Compressor Destruction :(

I have always been amazed by how quickly and destructively large powerful machines can fail.  I happened upon some pictures recently of an Industrial Combustion Turbine failure, brought on by icing.  The other type of combustion turbine is aero, like you might find on an aircraft.  Industrial (or "Frame") turbines are quite heavy and cannot be flown.

A gas turbine consists of three sections:  Compressor, Combustor, and Turbine.  At the inlet of the compressor, air pressure is reduced because the compressor is sucking air in.  This inlet pressure drop also causes a slight temperature drop, which is how this turbine's problem began.

Take a look at this video of an aircraft testing an aero turbofan engine.  You can see water vapor entering the engine.   This water vapor appears because the drop in pressure (a.k.a. suction) at the compressor inlet is also causing air temperature to drop.  The temperature is falling below the dewpoint - that is, moisure present in the air is condensing suddenly due to the drop in temperature.  Those swirls are condensing water vapor entering the engine.



Certain weather conditions, however, will cause moisture in the air to change from harmless vapor to ice.  Relative humidity must be high, ambient temperature has to be near the dew point, and temperature has to be near or below freezing.  Remember that air temperature drops as it enters the compressor, so the temperature doesn't have to be 32 degrees.  Inlet icing can occur with ambient temperatures as high as 38 degrees.

Icing is only an issue for the first row of compressor blades, because they are the coldest ones.  Water vapor can form, and freeze to the blades, if they are cold enough.   Icing is not a problem further inside the compressor, because as the air pressure increases, the air temperature also increases.   In fact, at the discharge of the compressor of a stationary gas turbine, the air temperature is about 700 degrees F. 

If necessary, a portion of this 700 degree air can be "bled off", and brought forward to the inlet.  This raises the inlet temperature just enough to prevent ice from forming. This is called the Inlet Bleed Heat system.  You take a small hit on engine output, but it definitley beats scrapping the engine.

Gas Turbine compressors are designed using complex fluid dynamics calculations to maximize airflow.  Maximizing airflow requires minimizing vortex formation, and minimizing interstage air leakage along the shaft and casing.  Thus everything inside is machined with very fine clearances for high efficiency.  For these reasons the blades are machined with pretty tight tolerances, and normally the airflow in the machine is in line with what the software calculated.

Inlet icing throws all that fluid modeling out the window, because if the blades are covered in ice, airflow will be erratic.   

This machine may have experienced ice buildup on the stationary compressor blades in the first row. The ice caused the air to flow unevenly onto the the rotating blades.  Each time a rotating blade passed a stationary blade with ice on it, turbulent/uneven airflow would cause a rocking motion as the rotating blade was buffeted by varying amounts of air.  This would actually be a very high speed vibration, since the rotor spins at 60 times per second.  The vibration created metal fatigue at the base of the blade (like bending a coat hangar over and over until it snaps in two) and eventually the blade came loose from the rotor. 

With the rotor spinning at 60 revolutions per second, the blade had plenty of centrifugal (centripetal) stress to shear it off, once the base had gathered enough cyclical fatigue to create a crack.

What followed after the compressor blade released compounded the failure.  It was pulled through the machine by the air stream and impacted most (perhaps all) of the other blades.  What a mess.  Moral of the story:  Don't allow the inlet to ice up.

Below: A new, clean compressor.  Inlet to the left side, discharge is near the circular flange to the right.

Below:  A severely damaged compressor.  It appears that all of this damage was caused by a single blade passing through the compressor.  (Inlet is to the left side, so the loose blade came from that row).  None of the blades on this rotor escaped harm.  It is not clear if the combustor section or turbine section were also damaged.

 

With the rotor removed, you can see that most of the stationary blades are also damaged.  The case will undoubtedly need some repairs as well.  Sad sad stuff here.