Calculations inquiry for naval gaming,

[Warship NWS]

Moved from SAS thread, posted by Johnus (http://forums.navalwarfare.org//member.php?u=8910)

This may be somewhat off topic, but Scott's last post reminded me of an issue. I don't have an answer, but I'm open for opinions.

Scott wrote about a look under the hood at the complications involved with Tony's tweaking of the air code. That triggers my question. Is it any more accurate to consider and track all of the myriad complications (which you can never do in any case) for any complicated historical "game" question; or rather to get the right result by simple result oriented calculations.

Let's use naval gunnery at Jutland, for an example having nothing to do with SAS. Who is more histrically accurate ?? Designer X, who tracks the flight of each shell fired by Sedlitz during a game, with consideration given to the flight path mathematics of the shell, the fire control system, wind, sea state, etc.; or Designer Y, who simply applies the historical results, 3% hits under such conditions and such range ?? Keep in mind that Designer X may have to "tweak" his complicated computations in the event said computations produce results which differ from Y's historical results anyway.

Again, I don't have the answer. Food for thought while waiting for the patch.

[Warship NWS]

First a note; complexities does not mean the same thing as complications in terms of "under the hood" for SAS-WW2.

As to your question regarding gunnery accuracy, a lot depends on the validity and historical accuracy of the variables and equations involved. Example; WC:NAW may seem simple due to the 3 minute scaled factors shown in the ship charts but the calculations are extensively detailed for resolving the gunfire results. Detail however is not a substitute for accuracy. A lot of research is required to try and determine if the results, or figures, are as accurate as possible - or at the least, plausible. Much of the research is also not just about which events could have happened, but also how possible, of often, as event to actually had a chance to happen - a study of historically plausible statistics in this case is often required.

It is a complex topic so I am simply offering basic concept response for what your asking.

Thanks.

[old_pop2000]

One of the problems in simulating history is that history usually didn't happen the way we believe or have been taught. It is my opinion that the engineering approach is the best and proper way to design games, not the historical accuracy method. We don't know where the facts and figures came from that we might use to ensure that a simulation follows historical statistics. We might not even know who prepared the stats.

However, we do know the physics of interior and exterior ballistics quite well. We know the physics of falling bodies like a MK82 500lb bomb. We know the physics of aerodynamcs. We are using the perfect platform to perform such complex calculations to ensure engineering accuracy. That is essentially what we are after; scientific and engineering accuracy. If we are trying to ensure that Hood blows up on the four salvo everytime, or that Richard Best hits the Akagi with a 1000 lb bomb just forward of the island, then we are not faithfully replicating the engineering and physical principles that governed those actions. We can include what I have termed; the fickle finger of fate factor. Quite simply, randomness. It governs almost everything in our daily lives. With randomness, the Blue might have seen the Japanese cruiser force. Or Lancelot Holland might have made the turn sooner, or the torpedo from the Swordfish would drop a little later and completely miss Bismarck, and she would have made it to Brest.

IMHO, designers should stick to the more difficult path; that of using science and mathematics, no matter how complicated or difficult. They should use good engineering practices with a little randomness thrown in. Leave history to the books. If I run the Battle of the Denmark Straits a thousand times, but Hood never blows up, so be it. Randomness wins again.

Thanks

[Warship NWS]

To Dennis.. historicaly accurate, and compiled via authoritative sources, facts still play a role - just in case we are not on the same page here. You have to know the historical weapon parameters to base any math on and then often tie that information in with statistical analysis and engineering concepts - the part I think your trying to address. If there was no historical accuracy in play we would have no historical simulations to compare notes with in terms of plausible results. As to the fickle finger of fate.. a lot of that could be considered the variable and complex dynamics of human involvement in warfare. Example, if the cruiser Sheffield had not been attacked by Swordfish prior to attacking Bismarck the torpedoes would have had magnetic rather then contact warheads and likely not worked at all. ;)

[old_pop2000]

