PDA

View Full Version : Summary: Design of a torsion testing system for an existingmaterials testing machine



John Costi
09-26-1996, 12:21 AM
About a week ago I posted a request for information on custom made torsion
testing systems for existing materials testing machines. I received a
number of quite interesting replies which are included below. I now believe
I have a design which will address all the criteria which I had originally
specified and I will briefly summarise the main features of it at the end of
this message. Thank you to those who replied to this message and if anyone
requires any further information or has any further suggestions I would be
interested to see them!

********ORIGINAL POSTING***********
Hi fellow Biomechanists!

We have an uniaxial Instron model 8511 servohydraulic dynamic materials
testing machine and I am currently designing a torsion testing system that
we can use with the existing setup. The Instron has a crosshead mounted
actuator (total stroke 100 mm) with a 20 kN load cell mounted on the base of
the machine frame.

We are looking to conduct sheep spinal torsion tests as well as other
torsion tests in the future.

There are certain criteria and constraints which I need to account for when
considering this design and they are :

1) Converting the linear actuator movement into torsion
2) Converting the torsion passed through the specimen back into a linear
force to be read by the load cell
3) Allowing for a pre-load to be placed onto the specimen prior to testing
4) Providing pure torsion only in the specimen during testing (ie no
additional compression or tension after pre-load had been applied)
5) Allowing the specimen to find its own centre of rotation during testing
(ie do not impose a fixed axis of rotation on the specimen)
6) Allowing for varying specimen heights

The major constraint that we have is that the load cell must be used in its
usual position on the base of the machine.

I have a design in mind which I think will work well, but I would be most
interested to hear of any suggestions you may have.

Thank you for your consideration of this design and I will post a summary of
responses received.

================================================== ============
John Costi
Biomechanical Engineer
Division of Surgery - Orthopaedic Department
Repatriation General Hospital
Daws Road, Daw Park
South Australia
AUSTRALIA, 5041

Phone : +61 8 8275 1126
Fax : +61 8 8374 0712
VoiceMail : +61 8 8468 2317
Email : john.costi@flinders.edu.au
OR scostjj@rgh.sa.gov.au

Feel free to visit our Orthopaedic Home Page :
http://wwwsom.fmc.flinders.edu.au/fusa/orthoweb/orthmain.htm
================================================== ============

************RESPONSE 1*****************

John,

I will be very interested to hear yours as well as other people's ideas about
using uniaxial machine to conduct torsional test. I have thought about the same
problem before. We fortunely updated our uniaxial machine to biaxial loading.
I think the key problem is how do you make sure that the specimen is on pure
torsional load without any bending effect. You definitely need mechanics to
transfer linear motion of the cross head into rotation of specimen (at one
end).
This is relatively easier part. The tough part is when specimen is rotated, the
mechanics should keep loading tengiential to rotation circle all the time. I
did
not quite figure it out.


Qi Liu
Developmental Engineer
Biomechincs Laboratory
Orthopeadic Surgery Division-San Francisco General Hospital
University of California at San Francisco

************RESPONSE 2*****************

John,
We too have an 8511 and have stuggled with performing torsional testing using
the linear actuator. The best answer is of course to buy the biaxial
configuration from Instron. Instron is also working on an add-on actuator
and load cell simply to releave axially generated loads while controlling and
reading in torsion only. After several attempts to perform torsional testing
of sheep and dog femurs and vertebrae, we finally sent the specimens off to
another lab with a biaxial system. If you come up with a method and fixator
to perform torsional tests, I would very much like to see it!

Karl H. Kraus
Associate Professor of Orthopedic Surgery
Tufts University School of Veterinary Medicine
200 Westboro Rd
North Grafton, MA 01536

************RESPONSE 3*****************

John,
while it is possible to modify a uniaxial machine to do torsion, I would
suggest a couple of things, first, consider purchasing a torsion cell or
a biaxial load cell, second I wouldnt try and convert your axial motion
to torsion, instead I would add a rorary component (either a stepper
motor or a hydraulic unit). Depending on you controler and software, you
may be able to set up a closed feedback loop. You will also want a
transduce to measure rotation (A RVDT which also requires an AC signal
conditioner) and wil need to integrate all of the above in your data
acquisition system.

