"The bullet is a mad thing; only the bayonet knows what it is about"
Assault is generally more complicated than shooting. It's more risky than shooting, but can win a lot of games.
Why is assault more complicated? Well, it involves a lot more numbers. When you shoot, you only have 4 numbers to worry about: your ballistic skill, your weapon's strength, your opponent's toughness, and their save.
In assault, you have to take both your weapon skill, and your opponent's weapon skill into account. You have to consider strength, toughness, and saves, just like shooting, but also initiative.
You also have to consider the placement of your troops much more carefully. It is one thing to line up a bunch of shooters so that they can see their targets and fire. Even placing heavy weapon options is pretty easy. With assault, you need to consider how many of your models will get to fight. You need to ensure that the models that you paid for weapon upgrades for get a chance to use them.
There's more to pay attention to, so there's more room to make mistakes. Or to exploit opponent's mistakes. If you understand how to maximize the returns on your assaults, you'll have a large advantage over someone who doesn't.
So, you have to decide if a charge is in your best interest. And you need to do this in your movement phase, because that's going to impact how you move.
First, is there an opponent in your charge range? You have six inches of charge room, plus your normal movement to get there. The worst thing is to move your troops in preparation for a charge, only to find that they won't make it, leaving them out in the open. So, estimate well. Difficult terrain can really mess with these estimates, so only assault through terrain if you have no way to move around it. If your opponent isn't chargeable, obviously you will need to do something else.
Second, decide if it is in your best interest to launch an assault. There are a lot of factors that go into this. Most importantly, which side is likely to win the first round?
Who wins a fight
"Fortune favors the brave."
To calculate this, you want to compare the number of casualties that you expect to inflict with the number of casualties that you expect to take in return. While the math for doing this is similar to the math for figuring casualties by shooting, you have to pay attention to initiative. For example (80 points): 10 eldar guardians (s3 t3 ws3 i4 Sv5+) launch an assault against 8 Tau firewarriors (s3 t3 ws2 i2 Sv4+). The eldar have a higher initiative, and so strike first. They all get a bonus attack for charging, so 20 attacks, hitting 2/3rds of the time, wounding 1/2 the time, and the Tau fail 1/2 their saves, means that 3.33 Tau are likely to die. That means that only 4.66 Tau are going to strike back. the Tau will hit 1/2 the time, wound 1/2 the time, and 2/3rds of the eldar will fail their saves. .77 eldar are likely to die as a result.
So, the eldar expect to win the fight rather handily.
But, that's not the only consideration. You also want to make sure that assaulting your opponent is the most advantageous thing you could do. Starting an assault creates an area where line-of-sight is blocked for both armies. Starting a fight in front of your heavy fire unit might make them useless for a turn or more. When you assult, you cannot Rapid Fire that round. Are you going to inflict more damage by assaulting or by shooting? Are you going to take more damage in the assault, or if you let your opponent shoot you the next turn?
Example 2: 7 sisters of battle run into the same 8 Tau firewarriors. If they assault, they get to swing first, and can be expect to kill (14 * 2/3 * 1/2 * 1/2 =) 2.33 Firewarriors. The remaining 5.66 Tau will swing back killing (5.66 * 1/2 * 1/2 * 1/3) .47 sisters. BUT, if the sisters shoot instead of assaulting what would happen? 7 sisters rapid-firing kill (14 shots * 2/3 * 2/3 * 1/2) 3.11 Tau - more than if they assaulted. But, because the Tau are not tied up in combat, they get to shoot back on their turn. The 4.88 Tau expect to return fire for (2 * 4.88 * 1/2 * 5/6 * 1/3) 1.35 kills. Assaulting trades 23 points of Tau for 5 points of sisters, a difference of 18 points, while shooting trades 31 points of Tau for 15 points of sisters, a difference of 16 points.
From this, we can see that, although the Sisters are expected to kill more Tau by firing, they expect to take more casualties in return as well. As such, they should probably forgo Rapid firing and charge.
