Su-27: Flight Range Analysis

TEST PURPOSE:

As I previously did with my series of tests investigating effective engine RPM settings against desired cruise altitude, for this test series I decided to investigate what sort of flight range would be realistically achievable at varying altitudes. As before I didn't set out to investigate theoretical maximum performance but rather what would be practically achievable in a combat scenario, i.e. carrying a nominal maximum air-to-air war load and taking off with a full fuel load of 9,400 Kg. The nominal war load used for each test was 2xECM, 2xR-73, 2xR-27ET, 4xR-27ER, 100% cannon ammunition, full expendables and 100% fuel.

TEST CRITERIA:

Each test was carried out under the following conditions:

1. A standard DCS day, i.e. 1200 hours on 01/06/2012.
   
2. Standard DCS meteorological conditions, i.e. +20*C temperature, no wind, no precipitation, no turbulence, no clouds.
   
3. Each test flight was a set course, taking off from Novorossiysk with 100% dry RPM, turning south to 180*, climbing to cruise altitude with Vv~ 25-35m/s, setting altitude hold mode at the desired altitude and cruising at a calculated efficient RPM setting until Batumi was exactly 90* to port then turning in to Batumi and again cruising until within a reasonable landing distance before descending to the designated pattern altitude and landing.
   
4. Remaining fuel was measured on shut down at Batumi.
   
5. I used Tacview 1.4 to measure the flight distance
   
6. I then used MS Excel to calculate the test statistics i.e. fuel consumption and what I'm going to call "steady cruise fraction."
   
7. Steady cruise fraction is the percentage of the total flight distance that was flown at the desired cruise altitude, i.e. not including climb to altitude or descent & landing.
   

Something that's important to consider is that for the various tested cruise altitudes I did not use a constant cruise RPM but rather used the most 'efficient' RPM setting that I previously calculated. Interested readers can find the test series that produced that data in this thread here: http://forums.eagle.ru/showthread.php?t=147556

To save you reading all 5 pages and the 15 or so associated data analysis graphs, here's the graph for efficient cruise RPM setting v desired cruise altitude. To understand how I arrived at these figures see the thread link above.

TEST FLIGHT RESULTS:

Having carried out test flights at steps of 1,000m altitude from 1,000 to 11,000m, the following represents the raw data produced by the test flights:

Here's a graph showing the calculated steady cruise fraction (SCF, in %) against cruise altitude:

As we would expect, the greater time taken to climb to higher altitude produces a smaller SCF value. The mean flight distance was 659 Km. An important consideration, which again is as we would expect, is that as total flight distance increases the SCF tends to --> 100%

Here's a graph showing the calculated flight range for cruise at various altitudes. Note that this graph is based on the raw data, i.e. NOT adjusted for varying SCF:

As we might expect the general trend indicates that cruise at higher altitudes provides a greater maximum flight range. There is some variation, in particular note that for the tests I conducted cruising at 11,000m would technically provide a shorter maximum range than cruising at 10,000m but I largely attribute this to the very long time it took to reach 11,000m altitude, i.e. a significantly lower SCF than for all other cruise altitudes. There were also severe issues with autopilot performance at 11,000m altitude, but more on that later.

In order to isolate the range calculations for the effects of varying SCF, I then normalized the calculated fuel consumption figures for SCF values. This produced normalized calculated range data shown on this graph:

This produces results more in line with expectations, i.e. calculated maximum flight range simply increases with altitude. Also note the calculated figures for cruise at high altitudes: the normalised figures produce range estimates much closer to observed real-world flight ranges that are known to be possible in the Su-27.

ANALYSIS & CONCLUSIONS:

Calculation of maximum range is far from simple and has to take in to account a lot of different factors. In 'reality' i.e. a given DCS mission maximum range will, as expected, depend heavily on chosen cruise altitude but things like time spent actually fighting, loadout taken, temperature and general weather conditions will all also have significant effects.

It is worth emphasizing that for almost any typical mission in the Caucasus theatre of operations, if you're flying the Su-27 fuel will not be a great concern for you. The Flanker is quite capable of flying 300 Km down in the weeds at below 1,000m altitude, fighting for 10 minutes and then flying another 300 Km back to base with plenty of fuel left for a safe landing or if necessary diversion to another airfield. If you have the tactical option of cruising to your designated target or operational area at any altitude, you can forget the problem of running out of fuel, in most cases it simply won't be a problem.

