I think you might be right.
--- old post above --- --- new post below ---
Ok we use 32c as our SFT and some of the curves are <100m radius (very tight)
The steel sleepers our crimped ended, but we are narrow gauge (1067mm) down here so the sleepers will be lighter also.
I think that's the problem. You've got basically an obsolete style of steel sleeper which has been proven to have poor resistance qualities on very tight curves. The SFT is the same as the UK for crimped steel sleepers.
Current UK standards for new track have no CWR to be installed on curves of less than 250m radius. For curves of 500m or less, Lateral Resistance Plates must be installed. This of course refers to standard gauge track
Basically, I'd try increasing the tracks resistance qualities initially. Lateral resistance plates, more ballast, big ass full depth concrete sleepers on the curve.
I'd consider conversion to jointed track with breathers each side (though you can't effectively manually pack steel sleepers; this may necessitate a change of sleeper type unless you've got Kango hammers and a tamper on call to deal with any joint related geometry problems.).
It's already proven the rail is wanting to move considerably in the colder months, so if you really anchor the track into the ballast, the rail is still going to want to contract and put itself under a lot of tension, if resistance is sufficient then it may just end up pulling itself apart. I've seen some incorrect overstressed CWR in the winter pull apart and it was so neat you'd think a skilled burner had used a gas axe. Correct stressing and regular ultrasonic inspections for weak points should help mitigate this I'd imagine.
Just my tuppence worth, I'm not the wisest of birds for PWay engineering but glad to throw something into the ring.