The same is not true for the Tau. If the Tau have the first turn, and they assault, their lower initiative means that they go second, but the sisters have only half as many attacks as they didn't get the charge bonus. Only 1.16 Tau dies in the charge, leaving 6.84 to make 13.68 attacks. They would kill 1.14 sisters. The combat would most likely end in a tie, with 11 points of Tau dead compared to 12 points of sisters. If the Tau fired instead, they'd kill 2.22 sisters (24 points), but the remaining sisters would then return fire and kill 2.12 Tau (21 points). Their lower initiative and Weapon Skill makes assaulting a poor choice, even when they could get the bonus attacks.
But wait, there is more to consider. One reason to get into assault is to protect yourself from shooting. As we discussed before, no unit should be operating alone (without good reason), so the question is not simply what will happen if the unit you're considering assaulting gets to shoot you, but rather, what happens when all the enemy units in range get to shoot you. If our Firewarriors have a three-to-one advantage on the sisters, then if they do not assault, they can expect to lose the entire unit to fire the next turn. This makes their decision a bit easier. Of course, sometimes, you are faced with a damned-if-you-do, damned-if-you-don't choice. Given the choice between assaulting the daemon prince or being assaulted by it next turn, both are likely to get your unit killed, so you simply want to maximize the damage you can do on the way out (which might be to ignore the prince and shoot at something else).
Ok, so we're making all these decisions about our assault, and we're still in the movement phase. And, lets say we have decided that it is in our best interest to assault. What do we need to do next? Well, first is to move in such a way that when you charge, you will have the right models in the places you want them. For some units, this isn't a big factor, but units that have a leader with a different weapon will want to consider where they place that leader. You want your best weapons near the front, so that they're included in the two inch zone of models who get to make attacks. If you have an independant character along, they need to be in base-to-base contact to get any attacks at all.
You also want to know what the goal of the assault is. Sure, you want to kill enemy models, but, do you want to have the best odds to kill the entire unit the turn you charged, or do you want to get into assault to protect yourself from enemy fire? If there are a lot of other enemy units in the vicinity, wiping out the squad may not be the best choice, as your unit will be targetted by massed gunfire the next turn.
If you want to kill the enemy unit completely, you want to use your movement phase to get as many models within six inches as possible, so that you can register as many attacks as possible. If, on the other hand, you want to make sure that the enemy is still, so you can consolidate into them and remain locked-up during your opponenet's turn, you might want to make it so that most of your models are further away than six inches, with only one or two who can get into base-to-base contact. Those two will probably not wipe out the entire squad, and the rest of your troops can pile in after the first round. The two who get sent ahead, however, may become casualties, as they will be the targets of all the enemies attacks.
You have to make moves like this during your movement phase. Once it is assault time, you cannot hold some models back, they all have to charge forward. The trick is to set them up so that only some will make it into contact with their charge move.
The next question is, do you want to shoot before you charge. Shooting can help soften up a target before the assault takes place, but, if you kill too many models against a smart player, they will lose the models closest to you, increasing the distance between your models and theirs. If you had a five inch gap to assault through, and you kill three models, that might become a seven inch space, and you'll find you cannot assault, and will be shot on your opponent's turn. For that reason, if it is at all questionable that you will actually make it into assault, it is probably wiser to sacrifice the turn of shooting for the safety of knowing you will be able to assault.
The ideal situation for all assault units is to kill all but one enemy model on the charge, and consolidate into that enemy. Killing that enemy during the assault phase on the opponent's turn, so that on your turn, you are free to move and assault again. This is unlikely to happen, as leaving one model behind is likely to see that model turn and run. However, if you can manage it, you want to get the odds in your favor as much as possible for ending all combats on your opponent's turn by maneuvering your models prior to their assault charge. Another benefit to this is that on your opponent's turn, the melee will block their line of sight, while on your turn the combat will have been resolved, and your other units will be free to fire through. As removing yourself from combat on your opponent's turn is the ideal, any unit with the "Hit and Run" ability (that allows them to leave combat whenever they want) is a good unit to have around, as they can lock an enemy in place long enough for you to move fire support into position, and then leave combat, letting the fire support finish the unit off.
