I found in a similar situation a while back that the best weld is no weld at all...

I spent a few years building mobile oil drilling rigs of modular design called "Boot Strap".
The modules were separated by size / weight so that they could individually fit on a flat bed trailer.
When delivered to an offshore platform, they assemble rather easily by crane through the use of hook and pin connections.
The hook connections, welded to the side of a module, were pairs of 2" thick T1 plate shaped with a slotted hole, open on top and spaced 3" apart to receive a horizontal pin dropped in from above.
The pin plate was a single 2 1/2" thick T1 plate, welded to the side of another module a placed so that it would fit in-between the hook plates and catch via a 3" dia 4130 pin press fitted transversely.
We found that placing a circumferential fillet weld around the pin to secure it to the pin plate was problematic due to cracking and clearance issues, especially when repairs were made by well meaning but unknowledgeable welders.
The solution was simple, press fit the pin to such tight dimensions so that the only way it could be forced to fit was by freezing the pin in order to make it contract a little. After fitting and expansion, the pin could not be removed with anything less than a cutting torch.
By eliminating the weld, we also eliminated the cause of failures / repairs.

This may not be applicable in your situation, just a point to ponder.

Tim

This is exactly why most welding standards require the proposed WPS be qualified by testing. Qualifying the WPS demonstrates the weld will perform as expected. If the test weld doesn't pass the required test, it's back to he drawing board to determine why. Corrections re made and another test weld made.

There isn't always a magic formula available. Sometimes it takes time and money to research and develop a viable WPS.

Al