Blood From Stone

[cloud camper]

While V ' s work is commendable and well motivated, a 2.6 joule theoretical gain over a 40 joule range has to be within the range of error for this resolution step and the accuracy of the Kutta Merson integration engine used by WM2D.

It may turn out that after further optimization a larger gain may be attainable (or not).

We will certainly need a larger gain than the approx 6% realized here just to overcome static and kinetic friction in a real world build.

Also if this gain is being produced in the first cycle, there should be no problem replacing the powered actuators with a simple spring, avoiding most of the potential integration inaccuracies with the current experiment.

This would add some confidence to the result.

Another confidence builder would be to repeat the experiment in a much higher end simulator such as Solidworks.

[Senax]

...
I've had a look at Sjack Abeling's wheel and consider it plausible, however when I saw what I thought looked like air tanks on the wheel in one of the pictures on his website, I decided to not get too excited over it.
...
silent

Please could you give me a link to what you "thought looked like air tanks on the wheel".

Thanks. 😎

(I notice that an emoji facility has been added to the website since my long break earlier in the year. Rt click and it's the first item on the list)

[MrVibrating]

Time to start building yet?

silent

Why yes, now that you mention it. Yes it is.


I think a real-life test rig would be very much in order, unless anyone has sensible objections to the above results..

Just come home from work, half-expecting a withering "you forgot to carry a 4" or "isn't gravity supposed to be vectored down, not up?", but if this isn't crass error, it's probably about as build-time as it's gonna get..


All that's needed is a weighted vMoI - the weight could be hanging off a cord that spools off the axle, doesn't have to be attached to the rotor. It really is just a simple case of moving the rotor mass out then back in, whilst the weight's still dropping.

I'll help out any way i can with calcs / sims / measuring momentum and energy..

The bill of material's pretty short:

• three weights of known mass, two of them equal

• a 'rotor' ain't gotta be 'round', could be a cross-shape, a 'T' shape, or just a simple beam spun up by a weighted ripcord..

• the rotor masses can be set to counter-balance one another using a simple pantograph / planar linkage, or pulleys, etc..

• input energy could be handled entirely by springs, but a basic electronic control system would be best

• a cheap laser tacho might come in handy, but any means of measuring speed is fine


Main consideration would be the builder's intentions - a physical test-rig designed to offer some testing flexibility should be the first priority, rather than attempting to design a PMM right off the bat.. but that latter option's firmly there on the table..


Should still prolly wait for some glimmer of confirmation by someone confident in the basic physics tho - a simmer, armchair theorist or anyone willing to stick their neck out...

[Fletcher]

You might consider a few things before a real build.

1. That it is possible that there is some inaccuracy in the sim Outputs, and that will either be found by yourself Mr V or by someone else looking behind the scenes when they get to it.

2. That any real gain should be able to result in an increase in Potential somewhere.

In the example of your experiment then two things could happen to test whether the gain is the real deal.

A. Start the equi-weights out from CoR at some small radius, and let then cycle normally. But use the predicted gain to push them together to CoR at zero radius. That would demonstrate that the gained energy has been used to increase Potential.

B. Somehow arrange the sim to do some Work, like lifting a mass of some type, or some sort of friction Load. The key is to add Load (energy usage as waste or rise in Potential) until it does not show OU. That Load is made up of energy losses to frictions such as bearings heat and air drag etc, and Load Lifting Potential. Usually I just do a quick and dirty Load by putting air drag on high (especially if few moving parts). Note that air drag can go up by linear, square or cube.

If the sim is still doing the business and handling some measurable Load then I would perhaps look at others simming it in perhaps other kinetic programs used in industry.

[ME]

CC... WM2D is pretty good.
You'll usually see differences in (say) the 4th decimal or so.
And if the gains are like mrV claims, then anything above zero (make it >0.1%) is still enormously good and reason for celebration.. even when a build need impossible accuracy right now..

Should still prolly wait for some glimmer of confirmation by someone confident in the basic physics tho - a simmer, armchair theorist or anyone willing to stick their neck out...

Here I go for the first 90 degree part.
Sigh, I really shouldn't... let's eyeball his first GIF in that post.


