Solve for the bending strength of the circular section with associated reinforcement.  Compare to a square section with the bending about the diagonal.  This is the weakest axis for the square section with equal steel in all four faces.

I have seen a book that mentions this, I will get back to you tommorow with the name of the method.

I don't believe there is any recognized conversion.  Hate to say this, but you probably whould be analyzing these piles as circular sections.  If one were to fail, you'd have a hard time defending a conversion analysis thats not officially recognized.  CRSI publishes a calculation method for circular sections in the CRSI Handbook.

It is called the charles whitney method from 1942.

2 layers of half the reinforcement 0.67Ds apart, and a concrete section 0.8D deep by (As/0.8D) wide?

This is an approximation, and as such you should always ensure that you have at least 10% or so up your sleeve for any inaccuracies. I would also not that I imagine it would be very inacurate if there were any significant axial loads.

You really need some design charts in this country, this problem would be done in 5 minutes with circular column design charts (like the ones I used in Australia) I havent seen any here.

Check out this web site to see if you can find a spreadsheet (check it before you use it):

http://civilengineer.webinfolist.com/structure.htm
 

Aren't interaction charts available for circular sections?  Do you not have a program such as enercalc?  It will design the reinforcement for a circular section rather quickly.  I did this recently.  The only conversion I had to make to a "rectangular section" was during the shear checks.  There is no clear 'd' for a circular section.  At that point, for the shear check, ACI gives guidance in 11.3.3.  
If you don't have a program like enercalc to design the flexural reinforcing, I would look for those interaction charts for circular sections.