Aircraft Maneuverability

[el cid again]

This turns out to be more importent than one might guess. It is used to decide if a plane gets to shoot - which makes sense. Probably it is relative maneuverability, or a differential. That means that a higher rating on EITHER side (attacker or defender) impacts the differential quite a bit. But what is it? Data analysis shows in many cases for 1 and 2 engine planes it is simply max speed divided by 10. Strangely, other cases indicate that

speed/10 plus roc/100 was used.

And for bombers and transports, the result was divided by 2 for twin engine planes and something like 8 for four engine planes!

As always - partly because of no stated definition and lots of added planes - the data is inconsistent. We will like relative results better if we use a consistent one.

I propose

[ speed / 10 plus roc / 100 ] / number of engines.

This will double the value of 4 engine planes (in most cases from 4 to 8) with a high performance plane getting a 9 and a low performance plane getting a 5, 6 or 7. It will keep twin engine transports and bombers about the same (in the neighborhood of 15) - but it forces other twin engine planes (fighters, night fighters, recon) down lower (in the 18-30 range).
Most single engine planes will be about the same or slightly higher than they were - those which had speed only ratings now gain a bit for ROC - while those with both are the same.

The results indicate that many of the planes we care about are remarkably similar. Most single engine fighters are in the 35-45 range - with high performance being around 51 (e.g. the P-51 is rated at 51 - which seems somehow fitting) - and low performance down around 25-35.

One must remember that maneuverabilty, firepower, range and durability are all different. A plane may be good in one but not in the others. Thus a B-29 is very similar to a B-17 in maneuverability - only slightly better - but it is quite a bit better in firepower and probably also better in durability. On the other hand an Oscar (or a comparable US light fighter like the P-63) are pretty maneuverable, but lack firepower and durability. As a general rule, multi-engine planes will have more payload and/or more range. This discussion ONLY considers maneuverability - now we understand that it is via this value that relative speed is implemented in the air combat engine.

[MkXIV]

ORIGINAL: el cid again

This turns out to be more importent than one might guess. It is used to decide if a plane gets to shoot - which makes sense.

SOmebaody did some testing a while back on the zero bonus, the results appeared to show that the bonus didn't really help the zeros get more kills, but it DID prevent zeros from getting shot down. So MVR may play a lager role in "Defense" then in "Offense"

[akdreemer]

ORIGINAL: MkXIV

ORIGINAL: el cid again

This turns out to be more importent than one might guess. It is used to decide if a plane gets to shoot - which makes sense.

SOmebaody did some testing a while back on the zero bonus, the results appeared to show that the bonus didn't really help the zeros get more kills, but it DID prevent zeros from getting shot down. So MVR may play a lager role in "Defense" then in "Offense"

If this is so then a fighter like the P-38 should have a higher maneuverablilty rating. There is more to maneuverability than jut how many engines a plane has. One needs to look at the entire system. I think a better solution would be to look at how the vaious air- to air simulators handle this issue. I would guess that there are a few forum readers out there who could shed some light on this.

[m10bob]

You rang ??[:D]

http://www.rdrop.com/users/hoofj/

[akdreemer]

ORIGINAL: m10bob

You rang ??[:D]

http://www.rdrop.com/users/hoofj/

thanks m10bob, but I have been informed that data on the Internet has no veracity as far as CHS or his own mod's standard of documentation as I have been recently informed by originator of this post ???

[el cid again]

Since

1) We wish to make air combat less lethal

and since

2) We wish to make durability generally less so as to increase operational attrition and AAA attrition - which probably will make air combat more lethal -

I now modify my proposal above by suggesting the first pass should be

[speed / 20 plus ROC / 100 ] / number of engines.

I am going to attempt to complete the allied aircraft review and then assign revised (and consistent) durability ratings. This means I have revised firepower (to make it accurate in range and proportional to round effect), maneuverability and durability. I wish then to do as broad a series of tests as possible to calibrate - and determine if a K factor is required? This variable - maneuverability - is probably the best place to put the K factor - because it is a statistical creature - but durability is also a possibility.