----------
It starts
A mass 1 kg, 2 meters above the axle: GPE=mgh = 19.6133 J
Apparently with an inital velocity (0.5 m/s): KE = ½mv² = 0.125 Joules
E.total: + 19.7373 Joules (as almost agreed by mrV)

----------
After 90 degrees:
The velocity at the rim: 2.02634 rad/s at 2m --> v=4.05268 m/s

A mass 1 kg, 0 meters above the axle: GPE=0 J
It's going at the rim with the calculated velocity: KE=½mv²= 8.2121 Joules

A mass ½ kg is now also at the rim doing that velocity: KE=½mv²= 4.1061 Joules. There are two of them.

The main wheel itself is weightless.

E.total: + 16.4242 Joules (as reported by the program 16.42415 J)
-------

Those numbers are agreed on by mrV.
As far as I notice this quickly: Nothing else is moving, no more potential-energies to calculate. Should I dare to calculate the difference?
In case someone wonders: That potential energy number moves linearly with height and the chosen base-line: that difference remains the same.

Shall I mention that the outwards motion of those weights may cause that drain, just like an ice-skater slowing down when the arms open up? Or does that sound too "snarky"?

I have yet to figure out where that magic +7.6 came from (he claims the actuators, I dare not to check)
Let's say, for the sake of (...) argument, those actuators do generated 7.6 Joules. Then why does it need to be added to the measured kinetic energy, as if it's potential?
In my best guess: it should already be part of the motion and included in the 16.4 Joules.

Conclusion... just build it if you really want to know.

[silent]

.

[Senax]

@Senax: http://www.mooieenergie.nl/en/

Scroll down and you can see the 2 air tanks in that photo where the guy is standing.
Maybe it's part of an air-brake system,
but what do you wanna bet it's part of the acceleration system at the top?

silent

I see what you mean. I thought they were some kind of piston -
but then I'm not a mechanical engineer - what would I know. 🤔

What's that blue thing by his chest. Is that the electrical generator?
It looks about the right size and a suitable colour.
https://www.youtube.com/watch?v=9UG1YU71a9w
I get the impression all the weights-going-round is to the left.
That must be the subject of the first photo.

I think I understand why he refers to it as Air Gravity.
Later I'll copy this discussion over to Buzz so as not to derail the thread.

[MrVibrating]

You might consider a few things before a real build.

1. That it is possible that there is some inaccuracy in the sim Outputs, and that will either be found by yourself Mr V or by someone else looking behind the scenes when they get to it.

2. That any real gain should be able to result in an increase in Potential somewhere.

In the example of your experiment then two things could happen to test whether the gain is the real deal.

A. Start the equi-weights out from CoR at some small radius, and let then cycle normally. But use the predicted gain to push them together to CoR at zero radius. That would demonstrate that the gained energy has been used to increase Potential.

B. Somehow arrange the sim to do some Work, like lifting a mass of some type, or some sort of friction Load. The key is to add Load (energy usage as waste or rise in Potential) until it does not show OU. That Load is made up of energy losses to frictions such as bearings heat and air drag etc, and Load Lifting Potential. Usually I just do a quick and dirty Load by putting air drag on high (especially if few moving parts). Note that air drag can go up by linear, square or cube.

If the sim is still doing the business and handling some measurable Load then I would perhaps look at others simming it in perhaps other kinetic programs used in industry.

These kinds of things are all down the road mate - they'll happen in time, if it's real.. at this stage i was just hoping someone could simply verify the measurement, its validity, just basic replication...

Left to my own devices i expect progress will continue to be incremental. First order of business is likely to be playing with the gain conditions, mapping out the dynamics in more detail - for instance i suspect that reducing the weight ratio will boost the gain margin - likewise, widening the MoI variation.

Then there's the issue of disentangling the two symmetry breaks (or however many it is) - if this suspected GPE asymmetry's real, then we've found our 'missing piece'..

..as previously noted, the KE gains from an effective N3 break follow that characteristic evolution of dissipation and momentum accumulation that means OU requires 3 to 5 full cycles, passing through an initial loss zone.

Whereas B's wheels were immediately OU before the first cycle was complete (since they contained no non-rotating weight reservoirs). So, we were evidently missing a trick..

..and now we appear to have found it. And it was the very, very last thing i expected, and it's been right under my nose for the last 12 months..


So first off i wanna verify this GPE asymmetry.

Then i wanna start cycling up, and seeing how the gain evolves over time - is it purely a function of the GPE asymmetry, or does the N3 break i was actually aiming for also start to kick in?

