The Savvy Aviator #46: Double Trouble
As many of you know, I've owned, flown and maintained a Cessna T310R for the past 20 years, powered by a pair of Teledyne Continental Motors TSIO-520-BB engines. I've achieved remarkable longevity from the cylinders on these engines -- 10 out of 12 are still originals with more than 4000 hours in service so far.
But not every aircraft owner is quite so lucky, and a few seem to be downright unlucky. Check out this email and photo I recently received from an owner:
I was flying westbound at 6000 feet MSL (3300 AGL) from Bismarck, N.D., to Lewistown, Mont., in good VFR weather. After about one hour of flight, while flying over the Dakota Badlands, about 20 miles from the Montana border and about 400 miles from the real mountains, I heard a loud "pop" and developed a slight engine vibration. Fortunately, the engine continued to run at reduced power. The manifold pressure, oil pressure and temperature remained within limits. Outside of the initial "pop," there was not excessive noise or vibration, although the engine was rough. In spite of the normal manifold pressure, I lost about 15 knots of airspeed, but was able to maintain altitude.
Fortunately I had chosen to follow Interstate 94 instead of flying direct. There really is not much civilization out here once you get away from the interstate. But I-94 provides a 500-mile-long emergency runway. I routinely keep track of emergency landing spots. It was only about 20 miles back to Dickinson, Mont., (KDIK), so I immediately turned back. The engine ran OK and I made a precautionary simulated-engine-failure approach and landing at Dickenson and taxied in.
After landing I noticed a lot of oil on the ramp and all over the aircraft. I had lost about three of the nine quarts of oil. It made a real mess at the ramp. When we removed the top cowling, we noted that the #1 cylinder head had separated, as you can see in the accompanying photograph.
Fortunately, Milt Purvis at Dickinson Air Services jumped right on the problem. With the help of FedEx, we had a replacement cylinder on the aircraft the next day. I test-flew the aircraft for about 0.3 hours and it was inspected after landing and found OK. We loaded up and continued on our way to Montana. After takeoff, I climbed to altitude in the traffic pattern, made sure everything looked good and turned on-course toward Montana, again following I-94.
About 40 miles from Dickinson, it happened again! Same symptoms! This time I was only about six miles from the Beach, N.D., (population 900) airport (an unattended airport along I-94 about three miles from the Montana border). Upon removing the cowling, we found a similar cylinder head separation on the #3 cylinder.
The Continental O-470 engine is about 400 hours from TBO. It had been running exceptionally well during the 30 months and 300 hours that I have owned the airplane, although oil consumption was relatively high -- a quart every 3-4 hours. The aircraft had spent much of its prior life in Florida and showed corrosion on the cylinder fins. I do not know if the cylinders were new or overhauled on the last engine major.
The engine is currently being replaced in Beach by the folks from Dickinson (60 miles away) with a gold-seal engine from Western Skyways.
I have not seen a failure similar to mine on a Continental O-470 engine, much less two failures within two hours! I do not know if the failure was from fatigue or corrosion. In hindsight, since all cylinders had the same history, I probably should have bit the bullet and replaced the engine after the first failure.
Actually, cylinder head separations are not all that uncommon -- although two separations in two hours on the same engine certainly is extraordinarily rare. This owner/pilot clearly did an outstanding job of keeping his cool and handling the in-flight emergencies well. He also did a great job of flight planning to make sure he always "had an out" nearby in case of trouble.
But could these catastrophic engine failures have been prevented?
There are two major causes of catastrophic engine failure: wear and fatigue. Failures caused by wear are almost always preventable, because wear events are usually detectable well in advance of failure using standard condition-monitoring techniques like oil filter inspections, oil analysis, compression tests and borescope inspections.
In contrast, fatigue failures usually happen suddenly, with little or no warning. They almost never "make metal" beforehand that could be detected through oil-filter inspections or oil analysis. Sometimes a sharp-eyed IA will catch an incipient fatigue crack in a crankcase or cylinder head during an annual inspection. But once-a-year inspections are simply not adequate to detect head cracks reliably before failure occurs. Sometimes we get lucky and catch head cracks before failure, but sometimes we aren't so lucky -- as our intrepid owner learned the hard way while flying over the Dakota Badlands.
Fatigue is the progressive structural damage that occurs when metal is subject to cyclic stress. The process starts with a microscopic crack (called the initiation site) that widens slightly with each stress cycle. As the part continues to undergo repetitive stress, the tiny crack begins to grow more rapidly. After many stress cycles, the crack grows to critical length at which point crack growth becomes unstable and complete failure of the part is inevitable.
Engineers use a graph called an "S-N curve" to characterize the fatigue properties of a particular material. The S-N curve plots the magnitude of repetitive stress S (in pounds per square inch) against the average number of repetitive stress cycles N the material can endure before it fails. The S-N curve reveals a substantial difference in fatigue characteristics between ferrous metals (like iron, steel and titanium) and non-ferrous metals (like aluminum, magnesium and copper). Ferrous metals exhibit a "fatigue limit" stress below which they can endure an infinite number of repetitive stress cycles without failing. Non-ferrous metals have no fatigue limit, and will always fail eventually if subjected to enough stress cycles.
Piston Engines And Fatigue
Fatigue failures can occur to various critical components of a piston aircraft engine, including the crankcase, crankshaft, connecting rods, pistons, and cylinders. The crankshaft, connecting rods and cylinder barrels are made of steel and are engineered to operate well below their fatigue limit, so they have an infinite fatigue life -- at least in theory. We do see a small number of fatigue failures of these steel parts, but only when the parts are improperly manufactured (e.g., bad steel), improperly assembled (e.g., incorrect torque), or stressed beyond their design limits (e.g., prop strike). History shows that fatigue failures of steel parts due to improper manufacture or assembly typically happen rather quickly after the engine enters service -- typically within the first 200 hours, and often quite a bit less. They are known as infant mortality failures.