[el cid again]

SOmebaody did some testing a while back on the zero bonus, the results appeared to show that the bonus didn't really help the zeros get more kills, but it DID prevent zeros from getting shot down. So MVR may play a lager role in "Defense" then in "Offense"

_____________________________

Do we have any clue what the "zero bonus" is supposed to affect?

[TheElf]

Cid,
How is relative Maneuverability translated in terms of game mechanics or "formulas" as it were?

[el cid again]

If this is so then a fighter like the P-38 should have a higher maneuverablilty rating. There is more to maneuverability than jut how many engines a plane has. One needs to look at the entire system. I think a better solution would be to look at how the vaious air- to air simulators handle this issue. I would guess that there are a few forum readers out there who could shed some light on this.

Actually, I can. I have worked as resident computer engineer at two Boeing SILs (Software Integration Laboratories), and as a field engineer at a dozen others. Aircraft and missiles in our day always fly for many years in software before any metal is cut, and similarly modifications are tested in software first. I began this process by writing to Joe - a professional programmer and a mathmetician. He suggested that I assume that in this system maneuverability may just be speed - and indeed in a large number of cases it is in fact exactly speed/10. Since this was not defined anywhere, it is not surprising it is not consistently used. I myself do not like using just speed, so I tried fitting ROC/100. It fit no planes in this system. I then tried ROC/500 and found many cases where it fit - maybe great minds think along parallel paths. OK - now I understood what is used - some planes are using speed/10. Some are using speed/10 plus ROC/500.

Now about engines. I wanted to penalize multi-engine aircraft. Clearly so did WITP designers - they have 2 engine bombers divide the value by 2 and 4 engine bombers divide by 8. In order to keep your P-38 - and other twin engine - fighters and night fighters as high performance as possible - they "cheated" and didn't divide them at all. I think this is the wrong way to model - so I "played fair" and divided them too. Anyway, what you said is backwards - more engines REDUCE maneuverability - they do not increase it. This is particularly true of propeller aircraft when the engines are separated. It actually is NOT true in rare cases where the engines are in line and driving the same propeller - or counter rotating propellers on the same axis - so if we had such a plane I would not penalize it by dividing by two. But P-38 is not such a plane. It gains all its positive values from its speed and ROC, and in other values - notably firepower and range - it will also benefit from the two engines.
But its maneuverability is not comparable to a single engine fighter - so I do not think that it should be treated any differently than a C-47 or a B-25. It is quite a bit better in maneuverability - about 10 points in my original scheme - and 5 points now - but that is because it is fast and climbs well - not because it is fantastic in maneuverability.

In more complex simulations, I do use a maneuverability rating which itself is more complicated. I consider wing loading, power loading, rate of climb, speed and turn radius. But the fact is that the composite value is very time consuming to determine, and it does not produce a value radically different than the present proposal. ROC does tend to be higher if you have lower wing loading or higher power loading, for example. It is a quick and dirty value, but it is surprisingly good. Unless you volunteer to do a lot of man-hours of calculating, it will do fine for this pass.

[el cid again]

thanks m10bob, but I have been informed that data on the Internet has no veracity as far as CHS or his own mod's standard of documentation as I have been recently informed by originator of this post ???

I looked at the site referenced. It is not directly applicable for our purposes. We have to rate a very large number of planes (it cannot be more than 249 and probably will be about 240 for RHS). More than a few of these planes are not available to test in the first place. Only a small set of planes were tested, many of them (e.g. Bf-109) not of our theater of interest. Further, we do not know how typical these planes may be.
These tests might be of interest if we wanted to measure the range of errors in references, but they would not be directly able to give us performance data for all the planes of interest, even if we devised a system to use all this data. Frankly there is no shortage of information - what there is a shortage of is time. I do not propose to spend a year on determining maneuverability - which I could easily do if I wished to get a very precise algorithm and then validate it. Right now I will settle for anything that is consistent, related to real data, and easy and fast to use.
I want something better than we have - not something perfect. There are problems in the code - going for industrial grade accuracy (which is in the 2 to 4 % range) is not appropriate until we get the model better than it is. I will be happy to have relative maneuverability ratings which are less than 10% different from what a very complex calculation would produce - and get them in the next 24 hours instead of 24 months.

