That question is one that many famous chess authors have grappled with.
It is not reasonable to ask people in the forum to give you an answer to such a question.
/ https://www.chess.com/article/view/storm-your-opponents-chess-castle
https://www.chess.com/article/view/tals-attacking-recipe
I was unaware that even famous chess authors struggled to answer this question, so apologies for my unreasonable actions (Little did I know, that the answer was nowhere to be found even on the internet!)
I am happy that I have not asked a silly question but rather a good one... But what worries me is that answers to such questions have not been found yet. It only makes me wonder, what else have we not discovered yet...
That sounded like what Kasparov said once but no, it really requires a lot of precise calculation in this case as it's an open position. For e.g. If the Q were at a5 instead of a4, it wouldn't have worked.
A general rule is if you have more pieces in a general area than your opponent, be on the lookout for sacrifices. There's also some imagination involved. I don't think I would've seen Bh6+ in your example, and even if I did, I would probably have thought the follow up was too slow.
The more you play, the more you know. I like to refrain from sacrificing, so my calculation abilities are not as strong as others with a similar rating, and my attacking intuition is similarly worse.
Nxg7 is a common enough sacrifice, but the real beauty of this is it's a double sacrifice (Bh6 immediatly after) and that typically you need 3 attackers, but in this case 2 is enough because g4 eventually allows both rooks to come into the attack. (After the 2nd sacrifice it seems white only has the queen and the bishop attacking)
As B1Z says, you usually count the number of attackers and defenders. In the starting position black only has a knight on the kingside, and white has a few pieces pointed in that direction. That lets you know a sacrifice is worth considering even if the details are very hard to get right.
That sounded like what Kasparov said once but no, it really requires a lot of precise calculation in this case as it's an open position. For e.g. If the Q were at a5 instead of a4, it wouldn't have worked.
Yeah, there are probably a dozen different very small changes that would make this attack fail, so it's really nice.
If the pawn is on d6, not only does the sacrifice not work, but black is winning no matter what white plays.
If the queen is almost anywhere else it doesn't work.
If the rook on f8 is almost anywhere else it doesn't work.
If the pawn isn't on f3 it doesn't work.
That question is one that many famous chess authors have grappled with.
It is not reasonable to ask people in the forum to give you an answer to such a question.
/ https://www.chess.com/article/view/storm-your-opponents-chess-castle
Ok, but there are some simple bits of advice:
In slower games, you remove pawns with pawn breaks to open lines (files, ranks, diagonals) for your pieces. The side that's better developed (or is more active in that area) benefits most (so it's possible to play a pawn break that helps your opponent).
Sacrifices are a lot less mysterious when you think of them in the same way. You're simply trying to open lines and have more piece activity in an area than your opponent. The main difference with a sacrifice is you're often on a timer i.e. if you can't reclaim your material investment in 10 moves or so, you're probably losing.
In an old interview, Kasparov said in his youth he had the habit of dividing the board down the middle (kingside and queenside) and counting the number of non-pawns on each side. As a rule of thumb, to hunt the king you need 2 more attackers than they have defenders.
The links you gave are good. Seeing common attacking patters really helps you spot them in your own games.
Ok, one last tip, and I'll stop spamming this heh.
A rook on f1 or f8 (if you've castled kingside) isn't much of a defender... in fact it often helps your opponent mate you because the king can't escape towards the center (and you can't use f1/f8 for a knight or bishop).
So when counting defenders, not only can you ignore that rook, but often, you can count it slightly in your favor 
dude ur 1100 rapid and 700 blitz if u honestly would see that move its because you sac too much
Is it still a sacrifice if you can calculate a path to victory from it? I prefer a sacrifice to be a leap of faith. Though more often than not, that's probably because I don't calculate deep enough.
A sacrifice without calculating to a Checkmate or material advantage might be in order to expose and chase the King, which often leads to a win. Another example would be sacrificing a trapped piece in order to recover something in return instead of simply waiting for it to be captured. You might also want to make a sacrifice in order to develop your pieces or to gain a positional advantage. For the latter, you have to trade off the material disadvantage with what you perceive the positional advantage to be. That's the difficult problem and I think it comes down to intuition. You have to decide how much you like the look of the board before and after the exchange has been made. At the end nodes of their search trees Chess engines use an analogous heuristic.
Sometimes it helps to get inspiration from the legends and famous games. I'm making a video on this exact topic right now. https://www.youtube.com/watch?v=_oebmb79L7U
That sounded like what Kasparov said once but no, it really requires a lot of precise calculation in this case as it's an open position. For e.g. If the Q were at a5 instead of a4, it wouldn't have worked.
Yeah, there are probably a dozen different very small changes that would make this attack fail, so it's really nice.
If the pawn is on d6, not only does the sacrifice not work, but black is winning no matter what white plays.
If the queen is almost anywhere else it doesn't work.
If the rook on f8 is almost anywhere else it doesn't work.
If the pawn isn't on f3 it doesn't work.
May as well add that if the queen-side bishop had been fianchettoed then f5 is a viable defense after Qh6 because the f8 rook is defended.
Hi! So recently I had come across this beautiful position
I thought and thought and thought... Found MULTIPLE candidate moves (candidate moves = multiple moves that can possibly be played in a position). I finally decided on a straightforward candidate move, that is c4... But the answer I found was nowhere close to what was played in the game...
Nxg7!! - LIKE A BOLT FROM BLUE!! Never in a thousand years would I have even 'considered' playing such a move like that... I never thought such a sacrifice would come from a position like this. Later, the white pieces sacrifice another piece to completely DESTROY the black king.
I was still in shock at what I had seen when a sudden question popped up I'm my mind:
"How do I know, WHEN to sacrifice?"
The position given above just seemed like any other middlegame position... But there was a hidden sacrifice that the white player spotted. What if, I get a position like this, but do not even consider of the option of "Sacrifice"? Are there any criteria I need to look at to find out whether I can sacrifice in a position or not?
The sacrifice in the game given above was to mate the king... But there are other types of sacrifices as well, (ex: - positional sacrifice)
How do I make use of a concept/Strategic idea, when I do not know when it should be used?
I felt that this was a very important question and that I needed to ask some other people about their opinion on it... Thanks!!