ANALYSIS Forum (IPACO) Forum Index

ANALYSIS Forum (IPACO)
Dedicated to the analysis of alleged UFO photos and videos

 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

The McMinnville case
Goto page: 1, 2  >
 
Post new topic   Reply to topic    ANALYSIS Forum (IPACO) Forum Index -> Analysis: photos and Videos -> Old cases - Discussion
Previous topic :: Next topic  
Author Message
elevenaugust
Administrator
Administrator

Offline

Joined: 20 Jun 2012
Posts: 67

PostPosted: 03/17/2013, 09:05 pm    Post subject: The McMinnville case Reply with quote

This topic is dedicated to the McMinnville case and the new analysis.

More to come...


Back to top
Publicité






PostPosted: 03/17/2013, 09:05 pm    Post subject: Publicité

PublicitéSupprimer les publicités ?
Back to top
jjflash
New member I
New member I

Offline

Joined: 18 Mar 2013
Posts: 1

PostPosted: 03/19/2013, 02:29 am    Post subject: The McMinnville case Reply with quote

Thanks for the report! Very interesting. Keep up the good work!

Back to top
Visit poster’s website
Larry
New member I
New member I

Offline

Joined: 20 Mar 2013
Posts: 9

PostPosted: 03/22/2013, 06:37 pm    Post subject: The McMinnville case Reply with quote

Just to establish my bona fides, since I am new here:  I am a professional Aerospace Engineer with more than 30 years of experience with a major space agency.  Back in graduate school, my major field of expertise was control of dynamic systems. Your hypothesis is that the well known Trent photos (MM1 and MM2) are the result of a small discoid model hanging from a thread about 3 feet below the lower of the two power lines seen in the photo. The physical dynamics of that kind of system is something I know something about and about which I would like to comment.
 
 
In support of your hypothesis, you present two pieces of photographic interpretation:
1)   You show that compared to MM1, the discoid image in MM2 is displaced about 12% further away from the camera, displaced upwards about 3.3%, and tilted toward the camera line of sight (LOS) about 25 degrees.  You point out (correctly) that these image differences between the two photos could be explained if a discoid model about 5 or 6 inches in diameter were suspended from the lower power wire by a thin string about 3 ft long AND the string had swung away from the camera about 17 degrees  AND the model had tilted forward (relative to the string) by an additional 8 degrees.  In addition, such a string would have to be thin enough and radiometrically dim enough to be below the resolution limit of the camera-film system; otherwise, it would already have been detected.
2)   Compared to a photo taken by a LIFE magazine photographer about a month later (mid-June, 1950) the lower power wire in the Trent photos appears to have sagged noticeably—perhaps an inch or two.  You postulate that the sag could have been caused by the weight of the presumed discoid model.
 
My critique:
 
Using the website: Weather History and Data Archive of the Weather Undeground, I searched for weather conditions on May 11, 1950 in the vicinity of McMinnville, OR.  At the time in question, it looks like the McMinnville airport did not have an active control tower and was therefore not keeping continuous surface weather observations.  The nearest airport that did have a tower appears to be Salem, located about 15 miles away.  Their records can be viewed at the url:
 
http://www.wunderground.com/history/airport/KSLE/1950/5/11/DailyHistory.htm…
 
As you can see, for the hour between 7:00 and 8:00 PM, the wind was relatively steady between about 10 and 8 MPH and coming from due West.  For a period of 3 years in the 1970s I lived about 20 miles north of McMinnville, while I was receiving training as a student pilot.  For that reason, I am very familiar with the weather patterns in that part of Oregon, which is known as the Willamette Valley.  It appears that on the day in question, the local region was dominated by a marine layer of air flowing East FROM the Pacific Ocean, inland TO the direction of 270 degrees.  After the winter storm season passes, sometime in mid to late Spring, this kind of stable on-shore flow is common. 
 