My personal recommendation would be for a lengthy cruise to target, start off with a gentle steady climb to 10,000m altitude, cruise at 85% RPM which will have you flying at a very respectable 983 Km/h TAS. Before you reach your CAP zone consider descending to 9,000m and reducing thrust to 83% RPM for a nice efficient loiter at an altitude at which the Su-27 becomes a bit more effective at actual fighting than it is up at very high altitude.

A NOTE ON AUTOPILOT EMPLOYMENT:

So, why not cruise at 11,000m or even higher? Well, the issue there is the autopilot. For the autopilot to produce stable flight, testing has shown that it must be engaged with an indicated air speed of at least 550-560 Km/h. Engaging the autopilot with an IAS any lower that that will produce unstable flutter or uncommanded bank oscillations.

At 10,000m altitude with the recommended 85% cruise RPM your IAS will be ~580 Km/h with the indicated full war load. This will then result in stable and care-free cruise behaviour. At any altitude above 10,000m on a "standard DCS day" as described above, your IAS will be below the minimum requirement for stable autopilot operation. You can easily fly above 10,000m of course, but for a long flight you will have to fly and trim manually for stable flight, something that I find very difficult to do. This is current behaviour as of DCS 1.5 beta version 1.5.1.46722.87

I've uploaded zipped archives of all the test flight tracks, Tacview ACMI files, the Excel work sheet and images of the graphs from this post. They should all be available from this link:

https://onedrive.live.com/redir?resid=D3293A5ED5656179!4801&authkey=!ANq91zV5EDw5QEY&ithint=folder%2cxlsx

:) You are having way too much fun! Interesting stuff. As far as the autopilot misbehavior at 11K meters is concerned, how much less is the IAS at 11K vs 10K. I wonder if it's just slow enough that you get the same misbehavior as you do at slow speeds at lower altitudes.

wow. impressive amount of work.

Thanks both for your kind comments :)

:) You are having way too much fun! Interesting stuff. As far as the autopilot misbehavior at 11K meters is concerned, how much less is the IAS at 11K vs 10K. I wonder if it's just slow enough that you get the same misbehavior as you do at slow speeds at lower altitudes.

Aaaand... Solved. Excellent suggestion! You're absolutely right, stable autopilot function at least in "H" mode depends entirely on the IAS at the point at which the autopilot is engaged. Using my recommended 85% RPM at 10,000m altitude produced an IAS of 560 Km/h. Engaging H mode at IAS 560 produced stable cruise.

At 11,000m altitude the IAS never increases above 530 Km/h even at full dry thrust (98% indicated RPM) which always results in unstable autopilot behaviour.

Conclusion? I'd say that ~550 - 560 Km/h IAS is the absolute minimum indicated air speed at which stable autopilot behaviour can be expected. Activation at any IAS below this will produce all sorts of issues. So, either Anatoly Kvochur flew the entire distance from Moscow to Paris manually, which I find hard to believe, or the in-game autopilot modelling needs some revision for low indicated air speeds.

In trying to understand why this happens I've read and re-read the relevant parts of the Su-27 flight manual (pages 110-117). The only thing I can think of is that for relatively low IAS values the yaw stability system is over-compensating and the yaw dampener is ineffective. What produces this behaviour is beyond my understanding of the FCS and how it's modelled in-game. That's probably involving delving deep in to control theory...

Attached is a track showing the effect of IAS on autopilot behaviour between 10,000 - 11,000m altitude.

Su-27 H-AP Test 10000-11000m.trk

Thanks both for your kind comments :)

 

 

 

Aaaand... Solved. Excellent suggestion! You're absolutely right, stable autopilot function at least in "H" mode depends entirely on the IAS at the point at which the autopilot is engaged. Using my recommended 85% RPM at 10,000m altitude produced an IAS of 560 Km/h. Engaging H mode at IAS 560 produced stable cruise.

 

At 11,000m altitude the IAS never increases above 530 Km/h even at full dry thrust (98% indicated RPM) which always results in unstable autopilot behaviour.

