Low speed propeller efficiency too optimistic in DCS?

I actually bought DCS for the jet modules initially but now with the coming addition of the Fw-190 D9 I decided to get the P-51D module and try it out to see how this sim fares in comparison to other WW2 flight sims.

First of all let me say I'm impressed by this simulation. I have over the years tried a number of different flight sims and my impression of the P-51D flight modelling in DCS so far is favourable.

That being said there is one thing I have noticed that IMHO sticks out and that is the low speed behavior: I started noticing this since the AI hang on the propeller in a way that looked really strange. This brings back memories of the early versions of IL-2 in which it was also possible to hang by the prop in low speed scenarios.

In DCS this property does not seem to be limited to the AI only because when I started to explore the low speed regime myself I found that this was also part of the player flight model. My suspicion is that the low speed propeller efficiency modelling, or more precisely the low advance ratio (J=v/(N*D)), modelling is a bit on the optimistic side.

To test if this was true, I set up a test scenario to test climb rates at different speeds. As the benchmark, I set climb at 61" 2700 rpm with 175 mph IAS. I then did another test with 120 mph IAS. I took the climb time from 1 to2 Km altitude out of TACVIEW and got climb time 1 min 2 s and 1 min 5 s respectively. Both tests with 9542 lb initial weight. (Track files attached)

Note that the difference in climb time in DCS is just 5%. I believe that this is to low. Granted, I don't have climb charts for the P51 showing the climb rate as a function of IAS but based on IRL data of other aircraft this should be more bell shaped with the climb rate being lower at both the higher and lower IAS speeds. IMHO it looks like this is true in DCS for speeds higher than the optimum climb speed but that the climb speed does not suffer as much as it should when going as low as 120 mph. (BTW: the reason I did not do 67" 3000 rpm test in DCS is because cooling is insufficient and I blew the engine in the 120 mph IAS climb scenario).

I have over the years developed a C++ simulation program that I have tinkered with and which gives rather good correlation in most cases and which I use to evaluate different flight sims from a performance perspective. However, I don't have the Packard Merlin 1650-7 boost 61" 2700 rpm case modelled so I had to compare the WEP 67" 3000 rpm case. With this boost I get climb times from 1 to 2 Km at 1 min 2 s for 175 mph and 1 min 16 s for 120 mph, i.e. a difference of 23% at 9542 lb weight. In addition, looking at tracks in general, it looks like this issue affects both the P51D and Fw190D9 and IMHO the low speed helicopter antics of the AI currently looks weird.

So I would be interested to hear the developers view on this and if they agree that the low speed propeller efficiency modelling right now is a bit on the optimistic side and if they have any plans to address this?

Pilum aka Holtzauge

Tacview-120mph.txt.zip

Tacview-175mph.txt.zip

I actually bought DCS for the jet modules initially but now with the coming addition of the Fw-190 D9 I decided to get the P-51D module and try it out to see how this sim fares in comparison to other WW2 flight sims.

 

First of all let me say I'm impressed by this simulation. I have over the years tried a number of different flight sims and my impression of the P-51D flight modelling in DCS so far is favourable.

 

That being said there is one thing I have noticed that IMHO sticks out and that is the low speed behavior: I started noticing this since the AI hang on the propeller in a way that looked really strange. This brings back memories of the early versions of IL-2 in which it was also possible to hang by the prop in low speed scenarios.

 

In DCS this property does not seem to be limited to the AI only because when I started to explore the low speed regime myself I found that this was also part of the player flight model. My suspicion is that the low speed propeller efficiency modelling, or more precisely the low advance ratio (J=v/(N*D)), modelling is a bit on the optimistic side.

 

To test if this was true, I set up a test scenario to test climb rates at different speeds. As the benchmark, I set climb at 61" 2700 rpm with 175 mph IAS. I then did another test with 120 mph IAS. I took the climb time from 1 to2 Km altitude out of TACVIEW and got climb time 1 min 2 s and 1 min 5 s respectively. Both tests with 9542 lb initial weight. (Track files attached)

 

Note that the difference in climb time in DCS is just 5%. I believe that this is to low. Granted, I don't have climb charts for the P51 showing the climb rate as a function of IAS but based on IRL data of other aircraft this should be more bell shaped with the climb rate being lower at both the higher and lower IAS speeds. IMHO it looks like this is true in DCS for speeds higher than the optimum climb speed but that the climb speed does not suffer as much as it should when going as low as 120 mph. (BTW: the reason I did not do 67" 3000 rpm test in DCS is because cooling is insufficient and I blew the engine in the 120 mph IAS climb scenario).

