An extensive combat analysis of ship hulls

Describe your experience with the latest version of FreeOrion to help us improve it.
Forum rules
Always mention the exact version of FreeOrion you are testing.

When reporting an issue regarding the AI, if possible provide the relevant AI log file and a save game file that demonstrates the issue.
Message
Author
User avatar
Kassiopeija
Dyson Forest
Posts: 212
Joined: Wed Jul 16, 2014 6:14 pm
Location: Black Forest

Re: An extensive combat analysis of ship hulls

#16 Post by Kassiopeija » Thu May 05, 2016 9:00 pm

Strangely, I commonly take the complete opposite strategy in shipdesign & do quite well with it:

All external slots are stuffed full of weapons, ignore any research into hitpoints-mods (which can be invested into hulls/shields instead), prioritize Reinforced Hull to give Robotics better survivability but switch fast to SelfGravi/Titanics ASAP.

Survivability comes from shields, which I prioritize throughout all game. The recent shieldnerf made it a bit harder, esp. since the upgrade from +3 to +5 isn't a big deal, but Gravitators have fairly enough hitpoints to still be invincible to anything the AI can muster.

I like playing a productive-heavy game so production costs aren't much of an issue, but research is always lacking and that's why I take this research path.

If you're ahead in technological development (and you really should against the AI) then there's not much point in exchanging weapons with hitpointmods. More firepower equates faster decimation of enemy forces. Once available even Robotic Hulls do get shields + 4*weapon mod. Expensive design but you pick only fights where you win and it'll repair fastly back. Later these will get decimated by planets on defense focus (=+40 defense) but at that stage you might as well get rid of them to keep fleet upkeep modifier down.

Generally, the strength of shields are dependant versus the strength of enemy firepower. So if you're going down that road you really have to prioritize them or you're playing suboptimally. It's possible to soak in most enemy firepower even with early hulls. Robotic Interface:Shields can be fun to play with if starlanes are sparse or you're making starlanes unpassably.

slv
Space Floater
Posts: 24
Joined: Thu Mar 17, 2016 12:12 pm

Re: An extensive combat analysis of ship hulls

#17 Post by slv » Fri May 06, 2016 4:30 am

Kassiopeija wrote:Strangely, I commonly take the complete opposite strategy in shipdesign & do quite well with it:

All external slots are stuffed full of weapons, ignore any research into hitpoints-mods (which can be invested into hulls/shields instead), prioritize Reinforced Hull to give Robotics better survivability but switch fast to SelfGravi/Titanics ASAP.

Survivability comes from shields, which I prioritize throughout all game. The recent shieldnerf made it a bit harder, esp. since the upgrade from +3 to +5 isn't a big deal, but Gravitators have fairly enough hitpoints to still be invincible to anything the AI can muster.

I like playing a productive-heavy game so production costs aren't much of an issue, but research is always lacking and that's why I take this research path.

If you're ahead in technological development (and you really should against the AI) then there's not much point in exchanging weapons with hitpointmods. More firepower equates faster decimation of enemy forces. Once available even Robotic Hulls do get shields + 4*weapon mod. Expensive design but you pick only fights where you win and it'll repair fastly back. Later these will get decimated by planets on defense focus (=+40 defense) but at that stage you might as well get rid of them to keep fleet upkeep modifier down.

Generally, the strength of shields are dependant versus the strength of enemy firepower. So if you're going down that road you really have to prioritize them or you're playing suboptimally. It's possible to soak in most enemy firepower even with early hulls. Robotic Interface:Shields can be fun to play with if starlanes are sparse or you're making starlanes unpassably.
As for research cost, the research price of Deflector Shields (300) is the same as of Diamond Armour. You have to research Zortrium for the later too, but 360 vs 300 RP difference is not that big and zortrium is useful in the process.

Let's say, you want to beat the typical AI fleet of 20 Robotic 2xLaser IV 2xZortrium ships (22 attack 47 structure per ship, 12 effective attack vs deflectors).

