The ingestion of foreign objects into jet engines powering military or civil aircraft can lead to severe structural damage of the fan and compressor airfoils; indeed, such foreign object damage (FOD) is a prime reason for maintenance and repair. In particular, the damage induced by small hard objects of millimeter size, in association with the typical load spectra experienced by airfoils, i.e., low-cycle fatigue (LCF) cycling due to normal start/flight/landing cycles superimposed with high-cycle fatigue (HCF) cycles due to vibrations and resonant loads, can lead to non-conservative life prediction and unexpected high-cycle fatigue failures. Specifically, the damage sites can provide initiation points for cracks, which then propagate under the HCF spectra of large numbers (>10⁹) of high frequency (>1 kHz), high load-ratio (R ~ 0.8) cycles. The objective of this study was to characterize (i) the size and shape of the indents and (ii) the level of microstructural damage induced by impacts of foreign objects, (iii) to examine the influence of foreign object damage on crack initiation and crack propagation that can lead to failure by HCF and (iv) how that crack propagation behavior is affected by microstructure. The primary study was performed on a fan/compressor blade processed (STOA) bi-modal microstructure of the titanium alloy Ti-6Al-4V. In addition, a b-annealed lamellar microstructure was investigated to isolate the role of microstructure.
 
 

The titanium Ti-6Al-4V alloy studied was delivered as a solution treated and overaged (STOA) forged plate. The plate came from a set of forgings produced specifically for the U.S. Air Force sponsored programs on HCF. Material and processing details are given in [1]. The plate material showed a bi-modal microstructure with a volume fraction of 60 % primary a (diameter ~20 mm) and 40 % lamellar a+b matrix (Figure 1a). In addition to the bi-modal microstructure, a b-annealed lamellar microstructure (Figure 1b) was investigated for comparison. To obtain a lamellar microstructure, the delivered plate material was solution treated in the b phase field at 1005° C for 10 min. This b-annealing resulted in large b grains with a diameter of approximately 1000 mm. Controlled cooling with a rate of 100-130° C/ min to match the same lamellae width of the bi-modal microstructure (2 mm) resulted in a a+b lamellae colony size of 500 mm. Finally, to ensure comparability with the bi-modal microstructure, the lamellar microstructure was given the same final heat treatment step of 2 h at 705° C.

The uniaxial tensile properties of both bi-modal and lamellar microstructures are listed in Table 1:

Foreign object damage by hard particles is simulated by firing chrome-hardened steel spheres on a flat specimen surface by using compressed gas to produce a single damage site. [2] The specimen geometry chosen for this study has a rectangular gauge section (3mm ´ 5mm) with cylindrical buttonhead grip sections (Figure 2). To provide a consistent, nominally stress-free surface, the gauge section was prepared by standard stress relief and chemical-milling procedures. In this study, spheres (diameter 3.175 mm) were impacted at an angle of 90º (normal impact) at velocities of ~200, 250 and 300 m/s. These velocities were chosen because (i) they represented typical in-service impact velocities on blades, and (ii) they provided different levels of damage (see below). Damage sites have been characterized to date in terms of the geometrical dimensions (shape, diameter, depth) and microstructural changes.

SEM micrographs of typical impact sites for the bi-modal microstructure are shown in Figure 3a (200 m/s) and Figure 3b (300 m/s). For the lamellar microstructure typical impact sites are shown in Figure 4a (200 m/s) and Figure 4b (250 m/s). For the bi-modal microstructure, corresponding cross sections of impact craters are shown in Figure 5a (200 m/s) and Figure 5b (300 m/s). While at 300 m/s (Figure 5b) and 250 m/s, tangential orientated shear bands (Figure 6) were seen emanating from the surface of the crater, such shear bands were not observed for 200 m/s impacts (Figure 5a). This result corresponds with the reported critical velocity of 214 +/-16 m/s to achieve shear bands in Ti-6Al-4V by an impacting 3.2 mm diameter steel sphere [4]. The amount and length of the observed shear bands increased with increasing impact velocity. In addition, the 300 m/s impacts caused a ridging along the crater rim (Figures 5b and 6b); this feature was not seen for the lower velocity impacts (Figure 5a). Figure 7 summarizes the relevant features in form of a schematic illustration of the crater profile. In the range of 100 to 320 m/s impact velocities, the diameter and depth of damage site increased linearly with increasing impact velocity.
 
