Success..?

[MrVibrating]

@Mods - can we get this thread cleaned up of intrusions..?

Or else is there a moderated sub-forum i should be posting in?

I'd just cleared my ignore list last week but this fraggle's making me wonder if i'm trying to do this in the wrong forum..

[MrVibrating]

My first efforts to access this asymmetry were back in 2017 (slooow progress eh?), using a pulley sim provided by Fletcher:




..and i'm thinking this will make a good 'V2' rig base for a simplified version of the present interaction.

Basically, this accomplishes a constant OB torque, and doesn't need to actively control the MoI to keep it constant; we can just assign the 1 meter radius 1kg disc an MoI of '1', make the drop-weight 1 kg, et voila, constant OB torque.


For the second MoI, we can just overlay a second, identical blank disc, connected coaxially by a motor and brake to that base rotor.

As before, then, the motor will apply the OB torque constant, spinning up the upper rotor, while perfectly counter-balancing the OB torque / base rotor, so the weight won't begin dropping until the first braking phase; from thereon, the motor will prevent the weight's gravitational acceleration, lowering it at constant velocity during the spin-up phase.


The final component needed is a basic vMoI attached to the base rotor; ie. a pair of identical, counter-balancing masses at equal radius, able to slide in and out in tandem.

Initially i think i'll control the vMoI via one or two linear actuators - if the same CF-PE gain shows up as before, then perhaps the actuator/s can be replaced with a simple spring, that gets progressively loaded with elastic PE over successive cycles..


There's only one further complication to solve:

• If the vMoI works as intended, it needs to get fatter in order to stretch the spring / actuator and thus express the CF-PE gain. No radial displacement, no gain harnessed..

..obviously though, this means the net MoI of the base rotor's increasing..

..thus if the upper rotor's MoI is constant, we're not going to be able to maintain a 1:1 ratio between the interacting inertias, so won't be getting uniform step heights on the velocity plots.. or optimum OU efficiency, if we're even still able to break unity..

Thus the only thing i can currently think of is to also give the upper rotor a variable MoI, and match it to that of the base MoI at all times..


If this does work, it has a number of benefits:

• as mentioned, the CF-PE gain could be harnessed by a passive spring

While this doesn't eliminate the possibility WM's winding us up, it reduces that risk..

• output GPE is easier to verify, as a simple function of linear GMH

• if both the interacting inertias are vMoI's then we admit the possibility of measuring the same CF-PE gain on the upper, as well as lower, rotor

• linear actuator integrals can be summed together, simplifying analysis

So we could have individual plots for each actuator, as well as a 'net actuators' plot; thus only that net integral needs taking to get a quick-ish efficiency measurement, leaving the discrete integrals for the most interesting results..

• In principle, the actuators / springs could even be external, connecting into the system via pulleys running through the axis..

[MrVibrating]

..here's a first draft of how the base rotor could look:



..having the actuator external to the system might actually be over-complicating things, due to the addition of the pulley system and any instabilities that may introduce...

..might just switch to a single actuator mounted on the rotor itself..


But you get the basic idea - this would form the 'base', weighted, rotor, and then the second rotor would just sit on top, attached centrally via a motor and brake.

The upper rotor would then match its MoI to that of the base at all times, whilst we spin'n'brake it..


If the s&b cycles result in the same CF-PE gains, they should get harvested in the actuator's P*t integral..

Note that although there IS active MoI control, it's not using reactive feedback, as the rotary solenoids were in the previous rig; there's only so much that can be done to mitigate the possibility of unmetered input energy, whatever types of constraint are used...

[MrVibrating]

..on 2nd thoughts i'm gonna mount dual actuators directly to the rotor, one for each vMoI mass, and just sum their integral - it simplifies the construction for the same net result, making it faster to sim and more robust..

[MrVibrating]

..like this:



..so i'll just use that 180° of rotation, 30.8 J of GPE for each run, spreading as many s&b cycles across it as will fit per their s&b speeds..

So now i just need to add the upper rotor + vMoI, motor, brake and target speed / braking torque inputs, and we're good to go..

[silent]

Looks good Mr.V! Just add the person to your block list and move along. Thanks for your animations and your insightful posts...good reading and thanks for sharing!

silent

[MrVibrating]

..just had a thought - you can see in advance that if the upper rotor is braked sharply, the base rotor will be jerked and the rope may briefly go slack as the base rotor out-accelerates the weight...

..thus the net KE may briefly exceed the output GPE, if you follow my drift.. something to watch for, maybe..

[MrVibrating]

Looks good Mr.V! Just add the person to your block list and move along. Thanks for your animations and your insightful posts...good reading and thanks for sharing!

silent

Cheers mate - like i say, i unblocked everyone last week, when i realised the chances were odds-on it's for real this time - if we're basically over the line already then no distraction's gonna slow me down.. just junks up the thread for anyone trying to follow tho..


I'll just say this one last time tho - anyone NOT lining their head-bra with suitable magnetic shielding is basically inviting me to steal their thoughts, so i'll make no apologies - lock it or lose it!

[MrVibrating]

..nearly there:



Red = 'base rotor'
Green = 'upper rotor'

Green MoI will mirror red's at all times, regardless of their respective speeds.

Obviously the brake's just jammed on at the moment, so all i need to do is add the spin'n'brake code, get the motor and brake metered up, add KE, velocity and momentum plots etc., and we'll very shortly be ready to pull data..


If those rotary solenoid integrals are valid, we'll see their results replicated in the above actuator integrals..

