Hi Tommy,
I truly hope that you had a quick recovery and you're doing better meanwhile!
I don't know if it's right to ask if I may add some humble words on this, since actually all has been said already, but however I'll venture it.
If I have really understood correctly how the parts are being manufactured, I guess there could be at least an assumption for a possible explanation from my side.
When I have read what you have explained on Al's additional request to detail the application, I thought by myself, that the problem may be based upon an effect, called in Germany "Mechanische Porenbildung".
I guess there will exist a technical term in English as well which I unfortuntely do not know. Thus I would like to translate the German term as mentioned above by using "Mechanically Induced Porosity".
Before resuming we should be keep in mind that "pores" are "nothing more" than gaseous inclusions mainly within the weld metal deposit or - in exceptions - in the HAZ (e.g. Microporosity along the grain boundaries). And thus it is clarified that these gaseous constituents "trapped" within the solid material have to "come" from somewhere.
Now as to my best knowledge "Porosity" and its mechanisms can be separated into two main fields:
1. Metallurgically Induced Porosity (Metallurgische Porenbildung)
2. Mechanically Induced Porosity (Mechanische Porenbildung)
The first named is including all cases of porosity based upon Al's - as usual - excellent explanation which was (quote):
"A little residual cutting fluid, machining fluids, forming oil/lubricant residuals on the tube or machined part, light rust, etc. could contribute to the problem. Without seeing the conditions of the parts it difficult to guess, however, porosity is typically caused by surface contamination, thus there are gas reactions that can result in porosity. Light rust can contain retain limited amounts of surface moisture and hydrates that can result in moisture." (unquote),
which means that the liquid weld metal is contaminated with "gaseous cleavage products" from hydrocarbon containing lubricants (generating hydrogen) or solids (generating in particular nitrogen or oxygen). By being split within the arc plasma, the solid or liquid contaminants generate gaseous constituents which are resolved within the liquid weld pool. As liquids have a higher "disolving power" than solids the gaseous constituents are "harmless" as long as e.g. the temperature does not drop. In case of welding whereas the liquid does - of course - solidify to form the weld seam. By decreasing the temperature the melt's "disolving power" drops as well and it tries to get rid of the gas.
Now for both - solidifying melt and escaping gas - it comes to a kind of "race against the time". This, since the crystallization front of the metal moves with a specific (depending to many different factors e.g. thermal conductivity,...) velocity. As in the amount of that the metal's solidification does proceed the gaseous contaminants solvability in the metal decreases.
And if now the solidifying metal's crystallization front is "faster" than the motion speed of the particular gas on its way towards the seam surface, it is entrapped within the solid metal and forms - pores.
This mechanism is drastically observable in pure metals (having no solidification range but a specific solidification point) and here again the best example is pure aluminum. The solvability for hydrogen in liquid aluminum is 20:1 compared with its solid state of matter. As pure aluminum has defined point of solidification much of the disolved hydrogen can be entrapped and yield porosity.
Nonetheless, metallurgical induced porosity can be created as well as by welding "older" steel grades having higher amounts of impurities or which are unknown by their chemical analysis (e.g. basic Bessemer steel etc.). But I guess to weld these steel grades is rather rarely nowadays.
But however, this all is certainly well-known. The reason I have mentioned it though is to compare these mechanisms with the above second named, the "Mechanically Induced Porosity".
Mechanically induced porosity is based upon the physical volume expansion of gases at higher temperatures. For instance, when welding circumferentially a fillet joint around two parts and between the two parts a cavity exists, the temperature rise by the arc can cause an evaporation of surface moisture, eventually both part surfaces are contaminated with, or can simply cause an extension of the volume amount of air within the cavity, existing between the two parts. I have prepared a s c h e m a t i c representation, please see also the attached Circumferential_Weld.pdf, for a better understanding of what I am trying to describe. By finishing the weld the extended air, gas,..., tries to escape the cavity and the only way is through the solidifying weld bead itself. By even its way across the weld seam's cross section it can thus be entrapped in the weld metal deposit.
Very often I have seen mechanical induced porosity in joining zinc coated thin sheet metals by having a lap joint geometry. By having a gap between the two sheets we have the same problem as mentioned above but severely aggravated by having the additional reactions of vaporized zinc!
By vaporizing zinc between the upper and the lower sheet, which can not escape across the gap between the sheets its way is determined to be through the molten and subsequently solidifying bead.
The result can be seen in the attached Mechanical_Pore.jpeg and I guess this picture says all.
Mechanical induced porosity must have thus a continuous connection to a kind of "reservoir" of gas, or gas generating substances which can be provided to the weld pool. Even this "gas generating substance" in the attached picture was the zinc, which was vaporized and could not early enhough escape through the solidifying weld metal. In other cases this can be e.g. moisture or even the substances Al has named in his excellent post.
Perhaps thus finally my assumption, might this be the case as well in your application and your problem was based upon mechanically induced porosity. Even as well that you have named the porosity found in your application "...under surface porosity..." let me assume that the way for the gaseous constituents trough the weld's cross section might be slightly too long before they can escape entirely from the weld metal.
Though by increasing the temperature (preheating + weld performance) you might have "done even right" since you have caused hereby - perhaps - a slightly extended period of liquid state which may allow the escaping gaseous constituents to "find their way" through the weld before being entrapped within.
So far my humble considerations, as usual please correct me when I'm wrong.
Best regards to you and everyone,
Stephan