Fretting Fatigue of Slot-dovetails in Turbo-generator Rotor

Fretting Fatigue of Slot-dovetails in Turbo-generator Rotor (From O&M Issues Discussed in Recent EPRI Meetings) H. Ito Toshiba Corporation 1-1-1, Shibaura, Minato-Ku, Tokyo, 105-8001 Japan Abstract-This paper describes the fretting fatigue of slotdovetails in turbo-generator rotor including typical examples and repairs of fretting fatigue cracks, results of fretting fatigue tests, factors affecting fretting fatigue strength, fretting fatigue preventive technologies, and UT inspection methods. I INTRODUCTION In the 70s, two 660 MW turbo-generators in Europe experienced shaft cracking due to fretting fatigue of the slotdovetails. The authors started the basic study of fretting fatigue of the shaft slot-dovetails in the late 70s. Inspection results showed that generator shafts, manufactured by the authors, also cracked in the slot-dovetails. High-speed fatigue test equipment was introduced to perform comprehensive fretting fatigue tests simulating actual machines. The authors succeeded in quantitatively defining dominant factors affecting fretting fatigue strength of slot-dovetails of turbo-generator rotor. A fretting fatigue preventive technology was established for the slot-dovetails based on the test results and review. A nondestructive inspection technology using UT was also developed. The fretting fatigue preventive technology has adopted in the new machines as a standard design application. It is also used in the installed machines as refurbishment. The machines with applied by these technologies are successfully operating for more than 20 years since application. This paper introduces examples and repairs of slot-dovetail fretting fatigue cracks, results of fretting fatigue tests, factors affecting fretting fatigue strength, fretting fatigue preventive technologies, UT inspection technology, and application of fretting fatigue preventive technology to the installed machines. II EXPERIENCES OF INSPECTION, CRACKED ROTOR SLOT- DOVETAIL AND REPAIR Fig. 1 shows experience of inspections and observed cracked slot-dovetails since 1979 by year. Nine out of 601 inspected rotors were found to have cracks. Table 1 shows operating history, description of cracks and repairs, and countermeasures. As a severe example, machine A* is described below in detail including crack profile, repair and countermeasure. Number of Inspected Rotor Unit 40 35 30 25 20 15 10 5 0 ; Cracked Rotor Slot-dovetail 1980 1985 1990 1995 2000 Inspection Year Fig. 1 Experience of Inspection and Cracked Rotor Slot-dovetail Year of Inspection Table 1. Experiences of Cracked Rotor Slot-dovetail Out-put (MW) Year in service A 1981 220 1963 B 1982 350 1972 C 1982 250 1971 D 1982 500 1968 E 1983 600 1973 F 1987 350 1972 G 1990 500 1973 A* 1990 220 1963 H 1991 375 1973 Description of Cracking Crack size: L10 x D3.2 mm Crack position: About the core center No. of cracks: 1 Wedge hardness: Equal to shaft Operation hours: 100,000h No. of starts and stops: 766 Crack size (max.): L3.5 x D0.98 mm Crack position: About the core center No. of cracks: 10 Wedge hardness: Harder than shaft Crack size: L2.5 mm Crack position: About the core center No. of cracks: 1 Crack size (max.): L6.6 x 1.5 mm No. cracks: 55 Wedge hardness: Equal to or harder than shaft Crack size: L4.5 x D1.4 mm No. of cracks: 9 Wedge hardness: Equal to shaft Crack size: L4.0 mm No. of cracks: 3 Wedge hardness: Shaft is harder Crack size: L4.0 mm No. of cracks: 28 Wedge hardness: Shaft is harder Crack size: L23 x D8.5 mm No. of cracks: 2 Crack size: L5.5 mm No. of cracks: 3 Repair / Modification Clearing of Cracking with Grainder. Make taper with radius at end of wedge shoulder. Replace with aluminum alloy wedges. Change wedge length and arrangement. Two cracks were observed at the core center of the slots by the pole on the joint of steel wedges. A larger one measured 23 mm in length and 8.5 mm in depth. Fig. 2 shows the crack in detail. The cracks were ground off to suitable shapes to prevent stress concentration and the surface was smoothly finished (Fig. 3). Wedges were bridged over the ground off area. The wedge material was changed from steel to aluminum alloy. Concentration of contact pressure was avoided by increasing the corner radius on the area where the wedge edge contacts the slot dovetails. The unit is running for 10 years since then without problems. 1