To Dennis.. historicaly accurate, and compiled via authoritative sources, facts still play a role - just in case we are not on the same page here. You have to know the historical weapon parameters to base any math on and then often tie that information in with statistical analysis and engineering concepts - the part I think your trying to address. If there was no historical accuracy in play we would have no historical simulations to compare notes with in terms of plausible results. As to the fickle finger of fate.. a lot of that could be considered the variable and complex dynamics of human involvement in warfare. Example, if the cruiser Sheffield had not been attacked by Swordfish prior to attacking Bismarck the torpedoes would have had magnetic rather then contact warheads and likely not worked at all. ;)

Absolutely, we have to have precise information on how the weapons were designed and used, if possible. Then the science and mathematics takes over along with the randomness. Randomness is the just the interaction of humans and nature. Nature does not always read the physics books, it tends to take its own path on occasions. However, the game must be based on proper science and engineering. Now, in testing, you can use historical statistics and facts to ensure, that the science and engineering is properly understood and duplicated, but once you have established that your science is within a reasonable accuracy, then the game is set correct. You cannot tweak a game to ensure that events work the way they did in real history. Let randomness take its toll.

[Johnus]

Excellent points gentlemen. Old pop2000, damn if I didn't think of Bismarck and Hood as I was phrasing the question.

Yes we have the technical tools. But how do you know if you are right in your calculations. The Battle of the Denmark Straights is a bad example because we don't know if Hood blew up due to her lack of armor; maybe it was that one in a million shot. The historical sample is too small. But if you did a game on Jutland and the British Battle Cruisers were bullet proof and the German Battle Cruisers kept blowing up you would know your calculations were somehow wrong and you would go back and revise them and revise them again until YOU THOUGHT you had found the error, and could then ensure more "historical" results.

Chris, maybe the bloody airstrikes of 1.03 are technically correct. Maybe they are overly bloody because that is how we and the AI play wargames. Maybe they could have fought to the death or exhaustion in the first few days of any campaign but they shrank from it. Maybe the limits (while undoubtedly required) need to be artificial to replicate both doctrine and the reluctance to shed blood, especially one's own side's. Or maybe I'm getting confused and overly philosophical in my dotage.

[Warship NWS]

To Johnus,

IMHO, the BCs at Jutland were lost as much due to human error as design philosophy and possibly even questionable battle tactics. I will stop there so as to avoid another elongated and often repeated debate about Jutland. ;)

At NWS we have been asked about the "Hood" event in regards to our simulation work. My reply is this - I would love to see someone come up with odds of what happened in detail including environmental effects, ship positioning during the turn, wave motion effects, ballistics, exact path through the hull after impact, the splinter effect after shell detonation - if it detonated, etc.. If Jurens, one of the most respected naval writers out there, could not do it what makes anyone think this would be a simple task? All we can do is try to apply as much historical fact, statistical analysis, and scientific calculations as we can to allow for such an event to occur within reasonable expectations.

SAS-WW2 v1.04 will have a lot of tweaks and adjustments included for the ais strike code including new calculations that will increase the level of realistic results. More on that part later. ;)

[old_pop2000]

Excellent points gentlemen. Old pop2000, damn if I didn't think of Bismarck and Hood as I was phrasing the question.

Yes we have the technical tools. But how do you know if you are right in your calculations. The Battle of the Denmark Straights is a bad example because we don't know if Hood blew up due to her lack of armor; maybe it was that one in a million shot. The historical sample is too small. But if you did a game on Jutland and the British Battle Cruisers were bullet proof and the German Battle Cruisers kept blowing up you would know your calculations were somehow wrong and you would go back and revise them and revise them again until YOU THOUGHT you had found the error, and could then ensure more "historical" results.

Chris, maybe the bloody airstrikes of 1.03 are technically correct. Maybe they are overly bloody because that is how we and the AI play wargames. Maybe they could have fought to the death or exhaustion in the first few days of any campaign but they shrank from it. Maybe the limits (while undoubtedly required) need to be artificial to replicate both doctrine and the reluctance to shed blood, especially one's own side's. Or maybe I'm getting confused and overly philosophical in my dotage.