 

Conclusion? I'd say that ~550 - 560 Km/h IAS is the absolute minimum indicated air speed at which stable autopilot behaviour can be expected. Activation at any IAS below this will produce all sorts of issues. So, either Anatoly Kvochur flew the entire distance from Moscow to Paris manually, which I find hard to believe, or the in-game autopilot modelling needs some revision for low indicated air speeds.

 

In trying to understand why this happens I've read and re-read the relevant parts of the Su-27 flight manual (pages 110-117). The only thing I can think of is that for relatively low IAS values the yaw stability system is over-compensating and the yaw dampener is ineffective. What produces this behaviour is beyond my understanding of the FCS and how it's modelled in-game. That's probably involving delving deep in to control theory...

 

Attached is a track showing the effect of IAS on autopilot behaviour between 10,000 - 11,000m altitude.

It is likley that the autopilot uses a Proportional, Integral, Derivative (PID) controller.

Short period oscillations often happen because the proportional gain is too high for the conditions- this causes the controller to overreact to the current error in position.

The Integral component of the controller tries to compensate out small errors by adding all past error together and adding that sum to the control input. If you're not careful about how you set it up then you can end up with a situation where the integral component winds up to very high values in one direction before the system starts responding. This then causes a massive overshoot and windup in the other direction, and so on so forth. That sort of situation tends to cause long period, very large oscillations, not a rapid, continuous wobble.

The Derivative component tries to prevent overshooting by measuring the change in error over time and subtracting that from the control signal.

PID controllers generally work well over a range of conditions for which they have been tuned. Because the tuning is a compromise, it may be that the autopilot is not tuned for flight below 530 km/h IAS out of an expectation that the aircraft is unlikley to spend much time flying at such low indicated airspeeds.

It is likley that the autopilot uses a Proportional, Integral, Derivative (PID) controller.

 

Short period oscillations often happen because the proportional gain is too high for the conditions- this causes the controller to overreact to the current error in position.

 

The Integral component of the controller tries to compensate out small errors by adding all past error together and adding that sum to the control input. If you're not careful about how you set it up then you can end up with a situation where the integral component winds up to very high values in one direction before the system starts responding. This then causes a massive overshoot and windup in the other direction, and so on so forth. That sort of situation tends to cause long period, very large oscillations, not a rapid, continuous wobble.

 

The Derivative component tries to prevent overshooting by measuring the change in error over time and subtracting that from the control signal.

 

PID controllers generally work well over a range of conditions for which they have been tuned. Because the tuning is a compromise, it may be that the autopilot is not tuned for flight below 530 km/h IAS out of an expectation that the aircraft is unlikley to spend much time flying at such low indicated airspeeds.

Very interesting, I learned a lot from your explanation :thumbup:

Leading on from there then, I would totally understand if the Su-27 autopilot / FCS system was tuned to largely ignore flight below ~550 Km/h true air speed, in fact that value for true air speed is right on the bleeding lower edge of the possible flight regime in which it's possible to maintain steady altitude at steady AOA, and then only at altitudes of 1,000m and below. What makes less sense to me is for the FCS / autopilot to not be tuned for that sort of indicated air speed flight regime. The Flanker service ceiling is well above 11,000m so that sort of indicated air speed, especially for the sort of long-range cruise that it would suit and which would surely involve use of the autopilot, should be expected and normal.

I wonder if the FCS has been tuned for IAS instead of TAS, and whether this might be a mistake or a bug?

I wonder if the FCS has been tuned for IAS instead of TAS, and whether this might be a mistake or a bug?

TAS isn't a great way of tuning the controller because 600km/h TAS at sea level is very different to 600 km/h TAS at 10,000m. IAS roughly approximates the forces on the wings and control surfaces regardless of altitude (assuming you discount mach effects), so your controller implementation will be more universal if you use IAS. Dynamic pressure (The pressure exerted on the airfram by its movement through the air, often written as Q), might be even better.

TAS isn't a great way of tuning the controller because 600km/h TAS at sea level is very different to 600 km/h TAS at 10,000m. IAS roughly approximates the forces on the wings and control surfaces regardless of altitude (assuming you discount mach effects), so your controller implementation will be more universal if you use IAS. Dynamic pressure (The pressure exerted on the airfram by its movement through the air, often written as Q), might be even better.

