Post Mortem

Well what a clusterfailure that was. Imagine if that vacuum and indication failure had all happened over the course of once turbulent flight in night IMC across the mountains. Suction failures are extremely dangerous. The suction-driven instruments are central to the instrument scan and experiments and studies of actual accidents show that they have a very high chance of killing you. The "ninety percent" figure sometimes quoted is misleading: accidents attributed to vacuum failures (about three per year in the US) result in fatalities 90% of the time. That's because there aren't many ways to crash an airplane due to spatial disorientation without it being deadly. More than ten percent of pilots who experience vacuum failure in IMC do get out alive, but I'm not going to guarantee I would be one of the survivors.

Here's what a vacuum failure can look like.

The gyrating blue and black instrument is the attitude indicator. It won't necessarily look that energetic in a failure. Its slide to one side or the other may look exactly like a functional instrument indicating a real wing drop, prompting the pilot to bank and pitch in an attempt to return the aircraft to straight and level flight. The pilot must fly by the turn coordinator, the tippy little white symbolic airplane at bottom left. It indicates rate of turn, not bank angle, so doesn't respond as instantaneously to changes in attitude.

I had difficulty identifying the failed vacuum system because its own indication system did not work properly. When I felt that the heading indicator was not working properly, I suspected the vacuum system, but the indicator lied to me, telling me the system was okay. When I shut down an engine, the associated vacuum failure indication appears immediately, leading me to believe that the same thing would happen in the case of a real suction failure. The two vacuum pumps are in two different engines, had widely separated serial numbers and had been installed by different people at different times. I had never heard of one vacuum pump failing and taking out another one.

If an electric instrument has a power loss, it immediately throws up a flag, so I know not to trust it. There is no failure indicator on the heading indicator or attitude indicator and without other reference, it is really easy to follow the dying instrument. The autopilot tried to follow it, too. With TWO failing vacuum pumps, there was no pressure differential across the shuttle valve to pull the autopilot out of the system. Flying across the mountains to the FBO where company has shipped the replacement pumps, I have post-it notes over the failed instruments, still the only indication, apart from their unresponsiveness, that they are not working. When practicing flying on instruments, if the pilot can see the real horizon, even a little, the task is much easier. Even if you try to ignore it, you subconsciously draw on a lifetime of doing what your species does: visually determining which way up you are. While it's easy to design an indicator that is invisible until electrical power is removed from it, I guess it's harder to calibrate one to not enough suction. Maybe it could depend on gyro rpm. This FAA document claims that "attitude indicators with failure warning flags are available in newer model aircraft and are slowly making their way into the general aviation fleet as retrofits," but I think they may be talking about replacing vacuum-driven AIs with electric ones.

The cause of the double failure is eventually determined to be poorly-installed air-oil separators. Oil was able to enter the dry vacuum pumps and damage the vanes. I'm not sure if the indication issue was because it was a gradual failure, or because oil got in the indicator, too. It is all functional now, but now the vacuum pumps are in phase, which makes me nervous, because that airplane has flown about 900 hours since that happened, and I'm back in it. This article seems to think that vacuum pumps fail every 500 hours or so, but our mandated time between overhaul is definitely 1000 h.