Keep ‘em spinning

Industrial plants throughout the world rely on critical fans for operation, but what happens when a fan rotor, rotating at a peripheral speed of 450 miles per hour, fractures?

The scene looks something like this: Pieces of the wheel rip away from the rotor, tear through the fan housing and fly off. The steel shaft bends and twists like a pretzel. The cast iron bearing housings break apart. The motor separates from its pedestal and lands on the foundation. The plant is shut down, and it takes weeks to clean up the mess, rebuild the fan housing and install a spare rotor (assuming the plant has one on hand).

This scenario has been played out in heavy industrial facilities worldwide. What are the causes of this situation? What steps can maintenance personnel and engineers take to minimize the chances of such a catastrophic failure?

Before we address what you can do to prevent fan failure, consider the real-world examples below when making a case for preventive action.

Coal dryer induced-draft fan

A 96 in.-diameter centrifugal fan had been in service for about three years on a coal dryer application in West Virginia. The fan operated at 1,180 rpm with a peripheral speed of 337 mph. Although there was an upstream cyclone separator to provide gas cleaning, some coal dust continually passed through the rotor. Erosion-resistant liners protected the fan wheel, but small particles of coal dust entrained in the high-velocity gas stream slowly eroded the unprotected welds at the edges of those liners. There was no planned inspection of the rotor and no routine monitoring of the bearing vibration. One morning, minutes after fan startup, the rotor ripped apart, completely destroying the fan and damaging nearby equipment. Fortunately, no operators were in the area. The plant was shut down for nearly three weeks while a new fan was fabricated and installed on an emergency basis (Figures 1 and 2).

Figure 1

This is the impeller from coal dryer fan following a catastrophic failure.

Foundry gas-cleaning induced-draft fan

This 88 in.-diameter, double-inlet, variable-speed centrifugal fan was operating at about 350°F. The operators noticed some increase in the noise and vibration at certain speeds, but once the fan changed speeds, the vibration seemed to subside a little, easing their concerns. The fan had no bearing temperature or vibration-monitoring equipment installed, and periodic vibration monitoring was spotty. The blades generated a pressure pulse when they passed the “cutoff” of the scroll-type centrifugal fan housing. The operators didn’t realize that at certain speeds, the fan’s blade passage frequency (and the resulting frequency of pressure pulsations) caused excitations that exactly matched the natural frequencies of some of the fan rotor components. The result was higher-than-expected deflection and stress levels. This caused high-cycle fatigue cracking of the rotor material near the toe of several fillet welds. Left undetected, the cracks propagated until they reached a critical crack length. Without warning, the crack growth rate increased dramatically, and the fan wheel flew apart (Figure 3).

Figure 2

Rotor failure damaged the fan housing and bearings.

These two examples demonstrate the significant effects that axial and centrifugal fan rotor failure can have on plant operation and worker safety. In addition to the obvious damage and plant downtime, similar situations have resulted in severe injury — and even death — of workers near the fan when it failed. So what can maintenance personnel do to protect operations and workers from catastrophic failures?

Continuous monitoring is your best protection

Figure 3

This 88-in. centrifugal fan shows the results of a rotor fatigue crack failure.

Preventive and predictive maintenance using frequent visual inspections, nondestructive testing of welds and periodic measurement of bearing vibration help, but these techniques alone can’t effectively prevent sudden and catastrophic failures like those described above. The best approach is continuous monitoring of vibration and bearing temperatures with the instrument signals linked to a fast-response automatic fan shutdown control. Here’s why:

• If erosion is a possibility, thinning structural components subjected to the high stress resulting from centrifugal forces can suddenly reach a failure point.
• Once fatigue cracks have been initiated, they can propagate faster and faster until there is a sudden, unpredicted failure.

Routine visual inspections, including nondestructive testing aids such as magnetic-particle, dye-penetrant or fluorescent-particle techniques, are helpful in establishing the presence of cracks, but these inspections can’t be done continuously during operation.

Similarly, predictive maintenance vibration monitoring is effective for establishing a baseline and determining dangerous trends (or step-changes) in vibration levels, but it can’t always protect against the risk of sudden and catastrophic failure.

Monitoring fans with rolling-element bearings