Jump to content
Maestronet Forums

An interesting graduation result


Johnmasters
 Share

Recommended Posts

I decided to convert my FEA model of a violin top from mm to meters in order to use elastic constants that were a problem to convert to mm usage.

I studied it all, got the elastic constants in and applied the forces at the ends.. from the neck and tailpiece. The thing did not distort much at all and seemed quite stiff.

I dropped the project to watch a bit of TV.

It occurred to me after a while that I had not changed the number indicating the thickness of the shell elements. These appear in the picture window as simple shell surfaces, They have no illustrated thickness.

Anyway, when I converted graduations from 3 to .003 (meters), things were much better.

How many of you have experimented with graduations 10 feet thick ?? :)

Link to comment
Share on other sites

I decided to convert my FEA model of a violin top from mm to meters in order to use elastic constants that were a problem to convert to mm usage.

I studied it all, got the elastic constants in and applied the forces at the ends.. from the neck and tailpiece. The thing did not distort much at all and seemed quite stiff.

I dropped the project to watch a bit of TV.

It occurred to me after a while that I had not changed the number indicating the thickness of the shell elements. These appear in the picture window as simple shell surfaces, They have no illustrated thickness.

Anyway, when I converted graduations from 3 to .003 (meters), things were much better.

How many of you have experimented with graduations 10 feet thick ?? :)

++++++++++++++

Conceptionally 0.0001 meter and 0.00001 meter is the same to me, " small " (in fact one is ten smaller)

Link to comment
Share on other sites

Please explain - Why did the stiffness increase in your 'protortionally increased' model? Would the thickness not be appropriate for the rest of the sizings?

Actually, there were FOUR things to adjust, x,y,z in shape and a thickness which is not contained in the geometry. Even in proportion, the elastic constants would change, as they have units of force per area.

The set of points (nodes) are in x,y,z but they only have the surface shape, outline and elevation of the arch. That is for shell elements. A triangular shell element is not a "thin brick", it has only three points. (its properties are handled by the program)

The thickness is entered in a different place where the shell elements are defined. One tiny little space gives the thickness. This is what I overlooked.

Because the picture on the screen simply shows a shell surface, it looked normal. But that one little entry said it was 3 meters thick !!

Also, even though it had been in proportion in mm, the elastic constants were in the MKS units and I did not want to try to change to mm. I did not know if the program would want mm, newton-force, seconds or some other force. There are also three constants that are a bit of a mystery to me. (Six of them are in units of pressure.)

There are 3 E's 3 shear constants, and 3 "poisson ratios."

Link to comment
Share on other sites

Never tried 10 foot thick graduations but I am currently working on an experiment to determine how much the curvature of the top plate changes the effective stiffness seen by low frequency modes. So far the effect seems bigger than I expected but I need more data points so I'll post on this in a few more days.

Link to comment
Share on other sites

Never tried 10 foot thick graduations but I am currently working on an experiment to determine how much the curvature of the top plate changes the effective stiffness seen by low frequency modes. So far the effect seems bigger than I expected but I need more data points so I'll post on this in a few more days.

If you are changing the curvature, I suppose you are doing it in an FEA program? I can supply any number of models, if you have student ABAQUS. If you have another one, the ascci input files can usually be converted with a change of format of the nodes and elements.

Happy to help. Let me know what you are doing. You can write jmluthier@sbcglobal.net

At least tell us how you are changing curvatures. By the way, the free plate modes have very wavy perimeters, and at the magnified scale of the view window in ABAQUS, one can easily see how any curvature may be not too important. I will post anything anyone wants to see.

Link to comment
Share on other sites

Simpler than that. Start with the top of a cheap chinese violin and slowly flatten it. I simply dampen the wood quite a bit and then lay the plate on a flat surface with an old calculus book on top of it. Once it dries out I measure the changes. The wood properties and mass will be pretty much the same thoughout the process so you can see the changes due to the arching. The problem is that wood doesn't like to stretch much so there are cracks along the edges that need fixing. It's not supposed to be an ideal experiment but it's quick, simple, and will finally give me an excuse to throw that old violin top away when I am completed.

I suppose you could also 're-arch' like is done with old Strads and Guarneris but that is a lot more work.

Link to comment
Share on other sites

It should depend on the original starting curve, if the original arch is very flat (large radius of curvature) then changing the curve a little bit will be a very small effect. If you apply the same amount of change to an arch with a very small radius of curvature then you will change things by a much larger amount. This implies that it would be harder to make a 'good' sounding violin with high arches since since small deviations from some hypothetical ideal arch would lead to relatively large changes in vibrational characteristics. Also it depends on the wavelength of the vibrations present, the effects of arching changes should be stronger in low frequency vibrations than in high frequency ones.

Cremer's book on violin physics has a section about how the curvature of the plates should change the vibrational propoerties from what would be obtained from flat plate calculations but I don't remember it too well. What I took from his results is that for the normal range of curvatures seen in violin arches it shouldn't be a huge deal, maybe a 5 or 10% change in eigenmode frequecies over the normal range if everything else were kept constant. The results of my experiment are showing a bigger influence than I had expected at this point but I should review his work to be sure. His results were from cylindrical and spherical shells but I would imagine that the results of a violin type arch should be something intermediate.

