PDA

View Full Version : Induced drag undermodelled in Il-2, FB and PC?



Holtzauge
12-04-2004, 04:57 AM
I really love this simulator, the graphics, the flight characteristics and rate it the best!

However, good things can always get better.

Having tested different climb speeds, it seems to me that the effects of the induced drag is undermodelled. In real life, you have to watch the speed real careful in order to maintain the max climb rate. Now it seems to me you can usually retain good climb performance even though you decrease the speed well below best climb speed. I have tried this in the 109 and 190 and I do not seem to be punished that much by slowing down below best climb speed. I even see G-vapour vortices at the wing tips and still climb decently. Now in real life I would be severly punished in terms of climb rate if I tried this.

It also seems amazing to me that after a head-on pass a Spit or Hurricane will have turned round in a jiffy and still amazingly have his speed left. In real life he would have traded the quick turn for speed and be out of energy.

Mayby this is done intentionally by the designers because it would not be fun for the average arcade jock who thinks flying is going full throttle and hauling back on the stick? He would probably complain that he could not turn properly in the game. Maybe this is needed for the sim to appeal to a broader audience?

However, I guess that I am not alone in wishing to have the sim as close to real life as possible. I want this hard core realistic.

It think it makes sense to work on rpm, engine temp, cooling louvre stuff and so on but please, fix the big things first!

I think undermodelling of induced drag is a big issue. It upsets the tactics you use and secondly, it should not be to difficult to fix in the code:

Cdi=Cl**2/(pi*A*e)

Where A=Aspect ratio= Span**2/Wing area
e= elliptical factor
pi=3.14.....

So as soon as the Cl is increased from the ideal at the best climb speed, I will be punished by the square of the Cl which will be severe.

To make things worse, the elliptical factor is highly dependant on Cl and as CL increases this will go down way below the ideal 1.

So the drag increases by the square of the Cl and you divide by a number below 1. It will all boil down to grief in real life but here you get away with it!

The Cl should be easy to calculate in the flight model and I hope they include the drag increase by the square. If this is so then what needs to be done is to decrease the e factor more at higher angles of attack.

So this is not rocket science, but basic aerodynamics!

Maybe the above sounds negative. But do not get me wrong. This is the best sim around and I love it!

However, I plead to the designers to fix the induced drag issue. Get it as close to real life as possible! Make a good thing better!

Extreme_One
12-04-2004, 05:02 AM
"It also seems amazing to me that after a head-on pass a Spit or Hurricane will have turned round in a jiffy and still amazingly have his speed left. In real life he would have traded the quick turn for speed and be out of energy."

Are you talking about a human or AI pilot?

AI cheats!

k5054
12-04-2004, 05:49 AM
You've got the right formulae, test it, track it, measure it, compare it with the calculated values.
I've done so and found it's in the right ballpark, bearing in mind that you can't separate induced drag errors from prop efficiency errors in the low speed area. At high speed you would do well to see how having a lightly loaded aircraft makes for far less energy loss at high speed. Of course having a heavy aircraft makes for less speed loss.
Take the 400mph 5g turn we have for the 109K, see how much speed it has lost in 180deg from your calculation, and compare with the flight test figure we have had posted here.

Climb rates below best climb do not in fact decline so much as you might think. The power required to stay in the air doesn't change much below best climb, therefore the power available only declines due to prop efficiency losses.


I have no doubt at all that the FM in this game uses exactly the formula which you cite. I don't know whether it has accurate prop efficiency curves, but nor has anyone else, the sim tries to get it right. Whether they always use the right constants for each aircraft to insert into the formula is also a matter of opinion. Oleg doesn't tell us, and he has the right not to. We, however, have the right to point out when it seems wrong, if we have data to back it up, and Oleg has often responded with a fix.

Willey
12-04-2004, 09:22 AM
Try this:

Take a D3A or A6M2-21. Climb at 180-200kph IAS. Then repeat the climb with wings folded, you have to pull extreme AoA then, but it's possible. Do NOT pass 210 IAS or your wings will rip off. Then just look which one was faster, you'll be stunned http://www.dierk3er.privat.t-online.de/JG5/Bilder/v.gif

Holtzauge
12-04-2004, 10:14 AM
I did some quick checking and I came up with the follwing sources:

1) Fluid dynamic lift, by Dr-ing S F Hoerner, 1965
2) Fluid dynamic drag, same author

In section 14-3 of ref 1, there is a complete drag breakdown of an 109G and some data on the Emil too.

There is a figure 9 on page 14-11 which depicts induced drag factor Cdi/Cl**2. It seems according to this that the Cl was about 1 at climb. I'm assuming about 250 km/h IAS for best climb.

If we for example compare what this would mean if we assume two different climb speeds, 220 and 250 km/h (13% speed difference) we get a Cl of 1.3 at 220 km/h.

Since the drag increases by the square of the Cl this will mean a drag increase of about 70% which should definitely show up on the climb rate indicator.

I have tried climbs (with the 109G-6) at 220 km/h IAS (and even lower) and at 250 km/h IAS and I do not think the difference is as marked as it should be.

By the way, where can I find the data on the 109K you referred to?

