Fatigue Cracking of Gas Turbine Diaphragms

N.A. Fleck, Dow Chemical U.S.A.


From: Handbook of Case Histories in Failure Analysis, Vol 2, K.A. Esakul, Ed., ASM International, 1992

Abstract: Several compressor diaphragms from five gas turbines cracked after a short time in service. The vanes were constructed of type 403 stainless steel, and welding was performed using type 309L austenitic stainless steel filler metal. The fractures originated in the weld heat-affected zones of inner and outer shrouds. A complete metallurgical analysis was conducted to determine the cause of failure. It was concluded that the diaphragms had failed by fatigue. Analysis suggests that the welds contained high residual stresses and had not been properly stress relieved. Improper welding techniques may have also contributed to the failures. Use of proper welding techniques, including appropriate prewelding and postwelding heat treatments, was recommended.

Keywords: Compressors; Welded joints; Welding parameters

Material: 403 (Martensitic stainless steel), UNS S40300

Failure types: Joining-related failures; Fatigue fracture


Background

Several compressor diaphragms from five gas turbines cracked after a short time in service. The vanes were made from type 403 martensitic stainless steel. Welding was performed using type 309L austenitic stainless steel filler metal.

Applications

The compressor is part of the combustion system in the gas turbine. It compresses air between the rotating blades and the stationary diaphragms and then feeds the air to the combustion system. The combustion system, after receiving the air, mixes in fuel and burns the mixture in order to raise the energy level that is then discharged to the turbine.
The stationary vanes in the compressor section are assembled into units called diaphragms (Fig. 1). The individual vanes are held in position in the diaphragm by being welded at each end into shrouds. The shrouds extend circumferentially around the inside wall of the easing, where they are supported in machined grooves. Each stage of the vane is split into two 180° shroud segments for easy access to the compressor.
Figure 1

Fig. 1  Compressor diaphragms removed from gas turbine

Circumstances leading to failure

The gas turbines were undergoing routine maintenance and part replacement (hot gas paths) when cracks were discovered on the compressor diaphragms (Fig. 2).
Figure 2

Fig. 2  Diaphragm, showing fracture surface after a vane was removed and two vanes with crack indications

Pertinent specifications

The specifications for manufacturing the compressor diaphragms were unknown. The parts were made by the original equipment manufacturer (OEM).

Specimen selection

Most of the cracks were located directly below the shroud/blade seal weld. Several welds, along with their respective fracture faces, were removed and prepared for examination.

Visual Examination of General Physical Features

Figures 3 and 4 show the locations where the cracks occurred most frequently. Visual examination showed evidence of poor welding craftsmanship. Some of the welds appeared irregular and sloppy; others appeared to have been made without use of a filler metal.
Figure 3

Fig. 3  Typical crack location at outer shroud seal weld

Figure 4

Fig. 4  Crack locations on inner shroud after removal of seal box

Testing Procedure and Results

Nondestructive evaluation

When cracks were discovered visually, the diaphragms were removed from the turbine and inspected using the techniques described below.
Liquid Penetrant Testing. Each diaphragm was cleaned of grease and dirt. The diaphragms were set securely in a rack, with the splits facing up. Several ounces of water-washable red dye penetrant were poured inside the openings on each end. The diaphragms were completely filled with water, and all vane welds were covered; the outer sections of the diaphragms were kept dry. After a minimum of 8 h, developer was sprayed over the outer shrouds and at the vane-to-shroud intersections. Most of the time, the cracked vane-to-shroud welds could be determined visually, because the penetrant would leak out after only a few minutes. All areas displaying cracks were documented.
Ring testing. was also conducted on the diaphragms, not only to determine the condition of the vane-to-shroud welds, but also to help locate cracks in the vanes themselves. Each vane was lightly tapped with a rounded metal hammer. The ringing that resulted was usually of a high to medium pitch and fairly consistent from one vane to the next. When a broken or cracked vane was tapped, a dull pitch resulted. If a definite change in sound occurred, the vane was inspected visually or using liquid penetrant.

Surface examination

Macrofractography. The shroud/blade seal welds, along with their respective fracture faces, were examined under a binocular microscope at magnifications from 10 to 50×. The results of this examination concurred with earlier observations that some welds had been made without use of a filler metal. After the fractures were opened, examination of the fracture surfaces revealed multiple ratchet and beach marks, indicative of fatigue failure. The cracks initiated on one end of the vane and propagated to the other side. In some instances, the cracks had propagated fully. The majority of the fatigue cracks initiated near the trailing edge (thinner end) of the vanes and propagated to the leading edge (Fig. 5).
Figure 5

