An alapin game, or, How to squander the initiative (tips for playing these positions?)

This is a correspondence time-controls game I played with the white pieces. Evidently, my calculation was a problem. But I would love to have some tips in general for how to play these positions (two pawns gambited for clear developmental compensation). I put some of my thoughts in the study.

Stockfish shows me that my opponent played very well; I did not.

@the_breadler #1
GM Tiviakov plays Sicilian Alapin nicely. I'm sure there's annotated games or videos for it, that you can find.
Here a very fresh game, Tiviakov - van Wely, 2018.

You exchanged 2 pawns for tempi so you had to keep Black's underdeveloped and/or make threats so that he'd be forced to make concessions (e.g. give back material, accept permanent weakenesses).

16. Bxd4 was terrible. 16. Nxd4 and you have some ideas (N[x]e6, Nf5-g7)...

The sequence 17. Be5 Qd8 18. Rfd1 e6 gives you nothing and he made an useful move. Your advantage has gone. A possibility was 17. Ne5 e6 18. Rfe1 threatening Nxf7 (Or 18. f4!?) which seems to be at least equal.

He gave you a chance with 19... h6 but you didn't play 20. Bb2 to avoid his 20... Ba3.

Thanks a lot. These comments are helpful.

Agree with @okgo.

16 Nxd4 was the right move. A knight becomes much stronger in the centre. A bishop is strong further down the diagonal i.e. on b2 as well.

On 16 Nxd4 Qc7 17 Rfe1 you have terrible threats.

The same indeed 17 Ne5 centralising the knight or 17 Rfe1 bringing your rook into play on an open file is much stronger than 17 Bb2 centralising the bishop. The bishop is fine on b2. The rook on f1 wants to go to an open file. The knight on f3 prefers to be in the centre.