 

After impact, the tensile specimens were subsequently cycled at 20 Hz (sine wave) with a maximum nominal stress of 500 MPa at a load ratio of R = 0.1. Periodically, the specimens were removed from the test frame and examined in a scanning electron microscope (SEM) to detect crack initiation. Once a crack had initiated, subsequent crack growth was similarly monitored using periodic SEM observations. Stress intensities for the surface cracks were calculated from the Newman and Raju linear elastic solution for semi-elliptical surface cracks, assuming a half-surface length to depth ratio of 0.9 (determined from fractographic measurements). [3] In this initial study, contributions to the local stress intensity factor values from the geometry of the indentation and the residual stress field surrounding the indent have been ignored. Current studies are focused on the use of micro X-ray diffraction methods to measure such residual stresses.

The effect of the simulated foreign object damage was to markedly reduce the fatigue life compared to that obtained with an undamaged smooth-bar specimen [5], as shown in Figure 9 for the bi-modal microstructure. At maximum stresses of 500 MPa (R = 0.1), surface cracks were initiated within ~3 ´ 10⁴ to ~5 ´ 10⁴ cycles, the failure of the impacted specimens occurred between ~4 ´ 10⁴ and ~8 ´ 10⁴ cycles. Specifically, fatigue lives were reduced by over two orders of magnitude following 200 m/s and 300 m/s impacts.

For the bi-modal microstructure, crack initiation sites of the small surface cracks are shown in Figure 10a (200 m/s) and in Figure 10b (300 m/s). First results show that the cracks tended to initiate at the bottom of the indent for the lower velocity impacts and at the crater rim for the higher velocity impacts. Corresponding fracture surfaces are shown in Figure 11a (200 m/s) and Figure 11b (300 m/s). The position of the crack front during crack extension is indicated schematically (Figure 11). For the 200 m/s impact, the initiation at the bottom of the indent can be explained by stress concentration due to the notch geometry. Crack initiation at the crater rim (300 m/s) might be related to circumferential residual tensile stresses at the crater ridge [e.g. 6], which overcompensate the stress concentration effect of the indent.

The presence of the tangential orientated shear bands do not appear to play a significant role in the initiation process of the fatigue cracks. This is thought to be due to the fact that at the center of the indent, the tangential aligned shear bands are parallel to the applied stress axis.

The growth of the small cracks originating from such impact sites are compared in Figure 12 with growth-rate data for large (>5 mm) [7] and naturally-initiated small (~45-1000 mm) [5] cracks for the bi-modal microstructure. Both the naturally-initiated and FOD-initiated small-crack velocities were within the same scatter band, initially up to an order of magnitude faster than corresponding large crack results. The small-crack data tended to merge with large-crack results above DK = 10 MPaÖ m as the crack size increased. However, in the limited data collected to date, no FOD-initiated cracks have been observed in the Ti-6Al-4V bi-modal microstructure below DK = 2.9 MPaÖ m; i.e., no FOD-initiated cracking was observed below the worse-case (large crack) threshold, DK_TH of 1.9 MPaÖ m, measured using constant K_max cycling at a final R ratio of 0.95 [7] Therefore, fatigue crack propagation thresholds of large cracks determined under conditions that minimize crack closure can be used as a lower bound for the threshold stress intensities for FOD initiated cracks.

To examine the influence of microstructure in addition to the a+b processed bi-modal microstructure, a b-annealed lamellar microstructure was also investigated. A typical crack initiation site of the small surface cracks is shown in Figure 13 for an impact velocity of 200 m/s. As shown previously for the bi-modal microstructure, cracks tended to initiate at the bottom of the indent for the 200 m/s and 250 m/s velocity impacts and that the presence of the tangential orientated shear bands did not appear to play a significant role in the initiation process. Further testing at higher velocities (300 m/s) is currently underway to evaluate crack initiation at the crater ridge.

Preliminary results of fatigue crack propagation tests of the lamellar microstructure are shown in Figure 14. It can be seen that the lamellar microstructure showed a slightly higher resistance against the propagation of small surface cracks as compared to the bi-modal microstructure. These preliminary results indicate that the resistance of the impacted microstructure against fatigue crack propagation is only minimally affected by significant changes in the microstructural dimensions (bi-modal vs. lamellar).
 