On the other hand, the crankcase, pistons and cylinder heads are made of aluminum alloy, so they have a finite fatigue limit. If a crankcase, piston, or cylinder head remains in service long enough, it will fail due to metal fatigue.
If the engine is operated within its design limits, the useful fatigue life of a crankcase, piston or cylinder head is something well in excess of recommended TBO -- generally at least two or three TBOs. Pistons are always replaced at major overhaul, which is why we very seldom hear of the fatigue failure of a piston. Crankcases are normally reused at major overhaul, so crankcase cracks are not uncommon. Fortunately, the crack growth rate in crankcases tends to be quite slow, so a careful inspection at each annual inspection is usually sufficient to detect crankcase cracks long before they reach critical length.
Historically, cylinders have often been overhauled and reused at engine major overhaul, so it's not uncommon to see engines with cylinder heads that have been in service for two or three TBOs. Such high-time cylinder heads are thought to be at far greater risk of fatigue failure than are first-run heads that have been in service for one TBO or less. Furthermore, crack growth in a cylinder head can progress quite rapidly, so a once-a-year inspection is not sufficient to assure that fatigue cracks will be caught before the head fails catastrophically.
Nowadays, many top-notch engine shops encourage (or even insist on) installing new cylinders at major overhaul, which greatly reduces the likelihood of cylinder-head fatigue failure. But it's not a guarantee. Recently, we've seen a rash of infant mortality fatigue failures of cylinder heads due to improper manufacturing. Thousands of ECi Titan cylinders were shipped with defective head castings that had inadequate thickness in certain places, resulting in cracks developing within several hundred hours. Recently, Superior recalled a large lot of Millennium cylinders whose heads were improperly heat-treated, resulting in a number of catastrophic, in-flight, head-to-barrel separations at low time.
In addition to time-in-service and manufacturing errors, there are a number of operational issues that can affect the useful fatigue life of engine parts. One of the most important is corrosion. Corrosion creates surface pits that can serve as initiation sites for fatigue cracks and greatly foreshorten useful fatigue life. In fact, serious corrosion can result in fatigue failure of ferrous metal parts even though they are operated within their design fatigue limit and should theoretically be immune from fatigue failure.
Fatigue life is also profoundly affected by the magnitude of the repetitive stress cycles. Because aluminum engine parts are normally operated in a very flat portion of the S-N curve (low S, high N), a small increase in stress can result in a large decrease in fatigue life. If we're talking about cylinder heads, stress S is a function of peak internal cylinder pressure, which is affected both by power settings and mixture management. We know, for example, that peak internal cylinder pressure is maximum at a mixture setting of roughly 50°F rich of peak EGT (ROP), and is considerably lower at richer mixtures (e.g., 125°F ROP) or leaner mixtures (e.g., 25°F LOP). Thus, cylinders operated at 50°F ROP (as recommended in many POHs) are more likely to suffer fatigue failure than cylinders operated substantially richer or leaner.
Detonation, pre-ignition and advanced ignition timing can all result in abnormally high peak internal cylinder pressures, and can drastically reduce the fatigue life of cylinder heads and pistons. Any time spark plug or borescope inspection reveals the tell-tale signs of detonation or pre-ignition, replacement of the affected cylinder and piston is prudent.
Was It Just Bad Luck?
Why did this owner's O-470 engine suffer two head separations in the space of two hours? It's impossible to say for sure, but we can certainly make some educated guesses.
It is clear from the photo that the engine compartment suffered from substantial corrosion while the aircraft was based in Florida. It also sounds more likely than not that the cylinders were reconditioned and reused when the engine was last major overhauled. The reconditioned cylinders may have been dimensionally restored to new fits and limits, but there's no way to restore the fatigue life of a cylinder head. As the saying goes, "Metal never forgets."
The particular O-470 engine involved was an O-470-U, a high-compression version of the engine with an 8.6-to-1 compression ratio (similar to IO-470, IO-520 and IO-550s). It therefore has substantially higher peak internal combustion pressures than the older O-470 variants (-K, -R, -S) that were designed with a 7.0-to-1 compression ratio to run on 80/87-octane avgas. High compression ratio engines are more efficient, but place higher stresses on cylinder heads and other engine parts.
We have no way of knowing how the engine was operated during the life of those cylinder heads, but since the aircraft didn't have an engine monitor and the owner didn't mention his leaning procedure, it seems likely that the engine was operated "by the book" at around 50°F ROP, which we know is the highest-stress condition.
Speaking of engine monitors, I believe that if this aircraft were equipped with one and the pilot trained to use it properly, there's good chance that the cracked head might have been detected prior to the point where complete head separation occurred.
The fact that the #3 cylinder suffered a head separation just two hours after the #1 cylinder failed is certainly extraordinary. It makes me wonder whether perhaps the engine suffered some sort of detonation or pre-ignition event that caused the cylinders to be stressed beyond their design limits.
Take a close look at the photograph at right. Is it my imagination, or is there some perceptible discoloration between the cooling fins on the #3 cylinder that might suggest that the #3 head is cracked and leaking? With the benefit of hindsight, I think it's quite likely that the #3 cylinder head was cracked at the time the #1 cylinder was replaced, but the crack in #3 escaped the mechanic's notice.
See you next month.