According to the diagram produced by Maccabee and reproduced in your analysis, the camera LOS when the two photos were taken was pointed approximately to the NorthWest (i.e., approximately at 45 degrees to the wind direction).  This  means that there would have been a component of wind velocity blowing directly toward the camera as well as a component blowing from the left to the right of the scene.  If the photo MM2 were really produced by a small model hanging from a thread and being displaced from the vertical by the wind, we would expect the image to be larger, not smaller (i.e., closer to the camera), showing more of the bottom, not less,  and being displaced to the right, relative to the image in MM1.  Obviously, we see none of those characteristics.  In short, the object, whatever it was, moved against the wind between MM1 and MM2.
 
With respect to the power wire sag:  the main parameter that determines how much a span of wire will sag is its temperature.  Changes in temperature, operating through the Coefficient of Thermal Expansion directly increase or decrease the length of the wire.  Since the distance between the end attach points of a power wire generally stay fixed relative to the pole or structure to which they are attached, an increase in wire length will show up as a sag in the wire span. 
 
Comparing the air temperature of May 11, 1950 with the air temperature a month later at Salem airport, we can see that the temperature was as much as 20 degrees F warmer on May 11.  The other factor that should be considered is whether the power wire was actually being used at the two different times in question.  If the power wire was carrying electrical power when the Trent photos were taken (therefore dissipating energy and warming up) and not carrying power when the LIFE photos were taken, that could completely explain the difference in the temperature of the wire between the two cases and therefore explain the sag.
 
 


Back to top
Larry
New member I
New member I

Offline

Joined: 20 Mar 2013
Posts: 9

PostPosted: 03/22/2013, 06:42 pm    Post subject: The McMinnville case Reply with quote

All:

Follow up to my previous post.  Sorry about all that format junk showing up at the front of my post--this was my first post on the forum.  I'll have to try to figure out how to suppress all that stuff.


Back to top
elevenaugust
Administrator
Administrator

Offline

Joined: 20 Jun 2012
Posts: 67

PostPosted: 03/23/2013, 12:11 am    Post subject: The McMinnville case Reply with quote

Hello Larry and welcome to our forum!

That's two good points that will be discuss in a more extensive way in the few next days:
1- The apparent discrepancies between the alleged/computed move of the object between the two shoots and the wind direction.
2- The power wire sag in the "epilog".

 In the meantime I would like to point out some notes:
- We haven't check any confirmation of the shoot order (i-e was MM1 really taken before MM2?) and your question raised up this important point that need to be confirmed or not. I'll ask Maccabee.
- Isn't the wind direction given by the airport weather station a general direction? I mean that we cannot exclude at this point some possible swirling or speed/direction local change, especially when there are houses around there.
- While no precise measurements can be done, Maccabee estimated that (it's own words) "the time between pix could have been as short as 10-15 seconds"
-
We are quite aware of the two points you underlined about the power wire sag between the Trent photos and those that were made one month later by Loomins Dean; that's why we have labeled it as following "the following lines do not constitute a scientific proof, but at least an interesting oddity."
- Any change in temperature would change the whole position of the wire, or at least a much longer part, and not a small part as it can be seen in our composition, wouldn't it?

I guess that all these data could be computed anyway.


Back to top
Larry
New member I
New member I

Offline

Joined: 20 Mar 2013
Posts: 9

PostPosted: 03/23/2013, 03:01 am    Post subject: The McMinnville case Reply with quote

The question was:

"Any change in temperature would change the whole position of the wire, or at least a much longer part, and not a small part as it can be seen in our composition, wouldn't it?"

OK, let's call the shape of the unperturbed wire (presumably as photographed by the LIFE photographer) as case "A".  Let's call the shape of the wire whose length is increased due to thermal expansion as case "B".   Let's call the shape of the wire whose length is increased due to suspending a weight from it as case "C".

Theoretically, shapes B and C would differ from shape A, along the entire length of the wire.  That's because either the presence of thermal expansion OR the presence of a weight suspended from the middle of the span would have the same effect of slightly increasing the length of the wire (even though the physical process by which the lengthening is created is different in the two cases).  A longer wire would then have to find a new shape in which the system of forces comes into equilibrium from one end to the other.

