What is the permability of mild steel
Lionel Titchener
I can get permability of magnet iron from the suppliers, but what is the
permability of various grades of mild steel?
Thanks
Lionel Titchener
Paul Victor Birke
>
> Where can I find the permability of metals other than soft iron.
>
> I can get permability of magnet iron from the suppliers, but what is the
> permability of various grades of mild steel?
>
Mild steel has a range of about 100 to 500. This range is for low
carbon types if I remember. But I am not sure exactly the values for
the grades.
Paul Birke EE
D.H. Kelly
L wrote:
>Although Mr Birke appears to be proud of his Electrical Engineer's
>status he omits a number of facts on magnetic materials.
>Almost all the physicists I've worked with and a number of electrical
>engineer's recognise the fact that no strongly magnetic material has a
>linear response to an applied field. Lionel : There are things called
>hysteresis curves supplied by most manufacturer's of magnetic materials,
>permanent or otherwise. All ferro- and ferri- magnetic materials
---------------------------snip------------------
You are right- permeability, per se, is not a useful working quantity. I
also agree with your following statements.
However, the use of the B-H curve for magnetic material is an engineering
approach and has been used by engineers from way back when-say in
Steinmetz's days. I doubt whether any engineering school which deals with
magnetic devices, deals with other than the "B-H curve" (AC or DC) except
for the case of a dominant air gap where the iron is negligable. The
hysteresis loop itself is of interest for permanent magnets and some ferrite
applications as well as in the design of the magnetic material (how much
silicon, what other additives, grain orientation etc.). but generally, core
loss/unit volume vs H data is more pertinent to the design of transformers,
etc..
--
Don Kelly
dke...@nanaimo.ark.combull
remove the "bull " to reply
L.
status he omits a number of facts on magnetic materials.
Almost all the physicists I've worked with and a number of electrical
engineer's recognise the fact that no strongly magnetic material has a
linear response to an applied field. Lionel : There are things called
hysteresis curves supplied by most manufacturer's of magnetic materials,
permanent or otherwise. All ferro- and ferri- magnetic materials
exhibit some form of hysteresis. If you know the field strength that
your material will be working under you can read the permeability from
that. Constant permeability does not exist for such materials. Your
manufacturer may be quoting initial permeability and this may not be
relevant to your app.
If you are applying an alternating field then you also have to allow for
a wider and squarer hysteresis curve than that typically published.
The book by C. Heck, Magnetic Materials and their Applications, (1974)
will give you all the info on the different meanings of the different
permeabilities used. This should be available in a good university
library - hopefully the reference section if you are not a member of
such an institution. Try to get it on an inter-library loan if your
place doesn't have it. There is also the Chemical and Physical Handbook
which quotes some figures of constant permeability for diamagnetic &
paramagnetic materials as well as the more usual iron compounds. I
can't remember if it states what permeability they are quoting.
As an example initial permeability is the permeability of a material
when first magnetised. This is no use for a transformer except at start
up - after which the material characteristics traverse the hysteresis
loop. Also no use in a magnetic circuit with a permanent magnet snce
any soft materials will never be in an unmagnetised state.
Try also Parker's Advances in Permanent Magnetism. Nowhere near the
value of Heck but worth a look at a fascinating subject.
Best of luck and let me know if you find the books of any use.
L
--
Skinticus Maximus :-(
Paul Victor Birke
As you have indicated, magnetic materials, to be used properly, exist
in, one could say, a nonlinear environment.
It takes some effort to solve a circuit for the correct results. In
fact, I have done this recently for a DC application and had to resort
to Nonlinear Programming because the solution was near the knee of the
B/H curve.
It was not my intent to oversimplify what is a rather complicated
subject by quoting very rough relative permeability ranges. In fact, we
have to know what Lionel is really try to do with the material. AC or
DC just for starters.
Anyways thanks for the very detailed response.
Paul
Paul Victor Birke
>
> Where can I find the permability of metals other than soft iron.
>
> I can get permability of magnet iron from the suppliers, but what is the
> permability of various grades of mild steel?
******************************************************************
An older but very good data text is:==>
Electrical Material Handbook
published by:==>
Allegheny Ludlum Steel Corporation
Pittsburgh, PA
1961
It is chock full of the data that you are looking for.
all the best,
Paul
Paul Victor Birke
s.
> However, the use of the B-H curve for magnetic material is an engineering
> approach and has been used by engineers from way back when-say in
> Steinmetz's days. I doubt whether any engineering school which deals with
> magnetic devices, deals with other than the "B-H curve" (AC or DC) except
> for the case of a dominant air gap where the iron is negligable.
