A bit of theory on what is called "upsetting" is helpful to understand how to control welding warp. If a concentrated heat is applied to one spot in the center of a plate, the surrounding metal acts just like a vise: the hot spot needs to expand, but it can't expand outwards because of the vise-like effect of the cold sheet around it. So, it expands upwards. If the heat is applied long enough for the part to become a dull red (and if the material had zero stresses to begin with), the metal is said to be "upset", and a quick quench of the area will intensify this effect.
The cold upset area is now greater in height and less in diameter than it was initially. So, now it acts like a winch and pulls inwards against the cold metal surrounding it. This creates the warp factor. A little bit of practice, and you can take advantage of this property to straighten a large tube or shaft.
Beware that lots of ready made materials have built in stress from cold working. C1018 shafting is bad for this, because heating a spot on it will have completely the opposite effect to what I just described. This is because of the skin stress created in the material by being stretched in processing. So a little bit of heat immediately relaxes one side of the shaft, and the still stressed skin on the other side pulls it away. If you mill C1018 cold drawn shafting or keystock on one side, you will also see this same effect of the stressed skin warping the piece as soon as you take it out of the vise.
So, back to the welding and upsetting. If you preheat the entire area, you reduce the upsetting effect. I recommend 500°F preheat in practically all circumstances (we use lots of propane up for preheating ) Now, the red hot weld zone is only 1000° hotter than the surrounding area, rather than 1500° (approximate figures). If you are fast with the ball peen hammer, you can "undo the upset" by peening the hot weld zone (while it is still red hot). And I mean hit it, not just tapping on it to make a noise
I'd start with a few good, but short, root welds on opposite sides, but I'd use a stick welder for this. It gives you a little more time to work. But, you use what you've got. It is important to work quickly to get opposing welds laid as fast as possible. Test whether your alignment shaft is binding up as you complete each pair of opposite welds. Its the first welds that have the most effect on the outcome of the alignment. You'll have to allow the temperature around the tube to equalize completely before you'll know how straight it is. This means that it is slow going at first. I use a digital thermometer to check the temp in several positions around the tube.
I'd take the precaution of wrapping the alignment shaft with some paper, so that if a bit of warp occurs, I won't get the thing seizing in the hole when I go to drive it out.
If you test the housing back plate flanges for wobble with the alignment shaft still in place, you can deduce where the final bend state is. Than, deliberate "upsetting" can be performed with an oxyacetylene torch in several places along the tube. I recommend oxyacetylene for this because it is very hot, and successful flame straightening requires fast heat to upset the spot quickly. Quench with a wet rag.
If you bend it the wrong way, I am not responsible
First you get good, then you get fast. Then grouchiness sets in.
(Note: The opinions expressed in this post are my own and are not necessarily those of CNCzone and its management)