Fig. 5  Fractograph of broken vane, showing area of crack initiation

Metallography

Microstructural Analysis. Microstructural examination revealed that the cracks initiated in the weld heat-affected zone (HAZ) (Fig. 6). Some sections taken from the inner shroud/vane joint revealed that the welds did not fully penetrate (Fig. 7), leaving defects beneath them. Such defects provide high stress concentrations and can lead to crack propagation. The base metal microstructures were martensitic, whereas the welds were composed of austenite (<80%) and ferrite, which indicated that they were made using an austenitic filler metal.
Figure 6

Fig. 6  Micrograph of compressor diaphragm seal weld, showing crack propagating along weld HAZ. 160×

Figure 7

Fig. 7  Inner shroud/vane joint, showing lack of weld penetration

Chemical analysis/identification

Material and Weld. Energy-dispersive x-ray analysis revealed that the base metal of the vanes was type 403 martensitic stainless steel. Analysis of the weld metal confirmed that the filler metal used was type 309 austenitic stainless steel.

Mechanical properties

Hardness. Microhardness readings were taken of the HAZs and base metals of selected welds. Figure 8 shows one of the inner shroud welds and corresponding microhardness readings. As in Fig. 8, all inner shroud welds exhibited different hardness gradients, indicating inconsistency from one weld to the next. Some welds actually contained more passes than others, which had a definite effect on residual stresses. The outer shroud welds did not exhibit any sharp hardness gradients through the transition zone. The microhardness values indicated that the diaphragms had been stress relieved.
Figure 8

Fig. 8  Micrograph of a diaphragm weld, showing base metal and weld metal microstructures, along with corresponding hardnesses. 122×

Simulation tests

Seven experimental welds were made on one of the broken vanes using different filler metals and postweld stress-relief temperatures. The same welding parameters (heat input, electrode size, etc.) were used for each experimental weld. Table 1 shows average microhardness results for the weld metal, HAZ, and base metal of the experimental welds and the actual welds. The results indicate that the welding parameters used to make the experimental welds were not the same as those used by the OEM. The weld that was made using matching filler metal (type 410 stainless steel) with a 675°C (1250°F) stress relief exhibited optimum properties.

Table 1   Average microhardness values

Label
Experimental welds
Actual
A
B
C
D
E
F
G
1
2
3
Filler material
309L
309L
309L
410
410
Inco A
410
309L
309L
309L
Stress relief
0
950
1250
1250
950
0
0
1100(a)
1100(a)
1100(a)
Weld hardness
47.5
40.0
34.5
28.5
41.5
84(b)
40.5
32.5
33.0
29
HAZ hardness
45.5
43.5
30.5
25.5
33.5
44.5
42.5
20.0
25.5
28
Base hardness
23.0
24.5
85(b)
98(b)
520.5
20.5)
98(b)
88(b)
85(b)
92(b)
Note: All hardness readings are in Rock well C units, except those noted. The base metal hardness readings were taken adjacent to the HAZ.
(a) Estimate stress relief temperatures from OEM.
(b) Rock well B.

Discussion

Use of an austenitic filler metal in the welding of type 403 stainless steel is not uncommon, especially in field weld applications when stress relieving or annealing is not feasible and when there is no requirement for the weld metal to have the same mechanical and corrosion properties as the base metal. Austenitic stainless steel filler metals are often used to obtain more ductile weld metal in the as-welded condition.
Martensitic stainless steels can be welded in the annealed, hardened, and hardened and tempered conditions. Regardless of the prior condition of the stainless steel, welding produces a hardened martensitic zone adjacent to the weld. As hardness increases, toughness decreases, and the zone becomes more susceptible to cracking. Preheating and control of interpass temperature are the most effective means by which to avoid cracking. Stress relieving or annealing the welds alleviates the residual stresses caused during welding. However, annealing cannot be fully effective when using an austenitic filler metal because of the difference in coefficients of thermal expansion between the weld metal and base metal when the compressor is operating.

Conclusion and Recommendations

Most probable cause

The compressor diaphragm seal welds failed because of fatigue cracks that initiated in the weld HAZs on the trailing edges of the vanes. The remaining defects, caused by either no filler metal or lack of penetration, also led to fatigue cracking. The weld HAZs contained high residual stresses because of:
  • Difference in coefficients of thermal expansion between the base metal (martensitic) and the weld metal (austenitic)
  • Improper postweld heat treatment (that is, stress-relief temperature was too low)
  • Inconsistent and poor-quality welding (lack of penetration, no filler metal, varying number of passes, sloppy beads)
Two other factors that may have contributed to cracking were the preheat temperature and the interpass temperature. However, no information was available (from the manufacturers) concerning these two parameters.

Remedial action

All inner shroud welds were ground out and rewelded to ensure 100% penetration. All other welds with crack indications were also ground out and rewelded. Type 410 stainless steel filler metal was used for all rewelding. The diaphragms (base metal) were heated to 230°C (450°F) prior to welding. After welding was completed, the diaphragms were stress relieved at 675°C (1250°F) for 2 h.

Related Information

Failures Related to Welding, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 156–191
Fatigue Failures, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 700–727