Is the existing machine the same type or is it a completely different machine? Are the machines calibrated for there output? Just curious why it would work on one machine and not the other if they are similar?
I can't help but to "push" the pulse capability on you if you have it as on option, the beads will come out very uniform in appearance. I found that when I switched alot of my welds to pulse there was less distortion in a lot of my sheet metal components. Here are a few pics of titanium and inconels welded with pulse (and wire).
Regards
JD369

Hello aevald and Stephan;

This perplexed me as well the first time I encountered it. Well, the situation was somewhat different. I was working with aluminum and qualifying the procedure using a heavy fixture plate that clamped the two pieces of aluminum plate quite tightly.

After welding the two plates the thickness in the HAZ was considerably thicker than the base metal. It didn't make sense until I noted how tight the welder was clamping the test plates in the fixture. I asked him to back off on the clamping device and the problem disappeared. What was happening was that the clamps prevented the plate from slipping as they heated up during welding, i.e., they expanded in the direction of the groove and thickened because the tensile strength of the hot aluminum is lower than it is at room temperature. When the clamps were loosened, the plates could expand and contract freely during welding (and cooling) and there was no thickening adjacent to the weld.

I saw the same thing happening when I first got involved in orbital welding. I expected the joint to thin in the area of the weld because no filler metal was added to the weld puddle. However, there was some face and root reinforcement once the weld was completed. Again there was a clamping device involved that prevented the heated tube from slipping through the clamps during welding. At the higher temperatures attained during the welding the tensile strength of the weld and HAZ is less than at room temperature, thus any expansion in the axial direction is accommodated by the weld and HAZ (more so in the weld that is in a "mushy state"). Upon cooling, the weld and HAZ regain their tensile strength quickly and the contraction forces of the cooling metal pulls the tube through the clamps.

Best regards - Al

Dear Allan!

Now, as already "threatened" a "little" time ago, here am I again to write down a few of my own thoughts on Boon's post. It's the first time since long, that I have found time to return to the forum and I really don't want to know how many interesting posts I certainly have missed over this period.

However, it's a very good feeling to be back..!

This sentence here was originally not intended to be written, but since I have exceeded the maximum text length size (shame on me!), I am going to divide the original text corpus by "2" and will try to create two single replies of actually one entire one. I beg your understanding...

With the attachments please note that I will attach all the pictures of which I find they might be helpful to elucidate the matter within the # 2 of my response. I.e. all what you read in a oherence with "jpeg's" you will find in the second part of the post. I truly hope that this will work!

Please allow basically to express that the subsequent considerations are my humble try to explain myself what kind of coherences are responsible for the sometimes different behavior the material shows in a direct relation to what Boon has stated to be his application in his initial post.

I am honest, I personally guess that this mentioned post is one of the very most interesting ones I have read on the forum.

Since I have considered long about all the different items, perhaps being in charge, I found out that it was as usual. The more one considers the more he is going to find out and the less he appears to know finally.

However, what I have done - of course under simplifying everything as best as I could due to the restriction of space in the forum - is, that I have tried to bring both what fascinates me personally (Arc- and Solid State Physics) together in a way that it could make sense finally for treating the particular topic - Autogenous GTA Welding of AISI 304.

First off however, I must agree with you and Dave! Al's explanation is again and as usually one outstanding good one.

When I have read what Boon has initially posted, my very first reaction was to say "Hmm, this should only be hardly possible!" But nonetheless, I am careful by knowing myself. When I am asking a question and somebody says "No, this is not possible!" something awakes in me and driving me to find out the pros and cons making something either "possible" or "impossible".

I must thank you for according me the time for this late, or better very late response. You know I was quite busy and thus the time for visiting the forums was extremely restricted, unfortunately!

Meanwhile however I had time enough to pay one of the scrap boxes of our company's welding training center a visit, trying to find some pieces of ~ 304 sheet metals GTA- a/o eventually Plasma-welded to see how the welding seam appearances were. I had luck by having found some "pieces of gold" :-).

Well, here we go..!

First question I asked myself after having read what Boon has posted was: "How can it be as far as it can even be?".

Next thought was going towards separating all the "things" being involved into the procedure making a weld seam even a weld seam. There we can state:

1.  (GT) Arc
2.  Base Material
3.  Filler Material

Since we know that no filler should be used in the case Boon has mentioned we can cancel item 3 again. This should simplify the entire affair appropriately (?).

So in fact what do we have? Actually merely two different things and - and I guess this is one of the most intricate but as well interesting process in physics - the interaction between the plasma (energy) and the solid state matter.

I have tried to continue my considerations on the basis of analyzing (surely quite basically):

1.  How is it possible to handle a Gast Tungsten Shielded Arc and its influence on the solid base material in a simple way?
2.  What are the results of the Arc's energy in particular terms of the used 304 base material

For (very partially) replying the question 1 let us have a short look upon the structure of a TIG Arc maintained under Argon as the used shielding gas. It has - normally - a bell shaped form with a brighter (i.e. "hotter") core and a lighter (i.e. "colder") zone around the core, see also Picture 1. Both criteria are specific and the results of indeed interesting but nonetheless very complex plasma-physically interactions. As an extract of both one can say that the shielding gas properties are responsible for the later shape and consistency of the arc plasma. Due to the specific properties in terms of thermal conductivity Argon as the shielding gas has a relatively low ability of conducting the heat through the arc volume which results again in a "hot" inner core and a "cold" outer area. What does "hot" and "cold" mean when we are talking of temperatures high enough or far beyond those ones able to melt metals. I hope you may agree when I say that this topic to treat herein could be a hard to realize undertaking, what's the reason for that I would like to avoid it furthermore. But - at least from my personal viewpoint - it is even unnecessary. Because what is much more important for having a possibility to deal with the application Boon has stated in his post, is the fact, that we should treat the arc plasma at our particular purpose, and as mentioned extremely simplified, as consisting of different electrically neutral and electrically charged particles being roughly mixed by pushing a current through a gap between the work and the electrode and therefore using the voltage. The actual major carriers of the electrical charge within the arc plasma are the electrons. The higher the concentration of electrons (i.e. the higher the current density) the higher again the temperature ("hot" core) and - and this is crucial - the higher the  p r e s s u r e  the arc causes upon the base material, or better, the weld pool.