However, shapes B and C would also theoretically differ from each other, from one end to the other.  Therefore, the Trent photos could theoretically be analyzed to see whether they better fit shape B or C.   

I say "theoretically" because 1) the expected difference between shape B and shape C would be tiny, and 2) there would probably be a lot of noise in the measurement.

In engineering, suspended wires are usually modeled with a number of simplifying assumptions.  The main one is that the wire has no intrinsic bending stiffness.  In other words, it is treated as though it were a limp string.  Under that assumption, a wire suspended between two uprights theoretically takes on the shape of a catenary.  When a concentrated load is applied at one point, the resulting curve theoretically becomes a parabola. 

The observed difference between either shape B or shape C and shape A is small.  The theoretical difference between B and C is even smaller.  So, trying to distinguish whether the difference between the Trent photos and the LIFE photos is more like a catenary or more like a parabola is looking for an effect that is second order small.

Also, I notice that when you look at either the Trent photos or the LIFE photos, there seems to be kink in the lower wire which doesn’t go away.  This tells me that the wire actually has some intrinsic bending stiffness—in violation of the simplifying assumption upon which engineering analysis is based.  In other words, the wire has a preferred shape regardless of the external loads that might be applied to it. This effect would amount to “noise” in the process of trying to figure out whether the wire in the Trent photo is a catenary or a parabola. 

This is why I’m skeptical whether that line of research would yield useful information.


Back to top
Gilles F.
New member I
New member I

Offline

Joined: 22 Mar 2013
Posts: 2

PostPosted: 03/23/2013, 11:40 am    Post subject: The McMinnville case Reply with quote

Hello Larry,
Nice to see you here!
Concerning your "wind" remark and the expected motion of the object if Trent used a model:

1. As already stated by ElevenAugust, wind data from airports are general directions: we can't exclude local differences.

2. Whatever the direction of the wind was, winds are never constant in force: if it blows rather towards the photographer & MM1 and MM2 were shoot in this order, the same as the negatives roll (obviously we have not absolute proof of that order, it is based on Mister Trent's testimony, as Bruce Maccabee noticed "the negatives were cropped - cut off - at some time after the newspaper publication and before he got them", so no absolute proof of the photos order), this does not exclude a pendulum back and forth motion - to be short - :
For example, if the wind stops or becomes weaker between the two shoots, and that it first "pushed" the suspended object rather towards the photographer, the object will go "back" in pendulum motion, and therefore "against" the wind. This is imho true whatever the wind direction was. So we can't exclude a pendulum motion, as we can't expect imho a second shoot with a model closer the photographer than in the first one.

3. Besides, if that was a suspended object, we do not know its properties, its weight in particular. Depending of the weight of it and what caused the light motion (ie: induced voluntarily or involuntarily cause motion during attachment), the force of the wind may have no (or a very low) impact on the movement (the "gravity force" resulting of the weight of the object can be much stronger than the force of the wind or its impact to this object).

Regards,

Gilles Fernandez


Back to top
Larry
New member I
New member I

Offline

Joined: 20 Mar 2013
Posts: 9

PostPosted: 03/24/2013, 03:01 am    Post subject: The McMinnville case Reply with quote

Gilles,  I will repeat the reply to your 3 points that I made at the "UFO Iconoclasts" website, for the convenience of readers here.

“…As already stated by ElevenAugust, wind data from airports are general directions: we can't exclude local differences.”
 
True, we can’t exclude local differences, however it is clear that on the day and time in question the predominant surface wind was blowing at about 10 mph (plus or minus a few mph) and from due West (plus or minus a few degrees) all the way from the northern border of Oregon (Portland) to the southern border (Medford).  I would refer to this as the large-scale, or global pattern. 
 
I agree, that there can be small-scale variations in wind direction and velocity, particularly in the immediate vicinity of surface obstacles, such as buildings.  Normally, this would take the form of eddies or trailing edge vortices that would occur on the lee side of the obstacles.  It is theoretically possible to estimate the location, size, and power spectral density of such vortices based on the Reynolds number of the buildings.  I have not done so, yet.  Overall, however, it should be kept in mind that local turbulence patterns must be compatible with the global momentum flow from West to East.  By itself, a vortex or eddy is not an energy, mass, momentum, or angular momentum source or sink; it is simply a rotary flow pattern superimposed on the global velocity field.

