Post by volkerboehme on Aug 10, 2008 9:20:22 GMT -5
The FAA runway design pdf documents linked from this thread are just that. They are aimed at runway constructors, not aeroplane operators. They provide guidelines to airport owners concerning the aircraft that may be 'attracted' to use a runway if someone ever funds it and builds it. The data could be used for air file design in the absence of superior data, but anyone contemplating creating a realistic air file will probably have access to similar or superior data from other sources that they acquired to code other parts of the air file.
The real issue is how MSFS works, not how real aeroplanes work.
ON_GROUND simulation in MSFS is simplistic compared to flight simulation. Water is the worst case, but each of the 16 different airfield surfaces coded by BGL authors has different Microsoft errors which the BGL author is stuck with. The FD author can mitigate those errors, but not remove them, precisely because they are BGL variables not an FS constant. The texture applied via a bitmap is of no consequence. There is no friction code in bitmaps. The friction is in the BGL which is why MSFS flight planners must read BGL and report runway surface coded *in the BGL* to those who intend to take off from that runway. They must plan accordingly. Gravel obviously has much higher friction (better retardation) that tarmac. That is especially true when wet, but the runways in MSFS are never wet. Gauge authors could make them wet, via a mouse click if not by superior means, but I don't think any gauge authors have ever bothered.
However none of the surface co-efficients of friction within sim1.dll seems to match either the FAA or ICAO definitions. I have no idea where MS got them from. Perhaps values specific and parochial to the Seattle region?
The air file author can make the distances 'correct' versus sim1.dll for a narrow altitude band and a narrow weight band for just one type of runway surface in MSFS. It will then be wrong for all other cases, or he can make it wrong for all cases but not too bad for any. This is anyway mostly a problem for AI aircraft producers.
Consequently take off and landing distance planning in MSFS is far from accurate all of the time. The trick is to use only runways that are plenty long enough in the current weather. If planning suggests we might need more than 75% of the runway, then that runway should be treated as 'illegal' in the current weather.
As to why a heavy aeroplane L1649A) needs more distance than a lighter one (L1049G) to stop it is the same reason that a truck needs more distance than a car which needs more than a pedal cycle. I would hope that anyone who ever operates a pedal cycle, motor cycle, car or truck already knows that stopping distance varies with mass and the square of velocity else real people are likely to die.
Force = mass * G
Momentum = mass * KTAS * KTAS
At first the force is only reverse thrust. Later it is runway friction and brake (drum/pad) efficiency via cadence wheel braking to preclude skid. Either way deceleration is inversely proportional to mass.
Then we must remember that heavy aeroplanes have higher KIAS = Vref and higher KTAS touchdown velocity to match. At double the velocity (KTAS) the same mass has four times the energy state (momentum). See Propliner Tutorial Part 1 (Basics).
So heavy things take much longer to stop from a higher Vref (or to accelerate to higher Vr). On the ground the lower the air density (higher the altitude) the more KTAS ( = velocity) exceeds KIAS (= profile drag) with no headwind, but the larger the headwind the more KIAS = Vref or Vr (profile drag) exceeds KTAS (velocity).
If we must achieve a profile drag of Vr=90 KIAS to rotate and the headwind is providing 40 KIAS of the necessary profile drag then we only need to accelerate to a velocity of 50 KTAS to achieve profile drag KIAS = Vr = 90 and rotate. The ASI only ever measures our profile drag (KIAS), never our speed (KTS), or our velocity (KTAS).
During acceleration to the profile drag (KIAS) = Vr the force applied is obviously screw thrust instead of reverse screw thrust, but the rolling resistance of the runway surface is deducted from that thrust instead of being added to reverse thrust during landing. So again gravel provides much greater retardation. What was a safety feature of gravel when landing, has become an added risk when taking off. Gravel is used where retardation of the landing is the critical safety requirement. MSFS has many runway surfaces. Plan accordingly.
Most MSFS aircraft are uploaded with impossibly effective ABS brakes which preclude the possibility of skidding after user mishandling of the wheel brakes. My dynamics tend to have more realistic values but do perform your cadence braking for you at a realistic cadence for expert human braking, (they have ABS before it was invented but at weak levels to match human proficiency).
Aircraft with skids can be retarded by digging the skid into soft surfaces with the stick of course, and on hard surfaces speed can be scrubbed off by adding sideforce to the tailwheel or tail skid with repeated left then right tail steering application at low velocity. Many small pioneer and vintage era aircraft have no brakes at all. Their momentum must always be scrubbed off. This is much easier on grass because side force is much higher than on say tarmac and we can dig the skid in better too. MSFS has all of this but the actual friction values and softness values for each surface type are quite 'odd' and the air file author can only correct the MS code errors in one of them.
