Eagle Dynamics FM discussion (WWII Aircraft)

It's very plausible. To calculate exact rate of climb with accuracy higher than, say, 10% you have to know exact jet thrust and radiator drag. Even not saying a word about exact prop efficiency measurement.

I still have not seen any evidences that they took all this factors in account.

As far as propellor effiency goes, I think that was taken into account for two reasons: one, they specifically wanted to compare the effect of the new 12199 propellor to the standard 12159 type, both of which were flight tested in a 109K at around the same time, albeit at lower power settings. Also the Mtt PB calculation specifically mentions that the difference between the standard/serial and project prop differences in level speed, but are (assumed to be?) equal in climb.

Secondly, a few moths back the same Mtt PB did calculations for gondola equipped G-14 and G-14/ASM as well, the latter having the same 12159 type prop as the serial 109K, and it was calculated as being some ~1 m/sec slower in climb that the one with the previous 12087 prop. With the difference in weight and power between the two 109G types are otherwise negligable (if anything, the low/med alt G-14 had more drag due to the larger bulges), this must come down to the poorer low altitude/speed efficiency of the broad blade 12159 prop

So in that test in all likelihood they calculated for the difference between the propellers efficiency for the 109K later on.

Radiator drag seems is probably accounted for, since the radiator opening is also specified as half open until VDH, then slowly closing and I see no reason why to mention this if its not factored in. Since the effect of this on climb rate was very significant and have varied considerably with radiator opening and altitude, as directly evidenced by Mtt 109G flight trials in mid-1944 and further hinted by Finnish and British trials of G-2 - and for this reason should be carefully followed in any DCS proof test to obtain the same testing conditions.

Exhaust and radiator jet thrust may have been not accounted for, I agree - this would also explain why there are so many Mtt climb calculations around for 109G that tend to yield lower climb rates than the actual flight tests! Also, I have yet to see any exhaust thrust figure for the DB 605 - the only one there is for the export 601Aa. It would be little wonder that they did not factor that in, if DB did not supply that to the engineer doing the calculations at Mtt PB, how is he supposed to do that without data...?

I can guess the method that allowes you to have your calculations close to pessimistic estimations...

 

As you mentioned, you use "A propeller" charts for Ct, Cm and efficiency. But is solidity factor of your propeller is close to the target propeller?

 

If not, the blades of your prop can not absorb the power at desired rpm and you have to use higher pitch, so the efficiency lowers.

 

I can trust to the method used in DCS because it gave exact results using only prop properties, engine model properties and airframe properties. Gathered together and with added radiator drag and jet thrust these parameters gave the result very close to test flights.

 

Using the same methods gives the results we can see now for 109: approxomately 25-26 m/s.

 

Anyway, I appealed to your common sense and asked if Me-109 is so bad that p-51 wins several hundreds kg for the same climb with the same power?

 

What will be your answer except "my calculations tells"?

 

Just two figures, please: what prop efficience you got for P-51 and 109 K max power climb conditions? I mean 1800 hp for both.

Beginning with the propeller modelling, I use two different methods, both of which certainly handle propeller solidity and also account for efficiency as a function of disc loading, advance ratio and tip Mach effects. One of which I have described in detail here:

http://forums.eagle.ru/showthread.php?t=127918&highlight=propeller

First of all I don’t think my calculations are pessimistic: I have modeled more than 20 aircraft over the period of 10 years and tuned this continously based on IRL data. The reason I think I get pretty close is simply that the results I get out of the simulations usually tab pretty well with historical data. When I say historical data, I mean an average of the data available. I tend to trust data like Allied pilot’s notes of pilot operating charts and German Kenblatts rather than individual flight tests. Why? Simply because tests even at the best of times data have a normal variance around a mean. There area also measurement errors, data taken at different tempertures etc. In addition, using individual tests opens up the field to cherry-picking, both high and low outliers depending on the agenda.

So yes, IMHO you can rely on tests but then you have to compile a statistically significant number, convert them all to the same weight, power, standard atmosphere, ensure that they include position error and Mach corrections etc. etc.

Getting back to the Me109K4: To me 25-26 m/s at 1.8 ata is too optimistic based on my modelling. True, there is not much IRL data to go on here either but I have modelled Steig & Kampfflesitung climb rates as well and they tab well with Kennblatt data on the K4. I also get quite close to the climb times here which are good indicators since they integrate the climb rates and any large deviations is bound to come up here. Here the DCS FM model seems too optimistic since both the K4 and Dora have way to good climb times compared IRL data.

