For the FDE Guru's and Tweakers

[Tom Clayton]

Have you ever taken a look at the formulas for calulating the MOI's for pitch, roll, and yaw? If you don't have a firm grasp on at least high school level algebra, they can be a pain. I'm not bad at it, but I still have to double-check my math once in a while. But MOI's are the first thing I check when I come across a plane with ground handling that doesn't "feel" right. So I decided to dust of my Excel skills and make a shortcut for myself.

The fomulas are taken straight out of the Aircraft Container SDK and then reformatted to work in Excel. Simply enter the required values into the first three cells in the second column and the formulas will do the rest.

 

[zmike]

I am afraid it should be obvious that the formula used by Microsoft is the problem not the solution. You should not look to Microsoft for aviation expertise. Study of relevant real world textbooks is required.

Each relevant MOI depends on distribution of mass *in the default MSFS load state* prior to addition of fuel. Repeating the Microsoft SDK error of describing MSFS default mass minus fuel mass as 'empty weight' is therefore also counter productive.

Each MOI depends on distribution of mass, not just total mass. The distribution of masses in a B-17 is different to a P-51. Estimation of roll and yaw MOI requires very different formulae for those two aircraft. High and low wings require different formulae. Swept wings require different formulae. Aircraft with tail mounted engines require different formulae, a fighter with, and without, underwing weapons in MSFS requires different formulae etc, etc.

In the absence of precise mass distribution data, MOI is exeptionally difficult to estimate to even two significant figures. Relevant textbooks propose proprietary and copyright 'solutions'. When applied to aeroplanes of unusual form the results from those different proposals in different textbooks can disagree by more than 100% and they are unlikly to agree to two significant figures for most aeroplanes of conventional form.

Consequently FD 'gurus' never use the MS SDK method of MOI estimation. They each tend to believe in the many required and different estimation formulae in one particular textbook and tend to stick with them. If they cause an error in MSFS they the FD author varies the 'formulaic' estimate until the obvious error is eliminated.

Ground handling should never have priority over in flight handling in any flight simulator. MOI should be coded for the in flight case. If ground handling errors exist they should be resolved using MSFS variables that have no influence on in flight handling. Base MOI variation should be used to 'fix' ground handling errors only as a last resort.

Fuel MOI addition is calculated separately in real time with MOI in each axis varying greatly and differently throughout the flight depending on whether fuel % of total mass is large and how it is displaced from CG, on this route, with the fuel we loaded for the weather today.

Many / most FD 'gurus' believe that additional MOI in each axis due to fuel placement and mass is miscalculated within MSFS. This may, or may not, be true when base MOI %, versus fuel remaining MOI %, is accurate, but it rarely will be. It cannot be accurate if today's payload is not default payload. The claim in the MSFS SDK that variable payload MOI is calculated by MSFS is also pure fantasy. 'Perhaps' it is calculated and varied by 'some' versions of MSCFS.

The total MOI in each axis is heavily dependent on *current* fuel mass and placement. Consequently correction of ground handling via movement of fuel masses versus CG is a better solution than coding false 'zero fuel' MOI, particularly in aeroplanes with many tanks. All of the problem may relate to fuel MOI in a single pair of outboard tanks contributing too much to total MOI, (in just one or several axes), only while those tanks have fuel in them. On many flights they must have none at take off anyway.

Failure to load and manage fuel in accordance with the real life requirements is a pilot error that the FD author can address only by encoding and demonstrating the unpleasant consequence. Failure to reduce to maximum landing weight, with specific fuel tanks always empty, and others with minumum fuel remaining, before attempting an approach, is a requirement often ignored by MSFS users. Attempting to taxi with the wrong fuel mass, in the wrong tanks, may indeed be a nasty experience.

During product development estimating base MOI in each axis is a complex problem with no correct solution. All real life estimation methods lead to significant errors. Evaluating whether MSFS is calculating total MOI in each axis correctly using real fuel placement, added to FD developer estimated base MOI, as fuel is used, (or never loaded), is even more complex.
The price of realism within flight simulation is complexity, both during product development, and during later use. 'Tweaking' of MSFS flight dynamics, by those who do not understand how they work, using formulae that cannot possibly work, is not something to encourage.

FSAviator.

 

[Milton Shupe]

It is always interesting to read your insight; always appreciated.

So now we are left with the question as to how to calculate MOIs to get the sim to work somewhat nicely in FS irrespective of the facts you highlight. We need some reasonable response around each axis that is somewhat believeable. Right now, all we have is what MS recommends for the sim, not real world, and I suspect that has to be what we live with.