 
“…we have not absolute proof of that order [in which the photos were taken], it is based on Mister Trent's testimony… so no absolute proof of the photos order)…”
 
I suppose it’s true that we have no independent evidence for the order in which the photos were taken, but we also have no independent evidence that Mr. Trent’s testimony was false on that point, and no apparent motivation for him to have lied.
 
“…this does not exclude a pendulum back and forth … This is imho true whatever the wind direction was. So we can't exclude a pendulum motion, as we can't expect imho a second shoot with a model closer the photographer than in the first one…”
 
I agree completely.  The IPACO analysis so far has been done as though it was a static system.  In reality, a small discoid model suspended on a slender thread being blown by the wind—which is the conjecture—would be a dynamic system described as a compound, spherical pendulum, subject to a temporal-and-spatially varying forcing function.  When I think about the equations of motion of such a system, I would model it with at least 5 degrees of freedom, all of which would be coupled together with kinematic and aerodynamic constraint equations.  Once such a system were set in motion, it is almost inevitable that all degrees of freedom would take on non-zero values that would evolve with time (more about that in a subsequent post).
 
The IPACO analysis already seems to show that that did not occur.  In particular, it has been shown that the relative motion between the two photos (regardless of the order in which they were taken) consists of a linear displacement along one axis only (the camera LOS) and a rotation about one axis only.  To me, as an Aerospace Engineer, that sounds much more like the motion of a 3-axis-controlled, free-flying object than the open loop, free-and forced motion of a spherical compound pendulum oscillator.
 
“…Besides, if that was a suspended object, we do not know its properties, its weight in particular. …the "gravity force" exerted by the weight of the object can be much stronger than the force of the wind or the impact of the wind to this object.”
 
True.  We also don’t know the moment of inertia of the object, which would figure in to the equations of motion.  But all those factors can be estimated (within reasonable error bounds) based on a few reasonable conjectures.  For example, it has been conjectured that the model was a metallic object (such a the rear-view mirror of a 1940’s Ford pickup truck) that Mr. Trent might easily have had access to.  We could find one and weigh it.  Alternatively, we could conjecture that the model might have been manufactured by Mr. Trent out of some easily available material (such as Douglas Fir wood). Personally, my hunch is that it would be very difficult to get a 17 degree pendulum displacement statically (i.e., by a wind that was theoretically perfectly constant in both direction and magnitude).  I may be able to estimate that by a simple calculation (which I also have not done yet).  If so, then if you want to pursue the pendulum hypothesis you pretty much have to presume that the system was dynamic (i.e., constantly moving) and the photos just happened to 'freeze' the motion at two points in time.

I have been thinking about constructing such a model, putting it on the end of a 3-ft string, and subjecting it to a 10 mph wind, just to see what it actually does.  Sometimes it’s actually easier and faster to build a real physical analog of a system than to code up the equations of motion in software.


Back to top
Gilles F.
New member I
New member I

Offline

Joined: 22 Mar 2013
Posts: 2

PostPosted: 03/24/2013, 01:11 pm    Post subject: The McMinnville case Reply with quote

"I have been thinking about constructing such a model, putting it on the end of a 3-ft string"

Hello Larry,

Interesting.
Allow me to suggest you -but probably you have already thought about it - to not forget to take into account or "to play" in your possible tests with different diameters of wires/fishing-lines.
Why? Because, in my minimal knowledge of the case (IPACO team could/will correct me), the grain size of the film was 5-10 microns. With this grain size, smaller objects (such as a thread) to the film granularity could be detected (probably objects with a size comprised between 1 and 2 microns?). After all, of course, it depends of the distance.
In the hypothesis of a suspended model under the wires, it was probably located 14/16 feet (4/5 meters) away from the camera.

