Measuement of clutch characteristics 2

[MikeyP]

OK, not one question but several related ones...
This is all related to *fully engaged* clutches - my brain isn't ready to start dealing with slip just yet.

1) Does anyone out there actually measure clutch/damper torsional characteristics?
2) Do you rely on manufacturers information for stiffness and hysteresis?
3) If anyone does measure them do you just measure at a single frequency of excitation do you measure the full torque vs displacement vs velocity restoring force surface?
4) Do Powertrain developers actually specify clutch characteristics as part of the design process or do they just stick an out-of-the box clutch assembly on their prototype engine and transmission, get a box of springs and play around until they get what they want by way of NVH?

TIA

M


--
Dr Michael F Platten

 

[SomptingGuy]

1) Yes. I've often seen measured static curves. I've even seen dynamic ones before.

2) ?

3) The measurements I've used have been full torque vs displacement from stop to stop. Plus hysteresis loops on the drive and coast slopes over small angles at different frequencies.

4) They will generally have a resonant frequency in mind and choose the basic characteristics to meet it. Not sure about the friction rings, preloads, idle stages though.

- Steve

 

[MikeyP]

Thanks Steve.

How much do the friction dampers play a part in the energy dissipation? Does most of the dissipation happen through hysteresis?

Basically if I took the results of a static test (stiffness and hysteresis) and applied them to a dynamic model, would I be 90% correct, 50% correct or 10% correct? I'm thinking mainly about simulating/controlling idle rattle issues here.

Thanks

M

--
Dr Michael F Platten

 

[SomptingGuy]

The clutch I modelled in detail (c. 1992) had a friction plate in it. The static tests showed a clear 10Nm offset between the forward/reverse parts of the loop. Our dynamic measurements had a 1deg amplitude. They still had the same 10Nm offset, but as the frequency increased (we had measurements at 50Hz, 75Hz, 100Hz), the loop became more eliptical. At 100Hz the effect of the damping was a "bulge" at 0deg (max loop velocity) of about 5Nm over the "static" loop. I concluded that the viscous damping term was important.

I concluded that to model this clutch damper, I required a spline for the static torque, plus a friction torque dependent on the direction of motion (with nicely rounded corners) and a pure viscous term. A model like this matched the measurements very well.

It was a transmission rattle model.

- Steve

 

[MikeyP]

Awesome. Just what I was looking for.

M

--
Dr Michael F Platten