Pressure..?
Of course. The arc causes a pressure upon the weld pool and all of you know that of course, since everyone of you has observed the fascinating spectacle a Gas Shielded Tungsten Arc is carrying out on the weld pool's surface when you have risen the current. The mentioned pressure is the result of different strong forces (electromagnetic, surface tension, buoyancy and plasma jet) Mainly responsible however is the electromagnetic or LORENTZ force respectively, which is proven to generate an inward flow of the molten metal within the weld pool volume. By the way, when you have a closer look upon Argon_TIG.jpeg, you can slightly recognize that the arc is causing an indentation into the base material. Why am I that strongly dealing with the pressure underneath the arc? Very simple... Since the pressure does arrange that the material is being distributed and furthermore displaced underneath the arc.

O.K. let's see what we do have by now?

A Gas Tungsten Shielded Arc established and maintained in an Argon atmosphere, does have a particular shape and a particular physical behavior.

Hmmm, but on what depends the physical behavior of the special shape of the arc? We have heard that the arc consistency, its temperature profile, its pressure gradient is amongst others - to be neglected herein - as well depending on the current density. What is the current density again depending on? Well, again amongst others it is depending on the tungsten electrode geometry, the electrode diameter, the electrode material's composition and other peripheral parameters (e.g. electrode to work distance, torch positioning angle,...).

How to make it possible to describe the pressure the TIG Arc is performing upon the weld pool? (Excuse me for not coming to the point but I hope that I will find a way to it finally.) Good, this is best possible by using the way of visualizing what's taking place while an arc is burning between the electrode (cathode) and the work (anode). But nonetheless I would like to make a short sidestep towards to describe how tremendously intricate it actually is what we are here talking about. So intricate that the theorists - as well nowadays - are not 100% sure about the complex interactions between the different physical phenomena existing within an arc plasma.  

Let the volume of a molten pool (V) be a function of - as we have heard - a large number of different parameters e.g. temperature (T); material density (rho); arc pressure (p); welding speed (v) and let us consider further that these parameters are - in a wide range - temperature depending again what complicates the prediction of the molten pool behavior additionally. I permit myself to decide that the latter should be neglected for a strong simplification to being able to reduce all the intricate combinations to a more simplified form where we can say, the conditions coming from the plasma side (arc) + the conditions coming from the solid state side (base material) yield a very specific molten pool condition which yields a  molten pool volume again. Now let us presume that - as already shortly mentioned - the main parameter affecting the height of the arc pressure should be the height of the welding current. Different - very interesting - investigations were conducted in the past with respect to find out more about the coherences of molten pool depression forces induced by the arc. Most of those surveys could confirm that the even stated (height of welding current) is the main parameter to be considered for calculating the arc pressure or its influence on the surface depression of the molten pool, respectively.

Why have I mentioned all this by now?

I hope you may agree when I say that - in a great extent - the pressure of the arc is responsible for the amount of the depth of fusion or - in case of welding autogenously - the depth of penetration. As I have said, please allow to neglect the various interactions - and thus variables - caused by electromagnetic-, surface tension- and other forces, taking place within the molten weld pool and being responsible for the convective flow being induced by these forces. What however is - from my point of view - important, in particular for the case of autogenous GTA-Welding and achieving both root- and surface reinforcement, is the fact of the composition of the base material. But hereunto I would like to come to a bit later on. Alright, let me coming to the visualization of how it might look like when the (Gas Tungsten) arc is autogenously deforming the weld pool. Please see the Surface_Depression_TIG.jpeg, for making it imaginable how a Gas Tungsten Arc is able to depress the weld pool surface and thus, in autogenous welding, making sure the depth of penetration in relation to the workpiece' wall thickness. What's interesting is the detail, that when GTA Welding autogenously the depth of penetration does not increase infinitely by increasing the current. This phenomenon is as well observable when increasing the welding time at constant current when welding with a stationary arc. Proportional to each other do increase both weld width and weld pool area. Many Investigations have been conducted in terms of the influence of surface depression and convection on the arc weld pool geometry. Here one has e.g. found out that there is a kind of "transition current" existing, where relative "shallow penetration" depth is increased to be changed to higher values by so-called "deep surface depression". Very interesting but I won't like to further treat this matter herein since the transition current ranges are far above those values Boon has stated in his post and which laid at 110... 130 Ampere. However, what we can conclude by here is that the surface depression of the weld pool is being influenced mainly by the height of current and other variables.

Since I would like to treat how the weld seam appearance is basically being formed I would like to coming now to the physical coherences in terms of how the weld bead geometry does react under the autogenous Gas Tungsten Arc. When having a look upon a weld pool being generated by a stationary Gas Tungsten Arc and when using relatively low current values (as which I interpret the values of Boon) - which is important since only here one can use the simplification of using a so-called "Gaussian" heat distribution (particular mathematical approximation method, see also: http://en.wikipedia.org/wiki/Normal_distribution) by considering the heat maximum in the center of the arc - the surface depression is as well relatively low. Let me now assume that - just as we have heard - increasing the welding time of a stationary working Gas Tungsten arc at constant current does not increase the depth of penetration but does increase the area of the weld pool on the workpiece at least as long there has no balance between heat addition and heat dissipation been generated. By the way, in my eyes the depth of surface depression should thus being reduced due to the greater pool volume. However, let us now presume to not use a stationary Gas Tungsten arc but to move our heat source, even to weld e.g. a square groove butt joint, as mentioned by Boon.

What does now happen to the surface depression?

