Mona,
first of all and once again I'm impressed by Al's post. It surely can not be better described.
Thus the following notes should be seen only as some addenda to Al's explanations.
In my opinion you are "right" when you think that when reaching the Curie Temperature of the base material there should no more exist any kind of (measurable) magnetism, but "wrong" on the other hand, when thinking there won't exist any kind of magnetism at all in DC arc-welding.
First of all one must know that the physical coherences of magnetic arc-blow are too intricate to be treated herein - at least from my personal point of view. To describe the exact sequences of the interaction between the electric current and the magnetic field being induced by it, it would be necessary to have as a minimum a diploma in electrical engineering since then it would be necessary to resolve the MAXWELL-Equations for stationary fields, which is far above my head. As you know I am even nobody who has ever pursued welding as a kind of career but rather as a kind of profession. Thus please understand that I am trying to describe the coherences only with the fundamentals of what a welder needs to know for "understanding" magnetic arc-blow and would like to avoid any mathematical formulae. Perhaps some others could explain it better...
Please allow, Mona, to recommend you as far as it hasn't happened yet to try to intensively find out the physical basics of a welding-arc and its behaviour in relation to different conditions. I guess, then you might have the chance for a better understanding when one says "...easy path..." or "...crossing the root opening...".
Very first of all I guess it would be reasonable to start with the magnetic field itself which is generated circumferentially around and acting perpendicular to the axis of current flow direction. As it is well-known, each electrical conductor is surrounded by a magnetic field. But most important is the fact, that the forces cohering to these fields try to displace the conductor's relative spatial position.
When we now imagine such a conductor - e.g. a welding lead - having a rotation-symmetric shape, the field lines around this conductor are shaped rotation-symmetric as well, please see also the Arc_Blow_1.jpeg.
Equivalent to this sketch or to the imagination of this rotation-symmetric conductor, can be seen our "tool" which consists of the stick-electrode and - yes - the arc itself, which is of course even nothing less than an electrical conductor, as well. And when everything works fine and symmetrically, the field lines, surrounding both the electrode as well as the arc-gap, are shaped symmetrical, too. This means again that all the differential values of the existing magnetic field line forces are symmetric in relation to their spatial direction. I personally interpret thus the statement: "...This is because the flux soon finds an easy path through the weld metal..." as an example for the symmetrical formation of magnetic field lines around both electrode and arc. The arc burns stable and no arc blow is to be observed...
Now let's try to imagine the opposite - i.e. arc-blow - and thus what happens when the symmetry of what we have spoken of is being disturbed by any peripheral means. Therefore let us basically suppose to use ferromagnetic material as steel - just as you are using it. What we assume is, that the "symmetry" is becoming an "asymmetry" in a special way, or in other words, the previously equalized values of field forces in space, must become different in relation to their spatial orientation. Allow me to let me use again our "picture" of the rotation-symmetric conductor but now "bending" it for resulting in such an "asymmetry" in the field force orientation, see also the Arc_Blow_2.jpeg. What we can see there, is that this bending leads to a kind of "compression" of the inner radius area field lines and a kind of "relaxation" of them at the outer radius area. The field lines having a higher concentration at the inner radius area, induce again an increased outward directed force. In case of real (Shielded Metal Arc-) welding we have now a quite comparable behaviour. Minimal discontinuities with respect to the symmetrical orientation of the field lines - how ever they should look like - lead to a deflection of the arc.
Let's resume...
We have now heard, that specific - even minimal - perturbations in the symmetry of the generated (invisible) field lines can lead to an observable asymmetry in the stability of the welding-arc. So far so good. This is what you could already more than once observe and this is the reason for that you have posted your question. And by the way, welding a butt-joint as you do, makes the issue more complicated compared with welding a fillet-joint, but more later.
What are now these mentioned different perturbations with respect to practical welding?
