Minnesota Crash

[jlunseth]

I went back and checked Scott's post. It is on page 10 of this thread. He says: "... there was a RAOB taken pretty close to the time and location of the site of the crash.  This shows the bases of this stratocumulus deck were ~1200 ft AGL and tops were ~2700 ft MSL; so the clouds were less 1,000 feet deep.  The sky was overcast which is also seen on the IR satellite imagery.  So an instrument approach would be necessary given the weather at FCM."

FCM is at 906 so 1200 ft AGL is 2100 MSL and the tops were 2700 ft MSL, thus the deck was 600' thick. Now, that was at FCM, not at Victoria, but they are only about 15 miles apart, or maybe a little less. The NTSB rep says in his statement that the wx at FCM at the time of the accident was 1100 OVC, so that would be a 700 foot deck, assuming tops at 2700. Not significantly different from Scott's report. The NTSB rep did not state tops.

[kortopates]

1 hour ago, 1980Mooney said:

 

OK - Scott says the data shows that the clouds were uniformly capped at 2,800 MSL.  That means that he was on top and in the clear at 17:38:48 and 3,300 ft. MSL when he slowed to 77 kts and started deviating from the ILS localizer with a turn to the NE.  And he was in the clear as he began his turn to the South over Lake Auburn at 17:39:13 - 19.  He dipped into the clouds briefly over Victoria while heading south at 17:39:19 - 30.   But then he popped up on top into the clear for 5 seconds.  During those 5 seconds in the clear he turned left nearly 70 degrees in 8 seconds - that is a pretty steep turn rate  of 360 degrees in about 40 seconds.  Over Church Lake at about 17:39:39 he went into the cloud layer starting his rapid descent.  He also tightened the left turn rate - turning about 80 degrees to the left in 5 seconds - that is a steep turn rate of 360 degrees in 23 seconds.  That would be about a 60 degree banked turn.  So he was in a 60 degree left banking turn while descending at an average of 9,500 fpm for the next 12 seconds (peak descent rates higher if you believe the data points in between).  The 60 degree turn put 2 G's on the wings.  G-forces while pulling out of dive unknown.  Also don't know if he was in the turn while trying to pull out of the dive or if it was all after he was straightened out nearly northbound towards Victoria.

The steep turn and descent must have been harrowing for all on board but once in the clouds they may have been disoriented and unaware until popping out the base.

Very good and illustrates he was having difficulties above the clouds but as you well point out, his difficulties began within 500' of the estimated cloud tops. I don't know how many feet plus or minus we can say the 2800 msl cloud tops are accurate too. But data is overwhelming for a thin layer but may have been a few hundred feet higher. Very interesting though if the pilot was already having problems deviating lateraly and verically before dropping into the clouds. Makes me really wonder where the actual clouds tops were. Hopefully the preliminary might be able to tell us from pilot reports from someone that flew the approach just before or after - but usually prelim's are pretty limited.  But from what you illustrate based on 2800 tops, perhaps the cloud tops were actually a bit higher - if the pilot was having difficulty well before dipping below 2800'. Afterall its hard to fathom a instrument pilot who would continue a descent into IMC when already deviating both laterally and vertically in VMC conditions.

Good points about the turn, but this was in an unusal attitude recovery mode i.e. nose pointing steeply down and therefore not loading the wing like a level unacelerated 60 degree bank turn loads the wing to 2 G's.  With the nose pointed down like that the wing should be unloaded, at least initially, till the pilot started pulling on the yoke to correct the descent - which is what we expect led to overstressing the airframe with too much airspeed. 

But FWIW, I also entirely agree with your points very early on in this thread, assuming I understood you correctly, that what ever caused the pilot to start deviating early on (loss of control) before the final big descent is the root cause of this accident. How the wing and elevator got overstressed at the end is of little consequence.  

Edited August 12, 2021 by kortopates

[Scott Dennstaedt, PhD]

3 hours ago, jlunseth said:

The NTSB rep did not state tops.

I can guarantee you that in the NTSB final report, the investigators will use this RAOB when defining the weather at the time of the accident. Heck, they sometimes will use RAOBs that are 100 miles away and many hours prior (or after).  So this will be the investigator's dream come true with respect to getting accurate data close to the time of the accident.  But it's still funny to read on a few of those "other" aviation boards pilots trying to suggest that this was due to some kind of microburst event or that the layer of clouds was several thousand feet deep.  Ugh!

[Scott Dennstaedt, PhD]

2 hours ago, 1980Mooney said:

OK - Scott says the data shows that the clouds were uniformly capped at 2,800 MSL.

