Community driven CFD for R-27 family of missiles (virtual wind-tunnel testing)

[Cmptohocah]

Hi everyone,

the reason for this post is my will to start an initiative that will perform a Computational Fluid Dynamics analysis of the R-27 missile family, so that we can provide the developers with the results in a scientific and transparent manner.
 

I have done some very basic exploratory research using Autodesk CFD software. It's fully functional for a trial period of 30 days, but can be used multiple times (instructions upon request in PM). The 3D model was created in Catia, but the CFD software accepts many different formats.

I would like to ask for help from any people that are knowledgeable in this area, people that are good in 3D modeling and basically anyone else willing to contribute, to help me out in setting up proper wind tunnel analysis and results post processing. I do have a Master's in Engineering, but I got my degree some time ago and I am quite rusty in fluid dynamics, meaning I suck at it.

Here are some pictures from the initial trials.

 

Figure 1.0 - velocity graph

 

Figure 2.0 - airflow

Edited December 21, 2020 by Cmptohocah
Formating

[TaxDollarsAtWork]

I don't know too much about this but good on you OP, it will give us something to compare the ED missile to and help keep them honest about its performance.

[GGTharos]

Are you planning to run this on your own PC?  What are your specs like?  (You play DCS so likely good enough ).

 

I've not set these up but I've interacted with people who have, this sort of stuff requires a lot of patience and fine-tuning.   Curious to see what you get, even if it's just a zero-lift drag coefficient which is useful in and of itself.

 

See if you can get a drag curve from M0.5 to 4.5 every half a mach number to start with.   If you want to get more detail, IMHO usually the most interesting area is going to be around M0.8-M2.0, the rest tends to follow a trend (Though of course down the road you want accurate numbers there as well)

 

My understanding is that for very accurate (and more importantly, accurate when the lift isn't 0, ie. there's some AoA) figures you need over 5 million cells, but don't quote me too much on this.

 

The second thing to work on would be a lift curve.   The moment you have those two you'd be able to plug them directly into the new FM, or compare it to the old FM (while it's not as accurate as the new one, it can still be made more reasonable with accurate data).

 

Also, realize that there is really only a couple of specific things that you can accomplish here (WRT DCS, I think you'll accomplish a lot as far as the experience with the whole thing goes):  By computing the drag curve, you might find that the missile in-game is too draggy ... adjusting to the new values will have the missile keep its speed a little bit better at distance, but don't expect miracles.   largest effect will probably be at high altitude.

 

By computing the lift curve, you'll see better maneuverability below mach 1 or so, because in the old FM this was typically set to pretty low - while you can set it higher the resolution of the curve in the old FM is very poor so you end up with some silliness at the edges, which usually doesn't have a practical effect in-game since the missile just can't catch up to its target, but it does change short-range behavior when under power, ie. say those dogfight shots.

Edited December 21, 2020 by GGTharos

[OpticFlow]

Would the lift coefficient change if the missile is rotated 45 degrees along the axis (to resemble X instead of +)?

[GGTharos]

I don't know enough to answer that question, except this:  You get more lift force when flying as an 'X', my understanding from very limited reading is that this is the preferred skid-to-turn configuration.  So, I don't know if it's the lift coefficient or surface area, or because of vortices or all of the above, but AFAIK X does mark the spot

 

If you look at METEOR it also seems to be a skid-to-turn missile.  There used to be some debate about skit-to-turn causing problems for the inlets, and some were expecting METEOR to be roll-to-turn.

Edited December 21, 2020 by GGTharos

[Cmptohocah]

1 hour ago, GGTharos said:

Are you planning to run this on your own PC?  What are your specs like?  (You play DCS so likely good enough ).

 

I've not set these up but I've interacted with people who have, this sort of stuff requires a lot of patience and fine-tuning.   Curious to see what you get, even if it's just a zero-lift drag coefficient which is useful in and of itself.

 

See if you can get a drag curve from M0.5 to 4.5 every half a mach number to start with.   If you want to get more detail, IMHO usually the most interesting area is going to be around M0.8-M2.0, the rest tends to follow a trend (Though of course down the road you want accurate numbers there as well)

 

My understanding is that for very accurate (and more importantly, accurate when the lift isn't 0, ie. there's some AoA) figures you need over 5 million cells, but don't quote me too much on this.

 

The second thing to work on would be a lift curve.   The moment you have those two you'd be able to plug them directly into the new FM, or compare it to the old FM (while it's not as accurate as the new one, it can still be made more reasonable with accurate data).

 

Also, realize that there is really only a couple of specific things that you can accomplish here (WRT DCS, I think you'll accomplish a lot as far as the experience with the whole thing goes):  By computing the drag curve, you might find that the missile in-game is too draggy ... adjusting to the new values will have the missile keep its speed a little bit better at distance, but don't expect miracles.   largest effect will probably be at high altitude.

 

By computing the lift curve, you'll see better maneuverability below mach 1 or so, because in the old FM this was typically set to pretty low - while you can set it higher the resolution of the curve in the old FM is very poor so you end up with some silliness at the edges, which usually doesn't have a practical effect in-game since the missile just can't catch up to its target, but it does change short-range behavior when under power, ie. say those dogfight shots.

