I'm not so sure how complex or accurate you want this system to be in the end.
As far as for what you're trying to accomplish, I've already seen a system that tries to take all sorts of ammunition & attack types into account.
After having read through the rule book (It was designed to be plugged into about twelve different RPG systems), I came to the conclusion that it was far too complex to use.
In about any tabletop RPG, combat is the absolute longest portion of play. - It's difficult to justify anything that makes combat even more complex.
---
That said, if you're wanting to gauge the approximate defensive value of a given material - the best I can think of is attempting to use specific density.
If you can find out how many pounds of force it takes to break/penetrate a given material, you can compute from muzzle velocity and bullet weight an approximate force applied to the target.
Of course, that would calculate for you how much material would be broken by putting the muzzle of the given firearm up against the material in question.
In reality, the velocity of the bullet will decrease /drastically/ as it travels further along its path due to friction from air resistance.
Unfortunately, ballistic curves are rather difficult to calculate without a decent calculus background (which I don't have), but it can be approximated using "drag coefficient". You do the math, and scale back the answer by a set percentage.
If it /doesn't/ need to be particularly accurate, I would take advantage YouTube and the large volume of people with the free time to shoot random things in their back yards. (If you want a /slightly/ more scientific approach, one of the better ones is the "box o' truth") - You would then see what a bullet penetrates and want it won't.
---
All in all though, the more I tried to make house rules to complicate d20 systems, I've noticed that the creators have spent a lot of time trying to boil down real word physics to something that can be approximated with relative accuracy and a decent amount of speed. - I wouldn't re-invent the wheel if I could avoid it. - Sometimes making minor tweaks or correcting minor oversights to a given rule are /much/ more rewarding than trying to come up with a new system on the fly.
As an example:
I tend to lean heavily towards d20/Pathfinder type systems.
1) I would simply have the attacker and the defender /both/ make opposed rolls. - You could stack on all of the armor you want, but if you roll a natural 1 on your defense roll or the attacker rolls a natural 20 on his attack roll, you're taking a hit. - Text-book accuracy suffers, but realism benefits. Any armor that allows for movement will have a chink in it somewhere.
2) I'd also simplify soft armor vs hard armor. Use hardness to deflect some of the incoming damage, and have the soft armor convert it from 'lethal' damage to 'non-lethal' damage. (Instead of suffering from a punctured lung you suffer from a cracked rib instead.)
You can then have the 'stacking' of armor be a function of how much damage it can deflect/absorb vs how much force it can take from a given weapon before it breaks. - It's /much/ easier to think in those terms on the fly than it is to run through a complex equation.
(EDIT) I would also look at making sure that no body armor /perfectly/ deflects all damage types. - In the middle ages, chainmail was great versus slashing weapons but tended to convert a sword strike into a bone-jarring bruise. - Full-plate armor was good all around, except that being struck with large hammers tended to transfer most of the force to the wearer- even if the armor didn't take a dent
If the armor nearly negates one form of damage, it should convert a significant portion of that damage to something else. - Nothing should be perfect.
3) I would lastly, take a hard look at cover/concealment rules. - If realism is the goal, most of the focus should be on minimizing opportunities to be shot at while maximizing opportunities to shoot at the other guy. - Armor stacking should be a secondary concern.
---
I hope that the my take on the topic gives you some ideas to help successfully create your new system.
I hope you don't take my minimal effort in 'dissuading' you from this path as an insult. - I've simply tried a fair amount of mucking about with realism vs text-book calculation in roleplaying games, and tend to always circle back to square one.
Your mileage may vary.
Good luck.
1) I disagree on the defender being powerless, but do concede that the defender has a built in (initial) disadvantage. - Your defender doesn't have to dodge the bullet, your defender only has to dodge the attacker's point of aim. - You would be surprised to look up security camera footage from various shootings that have occurred in enclosed spaces.
Sometimes a hail of bullets can be exchanged back and forth while neither party manages to hurt the other, or bystanders. (Of course, the exact opposite does happen often enough, sadly.)
I see it more as a stealth vs perception issue followed by an initiative vs technique issue.
If your defender isn't aware, the attacker takes the shot, given. If your defender is aware, I think that the attacker's ability to aim should be tested against the attacker's ability to dodge, at least to some limited degree.
2) Kind of like the hit location being a factor, but have tended to want to avoid tracking this except for critical hits/threats or deliberate called shots.
A possible compromise to this, that seemed to work well when I had done it, is to declare point of aim prior to a combat roll, such as "I'm going for the head", and have one number on the hit location die be the target the attacker was going for with the others being typically random, if the shot warranted tracking.
3) I like this quite a bit. I have tended to have my NPC's take 10 on certain checks as a simplification for judging outcomes, and to keep particularly powerful NPC's from neutering the PC's completely.
---
In order to parse this out, I propose a tactical example.
You have a family of defenders in a rural, two story farm house. You have a squad of attackers approaching from the cover/concealment of a forest. In between the house and the forest is a clear-cut yard, just large enough to prevent it from being covered in a quick dash.
In reality, there is a good chance that /neither/ side notices the other, at least not for a while.
How do you address something like this? Does it immediately become a battle of attrition as soon as someone makes the first mistake?
If someone on the attacking side gets bored and exposes themselves a defender can shoot, risking return fire from the remaining attackers. If the defender gets careless and silhouttes themselves in a window, the attackers risk return fire from the defenders when they give their position away.
Now it becomes about who can take advantage of the cover better. Who can expose themselves the least while getting the best shot. Who can react better, despite the attacker having the initial advantage?
Without introducing antipersonnel mines to the defenders and tear gas/flash bangs to the attackers, and assuming both have decent supplies of food and water, neither side has a perfectly clear advantage on the other in a tactical sense.
Here the attack/defense relationship becomes a little clearer. Without some kind of opposed check, this situation will (likely) result in an automatic stalemate.
---
In other words, who makes the first critical mistake?
I have a couple thoughts in mind on how to work on the mechanic, or maybe look at the how to resolve stalemate, but... It's late at the moent and not thinking of much in terms of elgant solutions. - Will get back to you on that later.
A couple of things to bring up, but not a really large post this time around either.
On the hit location front: It might be worthwhile to look up an old system called FUDGE. It had a pretty decent way of handling wounds that had overall impact on performance.
It's probably not exactly what you're looking for, but it might an overall look at the mechanics might give you a couple of ideas.
It isn't unlike the vitality/wounds variants of hitpoint for d20/Pathfinder, but carried it a bit further without being too overly complex.
-
As far as attacker vs defender, I think the "Army of Two" video game may have an insight, as far as the concept of 'Aggro' is concerned.
If it were d20, you could do away with the opposed rolls but throw circumstances bonuses or negatives out as Red Force and Blue Force engage each other. - Each person only has so much attention to go around, and the less attention someone has the less effective they are going to be at defending against attacks.
In that way, it isn't so much Hide and Seek (or Perception vs Stealth) as it would become Focus vs Distraction. - Not quite the 'Aggro' system that that particular video game has, but something that mike the be able to make use of the stealth mechanic that your game proposes, and bring it forward into situations where there is not stealth, but there definatley is a lot of "noise" vs "silence". You can make the question of who hits who, in part, a matter of who is able to focus on the task at hand better.
You were kind of already there in the last post in terms of opposed tactics checks, but figured I would put this out as a possible consideration.
It might make more sense as opposed to the typical "armor class" system, and help give a static (though varying) bar to the typical attack, always assuming that someone is trying to make use of what's available as cover and concealment when fighting.
---
Still not a very fleshed out thought, but... figured I would put it out there as an idea.