When a real L1649A lands in ISA conditions nil wind at sea level at max landing weight on a tarmac runway of default ICAO co-efficient of friction (as defined in 1964) and the aircraft is operated in a fully competent way by an airline pilot he can stop it from 114 KIAS at 50 QFE in about 6100 feet. How far it will travel in the hands of different FS9 users is highly variable and has very little to do with what was encoded by the FD author.
The same airline pilot would stop an L1049G in about 5250 feet with fully competent performance. The L1049 and L1649 have few handling criteria in common. As has already been explained in this thread the runway that real pilot may legally approach must be much longer than those fully competent handling values since real airline pilots also fail to deliver fully competent performance every single day. Real runways must be long enough for the world's worst airline pilot having his worst day, not the average airline pilot having an average day.
Real aeroplanes measure user error just like flight simulators and the runway length required is a function of that pilot error, not what the real or virtual flight dynamics author encoded. We only encode a possibility. Real and virtual pilots then have to extract the encoded possibility using their accumulated skill,
Unless explicitly written for AI use MSFS air files do not run a cartoon script. They measure user error and impose the consequence of user error, just like real aeroplanes. Of course the FD author needs to research and impose the 'possible' value, but it will never be replicated during use of MSFS as weather, user performance, runway surface type, altitude, and weight vary. Therefore runway length required depends mostly on pilot skill, then on the weather, then on runway surface type, and then on aircraft weight. Which is why handling notes never tell you how much runway you will require. The answer is far too variable and depends mostly on each users skill.
Not our skill to minimise the distances unrealistically by disregarding safety while rotating before Vr or slowing below Vref before reaching 50 QFE, but on our skill to operate the aircraft realistically to the real numbers. The handling notes must specify maximum landing weight, provide the real KIAS target numbers, the mandatory flap configuration and the typical cowl configuration (for average weather). For take off they must specify the elevator trim required for safe take off, The rest is up to us using our accumulated skill to achieve the real targets.
Like real airline pilots we will never achieve perfection. Like airline pilots we must learn to allow for our errors and we must preplan a remedy for the errors we will commit. We must never attempt take off from, or landing on, a marginal runway length in the current weather.
Remember data in real manuals will be useless in MSFS because runway friction assumptions differ.
FSAviator
The real issue is how MSFS works, not how real aeroplanes work.
ON_GROUND simulation in MSFS is simplistic compared to flight simulation. Water is the worst case, but each of the 16 different airfield surfaces coded by BGL authors has different Microsoft errors which the BGL author is stuck with. The FD author can mitigate those errors, but not remove them, precisely because they are BGL variables not an FS constant. The texture applied via a bitmap is of no consequence. There is no friction code in bitmaps. The friction is in the BGL which is why MSFS flight planners must read BGL and report runway surface coded *in the BGL* to those who intend to take off from that runway. They must plan accordingly. Gravel obviously has much higher friction (better retardation) that tarmac. That is especially true when wet, but the runways in MSFS are never wet. Gauge authors could make them wet, via a mouse click if not by superior means, but I don't think any gauge authors have ever bothered.
However none of the surface co-efficients of friction within sim1.dll seems to match either the FAA or ICAO definitions. I have no idea where MS got them from. Perhaps values specific and parochial to the Seattle region?
The air file author can make the distances 'correct' versus sim1.dll for a narrow altitude band and a narrow weight band for just one type of runway surface in MSFS. It will then be wrong for all other cases, or he can make it wrong for all cases but not too bad for any. This is anyway mostly a problem for AI aircraft producers.
Consequently take off and landing distance planning in MSFS is far from accurate all of the time. The trick is to use only runways that are plenty long enough in the current weather. If planning suggests we might need more than 75% of the runway, then that runway should be treated as 'illegal' in the current weather.
As to why a heavy aeroplane L1649A) needs more distance than a lighter one (L1049G) to stop it is the same reason that a truck needs more distance than a car which needs more than a pedal cycle. I would hope that anyone who ever operates a pedal cycle, motor cycle, car or truck already knows that stopping distance varies with mass and the square of velocity else real people are likely to die.
Force = mass * G
Momentum = mass * KTAS * KTAS
At first the force is only reverse thrust. Later it is runway friction and brake (drum/pad) efficiency via cadence wheel braking to preclude skid. Either way deceleration is inversely proportional to mass.