 

1. Yes, the specs are: i5-4960K clocked at 4.6GHz, 16 Gb RAM
2. How can I "plug" the lift curve into the FM?
 

23 minutes ago, OpticFlow said:

Would the lift coefficient change if the missile is rotated 45 degrees along the axis (to resemble X instead of +)?

Good point. I will also consider doing tests in the "X" orientation.

Edited December 21, 2020 by Cmptohocah

[GGTharos]

8 minutes ago, Cmptohocah said:

1. Yes, the specs are: i5-4960K clocked at 4.6GHz, 16 Gb RAM
2. How can I "plug" the lift curve into the FM?

 

You have two options:

Copy the say, AIM-7FM definition into the R27, then replace the lift and drag data with those from your simulation (probably not as easy as it sounds, but the absolutely best thing to do IMHO).

Another option is to try and fit the data you have using the old FM, which would require you to only change a small bunch of numbers.  Ping me when you're ready to try some time and we'll arrange a time and I'll walk you through the file location and format.

Edited December 21, 2020 by GGTharos

[Omz]

I just hope you jump onto the R-77 next cause that thing needs some serious work and its sad considering its the only active homing russian missile we got

[Skysurfer]

3 hours ago, Omz said:

I just hope you jump onto the R-77 next cause that thing needs some serious work and its sad considering its the only active homing russian missile we got

The base 77 will never compare to an amraam. Treat is like a slightly longer range, radar guided Archer.

[Cmptohocah]

8 minutes ago, Skysurfer said:

The base 77 will never compare to an amraam. Treat is like a slightly longer range, radar guided Archer.

Let's please stick to facts on this thread. These kind of "out of the blue" statements just flare up endless discussions and dilute the topic. It's like throwing a match in a pool filled with gasoline.

[BlackPixxel]

23 minutes ago, Skysurfer said:

The base 77 will never compare to an amraam. Treat is like a slightly longer range, radar guided Archer.

 

Don't let the current R-77 implementation in DCS fool you.

Once the R-77 gets CFD rework and lofting profiles (which it will) it will not be far from a Aim-120B.

[Omz]

2 hours ago, BlackPixxel said:

 

Don't let the current R-77 implementation in DCS fool you.

Once the R-77 gets CFD rework and lofting profiles (which it will) it will not be far from a Aim-120B.

I hope so any idea how soon that will be though?

Edited January 2, 2021 by Omz

[TaxDollarsAtWork]

8 hours ago, Omz said:

I hope so any idea how soon that will be though?

 

Don't hold your breath, its really one of those things on indefinite hold.

Originally they promised it for some time midsummer 2020 then autumn then some time around February/March of this year though they always say the F/A-18, F-16 and now Apache keeps delaying it

[Omz]

10 hours ago, TaxDollarsAtWork said:

Don't hold your breath, its really one of those things on indefinite hold.

Originally they promised it for some time midsummer 2020 then autumn then some time around February/March of this year though they always say the F/A-18, F-16 and now Apache keeps delaying it

Well then we're talking a year or two by the looks of it

[Cmptohocah]

** UPDATE **

 

Here are the results of initial CFD testing for the R-27ER's coefficient of drag and AoA=0:

v [km/h]	M	H [m]	Tt [deg C]	Rho [kg/m^3]	Fd [N]	A [m^2]	Cd
618	0.5	0	15	1.2	1187.71	7.423	0.0090
1250	1.0	0	72	1.2	4995.24	7.423	0.0093
1840	1.5	0	144	1.2	8330.82	7.423	0.0072
2450	2.0	0	268	1.2	10471.6	7.423	0.0050
3090	2.5	0	375	1.2	12003.2	7.423	0.0037
3708	3.0	0	533	1.2	14962.5	7.423	0.0032
4326	3.5	0	720	1.2	18827.9	7.423	0.0029
4944	4.0	0	936	1.2	23186.6	7.423	0.0022

where:

• v - true airspeed
• M - mach number
• H - altitude above MSL
• Tt - total air temperature
• Rho - air density
• Fd - drag force
• A - "wet" area

3D model used for flight testing is somewhat simplified in a way that the fins are not tapered, but are rather a simple flat surfaces, ie. no airfoil.

 

Image of the mesh used:

There were 10 layers used to simulate the boundary layer.

[GGTharos]

Are you able to create a speed profile for a given altitude?  Ie. time-speed (and therefore distance) for say 10000m.  

[HWasp]

Tried to roughly calculate average drag values in DCS measuring average deceleration of the missile at sea level, and the drag values don't seem to be vastly different from ones in the table:

Between 2800 and 2600 km/h your table suggests around 11000N drag, in DCS it seems to have around 13000N, so more but nothing extremely crazy.

 

What seems really strange in the sim, is that at low altitude the missile hits a fixed speed, and refuses to accelerate any further (around 1700 kts). It reaches that speed very quickly and then it just keeps that.

 

What is the thrust of the ER's motor? Is it in the 80kN range?

