The lessons from this case study have implications for any operation that has dedicated emergency standby equipment.

Standby engines that have a high ratio of thermal cycles versus operating hours could experience degradation in their bolted joints due to gasket creep and relaxation effects before they are to be replaced according to an operating-hours-based maintenance schedule.

Most maintenance plans are based on the expectation of wear and tear that occurs when equipment is operated. This is especially true of electric motors, pumps, engines, compressors and production line machines that are regularly operated. Consequently, most preventive maintenance schedules are based upon hours of usage.

What happens, however, when a piece of equipment like an engine-generator set is dedicated to an emergency standby mission? In such an application, the equipment may only be operated occasionally. In fact, it may not even be operated enough to warrant a regular overhaul during the life of the installation. Are there unexpected maintenance problems that can occur when equipment is not used very much?

During the monthly scheduled test of a standby, emergency diesel-generator set in September 2009, the mechanical overspeed governor was observed to be vibrating significantly. Dismantling the governor drive-unit and its attached components revealed that the flange connection was loose where the unit attached to the engine block with studs and nuts (see top image)—all eight nuts were loose. Oil was also pooled on the I-beam directly below the rear engine cover. The oil appeared to originate from the engine governor drive-unit flange.

Summary of investigation findings
A review of historical work documents found no indications that maintenance had been performed on the overspeed governor gearbox drive-unit connection. The flanged connection had not been disassembled since it was initially assembled and installed by the original equipment manufacturer (OEM) in 1974. The gasket used in the bolted joint was the original gasket. In response to inquiries, the OEM provided the following information:

• The original torque requirement for the 5/8" diameter NC studs and nuts was 50 ft-lbf.
• The expected service life of the gasket would be the same as the engine, about 40 years based on operation hours, provided that the proper nut tension is maintained and the joint is not disturbed.
• The usual degradation mechanisms for this gasket material are joint looseness or possibly long-term exposure to water.
• With respect to periodic maintenance for the gearbox-to-cam-house connection, the vendor recommended that nut tightness be checked if oil seepage is noted.
• The gasket was 1/64" compressed asbestos fiber with SBR, Buna-S binder.

The flange faces had no surface problems or deficiencies. The studs were not loose in the flange. No deficiencies were found with the tap-end stud washers. As shown in Fig. 2, three of the eight studs were noted to have abrasion marks on the threads. The governor overspeed housing had loosened sufficiently to drop down and "ride" back and forth on the studs.

The internal gearing of the overspeed governor drive-unit was examined. No unusual wear or damage was noted on the bevel gear teeth (Fig. 3), shafts and bearings—nor were problems noted in the governor itself.

Similarly, engine-block vibrations measured in the preceding two years indicated that no significant vibrations originated elsewhere in the engine—except those originating most recently in the loose overspeed governor and its cable mounting bracket. Witnesses to previous engine tests indicated that no similarly high vibrations had been noticed.

Prior to the event, the engine had 3513 hours of actual operating time and 1566 starts. This is an average of 2.24 hours of running time for each start. The oil temperature when the diesel-generator operates is 165 F. When the engine is in standby, the temperature of the oil is about 101 F. Thermal cycling of the gasket in question, therefore, was estimated to range from 101 F degrees during standby, to 165 F degrees during operation.

Because the overspeed governor trip-unit is located in a generally inaccessible area, there was no suspicion that someone had surreptitiously loosened the nuts. There were no specific preventive maintenance items for changing gaskets or checking nut torque in the manual. Gaskets were normally changed and nuts re-tightened when a part was removed and replaced due to normal-service time maintenance-inspection schedules.

A review of plant condition reports, though, found that there had been unrecognized precursors.

• Reports in March and April 2006 reported that the overspeed governor was leaking oil.
• A report in January 2008 documented that the oil sight glass on the overspeed governor fell off during a test. When the sight glass was examined, it had a fatigue fracture.
• A report in December 2008 documented that both of the diesel-generator intake-manifold butterfly-valve mechanical overspeed trip cables failed to meet pull-test requirements. The cables were replaced.

Analysis
An assessment determined that the following occurred:

• The overspeed governor drive trip flange-to-flange bolted joint loosened over time due to gasket creep and relaxation effects in response to the high number of engine thermal cycles.
• When the bolted joint on the overspeed governor began to loosen, oil leaked by the gasket.
• When the joint loosened more, vibrations increased and caused damage to the overspeed trip cables. The vibrations may have also caused some damage to the gasket due to internal fretting between the gasket and flange surfaces.
• When the nuts loosened to the point that they were "finger tight" or less, engine vibrations were sufficient to excite the overspeed governor assembly and visibly wobble back and forth.

Using the available information, the graph shown in Fig. 4 was generated to show the approximated loss of bolt load versus time. Inspection of the graph shows that the reduction in stud stress to 20,000 psi occurred after about eight years of service. The overspeed governor drive trip housing, though, functioned well until stud stress was below 5000 psi, or below a torque value of about 8.3 ft-lbf.

Discussion
The four-stroke, V-16, turbo-charged, 5000-kilowatt diesel engine used for the standby emergency diesel generator system (one of two such units at the site) is typically deployed in ships, locomotives and stationary power plants where it is run either in continuous or in semi-continuous service. The critical parts are the main engine bearings. When the main bearings are worn, the engine is dismantled and overhauled. A typical overhaul period is perhaps 25,000 to 50,000 operating hours. On a continuous operating basis, this is equivalent to 2.9 to 5.7 years.

