assigning velocity Initial conditions

[jisb]

Hi,

I have a question on regard of initial rotating velocity to specify before the STEP keyword. Abaqus supports both:

INITIAL CONDITIONS, TYPE=ROTATING VELOCITY,

and

INITIAL CONDITIONS, TYPE=VELOCITY.

For my problem, what's the difference between using these two options? If for the second one, can I specify the rotating velocity? How do I specify the rotating axis?

Thanks very much!

Sean

 

[xerf]

In the case of INITIAL CONDITIONS,TYPE=ROTATING VELOCITY , the traslational velocity is defined globally (more like the translation of rotating axis). This rotating axis is defined by the coordinates of two points on the 2nd data line folowing the INITIAL CONDITIONS keyword.


The definition of velocity in the case of using INITIAL CONDITIONS, TYPE=VELOCITY is related to the nodal degrees of freedom. The allowed degrees of freedom depends on the type of element you use. Some might have only translational degree of freedom. Such that to have an equivalent to a rotating velocity field (defined by INITIAL CONDITIONS,TYPE=ROTATING VELOCITY ) you should prescribe different velocity vectors at each node.

You can find the exact syntax in "ABAQUS Keywords Reference Manual"

 

[jisb]

Thanks for the clarification. Unfortunately, when I use TYPE=ROTATING VELOCITY, in abaqus 6.3.1, there's a memory leak when processing these initial conditions, and failed. I was wondering if you could elaborate a little more on TYPE=VELOCITY to specify an initional uniform rotating velocity. I was hoping this might help avoid memory problem. THanks very much!

Sean

 

[xerf]

What do you mean by memory leak ?

If you use INITIAL CONDITIONS, TYPE=VELOCITY then you'll have to define for each node the velocity vector such that
the resulting velocity field to correspond to a rotational motion with uniform rotating velocity( i.e. you have to compute by yourself the velocities vectors V|node=OMEGA x position vector (from rotation axis to node) , then you have to decompose each nodal velocity vector in components corresponding to the traslational degree of freedom , namely U1,U2 (and U3 for 3D problems)).

I don't think this is the common way for doing this task.

 

[jisb]

Got you, thanks!

The memory leak -- the preprocessor failed while failing to request additional memory than maximally allowed on the computer.

 

[johnhors]

That's not a memory leak. You either need more RAM in your computer, a larger swap file or both.

A memory leak is caused by poor programming, where a subroutine grabs memory when called by the main program segment and then fails to release this memory when the subroutine exits, if this subroutine is called many tiimes then all available memory resources are rapidly devoured! This is not your problem, your model simply requires more resources than are available for it to run.

I very much doubt if anything as bad as a memory leak will find it's way into a released version of Abaqus!

 

[jisb]

Right, thanks for the clarification.

My Unix machine is with Solaris 9 OS with 6G RAM. However, I was told that the maximum RAM addressible is 2G, and if I need to address a larger space, need to go back to Solaris 8. Is this true? Thanks!

 

[johnhors]

It's been some years since I last used unix and that was HPUX, anyway I thought all unix OS's (apart from x86 based Linux) are 64bit systems and that the 2Gb limit only applied to 32bit systems i.e. windows 2K, XP ecetera (2 raised to the power of 31 = 2.1 Gb).

I wouldn't down grade your OS, I doubt if that would fix your problem. Why are you still using version 6.3 when 6.5 is now the current version? Have you tried forwarding your problem to Abaqus support?