 

Yes, statocu will have a lumpy appearance (quilt-like) on top. Essentially if the air below has enough momentum (it was pretty unstable below the cloud base), it might overshoot (momentum) a little bit which creates that lumpy look (think about a pot of boiling water).  So it's possible that you may see some areas a few hundred feet deeper.  Pretty negligible though.  

[jlunseth]

@Scott Dennstaedt, PhD I have a question while you are here. I watched one of your videos on  Skew T Log P within the last month. I had always thought that the cloud tops would be where the temp and dew point lines separate. However, in the video you were describing how, at least in some circumstances, the lines could stay pinched together within a few degrees after they separate, and clouds would be possible in that area. I went back and looked at the RAOB you posted and saw that the lines did do that up to around 5,000, although as you said in your post at the time, I believe, there was a temp inversion that would cap the clouds where the lines separated. I do see in the RAOB that the temp line takes a small but abrupt jog to the right where the lines separate, meaning the temp increases a few degrees. Is there are chance that cloud tops could have gone further up the graph to where the lines started to divert further at around 5k, or is that small temp increase enough to cap the clouds?

[jlunseth]

For those looking for the ATC tapes, I found both the Minnie Approach tape and the FCM tower tape on LiveATC.net. To listen to the Minnie Apch tape, click on the archive link, chose Aug. 7, chose 2230Z and in the line that displays feeds, scroll way down until you find KMSP Approach 121.2 . As I told someone in a PM, that is the alternate approach frequency to 134.7. You will hear the communications with 9156Z starting at about 28:29 on the tape.

Unfortunately, it does not appear that we can get anything other than very approximate times from the tapes - off by minutes. The approach tape is stated to be from 2230-2300, for example, but the recording starts at 32:04 and counts down from there, so it is longer than 30 minutes. It is impossible to know if it starts at 2230 on the button or a couple of minutes later. The same with the tower tape, the times are only very approximate.

The content is about what would be expected and confirms what the NTSB rep said in his interview. At 28:29 on the tape, which is roughly 17:32 Z, he clears 9156Z direct to ZAMUD to join the final approach course. At 27:45, roughly 1:45 later he says 56Z descend and maintain 3000, cleared the 10 Right ILS approach. About 4:15 later, at 23:28 on the tape, he instructs 56Z to go over to tower 119.15 . None of the pilots responses are on the archive.

The tower tape then starts and has the same kinds of time problems. The animated recording very early in this thread appears fairly accurate, although missing some tower communications.  At 23:16 on the tape, which is roughly 36:44Z, the pilot checks in: “Flying Cloud Tower, Mooney 9156 Zulu with you. The tower responds with a clearance to land on 10R. At roughly 22:37:10 tower says Mooney 9156Z 10R cleared to land.” Pilot does not respond. Tower calls 9156Z 15 seconds later and the pilot says, Yeah go ahead. Tower again clears him to land on 10R and says altitude indicates 2700, pilot responds “Roger.” There are at least two more tower calls after that with no response.

Listening to the tower tape, I have to say that the pilot did not sound distracted, his responses were pretty much immediate except for one call and he did not sound confused. He sound like a pilot who did not understand he needed to repeat clearances as given and that is all.

Let me repeat, these times are off by minutes. The hand off from approach on the approach tape appears to be 17:38:36Z but the pilot’s first contact with tower is 17:36:44 on tower tape, which is two minutes before the handoff on the approach tape, so the times are only a very rough guide.

We know that he was cleared direct ZAMUD, told to descend and maintain 3000, cleared the ILS 10R into Flying Cloud, handed to tower, cleared to land by tower, and tower noted his altitude at 2700 at one point. All these match with the ADSB track into the S turns, except the times are just very rough. If anyone has a better way of establishing times of the calls, have at it.

[Scott Dennstaedt, PhD]

10 hours ago, jlunseth said:

I watched one of your videos on  Skew T Log P within the last month. I had always thought that the cloud tops would be where the temp and dew point lines separate. However, in the video you were describing how, at least in some circumstances, the lines could stay pinched together within a few degrees after they separate, and clouds would be possible in that area. I went back and looked at the RAOB you posted and saw that the lines did do that up to around 5,000, although as you said in your post at the time, I believe, there was a temp inversion that would cap the clouds where the lines separated. I do see in the RAOB that the temp line takes a small but abrupt jog to the right where the lines separate, meaning the temp increases a few degrees. Is there are chance that cloud tops could have gone further up the graph to where the lines started to divert further at around 5k, or is that small temp increase enough to cap the clouds?