 

I have over the years developed a C++ simulation program that I have tinkered with and which gives rather good correlation in most cases and which I use to evaluate different flight sims from a performance perspective. However, I don't have the Packard Merlin 1650-7 boost 61" 2700 rpm case modelled so I had to compare the WEP 67" 3000 rpm case. With this boost I get climb times from 1 to 2 Km at 1 min 2 s for 175 mph and 1 min 16 s for 120 mph, i.e. a difference of 23% at 9542 lb weight. In addition, looking at tracks in general, it looks like this issue affects both the P51D and Fw190D9 and IMHO the low speed helicopter antics of the AI currently looks weird.

 

So I would be interested to hear the developers view on this and if they agree that the low speed propeller efficiency modelling right now is a bit on the optimistic side and if they have any plans to address this?

 

Pilum aka Holtzauge

Im not a dev, but I did notice your inital zoom in the 120mph track is a good 2000fpm greater than in the 175mph. You also started 10 knots faster.

Im not a dev, but I did notice your inital zoom in the 120mph track is a good 2000fpm greater than in the 175mph. You also started 10 knots faster.

True, but I start off at 500 m and zoom to reduce speed but by the time I reach 1000 m I believe I'm pretty stabilized. So the actual measurement is from 24 s into the track until 1 min 29 s, that's where 1 min 5 s comes from.

It might be hard to get accurate climb times in such a short time as 1km to 2km. Your error will overlap the real difference in climb times.

I also know that that the best advance ratio for efficiency depends on propeller pitch, and we're dealing with a constant speed prop. How are you accounting for this?

Interesting post!

It might be hard to get accurate climb times in such a short time as 1km to 2km. Your error will overlap the real difference in climb times.

 

I also know that that the best advance ratio for efficiency depends on propeller pitch, and we're dealing with a constant speed prop. How are you accounting for this?

 

Interesting post!

I'm assuming a constant speed prop and that the revs are maintained at 3000 rpm in the C++ simulation.

I went back and checked my track and I did notice that that USARStarkey has a point and I do have a bit of a zoom initially and that was a bit sloppy but I guess anyone can reproduce the climb test and make up their own opinion of how 120 compares to 175 mph IAS climb. To me 120 mph climb rate seems to optimistic and I for sure think the prop hanging you can do today seems weird.

I'm assuming a constant speed prop and that the revs are maintained at 3000 rpm in the C++ simulation.

 

I went back and checked my track and I did notice that that USARStarkey has a point and I do have a bit of a zoom initially and that was a bit sloppy but I guess anyone can reproduce the climb test and make up their own opinion of how 120 compares to 175 mph IAS climb. To me 120 mph climb rate seems to optimistic and I for sure think the prop hanging you can do today seems weird.

I know for certain the ai can hang on the prop, amonst other things, far too often. This has been well documented. As for the human pilots, Im not sure I follow you. Note, I am not saying you are wrong at all, just that I'm not noticing the issue. I would like to point out though that at very low speed control surface effectiveness goes down alot more in DCS than in other sims, making it harder to aim etc if you were hanging on the prop. I have heard alot of ww2 accounts of fighter hanging on props at low speeds and rpms during turns, but I dont know the exact numbers so I cant really say one way or another. If you redid your test and compared it to some real life numbers it would be more revealing I think. at 67 inches 3000rpm, and P-51 at 9700lbs should get a max climb around sea level of 3600ft per min.

A constant 3000rpm is not the same as a constant propeller pitch.

Again, the primary problem as I see it is not in the absolute values as such but that the DCS propeller modelling seems to be too optimistic when the prop blade load is high, i.e. when the disc loading is high in combination with low prop advance ratios such as when WEP is applied at low speed.

Even if it's hard for human pilots to hang on the prop like the AI due to control issues in DCS, it's still perfectly possible to control the P51 at 120 mph which was the speed I did the tests at. The point here is should the prop efficiency degradation if you are slowing down to 120 mph IAS in a dogfight be 5 or 20-25%? I think it should be the latter. IMHO this also is a central issue in getting the sim as close to IRL performance as possible. If this is off, it will unrealistically favour those who fly in DCS using TnB verusus those who BnZ. Both the Pony and the Dora were as we know designed for the latter.