In a real game, as you said a player should have a technological advantage, i.e. having Plasma IV and Zortrium or Plasma IV and Deflectors. Assuming we are using self-gravitational hulls, with your approach you build
7x Plasma IV + Deflector ships with 108 attack and 100 structure for 350PP

Instead I will build
3x Plasma 3x Diamond ships with 54 attack and 154 structure for 204 PP

Now to have a tie against AI fleet you need 4 ships (everybody will die at the end of round 2) and to "win" need 5 ships. So it's 1400-1750 PP for you

I need 6 ships to have a tie (everybody dies at the end of round 2) and to win I need 7 ships. So it's 1220-1426 PP for me.

Unless I've messed up (I shouldn't have), then no-shield designs perform better than shielded ones.

Granted the difference isn't that big, and pretty sure you can still do fine against AI (in the same way you can be fine if you stop at Plasma III). It's still a bit less optimal. I know that I am a bit too nerdy on min-maxing and optimising, but wasn't I supposed to try to find best way?

Maybe we have a difference in a playstyle, but for me it seems that production points always matter, extra points go into new colonies, outposts, terraforms, more ships to have multifront war (I often play with many AIs so sometimes drag myself into numerous wars), etc.

Also note that no-shields designs were worse even given the technological advantage against the AI (deflector vs Laser). If there is actually another human player somewhere in the galaxy then deflector designs will be absolutely crushed by Diamond designs.

P.S. Choice of weapons doesn't matter here, with deathrays instead of plasmas you will get the same outcome, noshields>shields.

If we start comparing technological lead of Plasma shield against laser then you're right and shielded designs perform better, but I never managed to live that long, usually most resistance in the galaxy already became futile by then.

SkyCore
Space Floater
Posts: 23
Joined: Tue Apr 05, 2016 4:37 pm

Re: An extensive combat analysis of ship hulls

#18 Post by SkyCore » Fri May 06, 2016 9:48 am

You seemed to overlook the small asteroid(no armor required);
mass driver 4: 6*20/(60.5^2) = .0328
laser 4: 11*20/(83.8^2) .0313
plasma 4: 18*20/(107*107) = .0314
death ray 4: 30*20/(153^2) = .0256

Interesting, i just noticed the game version i have has different numbers than the ones you give.
slv wrote: Of course the production cost of the ship also contributes a lot to its combat efficiency, if your ship is twice cheaper than opponent's you can build two more ships, so your damage*structure will be four times greater.
You can only build ONE more ship, not TWO more as you stated.

slv
Space Floater
Posts: 24
Joined: Thu Mar 17, 2016 12:12 pm

Re: An extensive combat analysis of ship hulls

#19 Post by slv » Fri May 06, 2016 3:15 pm

SkyCore wrote: You can only build ONE more ship, not TWO more as you stated.
Whoops, you're right, I can build only one more ship, I was right tht damage*structure will be four times greater though.

Regarding small asteroids as you can see they have a lot lower efficiency so I decided to not include them.

As for numbers, I listed no-upkeep numbers, if you open the game and go to design page your ships have 4% upkeep. I listed the 0 upkeep prices (you can see them if you scrap all ships ad design on t2).

User avatar
Kassiopeija
Dyson Forest
Posts: 212
Joined: Wed Jul 16, 2014 6:14 pm
Location: Black Forest

Re: An extensive combat analysis of ship hulls

#20 Post by Kassiopeija » Fri May 06, 2016 3:59 pm

Well, couple of things.

I've the feeling that the defensive boost from Deflector-shield isn't a big deal. I mean you get a +3 initially, followed by +2, followed by +4. I've no clue how the design-intent is but before the shield-nerf these steps weren't so crude, but I just live with it that it's not so good & instead put more pressure to reach Plasma Shield ASAP.

