Wednesday, December 28, 2011

This continues from my last two (non-reindeer) posts, so read down to catch up.

I got a whole whack of information in this post from AvCanada, specifically the accident speculation thread. Stuff in quotes is taken verbatim from the discussion thread.

This is the ILS approach chart for Runway 35T into Resolute Bay. Do NOT use this to navigate, it's out-of-date now.

This is what the ILS into Resolute Bay looks like. The ceiling in this approach was 1,300':

On August 20th 2011, First Air flight 6560 was a 737 travelling from Yellowknife to Resolute Bay with 15 people on board, including four crew members. They called about 4 miles final for the runway, and then crashed a few minutes later. The aircraft appears to have been under control, and the surviving passengers reported nothing unusual right up to the point of impact.

The accident report hasn't come out yet, and the following stuff is entirely speculation - I want to make that clear. But I'm going to speculate.

METAR CYRB 202000Z 18009KT 8SM VCFG SCT003 OVC005 07/07 A2986 RMK

The weather at the time of the accident was foggy with low (300') scattered and (500') overcast layers. People on the ground couldn't see much at all, but flight visibility can be very different from ground visibility.

We do know some things: The crew was still in control on three mile final, the engines were running, the aircraft is oriented parallel to the runway and level with the horizon, and the wreckage is strewn across a large area. If it had stalled or a mechanical failure had caused it to rapidly descend it would leave a much smaller impact area. That has all the hallmarks of classic CFIT, or 'controlled flight into terrain'.

Here's a Google earth view of the area around the accident. The wreckage trail is represented by the yellow line. The gully leading down from the nose of the aircraft at the end of the wreckage appears to contain the same washed out sort of area in google and the photo. The two red circles outline similar geometric forms. One is the approach end of 35T. The other is a road structure on a slightly sloping flat area elevated above the airport by about 325 feet and one mile to the right of the centerline of the runway. The lake to the southeast of the wreckage trail could appear in certain cloud arrangements to be similar to the shoreline on the approach to 35T.

The aircraft would have been between two layers of clouds with approximately 200 feet between layers, with the bottom scattered layer at 475 ASL and the top broken layer at 675 ASL. Some of what was a scattered layer from the weather observation point may have been broken in the hill above.

Now here's the thing: Even with the local terrain being a bit confusing, there should have been no problem finding the runway and landing with any normal kind of ILS approach, with a 500 foot ceiling. ILS tolerances are pretty tight, and it would be really hard to conceive why the aircraft would have been a mile to the right of its course, and more or less on runway heading without any abnormal indications to the crew.

Let's add some more information: here's another picture of the accident site, this time with the Resolute VOR location plotted. This one chills my bones.

It turns out the ILS was functional on that day, but suppose the crew didn't get a glideslope indication and decided to fly the Localizer only approach. That takes them down to 540' ASL. The ridge they hit was 653' ASL.

The step down approach (localizer) really comes into the equation if the aircraft is mistakenly inbound on the 167T radial from YRB. In that scenario, the crew believes it is tuned to the ILS, can't get glideslope, and switches to a localizer approach with the final drop only 160 feet four miles back. They keep tracking the 167T radial as if it were the localizer. They don't have glideslope because they're tuned to 112.1 instead of 110.3. Their DME is coming off the VOR.

"In the mistuned VOR scenario, they have to have POKAN on the 167 radial at 4 DME, a mile east of proper track. Then do the procedure turn and fly back inbound on the false localizer. Maybe they left the VOR eastbound with 167 already set on the OBS, flew it needle centered to the false POKAN, did the PT and re-intercepted inbound. Get no GS, maybe call it in U/S, and fly the localizer approach instead. That scenario ends exactly where the accident happened.

I can't see a late tuning of the VOR and nobody noticing that the needle moved when the OBS was turned. If they'd been tracking 167 outbound to Pokan, and then set up the inbound course on the CDI, they wouldn't have noticed it so much as they would have been in the PT where they would expect it to be deflected."

An added factor now: Has anyone noticed that there is no missed approach point for the LOC/DME approach on the ILS/DME 35T plate? Not on the DND copies anyway. Take a look.

A second factor: Notice how the VOR isn't even depicted on the ILS chart? Why on earth would the crew have it tuned in? Well, a couple of possibilities exist. There was a temporary military control tower at Resolute that day, coordinating aircraft that were participating in a mock search-and-rescue exercise. The control tower was asking other aircraft for their radial and distance to the airport, and the accident aircraft had reported that information to the control tower fairly late in their approach. In order to report that information they would have had to tune in the VOR. The other possibility is that maybe they were using the VOR to navigate to the airport before commencing the approach.

