DCS World Flight Modelling Principles

That's pretty rich from someone who litterally doesn't understand what the chart says. I'd say learn to read such charts and understand what is ITR and what is STR before you criticize others of not reading it correctly. Getting a figure a tiny bit off (0.64 vs 0.67) is forgivable, but not even understanding what the chart is saying, like you, well sorry but spits you right out of the conversation.

 

 

RL ITR as pr. HAF manual:

26,000 lbs = 9.0 G @ 0.67

22,000 lbs = 9.0 G @ 0.62

 

This is what the DCS F-16 needs to match.

Nice try.

Thanks for your kind reply and sorry if I hurt you. But you were off by ~0.5g and that needs to be corrected, right? Reference page was wrong too, but shit happens.

Not an ideal thread to post definitions, but since you asked here again:

(This time transcripts from different reference books, you probably rather accept that source than in my own words.)

Str (region 1 in the chart)

„the region below the max t/w curve is one of sustained turn performance. This means that, in this region, the aircraft can maintain a turn rate without decelerating or loosing altitude...

In a sustained turn, aircraft maintains a constant altitude and airspeed and the turn radius, turn rate and load factor are constant...“

Itr (region 2 in the chart)

„In an instantaneous turn, the aircraft can not maintain these constant conditions, rather the turn entry conditions can only be maintained at the instant the turn is initiated. After turn entry, the airspeed, altitude, or both, decrease...

The region where both the lift and the structural limits are sufficiently high but the t/w is inadequate is a region of decelerating and / or descending flight. It is called the region of instantaneous turn performance. In this region high turn rates can be achieved for a brief period of time as the aircraft decelerates (or descends)...

From an energy perspective, the sustained turn is a constant energy maneuver, while the instantaneous turn is an energy loosing maneuver...

Doghouse

The doghouse plot is an aircraft specific turn performance chart that includes the aircrafts envelope boundaries. The overlaid ps=0 curve is the boundary of sustained performance.The aircraft has positive ps below ps=0 curve and negative ps above this curve. an aircraft has its highest turn rate at the intersection of the lift limit and load factor limit at the top of the doghouse(cp). This is an instantaneous turn rate, since ps is negative at this point.“

Hope that transcription helps you to get rid off your derangement(str @ ps-1000 and so on). If not, pm me, I will send you additional material.

Ps: itr is usually tested with wind up turns in rw...

I've heard before yeah, I have to say I don't agree with some of it on that kind of an exact level, it shouldn't be that simple I would imagine, and I'm just using logic, plus, my lifetime outsider knowledge of watching documentaries, following technical demos, NASA/ESA missions, and miscellaneous stuff.

 

The term "simulator" shouldn't mean the same, these are two totally different things, it is misleading if used improperly, I suggest using a prefix to differentiate, one is a hardware simulator versus a software one, it's a huge difference in so many ways and can't be compared apples to apples.

 

For military-grade hardware simulators those millions of dollars aren't spent on actual modelling and physics (the software side), but on the physical aspect of things being probably half of that IMO, but that's just the raw, not to mention other phantom costs as per consequence of the financial-business system such as the fact that it is a one-off solution or an extremely low amount of items, so demand is low and in business that means a much higher price automatically, second the customer is premium so they can be charged much higher and be acceptable, the seller is also taking a higher profit marging than it would if it sold to public just because they consider it a speciality, then there's all the research and support, all the pyhsical cockpit buttons and levers, all the screens, the hardware that powers the software, power delivery for struts and actuators, giant springs and hydraulics, and there's all kinds of side things I would imagine, so I would say software and specifically pyhsics modelling is probably a mere fraction of the total cost normally known in public, and it's that fraction that should be compared to DCS' phsyics modelling fraction. DCS's or any PC simulator's physics modelling fraction is much much higher than a military-grade hardware simulator.

