Strain gaging of Tube in torsion

[JordanL]

Hi Everyone,

I am trying to measure torsion in a thin walled tube (6.750" OD, 6.500" ID). The tube is also subjected to bending and compression.

I purchased some shear strain gauges and have nothing but nice things to say about the gages themselves. I mounted them 180 degrees apart on the shaft in line with the axis of the shaft in a 4 arm wheatstone bridge. I took great care to align these gages, and have no reason to suspect misalignment to be the cause of my problems.

When I hook up the gages to the strain meter, they register torsion just fine, and we have been able to calibrate our setup very accurately. The problem is that when we apply a bending load to this tube, we get a torsion reading. This seems non-physical as any shear from the bending load should be compensated out in a 4 arm bridge.

I have checked all my wiring, and tried mounting new gages, but got the same result.

Any thoughts on what might be causing this, or what I could try next?

Thanks!

 

[Engdoitbetter]

Hi JordanL,

If the tube is not rotating then I would try to put the gages on the neutral axis so that any influence of bending is eliminated.
Is that possible that a little misalignement in bending load could cause torsion?
Can you post a sketch/photo?

Hope it helps

Stefano

 

[rb1957]

you're right, that is odd ...

1st off, i don't think i've ever heard of "shear s/gauge" before. rosettes sure, but maybe these are "special" rosettes tricked to provide only shear stress output ? and then maybe your wheatstone bridge sums the gauges to output torsion ?

2nd off, it's pretty "odd" IMHO to try and read bending moment from shear stress. sure bendng induces flexural shear, and both gauges will be reading some shear (i'd've thought that the shear stress due to bending would be reasonable constant across the section with the opposite directions (CW or CCW) on the opposite sides of the tube (as opposed to torsion which'll have the same sense (CW or CCW) on opposite sides.

What sign convention are you using ? CW/CCW or up/down

 

[Engdoitbetter]

as far as I understand reading the catalogue, they're called shear SG since the wire is arranged at 45° so that, if you put them in line with the axis of the shaft, only linear strain due to torsion is read. I've also checked that using full wheatstone bridge, if the SGs are correctly positioned for torsion reading, then any effect of bending is self-eliminating.

Other points of view?

 

[BrianE22]

There will always be some cross-talk. What are the relative strains, torsion/bending?

 

[JordanL]

Engdoitbetter - I placed the gages 180 degrees apart on what I suspect is the neutral axis. I cannot see the inside of the tube to ensure it is concentric to the outside, but I suspect I have them placed on the neutral axis by virtue of geometry. The error I am seeing is quite high, and the gages seem to be placed straight. I am also under the impression that a 4 arm bridge correctly placed to measure torsion is insensitive to just about everything but torsion.

rb1957 - Here is a link to the strain gages I purchased, they are actually two gages at +45 and -45 degrees encapsulated in one package. I'm not trying to read bending from shear stress, I'm trying to read torsion. My problem is that I want my setup to be bending insensitive, but it is not.

BrianE22 - I don't have any relative strains handy, but an applied load of 1100 in-lb of bending reads anywhere from +500 to -2,750 in-lb of torsion, depending on where the load is applied. I can work out the strains if you would prefer them to the calibrated torque readout.

Some of my thoughts are this:
1) Could an eccentric neutral axis cause a bending load to read out as torsion? I cannot physically verify any internal dimensions of the tube.

2) My wiring is somehow wrong, despite being able to get torsion readings just fine (see sketch below).

3) My gages are misaligned. I would estimate that at most they are maybe a degree or so off.

Sketch of setup

Thank you all for your help so far, much appreciated!

 

[rb1957]

is one arm of your s/gauge reading poisson stress ?

it'd be nice if you could pick off the individual gauge arm voltages.

 

[Engdoitbetter]

Is bending the result of a pure moment? If not, then you should have shear due to the force applied.
In that case I believe the signal would be interpreted as torsion, wouldn't it?
Any thoughts?