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by k5054:
You've got the right formulae, test it, track it, measure it, compare it with the calculated values.
I've done so and found it's in the right ballpark, bearing in mind that you can't separate induced drag errors from prop efficiency errors in the low speed area. At high speed you would do well to see how having a lightly loaded aircraft makes for far less energy loss at high speed. Of course having a heavy aircraft makes for less speed loss.
Take the 400mph 5g turn we have for the 109K, see how much speed it has lost in 180deg from your calculation, and compare with the flight test figure we have had posted here.

Climb rates below best climb do not in fact decline so much as you might think. The power required to stay in the air doesn't change much below best climb, therefore the power available only declines due to prop efficiency losses.


I have no doubt at all that the FM in this game uses exactly the formula which you cite. I don't know whether it has accurate prop efficiency curves, but nor has anyone else, the sim tries to get it right. Whether they always use the right constants for each aircraft to insert into the formula is also a matter of opinion. Oleg doesn't tell us, and he has the right not to. We, however, have the right to point out when it seems wrong, if we have data to back it up, and Oleg has often responded with a fix. <HR></BLOCKQUOTE>

k5054
12-04-2004, 04:11 PM
There's an easier rule-of-thumb method. If best climb is 250kph, that's the speed at which induced drag will be equal to profile drag. If you go slower, one half of the total drag is increased by a percentage derived from the square of the speed
ratio, and one half is decreased by the same ratio. This always results in more drag, but not as much more as the speed ratio squared. Your 13% speed difference means 50% of the drag is increased by 27.7%, making 64%, and the other 50% reduced by 27.7%, making 39%. Total is 103%.
Not much of a drag increase. Your method is not exactly right, the induced drag CD varies as you stated, but it has to be multiplied by the speed squared again to get actual drag. Bear in mind also that the engine has to use power, that's thrust x speed, to overcome this drag. As its going slower, less power is needed to overcome the same amount of drag. However, changing this power into thrust requires the prop, which is probably losing efficiency at this stage.

A while ago we had a long thread about just this kind of subject. There was something wrong with just this induced drag in PF 3.0. If you aren't using the 3.01 or 3.02 fix you probably are seeing badly modelled low-speed drag or power.
Unfortunately I can't find the thread in question, it may have been lost or the search can't find it. That's also where the 109K chart was last posted, although it comes from months ago. I'll keep looking.

Blottogg
12-04-2004, 11:26 PM
Holtzauge, good sources. What may explain the small relative change in climb rates for your examples of 220kph and 250kph is that for level flight, Cdi is a relatively small portion of Cd. Only as you get close to the stall does Cdi start ballooning. In Dr. Hoerner's example, Cdi = 0.0025, while Cd = 0.036, meaning that induced drag is only about 7% of total drag (Cdi/Cd). Assuming power and parasitic drag don't change between the two speeds (they do, but not by enough to mention, and I don't have those numbers anyway), increasing Cdi by 70% only increases total drag by 0.7x(0.0025/0.036) = 0.049, or about 5%.

Like Simon mentioned, the Ai uses a simplified flight model, which can accelerate/decelerate better that the player's FM in most cases. For the player's FM, induced drag is modeled quite well. Getting behind the power curve on carrier approach (significantly slower than configured L/Dmax) has gotten me a mouthful of fantail on several occasions, despite the application of full power, just as it should have.

edit: grammar

WWSensei
12-05-2004, 06:14 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Assuming power and parasitic drag don't change between the two speeds (they do, but not by enough to mention, and I don't have those numbers anyway), increasing Cdi by 70% only increases total drag by 0.7x(0.0025/0.036) = 0.049, or about 5%. <HR></BLOCKQUOTE>

I'm not sure even this formula is correct. You have to use Amdahl's Law to calculate the efficiency of a sub-component on the overall impact to the entire system.

Amdahl's is typically used in manufacturing of computer components but it can be used to judge how increasing a sub-component (or decreasing a sub-component) where that sub-component is a fraction of an overall system and measuring the impact on the entire system.

So, assumming we are talking drag then we would use:

Total Drag Inc = 1/((1-f)+(f/s))

where f represents how much CDi is of the overall system and s = how much CDi increase.

so...

1/((1-.07) +(.07/.7) (assumming CDi is 7% of total and increases by 70%...

Overall increase = 1/(.93+.1)

= 1.0638 or about a 6% increase in overall drag when CDi...not far off of the rough 5% but it's an exponential scale and would not work if in this case CDi hadn't increased so realtively little...

Holtzauge
12-05-2004, 10:54 AM
When you look at Cd, Cdi etc in graphs or formulas you have to be careful about the reference used. Is it cross section area, wing area or the wet area of the a/c?

I think you are comparing Cd and Cdi based on different references.

However, if we look at figure 9 on page 14-11 in Hoerners book we are fortunate indeed. This gives the total drag (induced and zero lift drag)for a Me-109. In the graph the zero lift drag is as you say 0.036. However, on the x-axis the square of the Cl is given and the drag factor goes up from 0.036 to about 0.1 on the y-axis before stall effects start to be noticed.

This means that the induced drag is between zero and a factor of 3 depending on the lift coefficient even before stall effects begin.

I calculated the Cl needed at a 220 km/h and at 250 Km/h and read of the values of Cd=0.05 and Cd=0.07 respectively in graph 9 which states the total drag coefficient including all factors

So the drag increase is in the order of 1.4 i.e 40%

I think this is in line with my previous calculation which was a rough cut only to illustrate that the indiced drag went up in the order of 70%. I did not mean that the total drag went up 70%.

So I still think this should show up in the sim. I will now take a slight intermission to enjoy dinner. After this I will do some tests in the sim and log the results.