Good point. Tuning the FCS for Q sounds like a more accurate option all round. It's a shame that the manual doesn't say exactly how the FCS is tuned, or the specifics of the actual control system that's used. Unfortunately I can't imagine that a member of the Sukhoi engineering team will pop up & explain everything :D It would be interesting to have some input from someone on the ED team though about how the system is implemented in-game :book:

Thanks both for your kind comments :)

 

 

 

Aaaand... Solved. Excellent suggestion! You're absolutely right, stable autopilot function at least in "H" mode depends entirely on the IAS at the point at which the autopilot is engaged. Using my recommended 85% RPM at 10,000m altitude produced an IAS of 560 Km/h. Engaging H mode at IAS 560 produced stable cruise.

 

At 11,000m altitude the IAS never increases above 530 Km/h even at full dry thrust (98% indicated RPM) which always results in unstable autopilot behaviour.

 

Conclusion? I'd say that ~550 - 560 Km/h IAS is the absolute minimum indicated air speed at which stable autopilot behaviour can be expected. Activation at any IAS below this will produce all sorts of issues. So, either Anatoly Kvochur flew the entire distance from Moscow to Paris manually, which I find hard to believe, or the in-game autopilot modelling needs some revision for low indicated air speeds.

 

In trying to understand why this happens I've read and re-read the relevant parts of the Su-27 flight manual (pages 110-117). The only thing I can think of is that for relatively low IAS values the yaw stability system is over-compensating and the yaw dampener is ineffective. What produces this behaviour is beyond my understanding of the FCS and how it's modelled in-game. That's probably involving delving deep in to control theory...

 

Attached is a track showing the effect of IAS on autopilot behaviour between 10,000 - 11,000m altitude.

Yep.. this is an old story since the PFM of the Su27 is out. The autopilot use all the time the rudders at lower speeds as 560 Km/h IAS, without any need.

Why? I don't know.

Would be nice if we could use the autopliot like in any other plane. :cry:

:thumbup: Darkfire!

:thumbup: Darkfire!

Thanks :)

Having looked again at my figures, it's become increasingly obvious that next to cruise altitude, the steady cruise fraction has the most significant impact on maximum possible range. So I'll make another recommendation:

For maximum possible range, carry out a takeoff at full military thrust, set the throttle to 95% RPM and reach your cruise altitude as soon as possible, i.e. a maximum-rate climb. The Flanker will climb happily down at ~650 Km/h TAS which will get you up out of the dense air at low altitude as soon as possible. Climb to 10,000m, set your throttle to anything between 80 and 87% RPM (I recommend 85%) and cruise around forever wandering around the map :smilewink:

This might seem counter-intuitive in that climbing at a lower throttle setting would seem to be more efficient than climbing at higher throttle settings. What I'm finding though is that spending maximum possible time at cruise throttle setting at high altitude more than makes up for the additional fuel used climbing more quickly at a higher throttle setting.

It really is the case that the Su-27 can achieve ranges, with a full A-2-A war load, that no other fighter aircraft in the game could even dream of achieving. I used to envy F-15 pilots with their external tanks. Turns out in the Flanker you simply don't need any extra fuel.

I know that your experiments show that power settings above 95% do not provide any benefit in cruise conditions. But does that also apply to climbing? Can you climb faster at 100%?

And what about acceleration? Can you accelerate faster at full military power than at 95%?

I know that your experiments show that power settings above 95% do not provide any benefit in cruise conditions. But does that also apply to climbing? Can you climb faster at 100%?

 

And what about acceleration? Can you accelerate faster at full military power than at 95%?

Not sure to be honest, this is something I'll have to test. I expect that yes, using 100% RPM probably would give a small increase to acceleration and climb rate.

Edited to add: a very quick & dirty test. I took off, levelled off at 1,000m, waited until I achieved Vmax (1032 Km/h as predicted) then pulled up to 10* pitch, engaged attitude hold and observed the climb rate. 100% RPM gave between 50 & 54 m/s climb rate whereas 95% RPM gave between 47 - 50 m/s climb rate. This didn't account for decreased fuel load between the two tests (conducted in a single flight) but there does appear to be a small but observable difference.

