Induction motors don't have a consistent current vs. torque relationship. In order accurately control or limit torque, VFDs must determine an equivalent circuit for the motor then monitor and control the electrical parameters that determine the torque. I believe that VFDs that have a good torque limiting capability will have a specification stating the accuracy with which they can do that. They will only be able to state what can be done at the motor shaft. The variations in torque lost in the gearbox will significantly degrade the torque limiting or controlling accuracy. I suspect that finding out what to expect from the gearbox will be the most important and most difficult task in this situation.
Can you state what VFD models you have experience with and/or are considering for use in this situation?
Re link in comment.
If you click the "Specifications" tab, you will see "PowerFlex 700 AC Drive Technical Data," a document that you should download. On page 4 of that document, you will find:
You might find another manufacturer that can do better, but I doubt any will do a lot better. When a speed-controlled drive goes into torque limit, the motor will slow down until it reaches an operating point where the limiting value of torque is sufficient to drive the load. If no such point is reached, the drive will be at a standstill. At that point, the torque should not exceed the setpoint, but it may be less than the setpoint.
Re Answer Posted by Asker
I went through your answer, made a diagram and added some notes as shown below. The way I interpreted your numbers, my calculations gave slightly different results, but I put yours on the diagram. If you you used "best estimate" losses, I would suggest that you also do the calculations using the highest and lowest losses that you think might be possible. I believe that the lowest losses will show the smallest margin between normal operating torque and the failure level torque.
It seems to me that you might get better performance by sizing the drive for the maximum desired operating torque. You calculations seem to indicate it is sized at the failure level torque.