"Another Canadian carrier used to have an unofficial procedure on the 737-200. If you wanted to retain the DME display while doing an ILS, you would tune #1 to the ILS and #2 to the VOR. Then you would transfer the display [overhead switch] to "both on 1" Now, both pilots would have their HSI displaying the info from the #1 radio, and the dme would readout from the vor still tuned on the #2 radio. This practice was banned after a crew mistakenly switched "both on 2" during an approach to Prince George. Thinking they were tracking the LOC, the aircraft descended towards the YXS VOR and very nearly had an accident. [With both radios tuned to the ILS you would not have a DME readout and there is no DME hold switch on the 200] The practice of transferring the display was then banned and to be used only in case of radio failure. Keep in mind that was another air carrier, not First Air."

Here's a pic of the panel from the actual accident aircraft.

Hmm, same switches.

But wouldn't their GPWS (Ground Proximity Warning System) have saved them? The 737 in question was equipped with an older model, which basically gives no warnings once the landing gear is down. GPWS will only give you two calls "500'" and "Sink Rate", whereas the newer Enhanced GPWS will give 1000', 500' 100', 50', 40', 30', 20', and 10' above ground calls. This late in the approach, the gear would have been down.

So here's my speculation: With the knowledge that the military was asking for radials/DME bearings prior to the crash, maybe the VOR was tuned in, thinking they had the ILS frequency up. Now to those of you who fly IFR, how often have you made a late change in the approach? It happens to me once in a while. If you thought you had the ILS tuned in and once on the approach you had a G/S flag, it would be a fast and easy brief to re-brief for the LOC only approach and continue to the higher minumums (especially in an environment such as the Arctic airports)...if that was the case here, the LOC only minimums still would have put them into a hill if they were tracking the VOR.

One final pic:

This was taken on the LOC DME BC approach to rwy 17T (the accident runway, just landing in the opposite direction) in 2009. The distance to the rwy threshold is less than 1.5NM. Can you spot the runway? Imagine looking for that in low cloud and fog, let alone adding being on the wrong approach frequency.

Lots of links in the accident chain on this one, and again the final report hasn't come out yet - but if it went even remotely close to how I think it went, you can see the tragedy that resulted from a bunch of different factors that added up all at once.

Monday, December 26, 2011

Merry Christmas and Happy Holidays from Lisa, myself and the angriest and smelliest reindeer ever :)

Wednesday, December 21, 2011

I'm referencing my last post, so read down to get caught up.

Those of you who guessed there may have been some confusion with the Teterboro VOR and ILS frequencies, you win a cookie!

The inbound ILS course was set, but when the non-flying pilot read the chart and saw the TEB frequency in the 108 mhz range, he thought "108 is an ILS frequency", and tuned it and identified it. It's not quite that simple,so I'll paraphrase from Wiki:

VORs are assigned radio channels between 108.0 MHz and 117.95 MHz. Turns out the first 4 MHz is shared with the ILS band though. To leave channels for ILS, in the range 108.0 to 111.95 MHz, the 100 kHz digit (the first decimal place) is always even, so 108.00, 108.05, 108.20, and so on are VOR frequencies but 108.10, 108.15, 108.30, and so on, are reserved for ILS.

I didn't know that, but now I do. 108.4 would be a VOR because the .4 is even, where 108.9 would be an ILS because the .9 is not. I learn something new every day, and it's a good way to add a cross-check to the frequency being selected.

As we weren't expecting to see a glideslope indication (the G/S being out of service on this particular day), no warning flares went up right away.

Fortunately for us we were in good weather so we saw the airport a fair ways back, and got aligned with the runway, then discovered our mistake. Also fortunate for us, we were in a radar environment with ATC services, so if we had gone off-course in any major way (or even a minor way, this being among the busiest airspace on the planet), ATC would have let us know.

So how do we prevent this from happening again? Well, with SMS we ask ourselves a bunch of "whys".

Why were we off course? The wrong frequency was entered by the non-flying pilot on the approach.

Why was the wrong frequency entered? The non-flying pilot misread the chart and didn't see the correct frequency.

Why did he misread the chart? From his previous flight experience, he assumed a 108.** frequency would be an ILS (the desired approach), and didn't read further to the right of the chart where the ILS frequency was. It doesn't help that most people read from left to right, and the box on the left was the VOR, while the box on the right was the ILS. A contributing factor was that this was his first trip into the airport.