Hi Johnus:

The Battle of Denmark Straits is not a bad example, we know how the Bismarck's guns were built, the shell construction and just about everything else including the maneuvers. We know how Hood was built, her actions etc. We know the atmospheric conditions that might have played a part. The real problem is that while we know the penetration, range of an average 38cm shell fired from an average 38cm gun, we don't know how much exactly that shell weighed. We don't know the actual wear on the guns, if any. We don't know what position Bismarck was at, when she fired. By that I mean, pitch, roll, yaw etc. We don't know the actual wind speed at the moment she fired, or during the entire flight of the shell. We don't know what positon Hood was in. We don't know exactly how far into the turn she was. Basically, unless we can be absolutely precise in this information, we are working with averages and there is always a bit of randomness to all this. The shell , on the average, weighed 800kg. However, getting each shell exactly 800kg is impossible. That shell might have been 801 or 805 and still be within specifications. In other words, we have tolerances in all the parameters and those tolerance maybe enough to change the whole scenario and resulting action. Duplicating thoses changes due to tolerances is almost impossible, so we have randomness.

In the actual battle, if another shell, that was lighter had been used, maybe that shell would have missed the Hood completely or struck somewhere else due to atmospherics and physics acting on a lighter shell. We just don't know and it would take a computer like a Cray to run all the possible tolerances to determine the right combination.

Make sense, or am I confusing me. :D

[JMS]

This is the olde debate of "design for effect" vs "design for cause", both have their advantages and their disadvantages.

Wargamers usually, but not always, will measure the success of a game by comparing it with their historical counterparts and, in the case of hypotheticals, by comparison with either published specifications or common perception. Of course, professional wargamers should be less constrained, but that's not always the case.

Problem is, to achieve that you have to design for effect, so every 4th shell blows up the Hood, so German Panzers are always able to punch through Sedan and Barbarossa always ends up stopped at the doors of Moscow. On a tactical scale it may simply mean that the hit % was 3% and that's it. This approach is not necessarily wrong if the 3% hit percentage takes into account all conditions and changes to the conditions are taken into account.

To put an example, in 1978 USS Batfish trailed a Yankee class SSBN for 45 days at ranges between 7,000 and 10,000 yards according to the declassified patrol report and interviews with it's CO without being detected. (if curious google for Operation Evening Star)

Would it be more accurate to take into account sea conditions, computers, hydrophone sensitivity, irradiated noise, etc, etc, etc. or just take the plunge and assume that 45 days was time enough to provide an average N.A. acoustic conditions/technology of both sides (taking into account also that there was a storm during which contact was lost) and use it for everything else.

[Christian Schwietzke]

It seems I am somewhat less demanding on realism than some others here appear to me.

For me, realism isn´t about being as close as possible to simulating every aspect of reality. It is about making sure that using historical tactics (and strategies, if the game is on that scale) *can* work out to produce historical results - and that, if both players (and the dice) cooperate, it is possible to produce something that is close to the way things went historically.

So for the Denmark Straits example, a game engine that does not model the (perhaps rather remote) possibility of a "critical hit" that blows up a ship right away is not realistic in that respect... whether a game engine models such things by tracking the exact place a shell hits, or simply by rolling dice and saying that a 01 on a d100 (or whatever the probability is) is a catastrophic critical hit, that is a secondary consideration to me.

[thewood]

I agree on tactics being the deciding factor. The problem with using only engineering and physics is that is impossible to account for every variable, especially training and morale. Now getting realistic tactical results may mean a more detailed physics model, especially when you get into things like tanks combat. I want to be able to change tactics and see the outcome and how it differed from history.

[JMS]

It seems I am somewhat less demanding on realism than some others here appear to me.

For me, realism isn´t about being as close as possible to simulating every aspect of reality. It is about making sure that using historical tactics (and strategies, if the game is on that scale) *can* work out to produce historical results - and that, if both players (and the dice) cooperate, it is possible to produce something that is close to the way things went historically.

So for the Denmark Straits example, a game engine that does not model the (perhaps rather remote) possibility of a "critical hit" that blows up a ship right away is not realistic in that respect... whether a game engine models such things by tracking the exact place a shell hits, or simply by rolling dice and saying that a 01 on a d100 (or whatever the probability is) is a catastrophic critical hit, that is a secondary consideration to me.


Yes and no. The perfect simulation is the one that takes into account all factors and on top of the expected result adds a variable for random results (the shell duds or goes on to explode on a magazine).

