Hi cidme
I must agree with Tim, that you have some kind of contamination present. Either the base metal is just a very "dirty" casting, or your casting has experienced some preferential corrosion. If the pump has been in service for a long time, then it is possible that the material has experienced preferential corrosion of the ferrite phase, leaving a somewhat porous structure of graphite. This can typically be seen if the material seems to just "vapourize" when you strike the arc. When this has progressed very far, then welding is impossible. The material just dissappears and does not melt.
At any rate, if you have contamination, using SMAW should work better than GTAW. Especially if you use an ENi-CI electrode
Regards
Niekie Jooste
Fabristruct Solutions