Why didn't the other pilot catch that in the approach briefing? Because our SOP's dictated that the non-flying pilot is solely responsible for entering the approach frequencies into our radios. A contributing factor was the Glideslope was out of service, so nobody was expecting to see it, the lack of which normally would have been an early indicator of something not-quite-right on the approach.

A-ha! The first part of that last paragraph seems to show a weakness on the part of our procedures - we had just one pilot in charge of confirming the information in the box, and while it had served us for a decade, this incident showed us that it could be improved. I immediately issued a memo stating that from now on, both pilots shall confirm the proper radio frequency is entered into the navigation radios prior to the start of the approach. We have added standard phraseology to our approach briefings "Radio tuned and identified to ***** (whatever the appropriate frequency is), Confirm?" The other pilot will physically point to the approach chart with the desired frequency and then to the nav radio, then both will listen to the proper approach ident, followed by "Confirmed". This will put both pilots more in the loop when it comes to entering this data into the box, which will add an extra layer of protection into our approach procedures.

Tomorrow we will talk about another aircraft that *appears* to have made a chillingly similar error, but with considerably more tragic results. It's pretty fascinating, and from our recent experience I totally get how it could occur. I'm talking about First Air flight 6560, which crashed in August of this year. More tomorrow.

Sunday, December 18, 2011

Something happened recently that I want to share with you. We made a mistake, and learned from it. As part of our SMS (Safety Management System), when we make mistakes we try to identify the root cause, and then come up with a way to mitigate the situation so we don't make the same mistake in the future.

We fly to KTEB, Teterboro New Jersey, a lot. It's where most of the corporate jets go when they have passengers who wish to do business in downtown Manhattan. It's about a 30-40 minute drive from the airport to Wall Street, and the airport has less congestion than Newark, JFK or La Guardia.

A little while ago, the plane did a KTEB trip. It was an early-morning departure, and the weather was fine. The approach in use was the ILS (Instrument Landing System) to runway 06. On a full ILS approach, the radio signal lines you up with the runway and also tells you when to descend, and if you are high or low on your descent angle. The ILS approach is the most common instrument approach in most large airports, because most airplanes that use it can fly down to 200' above the ground without looking outside, and on some advanced ILS approaches, some sophisticated airplanes can use the radio signal to fly right down onto the runway without looking outside at all - pretty handy in places like Boston or Vancouver where it can get pretty foggy.

The only unusual variable on this trip was that the ILS for runway 06 had the glideslope radio signal out of service, so the approach only offered lateral guidance, ie it would only line you up with the runway, and you had to use alternate methods to calculate when to descend.

That's a link to the ILS approach onto runway 06, so you can follow along in glorious hi-res if you have a .pdf reader.

This next part is kinda technical, and I can`t figure out how to make it less so for non-pilots, but I`ll have a summary afterwards so bear with me.

The approach is pretty straight-forward - Air Traffic Control will give you radar vectors to intercept the inbound track, which is an angle of 060 degrees to the airport. They usually angle you so that you intercept the approach track at Vings intersection. Now there are a few ways to verify that you are at Vings intersection. One way is to put the waypoint in your GPS receiver, another way is to put the 294 degree radial off the JFK VOR and the 080 degree radial off Solberg VOR and fly over where they cross each other, but another way is to dial in the TEB VOR and when you are on the localizer and at 12.5 DME (taken from the VOR), then you are at Vings. The last way is the way we chose to identify Vings.

In our SOP`s (Standard Operating Procedures), the pilot flying (the guy in the left seat) will brief the approach and the non-flying pilot will tune and identify the radios. This was done, but as the plane approached the airport, it became obvious that the airplane wasn`t quite where it should be. The weather was good, so the pilot flying was able to see the runway about 20 miles back, and lined the airplane up with it nicely.

The thing is, the navigation radios were saying that the airplane was pretty far to the left of the runway, and this caused some confusion in the cockpit. I happened to be sitting in the back on this flight, watching the flight crew, and I was also confused for a few seconds.

Take a long hard look at the approach plate, and see if you can figure out the mistake that was made. Keep in mind that the glideslope was out of service for the ILS, so we weren`t expecting to see any glidepath information, nor did we.

//edit - to add a little information, the plane was showing off-track by only a few degrees, but our localizer indicator was showing nearly full deflection even though we were lined up with the runway. As mentioned earlier, the weather was fine so the briefing was for the visual approach backed up by the Localizer (the localizer is the ILS system without the glidepath system). The pilot flying had flown this approach probably a hundred times in the past few years, but this was the first time the non-flying pilot had been into Teterboro.//

I`ll talk about what happened next tomorrow.