This depends on the data you are plotting (e.g., RAOB vs Model), temperature of the environment (cold vs warm stratus) and type of cloud type (cumuliform vs stratiform).  What I posted for this accident was a RAOB, so it will very accurately depict the bases and the tops of the clouds since the sensor measures 100% RH as it ascends through the cloud.  It is quite common, however, in a stratocumulus cloud deck for there to be fairly moist conditions above the deck, but still have a very highly capped scenario where the clouds do not have vertical growth.  Here's an example. Notice the solid deck below and the rather dispersed clouds just above. In this case, the air above the stratocu deck was likely very moist nearing saturation. 

Nevertheless, it's pretty clear in the ROAB in my blog post that the tops of the primary cloud deck were defined reasonably well around 2,800 feet MSL. Could there have been a very dispersed area above?  Possibly, but not very likely.   

When looking at model forecasts on the Skew-T (called a forecast sounding), you can have very similar profiles for stratiform clouds, but cumuliform clouds are a different beast altogether. The Skew-T only predicts the environment for which the can grow, not the clouds themselves. So you can have a broken area of fair weather cumulus several thousand feet deep or even towering cumulus and they won't likely show up on a forecast sounding like you might think.  That's because they are created through ascending air. For that, you need to lift a parcel of air to determine their extent.      

[jlunseth]

Great lesson in Skew T Scott, thank you. 

[Shadrach]

On 8/12/2021 at 10:26 AM, jlunseth said:

I feel that flying an approach on autopilot actually is safer. However, I have had enough quirky things happen with the GPS + GPSS + analog AP system to know the importance of constantly checking that what you think you have commanded is what the system is actually doing. The biggest challenge is when something happens that requires flying by hand on a moment’s notice, or resetting a procedure on the GPS. So I practice both. A major issue, I think, is that these are not cohesive systems, they are separate components made by separate manufacturers and added in over time. No one, to my knowledge, has written a manual that says how the overall system is going to work. The pilot needs to intimately understand the quirks of each of them.

That may be true for each individual approach. What do you think the cumulative effects are on one’s instrument skills over time? There is no question that an auto pilot can fly a more precise approach than a human.  My sense is that this would be especially true for humans that almost  never hand fly approaches.  

[jlunseth]

To answer your question, I personally am concerned about the cumulative effects to the point where almost all of my weekend, keep-in-practice flying consists of flying approaches in VMC. I fly them about 30% on the AP and about 70% by hand. All kinds, ILS, RNAV, VOR, I used to even fly NDBs but got rid of the ADF about a year ago, there aren’t any NDB’s around here anymore. The first thing I do when the plane comes out of annual during the winter is spend about a month practicing like that, and getting myself approach hardened again. But the AP is still safer.

[A64Pilot]

19 minutes ago, 1980Mooney said:

The way things are going in the near future pilots won’t be using their instrument skills. Everything will be automated. 

Garmin AutoLand.  Garmin Smart Glide  

I suspect they have more in the pipeline. 
 
As automation becomes more capable more reliable and more available pilots will naturally rely on it more and more. 

At least for what I have left time wise on this world, it’s unlikely that Joe Average GA pilot will be able to afford those bells and whistles, and unlikely that the legacy aircraft will be able to be fitted with them, heck Joe Average GA pilot may not even be instrument rated.

Being a old guy I guess I don’t trust an Autopilot to fly an approach, enroute stuff, sure then you have altitude and altitude is time to correct for any malfunction it may have, I’ve had too many glitches with electronics, and it seems the more modern stuff glitches more often than the old stuff, I believe younger generations just don’t expect their electronics to always work. But then my go to approach is an ILS.

Look at some of the threads on this forum about avionics glitches.

 

I remember reading that the General public when automatic elevators were first invented wouldn’t ride in one, that stores that had them put an employee in them so that the public would feel safe.

Seems silly now, but as people like me die and go away there will be of course more and more trust in automation, there has been for years in the Commercial fleet, and I’m sure that it will be more reliable and safer than a human, but I still don’t trust it

 

[ilovecornfields]

I think most modern autopilots fly approaches better than most modern pilots…

[Hank]

38 minutes ago, A64Pilot said:

But then my go to approach is an ILS.

They are nice, but also relatively rare. Maybe available at 10% of US airports? Of the half-dozen non-towered places around here, only one has an ILS. It's also the only ILS on my frequently-visited-airports list.

[Guest]

On 8/9/2021 at 12:40 PM, 1980Mooney said:

He was Instrument Rated.

So much for privacy.

Clarence

[A64Pilot]

5 hours ago, ilovecornfields said:

I think most modern autopilots fly approaches better than most modern pilots…

Or old pilots, be real tough to beat a computer,  I’d even bet on a VFR cross country flight they will get you there faster and burn less fuel than hand flying.

