the new VRS web site:
http://www.s-92heliport.com/vrs.htm
jellycopter,
Thanks for the great post, it is this stuff that we need more of, where actual events are described in enough detail to help distill the real story. Let me try to describe what was probably happening aerodynamically to your helicopter, which I believe is entirely consistent with the data on the web site:
As you describe, the demo you normally give was really not full VRS, mostly, but nibbling (even biting) at its edges. The downwash velocity for a 13,000 lb Puma (Is this an OK value?) is about 32 ft/sec (1950 ft/min). If we follow the rule of thumb for VRS, we can see that we should start getting worried about VRS at about 50% of the downwash speed, and by 75%, katie bar the door.
With a torque fall=off of 20% in the demo you gave, you might get from 750 to 1000 ft/min descent, for most helos (sort of exactly the opposite as a forward climb, where the ETL gives you that climb rate.) So in most helos, if you shed ETL by decelerating, and make no collective/power adjustment, you will get 800 ft/min descent. For a Puma that is probably 40 to 50% of the downwash velocity, right at the edge of our calculated VRS speed. Note that one of the first effects of VRS nibbles is to increase the power required (see the cartoon I drew of the power required curve, where there is a hump at 7 knots as you descent at the .5 Vdown speed). This means that at this descent rate, the bottom begins to fall out, and you get increasingly more power shortfall, and therefore more descent rate (if you don't raise the collective). You ended the demos in the past with an acceleration forward, and all was fine.
The yaw wiggling is the rearward flight, mostly, but as you enter vrs, the yaw is in concert with the torque jumping, as the rotor flow changes rapidly enough to make the engines work up and down to keep up with the varying rotor torque. Those torque jumps are a fine indicator of VRS. Note the load factor trace on the V-22 as it gets into VRS, where the smooth slow oscillation (probably the product of a poor autopilot which is not optimized at that regime) becomes a choppy, random one. The load factor gets screwy as the rotor develops randomly oscillating lift and therefore oscillating torque.
In your one massive descent case, the initial descent was a bit more, and the conditions were perfect for the death spiral of descent/power required rise/descent to push you right into full VRS where the simple recovery you used was not enough. Your cyclic effectiveness loss was probably due to the rearward flight effects, where the horizontal tail might have stalled backwards, and the rear part of the disk also lost more lift that the front, leading to the inability to get the nose down. The Puma is a relatively low control power machine, anyway, and you were certainly outside of its normal envelope. Were it daytime, you could have pedaled it around so the nose pointed into the wind, which would have helped things (but at night, the disorientation would be a real problem!)
I believe the old pilot lore of lowering the collective is actually wrong, and more likely to harm you during recovery with greater altitude loss. If you have adequate power to climb vertically at your weight, raise the collective during the recovery, which increases your downwash speed and helps break the VRS. The old experienced Sikorsky test pilots who taught me said that, and it always worked for them. In the old days with piston powered helos, this was never the case, and increasing the collective could cause rotor droop, which is probably the last thing you need!
In a high powered helo, I would increase the collective to max power, but not droop, lower the nose and get out of dodge that way.
This set of events seems to show that the old rule of avoiding the 50% downwash region is a swell idea.