All good suggestions (some excellent). Personally, I'd ditch the 100% blind increases during the early rounds and go for a smoother blind structure using antes from R2. See below for an example of how to actually get more play out of fewer starting big blinds (8-12 players, T2000 stacks):
| Rd | min | sb | bb | ante | % incr |
| R1 | 20 | 10 | 20 | 0 | 0 |
| R2 | 20 | 15 | 30 | 5 | 0.50 |
| R3 | 20 | 20 | 40 | 5 | 0.33 |
| R4 | 20 | 30 | 60 | 10 | 0.50 |
| R5 | 20 | 40 | 80 | 10 | 0.33 |
| R6 | 20 | 50 | 100 | 25 | 0.25 |
| R7 | 15 | 75 | 150 | 25 | 0.50 |
| R8 | 15 | 100 | 200 | 50 | 0.33 |
| R9 | 15 | 150 | 300 | 50 | 0.50 |
| R10 | 15 | 200 | 400 | 100 | 0.33 |
| R11 | 15 | 300 | 600 | 100 | 0.50 |
| R12 | 15 | 400 | 800 | 200 | 0.33 |
| R13 | 15 | 600 | 1200 | 200 | 0.50 |
| R14 | 15 | 800 | 1600 | 400 | 0.33 |
Another thing I'd point out is that if you plan to have different lengths for different levels, they should typically
decrease as the event progresses, rather than
increase.
Think of it this way: you optimally want to be able to play the same number of
hands per blind level, and with fewer players as the event progresses, the number of hands played per minute typically
increases (fewer players to deal in, fewer players to act pre-flop, fewer players to see a flop, etc.). This means that you can actually
decrease the round length as the field gets smaller and still have the
same number of hands being played per blind round. To prove the concept, just look at the two extremes -- two players heads-up can play a lot more hands per hour than can an entire table of nine players.
Start the blinds at 20 minutes for the first six rounds (hours 1 & 2), then move to 15 minutes for the next eight rounds (hours 3 and 4). It should work out fine -- chances are your 12-player field will be heads-up by Round 12 or the 3-1/2 hour mark and finish shortly thereafter.