perform an aeroelastic analysis on the hummingbird in hovering flight, we must model the hummingbird in NASTRAN. Due to time constraints and the degree of difficulty, Ultra Flite and their sponsors, Dr. Stearman and Tifenn Boisard, discontinued the aeroelastic analysis of the hummingbird. The initial information researched by Ultra Flite, however, is included in this final report for future groups who will continue with the aeroelastic analysis on the hummingbird bone structure.

5.3.1 MSC/NASTRAN Ultra Flite has discovered two ways of modeling the hummingbird in NASTRAN. The first method includes importing the I-DEAS space-frame model directly into NASTRAN. The second possibility is to generate the model of the bird in NASTRAN. 5.3.1.1 Data Import/Export We have found it possible to import/export data from I-DEAS to NASTRAN. The only restriction is that the component needed to conduct the exchange is not readily available in the Aerospace Engineering LRC Lab at The University of Texas at Austin.

The component needed is called the SDRC I-DEAS Data Translator. It is capable of bi-directional exchange of data between MSC/NASTRAN and I-DEAS Finite Element Modeling (FEM) software. All data necessary to build and execute MSC/NASTRAN models for the majority of analyses are defined directly in I-DEAS FEM through a forms-based user interface. Data is directly output to the MSC/NASTRAN Bulk Data input file without going through an intermediate file. After MSC/NASTRAN execution, results data are recovered from the OUTPUT2 output file into the I-DEAS Universal file for reading into I-DEAS FEM [15].

At this time we have not yet decided if the purchase of the I-DEAS Translator is necessary. If it is not possible to purchase the translator continuation with the manual NASTRAN modeling is needed.

5.3.1.2 Manual Data Input Manually modeling the hummingbird in NASTRAN will be definitely less efficient but not impossible. Because the Aeroelastic NASTRAN software is new to Ultra Flite engineers, we did not complete this objective by the end of the contract deadline. Instead, Ultra Flite engineers became familiar with the syntax and command structure of the NASTRAN software. As a result, the following group must learn the process in order to complete the task of conducting an aeroelastic analysis which employs doublet-lattice aerodynamic modeling. 5.3.1.3 Current Work As of the Final, Ultra Flite has been able to successfully complete an example test run in NASTRAN. The example we followed was a version 65 NASTRAN run, HA21C [16]. The main objective in this example was to test for static aeroelastic characteristics of a fifteen-degree sweptback wing in a wind tunnel at Mach 0.45.

The purpose for conducting a preliminary NASTRAN run was to familiarize Ultra Flite engineers with the syntax and coding necessary to implement a NASTRAN analysis. We had to learn what each of the three main sections of the NASTRAN coding contained.

To begin with, Ultra Flite engineers decided the most efficient method for obtaining the example code was to use the HP scanner to save each page of the code into a different file. Upon completion of image transfer, we decided to use OmniPage v7.0 software to convert each image to text format. Conversion into text format allowed us to obtain the example code in a quick manner. Although there were some complications with the recognition of certain letters and symbols by OmniPage, we were able to process all the code. After detailed editing of the formatted text, we executed the example code in NASTRAN v70.

We began to face complications when executing our example code. The following are example error statements presented by NASTRAN:

^^^ USER FATAL MESSAGE 9002 (IFPL)
^^^ ERROR(S) ENCOUNTERED IN THE MAIN BULK DATA SECTION
^^^ SEE MESSAGES ABOVE.
ERROR ENCOUNTERED IN MODULE XSORT.
*** USER FATAL MESSAGE 307 (IFPDRV)
ILLEGAL NAME FOR BULK DATA ENTRY MAT
ILLEGAL NAME FOR BULK DATA ENTRY OA$
ILLEGAL NAME FOR BULK DATA ENTRY PAERO1
One problem encountered was the use of an illegal name for the bulk data entry. This statement told us that invalid commands exist in the code. NASTRAN was not able to read and execute those commands; therefore, an error resulted. In order to correct those mistakes due to oversights in the editing process, necessary changes in type were made.

Another problem was encountered when we finished correcting known errors. At first, we were not able to rectify the problem so we decided to find newer versions of MSC/NASTRAN manuals. After locating newer manuals, such as version 68, we found an exact resembling ours. After carefully examining the new code, we discovered that some of the commands used were no longer readable by our version of MSC/NASTRAN. Therefore, we decided to convert our old version of the example code over to the newer version of the example code. Through this process of editing and examining why certain commands no longer work, we learned that newer commands replaced the older ones. Also, some formatting in the older version of NASTRAN was no longer relevant in the new version of NASTRAN.