Well, before dealing with this interesting question we have to go back to an increase of welding current and exceeding the limit of where the "transition current" causes a deep surface depression. Here one has to not only to leave the range of relative shallow depth of penetration  but has also to leave the pattern of heat distribution ("Gaussian")and thus to achieve a changed current flow. Why is this detail that important? Even since the flow of current stands in a close relationship to the molten metal flow pattern. LIN et al have investigated this and have found out that by an increased surface depression there is generated a displacement of the field force density having similar potential (equipotential lines) and hereby it is assumed a "maldistribution" of current - or in other words an imbalanced heat density at the workpiece (anode) surface. This is by forming a deep crater after the "transition current value" is exceeded (deep surface depression exists) and the main share of both heat and current is being received by the crater side walls, but only a minimum is received by the bottom of the crater itself, for instance to increase the depth of penetration additionally.

Alright, we have now spoken of that there can be assumed a limit in penetration depth in relation to the height of welding current. And furthermore we can assume that when having fixed a specific arc performance to be coupled into the workpiece and presuming the generation of a thermomechanical balance between the heat added and the heat dissipated, we will achieve even a specific volume of (stationary) weld pool. It remains at even this specific size and is neither increased nor decreased in size. This means of course that the depth of penetration may not change as well. Finally concluded: "Specific (Gas Tungsten) arc performance being generated and coupled into a workpiece having even specific dimensions and (thermomechanical) properties yields even a specific depth of penetration."

Resume...

No is my question: "What does happen with the (depressed) weld pool when we switch off the welding current after having held the weld pool itself under stationary conditions (NO movement of the arc)?" And further: "Do we then have a surface reinforcement?"

Oh, I almost forgot to mention but I know you would have already asked. By now we have neglected all the little but often very important details (material composition,...) which can have a strong influence on the weld pool geometry and its behavior. But I hope to be able to treat them in a respective way later on, since actually these details are - at least in basics - not neglectable. 

Coming back to the questions of above...

YUSHCHENKO et al have investigated the behavior of "conventional GTA" vs. "A-TIG" (GTAW under activating fluxes) weld pools generated on a base material comparable the AISI 304 grade for finding a way to a theoretical approach of an explanation of the deep penetration effect in A-TIG-Welding. There it could be seen that - at least in these particular experiments - there is a slight surface depression of the solidified weld pool, see as well the jpegs Surface Depression_1 (transversal section) and Surface Depression_2 (longitudinal section). But what as well can be seen is - at least apparently - in the longitudinal macro section (jpeg Surface Depression_2) that the weld bead seems to be reinforced after the weld has been carried out, i.e. the seam itself should be reinforced slightly. Now one could ask: "Is the amount of molten material volume been depressed similar to the reinforcement?" This to reply is - only from the attached picture - surely hard to accomplish. But nonetheless it might be so. A bit better the effect of surface depression caused by the arc can be recognized by having a look upon the longitudinal macro section of an A-TIG welded material of same heat, see also the jpeg Surface Depression_3_A-TIG. Here one can see the "keyhole effect" - as YUSHCHENKO calls it - which is being physically induced by the activating flux. Here one can see, that the welded seam appears to be reinforced behind the crater which contains the "keyhole".

But, "Where does this surface reinforcement - as far as it is even there - come from?" and before replying this question one could ask as well: "Does the surface reinforcement occur as well when the weld pool is generated and being held stationary?" Can there even be a reinforcement of the seam - or better spot - after stationary weld pools have solidified? Hmmm, actually this is hard to imagine since there is actually no force which could cause the reinforcement of the melt. The only aspect which I could imagine to be responsible for a reinforcement of stationary weld pools is the natural thermomechanical increase in the weld pool's volume by increasing the temperature. What does this mean again? Let the "original" volume of the material (under room temperature conditions "T 0") be "V 0". When increasing the temperature it can be assumed that the materials volume does increase as well. Let now be the arbitrary increased volume of the specific element of molten material at an arbitrary increased temperature "T 1" be "V 1" and the volume of a specific element of the molten material at melting temperature "T Melt" be "V Melt" then we can write for the volume fraction: V 1 > V 0 << V Melt. But how to evaluate the volume of an element of molten metal? I hope that I am doing right now... The volume is the ratio of mass over density. When we now assume that the volume of a constant mass does increase as the temperature increases (what it evidentially does) then the density must decrease as the mass remains constant. I have tried to find some thermomechanical values of stainless steel comparable to AISI 304. First off, the specific weight of AISI 304 is - to my best knowledge - approx. 7.9 g/cm³ (1 g/cm³ = 0.036127292 lb/in³ and thus the density of AISI 304 equals 2,8540561 lb/in³). Let us now presume to have a volume of 5.0 cm³ of AISI 304. Then we obtain a mass of 39.5 g (~ 0,087 lb.). Now it's becoming a bit difficult since it was hard - at least for me - to find the values for the density of an AISI 304 melt. Since the density of a metal is, as already described previously, a function of the temperature and varies of course due to the lattice transformations. However, to not complicating the entire matter, let us presume that the molten metal has a density of ~ 7.3 g/cm³ (forgive me that I remain metric) at melting point ~ 1680 [K] (absolute temperature) what was an approximate value I have found. This means that we have a reduction in density of ~ 7.6%. By knowing now that the volume is supposed to increase since the density has decreased, let's see what should happen. Volume (V) = Mass (m) / density (rho). Thus 39.5 [g] / 7.3 [g/cm³] ~ 5.41 [cm³]. Thus we have an approximately increase in volume of ~ 0.41 cm³, which is an increase in volume of ~ 8.2 %. Although these are surely pure theoretical and inaccurate values we have calculated with but this however means in my eyes: When we could make sure that the area the melt is existing might be "separated" against the surrounding area of the non molten base material (e.g. by extremely cooling the adjacent areas of the seam) and thus generating a steep temperature gradient, I could imagine that the volume of the weld pool would be "depressed" by the adjacent areas and could - under as well presuming a rapid enough cooling of the melt - be - at least theoretically - reinforced in relation to the non molten base material. As I wrote you sometime that in the coherence with an IIW Commission XI Intermediate Meeting I've had the opportunity to visit the British TWI (The Welding Institute) in Cambridge. As you know perhaps the TWI is one of the worldwide leading research institutes in the field of Electron Beam (EB) Welding. When we were guided through the institutes laboratories we have visited as well the EB-labs. There I had the opportunity to talk to Mr. Chris Punshon who is with the "EB Applications Team". It was an enjoyment, by the way. Amongst others he presented us an EB-welded cylindrical steel component (square groove butt joint) having had a thickness of 40 mm. As you know, no kind of filler is (normally) used in EB-welding. However, what I have observed - I guess you know what comes now - was the fact, that although no kind of filler has been used to weld the 40 mm square groove, there was a huge reinforcement both root and surface side recognizable. This effect can be detected as well for larger wall thickness joints. For a better understanding I attach a picture coming from an IIW Commission IV paper written by Allan Sanderson who was the former head of the EB Welding department in TWI, see also the attached Electron_Beam_Weld.jpeg. Here one can see a 150(!) mm thick C-Mn steel square groove butt joint welded at 1 mbar (1 mbar = 2,0885434 lbf./in²) at 200 kV plate voltage, 300 mA current and a welding speed of 100(!) mm/min. Well, as I have now spoken with Chris Punshon (always kept the "Autogenous Welding" topic in mind :-)) I have asked his opinion about where the effect of reinforcement on both sides of the presented EB-Weld might come from. I am honest. He answered: "Huuh, that's a good question!" Then he explained, that the part does transversal shrink by 0.5 mm after whilst being welded. And this effect is supposed to be the reason of the detectable reinforcements. Very interesting as I find, isn't it? Although I had no more chance to ask him if the width reduction is 0.5 mm per each side of the part and thus would yield a total reduction of 1.0 mm (there was no more time to continue the discussion - unfortunately) it would probably pay off - and would be interesting as well - to calculate if the volume of solidified molten metal (even either recognizable reinforcements) would correlate the volume element calculable by: Thickness (t) x Width Reduction (delta W = W - W zero).

