Land and Hold Short

Canada is in the middle of a federal election (with no TFRs, proving to the U.S. that it can be done). This photo shows Stephen Harper, one of the four major party leaders, at my home airport in Ottawa, using the cockpit of his chartered jet as a campaign prop (not meant as an endorsement or an attack — I’m sure any of the other party leaders would do the same).

If you look very carefully at the middle of the top right quarter of the photo, you’ll see the best part — through the window, in the distance, is a poor line person, standing outside in sub-freezing temperatures, waiting to marshall the plane once the photo op is finally finished.

(The photo is copyright 2005 by the Canadian Press. I found it originally here.)

From aviation pundits with deadlines to meet and empty pages to fill, we hear a lot about the dangers of losing a vacuum pump (and consequently, attitude indicator and directional gyro) in IMC, and why IFR pilots need to (a) practice partial panel flight a lot, and (b) have a backup vacuum pump or electric AI.

As I’ve mentioned before, while it’s possible to find a couple of fatal accidents every year caused by loss of the vacuum pump in planes with retractable gear during a (legal) IFR flight, it is extremely difficult to find any in planes with fixed gear. In fact, I had failed to find any at all during my initial research. So thanks to Paul (N9002F) for drawing my attention to two from the NTSB files:

MIA98FA045

A Cessna 172 near Raleigh-Durham, NC on Christmas Eve, 1997

ATL91FA067

A Piper Cherokee near Hamilton, NC on March 18, 1991

A full report is available for the Raleigh-Durham crash, while only a summary is available for the Hamilton crash, but they both make interesting reading. To start with, as a brutal irony, both planes were equipped with functioning backup vacuum pumps, precisely what’s supposed to prevent this kind of accident. Even more ironically, it looks like the backup pumps themselves might have contributed to the accidents, at least in a small way.

Too much diagnosis, not enough flying?

In the first accident, the problem took place right after takeoff into low IMC. The pilot diagnosed the problem immediately and reported a vacuum failure to ATC, and then (as the NTSB determined from the wreckage) selected his standby pump. After that, the plane continued in a turn until it hit the ground.

The NTSB found that both the main and standby vacuum pumps were actually working and the gyros were undamaged — in fact, it is most likely that there was no failure at all. They also tested and discounted the possibility that a tube might have worked loose in flight, causing a (false) vacuum warning light on the panel. Furthermore, the flight lasted only 2 1/2 minutes, while it would have taken about 10 minutes for the gyros actually to spin down after a failure.

Why did the pilot report a vacuum failure? Could it simply have been a case of the spatial disorientation (his body disagreeing with the instruments), followed by the distraction of trying to troubleshoot a non-existant vacuum problem and select a standby pump? In my experience hand-flying my Warrior in IMC, initial climbout is by far the most difficult part of IFR flight, since the plane naturally wants to turn, and you have to keep strong rudder pressure to stay on course. Even the slightest distraction, like a radio call, and throw your course off 10-15 degrees if you’re not careful or a bit out of practice — in this case there was a lot more distraction than that, and sadly, having a backup vacuum pump to fiddle with (instead of flying the plane to a safe altitude first) probably made things worse.

Backup is not primary

The Hamilton, NC report does not give as much information as the other one, but it does mention the following:

• the pilot reported a pitot-static failure as well as a vacuum failure
• the pilot decided to continue the flight using the backup vacuum pump (and, presumably, no altimeter, VSI, or airspeed indication!!!)

After 28 minutes of erratic flying, the pilot finally lost control of the aircraft. The brief report does not indicate whether the backup vacuum pump also failed, but it does mention “improper use of equipment that affected the operation of the standby vacuum pump”. Did having a backup vacuum pump give the pilot the confidence to continue the flight?

Tentative conclusions

So neither of these is a clear case of a pilot losing control of a fixed-gear plane because of a vacuum pump failure. In the first case, both the primary and backup vacuum pumps (as well as the gyros) were all working properly, and we cannot know why the pilot thought otherwise; in the second case, the pilot was also facing a pitot-static failure, knocking out most of the panel, but decided to keep on flying for a half hour in IMC.

