Unless your system is as large as the Earth, the gravity force has a constant vertical direction, no matter how you orient the system in space

You can use datum plans to partition while assuming the Origon is the direction of the centripetal force by constructing geometry. All the datum plans must be limited at their centers and coincide with control points along their edges when the datum plans are defined. Create a plan in Sketch, then divide it up using the parameter manager. The file is attached for your understanding.

In addition to what I already said, the analytical field does not allow you to apply the force using the coordinate system you like. The analytical field's circle equation only works with Cartesian coordinates; the spherical case requires P-R-Theta. Utilizing the spherical coordinate system, you can orient it using polar coordinates to obtain the desired orientation (https://mathinsight.org/polar_coordinates).

Thank you! I will try it tomorrow. Tufail Mabood

That is exactly why I encounter this problem - this model is for the Moon. Denis Benasciutti

Tufail Mabood Thanks for your answer again! I have just checked the cae file. But it seems that the solution only works for concentrated force. For element-based body force, I cannot arbitrarily select the spherical coordinate system (csys-1 in this case). Please see the attached figure...

Abaqus doesn't offer alternative coordinate systems with body force. Body forces are always regarded as volumetric forces and can be applied with the global default coordinate system. Contrarily, concentrated loads are thought of as vectors with orientation in any direction, making it impossible to apply body force with the flexible coordinate system you're looking for. Also ineffective is the analytical field. When doing partitions, you must split the analytical field as a piecewise function since you need to create circle equations for analytical fields.

Thanks for the clarification. Then I think the spherical coordinate system cannot work for my problem...

But I think I have circumvented this problem by re-meshing more frequently and updating the body load distribution in new models. It is quite a tedious process, but at least works. Great to discuss with you! Tufail Mabood

How you overcome the problem with remeshing?

Tufail Mabood I first set up body force using Methods 1 or 2 in the original post. It has no problem except when the element deformation becomes significant and the body force becomes inaccurate.

I then apply re-meshing (more precisely, mesh-to-mesh solution mapping) technique to set up a new model with the initial states mapped from the previous model (using *MAP SOLUTION) but with the body force updated (using Methods 1 or 2). As long as the deformation within one single model is not too large, I would expect the solution is accurate enough (I will need to conduct some sensitivity tests later to verify this claim).

You can create two body forces: one for 1-component and other for 2-component. Attached figure shows creation of the first body force (BR). The expression for the analytical field corresponding to the second body force (BZ) is Y/sqrt(X*X+Y*Y) and Component 1=0, Component 2=-1 This way central oriented body force with no variation along radius is created. It is possible to multiply these expressions by any function of radius sqrt(X*X+Y*Y) to create body force varying along radius. I suppose that center C is (0,0).

That's kind of tedious Job.

I used this technique in Method 2). But it can not solve large deformation problem... Victor J. Аdlucky

Tufail Mabood You are right... That is why I was looking for a smarter method (つд⊂)