I haven't a direct answer for your question which is not a "Tank" matter.
I think you can consider any gap that assures complete penetration and complete fusion, as 5.1.5.5 & 5.1.5.6 require. So I think a gap of 10mm would be acceptable if is confirmed by a welding specialist.
A remark: 20mm thickness of annular plate would be a consequence of thick shell plate (and consistent stress in the first shell course)- see API650/5.5.3 (with the remark you may be asked for an elastic analysis if you are beyond H × G ≤ 23 m condition). Also, a consequence of tb=20 mm would be a great radial width of annular plate- following several requirements of API, including seismic appendix E requirements.
What I try to suggest you is to be sure tb=20 mm is indeed required by your specification/ calculation.
Best regards.
Edited by mariog (11/07/11 09:05 AM)