Just by <BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Blottogg:
Holtzauge, good sources. What may explain the small relative change in climb rates for your examples of 220kph and 250kph is that for level flight, Cdi is a relatively small portion of Cd. Only as you get close to the stall does Cdi start ballooning. In Dr. Hoerner's example, Cdi = 0.0025, while Cd = 0.036, meaning that induced drag is only about 7% of total drag (Cdi/Cd). Assuming power and parasitic drag don't change between the two speeds (they do, but not by enough to mention, and I don't have those numbers anyway), increasing Cdi by 70% only increases total drag by 0.7x(0.0025/0.036) = 0.049, or about 5%.

Like Simon mentioned, the Ai uses a simplified flight model, which can accelerate/decelerate better that the player's FM in most cases. For the player's FM, induced drag is modeled quite well. Getting behind the power curve on carrier approach (significantly slower than configured L/Dmax) has gotten me a mouthful of fantail on several occasions, despite the application of full power, just as it should have.

edit: grammar <HR></BLOCKQUOTE>

Holtzauge
12-05-2004, 12:14 PM
You are quite right about the the effect of the reduction due to the decrease in dynamic pressure and I'm not implying that the climb rate should be reduced by 70%. I was trying to make a point about a the increase in induced drag. I just answered another post with numbers derived for a graph (Hoerner fluid dynamic drag, page 14-11, figure 9) which showed the 109's total drag as a function of the Cl squared.

When I read of the Cd increase from the difference between the squared Cl's at 220 and 250 km/h this came out as roughly 0.07 and 0.05 respectively (Cdo=0.036).

So if we instead look at the total Cd this goes up 0.07/0.05=1.4 about 40%. If we then look at the total drag this as you say has to be reduced by (220/250)**2=0.7744.

This means that the actual total drag increase is around 1.4*0.7744=1.084 about 8.5%.

Unfortunately I am running 3.0 so mayby this is corrected? Great!

Here are the numbers I got in 3.0:

Me109G6 100% fuel climb from 500 to 3000m. I started at 100m, stabilised the climb and speed and triggered the clock passing 500 m and 3000m so there should zoom effect included.

at 250 km/h IAS: 19.45 m/s (average of two clims)
at 220 km/h IAS: 19.31 m/s (average of two clibms)

This is within the error margin 1%n and not near 8.5 %.

Now the clincher for me. I did a couple of tests at 180 km/h IAS and I got 18.66 m/s. Now 180km/h IAS is way off the best climb speed and there should be a marked difference.

I do not have 3.01 or 3.02. It would be nice if some kind person fortunate enough to have these clocked the same climbs so we could compare?

Unfortunately my practical experience does not include the 109 (alas!) but from flying private planes and it is of course difficult to extrapolate from this but I know for a fact that deviating from the best climb speed did have marked effects on the vertical speed indicator!

However, when climbing at 250 km/h IAS in Gustav I get the 19.45 m/s which is dead on the money compared to what Cpt Eric Brown clocked the "Beule" on (19.3 m/s)during trials in WW2 on a captured specimen!

My hat is off to the designers of this fantastic sim!

However, I still think Mr Brown would have clocked a much lower value had he tried the same thing at 180 km/h!

[/QUOTE]
There's an easier rule-of-thumb method. If best climb is 250kph, that's the speed at which induced drag will be equal to profile drag. If you go slower, one half of the total drag is increased by a percentage derived from the square of the speed
ratio, and one half is decreased by the same ratio. This always results in more drag, but not as much more as the speed ratio squared. Your 13% speed difference means 50% of the drag is increased by 27.7%, making 64%, and the other 50% reduced by 27.7%, making 39%. Total is 103%.
Not much of a drag increase. Your method is not exactly right, the induced drag CD varies as you stated, but it has to be multiplied by the speed squared again to get actual drag. Bear in mind also that the engine has to use power, that's thrust x speed, to overcome this drag. As its going slower, less power is needed to overcome the same amount of drag. However, changing this power into thrust requires the prop, which is probably losing efficiency at this stage.

A while ago we had a long thread about just this kind of subject. There was something wrong with just this induced drag in PF 3.0. If you aren't using the 3.01 or 3.02 fix you probably are seeing badly modelled low-speed drag or power.
Unfortunately I can't find the thread in question, it may have been lost or the search can't find it. That's also where the 109K chart was last posted, although it comes from months ago. I'll keep looking.[/QUOTE]

k5054
12-05-2004, 12:40 PM
Yes, there is a real problem with 3.0, 3.01 has the FM changes. Having said that, I would still expect climb rate not to vary too much a little way either side of the best climb speed. Expressed graphically the excess power is the difference between two parabolas, one convex up, the other down. Sometimes this can result in a fairly flat excess power curve, especially for a lightly loaded fighter. If the best is 250kph, its quite possible that any speed between, say, 220 and 280 would give a similar climb rate. Tactically a higher speed would be better anyway, even at the sacrifice of one or two m/s.
Compared to modern light planes WW2 fighters have much more excess power. In theory the loss of climb rate from the wrong speed will be the same number of m/s, but far less as a proportion of total climb rate.

Blottogg
12-05-2004, 06:01 PM
Sensei, you're right that the delta Cdi is exponential, and gets huge further from L/Dmax (the generic "ballooning" comment I made earlier.) Mine was just a back of the envelope approximation for a portion of the Cdi curve that was more linear (Amdahl's equation looks vaguely familiar, but my practical engineering is both rusty and in some areas obsolecent.) Still, I was pretty close.

