X-Plane 11 Kills Ground Wind Speed

[dts1 3]

Finally gave up on X-Plane 11. Its erratic taildragger ground handling and cutting of the wind speed
near the ground, as a work around for its inability to simulate taildragger handling, has finally fatigued my interest. It took a while to finally grasp why my RF5 Sportavia and other low speed aircraft dive into the ground on approach when flying in a wind. As the plane is flared and slowed within 10 feet of the ground, X-Plane 11 unexpectedly kills the wind and the RF5 stalls-dives irrespective of ground lift effect. This is the culmination of the simulators ground failures. 
 

[mSparks 2,086]

Just now, dts1 said:

X-Plane 11 unexpectedly kills the wind

not unexpectedly at all. Documented, well researched, and true to life.

ground effect is probably still a little off (and assuming you are using experimental flight model, and a tail dragger really tuned to 11.40 and not just pretending to be, prior to that tail draggers had numerous hacks that dont play nicely with the improved flight model)

[dts1 3]

On 6/17/2020 at 9:46 AM, mSparks said:

not unexpectedly at all. Documented, well researched, and true to life.

ground effect is probably still a little off (and assuming you are using experimental flight model, and a tail dragger really tuned to 11.40 and not just pretending to be, prior to that tail draggers had numerous hacks that dont play nicely with the improved flight model)

 

[dts1 3]

"not unexpectedly at all. Documented, well researched, and true to life. "

Yes, totally unexpected as mistakenly done in X-Plane 11. As this X-plane test shows, on a steady glide to the 952' elevation runway in a steady 15.13 knot wind, the bottom drops out at about the 962' level as the wind is abruptly linearly killed down to 6.82 knots on the runway - a drop of 8.31 knots. See bottom blue curve.

Correct reduction of windspeed is not abrupt, but smooth and exponential starting higher, typically, from the 30' level. At the 962' level = 15.13 knots, the curve continues translation to 11.2 knots at the runway - a drop of only 4 knots in 10'. See top red curve.

 

https://www2.jpl.nasa.gov/adv_tech/photovol/2016CTR/SNL%20-%20Est%20Wind%20Speed%20vs%20Height_1985.pdf

 

[Murmur 3,941]

Since you gathered accurate data, you could try emailing directly Austin about the issue. That kind of hard data is exactly what he looks for when someone reports issues. You never know, the issue could then be resolved in the next release.

[mSparks 2,086]

1 hour ago, dts1 said:

Correct reduction of windspeed is not abrupt, but smooth and exponentia

not in any aircraft Ive ever flown or flown in.

and if you watch the video I linked you will see the tests he did with the wind turbine guys, its the funnel on the left hand side of the whiteboard, what you have shown in that graph looks to me to be more or less exactly what he described came from their testing.

always 0kt windspeed at 0ft agl rising by a fairly simple formula to the windspeed measured 10 meters above the ground given in a metar.

ignore it at your own peril

 

Edited June 18, 2020 by mSparks

[tutmeister 695]

Austin googled it and didn’t do any research with ‘wind turbine guys’ but this phenomena has been common knowledge for engineers for many years as referenced by the above nasa paper which quotes the mechanical engineers handbook which was published in 1951!

I remember watching this video many months ago when it came out, thinking great, he’s finally addressing the cross wind issue. However, the above test shows that even though Austin recognises what should be happening, the sampled data in X-Plane is not following the curve and is stepping down too steeply. It is almost like X-Plane does not have enough fine points to allow a smooth wind gradient. It would be interesting to know how dts1 captured the data from in X-Plane as I would like to replicate it.
Austin tends to look at the maths and say, the maths checks out so it must be fine. It usually takes months of people presenting behavioural evidence before he will look at it. The ground friction issue is a good example of this and he even says it himself in the above video. He is definitely doing some good work but I suspect it is not quite refined enough to match real life yet.

You are right that aircraft devs need to ensure their model does not contain any physics hacks that were needed in the past though.

That draco crash was nothing to do with ignoring this issue, it was simply flying outside of minimus in a very gusty day. Mike Patty explained exactly why it happened in that and other videos he released. He was following criteria that he would not normally do and not taking off into wind.

[mSparks 2,086]

5 hours ago, tutmeister said:

is not following the curve

The "curve" shows someone expecting wind to be 11kts at 0ft agl.

It absolutely should be 0kts. So if anything only falling to 7 kts isnt fast enough - but is more likely to be a factor of where the actual measurement is being taken from.

But mostly 4kts difference between what someone is expecting and measuring when wind is 15kt is very light turbulence, certainly not enough to complain its causing it to be unflyable.

[mSparks 2,086]

5 hours ago, tutmeister said:

That draco crash was nothing to do with ignoring this issue,

It could have been a gust, it could have been the wind was just much higher a few feet agl than at ground level.

He ignored the correct procedure for flying in wind that always increases rapidly as altitude increases and nearly paid for it with his and his families lives, he doesnt have an Austin to complain about the world not simulating the wind properly.

Edited June 19, 2020 by mSparks

[soaring_penguin 428]

I agree with mSparks here, I have been a hang glider pilot much of my life, and in such tiny craft you really feel the air. And yes, the wind near ground level can fall away very abruptly, especially on a gusty or hot day with many thermals. Many have wacked in coming in too slow on a sunny afternoon.

Edited June 19, 2020 by soaring_penguin

[dts1 3]

In response to  tutmeister 's request for method.

I'm under no illusion of this abhorrent gradient implementation as an X-plane good faith effort to enhance weather veracity, but rather an aha! moment recognition of a way to both mitigate and cover for their failed taildragger crosswind handling. Drastically abruptly reducing runway wind speed independent of the physics involved serves their purpose - not starting from 30' using a proper exponential curve.
This won't change.
There is NO incentive to accurately portray wind gradients and re-reveal their handling failure.
BUT. they could give users the chance to opt out of the scheme.

To test X-plane runway wind gradients:
1) set up manual weather CAVOK
2) set wind at 45 degrees to the runway at a constant 16 knots with all turbulence and other variables at 0
3) In the DATA OUTPUT heading, choose (1) "Weather" and (2) "Latitude, Longitude, and Altitude".
4) Choose a low or mid wing glider.
5) set it back about 1/2 mile from and about a 300' altitude above the runway.
6) setup a constant very shallow glide into the runway at about 55 knots.  
7) afterwards, rerun using [ALT] [R]. Copy elevation versus wind speed data to see the relationship. Confirm your runway is flat and not affecting data.

[mSparks 2,086]

27 minutes ago, dts1 said:

BUT. they could give users the chance to opt out of the scheme.

I repeat, to spec windspeed at 0 ft above ground level is always 0kts speed in the real world and xplane 11.41.

What scheme do you want to opt out of?

If you want actual readings out you dont need to do any of that.

go to settings, the data out page, and have it save the data you want to disk as you fly. (dont forget to turn it off afterwards - if you do you will suffer the perf impact until you remember again)

replay does not give you accurate data - As I am sure you have seen from the instrument jumps and other artifacts when in replay.

This isnt to say there are not other factors at play that are off - but this really doesnt seem likely to be it, ive done a fair bit of testing on this for a lua based autoland of the 747, I havent seen anything unexpected in the values its using

 

Edited June 19, 2020 by mSparks