Let us begin with the most simple example:
· The Ground Clamp:
In DC welding you have - of course - two different polarities. Under "ordinary" circumstances you are using electrode negative and work-piece positive polarity. But it doesn't either matter if the polarity is changed, since the physical principle remains the same. Now let's imagine two electrical conductors, conducting the current in the same direction, as shown in Arc_Blow_3a_.jpeg. The magnetic field forces between both conductors "add" themselves and thus a reduction of their intensity in between the conductors can be observed (please remember the above mentioned "compression" and "relaxation"). For a better understanding see also the bottom section of Arc_Blow_3a.jpeg. Both conductors are attracted to each other since the field force density I between them is reduced compared to the outer areas.
Let's now come to the case where opposite Current Flow Directions are used, as to be seen in different welding polarities, and what is in general the case in arc-welding. The attached Arc_Blow_3b.jpeg shows both the schematic current flow and the adhering magnetic field lines. As closer both conductors and thus the field lines are approached to each other, the larger is the "compression" of those in between the conductors and the larger is the force trying to displace the conductors spatial position. What this means in practical application can be seen schematically in Arc-Blow_4.jpeg. When the negative polarized conductor (electrode + arc) are approached toward the positive polarized conductor (Ground Clamp) the magnetic field lines are compressed in between both conductors (compare Arc_Blow_2.jpeg) and outward directed (away from the Ground Clamp). As a result Arc Blow can be observed. See also the attached Arc_Blow_5.jpeg, which shows an arc burning between a carbon electrode and a steel sheet metal. This by the way, can be avoided by using two Ground Clamps and positioning them on both ends of the work-piece to be welded, as to be seen schematically in Arc_Blow_6_a.jpeg, and in practical behaviour in Arc_Blow_6_b.jpeg.
Time for a brief conclusion.
Up to now we have spoken about the main reason for Arc Blowing, i.e. asymmetric formed magnetic fields or -forces respectively, trying to displace the electrical conductor off its spatial position. Due to the arc is actually - as Al has described in a perfect manner - "nothing more" than an ionised column of gases this electrical conductor "arc" is appropriately sensitive against the magnetic field forces. Just as Al has also explained already.
But now let's have a look upon your specific problem. SMA-(Wet)-Welding of a butt-joint geometry and having excessive arc-blow phenomena, in particular within the groove.
Well, from my point of view, the most difficult in arc-blowing phenomena is that it is incalculable(!). Only by having gathered large practical experience - as we could impressively read in all the replies which were posted by Al, Gerald, Henry, Giovanni S. Crisi, bkoz, Bill, Martin,..., on your "Magnetic Permeability vs. Temperature" topic, one is able to "anticipate" what may happen when a DC-arc is burning in the adjacent area of large material accumulations etc.
This mentioned case is quite comparable with when you are talking about:
"...Easy path? crossing the root opening? I am sorry...I thought that when reaching Curie Point, there will be no magnetic field ...How come they will be concentrated in behind?"
To find a way off confusion let's start again with what we have already spoken about. We have heard that the stick-electrode + the arc are electrical conductors and thus being surrounded by circular magnetic fields or the adhering forces of these fields, respectively. What we also know is, that the - in particular Shielded Metal- - arc is sure an electrical conductor but due to it consists of a plasma, it is a weak one, sensitive against every little asymmetry in the formation of the field forces. Now you are right when you say, that the molten steel (which is the liquid weld-metal deposit) and all the work-piece areas elevated to temperatures above the CURIE-point are nonmagnetic. Thus - on the first view - the arc should not be deflected by being attracted of the nonmagnetic, since liquid, metal.
So what is the secret behind the butt-joint groove arc-blow..?
As far as I understand the subject (but you know I am always humble willing to learn), it is both a combination of magnetic field forces and the perturbations (asymmetries) being induced by the molten metal (where no magnetic field can be built up on), but where the anode spot is generated on since the electrical conductivity is higher compared with the surrounding air in front of the arc.