The problem with that approach is, as mentioned, wether all the variables are known and understood by the designer. OTOH, you can look at historical results and say that if the chance of a shell exploding a magazine was 1% of all hits, then use that as a guideline.

Both approached can give unrealistic results, the first one by missing one factor, the second one by having insufficient samples to generate a significant statistic.

The problem with wargames is that they don't happen in a void and people will expect to see certain things.

If you have the time, browse here for reviews of boardgames and how designer model cause/effect:

http://members.tripod.com/~RichardHBerg/ebrog.html/

[Warship NWS]

Lets not get too hung up on the "ballistics" or "physics" in terms of simulations as that is only part of the overall set of equations to consider for a naval combat environment. The mechanics behind chemical explosives (both above and below the water), splinter blast effects, KE calculations, metalurgy and quality of shell designs, armor integrity and quality vs various shell types, etc.. require quite a bit more understanding of combat environments then just "ballistics" or simple penetration values.

[old_pop2000]

Lets not get too hung up on the "ballistics" or "physics" in terms of simulations as that is only part of the overall set of equations to consider for a naval combat environment. The mechanics behind chemical explosives (both above and below the water), splinter blast effects, KE calculations, metalurgy and quality of shell designs, armor integrity and quality vs various shell types, etc.. require quite a bit more understanding of combat environments then just "ballistics" or simple penetration values.

There are many scientific and engineering principles at work in a combat environment, not just ballistics or the physics of explosions. As I did state, it would take a Cray supercomputer to account for all variables in a combat environment. All strategy and tactics are based on the weapons and those are designed using engineering principles. If I decide to advance through the Central Pacific like the advocate of War Plan Orange had planned, then I must have the ships and logistics capability to execute that plan. Those ships and that logistics capability will be a result of engineering and science. It will also be based on choices, choices of the kinds of weapons required.

Now, having said that, I also am fully aware of the intangible assets offered by humans in this endeavor. Intangible assets that are difficult to model in games but nonetheless, vital.

Anyway, that is my take.

[old_pop2000]

It seems I am somewhat less demanding on realism than some others here appear to me.

For me, realism isn´t about being as close as possible to simulating every aspect of reality. It is about making sure that using historical tactics (and strategies, if the game is on that scale) *can* work out to produce historical results - and that, if both players (and the dice) cooperate, it is possible to produce something that is close to the way things went historically.

So for the Denmark Straits example, a game engine that does not model the (perhaps rather remote) possibility of a "critical hit" that blows up a ship right away is not realistic in that respect... whether a game engine models such things by tracking the exact place a shell hits, or simply by rolling dice and saying that a 01 on a d100 (or whatever the probability is) is a catastrophic critical hit, that is a secondary consideration to me.


Christian, there is a reason why experts in ballistics and naval warfare call the kill shot at Denmark Straits, the "Golden BB". ;) But let's not focus too much on that incident,

[Ed Rotondaro]

Moved from SAS thread, posted by Johnus (http://forums.navalwarfare.org//member.php?u=8910)

This may be somewhat off topic, but Scott's last post reminded me of an issue. I don't have an answer, but I'm open for opinions.

Scott wrote about a look under the hood at the complications involved with Tony's tweaking of the air code. That triggers my question. Is it any more accurate to consider and track all of the myriad complications (which you can never do in any case) for any complicated historical "game" question; or rather to get the right result by simple result oriented calculations.

Let's use naval gunnery at Jutland, for an example having nothing to do with SAS. Who is more histrically accurate ?? Designer X, who tracks the flight of each shell fired by Sedlitz during a game, with consideration given to the flight path mathematics of the shell, the fire control system, wind, sea state, etc.; or Designer Y, who simply applies the historical results, 3% hits under such conditions and such range ?? Keep in mind that Designer X may have to "tweak" his complicated computations in the event said computations produce results which differ from Y's historical results anyway.

Again, I don't have the answer. Food for thought while waiting for the patch.

Chris:

If you can model the more complex dynamics of a shell flight, you will end up with a more accurate sim (he thinks?).