Yes I know, Boon does not use stationary process conditions and he does not use the Electron Beam. He is using a continuously moving heat source - even the Gas Tungsten Arc - for executing the Autogenous Welding of his 304 material. Yes, of course. But what I have liked to show previously is that it's worth to make a short sidestep and to have a short view if the discussed effects might be recognized in general in welding, and I would have liked to show the slight difference in the molten pool behavior between a "stationary" and a "moving" weld pool, although I would like to come back to the volume of a melt in another coherence a bit later on.

However, the stationary weld pool of a Gas Tungsten Arc on a stainless steel base material can be assumed to be more or less flat after having solidified - even since there is no complete temperature "separation" between the melt itself and the adjacent base material but there is a heat transfer caused by the well-known physical properties (conduction,...) from the highest point in temperature (melt) to its vicinity (room temperature base material). But what is then the reason for the basically surface reinforcement of a seam been performed by using a moving heat source - as to be seen in jpeg Surface Depression_2. I guess this is an important question to be replied, since there must exist particular forces making sure that even this reinforcement does occur. I honestly hope that I am on the right path by writing all this down, since you could surely ask me: "What the heck do you guy mean by telling all of this?" Hmmm, surely you are right, I guess it is likely not that easy to follow me :-).

O.K. let me try to explain what's driving me...

As I have mentioned already in the above regions of this try of a response I have had a look into our company's scrap box, where I have found some interesting pieces of stainless steel (304 grade) samples been welded by our apprentices and client's welders. Among others I found a piece which had been welded by using the GTAW process. As I could see, it was a piece where the welder - whoever it was - seemed to had carried out Autogenous Gas Tungsten Arc Welding on a square groove butt joint. Although it was in a thickness range (t = 4 mm) laying above that range Boon has used, I could detect an interesting - at least in my eyes - phenomenon. For a better imagination, please see also the attached jpeg Stainless_Surface_Depression_Imbalance. What can be seen there is both concavity and convexity. When I have seen this, I have asked myself what might have been the reason for this imbalance between the areas of surface concavity and convexity (reinforcement). Well, in my interpretation it appears explainable. It might be assumed that the reinforcement is caused by a direct change between the stationary weld pool's influencing forces (arc pressure upon the melt and forces being mentioned already above which induce a particular melt behavior again [surface tension driven = Marangoni]) and the displacement of these forces by inducing an imbalance between the forces being more or less "symmetric" arranged around the arc core and hereby making sure that the symmetry of the (stationary) weld pool is being generated. In other words: "The whole melt starts to be moved, thus showing a divergence in symmetry, thus shows a divergence in temperature profile, thus showing a divergence in solidification and thus showing a different seam profile. This behavior is, amongst others, being caused by the relatively clear to imagine fact. The welding arc as an electrical conductor is arranged always under strong influences of the respective conditions to be found upon the material welded. This means that by changing the material's conditions the electrical behavior (conductivity) of the base material is changed as well. By changing the materials behavior in terms of conducting the current - in other words changing the current density over the weld pool area -  one changes the pressure conditions on the one hand and - and this is much more important - changes the distribution of forces being responsible for the weld pool motion. This means, when having a high current density one has a high temperature and hereby one has a low surface tension. On the other hand when having a lower current density one has a lower temperature and thus one has a higher surface tension. When now changing the condition in terms of the heat source from "stationary" (and thus quasi symmetric arranged forces) to "moving" and thus towards a displacement in force arrangement one has a temperature gradient from "high" (in front of the arc) to "low" (behind the arc). This again yields a weld pool motion from the front to the back of the weld pool area, since the surface tension is higher behind the arc (lower electrical conductivity [of the molten metal] leads to lower temperatures and this leads to higher surface tension forces). Hereby, so at least my assumption, the "compression" of the melt - and thus the final "endeavor" for reinforcement - might be explained.