I noticed in my notes that such fishing lines/threads have a diameter between 0.03mm to 0.06mm and that the detection possibilities were computed/calculated by Maccabee:

 
 
Quote:



The most distant wires were probably over 60 m away. Using a wire diameter of about 0.6 cm (1/4"), the angular width of the distant wires would have been about 0.0001 radians . Experiments with detection of small linear structures (e.g., threads) by photographic means indicate that if there is sufficient contrast between the structure and the background a linear image structure much smaller than the grain size of the film can be detected. Since the grain size of the film used by the Trents was on the order of 5-10 microns, linear structures with images as narrow as 1 micron might be detectable, corresponding to angular sizes of about 0.001 mm/100mm = 0.00001 radians (where 1 micron = 0.001 mm). This would correspond to a thickness of about 0 .05 mm at a distance of 5 meters (about 16 ft) , which would have been the distance to the object if it had been hanging under the overhead wires. A typical thread is about 0.03-0.06 mm in diameter.





I hope here in France too, we would be able, to buy/obtain a "good" Roamer I camera (or a "similar" one), different models (mirrors or others), threads, etc., to establish a good test-protocol. But all of this is a allocation of resources (mainly "cognitive" ones, that's not so expensive) and must be thought with great attention and consideration in regards of the different parameters (known or speculative/deductive) to estimate the feasibility of such a test, and then if it will provide good informations.

Regards,

Gilles


Back to top
Larry
New member I
New member I

Offline

Joined: 20 Mar 2013
Posts: 9

PostPosted: 03/24/2013, 06:15 pm    Post subject: The McMinnville case Reply with quote

Gilles:

When I speculated about building a model and testing it, I was primarily thinking of testing its dynamic and aerodynamic properties.

Photography is not my strong suit, however, it occurs to me that there is one particular type of photoanalysis that perhaps someone at IPACO could perform which could prove very important. 

When you capture an image of a long, slender object (like a thread) with camera optics, the character of the image is qualitatively different depending on whether you scan along the length of the thread or perpendicular to it.  Here is my understanding of the physics of the case:

Every collection of optical elements (lenses, prisims, mirrors, etc.) has a lower limit of resolution determined by the diffraction limit, which is entirely a function of the aperture size, focal length, and the wavelength of the light that is being used for imaging.  I don't know what the diffraction spot size for the Roamer camera was, but it should be easily calculable.  It may be related to the minimum grain size, as calculated by Maccabee. 

The point being, that if the diameter of the putative thread is smaller than the diffraction spot size, then you could not resolve the image of the thread, no matter what.  Imagine sampling a line of pixels selected from a horizontal raster, cutting across the location where the thread is supposed to be, in the image. As you move from left to right along the line of pixels, you would not see any significant contrast between adjacent pixels as you cross the thread image.  The checking for contrast variations is done--I believe--with a spatial autocorrelation filter.

If, however, you applied the same spatial filter to a vertical line of pixels that happened to lie on top of where the thread was supposed to be, you would see a significant autocorrelation.

This is based on the physical principle that, even though the diameter of the putative thread may be below the resolution limit of the optics, the length of the putative thread most certainly would not.  When the image is captured by the optics, all the wavefront information necessary to perform this analysis resides on the film.

Does anyone on the IPACO team have the knowledge, software, and/or ability to do this kind of analysis?  In my mind, this kind of analysis would be conclusive, one way or another.  If there is information about a vertical slender thread residing in the image, that would be definitive evidence of a small model.  On the other hand, if there is no evidence for a slender thread--even after looking--that would also be pretty conclusive evidence for the existence of a free-flying object.


Back to top
louange
Administrator
Administrator

Offline

Joined: 20 Jun 2012
Posts: 2

PostPosted: 03/24/2013, 08:03 pm    Post subject: Detection of a thin thread Reply with quote

Hello!

This issue of "thread detection" is sometimes complex and surprising. It reminds me of 2 true stories:

1) In the early 80's, I was working with intelligence officers on early experiments in digital image analysis (using the ancestor of IPACO), with Landsat satellite images (resolution of ca. 60 m or so, I do not remember exactly). One day, we managed to locate the position of an electric power wire over Russia (wire's diameter being <<<< resolution), just because the ground was all covered with snow and the contrast of the dark wire was enough to shift by 1 or 2 the radiometric value of concerned pixels...