Link to comment
Share on other sites

Simpler than that. Start with the top of a cheap chinese violin and slowly flatten it. I simply dampen the wood quite a bit and then lay the plate on a flat surface with an old calculus book on top of it. Once it dries out I measure the changes. The wood properties and mass will be pretty much the same thoughout the process so you can see the changes due to the arching. The problem is that wood doesn't like to stretch much so there are cracks along the edges that need fixing. It's not supposed to be an ideal experiment but it's quick, simple, and will finally give me an excuse to throw that old violin top away when I am completed.

I suppose you could also 're-arch' like is done with old Strads and Guarneris but that is a lot more work.

The interesting modes would be the ones that involved in-plane stretching. Stretching the plates to flatten them would seem to not give helpful results.

I have a question if someone feels like taking it on.

What's the smallest deviation from a curve that will have a 'significant' (audible) effect on how the structure vibrates?

Is that a simple enough question or do I need to simplify it more?

I am trying to get my models in shape to look at different long arches with transverse CC. I will have back, sides, linings, and a top with bar.

I am almost finished with the model, and will be able to adjust various things.

It is a very general question, almost too simple. One may be able to see differences in normal modes in a model, but whether the results are interesting depends on where you think tone comes from.

Link to comment
Share on other sites

It should depend on the original starting curve, if the original arch is very flat (large radius of curvature) then changing the curve a little bit will be a very small effect. If you apply the same amount of change to an arch with a very small radius of curvature then you will change things by a much larger amount. This implies that it would be harder to make a 'good' sounding violin with high arches since since small deviations from some hypothetical ideal arch would lead to relatively large changes in vibrational characteristics. Also it depends on the wavelength of the vibrations present, the effects of arching changes should be stronger in low frequency vibrations than in high frequency ones.

Cremer's book on violin physics has a section about how the curvature of the plates should change the vibrational propoerties from what would be obtained from flat plate calculations but I don't remember it too well. What I took from his results is that for the normal range of curvatures seen in violin arches it shouldn't be a huge deal, maybe a 5 or 10% change in eigenmode frequecies over the normal range if everything else were kept constant. The results of my experiment are showing a bigger influence than I had expected at this point but I should review his work to be sure. His results were from cylindrical and spherical shells but I would imagine that the results of a violin type arch should be something intermediate.

It would be nice if you had some sort of model to work with. ABAQUS student version is apparently free now. Or it only used to cost $100. Aren't you a physics graduate student? You ought to be able to do this sort of thing.

Link to comment
Share on other sites

Aren't you a physics graduate student? You ought to be able to do this sort of thing.

Yes but that means I don't have the time.

I do have a related problem that I need to work on for my research though involving hundreds of coupled overdamped harmonic oscillators but I think I might be able to solve that on paper. At some point I will probably use some sort of discrete model for the system but I have access to some nice electronics simulations so I should make use of that.

Link to comment
Share on other sites

The interesting modes would be the ones that involved in-plane stretching. Stretching the plates to flatten them would seem to not give helpful results.

I'm interested in just the overall stiffness, whether that comes from the plate stretching or bending doesn't interest me so much at the moment. The calculations in Cremer's book were related to the extra energy involved due to stretching but there are other ways for a violin plate to have more stiffness than a flat plate than just because of stretching.

The plates don't really stretch when I flatten them, the central portion flattens out because it's cylindrical in shape so the middle bout width increases a bit with the flattening. The upper and lower bouts, however, are close enough to a spherical shell so that they won't flatten without cracking along the edges. I think I have the plate as flat as I can make it without using heat to help the bending. I've avoided using heat because I don't know how that might affect the materials properties. The plate just needs to dry out now.

Link to comment
Share on other sites

Yes but that means I don't have the time.

I do have a related problem that I need to work on for my research though involving hundreds of coupled overdamped harmonic oscillators but I think I might be able to solve that on paper. At some point I will probably use some sort of discrete model for the system but I have access to some nice electronics simulations so I should make use of that.

The solution you would need is worked the same way the program works it... Here is a suggestion that will save you many hours over your career ............

You don't need to learn how to use the model creation window in FEA. Except for very simple examples. When you solve a problem, you get an output that contains an ASCII file of nodes and elements and all the conditions, etc.

You work a simple problem and from there work in Wordpad to modify your situation. Then put it back in and simply solve with a couple of key strokes.

If you want to diagonalize a large matrix for eigenvalues (And eigenvectors) there is no easier way to do it............

The documentation does not tell you this simple thing. Most of the tutorials start with a very complicated instruction on how to draw the problem in the first section. That intimidates everyone.

Numerical solutions are a must, it seems, for most problems such as the one you have....

Link to comment
Share on other sites

I'm interested in just the overall stiffness, whether that comes from the plate stretching or bending doesn't interest me so much at the moment.