Doing this test accurately is going to be quite a challenge because climb rate is related to TAS and IAS much more strongly than it is to aircraft pitch attitude so I'd have to accurately and manually fly the same climb profile time after time. That's probably beyond the accuracy of my piloting ability at the moment. Acceleration might be easier to measure. I might well give that a go.

Thanks, so I guess the next question is: If the maximum range is the goal, does reaching the cruise altitude faster at 100% make up for increased fuel consumption while climbing?

(never mind, didn't mean to add even more to your plate :)) Great work, BTW!

Thanks, so I guess the next question is: If the maximum range is the goal, does reaching the cruise altitude faster at 100% make up for increased fuel consumption while climbing?

 

(never mind, didn't mean to add even more to your plate :)) Great work, BTW!

Thanks :)

Yes it does. Climbing at 100% RPM will use slightly more fuel than climbing at 95% RPM. What I believe these experiments show though is that this slight extra fuel consumption is more than made up for by increased time spent at cruise RPM at high altitude.

...

 

Doing this test accurately is going to be quite a challenge because climb rate is related to TAS and IAS much more strongly than it is to aircraft pitch attitude so I'd have to accurately and manually fly the same climb profile time after time. That's probably beyond the accuracy of my piloting ability at the moment. Acceleration might be easier to measure. I might well give that a go.

I wish I had actually marked the time it took but, inspired by this thread, I decided to take a quick flight up to 10,000 m. I took off a full mil power, leveled off and throttled back to 95%. Impatient, I only waited until my TAS airspeed was 830 before I started to climb adjusting my pitch to maintain 830 TAS all the way up. So I was fairly steeply pitched to start and a bit less steep toward the end. I was shocked at how quickly I reached 10,000 m. Unfortunately I never took a time check before I started the climb.

I wish I had actually marked the time it took but, inspired by this thread, I decided to take a quick flight up to 10,000 m. I took off a full mil power, leveled off and throttled back to 95%. Impatient, I only waited until my TAS airspeed was 830 before I started to climb adjusting my pitch to maintain 830 TAS all the way up. So I was fairly steeply pitched to start and a bit less steep toward the end. I was shocked at how quickly I reached 10,000 m. Unfortunately I never took a time check before I started the climb.

Cool :thumbup: At mil power it usually takes me in the region of 75 - 85 Km flight distance to reach 10,000m with Vv between 25 - 40 for the entire climb out.

The engine temperature re-profiling that happened a few patches ago made a HUGE and very positive difference to the way the Flanker behaves. Without being in any way an experienced pilot in the Eagle, I think these changes have brought the Flanker much closer to it in terms of performance, certainly in terms of climbing ability. They might actually have gone a bit too far: at ~13,000m altitude the Flanker will actually reach mach 2.6 at full afterburner... It isn't supposed to go that fast!

Cool :thumbup: At mil power it usually takes me in the region of 75 - 85 Km flight distance to reach 10,000m with Vv between 25 - 40 for the entire climb out.

 

...They might actually have gone a bit too far: at ~13,000m altitude the Flanker will actually reach mach 2.6 at full afterburner... It isn't supposed to go that fast!

Guess a bit more tuning needs to be done. :)

Just flew it again with slightly different parameters. Took off using full military power with 100% fuel and a medium weapons load. Leveled off at 300 meters altitude, throttled back to 95% and trimmed the aircraft as speed continued to increase. This time I was a bit more patient and waited for my TAS to reach 900 k/hr before I started my climb. Climbing, I altered my pitch to hold 900 k/hr TAS through 5000 m. After that I started to play pitch against TAS allowing myself to gradually slow to 800 k/hr TAS as I came over the top at 10,000 m. Time climbing: 265 sec. Total climb: 9700 m. Average Vv: 36.6 m/sec. I was 60 km downrange from where I started my climb.

Guess a bit more tuning needs to be done. :)

 

Just flew it again with slightly different parameters. Took off using full military power with 100% fuel and a medium weapons load. Leveled off at 300 meters altitude, throttled back to 95% and trimmed the aircraft as speed continued to increase. This time I was a bit more patient and waited for my TAS to reach 900 k/hr before I started my climb. Climbing, I altered my pitch to hold 900 k/hr TAS through 5000 m. After that I started to play pitch against TAS allowing myself to gradually slow to 800 k/hr TAS as I came over the top at 10,000 m. Time climbing: 265 sec. Total climb: 9700 m. Average Vv: 36.6 m/sec. I was 60 km downrange from where I started my climb.

