How do you convert this +3.5 position? I ended up in a losing position.

I'm looking for an explanation after move 21, on how to convert it to a win. I knew the c pawn was weak, but I wasn't sure how to attack it. I was considering e4, but after fxe4, Qxe4, and Qe7, I didn't know how to attack the pawns. I ended up playing g4 which was a mistake. After that, due to a piece blunder, I was in a worse position. Luckily, he blundered his queen to a skewer.

I spent like two seconds looking at this but I would go qd2 and then a4 

what why the opponents dark squared bishop could become active and black is just gonna pile on the isolated b pawn

The worst thing you did is trade the knight for the bishop, that light square bishop is terrible because most of the pawns were on light square. Keep the knight and try to add pressure on multiple weakness, not one Pawn break only work when you threaten to take, the g pawn break does help literally nothing as taking the pawn is just weaken your king. You should break with the e pawn, if they take, they have a new weakness on e6 now they have multiple weakness, you don't always have to attack the c pawn. If they don't take, you take them and use another break, d5 open up the position, you have the bishop pair, a more active queen, just attack the king and you should win

I see. What if he plays kf7?

The reason why I traded the knight is because I thought that if I trade the light squared bishop then his pawns would be weaker since the light squared bishop couldn't protect them. At what move do you recommend to play e4?

22.Rac1 looks good, threatening 23.Nxb7 or even 23.Nxa6 Bxa6 then 24.Rxc6 in some variations. And 22...Bxc5 23.dxc5 is obviously bad for Black, opening up the diagonal for White's black squared Bishop.

Yes, that's a good human response. Not trading off the bad bishop until at least get a pawn for it.

Then since have software, could look at what it is calculating for it's +3.

You have your own answer in your post. 

You see " c6" is weak, then why not reroute kt to a5 via b3 to attack c6 weakness.  It just takes 2 moves and worth it.  ( Sometimes you have to reroute kt up to 4 moves). 

Otherwise, run Stockfish for about 10-15 secs and you will see why +3.

The obvious move is Rac1 and black is already in trouble. I think black has to take the N on c5. Rerouting the N doesn't seem correct because that allows Rfc7.

It doesn't look +3. In such a position, +3 is equivalent to a simple win, since it means the positional equivalent of a minor piece up, so why not +99, if you're going to over-ice the cake anyway?

Alright, I have seen and played through everyone's diagrams. So basically, my two big mistakes were taking the bishop and g4. Rather than taking the weak light squared bishop, I should have kept my knight and kept on building up my position. Then, I should have took. Rather than playing g4, giving black some ideas on the kingside, I should have played e4 and saw that the e6 pawn was weak. Then, black would have two weaknesses, c6 and e6. After playing g4, as MoonWarrior pointed out, I should have done fxg5 first. Plus, after g5, I did miss a few tactical opportunities. Next time, after calculating the moves, I will look at them in a more general way. I didn't see a straightforward way to attack the e6 pawn, so I didn't play e4. I should have just seen it was big weakness. I will also ALWAYS ask myself which piece is better, before trading.

Agreed actually I had this same situation in one of my games I’ll have to find it later, but I had a knight outpost and I could have taken his bishop but I choose to do a different move (which turned out to be much better)

This. 22. Rac1 followed (after Qe7) by g2-g3 protecting f4 and intending e3-e4. Looks really grim for Black.

Queenside play. a4 comes to mind. Also, refrain from taking on b7. Note that he can’t move his bishop because his a pawn hangs. Therefore 22. Rac1! Rc7 22.Nxb7 Rxb7 23.Rxc6 for example.

21.Nc5 with 22.Nxb7 doesn't look right. Sure an engine will make it work, but that's not a very logical or human way to play IMO.

For example something like this is easy:

-

-

Chess.com gives the final position something like +0.5

My PC engine starts to give close to +3 after a while...

By the way that's a common mistake of lower rated players... if you want to pressure c6 then don't put a knight on c5 which blocks your pieces. I understand your idea was to capture the bishop, but trading a good knight for a bad bishop just to win a pawn... sure that's an idea, but usually a last resort. You're letting black off too easy because you can instead build up your position... which is another common mistake i.e. patience.

So think in terms of pieces blocking other pieces, think in terms of which minor pieces are most vs least mobile (rank them for both players), and when your opponent is passive be patient and think in terms of building up your position.

And looking at how the game went I see another common error... you traded off all your active pieces and went into a pseudo endgame with your b2 bishop. That's simply not ok.

22.Rc1 targeting the backward c pawn seems ok. 

Why did you trade off your good knight for a completely passive bishop?

Something i was taught, is to set up the position with just the pawn structure, and look for any weak pawns or squares.  You can then try and develop a game plan on what to attack, and where to play.