Dear Don
True. But in the past, at least when I was taught at UoToronto 1963-67,
there was only the piecewise linear approximation to the B/H curve.
Which misses describing the local curving at the knee point which tends
to be the solution range in many cases. For a better and more
computationally useful representation, I give a simple approximation to
a B/H curve I have fitted as:==>
Z(J) = X(1) * (1 - EXP(-((XLOG(J) - X(2)) / (X(3) - X(2))) ^ X(4)))
You can see for the curve to be somewhat properly specified one needs 4
constants to be determined as a Nonlinear Regression program. Are the
schools now teaching this or similar? What they should be educating the
students to fit exponential or polynomial approximations to the curve so
that one can actually use it properly in what even calculations you
love?
all the best,
Paul Birke, EE
L.
higher order polynomial regression to fit a BH curve altho it cannot be
difficult to apply the theory. I'm aware of the Frohlich-Kennelly
approximation
B= (H+Hc)/(a+b*(H+Hc))
where b is the inverse of the stauration of the material and a is
defined in terms of the coercive force, the residual induction and the
saturation. This expression works for permanent magnets but I found the
--
take the pxxx out of it to reply
still skint ...
D.H. Kelly
True. But in the past, at least when I was taught at UoToronto 1963-67,
there was only the piecewise linear approximation to the B/H curve.
Which misses describing the local curving at the knee point which tends
to be the solution range in many cases. For a better and more
computationally useful representation, I give a simple approximation to
a B/H curve I have fitted as:==>
You did this for a computer model of a B-H curve?- i.e. used published data
and then fitted a curve. I agree that your model is better than a piecewise
linear approx. but now the computer capability available makes this easy.
However, remember that the nice B-H model that you came up with really may
not mean more accurate modelling. With magnetic materials, the normal
variability between samples, the effects of mechanical impacts, etc, mean
that one is lucky to be within 5%. of the actual characteristic. Did you
ever try to get the magnetization curve of a generator (AC or DC) and "go
back" to pick up an in-between point?
Did you study under Gordon Slemon?
Paul Victor Birke
>
> Paul Birke wrote;
> True. But in the past, at least when I was taught at UoToronto 1963-67,
> there was only the piecewise linear approximation to the B/H curve.
> Which misses describing the local curving at the knee point which tends
> to be the solution range in many cases. For a better and more
> computationally useful representation, I give a simple approximation to
> a B/H curve I have fitted as:==>
> -----------------------------snip---------------------
> You did this for a computer model of a B-H curve?- i.e. used published data
> and then fitted a curve. I agree that your model is better than a piecewise
> linear approx. but now the computer capability available makes this easy.
> However, remember that the nice B-H model that you came up with really may
> not mean more accurate modelling. With magnetic materials, the normal
> variability between samples, the effects of mechanical impacts, etc, mean
> that one is lucky to be within 5%. of the actual characteristic.
Dear Don
Yes well this is true. But the nonlinear solution is rather interested
in the rate of change at the knee and I thing one can capture this will
some accuracy. But the absolute value, you are right, as these other
factors practically limit the accuracy.
Did you
> ever try to get the magnetization curve of a generator (AC or DC) and "go
> back" to pick up an in-between point?
Can remember! Maybe in the lab we plotted by this trick. We certainly
used this idea at Westinghouse to get a value at the knee by the
intersection of the two lines but in a bit! Hard to remember.
> Did you study under Gordon Slemon?
Yes, a most wonderful teacher. UoT was rather "heavy" on magnetics and
this was strongly due to Prof. Slemon.
A very gracious person as I had to deal with him as a class rep. He was
and is a first class teacher and person. A privilege to be a student
then under such a person.
Paul
Paul Victor Birke
Paul Victor Birke
>
> Paul Birke wrote;
> True. But in the past, at least when I was taught at UoToronto 1963-67,
> there was only the piecewise linear approximation to the B/H curve.
> Which misses describing the local curving at the knee point which tends
> to be the solution range in many cases. For a better and more
> computationally useful representation, I give a simple approximation to
> a B/H curve I have fitted as:==>
> -----------------------------snip---------------------
> You did this for a computer model of a B-H curve?- i.e. used published data
> and then fitted a curve. I agree that your model is better than a piecewise
> linear approx. but now the computer capability available makes this easy.
