convergence in abaqus is a frequent problem in abaqus. It can be due to contact, time step, convergence criteria etc. It is hard to say where is the problem from your model description. You have to look into the message and status files.
By the way, there is a abaqus user mailing on yahoo.
Anh Phuong

Yes, look at your message and data file to get an indication of what the problem is.  Also, have you look at your .sta file ?  It will show you how far you analysis has gone and how many steps it has taken.  The .sta file will show you how far from the final time you are.

The most reliable indicator is checking the message file (.msg).  Check any warnings for singularities, and especially check for high force/moment/field residuals (especially the locations where they are happening).

As already stated, there are many reasons in implcit nonlinear FEA why convergence may be difficult to achieve.  Not knowing the specifics of your problem, I can't give a more educated answer.

Possibly ask some coworkers for help; if you are entitled to ABAQUS support, call your support office and they can help iron out the issues (although you'll have to be more specific than you have been so far).

It is very difficult to give a specific answer to your problem without knowing the errors and warnings specified in the message file.
 
However here are a few thjings which may help you in solving the convergence problem.

Convergence will be a problem in contact analysis when the BC is not been correctly specified. Suppose the BC is specefied correctly then check for the mesh density in the slave and master region. It is always recommended to have a finer mesh in the slave surface as compared to the master surface.
If the elements you are using is tets then it is better to use C3D10M elements. If you are using Brick elements try to match the mesh density in the slave and the master surfaces.
Try to do the initial run without considering friction.