Are fixed-gear planes just a lot easier to fly partial panel, then? That’s what a 2002 ASF/FAA study suggests. Two groups of pilots were tested in actual aircraft rigged up to allow unannounced gyro failures — one group was tested in a Beech Bonanza (retractable), and the second group was tested in a Piper Archer (fixed gear). The results? The Archer pilots did a lousy job diagnosing the problem — it took them nearly 7 minutes on average to realize that something was wrong — but every single one kept control of the aircraft. The Bonanza pilots, presumably a much more experienced group of complex-aircraft pilots, diagnosed the problem faster — in less than four minutes, on average — but couldn’t all control their planes flying partial panel, and 4 out of 16 were judged to have crashed (i.e. someone without a hood had to take the controls).

I have two tentative conclusions from all of this, though I am far from an expert:

1. While installing an aftermarket backup vacuum pump in a fixed-gear plane might not hurt, it probably won’t help either, and might even be a dangerous distraction — investing the same amount of money in recurrent training, better maintenance, etc. probably makes more sense.
2. It’s a lot easier to maintain control of a fixed-gear plane flying partial panel, so upon losing gyros or the vacuum pump in a retractable, maybe checklist item #1 should be drop the gear to add drag and make the plane more controllable (if you’re above gear-extension speed, let that be the mechanic’s problem after you land). After all, the 25% fatality rate for retracts in the ASF/FAA simulation makes for lousy odds.

I’ll look forward to comments from people with other interpretations and suggestions, especially since I fly only fixed gear myself.

This morning, I read a CBC story about a control problem on an Air Canada Jazz Dash-8 flying from Kingston to Toronto on 2 September 2004. From the story, it was pretty hard to figure out what had actually happened, especially given statements like these:

• “It took the strength of two men to steady the control column of an Air Canada Jazz flight veering dangerously out of control” sounds like directional control problems — jammed aileron? asymmetric flaps?
• “pilots struggling to gain control of the plane from the moment after takeoff” directional control?
• “The control column, a stick used to manoeuvre and control altitude, was forcing the plane’s altitude higher as the craft continued to pick up speed” sounds like excessive up trim, but then why would the plane be picking up speed if the nose were up?
• “loose nuts caused bolts in the plane’s elevator spring tab - a part of the aircraft that helps maintain control - to move out of place and throw the plane’s balance and control out of whack” OK, then it’s elevator trim

I didn’t feel much better informed after reading the article, so I hunted down the actual TSB report. It makes terrifying reading, not because of the trim-control problem (as troublesome as that was), but because of the risk of catastrophic structural failure.

As far as I can understand from a quick reading, the Dash-8 has elevators that can move separately on the left and right sides, and each has its own trim tab — the left and right trim tabs have to be balanced with weights so that one doesn’t pull more than the other. When the incident plane was in for painting a while before the accident, the tabs were rebalanced. The TSB’s best guess is that the AME left the bolts loose on one side in case the weight needed to be removed for more balancing, and forgot to tighten them afterwards.

Eventually, the weight fell off, leaving the tabs badly out of balance. As the Dash-8 accelerated for takeoff from Kingston, it sought to trim nose-high. The pilot flying (first officer) noticed that very little pressure was required to lift off, and soon both pilots were pushing forward hard to maintain airspeed, even with full nose-down trim. After running through some checklists, the captain disconnected the copilot-side trim (I think — this part is a bit hazy), and then was able to trim the plane for cruise using controls on his side. Althought they had declared an emergency 30 seconds after takeoff, they decided to continue to Toronto (about a half hour away) rather than landing at CFB Trenton between Kingston and Toronto.

I can understand their decision. After all, trim is mainly just a convenience for the pilots, to save us having to push or pull the yoke constantly during flight. On such a short flight, and with one of the independent trim systems (apparently) working, why not just continue the last distance into Toronto, rather than dumping up to 50 passengers on the tarmac of a military base in the middle of nowhere?

Unfortunately, things could have turned out very badly. Because of the weight imbalance, the two elevators were exerting different forces — one was trying to push the nose down, and one was trying to pull it up. According to the TSB report, this caused a twisting force on the vertical stabilizer close to its structural limit. That reminds me of the force that snapped the tail off American Airlines flight 587, though the cause was rudder oscillations, in that case. In hindsight, we know that that was the real danger of the flight and that the trim problem was only a sympton.

If the flight crew had known the real risks, I don’t doubt that they would have set down in Trenton without a second’s thought. In my own flying, I’ll try to keep in mind that any anomaly I can actually detect may just be the tip of a very large iceberg.