As for pre load and constant force testing, this is easily done by running
the test in load control, at the pre load value. To allow for a free
center of rotation is a more complicated problem. You could add an x-y
bearing plae to one side, but this would be unstable in compression. I
assume that your crosshead currently adjusts so item 6 shouldnt be a
problem.

Good luck,


Ed

************************************************** *************************
* Ed Wachtel Fax: (510) 642-6163 *
* Orthopaedic Biomechanics Laboratory Phone: (510) 642-3787 *
* 2166 Etcheverry Hall e-mail:efw@euler.me.berkeley.edu *
* Dept. of Mechanical Engineering *
* University of California *
* Berkeley *
************************************************** *************************

************RESPONSE 4*****************

John,
Using a rack and pinion system will allow you to meet the requirements of
constraints 1 through 4. To meet 5 & 6 you could allow translation and
angulation of the "fixed" end (ie the end not being actuated). This could
be a relatively simple design problem depending upon the load levels to be
applied. If you'd like to run your ideas past me, please feel free.

Joel


*****************LIFE_IS_NOT_A_REHEARSAL********** *******

Joel M. Bach, Ph.D.

UC San Francisco Ergonomics Program
Phone:(510)231-9448 Fax:(510)231-9500
www.mother.com/~doc

*****************Another_Friend_of_Ishmael******** *********

************RESPONSE 5*****************

Dear John,

some time ago I designed a sorto of torsional jig that converted the linear
force into a torque. It was pretty simple but i found that for small
rotations it was quite effective and accurate. It consisted of a shaft
(supported on ball bearings) having a sort of arm to which the force was
applied (at known distance from the rotation axis). This system can be
mounted on top of you load cell (I had the same configuration), and you
only need to "zero" your load offset before starting the test. You might
find more details in an abstract that I published in the EORS conference
proceedings, 1995.
Also: Make sure you use a double universal joint (or equivalent) to avoid
any other force compoennt to be transmitted to your specimen).

Hope this helps. Kindest regards.

Luca Cristofolini

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Luca Cristofolini
Laboratorio di Tecnologia dei Materiali tel. 39-(0)51-6366864
Istituti Ortopedici Rizzoli fax. 39-(0)51-6366863
Via di Barbiano 1/10 E-mail:luca@tecno.ior.it
40136 Bologna, Italy
- - - - - - - - - - - - - - - - - - - - - - - - -
Experience is the name we give to our errors.
- - - - - - - - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - -


************RESPONSE 6*****************

John -

A worthy challenge, I think. I haven't worked with the spine but have
puzzled over similar issues for the human knee and ankle. It would be
interesting to see a sketch of the design you end up with.

Questions:
1) Is the centerline of the spine specimen oriented parallel or
perpendicular to the axis of the actuator? (Does it matter?)
2) Must all forces be generated by the Instron actuator?

Observations:
1) If compressed air is available, pneumatic cylinders can be a relatively
convenient and inexpensive way to generate forces. You might use them in
addition to the Instron actuator. They could generate the preload force,
for example, allowing it to be uncoupled from what the actuator does to
generate the torsion. The preload device would either have to move with the
torsion or pass through a thrust bearing to avoid interfering with the
torsional motion. Pneumatic cylinders are nice in that they can provide a
constant force over their travel. If you need some kind of cyclic
load/displacement or shaped load/displacement profile it would quickly get
much more expensive and complicated and other methods would likely become
preferable.

2) Using a separate device to generate the preload force might also be a
simplification with respect to compensation for varying specimen heights.

3) For some of our knee testing we apply torsional moments with minimal
restriction to center of rotation by using a torque ring on the specimen.
The ring is connected by ropes to pneumatic cylinders. The specimen can
move over a considerable range without affecting the applied moment. Torque
ring weight must be considered.

