Not Every Signal Is a Flaw: Ghost Echoes in UT of P/M Superalloy Rings

Not Every Signal Is a Flaw: Ghost Echoes in Ultrasonic Testing of Powder-Metallurgy Superalloy Rings

FGH4096 is a second-generation damage-tolerant P/M superalloy: compared with first-generation alloys it trades some strength for excellent resistance to crack propagation, with a service temperature of 750 °C. A note on designations before we begin: FGH4096 (GB/T 14992) is the Chinese counterpart of the US alloy René 88 DT, developed by General Electric for turbine disks and rings; the first-generation Chinese alloy FGH4095 corresponds to René 95, and the same damage-tolerant generation includes RR1000 (Rolls-Royce, UK).

Two forming routes are used for FGH4096 rings: direct HIP consolidation, and HIP followed by forging (HIP + F or HEX + F). Rings for an aero engine must combine high strength with low quench-cracking tendency, so they are solution-treated and quenched in a salt bath. If the salt-bath process is not tightly controlled, quench cracking occurs, typically as circumferential cracks or radial cracks on the ring end faces, both of appreciable length.

UT is the most widely used nondestructive method and readily detects cracks, delaminations, shrinkage, porosity, white spots, inclusions, and other metal-gas interface discontinuities; it can also characterize internal structure and stress. For salt-quench cracks in solution-treated rings, contact UT works well: circumferential cracks on the end face are easily found, while radial cracks — which run axially through the part and extend toward the center — are difficult to detect with a straight probe and require shear-wave examination.

The interesting case begins with a discrepancy. On directly HIP-consolidated rings, UT shows genuine crack signals: sharp defect echoes originating at the surface, with amplitude increasing toward the end faces and some back-wall loss. On HIP + forged rings, in contrast, defect signals appear around the entire circumference with almost constant position and amplitude, resembling crack indications, but the end faces show no visible cracks, the back wall loses little energy, and the signal amplitude is lowest near the end faces and highest toward the axial center. As this article — adapted from a study in the journal Powder Metallurgy Industry — demonstrates, those signals were not cracks at all. They were ghost echoes, an instrument artifact created by an over-high pulse repetition frequency acting on a low-attenuation material.

1. Experiment

FGH4096 powder was produced by the plasma rotating electrode process (PREP), then screened, electrostatically treated, canned, and welded before HIP. Two routes were compared. Route I is direct HIP forming: the HIP-consolidated blank is directly solution-treated and aged, at 1140 °C × 2 h with a salt quench. Route II is HIP + forging: the HIP blank is forged, then solution-treated at 1120 °C × 2 h with a salt quench, and aged. All rings examined by UT were in the solution-treated (salt-quenched) state, using the contact pulse-echo method with a portable Olympus Panametrics EPOCH-XT flaw detector.

2. Results

The A-scan waveforms of both rings, scanned along the outer circumference at different axial positions, show crack-like signals. The two routes differ in a decisive way. For Route I rings, the defect signal appears near the end faces with amplitude increasing toward them, some back-wall energy is lost, and the signal disappears toward the axial center if the ring is not cracked through. For Route II rings, the signal is present across the entire circumference, amplitude is highest at the axial center and lowest near the end faces — where it still exists — and the back wall loses little energy. Route I rings therefore contain real cracks; Route II rings show something else.

3. Analysis and Discussion

3.1 Why the Route II signal cannot be a crack

The circumferential cracks on Route I rings are intergranular, consistent with the oxygen-induced dynamic embrittlement mechanism established for René 95: in FGH95, large γ′ precipitates sit on grain boundaries and PPBs, surrounded by Al- and Ti-depleted zones prone to internal oxidation; residual oxygen in the powder oxidizes these zones during HIP or heat treatment, forming NiCr₂O₄ films around γ′ that act as crack sources under quench thermal stress, while the depleted and oxidized boundary paths provide preferred propagation routes. If the Route II signals were cracks, they would have to originate internally — but the mechanism above, and the absence of visible end-face cracks even after machining off 1 mm, rules that out.

3.2 Residual stress and sound velocity

Route II rings undergo recrystallization annealing before solution treatment, so both ring types are expected to show similar residual stress states: tension at the inner bore and the cross-section center, compression at the surface. Internal stress changes sound velocity — compression raises it and tension lowers it when stress and wave propagation directions coincide — so the velocities of both rings were measured radially and axially, as shown in Table 1.

The Route II axial velocity is slightly higher than the radial velocity, while Route I shows no clear difference; overall the differences are tiny, so the material’s sound velocity is not sensitive to its residual stress state at this level. Stress-concentration indications are known to resemble dense inclusion clusters — typically on shaft forgings that were cold-straightened after bending — but the waveforms here are crack-like, so the Route II indication is not stress-induced either.

3.3 Attenuation, pulse repetition frequency, and the ghost echo

After final heat treatment, both rings show comparable grain size, but Route II rings contain noticeably more twins. The radial thicknesses are 75.80 mm (Route I) and 71.83 mm (Route II), both larger than 3N (N = 15 mm at 10 MHz with a 10Z6N probe), so attenuation coefficients were measured radially, as shown in Table 2.

The Route I attenuation coefficient is slightly higher than that of Route II. A very low attenuation coefficient combined with an over-high PRF produces ghost echoes — spurious indications that appear before the first back-wall echo because the previous sync pulse’s multiple back-wall echoes are still ringing when the next sync pulse repeats. To test this, the PRF was changed while recording the A-scan. At the standard 800 Hz the Route II “defect” was present; when the PRF dropped to 600 Hz the signal disappeared, and at 900 Hz its position shifted. The Route I crack signal, by contrast, remained at the same position at both PRFs. This proves the Route II indication is a ghost echo, not a flaw.

The mathematics confirms it. The round-trip time is t = 2l/c, with l the thickness and c the sound velocity; a ghost echo of the n-th back-wall echo appears behind the next sync pulse when n × t > 1/fr. For Route II, l = 71.83 mm and c = 5996 m/s: at a PRF of 800 Hz the 53rd back-wall echo of the previous pulse appears at a displayed depth d1 = 59.62 mm, and at 900 Hz the 47th echo at d2 = 45.29 mm. The measured depths were 59.06 mm and 44.88 mm — very close to the calculated values. Thicker parts ghost more easily; Route I is slightly thicker but has higher attenuation, which is why only Route II showed the artifact. Absorption attenuation is proportional to frequency and should not differ between the routes; scattering attenuation depends on frequency, grain size, and anisotropy. Grain size is similar and frequency identical, so the lower attenuation of Route II is likely related to residual stress and dislocation state — a point worth further study, as is the observation that the ghost amplitude is highest at the axial center, where the residual stress distribution may differ.

4. Conclusions

Two conclusions follow directly from the experiments and the calculation:

• The defect signals found on the outer circumference of solution-treated HIP + forged FGH4096 rings are ghost echoes caused by an over-high pulse repetition frequency — not cracks, and not stress-induced indications.
• Compared with directly HIP-consolidated rings, HIP + forged rings have a lower attenuation coefficient and better acoustic transparency; at comparable thickness, they are therefore more prone to ghost echoes.

The practical lesson is broader than FGH4096. In ultrasonic acceptance testing of low-attenuation, high-quality materials, a “defect signal” must be proven before it becomes a rejection. Without the PRF experiment and the ghost-echo calculation, the Route II rings — sound, forgeable components — would have been scrapped as cracked. For NDT engineers and for buyers of P/M superalloy rings, the acceptance decision belongs to physics: set the pulse repetition frequency for the material, verify indications by changing it, and treat every signal as a hypothesis until the instrument’s own artifacts are ruled out.