Biel Corona Round Two

PLAYED a Swiss FM this afternoon in round two of the Biel Special Corona Amateur Tournament.
Daniel Borner (FM 2154) - Spanton (1855)
Sicilian Hyper-Accelerated Dragon
1.e4 c5 2.Nf3 g6 3.d4 cxd4 4.Qxd4 Nf6 5.Nc3
The main move is 5.e5 but the text is also popular with grandmasters.
5...Nc6 6.Qa4 d6
Ponomariov has played 6...Bg7!? but 7.e5 Ng8 (7...Ng4?? appears in ChessBase's 2020 Mega database) but 8.Bf4 seems to give White easy play.
7.e5
This is the energetic point of White's play.
7...dxe5
Now 7...Ng4!? is possible although Black's queen is a target after 8.exd6 Qxd6.
8.Nxe5 Bd7 9.Nxd7 Qxd7
White has won the bishop-pair on an open board. The only flies in his ointment are that Black is a smidgen ahead on development and has the sole central pawn.
10.Be3 Bg7 11.Rad1
I had a quick draw in a 2005 correspondence game in which White, who had a rating of 2226, played the less-direct 11.Be2.
11...Qg4 12.Qb5

How should Black proceed?

*****
*****
*****
*****
*****
12...Qb4?!
I rejected 12...0-0!? because of 13.Qxb7 Nb4 14.Be2 Nxc2+ 15.Kd2 Qf5 16.Bd3, but Stockfish11 and Komodo11.01 continue 16...Qe5 17.Bxc2 Rab8 18.Qxa7 Rxb2 19.Qa3 Rfb8 with play that they reckon is worth the piece Black is down. In this line if 17.Kxc2 then the engines give 17...Rab8 18.Qxa7 Nd5 with full compensation for Black.
13.Qxb4 Nxb4 14.Bb5+ Nc6
White must be better - he still has the bishop-pair and Black no longer has an edge in development.
15.h3
Presumably played to prevent ...Ng4 followed by ...Bxc3, but after, for example 15.0-0 Ng4 16.Bc5 Bxc6, the engines reckon White is much better with 17.Bxc6+ bxc6 18.bxc3. Certainly White has the better minor piece, and Black has problems getting his rooks into play. However, White has the advantage after the text anyway.
15...0-0 16.0-0 Rfc8 17.Be2!?
This may be over elaborate - the engines reckon White keeps a pleasant edge with 17.Rd2.
17...Ne5
More-or-less equalising is 17...Ne8!?, according to the engines.
18.Rd4
Preventing ...Nc4, but this was not much of a threat, according to the engines, who reckon a line such as 18.Rfe1 Nc4 19.Bc1 is better for White, who is clearly more coordinated.
18...Nc6?!
Probably better is the engines' 18...Ne8, eg 19.Rfd1 Nd6.
19.Rd2
White has gained a tempo on the position after 17.Be2.
19...Rd8?!
The black position is difficult but this belated challenge for the open file seems faulty. The engines reckon White's advantage is smaller after 19...Ne8.
20.Rfd1 Rxd2 21.Rxd2 Kf8 22.Bf3 Ne8?!
The engines marginally prefer 22...Rb8, which only goes to show how bad the black position is. The problem with the text, even though Komodo11.01 likes it, is that it makes White's play easy.
23.Rd7 Nd6
What else? 23...Ne5 24.Rxb7 Nxf3+ 25.gxf3 Bxc3 26.bxc3 gives White six isolated pawns, but with what seems a fairly straightforward plan of using his rook and bishop to advance his queenside passer(s). I rejected 23...Bxc3 because of 24.Rxb7, although the engines reckon 24.bxc3 and 24.Bxc6 are even stronger.
24.Bxc6 bxc6 25.Rxa7 Rb8!?
Hoping for practical chances. Best, according to the engines, is 25...Rxa7 26.Bxa7 Bxc3, but I found it hard to believe an FM would not win that (sure, his rating is now well-below 2300 but that is presumably because 'old age' - he was born in 1963 - is blunting his tactical skills, not his positional endgame abilities).
26.Nd1 Nc4 27.Bc5 Bf6 28.b3 Rd8 29.Ne3 Nxe3 30.Bxe3 Kg7 31.Kf1
White is 'only' a pawn up but his simple plan of pushing the a pawn, while Black has few prospects of activating his kingside majority in any meaningful way, gives White an advantage worth a minor piece (Komodo11.01) or almost a rook (Stockfish11).
The game finished:
31...e5 32.Ke2 e4 33.a4 Rd5 34.Rc7 Rd6 35.Bc5 Re6 36.b4 Be5 37.Rc8 f5 38.a5 f4 39.a6 f3+ 40.gxf3 exf3+ 41.Kxf3 1-0