Long Struggle

HERE is my round-five game today from the 'Paignton' premier.
Peter Kirby (1988/177) - Spanton (1881/168)
Queen's Gambit Declined Exchange Variation
1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.cxd5 exd5 5.Bg5 Be7 6.e3 0-0 7.Bd3 Nbd7 8.Nf3 Re8 9.0-0 c6 10.Qc2 Nf8

The main tabiya of the traditional QGD Exchange with Nf3

11.h3
This move occurs more than 2,200 times in ChessBase's 2019 Mega database, with White scoring a very creditable 62%. But the most common move, with more than 3,000 examples, is 11.Rab1, which scores 59%.
11...Ne4 12.Bxe7 Qxe7 13.Rab1 Bf5 14.Rfe1!?
This appears to be new. Normal is for White to get on with his Minority Attack by 14.b4.
14...Qf6 15.Ne5 Nd6 16.b4 a6 17.a4 Bxd3 18.Qxd3 Ng6 19.Nxg6 hxg6 20.Rb3 Nf5
With the not-so-subtle threat of ...Nxd4!
21.Reb1 Qg5 22.Qd1 Re6 23.Qg4 Qxg4 24.hxg4 Nd6 25.b5 axb5 26.axb5 Nc4 27.Rd1 Ra3?
I thought getting a pair of rooks off would help my defence, but instead I hand White the initiative. My main analysis engines, Stockfish10 and Komodo10, suggest connecting rooks with 27...Ree8, or fixing White's g4 pawn with 27...g5.
28.Rxa3 Nxa3 29.bxc6 bxc6 30.Ra1 Nc4 31.Ra8+ Kh7 32.Rf8
Also good was 32.g5.
32...f6 33.Rc8 Kh6 34.Rc7 f5 35.gxf5 gxf5 36.g3?!
Played to prevent ...f4, but it seems White could have done without it. The engines give 36.Na4, and if 36...f4, then 37.exf4, when Black has not gained anything.
36...Kg6?
I should have stuck with my first thought of 36...g5, when White is only a little better.
37.Ne2 Kh6
Black loses the g7 pawn after 37...Kf6? 38.Nf4 Rd6 39.Nh5.
38.Nf4 Rf6 39.Nd3 g5 40.Kf1 Kg6?
Better was 40...g4 as the text allows White to force a protected passed pawn.
41.Ke2?
White still has the upper hand after this, but winning is 41.Ne5+ Nxe5 42.dxe5 Re6 43.f4, eg 43...Kh5 (43...g4? is hopeless as the Black king cannot get across to defend c6, while if Black tries to counterattack with 43...gxf4 44.gxf4 Kh5 45.Ke2 Kg4, White has 46.Rd7 Re8 47.Rg7+ Kh4 48.Rc7 Re6 49.Rf7 Kg4 50.Rf6, winning the c pawn) 44.Ke2 Rg6 (44...Kg4? 45.Rg7 Kxg3 46.Rxg5+ and 46.Rxf5) 45.Kf3 Re6 46.Rd7, with lines similar to those after 43...gxf4.
41...Re6 42.Nb4 Na5 43.Nd3
White gains a pawn after 43.Ra7 Nc4 44.Ra6 Kf6 45.Rxc6 Rxc6 46.Nxc6, but the engines reckon White's winning chances are diminished, presumably because of the difficulties White will have in creating a passed pawn.
43...Nc4 44.Ra7 Kf6 45.Ra6 Ke7 46.Nc5 Rd6 47.Ra7+ Kf6 48.Rh7 Kg6 49.Rd7?
Tantamount to offering a draw. The engines give 49.Rc7, and if 49...Kf6, then 50.Nb7 Re6 51.Nd8, when White wins the c pawn with his knight. This is better than allowing an exchange of rooks, I guess, because with the extra material Black will have problems defending his newly isolated d pawn.
Instead of 49...Kf6, the engines suggest 49...Kh5!? 50.Nd3 g4 (ensuring that if an exchange leads to the d4 pawn becoming a passer on e5, it cannot be protected by White playing f4), and now 51.Nb4 Na5 52.Ra7 Nc4 53.Ra6 sees White win a pawn in a similar way to the line given at move 43. Black may have decent drawing chances, but White certainly has more winning chances than after the text.
49...Rxd7 50.Nxd7 Kf7 51.Nb8 Na5 52.Kd3 Ke6 53.f4?!
White can get away with this, it seems, but giving himself a weak g pawn like this is not advisable.
53...g4 54.Na6 Nc4 55.Nb4 Nd6! 56.Na6!
White has problems after 56.Nxc6 Ne4 etc.
56...Nc4
Or 56...Ne4 57.Nc5+ Nxc5+ 58.dxc5 Ke7! (only move), and neither side can make progress.
57.Nc5+ ½–½