Again, having some idea of what will happen when you launch an assault is the best way to plan your other actions around it. Just throwing your guys at your opponent's guys is likely to end up with a lot of dead models and no real advantage gained. Planning for what is likely to happen at each stage of the turn is likely to leave you in a superior position to take advantage of any gains you might make.
Optimizing your attacks
So the brave warriors have assaulted, and are now fighting a melee. Well, there's still more decision making to do to optimize the odds that you win the combat.
Most importantly is casualty selection. You get to choose your own casualties, with only two rules governing who can or cannot be picked. They have to be within the two-inch radius considered to be engaged in the combat, and they have to be picked so that the unit maintains coherency, if possible. Other than that, it's up to you who dies. And, this can be important, especially when dealing with combats with multiple initiative levels.
This is the simplest, and most obvious tactic behind casualty removal. You remove the models that will least-impact the number of attacks you get in return. If one model contributes his own attacks, while another contributes her own attacks, as well as including two other models within 2 inches, the one who only contributes his own attacks is the obvious choice.
Preparing to fallback
The first thing you want to decide is whether or not you think you will fail your leadership test, should you have to make one. Say you had ten guys, and three died because your opponent had better initiative. You're probably going to lose the combat. You're outnumbered, but not by much, so your troops who are normally Ld8 will be testing against a Ld7. So, they're only slightly more likely to pass their leadership test than not (58%). If you pass the test, you will all pile in and fight again next turn. But, if you lose the test, you're going to have to try and get away, and with the odds in your opponent's favor (because they have higher initiative), you're probably going to lose the squad.
But wait, you could minimize this effect. If, when it comes time to fail the leadership check, you have no one in base-to-base contact with your opponent, they cannot just massacre you. You get to escape 2d6 inches, and they get to consolidate 3 inches. Unless your fall-back move is extremely poor, your squad will survive. They might even rally the next turn, and shoot those guys who assaulted them!
Well, if only three of your models are in base-to-base contact with your opponent's models, you can choose those three as your casualties, and enable yourself to escape instead of being cut down, should you fail the leadership test. In fact, if you do this, you will want to fail the leadership test, as it will allow your other units to shoot the assault squad on your turn.
There is a tradeoff here. By removing all your models who are in base-to-base contact, you will not get any return attacks in the assault. So, doing this is best attempted when you weren't going to get a lot of successful attacks anyway. This can also be employed if you struck first, but were hit back harder than you expected. For instance, if you had initiative, and killed 2 of your opponents, but then they killed three of your models, you lost combat and will have to take the leadership test. In this case, other than special upgrades that you might not want to lose, there is no reason not to take your casualties from among the models in base-to-base contact with your opponent. That way, in case you fail your morale check, you can prepare to fall-back, instead of having to avoid being cut down.
Wasting your opponent's attacks
Sometimes, your opponent will have a weapon that strikes last (a powerfist, for instance). Now, if that weapon is wielded by a member of the unit, then it can strike anyone close enough to that model who is engaged. But, if that powerfist is wielded by an independent character, it can only affect models engaged with the independent character. If you have two models in base-to-base contact with the character, then when your opponenet's squad inflicts two wounds on you, you can remove the models in contact with the character. When it becomes the character's turn to fight, he is left unengaged, and his attacks are wasted.
Pouring more models into a losing battle
Sometimes, you're going to face a situation where a combat is expected to end on your turn. And, that's not good, because it means your opponent gets the first chance to move and shoot and assault again. If they are going to be able to assault your other unit anyway, you might as well pile that unit into the assault and prolong the battle, as well as getting the bonus assault attacks. This will buy more time for your other other units to get in position, as well as possibly turning the tide in the melee with the addition of more attacks.
Pouring more models into a winning battle
By the same logic, if you're winning an assault, but slowly, that's locking your unit up, and preventing them from doing anything else. They cannot move to grab objectives or address your opponent's other actions as long as they're stuck in the fight. So, take another unit, and charge them in too, to speed things up a bit.
One possible strategy is to take a small group of fast assault troops and hurl them at the enemy, causing a melee. Now that there is a fight to block line-of-sight, run a second, slower but larger assault force into the combat to finish it, before letting the fast group zoom off to rinse and repeat.