Propeller design is always a compromize between climb and speed characteristics: This can also be seen in WW2 NACA and German reports where prop designs are sometimes referred to as a climb or speed prop design. A climb prop will have a high activity factor (blade area/disc area) and relatively thick blades and camber to enhance high Cl operation. The high speed prop on the other hand will have a relatively thin blades and less camber to avoid compressibility problems due to the high prop tip speeds WW2 fighters had.

So how do you get the best compromize? Well you build in enough thickness and camber in the blades so that climb does not suffer too much. This means that you are operating on the higher end Cl (aOa) in the climb case, just before your profile drag starts to become excessive. In this way you have as thin a blade as you can get away with which will postpone compressibility effects and consequently give you good high speed performance as well.

So as long as you keep an IAS that is in line with the design in your climb you are fine. Now the problem that happens if you go to a lower IAS is that to absorb the power from the engine the prop blade aOa (not to be confused with pitch) has to increase to maintain RPM. This increses the prop blade loading outside what it was designed for and the drag of the blade increases a lot meaning you are expending torque to overcome the increased blade drag which is not helping your climb.

As I said before, my suspicion is that the DCS prop modelling is too optimistic here and that this effect (blade efficiency decreasing at high aOa) in the sim is not as large as it should be. At least that is what the 5% reduction in climb rate and prop hanging antics indicate to me.

Again, it would be good if someone involved in testing or design of the prop modelling in DCS could comment on this.

Sorry if I seem thick, but this is what I'm trying to get at:

For most blade angles the max efficiency only differs by about 5%. At different climb speeds the P-51's blade angle changes to maintain 3000rpm. So, between advance ratios of 0.8 to 2.0 or more, the max efficiency will be somewhere in the 0.82 to 0.85 range.

Please show me something in addition to your written explanations so I can better understand.

Sorry if I seem thick, but this is what I'm trying to get at:

 

 

For most blade angles the max efficiency only differs by about 5%. At different climb speeds the P-51's blade angle changes to maintain 3000rpm. So, between advance ratios of 0.8 to 2.0 or more, the max efficiency will be somewhere in the 0.82 to 0.85 range.

 

Please show me something in addition to your written explanations so I can better understand.

A perfectly valid question and this is a good diagram: First of all the Pony is limited to min 23 deg pitch AFAIK and taking this as an example (although we can do so only to show the principle) , if the advance ratio (J) in the picture goes to 0.5, then if we had this propeller we would get a prop efficiency of only 0.6 as opposed to the optimum 0.85 because pitch lower than 25 deg was not possible even to set. This illustrates the problem but note this diagram is only valid for a certain design and power loading and it would look different depending on the number of blades, blade size, profile and disc loading, i.e. what is the air density and what is the power being absorbed.

The problem in case of the Pony is that it may not even be possible to go to 23 pitch because the prop may not be able to absorb this power: By that I mean that if you fed the WEP power into the prop at 23 deg pitch it may overrev and go beyond 3000 rpm. So how to solve that? Well you increase the pitch which would result in a higher blade aOa.

So taking another example, say we have J=1. Looking in the diagram we would want to set pitch 25 deg and get 0.85% prop efficiency. However, say the constant speed governor needs to go to 35 deg pitch to absorb the power and limit the revs to 3000 rpm. Then the prop efficiency is reduced to around 0.72.

So again, the question is how is the propeller modeled in DCS? At low J values and high disc loadings like climbing at slow speeds like 120 mph, is the prop efficiency still the same as at 175 mph or is it reduced? If so by how much and how is it modeled in the prop hanging scenario?

Pilum, I am interested in propeller efficiency at high speed dive, such as 800 TAS. Still 85%? I doubt about that.

What if P51D 75% while Dora 65% efficiency at 800TAS?

Sorry for not replying earlier but I'm on vacation so I don't have access to my computer and consequently I can't check up either the advance ratio or prop efficiency in a high speed dive. However, as soon as I tire of Grecian wine and sun I'll post some numbers :smilewink:

I see too many "say" but I do not see "as a result of calculation" or "Fig XX in NACA report on a test of a 4-blade prop shows...", for example...

