Nice game.  You played very proficiently.  I think White gets a much better game with Be3 in the endgame.  You gave 19. Nxe3 a ?!, which I think it warranted for hanging the pawn, but also, the bishop is much better on the open diagonal of d3 instead of fxe3.  Consider the line 19. Bxe3 Bxf1 20. Bc5! and then either Black saves the bishop and loses the exchange, or saves the rook and loses the bishop.

Also, I agree that a2-a4 is a bad move.  If your opponent have a majority, the worst thing you can do is to march your pawns toward it!  Consider this:  If you've got a 2 on 1 majority and the pawns are in the starting position, it takes 5 moves to offer the exchange of pawns (a5, a4, b5, b4, b3 for instance against an a2 pawn) and two further moves to queen (not counting the tempo-neutral 6. a2xb3 a4xb3).  But if the minority pawn is on, say a5, then it's only 3 moves to offer the trade (b6 first, then a6 and b5).  That's a whole two tempo faster!  The moral of the story is to never spend a precious tempo moving your minority toward the majority.

Good game, and way to capitalize!

Thanks for the insights ChessSoldier.

In the game, I'd actually thought that 22. fxe3 was nearly forced since Bxe3 allows the expanding move 22...d4 with tempo, but perhaps it was better all the same to keep the bishop active, and the pawn structure intact.

I think you're right in general terms about the pawn push, but had the position been slightly different, I can see how a4 would have been a good move, since key to the position was the pawn on c5, and a potential pawn push on b6. There just wasn't time for a4 in this position, and it relinquished whatever pressure white had.

Playing Rxg5 was enjoyable! Wasn't too difficult to decide on, though, since I had three advanced pawns on opposite sides of the board, and it was pretty clear they couldn't all be stopped.

i still dont get why didnt you take his queen because you lost a knight for pawn and you could take his queen and a pawn for knight and bishop, 

Looking at it again, maybe you're right - it would have been a knight, both bishops and a rook for queen, rook and pawn:

17. exf6 Bxe2 18. fxe7 and the rooks are forked, so 18...Bxf1 19. exd8/f8.

In the game I assessed it as being tricky to play for black as although I'd have a queen for three pieces, those pieces could coordinate more easily than my queen and rook if he played something like c3, closing off the c-file. I was pretty tired, and so I took the easier advantage. Difficult to say which line was better.

thank you for enlightening me

ha, no problem!

Your opponent played the opening rather uninspiredly; after move 12 you were already slightly better. 15.Rb1 seems a better move for white, since the queen will have to move again as well. 16.e5?? actually loses, because after 17...Bxe2 18.fxe7 Bxf1 19.exd8Q+ Rxf8 20.Kxf1 white has three minor pieces for a queen and a pawn, which would be about equal since white has the bishop pair, but in this case his pieces are too uncoordinated: 20...Qb4 21.Be3 Qc4+ wins the a2 pawn, and the pressure on the queenside after Rc8 is too much.

After 17...Bxf6?? white is better because of his bishop pair, but 23.fxe3? was bad since this gave him an isolated and misplaced pawn. After 26.e4? you could simply have won a pawn with dxe4 because of Bxe4 Rd5 winning a piece. 30.Rd2?? was a blunder that abandoned the defense of c5 and allowed you to fork the pawns and activate your rook. With 30.Rc4 Rf8 Bd7 white would simply win the exchange and the game, while 30...b5 31.axb5 (31.cxb6ep Rxc4 32.Be6+ Kf8 33.Bxc4 d2 34.Bxd2 Rxd2 35.b7 Rd8 36.Bxa6 is about equal) axb5 32.Rb4 Rb8 33.Bb6 is also easily winning for white. 31.Bb6 might still have held after Rxe4 32.Bf3 Rd4 33.Bxc6 bxc6 when black is better because his connected passers and active rook outweigh the bad bishop, but perhaps not enough to win. After 31.Bc3?? (was your opponent in time trouble?) you won easily, but demonstrated some nice technique. Well done.

@Chesssoldier: "If your opponent have a majority, the worst thing you can do is to march your pawns toward it!"

That's usually not true, obviously; the 'minority attack' is a common strategy in many positions, such as those arising from a Queen's Gambit Declined exchange structure; if you can trade all the pawns on the wing, the one remaining will be isolated and weak. In this game, though, black had a rook while white had two bishops, and a rook is strong in supporting a passed pawn while bishops are bad at attacking it, so the strategy was kinda dubious.

Well, sure, if you take my statement out of context, then of course it doesn't work.  But I'm talking about tempi to make a passer and queen a pawn.  Moving your minority against a majority gives the opponent free tempi.

yeah if you can't blockade and win the remaining isolated passed pawn, you'll get in trouble :P