What computer are you using? Some of the comp analysis doesn't look right.
Lets look at the question moves in order:
8. Qd3 (mistake, comp likes Bc4)
A mistake but not because of your reasons. A move that is almost as good as Bc4 is Qb3!, which does exactly the same thing as Qd3 but also looks at f7 at the same time! It is easy to see the queen is better there.
After Qb3 black is almost forced to BxB (f7 is almost not defensible otherwise) which is very good for white. The computer does want BxB with advantage but In the game notice black did not need to play it due to the inaccurate queen.
9. Bc4 was not a mistake. You probably got confused with the lines.
11. Bf4? was a blunder. You didn't mention it in your analysis. Ng5 and Qf3 threats against f7 is winning as black either lose a piece or severely weakens structure and face mate threats.
20. Rfe1?! was an inaccuracy. Not much of one in my opinion.
22. Bd3?! inaccuracy, but my engine did not give Rab1 as best. My engine likes h4 best, and then we have about equal weight e5, Rac1 and Rab1 (the eval jumps back and forth). Anyways the black bishop is not threatening much on a6 so Bd3 just trades a good bishop with a bad one unnecessarily and the mentioned moves put more pressure on black.
26. e5?! inaccuracy, but your comp recommendation dxc5 was nowhere to be found and not even one of the candidate moves. Red1! is best and very sensible move.
33. a6? mistake. This one is obvious as the c-pawn is lost with Rxc5, which you should have protected.
Given the amount of times I run out of time in 15 | 10, or feel forced to move more quickly than I like, I've decided to move to 45 | 45. The following game makes me think this was a good decision, with half the number of blunders & mistakes from a computer analysis. I'm not sure, in every case, why the computer declared my moves a mistake, so any pointers would be greatly appreciated.
I'd highly recommend 45 | 45 to any other beginners who are feeling time strain!