OK, so to take an example and to be more specific and use some real numbers:

I'm assuming the following:

Altitude: sea level ( Using 0 Km altitude as an example to show the delta in climb rate. Results concerning the delta in climb rate due to reduction of prop efficiency due to lower advance ratio (J) and higher Power loading coefficient (Cp) are in the same order at 1-2 Km altitude)

P51D data assumed:

Power P: 1670 Hp at 3000 rpm @ sea level

Weight W: 4445 Kg

Prop dia D: 3.36 m

Activity factor AF=112

Prop reduction ratio: 2.088

Prop rps n: 3000/(2.088*60)=23.95 rps

Advance ratio J=v/(n*D)

Power loading coefficient (used in NACA report) Cp=P/(ra*n^3*D^5) =0.16

So using these numbers to estimate climb performance assuming prop efficiency is reduced according to NACA wartime report “The selection of propellers for high thrust at low speeds” (WR-L-483) figure 10 for Cp=0.2 for the 4 to 6 blade curve:

J=0.96 at 175 mph gives prop efficiency circa 0.8

J=0.67 at 120 mph gives prop efficiency circa 0.67

The reason for using an interpolation between the 4-6 blade curve in figure 10 is that the report is for blades with AF=90 and not like the P51 D which has AF=112. Note that according to the report author, it is perfectly reasonable to equate a larger number of thinner blades to fewer wider blades (third paragraph page 13 “Concluding remarks”). So five blades 5X90=450 solidity can be assumed to be equal to four blades 4x112=448.

At 120 mph (54 m/s) with prop efficiency assumed 0.67 : Climb rate circa 14.74 m/s (2902 fpm)

At 175 mph (78 m/s) with prop efficiency assumed 0.8 : Climb rate circa 18.35 m/s (3612 fpm)

Comparison of climb performance with prop efficiency assumed constant at 0.8:

At 120 mph (54 m/s) with prop efficiency assumed 0.8 : Climb rate circa 18.41 m/s (3624 fpm)

At 175 mph (78 m/s) with prop efficiency assumed 0.8 : Climb rate circa 18.35 m/s (3612 fpm)

So here there is not much of a difference with this assumption. This is also what seems to be the case in DCS, i.e. there is not much of a difference if one chooses to climb at 120 or 175 mph IAS which could be an indication that the prop efficiency is assumed to be the same at both speeds?

Is this the case or how is a prop efficiency at low speed modeled in DCS?

Link at NASA NTRS for report WR-L-483:

http://ntrs.nasa.gov/search.jsp?R=19930093644&hterms=wartime+report+483&qs=N%3D0%26Ntk%3DAll%26Ntt%3Dwartime%2520report%2520483%26Ntx%3Dmode%2520matchallpartial%26Nm%3D123

Pilum, I am interested in propeller efficiency at high speed dive, such as 800 TAS. Still 85%? I doubt about that.

 

What if P51D 75% while Dora 65% efficiency at 800TAS?

Well actually prop efficiency is not that much reduced at 800 TAS (At least not in my C++ simulation): This of course depends on altitude but I made a simulation diving from 500 Km/h TAS 20 deg dive from 9 Km alt and reached 800 Km/h TAS at 7.5 Km alt at which time the prop tip M=1.08 which is not too bad and prop efficiency is actually quite good at 0.8. However, a little later at 875 Km/h TAS at 5.7 Km alt the tip speed is up to M=1.11 and here the efficiency has dropped dramatically to 0.75. So things happen rather fast at these speeds.....

I don't get why the Dora should be so much worse? I get about the same results for the Dora with the same initial conditions. Why would the Dora's prop efficiency drop so low as 65% at 800 Km/h TAS?

Just checked what the model gives for the prop for DCS P-51 :

 

120 mph, advance ratio= 0.65, SL, 1640 hp, Cp = 0.146, Eff = 68.5%

175 mph, advance ratio= 0.94, SL, 1640 hp, Cp = 0.146, Eff = 80.4%

 

So, the report results match the model. THe model itself does not use tables for coefficients except the aerodynamic data bank for the blade.

 

I think that your measurements were not accurate enough to determine the difference in vertical speed. I think that more accurate way to measure could be full energy derivative method in horisontal or quasi-horisontal flight.

By the way, where did you get these vertical speed values from?