I tried to replicate this by waiting until my TAS was at around 1020 Km/h at ~500m before pulling up and only reached 10,000m by the ~75Km point. I was wondering what the hell I was doing wrong. When I watched the track I then realised that before doing the climb test I'd been doing a quick test of the new high-AOA sound effects and wing vibrations. I had in fact climbed the entire distance to 10,000m... With my flaps down... :disgust:

I'm sure this would have elicited some very choice words and a lengthy speech from my crew chief on landing due to flap damage, but apart from that it proves just how excellent the performance of the Su-27 is these days :weight_lift_2:

On a related note, I tried the autopilot performance again and found that not only will it produce unstable flight if engaged at under 560 Km/h but the unstable flight condition is produced if the IAS falls below ~560 Km/h while it's engaged. That's got to be a bug of some sort.

I had in fact climbed the entire distance to 10,000m... With my flaps down...

I'm sure this would have elicited some very choice words and a lengthy speech from my crew chief...

Not to mention the ribbing you'd get from fellow pilots. :)

On a related note, I tried the autopilot performance again and found that not only will it produce unstable flight if engaged at under 560 Km/h but the unstable flight condition is produced if the IAS falls below ~560 Km/h while it's engaged. That's got to be a bug of some sort.

I'm guessing that it is. You're supposed to be able to use the AP most of the way down to the runway. That would definitely mean you were flying in the slower speed ranges. So something is up. It's supposed to use the trim functions of the aircraft to do its work. So I wonder if perhaps it has something to do with our self centering sticks. Just a thought.

Not to mention the ribbing you'd get from fellow pilots. :)

 

I'm guessing that it is. You're supposed to be able to use the AP most of the way down to the runway. That would definitely mean you were flying in the slower speed ranges. So something is up. It's supposed to use the trim functions of the aircraft to do its work. So I wonder if perhaps it has something to do with our self centering sticks. Just a thought.

Definitely. If I set up an accurate approach I find it quite easy to fly the last half of the approach through flare to touchdown and roll-out with neutral stick, purely using the trimmer to make pitch adjustments. If it's possible to do it this way manually there's no reason why the autopilot can't do it automatically.

And yes, I've come to the conclusion that many of the 'issues' people have around trimming, not only with the Su-27 but also with the Ka-50, are caused by the fact that all (AFAIK) commercially available consumer-level sticks are auto-centring whereas the actual sticks on things like the Flanker and Black Shark are not, and that the trimming system on both does adjust the centre position of the control column. Everything I've read about people who've experimented with home-made friction-hold sticks with no centring springs has revealed that in each case their problems with trimming go away. A non-centring stick with a nice long tube would probably also make using stick-to-trimmer mode a very usable and accurate control method. It's a shame that the likes of Thrustmaster or CH Products don't make a "prosumer" stick that operates in this way.

So, as a sort of final comment on the range analysis I decided to see just how non-realistic a test I could perform. I took off from Novorossiysk with a full war load and full fuel. I immediately hit maximum afterburner, levelled off at 300m (985 feet) and let it fly until I reached 500 Kg of fuel remaining.

A 232 Km flight at a top speed of 1,327 Km/h (~ Mach 1.1 at that altitude) got me all the way to Sochi Adler before I had to land with less than 100 Kg of fuel remaining. This aircraft is capable of insane ranges!

Su-27 Low altitude AB range test ACMI.zip

Su-27 Low altitude range test.trk

It is. I have to say one thing though. In tournaments ive flown like Joint Warrior fuel has always been a concern. During the course of a three hour match we've often refueled at least once or twice having to coordinate refueling and rearming intervals between elements/flights. Combat flying is alot different!

It is. I have to say one thing though. In tournaments ive flown like Joint Warrior fuel has always been a concern. During the course of a three hour match we've often refueled at least once or twice having to coordinate refueling and rearming intervals between elements/flights. Combat flying is alot different!

Very good point. I guess in a combat scenario I'd rather treat the large fuel capacity as providing a good base capacity with an 'enhanced reserve' rather than being forced to rely on it all to get me through an entire mission.

man, I can't, no we, can't thank you enough.