Edited January 14, 2021 by HWasp

[dundun92]

3 minutes ago, HWasp said:

Tried to roughly calculate average drag values in DCS measuring average deceleration of the missile at sea level, and the drag values don't seem to be vastly different from ones in the table:

Between 2800 and 2600 km/h your table suggests around 11000N drag, in DCS it seems to have around 13000N, so more but nothing extremely crazy.

 

What seems really strange in the sim, is that at low altitude the missile hits a fixed speed, and refuses to accelerate any further (around 1700 kts). It reaches that speed very quickly and then it refuses to accelerate any more.

 

What is the thrust of the ER's motor? Is it in the 80kN range?

According to the missiles_data.lua, the ER uses a boost-sustain motor. Boost phase is 73.5 kN for 3.0 sec, sustain is 14.5 kN for 7.0 sec

The missile is probably hitting the sustain phase when it appears to hit this "speed cap", which is the point behind boost sustain; boost takes it up to a certain speed, then sustain "sustains"/slightly increases that speed for a bit.

[HWasp]

Thanks, that explains it then.

[BlackPixxel]

Nice work!

 

Something to consider (which is implemented with the new FM I think) is that as long as the motor is burning, the rear of the missile will not create a zone of low pressure.

 

So during the motor burntime, the drag would be reduced.

[GGTharos]

5 hours ago, Cmptohocah said:

** UPDATE **

 

Here are the results of initial CFD testing for the R-27ER's coefficient of drag and AoA=0:

v [km/h]	M	H [m]	Tt [deg C]	Rho [kg/m^3]	Fd [N]	A [m^2]	Cd
618	0.5	0	15	1.2	1187.71	7.423	0.0090
1250	1.0	0	72	1.2	4995.24	7.423	0.0093
1840	1.5	0	144	1.2	8330.82	7.423	0.0072
2450	2.0	0	268	1.2	10471.6	7.423	0.0050
3090	2.5	0	375	1.2	12003.2	7.423	0.0037
3708	3.0	0	533	1.2	14962.5	7.423	0.0032
4326	3.5	0	720	1.2	18827.9	7.423	0.0029
4944	4.0	0	936	1.2	23186.6	7.423	0.0022

where:

• v - true airspeed
• M - mach number
• H - altitude above MSL
• Tt - total air temperature
• Rho - air density
• Fd - drag force
• A - "wet" area

3D model used for flight testing is somewhat simplified in a way that the fins are not tapered, but are rather a simple flat surfaces, ie. no airfoil.

 

Image of the mesh used:

 

There were 10 layers used to simulate the boundary layer.

 

 

The subsonic value seems very high to me - generally it's much, much lower than the M1 peak and your Cd0 should not be reaching that low before M2 (usually past M2).

Could you re-visit that?

[Cmptohocah]

1 hour ago, BlackPixxel said:

Nice work!

 

Something to consider (which is implemented with the new FM I think) is that as long as the motor is burning, the rear of the missile will not create a zone of low pressure.

 

So during the motor burntime, the drag would be reduced.

That's true, but I don't know how I could simulate that bit. I guess even in the real wind tunnel this would be difficult to simulate. Or am I wrong?

 

1 hour ago, GGTharos said:

 

 

The subsonic value seems very high to me - generally it's much, much lower than the M1 peak and your Cd0 should not be reaching that low before M2 (usually past M2).

Could you re-visit that?

 

What is this based on? I mean, what should the values be and why?

[BlackPixxel]

Maybe it could be approached by giving it a very aerodynamic shape after the tail, but even then it would create some additional friction.

[GGTharos]

1 hour ago, Cmptohocah said:

That's true, but I don't know how I could simulate that bit. I guess even in the real wind tunnel this would be difficult to simulate. Or am I wrong?

 

It isn't that important.  This will affect top speed, but you could generate a graph just for the motor burn-out part.   The other trick is to reduce all drag coefficients by a ballpark of 22% when the motor is on.  It's not super-duper accurate but it'll get you close.

 

1 hour ago, Cmptohocah said:

What is this based on? I mean, what should the values be and why?

 

Look at the graph (Scroll down):  https://space.stackexchange.com/questions/22907/modelling-spacexs-lift-and-drag-versus-angle-of-attack-and-mach-number

 

This is what a typical coefficient graph looks like for an AAM.   Most missiles, really.   What you have now looks more like going up a slight incline and then you fall off the cliff.  You need to climb the mountain, then go down the mountain.   The exact shape of the mountain is what varies.

[nighthawk2174]

Concerning motor on and motor off differences this is what were talking about:

it's about a 20-22% difference.

 

On 1/2/2021 at 10:02 AM, BlackPixxel said:

 

Don't let the current R-77 implementation in DCS fool you.

Once the R-77 gets CFD rework and lofting profiles (which it will) it will not be far from a Aim-120B.

 

It already got a partial treatment and the drag nearly matches my own CFD stuff for it, within a few %.  Biggest concern I have is that the motor burn time fell by a decent amount.

 

Edit: graphs

To simplify, silver = my cfd with new ED thrust, blue = ED stuff with new thrust.  Yellow is Old ED drag with new thrust.  Orange is my CFD with old thrust.

Edited January 15, 2021 by nighthawk2174