Ok, that is interesting and if you don't use tables but actually calculate the efficiency from blade theory then that is certainly impressive, especially if you manage to capture the reduced Cl/Cd ratio that results due to the higher disk loading in a way that approaches the NACA measured data :thumbup:

Concerning my tests, I simply clocked the values between altitudes in Tacview so I may be off there and I'll do a few more and post some results later this week for verification.

Based on a reduction of prop efficiency from 80.4% to 68.5% in DCS I would expect to see a dramatic reduction in climb rate then between using 175 and 120 mph. Do you agree and what would be your estimate?

My climb rate estimate was in the order of 2900 fps at 120 mph and 3600 fps at 175 mph. So here there is a substantial difference.

Is this in line with what we should expect to see in DCS as well? i.e. a reduction in climb rate of around 20%?

Hey,

I don't want to interrupt your discussion which I think is a very very complex one but there is 1 thing I thought cannot be true as said:

"So five blades 5X90=450 solidity can be assumed to be equal to four blades 4x112=448"

You have to take into consideration that the more blades you attach and the higher the rpm gets the less effective the prop in general gets because the blades "life" in the shadow of the blade placed before it. The best propeller is a 1-blade propeller, any other blade added is a drawback

in efficiency. Just only some rare AC carry 1-blade props, I do remember some gliders that feature it.

Also, the less pitch the prop has the more severe this effect becomes.

The only good thing I know of with multi-blade props is that they slow you down like an anchor on approach.

I do not know to which extend it will actually effect your scenario but I can tell you from my own experience with propellers in R/C real life that it will have a measurable effect.

Apart from that, we also have a very complex calculation program for propellers, diameter, pitch, watts and climb rate.

It sometimes astonishes one how much impact small changes in this assemble have and how much calculations can differ from actual measurements taken while flying.

Wouldn't the variable pitch have a huge bearing on any calculations? Just how efficient are aircraft of the period at optimising pitch angles? It's a very fascinating subject, and I cannot pretend to know the first thing about aerodynamics.

I think so, because there is no reason for physics to behave in different way...

What method and data did you use to estimate the climb rate?

OK, good to hear that you agree about the expected difference in climb rate. I plan to do some more tests later this week to see if I get similar results as in the tracks I posted in the OP.

Concerning how I estimated climb rates I have two methods for this: One is using a C++ program I developed which can calculate speed, climb, turn, dive and acceleration performance etc. and another simpler Excel spreadsheet for ballpark calculations.

I post data from the spreadsheet below so you can see what assumptions I made for the 175 mph case. I could not paste in the table but I think you can see the assumptions made in the text anyway. Granted, Specific Excess Power (SEP) climb rate is only valid for small climb angles which is not really true for WW2 fighters but it is close enough to get the ballpark numbers for comparisons I think, at least to compare relative performance between 120 and 175 mph climb speeds.

P51D Mustang

Mass (Kg) 4445

Span (m) 11.26

Wing area (m*2) 21.79

Cdo 0.0176

Aspect ratio A 5.818614

Pi 3.141593

Cdi 0.01972

Cl 0.537017

Oswald factor e 0.8

Mach 0.229142

vTAS (m/s) 78

a (m/s) 340.4

ra (Kg/m*3) 1.225

q (N) 3726.45

loadfactor n 1

g (Kgm/s*2) 9.81

Propellerefficiency n 0.8

Engine power (hp) 1670

Tprop (N) 12606.36

Texhaust (N) 681.6327

Ttotal(N) 13287.99

D (N) 3030.397

IAS (m/s) 78

SEP climb estimate (m/s) 18.34845

SEP climb estimate (fpm) 3611.899

Propeller diameter (m) 3.39

engine rpm 3000

reduction ratio 2.088

Propeller rps 23.94636

Propeller advance ratio J=v(n*D) 0.96085

Cp=P(ra*n^2*D^5) 0.163208

Absolutely Yo-Yo,

I do know exactly about the problem of prop clearance and power absorbing and that a 1-blade prop won't take the power of a Merlin, unless its 50 feet long and has a counter balance weight the size of an aircraft carrier ;)

What I was to say is:

Adding a blade, even if you scale down the diameter, does effect your efficiency.

The less blades the better, the less pitch you have the more each added blade takes efficiency down.

I am not saying that it won't match the engine power etc... that can be adjusted by diameter and pitch of the blades.