 

So it's a lot more complicated than it sounds in that video/podcast, comparing these things would probably take a whole study worth of work to properly do, would have to disclose and compare the actual programming code or the results in debug outputs and graphs. You would have to correct all those dollars for inflation too, etc. :smartass:

 

So when we say and generally agree that the HW simulator "still doesn't feel like the real thing" isn't necessairly a good/appropriate argument for comparing simulation to reality, like in this case, it's tricky, it can boomerang, because those simulators may have not gotten that much attention on the software side when they were developed compared to so many years in DCS for example, their programming may have been from the time when programming it self was less understood and not as evolved as it is today, C++ has a lot of new features compared to 15 years ago for example, unless a HW sim is continiously updated like DCS is then it'll show it's weaknesses sooner or later. HW simulators still win because they can throw raw horsepower at it I suspect more than they are able to do on the software side, that however depends on the particularities of their focus on the software modelling and programming, they are certainly able to use a convenient function and the hardware just blasts through it even tho it's actually quite inefficient and would never be practical on PC. In other cases they are throwing so much horsepower at it but it can still be worse or not much better. The graphics is obviously not comparable because that's an actual decision for HW Simulators to not focus on them on purpose. (if anyone wonders about that)

 

The difference between HW simulators and DCS in pure raw modelling and that "right-feeling" perhaps already is much lower than some people expect (unless a huge new HW sim is developed that ups the bar higher ofcourse ***), however that low amount may seem high because of the high experience and someone like that see huge differences in something that is a mathematically small difference, so it's all relative, the closer to reality we get that last bit of percent difference while it seems small mathematically, it's actuall a whole world in terms of human practical difference and DCS on PC would indeed be lower on that scale, but not outside of that premium arena. It's not even about money or hardware computing power in some cases, but the feeling could be better if enough effort is put into it and the PC HW is used efficiently and could still approach the goal. That feeling may not require a proportional amount of dollars as it does in HW simulators, indeed.

 

So it's vital to not look at this with fixed proportions, it can end up in favor of DCS in some case for example you could do one thing really good, or you could simulate 1 aircraft at a time better, but not 100 ones as good as the HW Sim can do, for example.

 

*** == A huge new HW Sim that ups the bar higher and runs DCS as it's sofware component.

 

Those HW Sim Software components historically I presume were the one-off solutions done at the time for that specifically and never went into any other product, real specialites, but times have evolved since and there is something that can replace those solutions if it keeps evolving, even if it's a PC thing, and that could be your very own DCS (or a version of it) that replaces those unnamed HW Sim software solutions, even if they have unlimited resources, time is still not unlimited and to do something as all-encompassing as DCS in 2 years it's not just more expensive for at that time little gain, but also complicated, what a team of 2000 master programmers, who's going to manage all that for them to work efficiently, how do you put 2000 of them in a single git repository all working on their own code, the bigger the team the less efficient it becomes automatically, just mathematics and law of physics that works like this elsewhere.

 

This is again one of those posts of mine when I go to the left pocket with my right hand just to prove a point, yes I wanted to do it this way just to prove a point, going the reverse way, and finally mentioning there is already a military-grade DCS version if some of you don't know yet, so perhaps me knowing this ahead may make this post a bit biased, to have an earlier opinion of "HW integration can be feasible for DCS to achieve", but the point was to try to entertain the idea how this works in life and evolution, how some solutions override others unexpectedly, coming in from behind, it happens in the gaming world and other fields, when industry shifts engines, APIs, tools.

 

http://www.thebattlesim.com/

I really like what you write there.

Seasoned professional for numerical simulation methods in the automotive industry here.

What you say reflects perfectly my own experience.

Multi-million-euro car simulators are usually far below sim-racers for the PC when it comes to the driving dynamics side.

But the goal of a ‘professional’ car simulator is mostly just to create a certain workload on the brain of the test driver while he is testing new control elements or driving assist systems.

It is usually not meant for an accurate representation of driving dynamics behaviour of a car.

Fun fact: The so called multi body approach for sim-racers is very similar to the driving dynamics simulation tools in the automotive industry - in sim-racers they are just optimised towards performance because there you need real time performance.

The base for car simulators in the automotive industry is usually a ‘spread sheet’ simulation or in other words not more than a collection of look-up tables.

For professional flight simulators it is similar.

Civil aircraft (airliners) for example are never intentionally stalled - not for training and surely not for fun. Only test pilots do that during development.

Usually stall does also not take place in professional simulators. Pilots just get the warning signs recalled as a training (warning horn, stick shake).

Professional flight simulators which are capable of handling a so called upset recovery of an aircraft are at their very beginning:

https://trimis.ec.europa.eu/project/simulation-upset-recovery-aviation

Out-of-the-envelope behaviour of an aircraft in a simulation is the holy grail - it is yet to be developed.

At the moment outside-the-envelope behaviour is usually just 'scripted' or modelled as a rough approximation.

For the point of the “right-feeling" you mentioned:

Imagine taking a real car or a real plane and equip it with a remote control system and some cameras.

Show the camera pictures on pc-screens and add a gaming steering wheel or a gaming yoke or hotas.

Take test persons (professional test drivers and test pilots). Tell them what they see is a simulator.

Ask them how good it is.

They will most probably say something like:

It’s not bad but it’s clearly just a simulation. Does not feel like the real thing...:smilewink:

Short version of the stuff above:

Professional simulators are of a narrow scope and primarily procedural in nature. They skimp on details considered not relevant to the task at hand.

Interesting read, nonetheless.

On 8/5/2020 at 1:08 PM, Rifter said:

 

 

I really like what you write there.

 

Seasoned professional for numerical simulation methods in the automotive industry here.

 

What you say reflects perfectly my own experience.

Multi-million-euro car simulators are usually far below sim-racers for the PC when it comes to the driving dynamics side.

But the goal of a ‘professional’ car simulator is mostly just to create a certain workload on the brain of the test driver while he is testing new control elements or driving assist systems.

It is usually not meant for an accurate representation of driving dynamics behaviour of a car.

Fun fact: The so called multi body approach for sim-racers is very similar to the driving dynamics simulation tools in the automotive industry - in sim-racers they are just optimised towards performance because there you need real time performance.

The base for car simulators in the automotive industry is usually a ‘spread sheet’ simulation or in other words not more than a collection of look-up tables.

 

For professional flight simulators it is similar.

Civil aircraft (airliners) for example are never intentionally stalled - not for training and surely not for fun. Only test pilots do that during development.

Usually stall does also not take place in professional simulators. Pilots just get the warning signs recalled as a training (warning horn, stick shake).

Professional flight simulators which are capable of handling a so called upset recovery of an aircraft are at their very beginning:

https://trimis.ec.europa.eu/project/simulation-upset-recovery-aviation

 

Out-of-the-envelope behaviour of an aircraft in a simulation is the holy grail - it is yet to be developed.

At the moment outside-the-envelope behaviour is usually just 'scripted' or modelled as a rough approximation.

 

For the point of the “right-feeling" you mentioned:

Imagine taking a real car or a real plane and equip it with a remote control system and some cameras.

Show the camera pictures on pc-screens and add a gaming steering wheel or a gaming yoke or hotas.

Take test persons (professional test drivers and test pilots). Tell them what they see is a simulator.

Ask them how good it is.

They will most probably say something like:

It’s not bad but it’s clearly just a simulation. Does not feel like the real thing...:smilewink:

I see it this way. This is a game in the simulation genre after all!

On 8/5/2020 at 6:08 PM, Rifter said:

It’s not bad but it’s clearly just a simulation. Does not feel like the real thing...:smilewink:

There's a simple reason for this, actually: they're sitting in a chair and not in the vehicle.  We have a lot of options for a more realistic feel, up to and including full motion platforms (I'm gonna build myself one of those one day), but none of those allow you to experience more or less than 1G in total. Unless we develop artificial gravity or put our simulators inside centrifuges on a space station, this won't change, and even the latter case will feel funny at high Gs due to Corolis force.

In both real airplanes and real cars, the feeling of G forces is really important to achieving the right feel. This is also why seasoned simmers can routinely beat actual fighter pilots in the sim - the real pilot is deprived of some very important cues he's got in a real airplane, while the simmer is in his natural environment. 

3 hours ago, Dragon1-1 said:

There's a simple reason for this, actually: they're sitting in a chair and not in the vehicle.  We have a lot of options for a more realistic feel, up to and including full motion platforms (I'm gonna build myself one of those one day), but none of those allow you to experience more or less than 1G in total. Unless we develop artificial gravity or put our simulators inside centrifuges on a space station, this won't change, and even the latter case will feel funny at high Gs due to Corolis force.

 

In both real airplanes and real cars, the feeling of G forces is really important to achieving the right feel. This is also why seasoned simmers can routinely beat actual fighter pilots in the sim - the real pilot is deprived of some very important cues he's got in a real airplane, while the simmer is in his natural environment. 

When real fighter pilots have dueled simulator pilots in real aircraft using "laser tag" for gun shots, the simulator pilots often beat the professional pilots. Pro pilot training doesn't focus purely on ACM and doesn't provide nearly the hours of practice sim pilot addicts have. As drones supplant manned aircraft, the simulator pilots are going to become the aces of the future when the drone tech provides comparable of superior air superiority options that are not constrained to human pilot limitations. i.e. no only having higher g limits, but also not having to spend weight and space on cockpit provisions. 

Yeah, drones are superior to human pilots... up until enemy ECM crashes the datalink.  In real world, I just can't see them completely supplanting real fighters. For COIN operations? Sure, and this is already happening with the US drone strikes. For frontline combat against an enemy at similar tech level? Nope, they could, at most, be useful for aiding the human pilot, using short-range datalinks that would be harder to jam. Defensive operations over one's own territory could be doable with drones, but even then, you want human pilots on standby, for when things inevitably go tango uniform.

On 4/15/2021 at 1:51 AM, Dragon1-1 said:

There's a simple reason for this, actually: they're sitting in a chair and not in the vehicle.

Exactly! And also the fact that the input devices feel different from their counterparts in the real thing. There is more to building a simulator than just hitting the correct point on the mathematical/physics side of your driving or flight model.

In the transportation industry the basic situation for simulating accelerations is the following:

The perception of accelerations is done primarily in the inner ear, the equilibrium organ (in contrast to the perception of speed which is primarily accomplished on the basis of visual and acoustic information).

Because of the way the equilibrium organ works it is very good in sensoring the initial impulse of an acceleration. Whereas it is not very good in perceiving ongoing constant accelerations. In experiments pilots where seated as blindfolded passengers in an aircraft. The aircraft then flew different manoeuvres and the blindfolded pilots where asked to identify those manoeuvers. They where very good in recognizing the initial manoeuvres (like entering a dive or climb or roll). But they usually failed when for example the aircraft went into a sustained turn because their brains missed a certain rate of change of the acceleration. Here the visual information is crucial to actually identify the flight manoeuvre.

For driving and/or flight simulators that means it is more important to have the initial impulse of an acceleration (synchronized with the visual information) than the level of acceleration itself. This is the reason why motion platforms are doing an amazing job towards immersion even when they don’t reach the G-force level of the real vehicle.

Of course when the level of acceleration becomes part of the challenge itself (which obviously is the case for a fighter aircraft) you will have hard times in finding a substitute or surrogate for the real thing.

Advanced research in this area goes towards putting simulated pressure onto the body with for example pneumatic cushions and blowing air into the ears with sophisticated methods to create a fake impression of acceleration.

For the transportation industry there is lot of cost reduction potential for simulators in the future (cars, trucks, airliners) when it comes to minimize the technical effort by tricking the human brain rather than creating a technical/mechanical overkill by fully blown 6-DOF-platforms, especially in conjunction with VR-technology.

Though for simulating fighter jets I’m rather sceptical - the technology to get rid of the pilots themselves will probably advance faster than the technology to simulate fast jets in a way you could not tell the difference to the real thing...

2 hours ago, Rifter said:

Exactly! And also the fact that the input devices feel different from their counterparts in the real thing. There is more to building a simulator than just hitting the correct point on the mathematical/physics side of your driving or flight model.

To be fair, the FM is the part ED can do. Motion platforms, floor-mounted sticks and pedals with 40kg actuation forces, switch panels and all that jazz is up to us to provide (and just as well, I love to tinker with this stuff).  I would appreciate more native support for that sort of thing, though. The real obsessive simmers know these are the way to go.

On 7/24/2020 at 4:46 PM, Wags said:

Dear DCS Pilots,

 

DCS World Flight Modelling Principles document: https://www.digitalcombatsimulator.com/upload/iblock/dcc/DCS FM principles plus MiG-29 P-47 F-16.pdf

 

Perhaps the least understood by customers, and the most complex task for us, is the creation of aircraft flight dynamics for DCS World aircraft. To help shine some light on this mysterious subject, our flight model engineering team created the linked document to share the sources we use, the fundamental principles behind the engineering, and the techniques we use to create and test the most accurate flight models possible.

 

It is our hope that with a better understanding of the process that you will truly appreciate the time and effort that goes into creating a flight model for DCS World and why it is not a speedy or simple task.

 

Thank you and kind regards,

The Eagle Dynamics Team

If you did the CFM model, can you use it in the game also to show how they can modify the smoke and dust  when we fly trough them ? and make some nice smoke twist funnel. 

I think you are did it something similar for the jetwash turbulence but it is not visible on the smoke and dust , we just "feel" the effect.

On 4/15/2021 at 4:20 AM, Dragon1-1 said:

Yeah, drones are superior to human pilots... up until enemy ECM crashes the datalink.  In real world, I just can't see them completely supplanting real fighters. For COIN operations? Sure, and this is already happening with the US drone strikes. For frontline combat against an enemy at similar tech level? Nope, they could, at most, be useful for aiding the human pilot, using short-range datalinks that would be harder to jam. Defensive operations over one's own territory could be doable with drones, but even then, you want human pilots on standby, for when things inevitably go tango uniform.

Do AMRAAMs kill aircraft with a very high PK? Is anyone piloting them? Can they be lobbed at tremendous ranges? They get midcourse updates, but are autonomous. The aircraft is merely the first stage of a two-stage SAM. If you don't think an F-15 can be automated in the near future to rival if not beat human pilots, you don't understand where the technology is now and where it will be in the very near future. I expect drones will still be manned/monitored, but if jammed, they will at a minimum defend themselves and try to get back into comms range or fly home and depending on the situation may go offensive against their last designated target and/or the source of the jamming. I am not eager to see pilots replaced, but it is a matter of time before the drones, piloted and/or automated, will make up the bulk of the forces.

What an AI controlled aircraft needs is cameras / sensors in all directions and the processor power to understand and correlate that information in a way that is comparable and/or superior to human pilots. When it is there it will mix with dogfight AI that was just recently demonstrated to consistently beat the best real pilots in a simulator environment, where the only key difference is that the AI gets fed perfect information for making decisions. I am not in the loop, but I am sure there are some DARPA type people that are much further along with this kind of tech than is publicly admitted. The very old F-15 IFFC/Firefly system was a basic integration of flight controls with fire control. When the pilot pulled the trigger, the gun didn't fire, it gave the computer to fly the aircraft and fire the gun and it worked exceptionally well. Imagine where the tech is now.

1 hour ago, streakeagle said:

If you don't think an F-15 can be automated in the near future to rival if not beat human pilots, you don't understand where the technology is now and where it will be in the very near future. I expect drones will still be manned/monitored, but if jammed, they will at a minimum defend themselves and try to get back into comms range or fly home and depending on the situation may go offensive against their last designated target and/or the source of the jamming. I am not eager to see pilots replaced, but it is a matter of time before the drones, piloted and/or automated, will make up the bulk of the forces.

Oh, I know where the tech stands. AI is dumber than you seem to think, and it's not getting much smarter on a deterministic machine. No amount of processing power will fix that, because it's inherent in how computers work. No matter how much these systems seems like they're thinking, they're not. They're just getting better at pretending. It can learn how to fight in a jet, if given perfect information of a simulated environment. In real world? I don't think so. Even if they do get image recognition working (a big if, my experience with the tech in scientific context was very negative), a human with a real brain and real eyes will generally be more reliable at identifying and classifying targets. An AI could be a tool for a thinking human, nothing more, especially in a modern environment, with electronic warfare, civilians who might be close to the target, and complex situations that require thinking outside the current tactical situation. Pilots get briefed on these things for a reason. You can train the AI to play a dogfighting game, but not fight a war.

I also happen to know where the international law stands. And there are people asking very serious questions about whether what you propose should be legal. I, and many others, lean towards answering "no". A missile's life is measured in minutes, it goes after one specific target, and if its datalink is jammed early enough, it gets trashed (oh, and an AMRAAM's range is hardly "tremendous" compared to other missile types, a SA-10 would be a better example). The most that should be allowed for a drone is "RTB if datalink lost", which is something that you can get even in model airplanes these days. And that's not even getting into situation in which enemy might outright hack and steal your drone (Iran already did that with one US UAV). Hacking a human pilot is rather harder.

Have you considered using raw test data from sources such as NASA AMES? There is a raft of stuff in there you can use to develop aerodynamic derivatives. 

Pretty sure the low speed high alpha/beta stuff will have been Nerfed, for obvious reasons. But there are research papers going back to the 80s for early gen F16/F14 etc which may well help

It's amazing to see the actual processes involved not to mention a willingness to share such info, I'm guessing "moments" that are converted to dimensionless coefficients are the "look up tables"?

I wonder if we'll ever have enough compute power for bottom up circuit modelled flight sims or are we consigned to a top down process?  Due to the sheer amount of calculus involved with circuit modelling will we always be beholden to the resolution limit of the tables themselves?