> However, remember that the nice B-H model that you came up with really may
> not mean more accurate modelling. With magnetic materials, the normal
> variability between samples, the effects of mechanical impacts, etc, mean
> that one is lucky to be within 5%. of the actual characteristic.
***********************************************************************
Dear Don
Yes, well this is true! But the nonlinear solution is rather interested
in the _rate of change_ at the knee and I think one can capture this
with
some accuracy. But the absolute value, you are right, as these other
factors practically limit the accuracy.
Did you
> ever try to get the magnetization curve of a generator (AC or DC) and "go
> back" to pick up an in-between point?
Can't remember! Maybe in the lab we plotted by this trick.
We certainly modelled the knee regions at Westinghouse by the
intersection of the two lines and then plotting a asymototicall fitting
curve in a bit! Hard to remember.
> Did you study under Gordon Slemon?
Yes, a most wonderful teacher. UoT was rather "heavy" on magnetics and
this was strongly due to Prof. Slemon.
A very gracious person as I had to deal with him as a class rep. He was
and is a _first class teacher and person_!! A privilege to be a student
D.H. Kelly
But the nonlinear solution is rather interested
some accuracy.
I agree in that a piecewise solution (particularly a 2 piece one) isn't
much good in region around the knee. However, a piecewise solution using a
non-linear model in the region of the knee and one or more linear sections
on either side, works well - The advantage is that it is often easier to fit
a cubic or quadratic model in the knee region than it is to produce a
non-linear fit for the whole of the curve.
Whatever works for the application in mind.
---------------------------------
Did you
> ever try to get the magnetization curve of a generator (AC or DC) and "go
> back" to pick up an in-between point?
Here I was referring to an old student error. Going back to pick up a point
puts one through a minor hysteresis loop and the result is a "jog" in the
magnetization curve where the prior and post data don't quite meet.
Agreed as to your comments re Gordon Slemon.
L.
<snip>
> ---------------------------------
> Did you
> > ever try to get the magnetization curve of a generator (AC or DC) and "go
> > back" to pick up an in-between point?
> ------------------------------
> Here I was referring to an old student error. Going back to pick up a point
> puts one through a minor hysteresis loop and the result is a "jog" in the
> magnetization curve where the prior and post data don't quite meet.
>
Perhaps Paul should have got his head down to the books and his hands
onto the practical instead of playing with student politics from time to
time :->
Going back to pick up a point will only take u on a minor loop *if* you
don't run around the full cycle i.e. take the applied field all the way
through again. In *theory* at least one should be able to pick up the
major loop again and again. In practice, of course, accept that this is
probably not going to happen.
I have had the good fortune to spend the last 10 years of my life
involved in research into many aspects of magnetism and electrodynamics
including the computer modelling of hysteresis loops and the limitations
of some of the modern modelling techniques. Some people claim that the
knee of a material can be modelled reasonably well by decreasing the
intervals over which the p.l. approximation is taken. I prefer splines
where necessary or algebraic representations as developed by Frohlich
some time ago.
Recommend anyone interested to look at Gupta's papers. Not going to
give any further reference (go look for it!! ) and will only say that a
reasonable algebraic form of the Frohlich type is examined therein.
Gupta calls this the Frohlich-Kennelly form since he uses that form of
the SI system. It needs *some* manipulation to get a good approximation
to the 2nd quadrant of a permanent magnet's loop eg Alnico but when it
gives a good approximation it does so exceedingly well. I think I
managed to get the thing to also represent the demagnetisation of the
more modern rare-earth materials which have no knee at room temp at all.
L.
--
take the xx out of it to reply
still skint ...
D.H. Kelly
Going back to pick up a point will only take u on a minor loop *if* you
don't run around the full cycle i.e. take the applied field all the way
through again. In *theory* at least one should be able to pick up the
major loop again and again. In practice, of course, accept that this is
probably not going to happen.
I agree- it does take one on a minor loop but, if you do this, you will find
that the expected smooth curve, has an nice little jog in it. Annoying.
I agree (as does Paul) that splines or non-linear models are preferrable to
piecewise linear approximations - particularly for optimization problems-
avoids a lot of solution instabilities. It is not a matter of accuracy of
the representation but the jumps when the solution process bounces from one
side to the other of an inflection point (right term?).
My main comment in this thread was that B-H curves have limited accuracy-
too many factors can change the curve for a given core- test, drop it, test
again - different result - within 5% is bang on. Probably, for computer
models, a smooth curve is better than a more accurate model with
discontinuities at "piece" boundaries.
my 2 bits worth