A discussion of the practice of re-flux and reflow of BGAs to rework them has emerged within the "limits of flux residue" discussion.  I'm taking it out into a separate thread.  This rework process injects flux under the BGA, then reflows the BGA.  The part is not removed.  It will frequently make a part that wasn't working work again.  The heat damage to the board is less (fewer cycles) than a replacement process.  The question Joyce asked about whether there are standards and reliability studies has not been answered.  What is the proper procedure for this rework process?  Are there official standards?  Studies of effectiveness / reliability?