What does it mean all in all by now? Hmmm, I am fairly tempted to say that we stringently might need the heat source motion for even achieving a reinforcement when welding Gas Shielded Tungsten Arc autogenously. But what explains then the behavior to be seen upon the jpeg Stainless_Surface_Depression_Imbalance? Here we have both over the course of one seam. Concavity and Convexity. This is only an assumption and I do not know if I am right but... Let us assume that the "normal" behavior of a stationary GTA welded - in this case - stainless steel leads to a more or less "even" surface (i.e. neither concave nor convex). And let us now assume that - for achieving even a reinforcement - we need to move the arc (as described above). Then I assume that - as well as there is a threshold value in current for achieving a higher penetration grade - there is a kind of a threshold value for moving the weld pool in a way that the melt does solidify with a reinforcement. When now having a look upon the above mentioned picture, in my eyes it looks like as that the welder has exceeded even this specific threshold value in even welding this specific base material with even this specific thickness. By the way. I have not taken a picture from the rear side of the sample. What the welder has tried was to get a full penetration through the entire wall thickness. He has achieved that. There, where we can see a concavity on the front side, we could see a reinforcement at the rear side of the sample. On the other hand there, where we can see a reinforcement on the front side, we would see no reinforcement at the rear side. And even this is easily explainable. What - so my personal interpretation - what has the welder tried? When we are having a look upon the magnitude of the surface depression at the jpeg, we can see - or at least suppose - that the height of the surface depression is not that great. At least not great enough for achieving a full penetration over the entire sheet metal's thickness. Thus the welder had to reduce the welding speed in an appropriate amount for even achieving this full penetration. However, when he has "interrupted" the continuous (low) welding speed (eventually jiggling?) he has interrupted as well the entire process of weld pool motion, caused an imbalance in both current- and temperature distribution and this again led to a solidification (reinforcement) of an increased amount of molten metal behind the arc. So far even my humble explanation. By the way, there have been conducted investigations in terms of finding out even the GTAW threshold value for yielding a significant increase in the depth of penetration. What there could be found out is that the depth of penetration (for a specific height of surface tension [1300 dyne/cm]) is not that large. Up to a current of approx. 300 Ampere and using a 90° electrode tip angle one could see that the maximum depth of penetration was 1 mm(!) Not that high, isn't it? And thus, as well, when we do have a look upon what Boon has stated what the current values are in his application, one might assume that there must fit everything very accurately for achieving both a reinforcement at the front  a n d  at the rear side of the sample.

So far so good... Really..? ------>  Please see # 2 as the continuation of my response... (hopefully you'll read it) ;-)

For the moment all the best,
Stephan

Hi Allan,

I do honestly hope that you are in the mood to continue in reading what my humble thoughts on Boon's great post are.

So here we go, this is supposed to be the continuation:

-------------------------------------------------------------------------------------------
So far so good... Really..?

"No!" I guess. At least for me personally this is just a very little aspect been treated up to here.

So please let me coming now to another very interesting item, at least for me standing in a very close relationship to everything what Boon has inquired and... what Al has already described when he has posted his reply on Boon's post. Tensions and strains.

Let's thus have a closer look on what Al has explained by treating the mechanisms of tensions in welding a bit deeper. This in particular, since I am certain that what Al has mentioned has strongly to do with the measurable effects upon for instance forming a reinforcement although no filler was added.

Although I have truly considered if it would make any sense to deal with that kind of fundamentals - since I know of course everyone of you does know what is meant by "tensions" in welding - I have decided to do so, since it should lead to an overall view on my personal thoughts on this interesting topic. Thus the first question(s) to be replied should be:

"What is the origin of tensions in welding and how do they work?"

First of all it appears reasonable to distinguish the ways of how tensions can be originated within the material into the two most important ones. Namely longitudinal and transversal tensions being responsible for longitudinal and transverse shrinkage.

Let us consider a rectangular plate and let us assume that we are welding an autogenous bead on plate (BOP), which is situated symmetrically upon the plate. The basis of each remaining tension force after welding is an imbalance in the temperature distribution. When the base material has room temperature and we are welding a BOP upon it, we can measure a more or less steep temperature gradient from melting- to room temperature. The steepness of this temperature gradient (K/cm) again depends on the thermomechanical properties (thermal conductivity, thermal expansion) of the base material welded. This means, un-or low alloyed base materials (e.g. construction steels) differ from high alloyed steels (e.g. stainless steels) - and of course this is physically well approved. In terms of the mechanical properties of a base material it is well-known again the (yield-)strength of the base material depends again on its temperature, which means that the higher the temperature the lower - normally - the strength of the base material. Also it is well-known that when the material is heated up, its volume does increase - as already mentioned above. What does now happen when we have a temperature difference between the melting point on the one and the room temperature on the other hand? The material does expand where the temperature is increased but is restricted in this expansion by the "colder" areas of the plate. The result is a compression between those areas having a higher volume due to the higher temperature thus having a lower yield strength in the yet solid adjacent areas and those areas having room temperature and thus the "normal" base material's yield- and tensile strength, respectively. In other words. Each single point of the base material which is being affected by the welding heat undergoes a slope up- and downwards in temperature. Its temperature maximum depends to the distance between the weld seam (melting temperature) and the point detected. Every (shrinking) action we can observe subsequent to welding depends to even these - quite simplified - coherences. Concluded in one sentence one could say: "If the base material would show no thermal expansion, it would show no kind of contractions. And thus no kind of warping and tensions."