This is even more important than any weapon because it means, if you can effectively soak in most -if not all- weaponsfire, then it doesn't matter anymore if you win a fight in 1, 2 or 3 combatrounds, ie. you could as well only fight with Laser-weapons.

Another thing is that by that time I build Gravitators I've, most likely, found a Good-Pilot species and fight with +25% or +50% weapon-damage - which, in my design, is increasing the strength of 6 modules while in your design works only on 3 modules. This is an additional reason why to go all-weapons.

If you go down the Gravi/Robotic-line you could also easily get +5 HP from Reinforced Hull.

The FleetCostUpkeep-multiplier should be also in my favour since you build/loose more ships.

One internal slot is still empty - (and a core slot but at that point of the game there's usually nothing worthwhile to it) - I usually fill that internal slot with an engine or a stealth-module.
(Self-Gravitating hull has a -5 penalty to stealth but at +35 stealth I can successfully "first-strike" enemy fleets which helps immensely against planets on defense focus (against which shields are not really effectively...)).
Thing is that adding additional parts to fully complete the design will downgrade your design more than mine because, again, you loose more ships^^

Both speed & stealth do fit in my preferred playstyle & offer benefits I find better than existing alternatives - for example I don't bother with Stargates (it's expensive to research it and tiresome to fiddle around with the slider as well as you loose prod or res everytime you do it) instead I boost my ships speed by engines & lighthouses.
About 120 speed per turn gives you the ability to jump from one node to every other in a single turn, that ability seems to be crucial to be able to hit an adjacent AI without him trying to evade me by jumping at my lane or back (I see this behaviour constantly when I use slow Asteroid hulls - which forces me to separate into 2 fleets, one guardianing my current node (which is usually a chokepoint/entry into my own territory) while the other advances into his direction. Still he can jump backwards then, or try to fight it out against an enemy drastically reduced in firepower. With Gravitators you just jump at him with your complete fleet.

I can't say which version (speed or stealth) is better.
Stealth is cheaper to build but seems to take more time researching. As of now I'm unable to discern (before a combat starts) if I get firststrike or not - which is somewhat frustrating because I usually aim for lossless combat (once you've got Gravitators it's really easy... suffice to say I find your battle example a bit unrealistic... 20 RoboCruiser at one spot is highly unlikely, such a thing would only onccur if an AI is bottlenecked somewhere.... the AI usually is not really good at tactical placement of ships... if you ever encounter such a conglomeration just stay 2 jumps away and wait a few turns until he separates and then pick him apart...

On that note, usually AI fleets consists of fighters & also troopships & also scouts, the last 2 mentioned act as decoys & if you get unlucky they can account for an extensive soak in of firepower. More so if the AI uses Asteroid Troopers. Because of this, deliberately in such a scenario, I've go the feeling that the AI would be better off with all-attack ships (see screenshot taken from my current game...)

The stealth route also gives some additional benefits like covering planets which the AI is somewhat inept to react against so I usually refrain from using it. But if you go down that road it'll save you a boatload of research because you can ignore planetary defenses+shields+troops entirely. Which, sometimes, may even be counterproductive because maxing out shields & defenses can be a good way to form strongholds that are especially helpfull against certain roaming monsters.

Nevertheless, once you have the power to fight with Gravitators against RoboCruisers it's somewhat irrelevant what you do because you'll always win anyway. That might be different in Multiplayer without any technological advantages but I don't think there's a cookie-cutter best-build design there as well, since you need to play more reactively & observe what your opponent does every turn and then build a strategy to counter it. And that could be many things^^
Attachments
AIfleet.png
AIfleet.png (32 KiB) Viewed 378 times

slv
Space Floater
Posts: 24
Joined: Thu Mar 17, 2016 12:12 pm

Re: An extensive combat analysis of ship hulls

#21 Post by slv » Fri May 06, 2016 5:20 pm

Kassiopeija wrote:Well, couple of things.

