Vacuum Pump Failure (cont'd)

Two days ago I put up a post concerning the (really quite benign) circumstances surrounding the failure of N631S's vacuum pump and the subsequent installation of a new one. But perhaps the subject is worth a little more discussion.

It's an element of "common knowledge" in General Aviation that dry-vane vacuum pumps fail after around 500 hours in service. Monday's failure occurred at 576.6 hours. I had the predecessor component replaced prophylactically at 573 hours. So the question arises naturally: why did I not replace the unit that failed when it reached 500 hours in service a few months ago? Well, it's all Mike Busch's fault.

I've been a fan of Mike and his approach to aircraft maintenance since I ran into his "Savvy Aviator" columns on the AvWeb site. I've taken his seminar (probably the highest-ROI dollars and hours I've spent as an aircraft owner) and I read his articles first when the Cessna Pilots Association magazine shows up each month.

One particular series of three columns that Mike wrote in 2007 on the topic of Reliability Centered Maintenance is on point for the present conversation.

In these columns, Mike uses real world research and analysis to deprecate the time-worn concept of TBO. I can't summarize his work in a paragraph or two (please, go read the columns!) but I think it's fair to say that he makes a good case for maintaining "on condition" where tools exist to evaluate condition and where the relationship between failure probability vs. time and the consequences of failure can be well understood.

For example, we used to run an engine for 2,000 hours and then pull it off the airframe and send it out for overhaul. Now, we carefully examine oil filter media and do spectrometric oil analysis at each oil change. As long as we don't find alarming deposits in the filter, nor see unsettling trends in the oil analysis reports, we keep on running the engine.

We used to rework cylinders if the compression test result fell below some arbitrary value. Now we do a much "smarter" compression test and we rely on borescope inspection more heavily in judging whether a jug is actually in trouble.

But how should we deal with components that offer no good way to evaluate condition? Until recently, vacuum pumps fell into that category.

(Note: The pump that was just installed in N631S has an inspection port. It is now possible to open this port, align a vane with the opening, and insert a probe to measure the amount of wear the pump vanes have accumulated. The manufacturer recommends an initial inspection at 500 hours and subsequent inspections at 100 hour intervals, and provides a "go/no go" limit for wear that dictates when the pump should be removed from service. Older pumps lacked this feature.)

Inside the pump (See photo at left from Sacramento Sky Ranch Inc.), a carbon rotor spins within the ellipsoidal cavity. The carbon vanes ride in grooves in the rotor and are thrust outward by centrifugal force, against the cavity wall. The vanes push the air around, and of course over time they wear. Eventually they wear to the point that they come adrift, and the pump fails - abruptly and catastrophically. Lots of them fail at around 500 hours. Lots more will go for 700 or 800 or even 1,000 hours. Until the advent of the latest pump design, it has been infeasible to evaluate the condition of the pump in service, so the choices available have been:

• Replace the pump on a preventive basis at or near 500 hours in service.
  
• Run to failure, with the understanding that an in flight failure will have consequences that must be managed.
  

A vacuum system failure can be insidious and deadly. For example, this NTSB report describes a fatal 2004 accident where a vacuum pump failure killed an unprepared pilot. The fatal Carnahan accident in 2000, is well known and was probably central to Parker-Hannifin's decision to withdraw from the aircraft vacuum pump business. The FAA, for it's part, has published pamphlet P-8740-52 titled "Silent Emergency: Pneumatic System Failure", describing the insidious nature of vacuum failures and recommending protective measures. Rapco, the current vacuum pump manufacturer, recommends that those who install their products provide this pamphlet to the aircraft owner and enter the fact that they have done so in the airplane's maintenance records.

Back, then, to the question: why didn't I replace N631S's vacuum pump at 500 hours in service? After all, I had replaced the previous unit based on time in service. The answer is that in the intervening time I had absorbed Mike Busch's teaching on Reliability Centered Maintenance and had adopted the philosophy in maintaining N631S. In the case of the vacuum pump, I could not maintain "on condition", so I chose to develop a plan for managing the consequences of failure and then to consciously adopt a "run-to-failure" strategy.

The problem is two-fold. First you have to plan to recognize the failure and second, you have to plan to deal with the failure. For the first part, I had the Precise Flight low-vacuum annunciator installed - and it worked. When the pump failed, the light lit and I was aware of the problem in a very few seconds.

I was fortunate that the failure occurred in good VMC weather. But if it had occurred in IMC, I could have dealt with the failure using two resources that are fully independent of the vacuum system. First, the STEC System 50 autopilot bases its control of the aircraft on the output of the turn coordinator, an electrical gyro. (Those who fly with an autopilot that references the pneumatically driven DG or HSI need a different strategy.) Second, the GPS, backed up by the compass, provides the course guidance needed to get out of trouble. Therefore, I had confidence that I would recognize and have the ability to manage a pump failure in any set of flight conditions.

A vacuum pump can fail at any time, so this plan for managing failure of the pump remains in effect. However, I expect that when the new pump's 500th hour rolls around I will be exercising the new inspection capability to establish its condition and decide whether it needs replacement.