But let us firstly start with having a closer look upon what happens in regard of the generation of magnetic fields within the welding groove. We have heard that when having "only two" different polarities we have to expect "only two" different magnetic fields. These fields are with respect to their spatial orientation quite opposite to each other and thus by approaching anode (Ground Clamp) and cathode (Welding-Arc) - and thus compressing both fields - we can suppose to observe that the (weak conductor) arc is being deflected. What we also know is, that the arc is being deflected with respect to both directions - away from the Ground-Clamp and away from the work-piece's edge. The latter (in Germany we call it "Kantenwirkung") is explainable by the imagination of that due to the missing continuation of one work piece dimension (length) the field lines are compressed. By this we have an inwardly orientated increased magnetic force. Schematically this can be expressed by the Arc_Blow_7.jpeg.
Therefore let us name up to now two different increased fields of magnetic activity - or DC-arc-blow-phenomena - under presuming to prepare a bead on plate weld on a strip of steel sheet metal. The first one, as described under Arc_Blow_2.jpeg, is the conductor's "bending". And the second one is formed in the areas directly adjacent to the Ground Clamp and the "Edge"-area (in x-axis direction) of the sheet metal. Both phenomena take care for a compression, and thus a stringent asymmetry of the field lines.
And now it comes...
When you are now going to prepare a butt-joint-groove weld, a third magnetic field phenomenon must be added to the both already mentioned above. This is the field which is generated between the both sheets or plates respectively and which is a quite stringent one. Due to the gap (air) in between the both plates this field is additionally quite asymmetric formed and thus quite instable. So we can count the subsequent three different magnetic field phenomena acting on our weak electrical conductor (DC-arc):
Field "No. 1" - Start of the welding process on the edge:
Here the field lines are compressed by bending the "conductor" by ~ 90° (vertical stick-electrode - arc - horizontal plate). The "inner" or "concave" radius area having an outwards orientated field force.
Field "No. 2" - Magnetic influence of the work-piece' edge:
Compression of the field lines behind the arc. The base-material is - in contrast to the surrounding air - magnetized and the field lines are "absorbed" by the base-material.
Field "No. 3" - Magnetic influence of the groove bevels:
Compression of the field lines on both sides of the welding-gap. The magnetic forces are outwards orientated.
For a better understanding of the mentioned three fields please see also the Arc_Blow_8_a.jpeg. Here one can see, that the different fields interact while the seam is welded and strong field-asymmetries are taking place by that. The strongest force - when starting to weld on the work-piece' edge - is caused by "Field No. 2". Hereby the arc is quasi being pushed into the open groove. When continuing in welding and approaching the centre of the seam, Field No. 1 and 3 are getting more important and Field No. 2 disappears. From crossing the centre of the seam (Field No. 2 ~ 0) and advancing towards the Ground Clamp Field, No. 2 does increase again whereas Field No. 3 disappears. Now, the closer the electrode is approached to the Ground Clamp, Field No. 1 and 2 add themselves and by compressing the field lines in its front, the arc is deflected onto the already welded seam. For a better understanding of these descriptions please see also the attached sketch Arc_Blow_8_b.jpeg.
So far so good...
But what to do now to reduce arc-blow in groove butt-joint SMAW?
Well, I guess this is the reason for you to prepare the investigations and to write your PhD, isn't it? But as one humble hint from my side and only as an addendum to what the others have already recommended, I would try to intermittent tack weld the plate, just as to see in Arc_Blow_9.jpeg. This measure can reduce a stronger compression of field lines and thus reduce the asymmetry of the magnetic forces.
Might help as one item of several to solve your problems.
So far my humble contribution to your topic and as I already stated above, it might be of course that I am wrong by seeing the things as I have described them previously and I am always willing to be corrected.
I could imagine for instance, that our NDT colleagues who are experts in using often magnetic fields for non destructive testing, have some further detailed experience in the "magnetic-field-theories"..?
Best regards,
Stephan