I haven't tested low speed Cdi in the sim like Holtzauge has, but your numbers look a little high for 180 kph. If the Cd model is unchanged in 3.02, there may still be a problem. I had ascribed the reports of "prop-hanging 120kph climbs" to the Ai's simplified flight model. It may be due to an over-optimistic Cd curve, or that 1C is using only Cdi=Cl^2/(pi*A) (where A=aspect ratio) to calculate delta Cd, which remains linear plotted against Cl^2 (obviously) in Dr. Hoener's fig. 9 on page 14-11. The actual wind-tunnel data shows a Cd increasing non-linearly vs. Cl^2 at high Cl (for both the whole aircraft and the wing alone), due to parasitic and form drag going non-linear (separated, turbulent flow) at high AoA, like I mentioned earlier. If those changes aren't calculated, Cd will be too low, which may explain the discrepency. As my performance course professor told us "If you can come up with an accurate method to calculate the effects of turbulent flow, you'll be very rich." He also told us, "Yeah, I wanted to be an aerodynamicist, until I took my first graduate-level fluid dynamics course. I asked 'Where is that aeroelasticity syllabus' real quick after that."

You're also right that Cd's need to be an apples to apples comparision, using the same reference area S. That's one of the reasons I like Dr. Hoener's example, since it's all on the same sheet of music.

Holtzauge
12-06-2004, 01:22 PM
I am still curious if the high Cl induced drag issue is corrected in 3.01 or 3.02. Unfortunately, I'm stuck behind a modem and have not upgraded yet.

I also agree with K5054 that the effects on climb performance should not be to great if I deviate towards a higher speed. However, as you also point out, the minor performance drop going the other way to 180 km/h does seem a bit optimistic. At this speed Cl may be around 1.2 and if one looks in the figure 9 in Hoerners book we are in the stall region and the effects should be more marked than a decrease of around 1 m/s from the 19.5 m/s at 250 km/h that my test indicated.

I think the figure 16 on page 6-9 in Fluid Dynamic lift shows the slat operation of a 109 and using this figure I believe that at 250 km/h the slats will still be in due to the moderate Cl while at Cl=1.2 (180 km/h), the slats should be nearly fully deployed. I do not think that climbing with the slats out would pay off in real life. It would be better as K5054 says to go a bit faster than to end up on the slow end of the envelope with the slats out.

Incidently, the slat operation in the sim shows the slat closed at the 250 km/h climb while partly deployed at 180 km/h. A nice detail in the sim I think.

By the way, you can actually calculate the effects of turbulent flow for a complete a/c configuration at higher angles of attack accurately. All you have to do is solve Navier-Stokes equations. There is a small catch however, you probably need computing power similar to that of the NSA!

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Blottogg:
Sensei, you're right that the delta Cdi is exponential, and gets huge further from L/Dmax (the generic "ballooning" comment I made earlier.) Mine was just a back of the envelope approximation for a portion of the Cdi curve that was more linear (Amdahl's equation looks vaguely familiar, but my practical engineering is both rusty and in some areas obsolecent.) Still, I was pretty close.

I haven't tested low speed Cdi in the sim like Holtzauge has, but your numbers look a little high for 180 kph. If the Cd model is unchanged in 3.02, there may still be a problem. I had ascribed the reports of "prop-hanging 120kph climbs" to the Ai's simplified flight model. It may be due to an over-optimistic Cd curve, or that 1C is using only Cdi=Cl^2/(pi*A) (where A=aspect ratio) to calculate delta Cd, which remains linear plotted against Cl^2 (obviously) in Dr. Hoener's fig. 9 on page 14-11. The actual wind-tunnel data shows a Cd increasing non-linearly vs. Cl^2 at high Cl (for both the whole aircraft and the wing alone), due to parasitic and form drag going non-linear (separated, turbulent flow) at high AoA, like I mentioned earlier. If those changes aren't calculated, Cd will be too low, which may explain the discrepency. As my performance course professor told us "If you can come up with an accurate method to calculate the effects of turbulent flow, you'll be very rich." He also told us, "Yeah, I wanted to be an aerodynamicist, until I took my first graduate-level fluid dynamics course. I asked 'Where is that aeroelasticity syllabus' real quick after that."

You're also right that Cd's need to be an apples to apples comparision, using the same reference area S. That's one of the reasons I like Dr. Hoener's example, since it's all on the same sheet of music. <HR></BLOCKQUOTE>

Wallstein
12-06-2004, 03:05 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Holtzauge:
---"Maybe the above sounds negative. But do not get me wrong. This is the best sim around and I love it!

However, I plead to the designers to fix the induced drag issue. Get it as close to real life as possible! Make a good thing better!" <HR></BLOCKQUOTE>

I thank you for your writing and agree as well as support you! Let‚¬īs keep our thumbs up. The "BoB" is on the way...