[el cid again]

How is relative Maneuverability translated in terms of game mechanics or "formulas" as it were?

I do not understand your question. Or rather I have already answered your question when I decoded what the data is using. So I assume you may be asking something else. Maneuverability appears for many planes to be maximum speed in mph divided by 10 - period. For other planes it appears to be this value added to rate of climb divided by 500.

If you mean how is relative maneuverability used in the air air model - it is not possible to say given the code is secret and also scrambled so we cannot go look at a program listing for "air combat routine" and just read it for ourselves. However, it probably uses something like a differential - attacker maneuverability minus defender maneuverability - since that way the maneuverability of both is important. This differential might be indexed with a die roll (game talk for a random number in a computer) - for a model of this sort I would not expect anything much more complicated than simply that. There are probably modifiers to the die roll - things like altitude advantage, radar warning for interceptors and cap, etc. All we know is that maneuverability is the data that is used - our assumption it might be a set of factors was wring - and our assumption that speed would matter is sort of wrong: speed is the only component of maneuverability for some planes, or the dominant component for others - so it is still the factor - but changing the speed may not matter as such (since no one knew enough to change the maneuverability to match).

For the record this is working blind - it would be better if we knew how to enter data properly - and how it is used. It is somewhat deceptive to say "you can use an editor to mod the game" but not tell us the definitions for the fields. I think this ought to change - and if there is a cost involved then I think a way should be found to cover it. It is simply horribly inefficient to have us working on software without definitions and a support system (where we can go to get answers - even if we pay for them).

[TheElf]

If you mean how is relative maneuverability used in the air air model - it is not possible to say given the code is secret and also scrambled so we cannot go look at a program listing for "air combat routine" and just read it for ourselves.

That was the point of my question. If we don't know how the game uses manueverability what good will "reviewing" all the aircraft do? When you are done reviewing, and are certain you have the "truth data" for all the aircraft in WitP what will you do with it then? You can't just throw it into the database and expect a model we know nothing about to work better than it already does, or as well for that matter.

I'm not trying to rain on your parade Cid, just playing devil's advocate. I would like to see a more fleshed out A2A model too.

[el cid again]

If we don't know how the game uses manueverability what good will "reviewing" all the aircraft do? When you are done reviewing, and are certain you have the "truth data" for all the aircraft in WitP what will you do with it then? You can't just throw it into the database and expect a model we know nothing about to work better than it already does, or as well for that matter.

I would prefer a direct method:

1) Look it up in a technical manual
2) Read the code
3) Ask tech support and get a precise and complete answer

ANY of these would work for me.

But what I have is a more complicated and less precise situation.
We have no code access, no technical manual and no tech support willing to answer specific questions - even for money. But we DO have each other, and we DO have the editor and user manuals, and we DO get (unannounced) tech support. On top of that, I have a background in test, and I do intelligence analysis - so I can decode things by looking at data and by testing. I now know that the air combat routine uses maneuverability where I assumed it would use several variables - and that knowledge means we can control the model to a degree. The ONLY thing we cannot fix is the "more than 50 planes per side" issue and the "unlimited ammo" issue. It is not the way I like to do things - but it is the only option for now.

There is more to the story. I believe we will get some code changes. By Matrix and/or by others. And not in the next century either.
More like next summer. Things will get better. So I am fixing what I can for now - and focusing on getting inconsistent data consistent and well defined. Then when the code gets better we will really like what it does when planes are not rated wierdly relative to each other.

[el cid again]

You can't just throw it into the database and expect a model we know nothing about to work better than it already does, or as well for that matter.

The flaw in your sound reasoning is this: we know more than "nothing" about the model. We can figure out some things. And we are getting some help too. The two principles interact to mean we will figure out still more. And now I know the primary input variable, I can calibrate the model.