I would recommend review of the following document...
https://nickelinstitute.org/~/Media/Files/Technica...

Thanks metengr,
I've done the Schaeffler math so I know where we need to be. I just don't know what to expect for dilution.
My strategy is this: instead of going the higher alloy route (the conventional strategy), I would like to try and make ER312 work from below. At 30% dilution I calculate about a 28Cr-20Ni stainless deposit with around 6% ferrite.
The high alloy route means 52M and I don't need to tell you how grief-prone that is.

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

I would favor Inco 82 filler metal.

https://www.tandfonline.com/doi/abs/10.1179/136217...

OpEx (imaginary as it turns out, but that's another story) does not permit us to use this long-time preferred option.

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

Instead of guessing, perform one and section the weld. the macro will tell you the dilution (based on surface area).

---
http://be.linkedin.com/in/fusionpoint

kingnero,
Do you actually understand what dilution is ?
Probably good idea you do a bit of reading first and learn before commenting and making a very basic mistake.

random google image:


Please enlighten me what's wrong with this method?

---
http://be.linkedin.com/in/fusionpoint

Dilution of the fillet weld deposit will be maximum at the root. I would suggest a mock up of the actual fillet be made and after depositing the root pass take a chemical analysis of the deposit. With proper arc positioning and selected current you should be able to minimize dillution from the 690 alloy and maximize dillution from the 304L.

Apart from a small zone near the fusion line, dilution is fairly equal over the entire weld bead (speaking about single pass welds, like 1/8 fillets). By my knowledge, the only reason why DekDee would call this a very basic mistake, is because some elements vaporise during the arc transfer. But I'd say manganese is of minor importance here, apart from catching sulfides.
However, apparently I'm wrong, but I'd really appreciate somebody pointing me to where I'm mistaking.

kingnero,
My apologies - my response was a bit harsh.
My comment was based on dissimilar metal welding and the resultant chemical composition induced by dilution.
Maybe I am the one who needs to do more reading.
Regards,
Shane

That's all right - if I were easily offended, I shouldn't be online.

However your last reply still confuses me.
Even with DMW, using the calculated dilution (based on surface area of both base materials / total molten zone), this still gives you a rather exact indication of the chemical composition. Do you suggest otherwise?
Again, taking into account the loss of some elements (I should have the list here somewhere), but you should arrive within a few % of reality.
I'd say that's good enough for doing the Schaeffler/De Long/WRC20/Bystram math.

I am seeking actual knowledge, not guesses or formulae. My limitation is that I do not have access to a shop where running a test would be simple exercise.
'Welding development' is not R&D, it is trying to efficiently reproduce what 1,000 people before me have learned - I'm trying to ask one of those 1,000 people.
And for this case, a ferrite scope (or even a magnet) would be the best quick and dirty way to assess the composition and obtain proof of principle, and generating a little ferrite is my initial hope/objective.

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

The problem is that welding metallurgists would not recommend the filler metal selected, opting for a nickel alloy filler. In my 45 years experience, I have never seen it done. I have seen folks who attempted to use 309 with very bad results and they were forced to use accepted nickel base alloy fillers. You also have not indicated the thickness of either alloy to be joined. Is this a lap fillet or T-fillet?

With GTAW we have readily met less than 20% dilution, manually applied. With automated systems we have seen dilution very near 5%.

I understand that perfectly; an attempt with 309L already failed (all other people's work). But I believe 312 has a fair chance, and if 20% is anywhere near accurate then I am certain it is doable. Desperate times call for desperate measures, and 52M has desperately poor weldability, and I am looking to avoid it at all costs.

https://www.google.ca/url?sa=t&rct=j&q=&am...

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

Why would 52M and 312 be your only options? What about ERNiCrMo-3 (625) or ERNiCrMo-10 (C22) filler metals?

Kingnero,
I may be getting myself confused.
This is an excerpt from metengrs link regarding DMW.

"In fusion welding, the weld metal is a
mixture of the two metals being joined and
the filler metal. In arc welds made with
consumable electrode processes such as
SMAW, GMAW, SAW, and FCAW, the
weld metal is well mixed or stirred by the
arc action and the composition is quite uni-
form from one area to another. By sampling
any place in the weld
bead, the weld com-
position is determined and weld properties
reasonably predicted."

GTAW is not listed above so it is possible the weld deposit will not have a uniform composition in all locations.
How then will a macro show the approximate chemical composition of the weld deposit based on surface area ?

I am not looking to argue - just to learn a bit more about dilution.

All the above processes are current-carrying electrodes. There, Lorentz forces stir the weld pool and make it homogenous. In case of non-melting electrodes, Marangoni currents will determine movements in the weld pool (either of type A, outwards, or of type B, downwards and promoting penetrating) to the same effect.

I appreciate the enthusiasm but there are way too many tangents here. I think I framed the question in a fairly straightforward way.

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

I'd say 15-25%, 30 max, but it depends on the welder. Theoretically, GTAW has a 0 to 100% dilution range. However I'm surprised you'd suggest 40%. Seems high to me, however my opinion is probably worth what you paid for it... (as per your signature text )

kingnero,
It's an ongoing struggle to align my opinions with facts

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

I should place a caveat concerning >20% dilution. This was readily obtained when we implemented low dilution welding parameters to the WPS and to the welders qualifications and we inspected/observed/assured that welders were welding accordingly. I believe these issues will primarily affect your selection.

Left to their own devices and preferences, welders may weld with too high amps and slower travel speeds than appropriate.

ironic metallurgist,
It was me that took the post off on a tangent and for that I apologise.
I made an inappropriate comment and then spent subsequent posts trying to withdraw my foot from my mouth.
I am still confused but will now stay well away from metallurgical posts,
Cheers,
Shane

The WRC92 diagram will provide a better estimate of the ferrite content of the single pass fillet weld than the dated Schaeffler diagram. The chemistry of the completed weld can be assumed to be thoroughly mixed and thus homogenous.

Unless the base metal is very thin, I would start with an assumption that 15% of the weld chemistry comes from the nickel base metal, 15% from the stainless base metal, and the balance from the filler metal you select.

Nickel promotes austenite while chrome promotes ferrite. A common target for dissimilar base metals composed of carbon steels and austenitic stainless steel is 18% chrome and 8% nickel (minimum values) and a ferrite number between 3 to 10, 3 being the low and 10 being the high. Too much ferrite will promote Sigma phase, but if this is limited to a single pass fillet weld, it should not be an issue.

The Nickel Development Institute out of Toronto has a wealth of information on this subject. It may be worthwhile to review their literature.

Best regards - Al

gtaw,
I initially used just Schaeffler, which is more quick & dirty. Results were borderline, which is why I came here seeking hard experience.
I took your suggestion and plotted it on WRC-1992, and I was pleasantly surprised to find the predictions are significantly more favourable in terms of crack-resisting FN. However I did have to extend the isoferrite lines to cover the territory around about 28% Cr and 20% Ni.
`
30% dilution (15+15) predicts around 15% ferrite (higher than I normally would like).
40% dilution (20+20) predicts around 6% ferrite - just right.
So ER312 seems like a safe approach. Next step - find a way to get this tested at no cost to me

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

Except for the lower values, ferrite numbers are not the same as % ferrite. The WRC98 diagram provides values for the approximate ferrite number.

Best regards - Al

gtaw,
That is true, but I would estimate the error in my assumptions of compositions of the contributing alloys, variability due to welding conditions, and scatter inherent in the WRC-1992 diagram to each be greater. Of course dilution will always be the biggest variable.

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."