2) Around the same period, I was asked by a serious Italian ufological journal to write a paper about the use of computer image processing techniques. I wanted to demonstrate the possibility of detecting a suspending thread through correlation techniques such as what Larry is suggesting. I took a good camera, built a model with 2 glued paper plates and hanged it with a thin nylon thread. I got the photos scanned, then I spent hours with a powerful computer (for that time!) trying by all means to bring up evidence of the thread, with no success whatsoever. In the end, for the sake of my expected paper, I had to cheat and to modify a few pixels manually, so as to demonstrate the technique.

Concerning McMinnville pictures, I already tried several "high-pass filters" and examined closely the pixel distribution, and I am afraid we are in the same case as above. But since this question is raised again here, I will try again and let you know.

Regards  


Back to top
Larry
New member I
New member I

Offline

Joined: 20 Mar 2013
Posts: 9

PostPosted: 03/24/2013, 10:47 pm    Post subject: The McMinnville case Reply with quote

louange:

As I understand it, the use of an autocorrelation filter is based on a different principle than a high-pass filter.

The theory is that somewhere in the photo there is a line of pixels that have one thing in common--they all received photons reflected from the string.  An autocorrelation filter looks to see if there IS a string of pixels in the image where each pixel has more in common with its neighbors in a certain direction than it does with pixels selected at random.  A high pass filter looks at all the pixels that are brighter than the threshold value of the filter and asks whether--from that set of filtered pixels--there is an unbroken line of pixels somewhere in the scene.  In other words, one approach looks for pixels that are all similar to each other, by some measure; the other approach looks for pixels that are all different from the background, by some measure.

There's no reason to not try both approaches, in my opinion, assuming there are high quality digitized versions of the film photos available.  By the way, where are the best quality digital data files of the scanned photos?


Back to top
louange
Administrator
Administrator

Offline

Joined: 20 Jun 2012
Posts: 2

PostPosted: 03/24/2013, 11:14 pm    Post subject: The McMinnville case Reply with quote

You are right. I made it too short, but of course high-pass filtering is not equivalent to autocorrelation. However, a high-pass filter may be directional and look after discontinuities in a given direction (here vertical), which is sometimes largely enough to bring a vertical thread into evidence.
 
I already tried various types of convolution kernels, with no positive results so far, but I will think of other approaches along your interesting ideas. I must leave this forum now but will come back soon.
 
Concerning the quality of available scans, you can see in our report that we found a given pair of images more interesting for geometry, and another one for radiometry. But all of them are frustrating , since they all result from scans of paper prints, not of the original negatives. The quality is rather poor...

Regards


Back to top
Larry
New member I
New member I

Offline

Joined: 20 Mar 2013
Posts: 9

PostPosted: 03/27/2013, 03:31 am    Post subject: The McMinnville case Reply with quote

In a previous post, I mentioned that I thought it unlikely that a suspended model would show motion around only one axis of rotation.  This is a more extended discussion of why I think that.

The hypothesis is that the object in the Trent photos is a small discoid object (about 5 to 6 inches in diameter) suspended on the end of a thread that is about 3 feet long and so slender that the diameter of the thread is below the limit of resolution of the camera-film system.  Let’s do a gedanken experiment to predict some of the dynamic behavior we might expect from such a system.
 
First, we assume that the thread is maintained in the taut condition at all times by the weight of the discoid; this constrains the thread to always be straight and of a fixed length.  Second, we note that the thread is free to swing in any of two mutually orthogonal planes, simultaneously.  Because the thread is presumed to always define a straight line, we can define two angles (between the thread and the power wire to which the thread is presumably attached) that describe how much swing is present at any given time.  The analysis thus far has defined one angle, s, which is the amount of swing in the vertical plane containing the attachment point between the thread and the power wire (point ‘A’, in the diagram).  By symmetry, we can define another angle, r, that is the amount of swing in a vertical plane perpendicular to the 's' plane.  This is the classic definition of a spherical pendulum.
 