In addition, while DCS may be the latest and best simulator, I think there has been a general consensus about what figures to target in earlier high end siumulators like IL-2, Aces High etc. and while these may not have as an advanced FM as DCS, I have respect for their developers and the tuning they have performed over the years. To the best of my knowledge these developers have roughly the same opinion as me as to what the IRL figures to target should be and to me it looks like DCS as it is currently tuned a bit too optimistically when it comes to climb performance. How optimistic? Well I believe that my simulation numbers below are closer to IRL performance:

Sea level climb rate Fw190D9 at 1.8 ata MW50 W=4270 Kg around 22.5 m/s

http://forums.eagle.ru/showpost.php?p=2347279&postcount=64

Sea level climb rate Me109K4 at 1.8 ata MW50 at W=3362 Kg around 23.3 m/s

http://forums.eagle.ru/showpost.php?p=2397641&postcount=9

I’m not saying I have to be right. You can of course continue to claim that my figures are too pessimistic. However, you appeal to my common sense and I will do the same to you: What is more plausible? All previous high end flight sims targeting the wrong figures (which generally agree with mine) and IRL data from firms like Focke-Wulf and Messerschmitt not containing exhaust thrust and making no mention of it even though we both agree it has a significant impact on performance?

Or is the current modelling of the K4 and Dora in DCS simply too optimistic? Which is more plausible?

Finally, please note that this input is done with the best intentions and should not be interpreted as criticism: DCS is IMHO the best WW2 flight sim out there. I'm just trying to provide input which I believe will make it even better :)

@ Pilum. As an interested albeit uneducated bystander I am curious why you said this; "I think there has been a general consensus about what figures to target in earlier high end simulators like IL-2, Aces High etc."

They don't appear to be high end simulations to me :). Particularly IL2 .. with the possible exception of CLOD.

EDs FMs are not scripted and don't use tables like older ones have in the past. They are doing things with their FMs that nobody else has done so far that I have seen.

Rise of Flight does use rough fluid physics calculations (

So, I'd agree about "apples & oranges," since RoF isn't WWII, and none of the other ones use fluid physics, with the possible exception of CloD (which I haven't yet examined). Not much point in saying "but the older games did it such-and-such a way" to support the idea of making it so in DCS.

Just want to clear up some confusion: When I referenced other sims as high end I mean in terms of ambition: They explicitly targeted historical accuracy and in that ambition they strove to get as close to IRL performance as possible and I can't recall any of then defending figures that were far away from historical data that's all.

Some people here seem to think that just because one is using a more advanced FM this means it is inherently more accurate and therefore whatever results it produces must be right which is wrong. At least if the issue is the measurable climb rate and climb times like we have been discussing here.

Whatever computer simulation you have, this is a model of reality and is parameter controlled and a FM is no exception and it will give no better model of reality than the tuning it receives. So a CFD FM could be way off while a good old scripted 6DOF FM can peg down basic flight performance much better. It all depends on the parameter settings.....

I currently make a living running simulations in the telecom industry so I'm quite familiar with the issue and as a collegue of mine (who has a doctorate in Physics) jokingly puts it "Tell me what results you want and I will simulate it". :smilewink:

Jokes aside: Good to know the climb rate is being tuned and let's see what we get!

When I referenced other sims as high end I mean in terms of ambition: They explicitly targeted historical accuracy and in that ambition they strove to get as close to IRL performance as possible and I can't recall any of then defending figures that were far away from historical data

Well, there was that one time Mr. Maddox refused to acknowledge that the Browning M2 dispersion pattern was three times larger than RL test documents stated it should be, under any given conditions. Yes, that's a 300% margin of error; he apparently found that acceptable, given that he insisted that his insane figures were correct, all the way up to the point that the publisher stepped in & ordered him to fix it.

EDs FMs are not scripted and don't use tables like older ones have in the past

Technically speaking, i'd be really surprised if they didn't still use lookup tables for efficiency. There are a lot more tables involved because the airframe is divided into sections that each have separate characteristics. The magic is in obtaining the correct table values.

We do not use a lookup table for prop efficiency. Now you can start to be surprized. If you read the material on DCS site how Ka 50 rotor is modelled you would not be so surprised.

Not prop efficiency. Computational efficiency of obtaining lift force of the respective body/rotor/wing segment. :)

You had me scratching my head there for a moment. ;)

I only know / have used two other sims that use the same approach followed by DCS / ED - X.Plane and IL2 BoS. As far as I know, RoF also used BeT....

Whatever model is used, doesn't mean it is a certificate for accuracy / perfection in the obtained performance. X-Plane 10, IMO, does a lousy job in this area, and it's "ALL BET"....

OTOH, there are aspects of flight performance that I have never seen so well replicated in a GA sim as I did in ELITE, for it's fleet of GA aircraft, and ELITE uses pure table-based modelling !

So, it all depends, IMO, in what data is used for the models and how the necessary smplifications cut the full CFD model, or even how I is coded...