I’ll guess that 80% of my flying is done in continuous radio contact with ATC, either IFR, VFR in class B/C/D airspace, or VFR with flight following in class E/G airspace. This kind of flying has its own challenges, but one of the biggest ones is learning to read controllers’ minds (I’m sure that they’d tell you the same about dealing with pilots). If you can figure out what the controller is thinking — what her or his plan is to get you to the approach or the airport, why you’re being vectored, what traffic he or she is worried about, etc. — you can anticipate what’s going to happen next, and sometimes you can even help out a bit, especially if you can tell that the controller’s getting overwhelmed (again, just as controllers do with pilots).

Mind reading is also sometimes necessary, though, simply because of sloppy terminology. Here’s what happened to me yesterday — I was returning to Ottawa VFR, and a few miles from the airport, terminal control passed me to tower with a minimum altitude restriction of 2,000 feet still in force. Here, as far as I can remember, is what the tower controller said:

Tower: BJO, I’m going to have you follow the downwind for 25 first, then bring you around for the left downwind on 22 [my intended runway].

This is pretty normal when arriving at the airport from the southeast — they want to keep me out of the departure path of 25 and have me cross the middle of the jet runways at circuit altitude. But what about my altitude restriction? When you’re cleared to any leg of a circuit (in Canada, anyway), you are automatically cleared to descend to circuit altitude, but was I cleared for a downwind leg? I didn’t hear any clearance in the controller’s communication — feel free to leave a comment if you disagree — but my amazing powers of telepathy told me that because the tower controller used the word “downwind”, he thought he had cleared me; unfortunately, if I had acted on his assumption and ended up with a loss of separation with an IFR aircraft, the tapes would have shown me in the wrong. I decided to help things along a bit a couple of miles from the airport:

Me: Tower, BJO has an altitude restriction of 2,000 feet. Is it OK to descend to circuit altitude now?

Tower: Sure BJO, descend to 1,500 feet. For future reference, a clearance to downwind lets you descend to circuit altitude.

Yep, I read that one correctly — he thought he’d cleared me to downwind. I thanked him (no point wasting radio time on a long discussion), and counted myself lucky that he wasn’t wearing a foil hat to shield his brain waves.

The story goes (true or not) that 911 gets more than its share of calls about airplanes in distress over the practice area west of Ottawa. I don’t blame the people calling in — stall practice is (hopefully) too high to see details, but someone might still note the change in engine noise; a forced-landing practice down to 500 feet AGL, on the other hand, looks and sounds an awful lot like the emergency it’s trying to simulate, right down to engine-surging noises (advancing the throttle briefly every 500 feet) and the plane disappearing behind the trees for most viewers on the ground.

I think that might be what happened in New Brunswick yesterday. According to this CBC story, several people on the ground reported reported a low-flying aircraft with engine trouble, and, in this case, S&R took the reports seriously enough to dispatch a Hercules aircraft, a Comorant helicopter, and a Coast Guard cutter (!!) to investigate, possibly because local media had already picked up the story.

The lean-of-peak debate, which I’ve written about before, has just gone mainstream — check out this Forbes piece, part of a series of online articles about institutional stupidity. The focus is on Lycoming’s business practices (deny evidence that you cannot refute while insulting your customers in the process) rather than the technical fine points of LOP operation, but still, there it is, out of the pilot-geek closet.

Back in July 2005 (updated in September), Philip Greenspun published a detailed owner’s review of his factory-new Cirrus SR20, the 200 hp (cheaper) sibling of the Cirrus SR22. Philip has flown his plane pretty seriously around a huge part of the continent (including the Canadian arctic), and has a long list of both the good and bad aspects of it. There are dozens of annoyances — large and small — that you’d never know about after reading a glossy magazine review written on the basis of a single test flight.

Philip also speculates about the actual safety value of the chute. In his opinion, most or all of the cases where people have pulled the chute and survived would have been survivable in a non-chute equipped Piper or Cessna, while in the cases where the chute really was necessary, it failed and resulted in a fatal accident. I don’t know how accurate this analysis is, but it’s an interesting perspective, especially coming from a Cirrus owner.

This page is highly-recommended reading, even if (like me) you cannot imagine ever being able to afford a factory-new plane. The 6-8 week, USD 10,000/year annuals — while still under warranty — are certainly an eye-opener. I wonder how much he’ll pay for maintenance when the plane is a little older and the warranty has expired. It sounds like a pretty nice plane, though, on balance, and I certainly wouldn’t turn one down.