I've the feeling that the defensive boost from Deflector-shield isn't a big deal. I mean you get a +3 initially, followed by +2, followed by +4. I've no clue how the design-intent is but before the shield-nerf these steps weren't so crude, but I just live with it that it's not so good & instead put more pressure to reach Plasma Shield ASAP.
I actually would like to have shields be a bit cheaper, and Deflector to protect from 6 damage, not 5.
Another thing is that by that time I build Gravitators I've, most likely, found a Good-Pilot species and fight with +25% or +50% weapon-damage - which, in my design, is increasing the strength of 6 modules while in your design works only on 3 modules. This is an additional reason why to go all-weapons.
It doesn't work this way, getting Good pilots gives the same percentage increase of firepower for all designs. If you have 6 lasers you get +50%, if you have 3 you are getting same +50%. In both cases you need equally less ships to beat your enemy. In most recent builds with good pilots you are not getting +50%, but +2 weapon levels, but still the relative increase of firepower is constant and doesn't depend on how many weapons you put on your ship.

The FleetCostUpkeep-multiplier should be also in my favour since you build/loose more ships.
Fleet cost is indeed an issue and eventually laser-heavy designs become better, but not until you build ~40 ships, which is quite a lot. I want to disagree that I loose more ships with this approach, I don't. Having more structure per ship and more ships make the damage distribution more uniform (unless you fight with one ship). Damage discrepancies hurt less numerous fleets more and in case you loose a ship the loss is more costly.
One internal slot is still empty - (and a core slot but at that point of the game there's usually nothing worthwhile to it) - I usually fill that internal slot with an engine or a stealth-module.
(Self-Gravitating hull has a -5 penalty to stealth but at +35 stealth I can successfully "first-strike" enemy fleets which helps immensely against planets on defense focus (against which shields are not really effectively...)).
Actually for firststrikers it's indeed better to have all lasers on a design. Organing line is quite good on that due to the stealth bonus. In this case I don't think that placing any armour or shields is worthy, if your plan involve to kill all enemy ships on round one then you want glass cannons with just weapons (and stealth module/drives).
Thing is that adding additional parts to fully complete the design will downgrade your design more than mine because, again, you loose more ships^^
But I don't loose more ships, it's you who looses more PP iff the battle is close ;) .

As of now I'm unable to discern (before a combat starts) if I get firststrike or not - which is somewhat frustrating because I usually aim for lossless combat
It's actually possible to do that (at least in the latest test release). If you see an enemy planet hovering over its detection range allows you to know what their empire detection strength is (Active Radar=30 Neutron Scanner=50 Sensors=70). Whenever your stealth is greater than their strength you firststrike (there are exceptions with ion storms, and some others, but that gives an idea). Before it was still possible but you needed to watch detection ranges on enemy planets and notice when they are suddenly increased (that's when they research new detection technology)
20 RoboCruiser at one spot is highly unlikely, such a thing would only onccur if an AI is bottlenecked somewhere....
Well, 10 cruisers will give same results, there just will be higher damage discrepancy (which by the way will hurt your designs more). And the exact numbers doesn't matter that much, as you've said we as human players pick only those fight where we are going to win. Having slightly better designs will not matter in most cases, but sometimes, once per few games you will find a case when my fleet will win and yours won't.

On that note, usually AI fleets consists of fighters & also troopships & also scouts, the last 2 mentioned act as decoys & if you get unlucky they can account for an extensive soak in of firepower. More so if the AI uses Asteroid Troopers. Because of this, deliberately in such a scenario, I've go the feeling that the AI would be better off with all-attack ships (see screenshot taken from my current game...)
I am pretty sure decoys doesn't change the relative power of enemy fleets. In most cases troopships, colony ships and other soaks have less structure than attack ships in which case the only thing they do is they increase an effective structure of fireships.
... but I don't think there's a cookie-cutter best-build design there as well, since you need to play more reactively & observe what your opponent does every turn and then build a strategy to counter it.
There shouldn't be. But there is ;) . It's going to be changed with fighters invention, though.