In an analogous manner, the discoid model is capable of tilting about each of two orthogonal axes, as measured relative to the straight thread (independent of the two thread swing angles, 'r' and 's').  We may define one angle, 'a', as being the tilt of the model around a line that is horizontal, that passes through the point of attachment between the thread and the discoid (point 'O'), and lies in the vertical plane containing the camera LOS.  Again, by symmetry, we will define the other tilt axis, 'b', as being the horizontal line passing through  point 'O'  that is orthogonal to the 'a' plane.
 
Finally, we note that the discoid appears to be free to rotate around its own axis of symmetry.  Point ‘O’, is presumed to lie on the symmetry axis at the point where that axis intersects the model’s skin.  We will designate the angle of rotation around the symmetry axis as 't'.
 
Thus, our dynamic system contains 5 independent variables (or ‘Degrees of Freedom—DOFs').  Understanding the motion of the dynamic system is equivalent to understanding how the variables 'a', 'b', 'r', 's', 't', evolve with time.
 
Normally, you expect some or most of the DOFs of a dynamic system to be coupled.  That means that if you put energy into one DOF, you expect that DOF to start varying with time, but you also expect to see energy migrating from that DOF to the others to which it is coupled. So, one of the traditional methods used for understanding the behavior of a dynamic system is to consider its ‘impulse response'.  That is, you place the system initially at rest and then ‘whack' one of the DOFs with some kind of force or torque that is finite in its magnitude and essentially instantaneous in its application.  Then you watch the other DOFs; if energy starts migrating from the directly excited DOF to the others, then you know the DOFs are coupled.
 
Imagine that our system is initially at rest, and in perfect equilibrium with gravity.  In this condition, all 5 state variables take on the value of zero.  The thread is hanging perfectly vertically.  The symmetry axis of the discoid is perfectly aligned with the thread.  The discoid has not rotated around its symmetry axis. (Since it is necessary to choose an arbitrary index mark against which to measure rotation, let’s say that we use the camera LOS as the reference line, and imagine that we paint a dot on the outer rim of the discoid at the point where the LOS would intersect the rim.  At time = 0, the dot on the rim is exactly aligned with the LOS.)
 
The impulse we are going to provide is a puff of wind that is perfectly horizontal and moves from camera left to camera right and impinges on the discoid, edge on.
 
The first thing that happens is that the air flowing around the model creates what is called ‘form drag'; this is simply the result of the air flowing past the skin of the model and interacting through viscosity.  This shows up as a net force acting horizontally approximately through the Center of Mass (CM) of the model that would drive the model to the right.  Since the model is constrained to lie at the end of the thread this drag force would directly excite the pendulum mode and result in the DOF 'r' taking on a non-zero value. 
 
However, there is an offset between the model CM and the pivot point, 'O'.  Therefore, the drag force acting through the CM would create a torque that would directly excite the DOF, 'a' (the tilt of the model relative to the thread).
 
We note that the shape of the model is flat on the bottom and convex on the top.  This means that it is an airfoil (an inefficient airfoil, but an airfoil, nevertheless).  So, the puff of wind would also generate lift on the discoid. Many people are surprised to learn a fundamental fact of aerodynamics--that the sum of all aerodynamic lift forces on a wing acts as though the lift force vector were concentrated at the ‘quarter chord' (i.e., ¼ of the way back from the leading edge toward the trailing edge).  However, the CM of the discoid is at the ½ chord distance (i.e., at the geometric center of the model).  So, the offset between the lift vector and the CM creates a torque on the model that also shows up in the DOF, 'a'.  This torque is opposite in sign to the torque that is caused by the form drag.  The model will end up rotating around the ‘a’ axis in a direction that depends on which torque is greater.
 