Are the two breaks somewhat independent, or mutually intertwined?

As much as i'd like to make out this was all cool-headed reasoning and entirely expected, truth is i'm in shock. Did not ever anticipate a direct energy gain from gravity.

So do we also get the predicted gains from buying cut-price momentum - IOW, does the energy cost of momentum square up with speed, per ½mV²?

So i'm still a long way from trying to close-loop a design. Way too premature for me.

I only meant that with the above measurements independently replicated, prospects for build success are pretty much guaranteed. If that was the no. 1 priority, there's now sufficient evidence to expect a well-designed build to work as intended. Obviously, a real energy gain can be engineered to close-loop. But you do that to build a PMM, not simply to conclude that an F*d integral is valid.

We can check that ourselves against the CF force - that's what the meter's there for.. as noted, the instantaneous force on the actuators (the value WM is reading from them) should be the CF force plus the mass inertias during the accelerations / decelerations. Hence we can pause the sim at any frame and do a manual calc, or simply knock up another meter to bang out that answer on the fly...

..having confirmed that those force values are indeed valid, there's presumably no question the displacement data is too, and that's as near to certainty as physics provides. A build will definitely work.

So i'll continue as i am for now, offering support to builds but still very much in investigating mode myself. Hardly scratched the surface.. there's important outstanding questions to be resolved. The course of these investigations will hopefully further inform build options.


Just for starters tho, in principle, it should be possible to hold KE constant as the weight descends, by extending MoI at just the right rate (another use for reactive feedback control, maybe), thus converting all output GPE directly back into sprung PE, Joule-for-Joule as GMH ticks by.. How much greater than GMH can GPE as a function of time, be? Eh?, See, these are the questions... this is why i'm here (with no idea WTF i'm doing).

Likewise, maybe we could aim for a specific split of PE / KE, like 50/50 - so if we have a GMH of 100 J, try to convert that to 50 J KE and 50 J of PE - does this cause a 150% gain, or what? Etc. etc.

You're trying to think of ways of 'proving' the gain - but the only validation i want from anyone else, for now, is from their own confidence in their own knowledge and calcs - it's that confidence vote, not some kind of pronunciation ceremony, that would just be a small comfort. But all of that will come in its own time. No pressure or obligation on anyone, have fun spectating, but it remains ongoing research, not a PMM sideshow..

I'll proceed by verifying the gains myself using the above-mentioned checks, and then continue mapping out the dynamics, optimising ratios, timings etc. etc. Verify the GPE break, then see if the N3 break's there too..

[cloud camper]

More good stuff but still no indication how smooth mathematical curves in weight paths allow for a break in symmetry.

Smooth mathematical curves automatically imply a 1/1 conversion of one form of energy to another (before frictional losses) leaving no room for a gain.

Also a 6% gain per cycle (if real) is not enough to overcome even the actuator or possible spring losses before even considering static and kinetic friction.

Electronic actuators if used are activated by DC electric motors which are no more than 80% efficient. These are basically servo motors which also consume energy during idle conditions in order to maintain their commanded position.

Expensive AC brushless motors can achieve 90% efficiency but are not used in actuator applications.

So if a 6% gain per cycle is even real we're already in the hole 14% before taking friction into account.

Also, actuators are not designed to recover energy only provide it so on the outstroke of the radial weights no energy will be recovered so this will be a complete loss to heat as the motors are spun against back emf.

If the actuators are replaced by a spring we immediately run into the problem of how to effect timing of the radial weights as the electronic timing is gone.

We then need a dedicated stator to control timing which defeats the entire purpose of the experiment.

But a great mental exercise!

Not trying to be a Debbie Downer here but there are many other issues that come into play!

[MrVibrating]

Hooray it's the weekend! Let's make some energy!

Weight reduced to 0.5 kg (all three masses now equal)

First 90°:

GPE = 9.80665 J

Net KE = 7.71229 J

F*d = 6.60746058 J

Acts P*t = -2.35027846 J

CF P*t = -2.34672772



So the new F*d plot is kinda out there on its own, eh? Looks suspect to me..

I'm concerned it's measuring power rather than energy, since it's 'force times distance times time' (since it's F*d integrated over time).

If we multiply it by time again, does that give us energy? Or divide it by time? Dunno.. more testing required there..


