Exhaust valve failure analysis 3

[Azraelo]

Hi everybody! Please take a look at these photos. An exhaust valve failed in an automotive natural gas engine. The engine is new, just 1400 hours (around 28000 Kilometers). Operator complains about low output power and vibration. No remarkable fault codes logged in ECU (just 3 counts of oxygen sensor). Boroscope reveals failure. When disassembling, there was a lot of clearance between this valve and its guide. When taken out, bechmarks are noticed in the valve crack, consistent with mechanical fatigue. The valve guide diameter was fine at the top, but much much wider at the bottom. Seems that there was a a bad valve sitting (by the uneven pattern in the seat). But more interesting to me is the crack following a circular pattern and the order of the events. According to your expertise, what do you think happened here? what happened first? the fatigue failure or the out-of-round in the valve guide? Any positive suggestions are welcome. Thanks.

 

[gruntguru]

WOW!
Was the engine running OK when new?
If so, how long did the symptoms take to appear?
Is the O2 sensor code possibly due to lean mixture?
Have you inspected an exhaust valve from another cylinder?
Do all cylinder get the same AFR or is the gas multi-point-injected?

The guide wear might be a result of overheated valve. The seat is badly recessed for a new engine also pointing to a very hot valve. Some aspects of the failure look quite brittle - I would get a hardness test on the valve head.

je suis charlie

 

[Tmoose]

Are any of the other valves cracked ( as determined by mag particle or dye penetrant inspection) ?

Radial cracks on valves are often attributed to overheating or uneven heating.
In my mind sort of similar to cylinder heads cracking between valve seats or spark plug.
Get hot locally, metal becomes upset, and upon cooling tensile stresses are very high, too high. Once a crack starts, there are plenty of environmental factors to keep them moving inward and then across the valve head.

It is probably too late to determine How concentric the valve seat is to the (now worn guide), or how concentric the valve's seat was machined.
But I would check and quantify valve runout and seat concentricity on all the valves and seats.

Also check for tuning aberrations, like accurate ignition timing on all cylinders, vacuum leaks, etc.

 

[Lou Scannon]

Please advise, is this a stoichiometric, or lean-burn engine? Is this a fully validated, production engine, or is there anything experimental/unproven about the control system or the base engine conversion from a gasoline engine design (e.g. higher grade exhaust valve and seat materials than the gasoline engine)?

"Schiefgehen wird, was schiefgehen kann" - das Murphygesetz

 

[PJGD]

Download the Technical Information Catalogue from G&S Valve Ltd here:

http://www.gsvalves.co.uk/catalogues.html

Then take a look at pages 25 to 29 for some clues.

PJGD

 

[gruntguru]

Great question Hemi. When lean, NG burns very slowly and kills pistons and exhaust valves quickly if ignition timing, turbulence etc aren't right.

je suis charlie

 

[catserveng]

This appears to be a pretty good example of what we used to call a "chordal fracture", one of the pictures shows the fatigue fracture beach marks radiating from the initiation point, and the rougher area indicates the final fracture.

In most of the engines I am familiar with this type of failure was usually due to exhaust temperatures in excess of 1200 degrees F. So the question above about this being a stoich or lean burn is important. Do you have any form of exhaust temperature monitoring? Another possible cause was valve seats and heads not concentric, again as pointed out above.

A careful inspection of the parts that didn't fail can help you determine the actual root cause.

A lot of great questions got asked, if you could answer those and provide some better information about the engine and its application you may get some better answers.

MikeL.

 

[romke]

it looks like the valve never has been correctly seated. what happens then is that the stress on the valve face is not evenly distributed but that somewhere on the circumference a "stress raiser situation" occurs. that may lead to a initial crack that later on results in the damage seen. when the load is not evenly distributed also a bending moment will be introduced into the valve stem resulting in uneven wear of the valve guide.

 

[gruntguru]

Two best ways to have the seating that bad - misaligned guide to seat insert or bent valve.

je suis charlie

 

[turbomotor]

I would guess (and at this point I emphasize the word "guess") that the valve did not seat properly and an exhaust leak caused unequal circumferential heating of valve, with related hoop compressive yield when hot and crack propagation during the cooling cycle.

I think the pictures suggested by PJGD are very enlightening.

The hole in my hypothesis is a lack of evidence of an exhaust leak in the picture of the cylinder head and valve seat.

Was the valve designed to rotate?

 

[dicer]

Looks like running lean, the valve stem even discolored? Lots of heat, even what looks like recession on the face.
What material is it, and looks like hard material? Heat stress failure.

 

[Azraelo]

Hi again!

I will try to solve some of your questions:
It is a mass-produced, dedicated, non-modified engine running on natural gas.
According to specifications, it is single point injected, lean-burn engine. It can be as lean as 23:1 (40% excess air) depending on load conditions.
No temp sensor in exhaust side, so I'm not sure about exhaust temperatures. But in other natural stoichiometric gas engines, I have seen 700°C in exhaust (roughly 1400 °F). This engine was engineered for trucks or buses applications.
Oxygen sensor related fault code is described as "The measured air-to-fuel ratio is greater than the desired air-to-fuel ratio." Seems that oxygen sensor detected excessive oxygen in exhaust, as if it was a cylinder which is not properly burning their air-fuel mixture. I have seen this code when a defective coil, wire or sparkplug were not properly burning the mixture. Just 3 events of this fault code.

Take a look at the valve stem, seems that the exhaust gases passed through the seat and burned the stem far beyond the valve head.

ECU has logged an overspeed event It reached 3400 RPM and 4300 RPM in two different dates. The governed speed is 3100 RPM. At 3100 RPM, ECU cuts off fuel injection. The 3400 rpm event last less than a second, so I think was due to bad gearshifting. But I'm skeptical of this engine reaching 4300 RPM and be alive to tell the story, with just ONE cracked valve. No damages in block, pistons, liners, valves, crankshaft, camshaft, rockers or pushrods. Just piston and cyl head #2 are damaged, due to the fallen pieces of the cracked valve smashed between piston and cylinder head. Being the governed speed 3100 RPM, at 4300 RPM I would expect block shattering or piston destruction. Am I right?

Turbo also has a cracked blade in turbine wheel. Since the turbo wheels are spinning at 100 to 200 thousands RPM, Is it possible for the dettached valve pieces to hit the turbine wheel and damage just ONE blade? Does the valve debris broke down into small pieces before reaching the turbo? Most of them perhaps went out through the open wastegate and reached the exhaust pipe, and just a small one debris hit ONE turbine wheel. Not sure, due to high turbo speed.

I will try to upload more photos.

Thanks PJGD for your download link. Very interesting.
Thank you all for your replies, very helpful in order to trying to understand this failure.
--
Azraelo

 

[gruntguru]

Without knowledge of the engine it is impossible to predict the consequences of a 32% overspeed. Many engines will survive a brief event like that without bottom-end failure. It is possible however that valves have hit piston crowns (any evidence on the pistons?) and this valve may have been bent as a consequence. That would certainly explain just about everything.

Debris ingestion usually damages the other end (the tip) of a turbine blade and usually all of them - not just one. You need to get the turbine housing off to inspect the entire wheel. The damage is usually bent and eroded blades - not cracked.

je suis charlie

 

[dicer]

The other valves show no sign of contact with piston. The valve problem is strictly heat related. The over speed shouldn't do a thing, especially the very short time. Yes the pieces can do damage else where like shown.
What size engine etc.? What brand is it? How was coolant temps and what compression ratio? Piston must have a large dish chamber.