Benko gambit (declined) - middlegame

This follows up on Thread 78052 about Game 4064642. Please feel free to comment on anything on the game. I'll start a post on the endgame, that is probably the most interesting part of this game.



15.h3 Whte plays it safely, denying g4 as a pivot point for black in a possible attack on the kingside.
15...Nd7 This move may have many purposes. Black might continue with an attack on the kingside (after all white just weakened his castle) through something like f5. At the same time, black can just be there luooking at what will happen on the queenside, and the knight can switch there if needed.
16.Nb5 White occupies his outpost. This is probably fine and playable, but there were other options to take into account. One was to occupy b5 with the bishop instead. The bishop in c4 is acting like a big fat pawn, and on b5 he would have had more open lines. It still seems to me that the knight is preferable. It gains a tempo on the queen, and the pawn in c5 is fixed not allowing black to try to reduce his crampiness. Antoher option was to seek counterplay in the center, via Re1.
16...Qb8 The queen has to move, but Qb7 looked a bit better because it would have left b8 for the rook.
17.Ra3 A problem with the knight not being in b2 anymore is that b2 needs to be defended. Qc2 looke a natural developing move here, protecting both e4 and b2, freeing the c1 bishop and going towards connecting the rooks. This manouver of the rook is instead a bit clumsy, although probably playable.
17...Nb6 Attacking the bishop and seeking exchanges. It might sound strange that black is ready to exchange his knight for white's poor bishop, but black hi cramped and exchanges favors him. And white's bishop can became strong in the future if white starts playing in the center an menage to play e5, d6 at some point.
18.Qe2 White is ready to replace the bishop with the queen, and develop a piece in the meantime.
18...Nxc4 19.Qxc4 Ba6 Pinning the knight to the queen.

20.Rb3 Overprotecting the knight and freeing the bishop in c1 from the defense of b2.
20...Qe8 We see now why it was better to put the queen directly in b7, now black has to spend an extra tempo to bring the rook to b8. 21.Qc2?! Poor move. It unpins the knight, but the knight needs protection in any case, and if properly prapered the advance c4 is dangerous for white. It was high time for white to finsh his development with Bd2, planning to go to c3 and challanging black's Indian bishop.
21... Rb8 Black immediatly points out how poor white's previous move was.
22.Qe2?! White wants more space for the queen, but alas the lady's position is indeed in c4, blocking possible advances of c5.
22...Qd7 Black already had the option to a big exchange in b5, leaving white with a weak doubled pawn in b5. This move is good too because it makes room for the rook in f8 to enter the fight on the queenside.
23.Bd2 Black is still better, but white is finally getting his act together and mobalizes his pieces.
23...Rfc8 Offering a pawn in exchange of his chance to play c4.
24.Qc4 I like this better than taking the pawn, although this is the cristal clear proof of how many tempi white wasted in the previous moves. His position changed only in the development of the bishop to d2, which he could have done immediatley, while black had the time to increase the pressure on b5 and bring the other rook to the attack.
24...Rb6? Now, this is probably just an oversight. White gets a pawn with tempo. Black was anyway sacrificing his b5 pawn, but for more than enough compensation after (24. ... Rb7 25. Bxa5 Rcb8). A tempo can make quite a difference. By no mean black is losing now, the position is very complicated and white will be hard pressed in holding to the extra pawn.
25.Bxa5 Rb7 The rook goeas where it should have gone in the first place.
26.Bc3 The bishop challenge the Indian bishop, reducing the weakeness on b2.
26...Rcb8 27.Bxg7 continuing on his plan, since nothing can be done to not lose a pawn in the exchanges.
27...Kxg7 Avoiding complication and retaking immediatley.

28.Ra1 The a file will open and white wants it.
28... Bxb5 29.axb5 Rxb5 The pawn has been won back.
30.Rxb5 ?! Here it was worth while to play Qc3, defending the rook with tempo, and maybe follow up with Ra3. Ceratianly, this is a very stubborn defense, but white is worse here and has to do with the needs of the position. Black is now able to double pieces on the b file. 30...Qxb5!? 31.Qc3+ The best shot for white, leaving the pieces on the board. The pawn structure favors black, who has a very nice big island, while white has a weak b pawn.
31...Kg8?! This is in principle wrong. Although there is still quite some material on the board black should have kept his king closer to the center, ready for the endgame. f6 was better.
32.Ra2?! Passive play will bring white nowhere. Black already threathen to invade white's space with Qe2. Ra7 was better.
32...Qb3 This is probably ok, but Qe7 looked better. Black give a chance to white to regains some initiative with Ra7.
33.Ra7 Qxc3 I expected Qd1+, but probably black felt confident about the fact that my king is far from the pawn base that would form in c4, and he would have enough active play with his rook to force a win in a rook and pawns endgame.
34. bxc3

Here we are, black to play in this rook and pawns endgame. It will take me a bit to analyze this. Please feel free to comment on anything in the game, I'm looking forward to improve.

Thanks!

J34

34. bxc3 Kg8f8
35. Ra7a3 Rb8b2
36. f4 Rb2e2
37. Ra3a4 Kf8g7
38. Kg1f1 Re2d2
39. Ra4a7 Kg7f6
40. e5 (pushing this pawn looks like the slip up that cost white the win...as it loses a pawn eventually even though it at first appears that the pawn is regained 40.g4 looks better)
40....dxe5
41. fxe5 Kf6xe5
42. Ra7xe7 (looks as though white regains the pawn) Ke5f6
43. d6 Rd2xd6 (but it's gone)
44. Re7c7 Rd6d5