But just going with either of the other two metrics, it's still comfortably OU either way.. so no show-stoppers yet.

Playing it conservative, let's take the smaller of the two - CF power times time - as the safe-bet:

• 2.34672772 J PE + 7.71229 J KE = 10.05901772 J total GPE, from 9.80665 J of GMH

• 10.05901772 / 9.80665 = 2.5% OU

So IIRC this is worse efficiency than we got from the 1 kg weight. I had a suspicion that less weight might be more efficient - but apparently not.

So next, let's try the null hypothesis - raising it to 2 kg..

Don't assume it'll necessarily be better - the original 50/50 split of 1 kg gravitating vs 1 kg radially-translating could be the optimum.. we'll find out shortly..

[MrVibrating]

Eek - just remembered a minor detail: the 'kick-start' energy (otherwise the weight takes ages to topple) in that sim was 0.25 J.

Which means it's actually not OU at all (well, not really above noise anyway), at least for that initial cycle. IIRC, cycling up it does get OU - presumably the intended N3 break kicking in.

So in retrospect, the evidence for a successful N3 break is actually a lot stronger than that for a GPE break - which really boils down to that new F*d meter - the concept's solid, but the execution looks suspect..

Again, what i've done is to add the two actuators' forces together, and then multiplied by the radius, so with a range from 0 to 2. That much seems legit.

But then if we integrate force over displacement, that displacement's oscillating or 'reciprocating' back and forth, which means we'd be graphing concentric / overlapping loops, rather than a consistent contour - there'd be no 'curve' to take the area of.

So i've simply multiplied net F by d and integrated that over time. But this surely equates to dimensions of power, no?

Whatever, the integrals its putting out don't seem to match up to any other objective point of reference. The assumption of time-invariance of GPE should be applied, and thus the rational interpretation is that the putative GPE asymmetry is false.

If anyone can think of what's wrong with that meter or how ti fix it, we'll give it another shot.

In the meantime i'm gonna cut it, as i think it's leading us up a blind alley.

So with that, feck doing the 2kg 90° integrals - no point. GPE asymmetries are for dummies.

A simple test could be done by eliminating the actuators, and using a horizontal vMoI (ie. unaffected by gravity), torqued via a weighted ripcord:

• dropping the weight consistent height, measure a passive run - allowing the masses to be pulled outwards by CF

• then give the masses some initial KE at the outset, instead of using actuators; now we know exactly how much energy's been input

• let them slam into springs at the perimeter - we can measure their compression

• if what was happening was what i thought was happening, the system would be more OU the more initial radial KE was given - gravity, via CF force, would apply the same acceleration to the masses regardless of their initial speed / KE, hence cheating N3 / ½mV²

Might knock up just such a rig over the w/e, but i suspect the result will be negative.. see what i mean about getting sidetracked? Could spend two days on that, easy. Really, it's a back-burner, i think..

[MrVibrating]

So, priorities, priorities... What to proceed with first?

The F*d plot's on ice for now - the so-called 'GPE asymmetry' is gake and fay. For God's sake don't anyone waste a moment trying to build one.. there's just no rational reason it could be reading 6 J when the other two measures are saying 2, to within microjoules.

Ignoring the other two meters and just comparing the new meter against net energy was rash. My bad.

Those other two meters are showing unity for a single cycle. This is good. This is the intended outcome.

They're also showing kick-ass OU for multiple cycles. This is also good. And also the intended outcome.

So that pretty much determines what i'm gonna be doing for the rest of the weekend, then:

• cycling up, from 1 cycle to however-many-till-i-get-bored, and plotting the energy and momentum evolution

• if it really is free KE from an effective N3 break, we'll be able to see and isolate the causative momentum trades in high-def sims. Basically, the energy cost of momentum will not be squaring up with velocity.

If we can demonstrate exactly such a relationship, that's pretty much closure / certainty, right? Good enough to warrant build tests anyway.

[MrVibrating]

Can't be starting the w/e on a downer so here's a real (touch wood) energy gain to get us over that little wind-up:



Act. P*t = 48.4210353

CF P*t = 52.31357765


So this is the same unity cycle, now followed by a second, somehow resulting in a 4 J gain.

The answer to that 'somehow' is right there in the sim.