[WaltzCee]

Cheers mate - like i say, i unblocked everyone last week, when i
realised the chances were odds-on it's for real this time - if we're basically over the line
already then no distraction's gonna slow me down.. just junks up the thread for anyone
trying to follow tho..


I'll just say this one last time tho - anyone NOT lining their head-bra with suitable magnetic
shielding is basically inviting me to steal their thoughts, so i'll make no apologies -
lock it or lose it!

I've got some high-speed state-of-the-art shielding. Anyone publishing on the internet is
publishing in the public domain. And when you're publishing threats, you are inviting an
interview by federal agents.

Freedom has a price. Sneaky bastards do it behind your back. I'm in your face. :-)
Well, I'll let you get back to your original research. Strapping on the double D.

[MrVibrating]

First result's in:



..quarter-joule missing, apparently..

So, no more gain, but a quarter-joule loss seems above expected noise levels..

Strikes me now that simply widening the MoI isn't actually replicating what the last rig did, which moved one mass in whilst the other was moved outwards... so the 'gainy' part of the action might be related to the mass being slid into the center, rather than the one moving out, or else perhaps their interplay, or the fact that they're moving in and out at different, and varying, radial speeds, in order to lock the MoI..

The present MoI manipulation is too basic, i think - i'll try a few more runs at different spin-up speeds, see what else happens, but literally a dead loss for now..

There's other ways of applying the MoI-control trick, such as via more linear actuators - again using reactive feedback to 'sense' and correct MoI on the fly..

..but there's a passive way to do it, too... i just haven't figured it out yet.. like i say, the radial distance between each mass in a pair needs to vary by sqrt(2), implying that a simple sinusoidal function - such as might be implemented via a planar folding linkage or crank of some kind - could automate the process; then the radial translations could be powered by a single motor / actuator, without needing real-time feedback control...

Alternatively, perhaps i can find a simpler workaround, such as using reactive techniques to output the data and then just feeding it back in via a 'dumb' data table, with the control loop removed..



Yet another thing to try might be to re-run the 'gainy' configs, but with MoI control disabled - thus the distance between each weight in a pair will remain constant as they slide across the radius, causing the MoI of the OB axis to fluctuate, destabilising the per-cycle momentum steps etc. etc...

..this will undoubtedly eliminate the gain, but there's a small chance it could still be present in the linear actuators i suppose..

..if not, then if it is real, it depends upon that radial distance between weights in each pair changing, since that's where it's being harvested..

As for the 'V2' rig above, i could try further variations such as moving one mass in whilst the other moves out, then try doing the same thing but introducing constant-MoI control etc. etc.

[MrVibrating]

Seems a bit of a PITAS taking more measurements from a rig that obviously ain't working as hoped, however that last result was a loss, over and above the energy dissipated.

A 'non-dissipative' loss.

To wit, a non-thermodynamic loss...

Quarter-joule, vanished.


So i just took another measurement, this time at a higher spin-up speed:


V2 Rig

Turns: 0.25
TRS: 1 rad/s
Brakes: -100 N-m


Motor = 2.953637007

Brake = -0.752722488

Base Acts -0.026779835

Upper Acts -0.072898265

GPE = 5.205219

Total = 7.306455419

KE Rise = 6.388481

Diff = -0.917974419 J


..so to summarise, it seems that the V1 rig can only gain energy...

..whereas this one can only lose it..!

So we've managed to invert the asymmetry after all..

Both variations are producing an effective CoE violation.

Gonna try yet more variations...

[MrVibrating]

Fascinating.. i wanted to include some inbound radial action, as well as outbound, to more closely resemble the V1 gain conditions...

..so i inverted the sign of one of each pair of actuator controls, so that one's extending while the other's retracting; this causes 'classic OB'.

Furthermore each pair of actuators are now synced to the angles of their own rotors, and so have different MoI's at different times - so step heights are no longer constant over time..

Since they're now causing their own OB torque, the drop weight's become redundant, so remains pinned to the background for now.

Here's the result of 1 full rotation at 1 rad/s TRS:



V2 Rig

Turns: 1
TRS: 1 rad/s
Brakes: -100 N-m


Motor = 6.83657667

Brake = -3.397215396

Base Acts = 14.79641531

Upper Acts = 11.83659537

GPE = 0

Total = 30.072371954

KE Rise = 26.435029

Diff = -3.637342954 J


Bear in mind that all measurements in this thread, so far, are taken at the most painstaking precisions (v. time consuming!), which usually solve to within millijoules of unity..

The 3.6 J missing is either a particularly stupid error - as in, my fault, not the sim's - or else a massive non-dissipative loss...

[MrVibrating]

...as before, but with the upper actuators locked, so a fixed-MoI on the upper rotor:



Motor = 4.999835977

Brake = -2.505588097

Base Acts = 15.1864338

Upper Acts = 0

Total = 17.68068168

KE Rise = 15.652503

Diff = -2.02817868 J

So fairly clean-looking numbers here, showing significant losses.. GPE in / out can only be a zero-sum, so the work efficiency of the motor is substantially negative..

Note also that the energy dissipated is exactly half the motor's input energy; so the motor and brake integrals seem fine, in relation to one another.. ie. relative to the FoR of the 'base' rotor..

..but in the FoR of the ground / KE equation, less momentum - and, thus, rotKE - has been consolidated, relative to the momentum and energy value of the GPE interaction used to sink the counter-momentum.


So what might be tried to get back to gain conditions - if it's even possible with this rig - otherwise, what was the gainy rig doing that this one isn't?

[Fletcher]

That IS the question !