Wallstein

JG5_UnKle
12-06-2004, 03:17 PM
Take a look at this chart compiled by JG5_JaRa it shows 2 speed comparison for climb rates and the % difference :
First chart sorted by low speed climbrate.
Second chart sorted by high speed climbrate.

http://aa.1asphost.com/pinetrees/crd2.JPG

JG5_UnKle
12-07-2004, 04:21 PM
BUMP

JG5_UnKle
12-07-2004, 04:22 PM
BUMP BUMP

ZG77_Lignite
12-07-2004, 04:42 PM
Great post gentlemen, I really miss these from the old days http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

And great chart Unkle, tell JaRa he did an excellent job their (a LOT of work!). It looks to my amatuerish eyes that they conform more closely to what Holtzauge is looking for. Personally I'm surprised at the marked differences, I would have assumed that they might have fallen in with a little less deviation between models.

Kwiatos
12-08-2004, 03:23 AM
bump
Any send e-mail to 1C?

JG5_UnKle
12-08-2004, 05:21 AM
Yes Lignite, little variation between some models. Overall I would expect there to be MORE of a marked difference between the low and high speed climbs.

The average is about 20% difference and I would expect it to be more to be honest. I just can't see a powerful WW2 fighter climbing at full power at 150-180 Kph without just about falling out of the sky with Torque alone.

Nevermind still putting respectable climbrates in at less than optimal speeds, all IMHO of course http://forums.ubi.com/images/smilies/53.gif

Blottogg
12-08-2004, 07:05 AM
Holtzauge, you sound like an aero engineer (or student.) You're probably more up-to-date on the current Navier-Stokes computations than I am (I graduated in 1986, and I wasn't exactly the class valedictorian then.) From my old texts, Navier-Stokes equations are good (I never thought I'd ever say that), but only if certain assumptions are made. Such as incompressible flow, and time-averaging turbulent flow. Accurately predicting turbluent flow "on the fly" wasn't possible back in the 80's, which is what I think my old prof was referring to. And yes, it takes a big-hairy computer to crunch those numbers in real-time. Not something to be done on a PC while time-sharing duties with the graphics and Ai engines.

From "An Introduction to Theoretical and Computational Aerodynamics" by Jack Moran (the professor who also taught us the course) p.213:

"...Navier-Stokes equations constitute only three equations in six unknowns. The problem of closure -- of supplementing the time-averaged Navier-Stokes equations so as to form a closed set, with enough equations to determine all the unknowns -- is not completely solved (as of 1983). A typical approximation is to assume a formula for the Reynolds stresses in terms of the average velocity and its first derivative, with constants adjusted to fit experimental data. Such approaches work very well for flows not too different in character from the ones that constituted the database."

JG5_UnKle
12-08-2004, 07:56 AM
Holy moly Blottogg!

Could you explain it for the mathematically challenged? (like me) and how it might be useful in FB?

JG5_UnKle
12-08-2004, 09:07 AM
*BUMPAGE*

Fred_77
12-08-2004, 12:13 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Kwiatos:
bump
Any send e-mail to 1C? <HR></BLOCKQUOTE>

I sent in my findings from the other thread in ORR to Oleg. He dosen't think that induced drag is a major factor in slow speed high AoA climbs. He didn't eloborate. Interesting observation though, since 3.01 and the prop efficencies were changed, the planes don't climb so well in these situations as they did in 3.0, ie the climb difference between Vy and buffeting is no longer a single digit %.

S!
Fred.

Holtzauge
12-08-2004, 12:18 PM
Good point about the torque. It's easy to just get lost in the aerodynamics and forget that you have to retain control as well! In a 109 you might still have some roll authority due to the slats keeping the flow attached over the ailerons but a P-51D at 170 km/h, with WEP and high AoA sounds like a handfull.

My hat is off to JaRa for compiling the very informative table of climb performance above. I have assumed that the figures are sea-level climb rates.

By looking at it, I think that a number of issues raise the warning flag.

1) The test shows parity in max climb speed for a "Beule" and the Pony. Now all the data I have seen indicates that there should be a clear advantage to the 109G-6 (I've seen figures like 4560 ft/min versus 3500 ft/min). It is also interesting to note the the Spit 9 is better in the sim than the 109-G6 in climb. This does not compute with the real life numbers I have seen.

2) The Jug apparently climbs like a bird in the sim! JaRa's test puts it in the same class as the 109G-6 and Mustang. The best I have seen for the P-47D is 3200 ft/min with the improved prop. The sim test equates to 4260 ft/min!

3) The 109K climbs a wopping 6020 ft/min in the sim. I have seen the number 4829 ft/min for the K.

With the reservation that I have not studied all figures in the test I get the general impression that the climb performance is to optimistic across the board.

Now if there was just a translation of all climb rates this might not be to bad. However, it seems to me that this is not so and that the climb relations between the a/c are off. This is bad I think because it will mean that to use your plane to the best advantage in the sim you need to adopt a different technique than in real life. And we do want this sim a as close to real life as possible don't we?

I also still think that the around 20% decrease that JaRa's test shows for decreasing the climb speed between from 260 to 170 km/h is to low.

So there seems to be 3 issues that are off in release 3.02:

1) The performance loss when deviating from best climb speed is to optimistic

2)The absolute climb rates are to high

3) The relation of best climb speeds between the planes is off

On my wishlist these issues comes way before new features. Hope this will be fixed in the next release.

Is there a way to petition for changes?

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by JG5_UnKle:
Yes Lignite, little variation between some models. Overall I would expect there to be MORE of a marked difference between the low and high speed climbs.

The average is about 20% difference and I would expect it to be more to be honest. I just can't see a powerful WW2 fighter climbing at full power at 150-180 Kph without just about falling out of the sky with Torque alone.