4) I question the decision to use the Instron load cell as your force
measurement device. By the time the force passes through a linear-to-rotary
device (or devices) it will be a challenge to accurately determine (and have
confidence in) the relationship between what you measure and what the
specimen actually experienced. A suitable torque cell installed at the
specimen would be invaluable.

Regards,

Dale
----------
Dale R. Knochenmuss Dale.Knochenmuss@UC.Edu
Noyes-Giannestras Biomechanics Laboratories
University of Cincinnati
P.O. Box 210048
Cincinnati, Ohio, USA 45221-0048
Tel: (513) 556-4175, Fax: (513) 556-4162

************MY REPLY TO THE ABOVE RESPONSE**********

Hi Dale

Thank you for you reply to my challenge!! *grins*

To answer your questions:

>1) Is the centerline of the spine specimen oriented parallel or
>perpendicular to the axis of the actuator? (Does it matter?)

With the design I am considering, the centreline of the specimen will be
parallel to the axis of the actuator and yes you are right its orientation
may not matter except with respect to the design you have in mind.

>2) Must all forces be generated by the Instron actuator?

We would like to have all forces generated by the Instron and incorporate
Instron's technology for loop and feedback control etc. So I am aiming to
use the actuator to generate the forces and the load cell to indirectly
measure torsion.

>Observations:
>1) If compressed air is available, pneumatic cylinders can be a relatively
>convenient and inexpensive way to generate forces. You might use them in
>addition to the Instron actuator. They could generate the preload force,
>for example, allowing it to be uncoupled from what the actuator does to
>generate the torsion. The preload device would either have to move with the
>torsion or pass through a thrust bearing to avoid interfering with the
>torsional motion. Pneumatic cylinders are nice in that they can provide a
>constant force over their travel. If you need some kind of cyclic
>load/displacement or shaped load/displacement profile it would quickly get
>much more expensive and complicated and other methods would likely become
>preferable.

Unfortunately we don't have access to compressed air but I agree with your
idea Dale.

>2) Using a separate device to generate the preload force might also be a
>simplification with respect to compensation for varying specimen heights.

Yes I haven't looked at seperate devices as of yet as I am attempting to
incorporate this in my design. I feel I am not too far away from a solution
and the comments you have given me are what I have been looking for and are
definately thought provoking.

>3) For some of our knee testing we apply torsional moments with minimal
>restriction to center of rotation by using a torque ring on the specimen.
>The ring is connected by ropes to pneumatic cylinders. The specimen can
>move over a considerable range without affecting the applied moment. Torque
>ring weight must be considered.

As above

>4) I question the decision to use the Instron load cell as your force
>measurement device. By the time the force passes through a linear-to-rotary
>device (or devices) it will be a challenge to accurately determine (and have
>confidence in) the relationship between what you measure and what the
>specimen actually experienced. A suitable torque cell installed at the
>specimen would be invaluable.

This is a concern of mine as well Dale - the general idea I am looking at
will use ball screws to impart the linear to rotational movement and vice
versa for the load cell and yes a torque cell would certainly make this
measurement a lot more accurate as well. Although I don't anticipate much
loss of tranmission with these low friction ball screws with essentially the
reaction torque of the specimen being passed directly to the lower nut on
the screw and onto the load cell.

Once again thanks for your help and suggestions Dale, and I hope that one
day I may be able to be of assistance to you!!!!

Regards

John

*******DALES REPLY TO THE ABOVE*******

John -

With respect to the earlier question:

>>2) Must all forces be generated by the Instron actuator?
>
>We would like to have all forces generated by the Instron and incorporate
>Instron's technology for loop and feedback control etc. So I am aiming to
>use the actuator to generate the forces and the load cell to indirectly
>measure torsion.

