I can't check your model because of safety instructions of company. That’s why here some ideas to check your model for yourself. For further discussions maybe you could post some pictures of your model and results.
1. 2nd order shell elements with midside nodes (QUAD8 or TRIA6) have no stiffness at the rotational degree of freedom of the normal direction of element. Mostly that influence is opressed by solver but sometimes it "breaks through".
May be its the reason for a lot of rigid body motions of a model. Can you count the number of free rigid body motions or check for it?
As a solution you could use linear shell elements CQUAD4 or TRIA3. They do not have that influence because NX defines here an artifical stiffness by default for that degree of freedom and the results are not worther than with parabolic elements.
2. If you did remesh your fem maybe you have to merge your separated meshed surface areas afterwards. Check for duplicate nodes.
3. Did you really get rigid body motion modes or only modes with very small eigenfrequencies but suitable eigenforms? If the forms look like really eigenforms it also can depend on incorrect material or thickness parameters. Hence the eigenfrequency is in a simplified way propotional to square root of stiffness and proportrional to reciprocal of square root of mass a reason for the problem can also be found here.
4. A last chance to get rid of free degrees of freedom is PARAMETER "BAILOUT=-1". But its a mighty intervention because every free degree of freedom is getting an artifical stiffness value.
If your model is not connected in a right manner maybe you can’t see it with BAILOUT=-1.
Please check for duplicate nodes before.
I hope I could give you some ideas to get solved your problem. Best wishes for you and your thesis.