User avatar
Vezzra
Release Manager, Design
Posts: 4657
Joined: Wed Nov 16, 2011 12:56 pm
Location: Sol III

Re: An extensive combat analysis of ship hulls

#22 Post by Vezzra » Fri May 06, 2016 5:27 pm

Kassiopeija, slv, I've got one brilliant idea: why don't the two of you play a series of multiplayer games against each other? May the better strategy win... :mrgreen:

SkyCore
Space Floater
Posts: 23
Joined: Tue Apr 05, 2016 4:37 pm

Re: An extensive combat analysis of ship hulls

#23 Post by SkyCore » Fri May 06, 2016 10:11 pm

slv wrote:
SkyCore wrote: You can only build ONE more ship, not TWO more as you stated.
Whoops, you're right, I can build only one more ship, I was right tht damage*structure will be four times greater though.

Regarding small asteroids as you can see they have a lot lower efficiency so I decided to not include them.

As for numbers, I listed no-upkeep numbers, if you open the game and go to design page your ships have 4% upkeep. I listed the 0 upkeep prices (you can see them if you scrap all ships ad design on t2).
Actually the numbers i listed arent consistent with your version. I just downloaded the latest test branch and now the numbers look like:
mass driver(1) .0617
laser(1) .0536
plasma(1) .059
death ray(1) .0478

Small asteroids with plasma cannon shit all over most anything else in the early game.

Also i should list the crystalized asteroid with crystal armor(which are both unlocked from the same line of reasearch) for the mid-late game:
3x crystal 1x laser(1) 5*145/ 76.9*76.9 = .1226
2x crystal 2x laser(1) 10*118/99.8*99.8 = .118
3x crystal 1x plasma(1) 9*145/87.3^2 = .171
2x crystal 2x plasma(1) 18*115/120^2 = .143
3x crystal 1x death(1) 15*145/108^2 = .186
2x crystal 2x death(1) 30*115/162^2 = .131

Obviously you'd want to support all your asteroid fleets with a scattered asteroid or 2 for the flagship bonuses. Which also just happens to synergize VERY well with numerous cheap ships (rather than fully decked out large ships).

slv
Space Floater
Posts: 24
Joined: Thu Mar 17, 2016 12:12 pm

Re: An extensive combat analysis of ship hulls

#24 Post by slv » Fri May 06, 2016 10:56 pm

Hm, checked the numbers and they're right. It' kinda hillarious that mass drivers asteroids are as good a further versions (and better if you disregard shields). Also it's weird that plasma is the most efficient weapon.

Well, at least they have low speed, so it kinda balances out :). I will have to try to go for all in with asteroid one time, was a long time since I did it last time.

User avatar
MatGB
Creative Contributor
Posts: 3310
Joined: Fri Jun 28, 2013 11:45 pm

Re: An extensive combat analysis of ship hulls

#25 Post by MatGB » Sat May 07, 2016 2:22 am

Note on balance, especially shields.

PRedating my involvement, the current stats for weapons were set as follows.

Mass Driver starts at strength 3, and refinements add 1 each.

Laser starts equivalent to MD3 and refinements add 2 each.

Plasma weapons start equivalent to Laser3 and refinements add 3 each

Death Rays start equivalent to Plasma3 and refinements add 5 each (not sure why the jump)

When shields were introduced Geoff picked values and costs at a level he thought would be good for gameplay testing. I was always sure they were both too expensive and too powerful, so after extensive testing we (well, I, after discussion) toned them down.

Now each shield is set to block unrefined shots from a standard species of the equivalent tier weapon. That's deliberate, and those numbers aren't going to change unless the weapon numbers change.

The costs may or may not be wrong, I'm inclined to agree with slv on this point, unshielded designs (especially with engines) are my generally preferred route until I get to REALLY big ships.

