WM2D

[Bill_Mothershead]

Thanks, Fletcher.

It had to be something like that.

Certainly will start using damper component,
and clean up obvious lack to precision,
silly weight values, etc.

Still trying to wrap my head around oscillation.
Will try to avoid it....but if it is what is causing
rotation...maybe it might be a good thing?

Hope to post an updated model (if it still works).

[ME]

What I do to manually position things accurately.

For example 8 mechanisms on a wheel all need a single connection:
(this method could be used with any collection of mechanism)

I use SI(degrees) as a Unit system.
'Glue' all moving things temporarily with a rigid joint to the wheel.

-Pin some small circle to your wheel at the appropriate position as a place holder.

[A]
-Copy/Paste that circle, and move it to overlap the original (don't pin it yet).
-Rotate the wheel to 180 degrees (that's why SI-degrees: for accurately input a rotation)
-Pin the copy to the wheel.
-Now there are two placeholdercircles pinned on the wheel 180 degrees apart.


-Now copy/paste the twp placeholders, and let them overlap the original
-Rotate the wheel to 90 degrees.
-Pin those two.

[C]
-Now copy/paste the four, and overlap.
-Rotate the wheel 45 degrees
-Pin those four.

See the pattern?
-Need 16? Copy 8, rotate 22.5
-Need 32? Copy 16, rotate 11.25
or
Need 12?
At step [C]
-Copy/paste the four, and overlap
-rotate 30 degrees, and pin
-Paste those four again (that info is still on the clipboard)
-Let them overlap those previous four
-rotate the wheel to 60 degrees (or -30)
-Pin those four again.

Rotate the wheel to 0 to get to the base situation, and remove the temporary rigid joints.

Now those circle-placeholders can be used to pin your ropes or springs, they'll act like snap-points.
As those placeholders are symmetrically distributed around the wheel, they'll not create any imbalance. Perhaps only add some weight.

I think it should be obvious that this trick should be done as early in design as possible to avoid a complicated mess of pinning and unpinning all kinds of loose mechanisms....

Hope it helps,

-- perhaps someone knows an easier way (besides scripting)? --

[Bill_Mothershead]

OK, that was too easy (and a little scary?)

Using dampered springs K=85 damper=1.0

weights (colored) = .1 Lb
supports (gray) = .001 Lb

all ropes same size
all springs same size

NO vibrations / oscillation

Still turns (maybe even better)

So what's wrong now ??!!

Is is the damper value?

[MrVibrating]

It's stable when when centered, as Fletcher advised (see the enclosed sim - would advise saving its preferences as your defaults).

The motion is due to an internal integration (rounding) error arising between the rebound and damping integrals, due to the slight offset of the axis.

If you keep everything else the same in that last sim you've posted, and only increase the accuracy of the integration error, the sim crashes with a buffer overrun (too many significant digits). But rounding that error over fewer digits - as your current accuracy settings do - results in an erroneous gain.

Generally, to scrutinise a system's behaviour you want to increase the resolution of the integrations in the accuracy settings. If that results in a crash then there's some kind of assembly error. Relying more on grid-snap and coordinate data during assembly will help keep things mathematically perfect...

[Bill_Mothershead]

It might help a little if you

.... SET GRAVITY to ON !

Here is an updated model
set to SI metric,
Grind snap set ON + central hub set to 0,0 ...
frames per second set real low to speed up animation

Apparently.....it still runs when centered.

[MrVibrating]

Oops you're right, i left gravity off... (duh)

All i did was range, copy and paste the model into the settings environment i'd been using for something else (w/o gravity), deleted the off-center pin joint axis, centered the model at the 0,0 origin and replaced the central pin joint.

Re-enabling gravity in the centered sim however just makes it keel...

If you follow Marchello's technique and rebuild it perfectly symmetrical, it will be balanced at any angle.


ETA: enclosed a few examples of similar false positives.

All of this touches on a question that's been much discussed here; of whether WM2D is even capable of simming a working design, as opposed to enforcing CoE in a top-down manner. My guess is that if it can show erroneous gains, it can probably reproduce real ones too...

[Bill_Mothershead]

Thank you for posting the 4 WM2D files.
I had not seen them before.


1 is an example of the program recovery from collisions.
As it is computing / stepping through to new positions,
it suddenly finds an object INSIDE another object.
It tries to gracefully recover by placing the object
safely outside...but there is the matter of recalculating
momentum / force and everything else. It does not
always do a very good job.

The work about is to attach a rope that prevents
the collision. See attached file.

