I'll go out on a limb here and say that probably when you put a 635d.C PWHT on it in open air, you probably contaminated the whole piece you were welding...Also, when you removed the crack intially, did you use dedicated tools, ie. grinders, files, etc, that had not come into contact with anything other than TI as so not to contaminate them? I'm not a titanium welder, so I'm just giving information that I have read. (my disclosure statement)
Regards,
John

Hi Alfred!
A few questions need to be asked before any of us can even come close to giving you some possible causes for the cracks in your "case"...

1) Is the AMS 4951 Ti filler metal the most compatable filler metal for a part consisting of AMS 4901 Ti base or parent metal?

2) What is the thickness of the metal and is it tube, pipe or plate?

3) Did you back purge the base metal prior to welding the initial crack?

4) How deep and what size groove did you prepare the crack for welding?

5) During the welding of the crack, did you use a trailing shield along with the shielding gas of the TIG/GTA torch?

6) Do you know what color was the weld at the face, HAZ and possibly at the root of the weld prior to stress relieving and what was the color of the metal in the stress relieved zone after the metal was allowed to cool to room temperature?

7) I do'nt mean to be rude but, how much do you know about Titanium?

I ask this because as Boilermaker suggested, stress relieving Ti, especially a Ti alloy in open air at 635 degrees C or approximately 1175-80 deg. F and then letting it cool to room temp without any protection from atmospheric contamination was not very thoughtful of you or whoever performed and was in charge of the work - to say the least!!!
Note: The max annealing (Not the annealed!) or stress relieving temp should be no more than 1000 - 1100 deg. F (538 deg. C - 593 deg. C) in open air for 30 minutes... Reference from Allegheny Technologies webpage for CP grade 4 Ti AMS 4901/Allvac70 CP-4 Titanium/UNSR50700.
Here's the website: http://www.allvac.com/allvac/pages/Titanium/UNSR50700.htm
When you get there scroll down to the Allvac CP series technical data sheets that are available in either .html or .pdf format; Once you get to the data sheets, scroll down to "Heat Treatment" and "Special Precautions" and read what is written...
This also probably was the cause of the new crack outside of the weld area... When you tried to repair the new crack in the incorrectly stress relieved zone, another crack was generated because you did not initially protect the stress relieved zone from atmospheric contamination prior to and while the zone was at the described temperature (635 degrees C?) also - while cooling down to room temperature!!!
Therefore, most likely the root cause of your problems encountered after your initial weld repair was due to the lack of atmospheric protection (PURGING OR SHIELDING) at the stress relieved zone of the base metal!!!
Why? The base metal alloy's microstructure changed (It gets excited!) during stress relieving which is normal but because that area was not shielded or purged from the AIR, just think about it for a minute -either oxygen, nitrogen, carbon dioxide and most likely hydrogen got in between those crystalline structures and at least with the oxygen, carbon or hydrogen, reacted with the titanium!!! Voila! Not only did you get a reaction from the oxygen or nitrogen, you probably also got some hydrogen or carbon entrapment or embrittlement as the base metal in the stress relieved zone cooled off!!! The metal never had a chance (stress relieved for too long!) to change properly again back to the proper grain structures (Abnormal & uncontrolled grain growth and alot of voids in between!) and therefore maintain it's mechanical properties, plus the fact that you had too much of what was already in the original crystalline structures which only had minute amounts of oxygen, nitrogen, carbon and hydrogen!!!
I use the word "PROBABLY" alot in my assessment and for good reason!!! It's because I do not have all the necessary details on how you actually performed the repair and the other details such as some the ones listed above in the form of questions!!!
The only way to know for sure is to perform a detailed metallurgical study for sure but based on the information you already provided, I'd say that the root cause is leaning heavily towards the fact that you did not shield or purge the stress relieved zone prior to, during and after "stress releasing" the part!!! Titanium and it's alloyed variants make a REACTIVE metal so careful consideration should be taken when working with this metal and it's alloy variants!

Respectfully,

SSBN727 Run Silent... Run Deep!!!

Don't have time for a full length reply but here is some good data from "Timet"

http://www.timet.com/fab-p33.htm

The questions about purge/shielding of top and bottom of the weld and the color associated (if any) are vital.

I am curious about your "like a lamp" equipment for stress relief.
While your temperatures appear to be within norms, is it possible that heating was uneven? If the surface exceeded your stated 635 C in order to get a full soak you may have indeed created a soluable condition that is detremental to the integrety of your part.

Here is a little excerpt from Timet
"TIMETAL Code-12 grades, are typically stress-relieved at 900-1000 degrees F. for 45 minutes and annealed at 1300 degrees F. for two hours. A slightly higher stress relief temperature (1100 degrees F., 2 hrs.) and annealing temperature (1350 degrees F., 4 hrs.) are appropriate for the TIMETAL 6-4 alloy. Air cooling is generally acceptable."
"Although no special furnace equipment or protective atmosphere is required for titanium, a slightly oxidizing atmosphere is recommended to prevent pickup of hydrogen. Direct flame impingement for extended periods, leading to temperatures in excess of 1200 degrees F, should be avoided because of the potential for contamination and embrittlement. Hydrogen or cracked ammonia atmospheres, also, should never be used, because their use would lead to excessive hydrogen pick-up, and embrittlement."


Knowing the joint thickness and geometry will also help determine if the secondary crack was born of internal stress or contamination.

More data will help provide higher quality advice.

By the way. Are you working to a published specification? If you are please tell us that also.

Lawrence