Since we know, that the forces, acting while the welding heat affects the base material are elementary and are not to repress by any means, there is one question which busies myself in an extraordinary way: "How strong does the change in the material's temperature (while welding) - and hereby the change in the mechanical base material properties - does affect the formation of effects as e.g. reinforcement?" Or in other words: "Do the shrinkage tensions influence the formation of a reinforcement in autogenous GTA-Welding?" Well, you again may ask: "Where does this consideration come from?" or "What measurable effect should be stated by even these forces?" As Al has already explained in a wonderful way. What he has experienced when he has qualified the aluminum welding procedure. The fixing of the parts by using strong clamps could finally not repress the thermomechanical sequences. It could only change its measurable effects. What does it mean? Well, let's have a look upon how the "seam" affects both theoretically and practically our rectangular plate after it has been welded. As already mentioned different (special oriented) tensions can be found after welding. Longitudinal tensions for instance are generated by a longitudinal contraction while the seam itself is cooling down. In other words the seam itself is longitudinal or length reduced, respectively. Hereby one can measure a material specific distribution of tension forces, as to be seen in the jpeg Longitudinal_Tension_1. Interestingly both ferritic (unalloyed) and austenitic (high alloyed) steels showing the same behavior in the effects of longitudinal tensions subsequent to welding. What can be seen in the mentioned sketch are tensile stresses (+) in height of the yield strength being situated in the center of the seam. Besides the seam one can recognize compressive stresses (-) of lower heights being damped continuously with increasing distance from the seam's center. The tensile stresses do arrange the reduction in length, what can be seen very clear when having a look upon the workpiece which I have found in the scrap box and what I have already treated in terms of showing both "concavity" and "reinforcement" (jpeg Stainless_Surface_Depression_Imbalance). When the tensile stresses can act in a free way one can very clearly recognize a "longitudinal warping". Please see also the jpeg "Longitudinal_Reduction". A reduction in warping means an increase in stresses and vice versa. Before I come to the treatment of the question above on how the stresses might influence the formation of a reinforcement I would like to deal shortly with the other important stresses namely the "transversal stresses". These kinds of stresses are formed by a transversal contraction of the seam being cooled down. In other words. The areas adjacent to the seam undergo a reduction in width by even being clinched by the colder areas of the workpiece. When I understood correctly what Al has stated, this was what he has experienced when he has qualified the aluminum welding procedure. By having had fixed the two parts by using strong clamps there was a thickening of the HAZ, even by the "damping" or "clinching" effect of the colder areas on the one hand and the clamps on the other hand, respectively. By removing the clamps the part could work and the thickening disappeared. But I guess there might be a width reduction been observable finally. Of course not as high as when welding stainless steel - since the relatively high thermal conductivity of aluminum and its alloys takes care for a relatively shallow temperature gradient. How the "reduction in width" did look like on our autogenous GTA welded plate can be seen by the attached jpeg Transversal_Reduction. I would like to mention an important fact. The welding speed being used for welding a seam does affect the generation of stresses in a wide range. This means. Depending on how fast I weld a seam on a rectangular plate I can obtain different fields of stresses.
Coming now to the coherences in terms of how the stresses or tensions might affect the formation of a surface and root reinforcement when using no additional filler. The question is actually based on a comparison which I was able to accomplish. A comparison between the part which I have used by now for showing some effects of the GTA autogenous welding on an AISI 304 grade steel having a thickness of 4 mm and a keyhole Plasma welded AISI 304 grade steel having the twice thickness, namely 8 mm. More a bit later on...

Basically one could mean or mention by here, that the "reduction in width" affected by tensions might be considered as for being the main reason for a root- and surface reinforcement. Even since the parts - in relation to one another - or better the fusion zone founded on changed thermomechanical material properties (e.g. the yield strength at elevated temperatures) do undergo a shrinkage process and hereby the distance between the parts is reduced as well, there might be generated a reinforcement on both sides (surface and root).

Hmmm, but... "Which distance?"

I have thought there were "no" distance between the both parts prepared as for being a square groove butt joint. O.k., of course there is a distance between both parts, of what Dave has already spoken about. But let's this "distance" be considered as being not "existing", at least under normal circumstances. This kind of condition we call in Germany "Technischer Nullspalt", what might most likely be translated with "technical zero gap". We express hereby that there is of course always a gap when joining two sheet metals one to another by using a square groove preparation, but this "gap" is to little to measure it "macroscopically". But please remember the little EB-Weld story mentioned, there was no "macroscopic gap" between the thick plates been EB welded, nonetheless there were a huge reinforcements detectable after the parts have been joined. This means in my eyes that at the earliest here at this point we should begin to consider, that only the shrinkage (width reduction) alone might be a bit too less to be responsible for the "reinforcement effect" in autogenous welding operations using no additional filler materials.

Just as you Allan have explained this perfectly when you have posted (quote): "I was told that the resulting reinforcements were from the contraction of the weld metal prior to complete solidification of the weld pool as the welding progresses." (unquote). This appears basically reasonable. The weld metal contracts and this yields a both side reinforcement. And perhaps - what even I do not know exactly - this might have been the true reason for the very particular reinforcement you have witnessed. But my considerations go toward if there is perhaps a way for that it can be basically "calculated" whether a reinforcement can be expected subsequent to autogenous welding or not. And for the case of the latter, what the details were being responsible for the "incalculability" of this complex process. So please let me mention that there was a famous German researcher in the 1930's until the late 1960's who has tried to find out and to explain all the peculiarities of shrinking and tensions induced by welding and cutting. Although I truly don't want to treat all his famous papers etc. he has found out a very interesting fact in a very specific welding application. Based on the - at that time - valid predications in terms of welding shrinkage and warping which were commonly based on the statement: "Due to the weld deposit shrinks while being cooled down, warping and part distortion is induced." To evaluate if this commonly statement was right he has SMA welded a V-Joint, prepared by using two 12 mm thick unalloyed (30°) beveled steel plates. And yes, he has found out that there was - of course - a width reduction, just as mentioned above, in this joint. By having welded 5 layers to fill the cross section the parts were contracted approx. 1.6 mm to one another. But what he has found out as well was the surprising fact, that the weld deposit itself could contribute a share of only approx. 2(!) % to the entire reduction. This is actually just 0.032 mm of the total contraction. Please don't ask me where he has got these values from, since I honestly do not know. And by the way, particularly the 2% are somewhat strange for me, compared with the values we have assumed above by using the. A pure personal assumption of mine might be that he has calculated by using the linear coefficient of expansion and not the cubical one and he thus achieving results (~2%) corresponding 1/3 of the actual shrinkage volume. However, this is only an assumption and I have unfortunately no detailed information on this. Although this is an entire different matter on what MALISIUS (the researchers name) has measured (using filler in SMA welding thicker plates and larger cross section) I have asked myself: "Could it really be, that only the share of the contracting weld deposit might arrange the reinforcement?" although as we know the material properties of stainless steel materials are different to those ones of unalloyed steel materials. Or, so I have asked myself, is it as in many other cases as well? Cases in where not only one factor takes care for a specific measured physical behavior, but where the interactions of many different factors together play an intricate game with the observer (even us) and thus impeding him to predict what might happen in very specific relation to a very specific application?

You see questions over questions waiting for answers over answers.