There is almost certainly a good argument to be made that the relative strengths (and definitely the build costs) of the various weapon parts needs a balance pass, the costs of armour plating definitely does (far too cheap currently). That's not going to happen before the next Release, and probably not going to get touched on until after Fighters are integrated properly and we have feedback on that (ie during the next Release cycle).

@Kassiopeija, a recent change (and a very useful one) put the AI detection strengths into the Empire window (which is now actually useful so I leave it open in the bottom corner and have it just jut out from behind fleet windows). If the AI detection strength is below your fleet strength, they can't see you, and you get the first round of shots for free. Using stealth does take getting used to and it is very weird. Plus, the AI sometimes does annoying things attacking places you weren't expecting as the AI thought there was nothing there. But it is fun to try.
Mat Bowles

Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.

User avatar
Kassiopeija
Dyson Forest
Posts: 212
Joined: Wed Jul 16, 2014 6:14 pm
Location: Black Forest

Re: An extensive combat analysis of ship hulls

#26 Post by Kassiopeija » Sat May 07, 2016 2:47 am

MatGB wrote: @Kassiopeija, a recent change (and a very useful one) put the AI detection strengths into the Empire window (which is now actually useful so I leave it open in the bottom corner and have it just jut out from behind fleet windows). If the AI detection strength is below your fleet strength, they can't see you, and you get the first round of shots for free. Using stealth does take getting used to and it is very weird. Plus, the AI sometimes does annoying things attacking places you weren't expecting as the AI thought there was nothing there. But it is fun to try.
Thanks for the info. I'm still playing a somewhat older version and will upgrade once I finish the current game. Has there been any progress related the difficulty setting? I remember I tried to enforce to give the AI some bonuses.. it would really make the game more interesting if it's more challenging (even hopeless!) to win the map

wobbly
Space Floater
Posts: 30
Joined: Thu Oct 10, 2013 6:48 pm

Re: An extensive combat analysis of ship hulls

#27 Post by wobbly » Wed May 11, 2016 4:46 pm

So in my current game I've been using a fair few static multicellular hulls & finding them to be awesome little beasties & highly useful. I barely built any basic organic hull ships, all the utility ships I'd usually build as organic (colony, troop etc.) were the faster moving multi-cellulars. Looking at my turn log: organic hull - turn 17, orbital incubator - turn 25, multicellular - turn 28.

So some commits on the analysis:
I'd be a bit worried if asteroid hulls didn't outperform them in raw firepower+protection/pp. Why would I ever build clunky slow moving asteroid ships if I could get the same raw grunt from much faster ships?
The fact that they're not the most efficient combat vessels doesn't bother me, as I don't consider them front-line combat vessels. Robo-ships are so cheap to research & right on the production tech line if I need to mass churn out guns.
The speed difference between them & a regular organic is 20 instead of 10 once you have improved engine couplings because of the extra internal. If it's a colony ship the new colony will easily pay the difference if you get it down sooner. If it's a radar or troop ships it's less time your fleet is sitting & waiting for a critical ship.

What they're good at is I need an X at Y fast. Both in the time it takes to build 1 & the speed they move. The fact that they come already at full health is a bonus here if it needs to see immediate combat.

Here's an odd little multicellular I have lying around: 1 laser, 1 zortrium plate, active radar, deuterium tank, improved engine couplings. Doesn't matter that it's not highly efficent as I'm not mass producing it & it was cheap enough. If I suddenly need a radar somewhere it moves fast. If a scout or troop ships enter my zone of control the laser is enough & the speed gives it good spatial control. If I need a radar in combat (say laenfa) it's solid-ish & if it dies it was cheap. In an emergency it's an extra combat vessel that can get somewhere quick & absorb a little damage.

Edit: A small note about the AI. If it's going to continue to leave it's scouts unguarded & play them suicidal, it may be better off with this sort of layout. It's not a whole lot more to chuck a basic gun on an organic or a spatial flux & at least it'll kill/blind any unarmed scouts it meets.

Post Reply