Nevermind still putting respectable climbrates in at less than optimal speeds, all IMHO of course http://forums.ubi.com/images/smilies/53.gif <HR></BLOCKQUOTE>

Holtzauge
12-08-2004, 12:42 PM
Yes, your observation is correct, in my youth I was happily engaged in the noble art of aeronautics, took a degree, got a job in the industry.But alas, ever since our neighbour in the east stopped churning out Su-27 Flankers and began producing more peaceful things like this sim I have been banished into the cursed IT industry where I am now bogged down.

I used to go to Farnborough and LeBourget and kick the wheels! Those were the days....

Seriously, I also took my degree in the middle eighties so it was intersting to read your piece on the equations! With Moore's law on computing power in mind, maybe we will fly a Navier-Stokes powered flight model when we get our retirement!

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Blottogg:
Holtzauge, you sound like an aero engineer (or student.) You're probably more up-to-date on the current Navier-Stokes computations than I am (I graduated in 1986, and I wasn't exactly the class valedictorian then.) From my old texts, Navier-Stokes equations are good (I never thought I'd ever say that), but only if certain assumptions are made. Such as incompressible flow, and time-averaging turbulent flow. Accurately predicting turbluent flow "on the fly" wasn't possible back in the 80's, which is what I think my old prof was referring to. And yes, it takes a big-hairy computer to crunch those numbers in real-time. Not something to be done on a PC while time-sharing duties with the graphics and Ai engines.

From "An Introduction to Theoretical and Computational Aerodynamics" by Jack Moran (the professor who also taught us the course) p.213:

"...Navier-Stokes equations constitute only three equations in six unknowns. The problem of closure -- of supplementing the time-averaged Navier-Stokes equations so as to form a closed set, with enough equations to determine all the unknowns -- is not completely solved (as of 1983). A typical approximation is to assume a formula for the Reynolds stresses in terms of the average velocity and its first derivative, with constants adjusted to fit experimental data. Such approaches work very well for flows not too different in character from the ones that constituted the database." <HR></BLOCKQUOTE>

Sig.Hirsch
12-08-2004, 12:52 PM
Excellent thread , and interesting observations (It's rare in these days)

BUMP!!!! http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

Oh , and thx JG5 Jara for your work http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif

k5054
12-08-2004, 05:04 PM
Control problems aside, there ought to be plenty of excess power when a WW2 fighter is at or just above the stall. In fact although the induced drag goes way up as you slow, the form drag gets small (yes, its Cd may increase with altered flow, but the speed decrease makes more difference). If you are at half of best climb, the drag will be 2.25 times what it is at best. But the power produced by the engine is the same, and the power used to overcome 2.25 times the drag is only 1.125 times what was need at best climb, because thrust increases inversely with speed. (That's why props are so much better than jets on the back of the drag curve, jets have constant thrust, props constant power). So, the power required to stay airborne doesn't increase much as you approach the stall. And the excess power doesn't decrease much either. The question is, can you use it? First, the prop has to turn it into thrust efficiently, which it doesn't do too well (but not all that badly either, or you would never take off!). Then you have to control that power with flight controls which are also struggling with several kinds of asymmetric forces, and don't work well at slow speeds.

But if you think this is all wrong, think about it from a different aspect. When you take off, you expect to climb away, and to accelerate to best climb speed, while you are bringing up flaps and gear. This would not be possible if you didn't have a lot of excess power even in a light plane, never mind a 2000hp fighter.

I don't think the sim models the control problems very much, and I guess it never will.

Wolf-Strike
12-08-2004, 08:36 PM
[\QUOTE]Originally posted by Holtzauge:

I also still think that the around 20% decrease that JaRa's test shows for decreasing the climb speed between from 260 to 170 km/h is to low.

So there seems to be 3 issues that are off in release 3.02:

1) The performance loss when deviating from best climb speed is to optimistic

2)The absolute climb rates are to high

3) The relation of best climb speeds between the planes is off

On my wishlist these issues comes way before new features. Hope this will be fixed in the next release.

Is there a way to petition for changes? [\QUOTE]

Will we ever see this global change that you write about above?What will it bring to the table for us.Will dogfights be dictated very strongly by what they were in real life......who has the altitude advantage????

Will this make it boring??Having someone dive down at you at high speeds will be very dipleasing as you try to gain alt.Youll just be bogging down and pushing nose down to get some speed which of course will make your alt lower stillhttp://forums.ubi.com/groupee_common/emoticons/icon_smile.gif))

Imagine though,the thrill of being the Butcher bird diving downhttp://forums.ubi.com/groupee_common/emoticons/icon_smile.gif))

Ugly_Kid
12-09-2004, 12:02 AM
There's another side to the coin. There was a similar discussion some time ago in ORR (E-Bleed stuff).

If you're unsure about the role of the thrust you have to go for no thrust tests. Testing glide ratio with engine off or engine on idle and with different speeds seems to indicate there is little difference at low speed vs. high speed performance, aerodynamically. There is now a possibility to pull the nose-up for a while and let the prop to stop completely. This way measure a steady glide of some 1000 m close to SL and you will notice.

It would appear as if the climb rate reduction is controlled with prop efficiency alone (well, to state this one would need to generate a similar chart for glide).

