You'd need 1 dummy cell, not 3 -- you can only fit two stacks of 18650 in a D Mag (and that only if it's bi- or quad-bored), and 18650s being just a hair longer than a D cell, you'll (probably) get 2x3=6 in a 3D, so only one dummy cell. or possibly offset the 5th cell to the centerline (would require a spacer with contacts/wires to maintain electrical contact, and a centering tube for mechanical support) and protruding into the tailcap. I think you can make it with the simpler no-offset option if you trim the switch body aggressively, but it all depends on the specific cell dimensions.
As for your runtime numbers, add cells in parallel to increase capacity, or in series to increase voltage. Since you need the 16V or so, you're adding them in series; that means the entire pack still has a capacity of 1100 (or 1500) mAh. At the sort of discharge rates you're considering, you'll want to go with good cells, such as A123s; while I'm not familiar with the LiFe options available, I suspect the 1.5Ah cells you found probably can't handle the 12A current.
But I'm not sure where you came up with a runtime of 4.5 minutes, either; if you had a 5.5Ah pack (as you mistakenly figured), you should get 5.5Ah/12A = almost a half-hour; a 1.1Ah pack yields 5.5 minutes.
As for your runtime numbers, add cells in parallel to increase capacity, or in series to increase voltage. Since you need the 16V or so, you're adding them in series; that means the entire pack still has a capacity of 1100 (or 1500) mAh. At the sort of discharge rates you're considering, you'll want to go with good cells, such as A123s; while I'm not familiar with the LiFe options available, I suspect the 1.5Ah cells you found probably can't handle the 12A current.
But I'm not sure where you came up with a runtime of 4.5 minutes, either; if you had a 5.5Ah pack (as you mistakenly figured), you should get 5.5Ah/12A = almost a half-hour; a 1.1Ah pack yields 5.5 minutes.