However. What I would like to is to find a way for having finally a clue on how I may be enabled to predict if a reinforcement in autogenous GTA welding can be foreseen or rather not. And to discuss my thoughts on Boon's topic with you great experts. Who else should point me in the right direction if not - you?!

O.k. I have interrupted above where I have spoken about a comparison between "our" meanwhile well-known 4 mm thick square groove AISI 304 butt joint, been manually autogenous GTA welded, and an AISI 304 square groove butt joint been autogenous Plasma welded by using the keyhole mode, but this part having a thickness of 8 mm. First of all I would like to attach some pictures which I have taken from the Plasma welded joint. And these pictures show three different regions of the part's surface. The first of those three is the starting area of the joint, please see also the attached Autogenous_Plasma_Keyhole_Start_Area.jpeg. What can be seen on this is that there is no kind of reinforcement (convexity) recognizable, but rather a slight concavity. More to that later on. Then I'd like to attach the Autogenous_Plasma_Keyhole_Area_Adjacent_to_Start.jpeg. Although similar welding parameters were used one can see that the formation of a slight surface reinforcement has begun. This reinforcement can be detected until a specific distance before the plates end has been reached, see also the attached Autogenous_Plasma_Keyhole_Finish_Area.jpeg. What likewise can be seen here is that where the welding process itself has stopped (keyhole area) a slight reinforcement of molten and subsequently solidified weld(base) metal can be recognized. This finally again by the forces as already mentioned and described above. But however, it's a fact, that we can detect a different molten metal or weld pool behavior respectively, relative to the position of the region of the welded plate. We have an area where we can recognize a surface reinforcement (region between the starting and the finish area) and even regions where this reinforcement is rather not to be detected and where we achieve a flat or even concave surface. And all this, by the way, although the root shows a continuously reinforcement, as to be seen on the attached Autogenous_Keyhole_Root_Side.jpeg. What do I mean by this, you might ask and I would understand you absolutely. O.k. for an explanation please let me... resume.

Please remember the appearance of the part having had a thickness of 4 mm. We could see there a severe warping and distortion, respectively. Both in transverse and longitudinal direction. And what could be seen as well was, that - although there was a root side reinforcement (what I unfortunately haven't photographed, I apologize!) - there was no surface reinforcement. As for comparison of both parts (4.0 mm GTA- and 8.0 mm Plasma) please see also the attached GTA_Plasma_Comparison.jpeg's.

To first of all trying to reply myself the question why there is a different behavior in the Plasma Keyhole welded 8 mm thick AISI 304 part (no reinforcement at the start- and finish area, but reinforcement adjacent to those areas) I have tried to consider to finding out, if - within a longitudinal welded butt joint - as well different residual stresses can be detected. And indeed! The welded part underlies different stress distributions over the course of the seam. There have been conducted some very interesting surveys on this over the course of time. In conclusion: The longitudinal welded part shows different amounts of shrinkage relative to the position. This means, the start and the end of the joint show lower shrinkage amounts whereas the "middle" (longitudinal) seam area does show larger values. This can be simply measured by comparing the different areas in regard to their width reduction after welding. And now it comes... Calculations have shown that there a lot of different parameters have to be considered for approximately assessing the tensions behavior while and after welding and thus to predict the "geometry" of the part subsequent to welding. Even just as already supposed somewhat above. Not only the physical parameter of the solid state (part), but as well the physical parameters of the energy source (arc,...) interact in a most complex way. However, it might happen, that the welded part is afflicted at "start- and finish region" by residual compressive stresses (working against the shrinkage) and the areas between them by residual tensile stresses (causing the shrinkage). This again leads to an appropriate difference in both total height of the stresses and amount of shrinkage (width reduction), respectively. Please permit to avoid the partly difficult details being the reason therefore and the tricky calculations to prove this. However, there are quite complex stress and strain interactions within the material while it's being and subsequently to when it's been welded. This however could explain - as my very personal assumption or hypothesis - why we have detected different amounts of convexity (reinforcement) and concavity (underfill) at the (rather more rigid = no significant distortion) Plasma Keyhole welded part. All the different - but of course important - parameters (welding process, heat input, welding speed,...) are supposed to be neglected in this assumption.

Okay, then I have asked myself:

·  "Is there eventually a close coherence between the stresses responsible for warping and distortion and the size of reinforcement in autogenous (GTA) welding?"
·  "Is it eventually possible that the forces are even the origin of both (Warping  a n d  Reinforcement), and the detectable result (Warping  o r  Reinforcement) depends to how they can act within the material?"
·  If the latter applies, what are then the physical reasons, responsible that either the one or the other can take place? And finally:
·  "How can I find a way to conclude all these personal considerations in a way that no one of you may laugh on me after having read what I have written?" But this comes last!