10 8.5 8 5 Cracking equipment. The specimen is made of Ni-Cr-Mo-V steel, which is the same as the shaft material. The pad is made of S55C (quenched and tempered) (AISI-1055), which is the same as the material of steel wedge. Table 2 and 3 show chemistry and mechanical properties of materials of specimen and pad, respectively. Fig 2. Detailed Cracked Rotor Slot-dovetail of A* Unit (10) (16) Fig. 4 Schematic Illustration of Fretting Fatigue Test Method Fig 3. Repair and Correction for A* Unit III EXPERIMENTAL STUDIES Fretting fatigue of turbine and turbo-generator rotors have been studied by many researchers for many years. Areas of concern are blade roots and coupling-to-shaft shrinkage fit. The conventional fretting fatigue tests used only one pad and adjacent wedges such as in the slot-dovetails of a turbogenerator shaft were not simulated. We developed a high-speed fatigue testing equipment incorporating two double-pads that simulate actual arrangements of wedges in slot-dovetails. We quantitatively evaluated factors affecting fretting fatigue strength of slot-dovetails. The evaluated factors included proximity effect of pads (wedges), contact pressure, relative slippage (wedge length), pad (wedge) materials (hardness, Young s modulus, rigidity) and repeated stress (shaft bending stress). We studied the fretting fatigue preventive measures and fatigue life evaluation method for slot-dovetails based on the test results and review. Test Equipment and Method Fig. 4 shows a schematic test arrangement of the fretting fatigue testing system. Fig. 4-(a) and 4-(b) are conventional testing systems. Fig. 4-(c) is the double-pad type test system, which is newly devised to define the proximity effects of wedges. Fig. 5 shows a specimen and a pad. Fig. 6 illustrates the testing equipment. Fig. 7 is a photograph of the test Fig. 5 Specimen and Pad Fig. 6 Fretting Fatigue Test Equipment 2

Specimen Pad Fig. 8 Specimen and Pad for Pad Materials Effect Test Fig. 7 Photograph of Setup of Fretting Fatigue Testing Table 2 Chemical Composition of Specimen and Pad C Si Mn P S Ni Cr Mo V Speci 0.24 0.10 0.32 0.00 0.00 3.71 1.60 0.27 0.12 men 7 5 Pad 0.56 0.21 0.75 0.01 0.01 - - - - Table 3 Mechanical Properties of Specimen and Pad Reduction 0.2% proof Tensile Elongation Hardness of area stress MPa stress MPa % Hv (1.0kg) % Specimen 790 895 25.3 71.3 269 Pad 634(*) 811 19.5 39.6 265 (*) Yield stress Fretting fatigue tests were conducted in the atmospheric air at room temperature with load control of pulsating tension at 15 Hz. Pad clearance C was 0.25 and 4 mm and infinity (single pad). Nominal contact pressure was 50, 100 and 200(equivalent to actual machines) MPa. Relative slippage of the pad edge was 5 to 6 µm (equivalent to actual machines). Fatigue tests to evaluate the effect of pad materials were conducted using the specimens and pads shown in Fig. 8 and the testing equipment shown in Fig. 9. The specimen materials were Ni-Cr-Mo-V steels or the same as the above specimens. The pad materials tested were 5 each ferrous and nonferrous materials as shown in Table 4. Fatigue tests were conducted in the atmospheric air at room temperature with load control of pulsating tension at 100 Hz. Nominal contact pressure was 200 MPa constant. Fig. 9 Test Equipment for Pad Materials Effect Test Table 4 Mechanical Properties of Pads for Pad Materials Effect Test Contact pad σ 0.2 σ u δ ϕ E σ wf Hv μ material MPa MPa % % GPa MPa 3.5NiCrMoV-QT 790 895 25.3 71.3 269 205 0.7 75 3.5NiCrMoV-Q 1103 1660 13.7 46.4 403 207 0.7 80 S55C-QT 561 933 21.5 45 255 201 0.68 75 S25C-N 231 430 27.7 58.1 195 201 0.7 90 SUS304 247 663 69.9 81 200 190 0.7 80 Cu-Be-Ni 507 760 15.5 31.5 239 130 0.55 100 Cu-Cr 380 450 26 51.2 150 120 0.65 130 Cu 245 264 30 85 102 110 0.65 140 A2024-T351 296 477 13.7 17.2 141 75 0.7 130 Ti-6Al-4V 941 1035 15 35.5 365 115 0.68 100 σ 0.2 = 0.2% proof Hv = Hardness σ u = UTS E = Young s modulus δ =Elongation μ = Coefficient of friction ϕ = Reduction of area σ wf = Fretting fatigue limit (amplitude at 2x10 7 ) Experimental Results and Review Fig. 10 shows the effect of pad-to-pad clearance on S-N curve. N represents the number of repetitions at the initiation of a 3

crack. Fig. 11 indicates the relationship between pad-to-pad clearance and fretting fatigue limit. As Fig. 11 shows, fatigue limit decreases sharply with the decreasing pad-to-pad clearance. Fretting fatigue limit by single pad was 40 MPa, which is 1/8.8 of the ordinary fatigue limit of 350 MPa. Fig. 12 shows the initiation and the propagation of cracks. A number of small cracks(a) were observed on the pad edges at the initial stage of fatigue. Along with the increase in the number of repetitions, the pad edges were worn out, and then the major cracks(b) initiated and propagated at the inside of the pad edges. C=0.25mm, σa=15mpa Fig. 12 Fretting Fatigue Cracks Observed in Specimen Fig. 13 shows the effect of contact pressure on S-N curve. Fig. 14 indicates the relationship between fretting fatigue limit and contact pressure. For the contact pressure up to 100 MPa, fretting fatigue limit decreases rapidly with increasing contact pressure but for the contact pressure above 100 MPa, decrease in fretting fatigue limit is very small. Fig. 10 Effect of Pad-to-Pad Clearance on S-N Curves Fig. 13 Effect of Contact Pressure on S-N Curves Fig. 11 Relationship between Pad-to-Pad Clearance and Fretting Fatigue Limit 4