During an overhaul, it is normal to disassemble the unit, inspect items that wear, and change out gaskets that are often damaged during disassembly—then reassemble the unit. This includes re-bolting gasket joints to the specified torque specifications. In other words, when the engines are operated as designed on a continuous or semi-continuous basis, most gaskets are replaced and their nuts are re-torqued perhaps every three to six years.

This standby emergency diesel-generator, however, had historically been operated about 100 hours per year and thermally cycled about 45 times per year. On a continuous basis, the engine had run the equivalent of about four days per year for 35 years. At this rate, it would require 214 more years for the engine to reach the typical lower overhaul period of 25,000 operating hours. In short, degradation was not occurring because of operating wear. The engine had hardly been run at all.

The phenomenon of gasket creep and relaxation—which primarily affects soft-gasket bolted joints—is a combination of several effects.

An initial relaxation of bolt-load stress occurs directly after the bolt and nut are tightened due to plastic flow of the soft gasket material under compression and embedment of the joint components. Because the gasket is not constrained at the ends, it can squeeze out the sides of the flanges in response to compression. This is the primary reason why many bolting specifications require that a second tightening be done four or more hours after the initial tightening.

After the initial tightening of the joint, two other time-dependent effects come into play. One is cyclic stress loading that occurs during operation, usually due to vibrations. These resulting alternating stresses—if high enough—cause soft gaskets to plastically flow. Because of hysteresis effects, the gasket never fully regains its original position after each cycle. It slowly flows out the ends of the flanges and gradually becomes thinner between the flanges, which causes the initial bolt pre-load stress to diminish over time.

In this particular case, though, vibration was initially not a significant factor. If it had been, the joint would have loosened right away.

The measured amplitude of engine vibrations in the area of the overspeed governor was not high enough and the duration was not long enough to cause an immediate problem. This is why the joint performed well for 32 years.

A second effect is thermal cycling, wherein expansion and contraction affect the thickness of the gasket. Alternate heating and cooling of the joint causes both the bolts and gasket material to alternately expand and contract. The net effect is similar to that due to alternating stresses due to vibrations. Again, in an alternating expansion and contraction cycle, hysteresis effects cause the gasket material to not fully regain its original dimensions, and the gasket becomes thinner.

When there are enough thermal cycles to cause the load stress to sufficiently decrease, engine vibration effects then become significant. As the torque on the nut decreases, a threshold will be reached where the alternating stresses induced by engine vibrations are significant as compared to the remaining load on the stud or bolt. When this occurs, engine vibration effects cause the nut to loosen and back off.

The graph in Fig. 4 indicates that the pre-load stresses in the studs were perhaps 3000 psi when seepage of oil was first observed. This occurred after the engine had undergone over 1400 thermal cycles—this is a high number of thermal cycles. If the engine had been regularly run eight hours per day, for example, the engine would have already been overhauled and the problem would not have occurred.

A review of the preventive maintenance tasks for the engine found that they were primarily based upon operating time, as recommended in the manual provided by the OEM. To correct the condition, a program to check all bolted joints on the emergency standby engines was initiated. MT

---

Popular contributor Randall Noon is a root-cause team leader at Cooper Nuclear Station. A licensed professional engineer, he has been investigating failures for more than 30 years. He is the author of several articles and texts, including: The Engineering Analysis of Fires and Explosions; Forensic Engineering Investigations; and, most recently, Scientific Method: Applications in Failure Investigation and Forensic Science (CRC Press), a chapter of which was excerpted in the June 2009 issue of Maintenance Technology. E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

References

1. EPRI Technical Report 1010639, dated January 2006, Non-Class 1 Mechanical Implementation Guideline and Mechanical Tools, Revision 4. This technical report contains an excellent description of gasket creep and relaxation, and how bolts loosen due to vibration.
2. "An Experimental Investigation of the Factors that Contribute to the Creep-Relaxation of Compressed Non-Asbestos Gaskets," by Jose Veiga, Carlos Cipolatti, Ana Sousa and David Reeves, published in the Proceedings of the ASME 2007 Pressure Vessel and Piping Conference, July 22-26, 2007, San Antonio, TX. This paper, while primarily discussing non-asbestos gaskets which were used in the testing, nonetheless provides an excellent description of the phenomenon of gasket creep and relaxation, especially in response to thermal cycles.
3. Machine Design: An Integrated Approach, by Robert Norton, Prentice-Hall Inc., 1998, pgs. 914-939. This text contains a detailed explanation of how to perform a standard calculation to determine pre-load in bolted joints.
4. "Heat Exchanger Gaskets Radial Shear Testing," by Jose Veiga, Nelson Kavanagh and David Reeves, Proceedings of PVP2008-61121, 2008 ASME Pressure Vessel and Piping Conference, July 27-31, 2008, Chicago, IL. This reference is noteworthy because it provides a qualitative sense of how thermal cycling affects the preload on a bolted joint in several types of gasket materials.
5. "Advantages of Shortening Overhaul Periods," by Randall Noon, ASME Paper 79-WA/PEM-1, Winter Annual Meeting, December 2-7, 1979. This paper notes typical overhaul periods in stationary engines, and why it is sometimes better to overhaul engines before their normal turnaround time.

EDITOR'S NOTE: Gaskets played a big role in the root cause of the problem referenced in this article. To learn more about these important components and how their application (or misapplication) can affect equipment reliability in your own operations, check out the Pre-Conference Workshop "Best Practices in Compression Packing and Gasketing" that will be offered at MARTS 2010. Sponsored and presented by member companies of the Fluid Sealing Association (FSA), this Workshop promises to be an invaluable learning experience for anyone involved in the selection, installation and/or maintenance of these types of sealing solutions. To register, go to www.MARTSconference.com