Now it becomes interesting. As I have mentioned sometime I have spoken with some of my colleagues being real experts in "GTA Orbital Welding" (which I am honestly not!). I have asked them on their personal experiences in observing "root + surface reinforcement" when autogenous GTA Orbital Welding at stainless steels. They reported to me as well as you have done, that it appears quite possible to get both root + surface reinforcement although autogenously welding is used. But what they told me as well. They have found out over the course of the years, that in particular when welding stainless steels it is often a question of incidence whether a reinforcement can  c o n t i n u o u s l y  be obtained. Hmm, I thought by myself, this doesn't make me really happy, since I would have liked to know if the effect might be certainly predictable by knowing what kinds of parameters are influencing the effect. My next thought was - you can imagine: "Nil report!" Since if it might be a question of incidence how can I then understand what's going on when the effect occurs? But we have continued the discussion - you know me :-). After I asked if there are some very specific conditions to be detectable when the effect of root + surface reinforcement does take place, my good fellow Andi Fischer told me: "Yes, we have detected that - beside of course the accuracy of fixing or clamping the parts relatively each to one another (which is in my opinion anyhow a peripheral or external and no true "physical" parameter) - the thickness of the parts appears to have stronger influence on the effect." He explained that it seems to exist a more or less approved limit in wall thickness, affecting the formation of both root + surface reinforcement in autogenous Orbital GTA Welding. And this limit - so his personal statement - might be set at ~ 1.5 mm (~ 0.059 inch). Above this limitation it appears difficult to obtain both root and surface reinforcement. So far so good. Then I have asked Andi if he has no samples for me, suitable for reconstructing what he has told me. And he had. I will attach thus some pictures I have taken from parts been autogenous GTA welded. The first picture shows a stainless steel tube, comparable with what you have posted Allan. It has a wall thickness of 1.0 mm (~0,039 inch), a diameter of ~ 120 mm and has been longitudinal autogenously GTA welded, see also the Autogenous_GTA-Welded_Tube.jpeg. The next picture shows the same tube, though by having a closer look upon the seam surface appearance itself, see also the attached Autogenous_GTA_Welded_Tube_Surface_Reinforcement.jpeg. I hope one can recognize that there is a slight surface reinforcement detectable. Unfortunately I have no further details on used parameters etc. Sorry for that! The next picture I would like to attach, I have taken from an orbital autogenously GTA welded tube having a wall thickness of 1.5 mm and a diameter of ~ 38 mm (~ 1,5 inch), see also the attached Autogenous_GTA-Orbital_Weld.jpeg. Here as well - I beg your forgiveness - I have no parameters available of which have been used to weld the part or how the parts have been fixed one to another prior and whilst welding. I have marked however the details which are important - at least for me - to consider. Over the course of the circumferential weld we can see a planar region (no reinforcement could be obtained upon the surface but at the - purged - root side) where the weld metal has solidified even in relation to the base metal surface, and partly a concave (underfill) region. I guess these are some quite interesting samples for the representation of Boon's problem. The most likely  g e n e r a l  incalculability of predicting how a seam might appear after been GTA welded autogenously. Although I guess that this might be the case for all materials, I guess as well that this might be the case in particular for stainless steel grades. This can - in my opinion - be traced back on the very intricate conditions in terms of surface activating elements - especially sulfur is supposed to be named in this coherence. I can remember that we have discussed this topic within the AWS forum a time back. These elements are strongly influencing the (surface tension) forces which are again a strong influencer of the bead motion which is again the basis for the later melt solidification - and thus the seam appearance. This by the way is though a topic being worth to be discussed separately, thus I would like to avoid to deal with this - crucial - detail herein.

Well, want to coming to an end and thus conclude my observations, considerations and assumptions(!) as follows:

·  When joining two parts together by using a square groove butt joint preparation and presuming to have a "Technical Zero Gap", the parts can not move towards one to another.
·  Nonetheless one can detect - under particular circumstances - both a root + surface reinforcement when GTA welding autogenously AISI 304 sheet metals.
·  These "particular" circumstances are (personally) supposed to act as follows:

1.  The material is heated up while being welded.
2.  By an increased temperature the mechanical resistance against deformation (yield strength at elevated temperatures) is decreased.
3.  The metals volume does increase by increasing the temperature.
4.  The largest metal volume can be supposed to be found at melting and liquid conditions, respectively.
5.  The heated up areas of the parts try - depending to their thermomechanical properties - to expand (thermal expansion).
6.  The colder part areas impede this expansion.
7.  The material - impeded in its expansion - is compressed.
8.  The share of compression should equal the amount of both residual stresses + deformation (Warping/Distortion) after the part has been cooled down to room temperature again and can cause a reduction in width or length (or thickness) respectively.
9.  In autogenous GTA welding of (AISI 304) stainless steel (others are supposed to not being treated here) having a particular thickness under using a square groove butt joint preparation (others are supposed to not being treated here) the thermal expansion can be absorbed only (since the parts are positioned to have a technical gap of "Zero") by the materials regions having a low yield strength (at elevated temperatures) and thus being malleable.

·  This means to me to ask the following questions:

1.  Is there a minimum stress necessary to obtain even this plastic deformation of both malleable metal and base metal within the elevated temperature material's areas while the part is welded? And thus causing the "reinforcement" on both sides of the welded parts? For me personally it appears to be so.
2.  Is there a strong coherence between the physical base material properties (thermomechanical [thermal conductivity,...] + geometrical [thickness,...]) and the physical properties of the heat source (Gas Shielded Tungsten Arc) as well as the peripheral or external parameters (welding speed, thermal ,...) which playing a major role for obtaining a reinforcement on both sides of the part? For me personally it appears to be so.
3.  And now the for me most crucial question. Is there a both sided reinforcement even possible as far the tension forces, induced by the welding process, can be relieved even under their "minimum height"? This by the possibility of free warping and distortion (visible deformation of the parts)? This means, if there is a particular stress force necessary to "create" the reinforcement by contracting the malleable material areas and even this stress force is not being used completely for this plastic "reinforcement process" but can rather act as deforming the entire part geometry (warping/distortion), is there even a reinforcement possible when we have stronger distortion or deformation?

By the way, this last hypothesis or assumption respectively, has grown in my little head when I have seen the very first time the Plasma Keyhole welded part. In particular after I have found out that the stress over the course of a longitudinal seam can change its character by changing its algebraic sign (+ = contraction / - = compression) and by using this fact for the interpretation of the Plasma welded part (areas of underfill and areas of reinforcement + its more rigid geometry compared with the warped GTA welded part) I had the idea that, to obtain a reinforcement on both sides a very difficile balance between the forces acting within the material (being malleable only for a very short period) and the outer parameters (parts fixation, welding process energy density and welding speed respectively,...) might be necessary. And this, I am certain, might be reached as long one can assure that the - most likely - empirical parameters for obtaining even this accurate balance between all the influencing factors are remaining constant. Not to speak of the important factors coming from the base material side (content of surface active substances). These as well should - at least in my humble opinion - be held constant and should not vary too much for obtaining reproducible results in getting a both side reinforcement.

However, now I'm done! And as always the only thing remaining for me is to ask your criticism, suggestions and appreciated comments on what my humble thoughts are!

This is, what I am looking forward to...

So far, may God bless you all and best regards from Germany,
Stephan

P.S. I truly hope to future have more time again to attend this great forum - although I guess it might unfortunately just remain a hope!