[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [femm] transient analysis

William,

This is in reply to your last but one post I'll have a look at the web pages asap.

I suppose it depends on what you are trying to do. If you are trying to produce a fairly gentle acceleration up to not too high speeds then the machine is effectively a linear induction motor and iron would certainly make the thing work better. These things have been built ( a colleague of mine has built several as student projects) and there are journal papers describing them. However, if you are trying to produce very high accelerations (like a gun) then the presence of iron is questionable. That said I have used steel projectiles and found that they work at high energies when the electrical accelerating forces are significantly bigger than the ferromagnetic forces. In this situation steel is just another conducting medium.

The launchers than I have built where all intended to move a solid conducting mass to the highest speed possible in a single stage of acceleration. One of the most difficult problems I had to contend with was making the current change fast enough; often the projectile is long gone before the current hits its peak. In the fastest launchers I was getting accelerations such that the projectile is transonic within the first millimetre of travel - I wouldn't have been able to do that with a tubular iron cored device.

I tried steel projectiles and they worked ok at high energies where the accelerating forces

Tubular launchers are by their nature inefficient, mostly because they expend a lot of their energy trying the crush the projectile rather than push it. If you do the sums I think you will find that the forces radially inward can be substantially bigger than the axial accelerating forces.

At 11:01 16/06/2003 +0930, you wrote:
Stoned koala bears drooled eucalyptus spit in awe as Keith Gregory declaimed:

>If I was building a capacitor discharge induction launcher, i.e. one that
>produces current in the projectile by induction (and I have built a few) I
>wouldn't have iron anywhere near it. The performance of an induction

<snip>

yes, I understand this. The armature current is also reduced by the increased
inductance of the armature. However, I think the same amount of energy is
transferred to the armature - the current builds up slowly to a lower level,
but B.H is higher due to the iron inside the armature. Net effect is a much
slower discharge, with force spread over a longer time (the L-R decay of
current in the armature takes longer). I am yet to verify this (see
below) but should have presentable results Real Soon Now (tm).

I suppose it depends on what you are trying to do. If you are trying to produce a fairly gentle acceleration up to not too high speeds then the machine is effectively a linear induction motor and iron would certainly make the thing work better. These things have been built ( a colleague of mine has built several as student projects) and there are journal papers describing them. However, if you are trying to produce very high accelerations (like a gun) then the presence of iron is questionable. That said I have used steel projectiles and found that they work at high energies when the electrical accelerating forces are significantly bigger than the ferromagnetic forces. In this situation steel is just another conducting medium; its conductivity is not great though and we have usually backed steel projectiles with aluminium or copper.


I spent much of yesterday running FEMM sims on different geometries, mostly
varying the amount of iron around the place: none, a core inside the coil, a
long central core that the armature travels down, fully shielded (coaxial core
& tube), etc. The results were nonsense (mA induced in the armature), but now
I realise I forgot to divide delta-flux by the time delta when calculating
armature EMF. So everything is out by 6 orders of magnitude; I'll fix that
tonight and see if I get something meaningful.

>I would suggest that you just use FEMM to work out the system inductances
>and their variation with projectile position and then use those
>characteristics in a MatLab simulation to work out the forces. The

this is exactly what I've been doing. see below for something more explicit.

>simulation would be able to make a reasonable estimate of the projectile
>current too. It might be possible to determine from AC FEMM simulations an
>effective "shorted-turn" representation of the iron so that eddy-current
>effects could be included in a MatLab simulation.

I've been ignoring eddy currents for now (the shielding will be laminated in
the appropriate direction); I realise that the numbers won't be exactly right
but they should be OK for comparison purposes.

>I have found in the past that static AC simulations - where the projectile
>is assumed not to move - made at the approximate ringing frequency of the

how do you find its ringing frequency? Do you mean the LRC time constant of
the power supply + coil or some magnetic time constant?

Yep. If you can get an estimate of the system inductance an resistance and you know the capacitance then just use the standard equation. If the coil inductance is low then the power supply parameters (things like capacitor equivalent series inductance and resistance and the cable parameters are vastly important).

What I've done is this (please tell me if I appear to misunderstand something;
its 5 years since my eleceng magnetics classes and I haven't done any since
then):

blockintegral(A.J) in the power coil with varying armature position to get
power coil inductance.

blockintegral(Bz) over the armature and the region of iron (or air) inside it
with zero armature current and a non-zero test coil current to find the flux
linkage into the armature with varying armature position.

measured lorentz force on the armature with test current in it and the coil.

measured blockintegral(A.J) in the armature with a test current (no coil
current) to get armature inductance varying with armature position. uniform
J was probably a bad idea, but I'll re-run this part with a specified I so
most of it will flow nearer the centre of the armature.

measured resistive power loss in the armature with a test current to get
armature resistance.


In matlab, the coil current is simulated as an LRC system with zero initial
current and a specified capacitor size & voltage. R is chosen for very slight
underdampedness. Flux through the armature is determined and its rate of
change (when I remember to divide by time) gives me EMF in the armature. The
armature is simulated as a series L (position-varying), R and voltage source
(EMF).

I'm not sure that I would calculate any forces in FEMM other than for comparison. I would do the force calculation in a separate simulation, in this case the MatLab one. To calculate the forces you need to know the derivatives of the self and mutual inductances with respect to armature position. These can be got from FEMM pretty easily. Also, I'm not sue why you would want to calculate the armature emf directly, it should come out of the MatLab calculations. We clearly do this stuff differently, I could send you some descriptive stuff if it would be of any interest, but it might be better off list.

Keith.