Honestly, I think there are aspects of DCS in the modelling of ( powerful ) prop aircraft that I find unique among all the sim I have used, but there are also some exaggerations, IMHO, like the way P-factor is modelled, accounting for instance for that pitching up when light trim adjustments are operated, even at cruise... or that wobbling of the nose of an aircraft falling at near 90º with it's engine / prop fully stopped...

like the way P-factor is modelled, accounting for instance for that pitching up when light trim adjustments are operated, even at cruise...

or that wobbling of the nose of an aircraft falling at near 90º with it's engine / prop fully stopped...

PS / EDIT: You say the engine is broken. Any other damage you noticed? Because the wobble can result from all kinds of things actually.

Beginning with the propeller modelling, I use two different methods, both of which certainly handle propeller solidity and also account for efficiency as a function of disc loading, advance ratio and tip Mach effects. One of which I have described in detail here:

 

http://forums.eagle.ru/showthread.php?t=127918&highlight=propeller

 

First of all I don’t think my calculations are pessimistic: I have modeled more than 20 aircraft over the period of 10 years and tuned this continously based on IRL data. The reason I think I get pretty close is simply that the results I get out of the simulations usually tab pretty well with historical data. When I say historical data, I mean an average of the data available. I tend to trust data like Allied pilot’s notes of pilot operating charts and German Kenblatts rather than individual flight tests. Why? Simply because tests even at the best of times data have a normal variance around a mean. There area also measurement errors, data taken at different tempertures etc. In addition, using individual tests opens up the field to cherry-picking, both high and low outliers depending on the agenda.

 

So yes, IMHO you can rely on tests but then you have to compile a statistically significant number, convert them all to the same weight, power, standard atmosphere, ensure that they include position error and Mach corrections etc. etc.

 

Getting back to the Me109K4: To me 25-26 m/s at 1.8 ata is too optimistic based on my modelling. True, there is not much IRL data to go on here either but I have modelled Steig & Kampfflesitung climb rates as well and they tab well with Kennblatt data on the K4. I also get quite close to the climb times here which are good indicators since they integrate the climb rates and any large deviations is bound to come up here. Here the DCS FM model seems too optimistic since both the K4 and Dora have way to good climb times compared IRL data.

 

In addition, while DCS may be the latest and best simulator, I think there has been a general consensus about what figures to target in earlier high end siumulators like IL-2, Aces High etc. and while these may not have as an advanced FM as DCS, I have respect for their developers and the tuning they have performed over the years. To the best of my knowledge these developers have roughly the same opinion as me as to what the IRL figures to target should be and to me it looks like DCS as it is currently tuned a bit too optimistically when it comes to climb performance. How optimistic? Well I believe that my simulation numbers below are closer to IRL performance:

 

Sea level climb rate Fw190D9 at 1.8 ata MW50 W=4270 Kg around 22.5 m/s

 

http://forums.eagle.ru/showpost.php?p=2347279&postcount=64

 

Sea level climb rate Me109K4 at 1.8 ata MW50 at W=3362 Kg around 23.3 m/s

 

http://forums.eagle.ru/showpost.php?p=2397641&postcount=9

 

I’m not saying I have to be right. You can of course continue to claim that my figures are too pessimistic. However, you appeal to my common sense and I will do the same to you: What is more plausible? All previous high end flight sims targeting the wrong figures (which generally agree with mine) and IRL data from firms like Focke-Wulf and Messerschmitt not containing exhaust thrust and making no mention of it even though we both agree it has a significant impact on performance?

 

Or is the current modelling of the K4 and Dora in DCS simply too optimistic? Which is more plausible?

 

Finally, please note that this input is done with the best intentions and should not be interpreted as criticism: DCS is IMHO the best WW2 flight sim out there. I'm just trying to provide input which I believe will make it even better :)

In the course of multiple discussions over several games FM it has been pointed out to you that your estmates are consistently pessimistic for certain aircraft. The conclusion myself and others reach is your modeling of the propeller. That is why I gave you the NACA work on disc theory.

Yo-Yo is trying to help you out. Please listen.

Doesn't necessarily have to do with the P-factor at all, but rather with how coarse an input device is used. F.ex. how large a step does, say, 1 short keyboard press produce for the sim to interpret/process?

 

Nose wobbling when the AC is in a perpendicular dive towards the earth? Haven't seen this, AFAIR at least. Which aircraft?

 

PS / EDIT: You say the engine is broken. Any other damage you noticed? Because the wobble can result from all kinds of things actually.

I've experienced this with the P51d, and then with the K4 too..

Damage is "simply" engine damage, due to overheat for instance... Engine stops, prop too, then, rollover to inverted and start a descent... The nose will begin to circle ( clockwise if I can still recall correctly, since I no longer use DCS or any other combat sim... )

In the course of multiple discussions over several games FM it has been pointed out to you that your estmates are consistently pessimistic for certain aircraft. The conclusion myself and others reach is your modeling of the propeller. That is why I gave you the NACA work on disc theory.