If everyone can get over trying to see it as a potential machine, and just view it as a measurement of a pure interaction, the meters should be fully comprehensive - power times time, derived from force times velocity, by definition makes no consideration of what powered the accelerations or how; only the energy required or produced by the masses themselves during that acceleration / deceleration matters. The 'actuators' are just a visual representation of the mass accelerations they facilitate, but the costs and benefits of those accelerations depend fundamentally only upon the accelerated masses themselves.

They're not real actuators, it's all just a bunch of numbers with a lick of paint. All that determines what energy the masses receive, or giveth, is their mass and accelerations..


...and that's what 'force times velocity times time' is measuring!!!

Please, anyone with doubts, let that point sink in.. measuring the actuators themselves as if they were real items is kinda redundant.. at this stage, anyway.

The sim's designed to be all but hermetically-sealed with regards to momentum and energy. Nothing should be able to get in or out without tying up neatly and consistently with everything else.

So the answer to where this energy is coming from is not going to be found in an error or oversight.

It's right there in the meters, in the data, and in the principles elaborated throughout:

• First off, the form of the 'gain' is a reduction in input energy, rather than an excess of KE over and above ½mV². So let's be crystal clear about that.

52 J is the right amount of KE for the system velocity. It doesn't (cannot) have 'too much' KE. The very notion of 'excess KE' is almost oxymoronic.

The 'gain' is that we've only paid 48 J.


So if this were error, how could it only affect the second cycle? How could the first be at unity? The sim doesn't 'know' one cycle from the next, the metrics of input and output energy haven't changed, my system isn't blue-screening every 5 mins, i mean, there's just no rational explanation for input energy decreasing, besides an effective N3 break - the whole friggin' raison d'etre of the config's inception - we were trying to pre-empt CF workload rising with velocity, remember? So mebe it's finally friggin' working, eh?

I'm right innit? You know i'm right.. there's just no better alternative than the central hypotheses - we're finally beating ½mV²!


And the proof of that must be right there in the data. Specifically, in the relationships between energy and momentum.

Check out the CF/CP work plot above - see that negative portion below the line? How pronounced it gets going into the second cycle? That's where we'll find our answers...

That negative work integral, and the corresponding chunk of momentum being introduced from gravity, should be the key to framing this gain in a precise, snug fit, completely eliminating any possible room for error..

Getting that Chistmas-y feeling yet..?

[MrVibrating]

..and hey look, we've got OU down to 2 cycles!

5 would've been great. 3, all the better. But i didn't previously think 2 was possible. (we're forgetting about yesterday's dodgy ½, like it never happened - we're cheap, but not that cheap).

2 cycs to OU i think we can just about make do with, if we must.. 4 mechs for OU in 180°, 8 for OU in 90°. Do-able, surely.

[Fletcher]

Hi Mr V .. spent a few hours today looking behind the scenes of your sim. Not trying to shade you. Just some thoughts on fundamentals.

The Actuators displacement is controlled by Length of Actuator formula (for those that can't see this) which is dependent on 2 relative points in the wheel. So independent of any capacity to do Work formula.

I dumped a lot of the MOI stuff and went back to basics of Green Driver GPE/KEt and Red Loads KE translational vs System KE to see how things lined up.

I note that the extra energy you apparently see comes from that Actuator length formula controlling the position of the Red Loads moving radially out and then in. And that it appears to close at first glance more quickly than it opens. This means it requires more Power to close quickly like that. And at the same time fighting against Cp's (tangential inertia). I think there must be a question mark about this.

ATEOTD since the Red Loads move equal distances radially their Net GPE remains zeroed. So we only have Driver GPE loss to convert to Driver and Loads Translational KE (KEt). Since the Driver always has some residual velocity (KE) then we don't have the full compliment of GPE loss to use doing Work on the Loads to bring them to CoR against Cp's.

This is where I think there is a problem in your Outputs. My instinct tell me that it takes more Work (or Power) to close the Loads than GPE Driver loss even if all was available to do the Work which it is not because the Driver still has velocity and KE.

It ain't easy to drive those radial loads back to CoR in an unassisted system, IMO.

ETA: Energy is said to be the capacity to do Work. It is the currency for Work. That's why I think you need to swap out the actuator length control formulas which currently dictate where the Loads are located at any time, to reflect real energy or f x d requirements an actual move would entail.

JMO's.