I suspect that the switch from sorbitol to dextrose based propellant was the reason the TR-1 motor experienced a cato. More specifically the fact that most Dextrose is the monohydrate form which contains molecularly bound water, Even more specifically, the fact that I didn't remove all that water is the reason. I believe that the extra water lead to grain slumping which was the ultimate demise of the TR-1 motor. Fortunately I cast an extra segment and as that one wasn't used by the end of the day of testing it had been exposed to the ambient temperature of the day which was in the upper 90s. It had been in the same location as some similarly sized KNSB segements cast by Rick. The KNSB segments looked completely fine while the KNDX segment had the consistency of taffy and the core had all but disappeared. I had some concerns after casting the segments. There were a number of observations made during the casting process that were likely a result of the water content. It melted quickly and poured easily, but the most obvious and concerning indicator was that the burn rate couldn't be measured because burning was so unstable that it kept self extinguishing, I assumed that wouldn't be the case when fired in the motor burning under pressure. I didn't consider at the time that the water content may effect the dimensional stability of the segment.
I assumed that most of the water would be driven off during casting but that was apparently not the case. More on that in a bit.
As far as dextrose based propellant (KNDX) goes I decided I needed to get more experience with it. The first step was to remove the water from the dextrose and from anhydrous dextrose. This was done (as suggested by Richard Nakka's site) by dessicating the dextrose in an oven at 175 f for two hours. The sample weighed 901gr prior and 825gr after, approximately a 09% reduction.
The dextrose came out of the oven caked together. The clumps were pretty tough; initially I thought I could push them through a sifter but ultimately had to use a coffee grinder to break them up. I cast a small sample to measure the burn rate from. Melting took much longer to commence and the resulting propellant was noticeably more viscous. The resulting propellant sample burned well and the resulting burn rate was in the expected range.
Back to the TR-1 grain segment. Those segments we approximately 5lbs each and took about an hour to melt and cast. One hour at a nearly twice the temperature used in the desicating process seems like enough to remove the water. The problem though is that not all of the mixed dry ingredients were added at the same time. Typically small amounts are added and once melted more dry mix is added. The result is that the last amount added spend much less being heated. I'm sure some water was driven of those initial TR-1 segments but clearly not all.