[Warship NWS]

Chris:

If you can model the more complex dynamics of a shell flight, you will end up with a more accurate sim (he thinks?).

Are you sure? Depends on how accurate your dynamics are and if one variable fails it can easily have a negative cascading effect. You also need to have enough knowledge, understanding of weapon mechanics, and accurate information to make sure that does not happen. You can't just go "I know ballistics" and think that is the whole equation to accurate weapon modeling. Your still plowing down the road of probabilities and statistical analysis to determine the final results - one method however is just from a different approach, but not necessarily always a better approach - to resolving the complexities of naval gunnery. The plausibility fo the results is a critical consideration.

Thanks.

[djcyclone]

The biggest problem I have with War Games, is the computer always has the advantage. You see it in games all the time.

Take CFS I, II, &III. In every single one of them, aswell as in every single other flight simulator, the computer always has the advangtage.

They can always out manuver you, they have unbelievable accuracy. I mean I have done dog fights, where I would chose a plane, then let the computer chose a plane randomly. During the fight the computer always does circles around me, so okay. Now I do the same fight, only switch aircraft. The computer will fly the plane that I was in with unbelievable abilities. The best way to describe it is as if the computer is flying a space jet (no gravity, no blackout, no redout). They can do anything they want to.

Take simulations like Silent Hunter III. I can guarantee you, that if you have only one torpedo left, the ship will not sink. It does not matter where you hit it, the ship will always limp away.

If you have 2 or 3 torpedoes, you can always break a ship in half with one torpedo. Game designers do this because they are under the false impression that it makes the game challanging, (it doesn't). It only makes the player angry, and then they stop playing.

Players like me are looking for realism. I want to be on equal terms with the computer, and if I get the upper hand due to certain cercumstances, then that is just the way it worked out. If the computer gets the upper hand, then I can live with that. But when it happens every single time (Come on?).

That is why I always cheat on games. I know that the computer has unbelievable advantages, and I cannot stand that.

[Warship NWS]

To DJ, I would not lump sum first person simulations with wargames as they are different genres even if they are covering similiar topics. ;)

[Ed Rotondaro]

Are you sure? Depends on how accurate your dynamics are and if one variable fails it can easily have a negative cascading effect. You also need to have enough knowledge, understanding of weapon mechanics, and accurate information to make sure that does not happen. You can't just go "I know ballistics" and think that is the whole equation to accurate weapon modeling. Your still plowing down the road of probabilities and statistical analysis to determine the final results - one method however is just from a different approach, but not necessarily always a better approach - to resolving the complexities of naval gunnery. The plausibility fo the results is a critical consideration.

Thanks.

Chris:

Apparently all major navies at least tested the accuracy and penetrating powers of their major caliber shells on obsolete warships to at least establish a baseline of what could be reasonably expected from the ship's armor and the weapon opposing it. The problem is we have tons of examples of air to air combat or tank to tank combat or artillery against fixed positions. We have very little for naval combat in comparison unless reviewing the lighter combatants (CAs, CLs, DDs) or air attacks. The best you can do is use the empirical data from tests to see if it can replicate anything historical.

[Warship NWS]

Chris:

Apparently all major navies at least tested the accuracy and penetrating powers of their major caliber shells on obsolete warships to at least establish a baseline of what could be reasonably expected from the ship's armor and the weapon opposing it. The problem is we have tons of examples of air to air combat or tank to tank combat or artillery against fixed positions. We have very little for naval combat in comparison unless reviewing the lighter combatants (CAs, CLs, DDs) or air attacks. The best you can do is use the empirical data from tests to see if it can replicate anything historical.

Note, navies did not have readily available access to potential, or known, enemy armor designs and structural configurations so at best they could only guess what their weapons could do vs enemy designs scaled using non-combat results vs whatever armor and guns they had on hand. Nathan Okun even noted that KM tested their ordnance vs older armor types and had to rescale his figures to compensate for a slight deduction in penetrating power for their AP shells. However, this still does not account for AP cap designs and metallurgies and how they will respond dynamically after impacting various armor types, angles, multiple layers of armor, etc. Point being, it needs to go a lot further then simple penetration values and ballistics to gain any reasonable, or plausible, level of accurate results.

Thanks.