You don't need Cremer, work from first principles perhaps. There are other books on shells, you may as well forget violins.

As for your problem: Put three or four connected beams in FEA and solve it for vibrations. (Which is the easiest solution for any problem...... no boundary conditions or loads)

All of the damping etc will be in the materials description.. They lead you through that by filling in boxes. Take the input file which will be created by the solver. Then work in Wordpad.

You will be glad you did it. BELIEVE ME

Link to comment
Share on other sites

If given the choice between hours of computer programming and a simple experiment that I can do in minutes per day, I will just do the experiment. I'm not convinced (due to the stuff I read in Cremer) that stretching is all that important when comparing two plates with slightly different archings so unless something major comes from this experiment I won't spend much time on worrying about shells. In the next few weeks I'll be learning to use enough new software, if one of those happens to involve FEA then great but otherwise I won't have the time or patience for more new programming.

Link to comment
Share on other sites

If given the choice between hours of computer programming and a simple experiment that I can do in minutes per day, I will just do the experiment. I'm not convinced (due to the stuff I read in Cremer) that stretching is all that important when comparing two plates with slightly different archings so unless something major comes from this experiment I won't spend much time on worrying about shells. In the next few weeks I'll be learning to use enough new software, if one of those happens to involve FEA then great but otherwise I won't have the time or patience for more new programming.

There IS no programing involved. You can do what you want. I have been following your comments for some time. You are in a purely experimental physics environment, are you not? You can get a PhD in physics that way, but you may not know very much when you are finished.

If differences in stretching are not that important, than neither will be bending. Your results will be hard to interpret and close to null.

Link to comment
Share on other sites

There IS no programing involved. You can do what you want. I have been following your comments for some time. You are in a purely experimental physics environment, are you not?
No.
You can get a PhD in physics that way, but you may not know very much when you are finished.

And what do you learn from tinkering about with an FEA model?

Three years ago when I was working on acoustics elsewhere a physicist told me something important that has stuck with me. Don't expect to learn anything new from FEA models. All of the physics was inserted by whoever wrote the package.

Link to comment
Share on other sites

I didn't have a double physics and math major for the first ~12 years that I built either. I don't expect it to help me make better violins, I want to use it to help automate making consistantly really good violins. One potential approach to doing this is to try engineering a material and design which works well, as is done with carbon fiber instruments. Another approach is to figure out how to modify the violin's design to account for the properties of each piece of wood, this is what I'm trying to do.

Another way to look at your statement is this, sure Strad and Guarneri didn't use physics to design violins but neither did the thousands of builders who built really, really bad violins.

Link to comment
Share on other sites

No.

And what do you learn from tinkering about with an FEA model?

Tinkering with FEA in general. It is a good way to solve a number of mechanical problems. Forget violins, I keep saying that.

IF you want to look at small changes of violins, you can get a good qualitative picture from FEA. I have made violins 4 times as long as you. I have thought a lot about making experiments. "Tinkering" is a bit dismissive, don't you think?

Three years ago when I was working on acoustics elsewhere a physicist told me something important that has stuck with me. Don't expect to learn anything new from FEA models. All of the physics was inserted by whoever wrote the package.

Your friend is right, the general formulation of the lagrangian approach is what it solves. What else do you use to solve many-body problems? If there was nothing to learn from FEA, Boeing, GM, and all the rest would not buy million-dollar packages. I think this posting was the dumbest you have made so far.

I didn't have a double physics and math major for the first ~12 years that I built either. I don't expect it to help me make better violins, I want to use it to help automate making consistantly really good violins. One potential approach to doing this is to try engineering a material and design which works well, as is done with carbon fiber instruments. Another approach is to figure out how to modify the violin's design to account for the properties of each piece of wood, this is what I'm trying to do.

Another way to look at your statement is this, sure Strad and Guarneri didn't use physics to design violins but neither did the thousands of builders who built really, really bad violins.

"It" meaning your physics. You can't get there without numerical solutions.. Engineering a material?? That is a lot of experiments, especially if you don't know what you are looking for. You are not going to reinvent the wheel. The way to account for wood properties might be an ideal application of FEA. You can make a model and change materials in small ways. See the differences.

If you are looking for "tone" with no attention to the dynamics, forget it.

I don't care about those old dead guys. For all I know, their new violins were about like my own. I don't think there are many really, really bad instruments by people with names. Their intrinsic value means that they would have been worked over and optimized.

I just suggest you learn to do a simple task and THEN decide if it is helpful...Your hubris is astonishing.

Link to comment
Share on other sites

Changed my mind, I've argued with you enough.

One of my favorite movie scenes is at the end of the movie THX1138, where the robot cops are chasing THX (Robert Duvall) up the ladder, to the surface, and to an uncertain freedom - but - mere moments before capture, they stop and retreat, as, apparently, at that percise moment, the expense of his capture exceeds their pre-determined budget.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
 Share

  • Recently Browsing   0 members

    No registered users viewing this page.



×
×
  • Create New...