Wolf-Strike
12-09-2004, 10:34 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Testing glide ratio with engine off or engine on idle and with different speeds seems to indicate there is little difference at low speed vs. high speed performance, aerodynamically. <HR></BLOCKQUOTE>

Yes,I have also done some quick tests with engine off to test zoom climb abilities.It seems as if all the planes zoom climb to the same exact amount which shows that there is also no difference between planes in weight.So like you said,climb is more to do wuith prop efficiency.

It seems like they did the best they could with IL2 engine....I cant wait to see what BOB has in store for us.,I imagine that were gonna get a really big boost in "feel" of the different weight and HP ratings of the planes.

k5054
12-10-2004, 03:12 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR> Yes,I have also done some quick tests with engine off to test zoom climb abilities.It seems as if all the planes zoom climb to the same exact amount which shows that there is also no difference between planes in weight.So like you said,climb is more to do wuith prop efficiency. <HR></BLOCKQUOTE>

If there was no drag, all aircraft would zoom to the same height independent of weight. The differences in zoom climb between RL aircraft are down to the energy used in changing direction to an upwards flightpath, and the thrust minus drag at various stages in the climb. Not all aircraft should zoom to zero-speed height, but the tactical situation should probably be the determing factor, not the different a/c physics. Good zoom climb should be expected from a/c which have a good speed for their power, that is low-drag aircraft, and if they are also dense, that's good too. The same ballistic coefficient applies as in the dive, indeed the physics for a dive, the equation you use, is the same as for a zoom climb, its just the angle is different. Any flight sim ought to be using just one equation for climb, dive and level flight, with different CD, areas and weight etc inserted for each a/c type.

I'd be interested in any glide ratio tests, but IIRC we ran into the problem of what the prop is doing, depending on the pitch it was set for.

karost
12-10-2004, 05:05 AM
I don't have knowledge about airplane engineering but would like to learn/know about 1G stall speed ( same direction from Ugly_Kid point about "glide ratio with engine off or engine on " ) but for me I mean ratio when we lose altitude "vertical speed fall when stall" .... something like when I try to make a glide landing three point on the carrier but I miss by pull alot AOA for glide speed then suddenly the plane drop fast and hit the carrier... ( but this sim seem more easy http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif )

S!

Holtzauge
12-10-2004, 12:22 PM
I do hope we will see this change. At least the one about the relative climb ratios. I think it is as important as any other aspect of the performance.

As a blatant example, I do not think the s/l climb rate of the Spitfire 9 should be better than the 109G-6 in the sim!

If the climb ratios are off, we might as well have the same situation with turn-rates and have P-47's out-turn Spitfires!

Since I prefer to fly energy tactic planes like the 109 and 190 because I like the style of flying I expect the sim model these planes in a way so I can fly them like they were intended.

Now an important part is the climb rate. This can be used to regain energy faster than your opponent. I can also use it to climb my way out of trouble by the spiral climb used by the LW "experten". If you are in a 109G-6, the climb rate is the ace in your sleeve.

Now the climb rate issue at lower speeds that was discussed earlier in this thread fades in importance when you consider that the relative climb ratios between the planes are as bad as they seem.

I think this is a fundamental flaw in the modelling which needs to be fixed.

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Wolf-Strike:

Will we ever see this global change that you write about above?What will it bring to the table for us.Will dogfights be dictated very strongly by what they were in real life......who has the altitude advantage????

Will this make it boring??Having someone dive down at you at high speeds will be very dipleasing as you try to gain alt.Youll just be bogging down and pushing nose down to get some speed which of course will make your alt lower stillhttp://forums.ubi.com/groupee_common/emoticons/icon_smile.gif))

Imagine though,the thrill of being the Butcher bird diving downhttp://forums.ubi.com/groupee_common/emoticons/icon_smile.gif)) <HR></BLOCKQUOTE>

at_evans
12-10-2004, 01:12 PM
Bump http://forums.ubi.com/images/smilies/354.gif

Blottogg
12-10-2004, 03:14 PM
I'm back (had computer problems for a day or two.)

Holtzauge, you're starting to scare me. I think we had the same job, on opposite sides of the pond.

JG5_UnKle, I won't pretend that I understood all of Prof Moran's book (either now or when I took his course) but I pulled that quote to illustrate that Navier-Stokes equations have more variables than equations to solve them, and the practical solution is to assume some of the variables are constant (incompressible flow, and turbulence that is averaged over time instead of trying to calculate each vortex as it is shed), and to take experimental data, and massage the equations to fit. It would be nice to be able to predict vortices accurately, from a stability and harmonic standpoint. If that were possible, F-18's might not need external reinforcement of their vertical tails, to keep them from vibrating off at high AoA. But it isn't, and they do.

Good point about control at low speed. Since the sim models torque rather gently, and P-factor not at all, low speed control tends to be better in the sim than real life. I hadn't considered the possibility of minimum climb speed being control limited instead of power, but that is probably the case IRL for these high torque planes (with exceptions such as the P-38.)

Lacking climb curves for these aircraft (the experimental data Prof Moran mentioned in getting the calculated data to match actual aircraft performance), I'm not real confident making low speed climb calculations. One thing I will mention is that while JaRa's data is appreciated, and painstakingly even on speeds from one aircraft to another, it doesn't necessarily reflect each aircraft's best climb speed. Where an aircraft's Vy occurs in relation to the two sample speeds will affect the difference in it's two sampled climb rates. The low wing loaded planes will tend to have Vy's lower than the high wing loading planes. But as Holtzauge mentions, a climb rate 2000 ft/min higher than the historical maximum is wrong, regardless of the speed at which it occurs.