Stefano

 

[Engdoitbetter]

In addition, remember that shear stress is max where bending stress is zero and viceversa.

 

[rb1957]

flexural shear should be in the same direction on both sides of the tube (as opposed to torsion which will be in opposite directions).

i believe the bridge is effectively subtracting the two gauges, 1 + (-2), which should give zero for bending (or shear) and double for torsion; yes?

 

[JordanL]

I believe that one arm of each gage would be measuring poisson stress in the +45 direction (-45 if torsion is applied in reverse).

 

[JordanL]

The idea is that shear caused by bending is equal for both gages, and gets cancelled out in the wheatstone bridge. Practically speaking, something is causing this not to happen.

 

[btrueblood]

Is the tube as short as shown in your sketch? If so, then 3d effects of the bending load may give significant deviation from "pure theory" - dunno, but I'd try looking at a FEA analysis. Wondering how you apply the bending load - as a pure couple, or by just applying a tensile/compressive load at one point on the edge of the flange?

 

[rb1957]

"I believe that one arm of each gage would be measuring poisson stress in the +45 direction (-45 if torsion is applied in reverse)."

thinking about this for a second longer, and remembering mohr's circle for pure shear, the stress along arm 1 direction is +ve shear, and arm 2 -ve shear, so the strain in direction 1 is stress/E*(1+.3) and in direction 2 is -stress/E*(1+.3) ... yes?

does that mean that your output voltage is 5.2*shear stress ? (you're summing the voltages from all four arms, yes?)

but this doesn't explain why you're seeing voltage from bending ...
is it a significant voltage ?

 

[dhengr]

Engdoitbetter:
You may want to reconsider your post of 9NOV12, 12:31, I’m not sure that it is universally true. And, in this case it is probably not true. However, from JordanL’s sketch, I don’t understand how he imparts a bending moment in the cantilevered/fixed ended shaft. I do see how the braking system will impart a torsion.

The direction of the torsional shear stress or strains which can be read will be parallel to one of the legs on your strain gage and will show a strain reading; while that stress or strain is almost perpendicular to the other leg on that gage. Torsional shear stresses tend to wrap around the shaft at about 45°, don’t they? What does this do to your results? And, which leg on the other gage (gage B) is the primary leg, which is the perpendicular leg, and how are they all wired? Reverse the direction of the torsion and the primary leg on each gage will change/switch. This shear stress is max. on the surface of the shaft. How do these gages measure normal stress/strain due to bending and the associated shear stresses? Certainly, these bending type stresses/strains are arranged/oriented differently on the surface and through the wall thickness of the shaft.

 

[desertfox]

Hi JordanL

Look at how they position gauges for torsion here:-

http://www.omega.com/faq/pressure/pdf/positioning.pdf

 

[Engdoitbetter]

Desertfox,

The same SG arrangmeent is used for torsion and shear, so if we have only bending and shear we won't read bending but only shear.
Now JordanL should show us the way he applies bending to the tube...

 

[JordanL]

btrueblood - There is a section of the pipe that has been machined down for attaching the gages. I did not show this on my sketch for clarity. I will give FEA a try as the length of the pipe as compared to the diameter is about 3:1.

desertfox - My gauges are positioned just like in figure F in the .PDF you linked to.

Engdoitbetter - Irregular brake friction is suspected to be causing the bending loads (this may be difficult to visualize). There is also a bit more to the actual setup that I am not showing in my sketch for clarity. For various reasons outside of my control, these bending loads cannot be removed by redesigning the design.

The same strain gauge setup can be used for bending and shear, but the difference is in the wiring. For the torsion setup (figure F), the shear seen by each gauge is in the same local direction. The shear strain setup (figure E), will have each gage experiencing opposite local strain.

Also, in the same .PDF, the table at the bottom says a full bridge "Shear & torsional" setup will have axial and bending compensation.

I want to thank all of you for your help so far, I really appreciate it! I think this problem is not a strain gauge problem, but something else. For curiosities sake, I'll update this thread if I ever find out what it is.