Recovering from a wreck

Darayus Pardivala, pres­ident, Sulzer Turbo Ser­vices, took a call in his La Porte (Tex) office last sum­mer from a North American custom­er that had recently suffered damage to one of its 501FD2 gas turbines. The caller, who asked that neither he nor his company or plant be men­tioned publicly, said an R1 turbine blade fractured and traveled down­stream. He was afraid that the rotor might have been severely damaged in the accident and wanted Pardiva­la’s team to assess its condition and report its findings within 10 days.

The owner removed all compres­sor and turbine blades and shipped the rotor to Sulzer. The catastroph­ic failure trig­gered a period of high vibra­tion before the unit tripped and the rotor stopped. Some physical dam­age was seen by experts as they turned the shaft slowly in a lathe upon receipt (1). For example, seal areas and journals were scratched (2). Significant runout also was in evidence—greatest at the midspan between the journals, as expected. The rotor was bowed as well.

International Sales Manager Jaime Valdez and Elmer Williams, superintendent of what’s known by employees as the Big Bay, told the editors the first thing done was to conduct a thorough visual inspection and photograph the findings. Part of the exploratory work involved put­ting the shaft on turning gear and rotating it at 2 to 3 rpm for three or four hours to “stabilize” the rotor. “Stability” is achieved when the phase angle and amplitude recorded by the vibration monitor don’t change significantly over time.

Next step: Determine runout. It was 18 mils at the shaft center—six times the OEM’s maximum allowable of 3 mils. Williams said such a large runout on an F-class machine could never be corrected with balance weights. After inspection and cleaning, the cus­tomer authorized Sulzer to bring the rotor back into spec.

To achieve that goal, the rotor had to be dismantled. Most users with large frame experience might think the overhaul would begin by break­ing the marriage coupling to discon­nect the turbine and compressor sec­tions. It does for many frames, but not the 501F, which does not have a marriage coupling.

The way you dismantle this rotor, Valdez and Williams said, is to stand it up (3), remove the nuts from the 10 turbine through-bolts, and lift off the four wheels one at a time—R4, R3, R2, and R1. However, before remov­ing the nuts, technicians should “mike” the bolts and record stretch data to determine if there’s a prob­lem caused by uneven tension or a cracked bolt.

The component you see after removing the first-stage turbine wheel (4) is the curvic clutch adapt­er, which resides inside the air-sepa­rator sleeve (5, 6). Note the accumu­lation of rust in the first-stage wheel in (4). The flange at the bottom of the air-separator sleeve bolts to the torque tube, which, in turn is bolted to the compressor section (5).

Viewing the components indepen­dently offers supporting perspective. The curvic clutch adapter is shown alone in (7). It sits on top of the torque tube (8) and the air separator sleeve slides over both and bolts to the flange on the torque tube.

The air separator sleeve and cur­vic clutch adapter removed, the com­pressor section, with torque tube attached, was checked for runout (9). The result: severe runout (7 mils). Next, Williams’ crew removed the compressor section from the lathe and positioned it vertically so the individual disks could be removed. The 12 through bolts in the compressor section all were checked for stretch to verify their integ­integ­rity and see if they were a source of runout.

Rows 4 through 16 are character­ized by individual disks and they were removed (10); the first three stages are integral with the stub shaft (11). The contact faces of some disks were found slightly damaged from the inertial shock of the acci­dent. Disks and other parts were cleaned with a glass-bead blast and put through a rigorous NDE regimen. Magnetic particle inspection was used on the wheels because the entire rotor is made of carbon steel.

Results: Compressor section fine, turbine section fine, male/female rabbet fits between compressor disks fine, all bolt and dowel holes were measured and found within spec. There was some variation in quality among the through-bolts, but it was not sufficient to cause the runout. In fact, all of the through-bolts and their nuts were reused. Only the disk faces were off-spec. However, there was no fretting between the faces; disks just were slightly out of parallel.

  • s1
  • s2
  • s3
  • s4
  • s5
  • s6
  • s7
  • s8
  • s9
  • s9a
  • s9b
  • s9c
  • s9d
  • s9e

After the accident, the rotor came to a stop when hot and it was not possible to engage the turning gear. Thus, at least some of the dam­age was caused by the stationary cool-down. Face displacement measured a maximum of five tenths (0.0005 in.). A proprietary software program was used to match the highs on one disk with the lows on the adjacent one and set them accordingly. In the end, only a skim grind between the integral third stage and the fourth stage was needed to achieve success.

The compressor section was stacked, through-bolts stretched to spec, and returned to the lathe for what Williams called a “true check” runout measurement. Some runout was identified at the torque tube face. A skim cut fixed that and the final runout came in at less than the 1.5-mil limit established by the OEM.

The compressor was removed from the lathe and positioned verti­cally to receive the torque tube, cur­vic clutch adapter, and air separator sleeve (12). The curvic teeth are ground into each turbine wheel forg­ing to center and hold the stages in place. All teeth on each wheel were blue-checked to assure a proper con­tact pattern and no cracking or fret­ting. None found.

Once all wheels were installed, (13) and through-bolts stretched, the completed rotor was moved to a lathe (14) for a four-hour stabilizing roll. Then runout was checked at less than 2 mils. A slow-speed balance was next, and journals and seal areas were dressed.

Valdez said the customer poured new babbit to assure proper bearing clearances and also took responsibil­ity for installing new turbine and compressor blades onsite. Field tests were successful. Actual shop time for inspection and repair was less than 60 days on a routine repair basis. ccj