You didn't indicate what control mode you would be using with the Instron
(load, displacement, or strain control) but I suggest you tread carefully
when you first start testing if using something other than displacement
control. The control loop dynamics can change quite a lot as the complexity
of mechanical fixturing increases (backlash, mechanical resonances, and all
that). We didn't anticipate this and got surprised.

>>4) I question the decision to use the Instron load cell as your force
>>measurement device. By the time the force passes through a linear-to-rotary
>>device (or devices) it will be a challenge to accurately determine (and have
>>confidence in) the relationship between what you measure and what the
>>specimen actually experienced. A suitable torque cell installed at the
>>specimen would be invaluable.
>
>This is a concern of mine as well Dale - the general idea I am looking at
>will use ball screws to impart the linear to rotational movement and vice
>versa for the load cell and yes a torque cell would certainly make this
>measurement a lot more accurate as well. Although I don't anticipate much
>loss of tranmission with these low friction ball screws with essentially the
>reaction torque of the specimen being passed directly to the lower nut on
>the screw and onto the load cell.

Will you have any way to insert a test piece into the system to try to
measure the frictional and other mechanical effects before testing? Might
provide some peace of mind regarding the accuracy of your data.

Good luck.

Dale
----------
Dale R. Knochenmuss Dale.Knochenmuss@UC.Edu
Noyes-Giannestras Biomechanics Laboratories
University of Cincinnati
P.O. Box 210048
Cincinnati, Ohio, USA 45221-0048
Tel: (513) 556-4175, Fax: (513) 556-4162

*******END OF RESPONSES POSTED********

*****MY SOLUTION TO THE PROBLEM*******

Before I start please forgive this crude diagram you see below. It may
probably look a mess for those not using Eudora!!

The diagram below is a simple representation of the design I have come up
with. Essentially it consists of a ball screw (1) to convert linear
actuator motion to torsion. Along with the torsion there will also be
compressive or tensile forces due to the nut moving along the screw. These
forces are undesired as I require pure torsion through the specimen. An X-Y
linear bearing table (2) will be used to allow the specimen to find its own
centre of rotation during testing. To remove the undesired compressive or
tensile forces a "free floating" thrust plate (3) will be used. The aim of
this plate is to allow unrestricted X-Y planar motion whilst still
transferring torsion through to the specimen. Ball transfer units are to be
used to allow unrestricted X-Y motion and will remove the compressive or
tensile forces passed through from the nut. Note that there is a frame
supporting the ball transfer units which has not been drawn for simplicity.
This frame is bolted to the base of the machine. These ball transfer units
are needed on both sides of the thrust plate to account for both tensile and
compressive actuator movement. Pure torsion is then applied through the
specimen via the thrust plate. A torque or torsion cell (4) will be used in
place of the load (force) cell to directly measure torsion from the specimen.

Actuator
| |
| | 1. ball screw
__|_|__
|_______| nut
|| |_| ||
|| || spacer
__||_____||__
|_____________| 2. X-Y table
_o__|_____________|__o_
|_______________________| 3. thrust plate mounted on ball transfer units
(bearings).
o |_________| o Ball transfer units mounted on above and below
thrust plate
| | to support tension or compression forces.
|xxxxx|
|xxxxx| specimen mounted in cups
|xxxxx|
_|_____|_
|_________| mounting cup
4. to torque cell

I hope that this design is easy to follow and understand and I look forward
to any further comments or requests for more information on this design.


Thank you

John Costi

================================================== ============
John Costi
Biomechanical Engineer
Division of Surgery - Orthopaedic Department
Repatriation General Hospital
Daws Road, Daw Park
South Australia
AUSTRALIA, 5041

Phone : +61 8 8275 1126
Fax : +61 8 8374 0712
VoiceMail : +61 8 8468 2317
Email : bejc@flinders.edu.au
OR scostjj@rgh.sa.gov.au

Feel free to visit our Orthopaedic Home Page :
http://wwwsom.fmc.flinders.edu.au/fusa/orthoweb/orthmain.htm
================================================== ============