 

Yo-Yo is trying to help you out. Please listen.

No, you have it backwards: I agree with the first part: That over the years it has become clear that we disagree on a lot of issues. However, I think we draw very different conclusions from this fact and what the problem might be. :smilewink:

Anyway, if you have anything constructive to contribute then post it but please spare us your analysis of who understands what because that will lead nowhere.

It is a matter of what parameters stated in documents you want to presume right. Speaking about the tests you omitted, that generally all tests excluding brief estimation tests ALWAYS generalise results to MSA condition. Even in NII WWS :) during the war. Otherwise, the results of this test costs NOTHING. It's a rule of thumb, and most of reports that do not generalise atmosphere condition mention if they are not standard.

Moreover, some reports even mention that airplane mass is converted to constant.

Theoretical calculations, in its turn, could have many possible levels of detailing the data and methods they used. That's why 10% of accuracy was stated for German climb rate estimations, for example.

 

A complicated FM methods including complicated models of aitframe, engines, props, coolers, etc have a significant advantage: if you can use them right they can predict aircraft performance even if some basic parameters are not available. For example, starting to model A-10 we have no valuable lift / drag figures for it. WE have only performance curves and our model of the TF-34 engine. Using these model as a trusted source of the thrust curves made possible to get polars at different M-numbers. Later, we got exact data, and I was surprised how accurate our estimation were.

 

In its turn, data set for the engine model suffered from the lack of altitude-M curves for the thrust, and the model was tuned using only SL M=0 data and fuel consumption for different speed/altitudes. All the rest of the work does the true thermodynamics model itself.

 

Something similar was for P-51 performance.

 

That's why I trust these methods more than calculations of 40'... especially if I do not know what methods and data they used, so I prefer to have only measured values (but regarding their potential accuracy).

 

But DCS FM methods do not allow to bend performance as you wish as less complicated models allow...

Well, concerning accuracy, I have the same experience with my model: In the beginning (about 10 years ago), the assumptions I made and parameter settings I had in my C++ simulations did not always match IRL data so well. This was partly due to parameter settings but also due to that some effects which turned out to be important were not included in the beginning. However, simulation being an iterative process, this was ironed out and the number of lines of code grew.

Consequently, when I today add a new plane, I just add the basic input parameters and the results tab well with IRL data from the start. Like you, I am pleasently surprised that when new data arrives that I did not have before, this usually confirms my modelling based on the limited set of input parameters I had when I did the modelling. As an example, one aircraft I modeled based on the only data I had at the time (speed and climb) turned out to have the same acceleration and turn performance as measured in flight tests. Another example is actually the K4: I did the basic modelling based on climb and speed data. Then later on I got hold of diagram data calculated by Messerschmitt for the momentaneous turn performance of a K4 at 6 Km altitude which was quite close to my estimate.

So in the same sense as you are reassured by the A-10 data, I have over the years been reassured by the correlation between my results and the IRL data I have gathered so this is the basis why I think I’m pretty close in my estimates now for the Dora and K4.

On the subject of data from the 30’ and 40’ we seem to have different opinions: I think most calculated data by respectable firms is pretty close. At least the calculated engineering data that is unadjusted by marketing and management in order to secure contracts. So if anything, the calculated estimates you see will most likely be on the high, not low end if they have received TLC from marketing and management. This is why I doubt the calculated K4 and Dora climb data by Messerschmitt and Focke-Wulf we have discussed previously would be on the low end.

Getting back to wartime calculated engineering data: I have the greatest confidence in their estimates: It is the same here: Just because we have computers and CFD today does not make us more accurate: As the saying goes in simulation: crap in gives crap out so it all depends on the tuning. So a 10% deviation may be just as applicable to CFD or in flight tests due to measurement errors as for calculations. I believe you can get very good results using the slide rules and nomogram engineers used at the time. Again, it all depends on the assumptions and simplifications you make in your calculations and I have a humble view here: The guys that did the diagrams and figures we today post were professionals in their field and they did this on a daily basis and they did it for a living. So I think we should be very careful before we dismiss data from the 30´and 40's just because it was calculated. And before someone post a calculation from the 40’ that is obviously wrong to prove I’m wrong, I am again talking about a judicious compilation of data to find the mean, i.e filtering out outliers, be they flight measurements or calculations.

Finally, sorry for teasing: But if you are convinced that the engineers back then compensated for atmospheric effects and even aircraft mass without mentioning it why would they treat exhaust thrust any different? Because performance results without exhaust thrust mean NOTHING! :smilewink:

Anyway, if you have anything constructive to contribute then post it but please spare us your analysis of who understands what because that will lead nowhere.

Fascinating read, guys. I had to put my thinking cap on. :)

Interesting read!

Pilum do you publish your simulation results anywhere?