Thanks

 

[Tom Clayton]

Just like the MS formulas, my little toy here isn't designed to find the "final" figure for any aircraft. It's simply a shortcut for people that don't like Algebra.

I fully understand that planes with similar dimentions can have vastly different MOI's. As I understand it, the A330 and A340 have essentially the same wing. (Don't flame me if I'm wrong - this is just an example.) But with two extra turbofans hanging outboard from the first two, the roll and yaw MOI's will be quite a bit higher. But then, I also understand that Airbus programs all of their fly-by-wire systems to "feel" as close to their other planes as possible, making cross-training easier. So it's all very much a subjective matter, and the only way to really get it right is with the cooperation of a real-world type-rated pilot that's also intimately familiar with the inner workings - and limitations - of Flight Sim. Without that kind of input, all anyone is really doing is making a WAG anyway.

 

[zmike]

<<So now we are left with the question as to how to calculate MOIs to get the sim to work somewhat nicely in FS irrespective of the facts you highlight....... Right now, all we have is what MS recommends for the sim, not real world, and I suspect that has to be what we live with.>>

We have known since 1997 that the Microsoft inertia equations are nonsense. The reason there are so many bad FD 'out there' is the defeatist conclusion that 13 years later FD developers still have no choice but to copy Microsoft errors.

If Microsoft modelled a default aeroplane the wrong shape does every subsequent MDL author have no choice but to repeat the error? How many third party MDLs would we have if every MDL author refused to buy or borrow real books from a library for study before bothering to develop an error free MDL?

Don't FD authors have an equal obligation to eliminate Microsoft's errors instead of perpetuating them? Nobody needs to guess how flight dynamics work. The source of the problem is well known. The solution is well known. It sesm to me that what is missing is the resolve to eliminate the problem unless it is a free download.

The two 'most favoured' inertia estimation techniques within the FD author community are USAF DATCOM which 'seems' to have been adopted as a 'standard' by the CFS community, and which 'may' be best suited to their needs. The USAF DATCOM methodology 'appears' to be 'freely available'.

http://www.holycows.net/datcom/Digital_Datcom_Users_Manual_1.2.pdf

http://www.holycows.net/datcom/AIAA-1985-4070-141.pdf

http://rapidlibrary.com/index.php?q=usaf+digital+datcom

For MSFS use the Roskam methodology is probably superior. I am not entirely sure that Abacus Software had copyright permision to make it freely available, but they did, and it is here.

http://www.flightsimdownloads.com/pub/FlightDynamics.pdf

Remember that 'empty weight' is irrelevant and that the relevant mass is default MSFS mass with zero fuel.

So now there should be no reason to use nonsense from Microsoft to evaluate 'base' MOI.

<<I also understand that Airbus programs all of their fly-by-wire systems to "feel" as close to their other planes as possible, making cross-training easier. So it's all very much a subjective matter,>>

Q-feel is something a computer imposes on a pilot. The inertia of the aeroplane is something else.

Inertia controls, (along with some other variables), how the aeroplane responds to the weather (MSFS weather model) with no crew input at all. How much force is required to deflect a control surface depends on the profile drag (IAS) the pilot is abusing it with, not aeroplane inertia.

<<and the only way to really get it right is with the cooperation of a real-world type-rated pilot that's also intimately familiar with the inner workings - and limitations - of Flight Sim.>>

A large enough sample of real pilot opinions can deliver a bell curve of opinions. If a question about some variable not declared within the real aircraft documention is sensibly formulated, with dozens of sampled opinons, it may be possible to work out a rough answer from the bell curve. Using an estimation technique from a relevant textbook will always produce a more accurate result because the textbook solution is never based on a few opinions.

Pilots have no idea what the force required to deflect a control surface to angle alpha is at a specified level of profile drag (IAS) abuse, and they have no reason to. If they needed to know it would be in the real documentation and we would not need to ask them. An Airbus pilot does not even know how much of his input was rejected by the flight computer.

Inertia, stability and damping in each axis have no relevance to the feel of control inputs. They are about resisting the weather (model) with no control inputs at all. Pilots are not involved. We are discussing evaluation of a mathematical attribute of a distributed mass, in a turbulent fluid, not something that pilots invoke or evaluate.

These are exactly the type of misconceptions which lead to broken FD. Those who hope to encode flight dynamics must study which variables, in which equations, control or modify which inputs and which outputs. I do not agree with everything in the texts cited above, or warrant that they are the most accurate methodologies within the real world, but those who seek a greater understanding of (MSFS) flight dynamics will benefit from studying them.

FSAviator.