Then we must remember that heavy aeroplanes have higher KIAS = Vref and higher KTAS touchdown velocity to match. At double the velocity (KTAS) the same mass has four times the energy state (momentum). See Propliner Tutorial Part 1 (Basics).
So heavy things take much longer to stop from a higher Vref (or to accelerate to higher Vr). On the ground the lower the air density (higher the altitude) the more KTAS ( = velocity) exceeds KIAS (= profile drag) with no headwind, but the larger the headwind the more KIAS = Vref or Vr (profile drag) exceeds KTAS (velocity).
If we must achieve a profile drag of Vr=90 KIAS to rotate and the headwind is providing 40 KIAS of the necessary profile drag then we only need to accelerate to a velocity of 50 KTAS to achieve profile drag KIAS = Vr = 90 and rotate. The ASI only ever measures our profile drag (KIAS), never our speed (KTS), or our velocity (KTAS).
During acceleration to the profile drag (KIAS) = Vr the force applied is obviously screw thrust instead of reverse screw thrust, but the rolling resistance of the runway surface is deducted from that thrust instead of being added to reverse thrust during landing. So again gravel provides much greater retardation. What was a safety feature of gravel when landing, has become an added risk when taking off. Gravel is used where retardation of the landing is the critical safety requirement. MSFS has many runway surfaces. Plan accordingly.
Most MSFS aircraft are uploaded with impossibly effective ABS brakes which preclude the possibility of skidding after user mishandling of the wheel brakes. My dynamics tend to have more realistic values but do perform your cadence braking for you at a realistic cadence for expert human braking, (they have ABS before it was invented but at weak levels to match human proficiency).
Aircraft with skids can be retarded by digging the skid into soft surfaces with the stick of course, and on hard surfaces speed can be scrubbed off by adding sideforce to the tailwheel or tail skid with repeated left then right tail steering application at low velocity. Many small pioneer and vintage era aircraft have no brakes at all. Their momentum must always be scrubbed off. This is much easier on grass because side force is much higher than on say tarmac and we can dig the skid in better too. MSFS has all of this but the actual friction values and softness values for each surface type are quite 'odd' and the air file author can only correct the MS code errors in one of them.
When a real L1649A lands in ISA conditions nil wind at sea level at max landing weight on a tarmac runway of default ICAO co-efficient of friction (as defined in 1964) and the aircraft is operated in a fully competent way by an airline pilot he can stop it from 114 KIAS at 50 QFE in about 6100 feet. How far it will travel in the hands of different FS9 users is highly variable and has very little to do with what was encoded by the FD author.
The same airline pilot would stop an L1049G in about 5250 feet with fully competent performance. The L1049 and L1649 have few handling criteria in common. As has already been explained in this thread the runway that real pilot may legally approach must be much longer than those fully competent handling values since real airline pilots also fail to deliver fully competent performance every single day. Real runways must be long enough for the world's worst airline pilot having his worst day, not the average airline pilot having an average day.
Real aeroplanes measure user error just like flight simulators and the runway length required is a function of that pilot error, not what the real or virtual flight dynamics author encoded. We only encode a possibility. Real and virtual pilots then have to extract the encoded possibility using their accumulated skill,
Unless explicitly written for AI use MSFS air files do not run a cartoon script. They measure user error and impose the consequence of user error, just like real aeroplanes. Of course the FD author needs to research and impose the 'possible' value, but it will never be replicated during use of MSFS as weather, user performance, runway surface type, altitude, and weight vary. Therefore runway length required depends mostly on pilot skill, then on the weather, then on runway surface type, and then on aircraft weight. Which is why handling notes never tell you how much runway you will require. The answer is far too variable and depends mostly on each users skill.
Not our skill to minimise the distances unrealistically by disregarding safety while rotating before Vr or slowing below Vref before reaching 50 QFE, but on our skill to operate the aircraft realistically to the real numbers. The handling notes must specify maximum landing weight, provide the real KIAS target numbers, the mandatory flap configuration and the typical cowl configuration (for average weather). For take off they must specify the elevator trim required for safe take off, The rest is up to us using our accumulated skill to achieve the real targets.
Like real airline pilots we will never achieve perfection. Like airline pilots we must learn to allow for our errors and we must preplan a remedy for the errors we will commit. We must never attempt take off from, or landing on, a marginal runway length in the current weather.
Remember data in real manuals will be useless in MSFS because runway friction assumptions differ.
FSAviator