[MBDyn-users] Deformable joint
masarati at aero.polimi.it
masarati at aero.polimi.it
Fri Oct 30 21:49:31 CET 2009
> See comments below...
>> From: Pierangelo Masarati [mailto:masarati at aero.polimi.it]
>> Sent: 10/26/2009 9:14 AM
> <snip link to development HEAD>
>> The new element is called "viscous body".
>> The syntax is defined here
>> The formulas are defined here
>> <http://www.aero.polimi.it/masarati/tecman.pdf> (Section 9.3)
>> The element provides a force and a moment that depend on the absolute
>> linear and angular velocity, projected in the reference frame of the
>> node. The angle of attack is defined as the local tan^-1(V_z/V_x), so
>> aerodynamic forces
>> L = 1/2 * rho * V^2 * S * C_L( tan^-1( V_z/V_x ) )
>> when linearized yield
>> Delta L ~= 1/2 * rho * V_x * S * C_L_/alpha Delta V_z
>> when V_z << V_x.
>> The aerodynamic moment behaves similarly.
>> You can probably produce this behavior, when the velocity does not
>> change too much, using a linear constitutive law. Modeling the flight
>> mechanics of an aircraft would probably require an ad-hoc constitutive
>> law. As far as I can tell, there is no general model that can directly
>> suit your needs, you'll need to develop it yourself based on your
>> specific needs.
>> In tests/joints/viscousbody/ there is an example to test the element and
>> to serve as a starting point for your model.
> The new element you created works great for my needs. Many thanks!
> I find that 'position' is not optional, so we will probably want to change
> this to be consistent with the documentation.
Right, the offset is not optional, but the "position" keyword is, for
consistency with other joints. I've fixed the manual.
> I appreciate your comments regarding aerodynamic forces. I think that in
> general I would prefer having a viscoelastic element, and use a strain
> constitutive for aerodynamic forces. The aerodynamic angles can be
> absolute angular displacements projected in the reference frame of the
> same as the viscous constitutive. If you agree and offer some guidance, I
> can attempt to code it.
the "elastic" part makes little sense: the angle of attack is defined with
respect to the absolute velocity, projected in a reference frame attached
to the body. The elastic part would necessarily need to be referred to
the absolute reference frame, which is definitely not physical.
> For my immediate needs, the strain equations have the form Y = k1*roll, X
> k2*pitch, and Z = (k3*roll + k4*pitch), so a collection of force elements
> might be more expedient and possibly would perform better than a general
> viscoelastic solution. I would appreciate your thoughts on using node
> private data to implement these equations with force elements.
Please define "pitch" and "roll"; extracting that information from the
motion of the node may or may not be trivial.
More information about the MBDyn-users