Fig. 14 Relationship between Contact Pressure and Fretting Fatigue Limit Fig. 16 Effect of Pad Materials (Ferrous) on S-N Curves Fig. 15 illustrates the effect of pad contact length on S-N curve. Fretting fatigue life decreases with increasing contact length. Fig. 17 Effect of Pad Materials (Non-Ferrous) on S-N Curves Fig. 15 Effect of Contact Length of Pad on S-N Curves Figs. 16 and 17 show the result of fretting fatigue tests of various pad materials. Fig. 18 shows the relationship between fretting fatigue limit and hardness of pad materials. Generally, fretting fatigue limit decreases rapidly with increasing hardness of pad materials. In the high hardness region, however, pad materials of the same hardness do not have an identical fretting fatigue limit. Young s modulus seems to be another factors affecting fretting fatigue limit in addition to hardness. Fig. 18 Relationship Between Fretting Fatigue Limit and Pad Hardness 5

IV IMPROVEMENTS IN DESIGN AND PRODUCTION Fretting fatigue preventive technologies have been established based on the results and review of the extensive fretting fatigue tests simulating actual machines as described in Chapter III. The preventive technologies are summarized below. (1) Fretting fatigue strength decreases rapidly with decreasing pad-to-pad clearance. As a countermeasure, the area of the wedge edge that contacts slot-dovetails was tapered to practically increase the wedge-to-wedge clearance and the corner radius of the wedge edge was enlarged to prevent the concentration of contact pressure. (2) Fatigue life decreases with increasing wedge contact length. As a countermeasure, several different wedge lengths in use were unified into the shortest length. Furthermore, long wedges of a whole length or several split lengths were adopted with wedge joints to avoid meeting the rotor core center. (3) Fatigue life decreases with increasing hardness, rigidity and Young modulus of the wedge. As a countermeasure, wedge materials of a lower hardness than shaft materials were selected when using ferrous wedges. Analysis of main flux distribution indicated that no operational problem occurs when using nonmagnetic wedges in the slot by the pole. In view of this finding, aluminum alloy wedges may be used in all slots in some cases. Residual stress in the shaft is one of the known factors affecting initiation and propagation of fretting fatigue cracks in the slot-dovetails. Residual stress originates at the stage of shaft forging production and in the process of slots making. The former is controlled to a level of several kg/mm2 or below by improving steel making process in the shaft forging production or by stress relieving annealing after steel making. For the latter, thickness of the residual stress layer is controlled at several tens µm or less by using carbide cutter and improving the slots making process. V INSPECTION AND REFURBISHMENT Inspection Fretting fatigue cracks in the slot-dovetails are easily detected by magnetic particle or eddy current inspection when the wedges are removed. When the wedges are in place into slots, ultrasonic tests are used but testing accuracy with standard probes is often insufficient because of the short distance between the rotor surface (inspection surface) and the crack and inclination of slot-dovetails relative to the rotor surface. We developed a special probe featuring an optimum angle of refraction and frequency. Model teeth simulating actual shaft teeth were prepared and artificially flaws to confirm the accuracy of detection. It was confirmed that the probe detects a minimum of 1.0mm deep defect. Fig. 19 shows the special probe and the test using the model slot-dovetail. We recommend UT inspection of slot-dovetails at major outage (with the rotor taken out). 361 turbo-generator rotors,as retaining rings and wedges installed, have been inspected by UT inspection at site so far. Fig. 19 UT Probe for Shaft-Dovetail Refurbishment Permanent countermeasures against fretting fatigue cracking introduced in Chapter IV were applied to approximately 150 installed generators so far. No crack event is reported, in the follow-up research, since early 1990s when the refurbishment was almost completed. This means that the validity of prevention measures has been confirmed. VI CONCLUSION In the 1970s, it was reported that generator shafts were cracked due to fretting fatigue in the slot-dovetails. We studied the countermeasures and have developed fretting fatigue preventive technologies and UT inspection technology. These technologies were applied to new and existing machines as a standard design application and refurbishment. The machines are running for more than 20 years successfully since then. Thermal power turbo-generators are, however, running under increasingly diverse and severe operating conditions in recent years. It is not rare that units are operated over the original design conditions. Fretting fatigue life of slot-dovetails may be affected also by operating conditions. We will keep close contact with users to prevent fatigue damage and contribute to stable power supply. 6