The bubble powder was not Bullseye.
Going to 1700 would imho be exceeding where you want to be with Bullseye glass.
The 1700 would have been air temperature read by thermocouple inside the kiln.
the glass would have been a heat sink on the way down and would keep the glass at an elevated temperature for much longer than you think.
Given the anneal temperature was once again air temperature and only for a short period of time not sufficient to dissipate all the elevated residual heat, I'd suggest a lack of annealing.
So you have incompatible material in the bubble powder perhaps used in excess, a high temperature above that needed to fuse Bullseye glass which may have changed the COE's and a lack of annealing.
Did you take into account the thickness and thermodynamic properties of your kiln shelf in determining your short annealing schedule? Obviously not.
You did not specify if the Bullseye glass had been fired before and had thus already experienced heat work.
Simply put, ditch the bubble powder, fire to lower temperatures, use virgin Bullseye glass and anneal longer.
Might want to go and read Graham Stone's firing schedules for glass cover to cover a couple more times, not for the schedules but for the other information on all the pages.
The glass has told you there are limits to what it can take.