2 is an example of a friction less chaos pendulum.
Set low speed air friction to some very tiny value
( .001 ) and it will run more realistic.

3 looks good...there are no settings that need changes.
Having larger weight values for the box will actually
make it run better....up to a point.

I am thinking that as the flywheel gains energy there
will be a speed that the box starts swinging and the
effect MIGHT not work so well.
In summary...seems to work but would be nice
if somebody would figure out how to scale it so
it turns faster.

4 is similar to 1....an example of what happens
when an object goes through a pulley attachment
point. Fixed it by making the small circle object
collide with the rectangle object.

=====================================

I am still working on why the "springs" models wont
work in your "world" settings. There is obviously
something different and I will learn a lot by
looking through everything. I have suspected all
along that the springs simulations were bogus
and someone would tell me to adjust a simple
menu item and it would all go away.

Your world settings are close to that.
Now, if I can just figure it all out.
Will post if I find out what it is.
(Could be you....or maybe not!?)

[Bill_Mothershead]

.
.
.
OK....found the problem !

All my models run with "High speed" air resistance ON !
(with some obscure default value of .3 )

If Air resistance is set to NONE then nothing runs.

If you download "centered.wm2d" above and
click "high speed" air resistance ON then it will run.

What this all means....I have no idea !

I would think that "resistance" would mean that
it would take energy away of something that is
moving....and tend to keep it from running.

This seem to be doing the reverse !

Help ! Anybody explain what is going on?

[WaltzCee]

But be warned, the underlying programing of WM2D is based upon conservation of energy. Thus it (in theory) will never show a positive output for a PM wheel. Some people dispute this and think it would show positive output for a real PM wheel

What do you think WM would do with the model of a working wheel, Jim?

My guess is you could make a 2 inch radius wheel with appropriately sized masses and it would take forever to calculate a single frame, then as it tried to resolve the calculations it would begin to break constraints and stretch ropes forcing the model into a shape it calculated it should be in. All this in less than 1/1000 of a second.

Finally the model would just spin as everything designed to move was frozen to the substrate. Just a guess.

What do you think?

[WaltzCee]

I also think parts of your model would begin to levitate.

If any supposes they have a viable model, I'd recommend they only use wm2d on a machine off the net. Also, wm2d can't be trusted to do some very basic calculations and even the ones it does can't keep pace with the speed of reality.

Nothing can replace the analog computer of a build. Now what will your computer prove? I don't think it will disprove the conservative nature of gravity. Then what? I speculate it will prove creation; that a single force or entity, in this case gravity, can create.

Also, nothing is more powerful than an idea whose time has come. I think it's time.

Freight's moving past your burning shack right now. The bitch is you set fire to the shack then went to sleep.

[graham]

Hello all,
Is the attached WM2d result correct?

I have an inclined surface and am representing atmospheric pressure applied to the entire surface by the force arrow to its midpoint.

I was expecting the 42lbs of pressure to be halved in the X & Y directions giving 21lbs equally.
As you can see WM2d is giving me 30lbs instead.
I don't always trust results with the program due to my own lack of experience with it.

Your comments please
Thanks
Graham

[ME]

Sqrt(30^2+30^2)=Sqrt(1800)=42.426

[rlortie]

Grahm,

I am not the least bit acquainted with the use of WM2D, I do however have a little knowledge in rigging and "sling loads" which is what your simulation suggests.

See: https://www.youtube.com/watch?v=4ESdgrCJrek and also the related videos found on this link.

The angle of your depicted stress may be correct if the program is reading the angled stress on the atmospheric load as it would be expected to do if based on the load being lifted.

Try the formulas found in the above link or those related to see if you are getting the correct answer from your program. As the angle is reduced the stress increases!

Ralph

[ME]

Perhaps my answer was a bit too short:

I was expecting the 42lbs of pressure to be halved in the X & Y directions giving 21lbs equally.

a Polar vector : 42.426, at angle 225 deg.
Can be split into, or is the combination of, two cartesian vectors:
Fx= 42.426*Cos(225°), Fy= 42.426*Sin(225°)

I think this was the 'oops' :-)
Sin(225)= - Sqrt(0.5) = -0.7071


---
Add: Which also applies to the slingload-video-formula:
For one 'sling' the Fy=4000/2=2000, and the 'side stress' Fx=1500, The total stress is sqrt(2000^2+1500^2)=2500, hanging at an angle of 36.87°, and attached at 53.13°.
Sidenote for if one wonders: That video-formula is correct (not some rule of thumb)

[graham]

Thanks for the feedback guys,
One more question. What is the difference between a cartesian and a polar vector

Graham