Bishop Pair v Pawn Structure

OFTEN a player has the chance to make the capture BxN, giving up the bishop-pair but damaging the opponent's pawn-structure.
It can take fine judgement as to which feature is better, and sometimes it may be just a matter of taste.
I used to generally value pawn-structure highly, which is why I almost invariably played, when I had the opportunity, the Exchange Variation of the Ruy Lopez (1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6).
My score with it is +37=18-23, which works out at a healthy 59 percent.
These days I am at least as likely to retreat the bishop, ie to play 4.Ba4, although I still believe the Ruy Exchange is a strong weapon against players graded under 150ECF (1825Fide) - it's something to do with needing a certain level of chess sophistication to properly use the bishop-pair.
Conversely, when facing the Trompowsky (1.d4 Nf6 2.Bg5), I have usually been happy to play 2...d5, and so allow White to double my pawns with 3.Bxf6.
My score here has been +1=5-3, for a rather miserable-looking 39 per cent. Maybe I do not have the necessary sophistication for handling the bishops.
Actually, the picture is not as bleak as the small number of games would suggest. I was outrated in eight of the games - my sole win coming against an ungraded opponent.
I am writing about this now because the issue of bishop-pair v pawn-structure came up in rather interesting ways in two of the team games I played this month.

Spanton (163ECF) - Robin Wilson (156ECF), Kent League
1.c4 f5 2.Nc3 Nf6 3.Nf3 e6 4.d3 Bb4 5.Qc2
Played not so much as to be able to recapture on c3 with the queen, but rather to attack Black's Dutch set-up with the pawn-thrust e4.
5...0-0 6.e4
This may be a new move. Previous games saw 6.g3 and 6.a3.
6...fxe4 7.dxe4 Bxc3+!? 8.bxc3
Generally speaking, the player with the bishops should open diagonals to try to gain a significant middlegame advantage, while the opponent hopes to exploit his better pawn-structure in an endgame. In this instance, perhaps surprisingly, Black's position is already critical.

Position after 8.bxc3. Black has to tread very carefully

8...b6?
This natural-looking move is an instant loser.
Black had to play 8...d6, although White has the initiative after 9.e5 dxe5 10.Ba3 Re8 11.Nxe5.
The game saw:
9.e5 Nh5
9...Ne8 10.Bg5 or 9...Ng4 10.Qe4 is even worse.
10.Bd3 h6 11.Bh7+ Kh8 12.Qe4
RW said he probably would have resigned after 12.Bxh6, but during the game I did not trust the sac (and anyway the text wins a piece).
My suspicions were correct. After 12...gxh6, the obvious 13.Qg6 loses to 13...Qe7! 14.Qxh6 Qg7! (Stockfish8).
After the text, the game continued:
12...Nc6 13.Bg6 Nf6 14.exf6 (1-0, 26 moves).

Rafal Walczak (175ECF) - Spanton (163 ECF), Central London League
1.e4 e5 2.Nf3 Nc6 3.c3 d5 4.Bb5
4.Qa4 is much more popular, and has been played by Nakamura and Morozevich. The text was a favourite of Harrwitiz, and has been played by Yusopov.
4...dxe4 5.Nxe5 Qd5 6.Qa4 Nge7  7.Nxc6!?
The mainline runs 7.f4 exf3 8.Nxf3.
7...Nxc6 8.Bxc6+ Qxc6 9.Qxc6+ bxc6
Black has three isolated pawns, and two of them are doubled. But White will have difficulty developing his queenside, let alone targeting Black's weaknesses.

Position after 9...bxc6. White has a difficult positional decision

10.f3?!
RW chooses a radical solution to his problem of development.
10.0-0 was played, unsuccessfully, against Steinitz in 1867.
Stockfish8 suggests 10.b3, but prefers Black.
10...exf3 11.gxf3 Bd6 12.d4 Be6 13.Be3 Rb8 14.b3 f5?!
My analysis engines are happy with this move, but I wonder if it is where I start to go wrong. The move helps restrict White's bishop, but it allows White to partly block the kingside, and leaves a hole at e5 for the knight.
15.f4 Kf7 16.Kf2 Rhe8 17.Nd2 g5?!
A "clever" move, but probably not very good.
18.Nf3 gxf4 19.Bd2 Rg8?
I continue to lose the thread. Stockfish suggests 19...Bd5 or 19...c5, but believes White has full compensation for the pawn, and it is hard to disagree.
20.c4 c5
Now 21.d5 would have given White a substantial advantage, thanks to Black's restricted bishops and six(!) isolated pawns.
Instead, 21.Ne5+ was played, after which inaccuracies on both sides led to a winning position for White, but the game was eventually drawn in 45 moves.