After all the corrections, we were able to execute the new version of the example code. The code was executed and all the results are listed in the appendix of this report. The output of the NASTRAN run included four plots of the wing. Of the four plots, two plots illustrated the aerodynamic elements of the wing and the other two depicted the structural elements of the wing. Other output included physical responses such as displacements and stresses located at each node. All output sections can be seen in the appendix of this report.

In conclusion, executing the example NASTRAN v68 run familiarized Ultra Flite engineers with the requirements necessary to implement a NASTRAN analysis. Although the example used was a static model, future groups must determine how to model a dynamic model of the hummingbird in hovering flight. The goal is to analyze the unsteady flight loads on the hummingbird by using an aeroelastic analysis that employs doublet-lattice aerodynamic modeling. In order to complete this task, further research in NASTRAN aeroelastic analysis will have to be conducted.

5.4 MAV Relations The initial goal of the microflyer program is to design a machine that can take off, land, and follow simple instructions while aloft. The five main areas that will present the most obstacles to this goal of advancement for MAV technologies are size, autonomy, flight controls, power sources, and weight. "You can't just shrink a 747 proportionally down to six inches and expect it to fly," says Dr. Samuel Blankenship, a principal research scientist at the Georgia Tech Research Institute (GTRI) [4]. The MAV designs currently under development could be affected by wind, rain, and even air itself due to its small size. Unlike that of large jets, the small wing area and resultant low Reynolds numbers present many aerodynamic challenges to future designers [3].

For an MAV to function properly, it will need to be at least partially autonomous. A human cannot react in time to make the necessary course adjustments that will be necessary during flight. Proposed means of making the microflyer autonomous include using a geographic information system (GIS), which employs a satellite to map the location of the flyer. Another option would be GPS, although they are only useful outdoors. Even the smallest GPS receivers, however, are heavier than a complete MAV and would consume all the available power.

The size of the vehicles’ control system will be dependent on the amount the control surfaces must react to induce attitude changes in the vehicle. Traditional planes use weighty motors and hydraulic machinery to maneuver control surfaces in the wings and tail. This will be impossible for the low weight requirements of the MAVs. Therefore, altered wing designs may help provide control without extra weight and allow the small aircraft to fly under control at very low speeds not possible with conventional wings and control surfaces [4].

Due to its small size, separate modules for each function of the aircraft will consume more volume than available. Therefore, the multifunctionality required by the MAV weight and power restrictions may be achieved only by a highly integrated design, with physical components serving multiple purposes or accomplishing multiple functions. For example, the wings may serve as an antennae or as sensor apertures, or the power source may be integrated with the footage structure. Each of the craft’s structures must do as many jobs as possible.

There have been many proposed options for power sources of the MAV, but since a drop of gasoline has more energy potential than current batteries of the same size, early models will probably use fossil fuels. Fossil fuels are probably the most promising power source because they have greater energy densities than stored electrical energy for driving a system. The internal combustion engines could provide electrical power from alternators. Alternatively, thermocouples could turn heat from burning fuel into electrical power [10].

Alan Epstein, an Aerospace Engineer at The Massachusetts Institute of Technology, has built a jet engine that is little more than one centimeter across. It is a scaled down version of conventional jet engines, and it works by using rapidly spinning turbines to compress the incoming air. Next, fuel is added and then burnt allowing the exhaust gases to provide the thrust for the MAV.

Rob Michelson, an Electronics Engineer at The Georgia Tech Research Institute (GTRI), has proposed an alternative to power supply, the pulse jet engines. The pulse jet engines are hollow tubes with a flapper valve that admit air and contain a hole on the side for injecting fuel. Exhaust gases are created by burning the fuel which power the engine forward and allows for the expanding gases to push open the flapper valve. This allows for a quick burst of air which then mixes with the fuel producing the next spark.

Another source of power is the Reciprocating Chemical Muscle created also by Michelson. The RCM is used to generate up and down motion for flapping wings by injecting fuel into a chamber. This causes a push-pull motion that not only causes the wings to move, but it also generates electricity as a byproduct [11].

Hummingbird MAV Study of Hummingbird Aerodynamics in Relation to Micro Air Vehicles

Hummingbird MAV
Study of Hummingbird Aerodynamics in Relation to Micro Air Vehicles