‘Until they don’t, my concern isn’t when they are working, it’s when they don’t.

Truth is I don’t fly IFR anymore, I was very good at it, you learn to flying a helicopter IFR, but I’m not current now, but that’s easy to get back, what I’m not is proficient now and I don’t think I would fly hard IFR enough to maintain proficiency, so it’s likely I’ll not pick it back up.

‘Back when I did fly IFR frequently I hand flew it all. it kept me proficient. Now it’s not like I was flying it everyday, I wasn’t, usually not many good IFR days in Summer so I may go a long while between.

[A64Pilot]

15 minutes ago, M20Doc said:

So much for privacy.

Clarence

Yeah,I’ve had a heartache about that forever, your tail number gives out your home address etc.

Personal data from an A&P license number, which has to be on any logbook entry an A&P makes is just as easy to find.

So why is that?

[Guest]

1 minute ago, A64Pilot said:

Yeah,I’ve had a heartache about that forever, your tail number gives out your home address etc.

Personal data from an A&P license number, which has to be on any logbook entry an A&P makes is just as easy to find.

So why is that?

I find it odd that the FAA makes this information public.  Transport Canada doesn’t.

Clarence

[jaylw314]

3 minutes ago, M20Doc said:

I find it odd that the FAA makes this information public.  Transport Canada doesn’t.

Clarence

Yes, I don't know how hard it is to keep your home info private on the Airmen Registry.  I had a pretty slam dunk reason so they gave me an exception, but I don't know how many people they've turned down.

[jaylw314]

15 minutes ago, A64Pilot said:

Or old pilots, be real tough to beat a computer,  I’d even bet on a VFR cross country flight they will get you there faster and burn less fuel than hand flying.

‘Until they don’t, my concern isn’t when they are working, it’s when they don’t.

Truth is I don’t fly IFR anymore, I was very good at it, you learn to flying a helicopter IFR, but I’m not current now, but that’s easy to get back, what I’m not is proficient now and I don’t think I would fly hard IFR enough to maintain proficiency, so it’s likely I’ll not pick it back up.

‘Back when I did fly IFR frequently I hand flew it all. it kept me proficient. Now it’s not like I was flying it everyday, I wasn’t, usually not many good IFR days in Summer so I may go a long while between.

I recall reading a NASA study suggesting we should be hand flying about a third of the time, and using automation two-thirds of the time.  Like many NASA studies I've read, though I haven't been able to find it again

[tigers2007]

Yes, I don't know how hard it is to keep your home info private on the Airmen Registry.  I had a pretty slam dunk reason so they gave me an exception, but I don't know how many people they've turned down.

Mine has been private for years. All I asked them was to make it private. I didn’t give any sort of reason. On the other hand my aircraft registration is not. Nor is FCC address data.

[Mooney Dog]

1 hour ago, M20Doc said:

I find it odd that the FAA makes this information public.  Transport Canada doesn’t.

Clarence

 

1 hour ago, jaylw314 said:

Yes, I don't know how hard it is to keep your home info private on the Airmen Registry.  I had a pretty slam dunk reason so they gave me an exception, but I don't know how many people they've turned down.

It is an opt in option to hide your home address on the airman's registry. For new pilots they even ask during sign up on the computer. Mine was a letter saying "if we dont hear from you in 60 days, you address will be public online." Opt out of that real quick.

 

...now if only my N number could be the same. 

[A64Pilot]

I formed an LLC when I bought the Mooney to keep my info private, ADSB makes it even worse.

‘Not that I’m engaged in any illegal activities, I just don’t think my movements and location ought to be public knowledge. 

If someone really wanted to they can still find me, but it takes a little more than just Googling my N number.

 

[Will.iam]

10 hours ago, ilovecornfields said:

I think most modern autopilots fly approaches better than most modern pilots…

You have to define modern. Our A320 routinely cross star arrivals 200 ft high and 20 kts fast so much so the controller doesn’t  even query us about it and it’s an accepted norm. Gusty crosswinds for autoland forget it but I don’t complain though as that is job security. But the computer already has us beat on 0/0 vis landings. 

[jaylw314]

4 hours ago, tigers2007 said:

Mine has been private for years. All I asked them was to make it private. I didn’t give any sort of reason. On the other hand my aircraft registration is not. Nor is FCC address data.

Yes, I use a PO box for my aircraft registration.  I didn't realize the Airman Registration info could be kept private just by opting out, I thought I remembered them asking for a letter explaining why initially

[toto]

Looks like you can change your address “releasability status” online here..

https://amsrvs.registry.faa.gov/amsrvs/Logon.asp