How I sacrificed my queen positionally to defeat a 2400 International Master OTB!

I just finished playing a week long tournament. I did quite well. 5.5/9 losing only to a GM, IM and FM. 

(Playing a GM)

I did however manage to get one of my best wins so far against a 2400 IM and the game happens to feature a positional queen sacrifice, so I thought I may share it here.

I thought that I could just play it slowly by improving my position little by little with moves like Be2-c4, Qc2-e2 and trying to play for the c6 square with Nf3-d4-c6. But that was exactly what probably my 2400 opponent wanted. I wanted to find something double edged, which would take his out of his comfort zone and create immediate chaos on the board while my pieces were still placed better than his. And suddenly I came up with a very surprising idea. 

And then came the even more shocking follow-up:

I was very happy that I got the courage to play this sacrifice. I didn't know if it was fully correct at the time, but intuitively white seemed to have huge compensation. Already I had a rook and minor piece, which in itself is already quite good, but I additionally had an extremely strong passed pawn on d6. The concrete calculations only confirmed my enthousiasm for the sacrifice. 

The more my opponent thought, the more I liked my prospects. Just skimming through variations I started to see more and more tactical ideas. The benefit of being on the side with the initiative, in contrast to the defending side is that even if I miss some tactical idea, it doesn't matter as long as I find something else. On the contrary if the defending side misses even a small tactical detail, he risks losing the game on the spot. 

The game continued: 

Now we ended up in this endgame where I have a rook, bishop and knight for the queen. It is obviously winning for me, but to my opponent's credit he showed some solid defensive skills. 

I wasn't sure how I would tackle this endgame, but I did not feel the hurry to push too much immediately. After all since I had plenty of pawn moves left, which meant that I could easily delay the 50 move rule for hundreds of moves. There was 30 second increment. 

The idea to torture my opponent for hours seemed even better that trying to win as fast as possible. Of course I would have loved to win quickly if the opportunity arrived, but there was no reason to take any unnecessary risk.  

After 43...g5 I have an important decision to make about what pawn structure I was heading for. 

I considered the following options: 

1) leaving everything as it is (either after exchanging on g5 or not)

2) pushing h5, keeping as many pawns on the board

3) pushing g4 with or without trading on g5.

The problem with plan 1 is that my rook can never leave the fourth rank due to the threat of g5-g4 which means that I would only be able to move my bishop, which is out of question.

I was worried that plan 2 could backfire as the h5 pawn could be a weakness after something like f7-f6 and Qe8 attacking that pawn.

The third option was the most pleasent to me.

I played the last 10 moves just to gain time on the clock.

Now the situation became much more serious. I ran out of pawn moves! This means that it's the last 50 moves I get to play, so I really need a winning plan.

I need to attack black's king, but how?

If I bring 2 pieces to the attack, it will probably not be enough... I need 3 pieces to create real mating threats. But if I bring my rook, bishop and knight to the attack, then my king will be weak and will get checked forever by the black queen... 

Then the winning plan hit me: 

I need to bring my king along all the other pieces so that they all support each other like one giant snowball rolling down the board. 

Here my king is well protected and fully active as well, which I believe is the only way to win this endgame.

Here's the full game in its entirety: