Gaming Xplane: Accuracy of Simulation

[Mojave]

The following is a thread to discuss Xplane's accuracy of simulation compared to other flight simulators, as well as an initial platform for Thorsten to voice his opinion on blade element theory. The context is listed below:

Yeah, I know people who have spent thousands upon thousands of dollars to get FSX up to any kind of real standard. It's a great simulator! It just misses the mark as the best. Honestly, if I had a rig worth its weight in gold, I'd be running Xplane. All that sweet adherence to shape, and forces simulation.

I have an opinion about the accuracy of that as well (to my knowledge, X-plane never got an FAA certified simulation running anywhere, which is kind of the gold standard for accuracy) but that's leading too far off the thread - if you're interested in my take on blade element theory, feel free to bring up the question elsewhere.

NOTE: Quotes taken from the original thread found here: http://www.orbiter-forum.com/showthread.php?p=558628&postcount=26

 

[Thorsten]

Okay, so what Blade Element Theory claims to be able to do is to derive the forces and moments acting on an aircraft 'with incredible accuracy' (X-plane website) from the shape of the aircraft 3d model in real time (i.e. in single frame updates, i.e. 1/60 of a second).

In the reality of it, there's a complicated airstream around the aircraft which may in places get turbulent. The direct way to solve that physics problem is to take the equations of fluid dynamics and solve them around the body, thus get the velocity field and the forces and moments acting on the embedded 3d model. That has to be done numerically.

Now, unfortunately turbulence is one of the things which are not really solved in computational fluid dynamics (CFD), there's parametric solutions to it, but because of the cascading of eddies to different scales it's very difficult because conceptually you need near-infinitely fine grid sizes.

Talking to aerospace engineers who have spent a professional lifetime with prototyping aircraft or coming up with simulations, they all told me CFD works to a point, but always has to be augmented/verified by good old wind tunnel testing, because only that gets the turbulence right.

CFD is not a thing that runs in 1/60 of a second - it's usually something that takes a few hours for a configuration.

If you look into design documents or aircraft test data (NASA has plenty in public domain, Saab has published a detailed set on the Viggen,...), they do all the same: Parametrize wind tunnel and CFD data into the form of coefficient functions which in turn depend on multiple variables. For instance there might be a yawing moment coefficient as a function of sideslip angle, AoA, Mach number, rudder deflection etc.

Such coefficients represent CFD or wind-tunnel accuracy for steady-state airstreams (you can derive their existence from the basic equations using the assumption that the unperturbed airstream on the nose of the aircraft doesn't differ from that at the tail - so the assumption is wrong during the time the aircraft is passing through a shockwave as caused by an explosion, but it's really good otherwise)

Based on such a coefficient scheme, one can build a real time simulation which (apart from external shockwaves) is as accurate as wind tunnel testing and CFD can make it.

Now, Blade Element Theory is an approximation of CFD which allows to get something like the real solution in a fraction of a second rather than in hours. From the difference in time, you can start to suspect that there are limits to the accuracy this can achieve. It's going to be reasonably good for non-turbulent air and deliver nice and fast results for aircraft inside their nominal operating profile - but it's going to be massively off where CFD starts being off, i.e. when the airstream gets turbulent, when the onset of stall is reached,... - the region where wind tunnel testing is most important (I think the company isn't called 'Laminar Research' for nothing - turbulence is not their thing...)

My comparison example is that you can (for reasonably small angles) replace sin(x) by x - and this is going to compute much faster and easier in complex expressions and will give you almost the same result. Except when it does not, when x gets large - then it will produce amazing crap. And I think that's about the relation of Blade Element to CFD/Wind tunnel - great and much faster if the airstream is largely laminar, uncontrolled when not.

There's also small details, like the people who do CFD modeling of aircraft also derive the flight characteristics from a 3d model - but they also comment that the accuracy has to be sub-millimeter to get good results (which isn't really the case for most X-plane 3d models) and that surface characteristics such as whether there's a smooth paint or a rougher surface also matter quite a lot.

From that it would seem that the input data used by Blade Element Theory often is not sufficient to get really good results - and in practice people who have created flight dynamics models for X-plane confirm that Blade Element Theory get you the basics, and if you want to get better it's gets hard because you need to do all sorts of hacks and fudges to tune it.

So, in summary, the aerospace industry seems to go with wind tunnel data /CFD in coefficient schemes for real applications, I understand why that this makes sense by deriving the scheme from the fundamental equations using a steady-state approximation, and Blade Element is worse than CFD and certainly much worse than wind tunnel, so I don't see how the label 'realistic' applies in general - it's a good approximation inside the nominal envelope, but that's just it.

And I suspect that the fact that there's no certified trainer simulation utilizing X-plane is caused by just that - you can't trust that the simulation teaches you to handle off-nominal situations correctly.

Hope that explains my stance.

 

[Ravenous]

Hmm aren't turbulent conditions part of the norm for some aircraft? I'm not knowledgeable on this at all but I read somewhere that relatively sharp airfoils trigger vortices that contribute quite a lot of the lift. (Examples - Concorde's leading edges, SR-71's "chines" at the nose, etc.)

Admittedly these planes are extreme examples but the effects are claimed to matter at relatively low speeds, such as landings.

So I wonder if this theory actually works at all for airframes like that...

 

[Thorsten]

So I wonder if this theory actually works at all for airframes like that...

I would think not really, and I suspect there are ad-hoc tuning factors in X-plane for situations like that. But admittedly I haven't studied how such a plane is actually implemented in X-plane.

 

[Xyon]

So - pardon my ignorance - but how does FSX differ here to be more realistic (and I assume it is, because of the comment about FAA approvals up top)? The issue with doing CFD calculations in sub-second intervals persists, so how is the FSX solution better?

 

[dbeachy1]

I'm no expert, but as I understand it, FSX can have "more data included in a given aircraft's flight model" because it's computed independently of the mesh, so it can have real-world data from wind tunnel testing for that aircraft integrated into it. X-Plane, conversely, tries to infer the flight model based solely on the aircraft's mesh's shape -- which, as Thorsten eloquently pointed out, has significant limitations when flying in less-than-perfect conditions. But of course I could be over-simplifying this.

 

[Majid]

I'm no expert, but as I understand it, FSX can have "more data included in a given aircraft's flight model" because it's computed independent of the mesh, so it can have real-world data from wind tunnel testing for that aircraft integrated into it. X-Plane, conversely, tries to infer the flight model based solely on the aircraft's mesh's shape -- which, as Thorsten eloquently pointed out, has significant limitations when flying in less-than-perfect conditions. But of course I could be over-simplifying this.

I think you're right. This thread has more information: http://forums.x-pilot.com/forums/topic/4614-comparable-flight-dynamics/

But according to Cameron who is the founder of X-Aviation:

This would be the .acf files. In X-Plane, the geometry in most cases defines the actual flight model, unlike MSFS which uses tables. That said, you would use the Plane Maker program located in your X-Plane directory to open an aircraft and edit the parameters for it.

 

[Thorsten]

So - pardon my ignorance - but how does FSX differ here to be more realistic (and I assume it is, because of the comment about FAA approvals up top)?

Elaborating on dbeachy1's answer:

I have to be nit-picking here - it's difficult to speak of 'FSX being more realistic' in general in face of the totally unrealistic flight dynamics you get with the base package.

Rather, the simulation scheme (coefficient function) allows you to potentially have more realistic flight dynamics - if you feed it with the right data. Which you need to have (and there lies the rub).

If you have a case like the Saab Viggen mentioned above where you have pages upon pages of coefficients from wind tunnel measurements and use all these in a coefficient function scheme, the resulting flight dynamics will be as realistic as it can get - it's basically based on reproducing all measurements taken on the plane everywhere (I have had the pleasure of flying this simulated Viggen, and the high-AoA behavior, the departure from stability and the recovery is just... unbelievable - it's a joy to fly. I have piloted a couple of airplanes for real (gliders and single prop GA craft, no fighter jets of course) and there's something to real piloting which is difficult to capture in simple schemes, a certain 'feel' in the way they handle - hi-fidelity wind tunnel data gets this just right) ).

Conversely, if you have a classified plane (say the B-2) for which you can't readily get tables of data, the Blade Element approach is actually not so bad because it will deliver you something very reasonable based on what you know (the exterior shape of the plane).

Though, theoretically, you could put that same data into a CFD code and compute the flight dynamics with higher accuracy than Blade Element theory could. But running your own CFD code isn't for the casual plane developer, you really need to do this (semi-)professionally to get something good out of it and I suspect that's not regularly done for desktop sims.

CFD vs. Blade element is a bit like comparing raytracing against real time rendering. Raytracing is going to give you better results because it can take half an hour to compute, real time rendering has to finish within a frame. Likewise, if set up properly CFD will always beat Blade Element because it has much more time to solve the same problem and can hence delve into the details. Computing stuff offline rather than within a frame and then reading it from a table in-sim generally has (much) higher accuracy.

But you have to know what you're doing, and most of us can't set up wind tunnel tests or run a CFD code.

 

[martins]

I'm no expert, but as I understand it, FSX can have "more data included in a given aircraft's flight model" because it's computed independent of the mesh, so it can have real-world data from wind tunnel testing for that aircraft integrated into it. X-Plane, conversely, tries to infer the flight model based solely on the aircraft's mesh's shape -- which, as Thorsten eloquently pointed out, has significant limitations when flying in less-than-perfect conditions. But of course I could be over-simplifying this.

Yes, that is my assumption as well. Essentially there are two fundamental methods of describing the behaviour of an aircraft in the simulation:

1. You build a forward model of the physics of the problem, usually by a complex numerical simulation (FEM, CFD), that describes the geometry of the vessel, its elastic properties, the physical properties of the airflow around it and their interaction, etc. Such simulations can be quite computationally expensive, and their accuracy may depend on the input parameters and the specific conditions of the simulation.

2. The other method is to sample the behaviour of a real aircraft as a function of relevant parameters (airspeed, incidence angles, air pressure, load distribution, etc. etc.). The simulation essentially then simply consists of an interpolation of these sampled data in the multidimensional parameter space. This second method is entirely independent of the physics of the problem. It is just an elaborate interpolation mechanism. Provided that the sampling is sufficiently dense, and covers all parameter ranges encountered in the simulation, it can be quite accurate. The magic lies entirely in the acquisition of the samples.

Of course you could combine the two methods. You can have a high-fidelity physics model to generate the samples on a supercomputer, and then use those to instruct the interpolation model that can be run at real time on a moderate platform.

A good analogy for this audience may be the computation of planetary ephemeris. You can write a highly accurate dynamic multi-body model that propagates in sufficiently small time steps to keep the errors under control, but is too slow to use in a real-time (or time-accelerated) simulation, and can't give you planet positions at arbitrary times.

Or you can use a series expansion of the state vectors that interpolates observation data (or data generated by the above high fidelity model). That model is very fast, generates data at arbitrary times, and allows to control accuracy by the number of coefficients included. Of course, such a model can only be as accurate as the data that were used to generate it.

 

[n122vu]

Okay, so what Blade Element Theory claims to be able to do is to derive the forces and moments acting on an aircraft 'with incredible accuracy' (X-plane website) from the shape of the aircraft 3d model in real time (i.e. in single frame updates, i.e. 1/60 of a second).

X-Plane, conversely, tries to infer the flight model based solely on the aircraft's mesh's shape

This is a common misconception about X-Plane. The only time the flight dynamics of an X-Plane aircraft are directly tied to the 3D model is when you first save the 3D model and the .acf file is generated. From this point on, you can make modifications to the .acf data using PlaneMaker, completely independent from the 3D model.

From the PlaneMaker manual:

X‑Plane does not look at the shape of the wing and then decide how much lift, drag, etc. the foil will put out—X-Plane is not a computational fluid dynamics program. Instead, X‑Plane uses pre-defined airfoils that list the performance of any airfoil (lift, drag, moment) to predict how the plane will fly with that foil.

No, X-Plane does not look at the 3D model in real-time and decide how the plane will fly or how it will react to the wind. It looks at the predefined airfoil and parts definitions (fuselage, gear struts) generated by PlaneMaker and decides how each individual part would react, then averages how they would affect each other as a whole, and simulates the results accordingly.

I've had this discussion several times on my Twitch stream. People on the X-Plane side of the 'argument' chide FSX and ESP+ derived sims as basically a "flying spreadsheet." My answer to that is, FSX is to a single spreadsheet as X-Plane is to a multi-sheet workbook. Basically, when you edit a plane in PlaneMaker, what you're essentially editing is different tables associated with that aircraft's performance, right down to the shape of the fuselage, the length of the prop, the horsepower of the engine. It's no different, it's just more data sources at runtime.

While Plane Maker is perfectly adequate for entering the design of the plane-it is reliable and easy to use-Plane Maker is not a 3-D modeling program. Instead, it is used to simply lay out the basic aerodynamic shapes and properties of an aircraft. This results in a model that looks okay, but not great. Plane Maker cannot make the complex model required for a 3-D cockpit, and it cannot create a highly detailed model of the craft as a whole to overlay the basic Plane-Maker model from which flight physics are calculated.

The last sentence says it all. Once the .acf file is created, you can use a third-party 3D modelling program to generate a high-quality 3D model of the aircraft. X-Plane will completely ignore this 3D model and determine the flight characteristics based on the .acf file.

 

[Parnikkapore]

AFAIK (I've used X-Plane Demo before), the blade element thing calculates flight dynamics off the model for the fuselage,but off the specified values only for the wing segments.
I could be wrong, I haven't really made a plane.

 

[Thorsten]

This is a common misconception about X-Plane. The only time the flight dynamics of an X-Plane aircraft are directly tied to the 3D model is when you first save the 3D model and the .acf file is generated. From this point on, you can make modifications to the .acf data using PlaneMaker, completely independent from the 3D model.

This 'common misconception' is prominently advertized by the X-plane website:

X-Plane works by reading in the geometric shape of any aircraft and then figuring out how that aircraft will fly. It does this by an engineering process called “blade element theory”, which involves breaking the aircraft down into many small elements and then finding the forces on each little element many times per second. These forces are then converted into accelerations, which are then integrated to velocities and positions… Of course, all of this technical theory is completely transparent to the end user… you just fly! It’s fun!

Or, as they go further down:

With blade element theory, though, as used in X-Plane, you can enter the shape of an airplane and let X-Plane figure out how a plane of that shape, weight, power would fly!

The way this is sold to the general public (and advertized as superior to 'mere tables') is that there's a direct correlation to geometry.

I don't think my characterization of, say,

The airfoil data entered in Part-Maker is 2-dimensional, so X-Plane applies finite wing lift-slope reduction, finite-wing CLmax reduction, finite-wing induced drag, and finite-wing moment reduction appropriate to the aspect ratio, taper ratio, and sweep of the wing, horizontal stabilizer, vertical stabilizer, or propeller blade in question. Compressible flow effects are considered using Prandtl-Glauert, but transonic effects are not simulated other than an empirical mach-divergent drag increase. In supersonic flight, the airfoil is considered to be a diamond shape with the appropriate thickness ratio; pressures behind the shock waves are found on each of the plates in the diamond-shaped airfoil and summed to give the total pressures on the foil element.

as doing a fast approximation of fluid dynamics in real time is in any way wrong or misleading. And I don't think it's misleading to state that e.g. 'an empirical mach-divergent drag ratio' is not on the same accuracy level as genuine wind-tunnel data.

And I think it has also been stated that X-plane allows you to apply fudge factors after the Blade Element has been applied.

It's no different, it's just more data sources at runtime.

It's in fact pretty different whether you have a set of coefficients from measurements//computations on the whole plane, or whether you try to build them from from a sum of the coefficient of individual components.

The latter is less accurate because perturbations to the airstream generated at the nose of the plane may affect the airflow downstream. According to the above description link, X-plane accounts for propwash or downwash empirically, but again this is not on par with actually calculating the perturbations in the airstream by CFD or measuring them. Incidentially, this is where turbulence comes in again, because often the downwash is to some degree turbulent.

For that reason, the industry standard is wind tunnel tests or CFD and coefficient sets on the whole body of the plane. Having data for all wing elements formally seems like having more, but in fact it's less information (unless you break the wing that is) because the correlations between what you have can't be reconstructed, whereas a full body coefficient table has it all.

For the audience here, there's an interesting space story illustrating this: One reason the Space Shuttle launches in inverted flight is that the wing bending moment during atmosphere passage in heads-up attitude was found to be too large. Engineers measured this in the wind tunnel, but it was only a decade later when CFD was viable that they discovered the reason was the equipment box structure on the SRBs which caused a shockwave hitting the wings, thus much increasing the wing bending moment of the whole launch stack over what the orbiter alone would have.

Good thing they did full launch stack wind tunnel testing and not just summed individual elements!

 

[boogabooga]

I'm not very familiar with X-plane and I've only heard of "blade element theory" in the context of X-plane. But, I have an informed opinion about this stuff in general, so I would like to weigh in.


First, a reminder about CFD. Real CFD grids are negative space. That's the opposite of the 3D model mesh that simulation/ game software uses. Also, the grid has to be specially constructed. So X-plane couldn't do CFD even if it wanted to and had the resources- it lacks the grid information.


I don't think Thorsten's and n122vu's interpretations of "the common misconception about X-Plane" are mutually exclusive. First of all, the kind of information that you get by looking at the geometry of an airfoil is more along the lines of coefficients of lift and drag as a functions of angle of attack, Reynolds number, etc. That's not the same as lift and drag. You still have to multiply by dynamic pressure (i.e. how fast you are going and how dense the air is). The software still has to "find the forces on each little element" in real time whether the Cd-Cl response is pre-defined or not.


I think what they are saying is this. You start with pre-defined airfoils with a known Cd-Cl response. Then you assemble those into the "geometric shape of any aircraft". The program makes some kind of corrections to the Cd-Cl response to account for various interactions, finite span, etc. Then in the actual flight simulation, the forces on each "airfoil" are found in real time based on the corrected Cd-Cl response and the current dynamic pressure. All forces are summed, and the a/c is accelerated accordingly. Not as accurate as CFD with a good turbulence model, sure, but better than nothing in the absence of data.

Interestingly, there IS a method of looking at a 2D airfoil geometry and very quickly determining the Cd-Cl response without doing full CFD. It's called the Vortex Panel Method. It's basically the CFD hack that Thorsten though "blade element theory" was.
http://www.engapplets.vt.edu/fluids/vpm/vpminfo.html
http://web.mit.edu/drela/Public/web/xfoil/

 

[Reticuli]

The following is a thread to discuss Xplane's accuracy of simulation compared to other flight simulators, as well as an initial platform for Thorsten to voice his opinion on blade element theory. The context is listed below:





NOTE: Quotes taken from the original thread found here: http://www.orbiter-forum.com/showthread.php?p=558628&postcount=26

FAA-Certified X-Plane | X-Plane

Any copy of X-Plane purchased from X-Plane.com can have the features required for FAA certification unlocked using a special USB […]

www.x-plane.com

FAA-certified X-Plane is not exactly new or anything. Been around for some time.

The claim that database/tables can do extensive edge-of-envelope flight dynamics stuff well would be incorrect, obviously because the majority of the data these simulations are based on is not extremis/novel data. This is the reason Flight Unlimited's reliance on CFD did aerobatics well, because it could get the edge-of-envelope performance of any complex shape right if you dropped or propelled it in its environment, but basic landings/takeoffs exhibited slightly goofy ground effect behavior sometimes and was arguably too forgiving of hard touch & goes even with its amazing damage model. It also means X-Plane does aerobatics worse than Flight Unlimited but better than FSX/P3D does. And of course it means the 'tinker toys' aspect of X-Plane means you can actually make your own plane in Planemaker and see how it might roughly perform in the real world... something that would have been really interesting to see in a genuine FU sequel if Looking Glass hadn't been so stupid and fired Seamus Blackley.

 

[Sbb1413]

An android version of X-Plane is available, but it is too sizey to use in my android phone (1.12 GB). I will look for the PC version of it.

Link: https://play.google.com/store/apps/details?id=com.laminarresearch.x_plane10

Addendum: I have installed the free demo version of the simulator. I shall* purchase the full version of it in the future.

* You may be thinking why I use "shall" instead of "will" in some cases. Because in traditional grammar, "he shall" represents that he must do something in the future, while "he will" represents that he may or may not do something in the future. However, "I shall" represents that I may or may not do something in the future, while "I will" represents that I must do something in the future. Therefore, "I shall purchase the full version of it in the future" means "I may or may not purchase the full version of it in the future."

 

[Moach]

AFAIK, The MS-FS series is based on a fully parametric flight model.

That is: It solves flight dynamics for the entire aircraft as a unit whole, using a wide array of config parameters to specify just exactly how each particular type performs across its flight envelopes.

Basically, it turns to scalar values and lookup tables for pretty much everything, the only non-graphics consideration given to aircraft shape is the definition of control points for various items around the airframe (like wheels and contact points)



To my understanding, X-Plane uses a very different approach, which is "Finite Element Theory"


This method is closer to how my own MotorWings flight model works. It seeks to approximate the influence of discrete bits of the airplane on the air flowing around them, which in turn affects the airflow on parts behind and around.

This is technically not a CFD-based solution. Which means, there is no actual simulation of air particles or volumetric mass flows around the aircraft. All deflections of the air stream are accounted for only where trailing airplane bits encounter these flows. Air flow in spaces between parts is approximated by parametric functions, which are then tuned up to yield authentic results.

The resulting forces from each part are then applied through the physics engine and we move on to the next frame... PhysX sorts out the aeroelastic forces of the joints between those parts in the meantime, so our "collection of parts flying in close formation" actually handles like a single airplane. (at least until you snap off a wing, which is quite easy to do, rly)

debug visualization screenshot: cyan lines indicate lift output from each wing segment - tip vortices are shown in black


Well, that's how my sim works, at least.



X-Plane is surely more "sciency" about it, whereas MotorWings (being a fictional world and all) takes some creative liberties and thus cuts a few corners by going a bit more towards how FSX works. - As it evolves, I cycle back and replace these empirical "short-hand" solutions with something more theoretically sound as to keep up with other developing aspects of the sim.




What I've found, from over 20k hours of flightsim experience across all levels of complexity, and then developing my own sim - No matter how intensely scientific you get about it, in the end you'll inevitably end up having to fine tune a bunch of "tweak values" to account for all the assumptions used in the flight model.


Getting these right inevitably becomes a subjective matter which can only be managed by lengthy trial and error adjustments.

A "Perfect Simulator" should have no need of any tweak settings. It should all work itself out mathematically... right?


Except that... Such a "perfect sim", would require that every quantum of matter gets solved for with 100% precision. (chaos theory and all) No present-day or foreseeable future computer can possibly do that in real time. There is also the problem of the basic theory behind it all being still beyond the full understanding of human science.


Because of this, there is only one simulator that we know of that can actually be deemed "totally realistic" - We call it "The Universe". It's graphics are great, but the gameplay is really quite frustrating.

 

[Sbb1413]

Except that... Such a "perfect sim", would require that every quantum of matter gets solved for with 100% precision. (chaos theory and all) No present-day or foreseeable future computer can possibly do that in real time. There is also the problem of the basic theory behind it all being still beyond the full understanding of human science.

Thanks to the way technologies evolve, (the evolution curve of technology is exponential) we can expect a simulator that could be 99.9999999991% accurate in quantum level at the end of the 21st century.

Because of this, there is only one simulator that we know of that can actually be deemed "totally realistic" - We call it "The Universe". It's graphics are great, but the gameplay is really quite frustrating.

Life is a game, LOL. If there are multiple universes, (according to several unverified multiverse theories) then we wouldn't be sure which one can be deemed "totally realistic", which will bring us into questions like "what is real and what is not real? Is there an absolute truth?"

 

[Reticuli]

Life is not a simulation, unless you're willing to posit that either there's only one user in the simulation (which I'll claim is me, obviously) or that relativity doesn't actually work. Multi-user real-time simulations cannot have relativity.

As for having to tweak the simulations to meet the failings later on, it's still curious that in the inverse, the 25 year old original Flight Unlimited 1, if you'll allow me to digress back to that old goldie since CFD was brought up in this thread earlier, had such an astounding flight model that issues with how aircraft seemed to be performing were nailed down to deflection rates of control surfaces, exact 3D geometry of the airframes, etc, and once they got that right, the performance aligned with what it should be. Impressive one-off achievement in the field of flight simulations.

 

[Linguofreak]

* You may be thinking why I use "shall" instead of "will" in some cases. Because in traditional grammar, "he shall" represents that he must do something in the future, while "he will" represents that he may or may not do something in the future. However, "I shall" represents that I may or may not do something in the future, while "I will" represents that I must do something in the future. Therefore, "I shall purchase the full version of it in the future" means "I may or may not purchase the full version of it in the future."

Not quite. "will" originally meant "want to", and "shall" meant what "should" does now (with should being at that time the past tense, both indicative and subjunctive, of shall). As both words became primarily auxilliary verbs for the future tense, will retained an implication of intent, and shall retained an implication of obligation or advice. The rule of thumb that "will" is stronger in the first person and "shall" is stronger in the second and third persons comes from the fact that when speaking of oneself, indicating ones intent tends to be more emphatic, while when speaking of others, indicating the other party's obligations tends to be more emphatic. This usage remains in England, but in North America (and, as I've heard, in Scotland), "shall" is used only in stock phrases and legal language, and "will" is used as the future auxiliary in everyday speech, and rarely carries any implication of intent.

This results in a joke about an American (or Scotsman) who, not knowing the proper use of "will" and "shall", drowned in an English lake screaming "I will drown, no one shall help me!"

 

[Padre Pedro]

For the past 6 months or so I found that I am extremely into gaming simulators. I used to be a fan of PoE and bought a lot of currency playing it but now my gaming tastes change greatly. That is awesome