 

• Foreign object damage, simulated by hardened steel spheres impacted at 200-300 m/s on a flat surface, provides sites for the initiation of small fatigue cracks.

• At applied stresses some 10 % below the smooth-bar fatigue limit (10⁷ cycles), crack initiation lives were less than 4 x 10⁴ cycles, many orders of magnitude lower than lifetimes for un-impacted samples.

• Subsequent propagation of small cracks from the damage sites was faster as compared to the propagation rates of large cracks subjected to the same applied DK level.

• No crack growth from the damage sites has been observed to date at DK values less than 2.9 MPaÖ m. The "worse-case" large crack threshold value could thus be used as a lower bound to FOD initiated cracking.

• Preliminary results of the microstructure comparison showed that the resistance against fatigue crack propagation was only minor affected by a drastic change of microstructure (bi-modal vs. lamellar).

This work is supported by the Air Force Office of Science and Research, Grant No. F49620-96-1-0478, under the Multidisciplinary University Research Initiative on "High Cycle Fatigue" to the University of California, Berkeley.
 
 

1. D. Eylon, "Summary of the Available Information on the Processing of the Ti-6Al-4V HCF/LCF Program Plates", University
    of Dayton Report, Dayton, OH. (1998).

2.. R. O. Ritchie, S. Suresh, J. W. Hutchinson, W. W. Milligan, A. G. Evans, A. W. Thompson, "High-Cycle Fatigue and
     Time-Dependent Failure in Metallic Alloys For Propulsion Systems", AFOSR F49620-96-1-0478, Progress Report
     (1997).

3. J. C. Newman, Jr., and I. S. Raju, Eng. Fract. Mech. 15, 185 (1981).

4. S. P. Timothy and I. M. Hutchings, in Proceedings of the 3^rd Conference on Mechanical Properties at High Rates of
    Strain, The Institute of Physics, 1984, 397.5

5. J. A. Hines, J. O. Peters and G. Lütjering, "Microcrack Propagation in Ti-6Al-4V", in Fatigue Behavior of Titanium
    Alloys, R.R. Boyer, D. Eylon, J.P. Gallagher and G. Lütjering, TMS, Warrendale, PA, 1999.

6. C.J. Studman, J.E. Field, "The Indentation of Hard Metals: The Role of Residual Stresses", J. Mater. Sci. 12, 215 (1977).

7. B. L. Boyce, J. P. Campbell, O. Roder, A. W. Thompson, W. W. Milligan, and R. O. Ritchie, Int. J. Fatigue (1999),
    accept for publication and in press.
 
 

                                                
                a) Bi-modal                                        b) Lamellar

                Fig. 1: Microstructures (LM)                                                                   Fig. 2: Specimen geometry
 
 

                                                          
                                 a) 200 m/s                                                 b) 300 m/s

     Fig. 3: Bi-modal microstructure. Impact damage sites (SEM), Impact conditions: 3.2 mm steel sphere, 90º.  
 

                                  
              a) 200 m/s                                                  b) 250 m/s

Fig. 4: Lamellar microstructure. Impact damage sites (SEM), Impact conditions: 3.2 mm steel sphere, 90º.
 
 
 

                    
    a) 200 m/s                                                             b) 300 m/s
 

Fig. 5: Bi-modal microstructure. Cross sections of impact craters (LM). Impact conditions: 3.2 mm steel sphere, 90º.
 
   
                           
    a) shear bands at crater bottom                                     b) crater ridge

     Fig. 9:  Bi-modal microstructure. Influence of FOD on fatigue behavior, compared to S-N smooth-bar fatigue data after
                 J.A. Hines, J.O. Peters and G. Lütjering (1999).

  Fig. 11: Bi-modal microstructure. Fracture Surfaces (SEM), Position of crack front as a function of number of cycles during
               crack extension is shown schematically
 
 
 

                                

Fig. 13: Lamellar microstructure. Crack initiation site at the specimen surface for a 200 m/s impact (SEM): s_max = 500 MPa,
            R = 0.1, 20 Hz.
 
 

Fig. 14: Fatigue crack propagation behavior of small cracks emanating from simulated FOD sites. Comparison of bi-modal and
             lamellar microstructures.