JG77Von_Hess
12-10-2004, 06:43 PM
http://forums.ubi.com/images/smilies/16x16_smiley-happy.gifThis thread needs a big bump!

Regards.

VH.

S.taibanzai
12-11-2004, 12:16 AM
wel i am not a engineer but this must be reported to Maddox and be fixed


thx for starting this topic

JG5_UnKle
12-11-2004, 02:30 AM
RGR That Blottogg http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif especially the "best" climb speed element. JaRa has done many tests to determine best climbs and as you know this was just to compare a generic "low" speed climb to a "close to optimum" climb http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif. It really does show some of the limitations of the engine IMHO.

If it can be changed at all (I have my doubts) I don't think we will see it in this sim. I feel the IL-2 engine has been pushed way beyond it's original concept (IL-2 mud moving sim) into something that really begins to show it's limitations.

Proper torque, P-Factor and a more complex stall model are hopefully some of the things to look forward to in BoB along with a smaller planeset which I think is another thorn in the side of the whole series.

Anyway nice thread from all involved http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif

Holtzauge
12-11-2004, 03:43 AM
Excuse me for my ignorance but BoB? What you describe sounds good. Could someone who knows what this is elaborate a bit what BoB is and where one can find some info about it?

I'd would rather fly a sim with a limited number of planes as long as they are modelled correctly.

I'm really devastated by the info that the Jug and Pony climbs as well as the "Beule" in this sim. I wonder what Heinz Knoke (father of the spiral climb evasion) would have said about this.

I think he would have been rather suprised to see a Spit 9 creep up and blast his 109G-6 out of his spiral climb evasion.....

If this is not fixed I'm gonna start looking at alternatives. Mayby BoB, whatever it is, is it?

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by JG5_UnKle:

Proper torque, P-Factor and a more complex stall model are hopefully some of the things to look forward to in BoB along with a smaller planeset which I think is another thorn in the side of the whole series.

Anyway nice thread from all involved http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif <HR></BLOCKQUOTE>

Blottogg
12-11-2004, 05:50 AM
BoB is the Battle of Britain simulator 1C is working on for a 2005-ish release. We saw some development work posted on the sight before PF kind of hip-checked it to the side. BoB will have a new game engine, which should be more detailed and accurate. Now that PF has shipped, we're hoping to get more info on BoB from Oleg and crew.

JG5_UnKle
12-13-2004, 01:08 AM
Yes, sorry I should have elaborated more BoB is short for Battle of Britain and promises more fidelity and greater detail for a smaller planest.

We hope anyway http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif

JG77Von_Hess
12-13-2004, 07:43 AM
http://forums.ubi.com/images/smilies/16x16_smiley-indifferent.gifI sertenly hope you are right Unkle, but compare the interrest in this thread with this whine o rame. http://forums.ubi.com/eve/forums?a=tpc&s=400102&f=26310365&m=9951034342

Thats a sad indication of what people really wants.

My hopes are still high for BOB, a few hifi modelled planes in a hifi envioment is what i really dream for.

Regards.

VH.

BlitzPig_DDT
12-13-2004, 10:32 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by JG77Von_Hess:
http://forums.ubi.com/images/smilies/16x16_smiley-indifferent.gifI sertenly hope you are right Unkle, but compare the interrest in this thread with this whine o rame. http://forums.ubi.com/eve/forums?a=tpc&s=400102&f=26310365&m=9951034342

Thats a sad indication of what people really wants.

My hopes are still high for BOB, a few hifi modelled planes in a hifi envioment is what i really dream for.

Regards.

VH. <HR></BLOCKQUOTE>

I see. The US shouldn't be allowed any late war or '46 planes, but everyone else should.

People with opinions like that just plain suck.

We have the Ta-152, and a few '45 VVS crates, as well as the A6M7 and Ki-84Ic. Get over yourselves.

(although, it's arguable that we already have the F4U-4 in every single corsair we currently have. That thing flies in a **** vacuum with moon gravity.)

JG77Von_Hess
12-13-2004, 11:06 AM
http://forums.ubi.com/images/smilies/16x16_smiley-surprised.gifBlitzpig re read my post.. just switch on your brain this time, compare the numbers of replies/views to this thread with the one im linkin to, Popular demand=$=endproduct.. simple

You are *****ing over 1 plane!!!
Im talkin about the complete invioment we fly around in!!

As for the Suck part, no thanks for the offer, but im Hetro.

If this dosent explain things, then ok.

VH.

BlitzPig_DDT
12-13-2004, 11:51 AM
Nice turn around attempt, but no dice.

Nobody wants the environment (as you put it) to stay hosed. You didn't point out the requests for any Axis, or even UK or VVS superplane, just the US one.

As for reading....If I had your, obviously superior, level of intelect and reading ability, I'd be calling you JG77.

JG77Von_Hess
12-13-2004, 12:14 PM
http://forums.ubi.com/images/smilies/16x16_smiley-very-happy.gifThanx for the kind words, Du fatter stadig ikke en skid din lille fedtede tr√¬¶kker-dreng i tunolie http://forums.ubi.com/images/smilies/59.gif

Regards.

VH.

effte
12-15-2004, 05:39 AM
Why did you post the "you still don't understand ****, you small greasy..." etc part in Danish, Von_Hess?

Rude and childish, and an abuse of the language.