It is often the case that there can exist singular solutions to the equations of motion which are mathematically allowed, but not physically meaningful.  For example, I could imagine a system consisting of 5 perfectly spherical ball bearings stacked one on top of another with each CM perfectly placed on a common vertical axis.  If I simulated such a system mathematically with all of the displacements exactly equal to zero at t = 0 and ran the simulation for a finite amount of time, nothing would happen.  The simulation would predict no dynamics, even though we know intuitively such a system is unstable and would quickly collapse under its own weight.

In order to avoid this kind of  unreal solution, it is customary to imagine  what happens if the system starts from a condition that is slightly off (‘perturbed') from the nominal condition.    So imagine that when the puff of wind occurs, there were also slight, initial, non-zero values for the DOFs, 'b' and 't'.  An initial value for the DOF 'b' is analogous to an aircraft having a bank angle relative to its relative wind.  So, when the puff of wind hits the airfoil, there is a component of lift that is perpendicular to the vertical.  This component tends to push the CM of the model either into or out of the 'r plane, thus exciting the 's' DOF.  Simultaneously, because the center of this lift vector is at the ¼ chord location (and therefore ahead of the CM), it creates a torque around the 't' axis. This creates rotation of the model around its axis of symmetry.
 
We would expect that motion around the 'r' and 's' axes would be oscillatory because gravity acts as a restoring force; this is the main effect that creates the pendulum mode.  Displacements around the 'a', 'b', and 't' axes do not, in general, experience restoring forces and so would probably not be predominantly oscillatory.    We would expect displacements around those axes to either follow the forcing function in a proportional manner or to experience exponential divergence (i.e., be unstable). 

Most of the motion of the system should show up in the primary mode--which is the pendulum mode (DOFs 'r' and 's')--and in the direction of the wind, but we would not expect the other modes to be completely unexcited.  The bottom line is that we would expect noticeable motion in all the DOFs after a period of time.


Back to top
Haraka
New member II
New member II

Offline

Joined: 21 Aug 2012
Posts: 12
Localisation: Tanzania

PostPosted: 03/28/2013, 12:00 pm    Post subject: The McMinnville case Reply with quote

 Hi Larry,
Your helpful and informed input is very much appreciated and certainly it seems  unreasonable to be able to attempt to quantify the movement of the disc.
 
The motion of the disc is certainly probably going to be irregular and five variables ( not including possible  "bounce" as you noted on the wire) doesn't help.  This I think makes a resolution  virtually impossible. 
 I'm more used to centre of gravity ( C of G ) as a term  and presume that is equivalent to your Centre of Mass. Aerodynamically  of course the centre of pressure ( C of P)  typically moves forwards against the direction of airflow as the  the  Angle of Attack  ( Alpha) increases, right up to the stall of the airfoil at around 15 deg. Increasing Alpha  invokes a consequent  increase in the Coefficient of Drag ( probably adding both to "r" and "s" components in this case) which again suddenly drops  as the disc stalls ( C of P moving rapidly  aft) , building up again as it recovers and adding to the overall oscillatory interaction .  We don't know the mass of the disk either ( as you have also noted)  which doesn't help in predicting the behaviour of this complex system. 
My primary concern in this case has always been the age and provenance of the imagery available.  The subtended angle of the diameter of the supposed line  when considered in relation to the size and irregularity of the film grain will probably  defeat any digital resampling and analysis. Certainly sub-pixel features spatially ( especially linear ones) can be pulled out by post processing, indeed the human visual system does it rather well as Francois has indicated. To me , in this example, " absence of evidence is not necessarily evidence of absence" will probably remain  the verdict regarding the probable  suspension line.


Back to top
Contenu Sponsorisé






PostPosted: Today at 01:00 am    Post subject: The McMinnville case

Back to top
Display posts from previous:   
Post new topic   Reply to topic    ANALYSIS Forum (IPACO) Forum Index -> Analysis: photos and Videos -> Old cases - Discussion All times are GMT + 2 Hours
Goto page: 1, 2  >
Page 1 of 2

 
Jump to:  

Index | Create free forum | Free support forum | Free forums directory | Report a violation | Conditions générales d'utilisation
Powered by phpBB © 2001, 2005 phpBB Group