Derg:
actually both two-spool and three-spool engines have shafts inside shafts
These shafts are subject to TWIST forces and are located by splines. The T-970 series shafts are shorter than the existing shafts in the RR RB-211 seign so far. We also know that this engine is designed to have a MINIMUM MASS for a MAXIMUM THRUST.
Now you can imagine that the twist forces are not constant. As the engine spools up the forces get progressively greater. The shaft seeks to TWIST and moreover the shafts VIBRATE.
I hope now we can see why the splines have been under so much stress. Not only the expected stress from the tube but also the UNEXPECTED self destructive stress from VIBRATION.
when discussing different load components acting in any joint it is important to evaluate actual values of these load components. If not possible to measure than standard evaluations are taken assuming that the loads would be of similar type to existing known solutions (trent 800 et al). The twist load is definitely calculated and appropriately taken care of. Even if combustion may not be constant process but the force to the shaft is transferred via hot air which is a lot more compessible than stiff disc so the total torsional shaft load is very much smoothered by this.
So the question now is: How come RR could not mathematiccaly model the variables with computer software? Surely these harmonic, self destructive vibrations could have been easily predicted.
not really and for two good reasons:
(1) difficulty to define the vibration characteristic of the vibrating parts - especially when we have to take into account non ideally stiff (so non linear) elements like spline coupling and bearings
(2) difficulty to define the source, type and size of excitation: bearings, other rotating elements and their relative influence
You might say..surely NO! How could this be? How could they have made such a basic fundamental error? That is the question we all want to know. RR has been making this basic design for 50 years, this phenomena is not unknown.
yes, the phenomena is well known but the new solutions are being developed using new techniques: instead of making models and improving them RR try to digitally model and improve the digital model. But if it is easy to model the KNOWN forces like thrust and imbalance tolerences than the vibrations and wear are the real problems, especially if they develop into not foreseen way.