Gas turbine coatings eddy current

quantitative and qualitative evaluation

Cite as: AIP Conference Proceedings 1806, 110022 (2017); https://doi.org/10.1063/1.4974700
Published Online: 16 February 2017

Remo Ribichini, Carlo Giolli, and Erica Scrinzi

ARTICLES YOU MAY BE INTERESTED IN

Development of magnetic eddy current testing techniques

AIP Conference Proceedings 1806, 110021 (2017); https://doi.org/10.1063/1.4974699

Electromagnetic nondestructive evaluation of defects in ferromagnetic structures

AIP Conference Proceedings 1806, 110024 (2017); https://doi.org/10.1063/1.4974702

Model development and validation of geometrically complex eddy current coils using finite
element methods
AIP Conference Proceedings 1806, 110015 (2017); https://doi.org/10.1063/1.4974693

AIP Conference Proceedings 1806, 110022 (2017); https://doi.org/10.1063/1.4974700 1806, 110022

© 2017 Author(s).
 Gas Turbine Coatings Eddy Current Quantitative and
Qualitative Evaluation

Remo Ribichinia), Carlo Giolli and Erica Scrinzi

GE Oil & Gas Turbomachinery Solutions. Florence, 50127, ITALY

a)
Corresponding author: remo.ribichini@ge.com

Abstract. Gas turbine blades (buckets) are among the most critical and expensive components of the engine. Buckets
rely on protective coatings in order to withstand the harsh environment in which they operate. The thickness and the
microstructure of coatings during the lifespan of a unit are fundamental to evaluate their fitness for service. A frequency
scanning Eddy Current instrument can allow the measurement of the thickness and of physical properties of coatings in a
Non-Destructive manner. The method employed relies on the acquisition of impedance spectra and on the inversion of
the experimental data to derive the coating properties and structure using some assumptions. This article describes the
experimental validation performed on several samples and real components in order to assess the performance of the
instrument as a coating thickness gage. The application of the technique to support residual life assessment of serviced
buckets is also presented.

INTRODUCTION

Heavy-Duty Gas Turbines (Fig. 1) are widely employed for power generation and mechanical drive applications
in the Oil & Gas industry [1]. Reliability is one of the most important features of the engines as often the operation
of a whole plant depends upon the power provided by a few units; an unplanned outage would lead to loss of
production and significant economic damage for the operator.
One of the most critical components of the engine is the first stage turbine blade (or bucket) which must
withstand a combination of high temperature, severe stress and aggressive environment [2]. In order to protect the
bucket material from corrosion and oxidation, which would rapidly reduce its mechanical strength, high temperature
coating are employed. High temperature coatings must be used when the operating environment temperature would
overcome the inherent oxidation resistance of the bucket base material. Two main types of coatings have been used
by GE [2]: platinum aluminide diffusive coatings (PtAl) and overlay MCrAlY coatings (where M stand for Ni and
or Co). The function of coatings is to provide a reservoir of elements that will form protective oxide layers thus
avoiding bucket base material damage.
Given the importance of coatings for gas turbine engines reliability, their assessment by means of Nondestructive
Evaluation (NDE) methods is highly desirable, both for production quality control and for in-service assessment.
Several investigations have been performed using eddy current techniques [3-6]. Since the electromagnetic skin
depth varies with frequency, it is very attractive to measure depth-dependent electromagnetic properties (either
using pulsed or frequency scanning methods) and then infer the testpiece structure via inversion techniques [4, 7-
10]. Most of previous work has been performed at research level, employing laboratory equipment in ideal
conditions; the present work focuses on field applications, using an off-the-shelf commercial system that can be

43rd Annual Review of Progress in Quantitative Nondestructive Evaluation, Volume 36

AIP Conf. Proc. 1806, 110022-1–110022-9; doi: 10.1063/1.4974700
Published by AIP Publishing. 978-0-7354-1474-7/$30.00

110022-1
easily used even by relatively unskilled field operators and it is affordable for site tests. For these reasons FSECT, an
instrument supplied by CESI was selected.
The present work will first review the physics employed to evaluate coatings structure; then the eddy current
experimental measurements are compared against destructive tests. Experimental tests on serviced samples are
presented. Finally, the results are analyzed and compared to literature laboratory results, taking into account some
field experiences. It has to be noted that the scope of present work is limited to MCrAlY anti-oxidation coatings
only; diffusive coatings may be investigated in a future work. Also thermal barrier coatings TBCs, which are
typically nonmetallic, are not in the scope of work.

FIGURE 1. GE Oil & Gas MS5002E gas turbine cross-section. From [1].

FIGURE 2. Typical structure of MCrAlY coating after aging.

110022-2
 FREQUENCY SCANNING EDDY CURRENT TECHNIQUE
For a homogenous half-space, eddy current standard penetration depth į can be computed with the well-known
relationship [12]:

2
δ= (1)
ωσμ

where, Ȧ denotes the angular frequency, ı is the electric conductivity and ȝ is the magnetic permeability of the
conductor. This relationship ignores any finite-probe effects.
For a coating-substrate system an electric conductivity profile ı(z) exist. However, for multilayer MCrAlY
coating as the one shown in Fig. 2, the conductivity profile ı(z) can be simplified into a number of discrete
homogenous layers with electric conductivity ıi [4,5]. For such cases, the complex impedance measured by an eddy
current coil can be described by the analytical solutions found by Dodd and Deeds [13].
From the experimental point of view, the complex impedance spectrum is acquired between 0.4 and 8.0MHz
using a §6mm diameter flexible probe and the inbuilt electronics of FSECT. The calibration procedure is essentially
an extension of the four-point linear interpolation procedure [7]. Two reference samples with known electrical
conductivities ıA and ıC similar to those of the coating and the substrate are used, together with two calibrated shims
of different thickness. Six measurements are taken for each acquisition frequency: three measurements per sample
with three different lift-off levels. Since the conductivity gradient between the layers is relatively small, an
interpolation can be used in the complex impedance plane to obtain the apparent eddy current conductivity (AECC).
A typical AECC spectrum is shown in Fig. 3. For high frequencies, corresponding to shallow penetration depth (Eq.
(1)), the AECC converges toward the electric conductivity of the coating only, whereas for low frequencies (i.e.
large penetration depth), the AECC converges to the conductivity level of the substrate.
Once the AECC spectrum has been measured, significant physical and geometrical properties (namely the
electric conductivities ıi and layer thicknesses di) can be obtained by using a suitable inversion technique. FSECT
employs an inbuilt iterative minimization algorithm that assumes a testpiece made of homogeneous parallel layers.
The software takes as an input both the experimentally measured spectrum and some prior knowledge of the coating
structure, i.e. the number of layers and a guess range of their conductivities and thicknesses [14].

1.0
0.9
0.8
0.7
AEEC [arb.]

0.6
0.5
0.4
0.3
0.2
0.1
0.0
5.0E+05 5.0E+06
Frequency [Hz]

FIGURE 3. Typical AECC spectrum of a MCrAlY coating. A transition from a low conductivity substrate at low frequencies
(high penetration depth) to a higher conductivity coating at high frequency (low penetration depth) can be observed.

110022-3
 COATING THICKNESS MEASUREMENT
A large number of experimental tests has been performed to evaluate the performance of the technique as a
thickness gage. Most tests were performed in field or workshop conditions. For all the samples and components
tested the following procedure has been followed:

• Non-destructive measurement on a significant number of points

• Destructive analysis of the measured points (EDM cutting and microscope measurement)
• Comparison of the results in terms of coating thickness.

Overall, around 200 data points have been acquired and validated against the relevant destructive measurement
(Fig.4). The test objects under examination were either new coupons and buckets or serviced buckets with negligible
alteration of original metallurgical properties, such as beta-phase depletion. The latter condition was verified
through destructive metallurgical analysis. Some combustion hardware components whose substrate materials and
coatings are similar to those of buckets were also included in the study. The inversion model used was a simple bi-
layer structure, neglecting any fine details such as inter-diffusion zone or surface roughness (Fig. 2). The standard
deviation of thickness error is 15% with nearly zero average. Given that the measurements were taken in field
conditions with a commercial instrument, the results are satisfactory and yield useful information for condition
based maintenance strategies.

As New GT29/GT33
Serviced FR61 S1B 64khr
Serviced FR91 S1B 48khr
Serviced Liner
Serviced XF Tube
Thickness - ECI [a.u]

Serviced Hula Seal

Thickness - Destructive [a.u.]

FIGURE 4. Summary of coating thickness validation results. Eddy current FSECT measurements are plotted against the
thickness measured destructively.

110022-4
 ASSESSMENT OF IN-SERVICE COATINGS

MCrAlY coatings show a biphasic structure (Fig. 2) [2]: a Ȗ-phase Ni,Co,Cr solid solution matrix and ȕ-(Ni,
Co)Al phase. The latter phase provides the aluminum needed for protective oxide scale formation. During gas
turbine engine operation, ȕ-phase is depleted: Al diffuses outward to form the oxide scale protecting against
oxidation/corrosion. The aluminum also diffuses towards the coating-substrate interface due to Al concentration
gradient; this creates an interdiffusion zone that grows as a time-temperature dependent phenomenon (Fig. 5). A
typical depletion phenomenon is shown in Fig. 5: over time the outer and inner depleted layers grow at the expense
of the ȕ-phase rich layer. The residual ȕ-phase content is thus strictly related to the remaining life of a coating.

FIGURE 5. Aging of an MCrAlY coating. Exposure to high temperature over time ((a) to (d)) leads to the depletion of the ȕ-
phase rich layer.

Tests have been performed on coupons artificially aged by being exposed at a given relevant temperature T for t
hours in a furnace. Eddy current and destructive tests were performed at given time intervals to evaluate the
depletion phenomenon. The initial AECC spectra of six coupons are shown in Fig. 6 (a). Fig. 6 (b) shows the time
evolution of AECC spectra: initially (e.g. below 2t ) no major changes occur, whereas after e longer exposure time
(e.g. above 9t) the spectra flattens, due to the lower content of highly conductive Aluminum, indicating a ȕ-phase
depletion phenomenon.

110022-5
FIGURE 6. AECC spectra of coated coupons. (a) As new, before any aging process. (b) Following high temperature ageing in a
furnace. Aging time is express in multiples of a reference time t.

In order to provide a quantitative evaluation of the degradation, a simplified 3-layer inversion model was used to
invert the experimental data. The model is shown in Fig. 7; it has to be noted that the internal depletion, the
interdiffusion zone (IDZ) and the bucket substrate are all aggregated into a single layer. The residual ȕ-phase is then
defined as the percentage ratio between the layer thicknesses (d2 -d1) over the total coating thickness. Figure 8 shows
a comparison between the eddy current results against the destructive tests on 5 coupons. For residual ȕ-phase
between 70% and 100% the agreement between the NDT measurements and the micrographs is satisfactory;
however, for more severe depletion, the discrepancy is significant. The main reason for such disagreement is
probably due to the hardware limitation of FSECT instrument: even at the highest excitation frequency (around
10MHz) the standard penetration depth į is too large to resolve effectively a thin (less than few tens of microns) ȕ-
phase rich layer. This hypothesis to explain the mismatch between NDT and destructive tests is further supported by
previous work in laboratory conditions [4, 5], which showed that using driving frequencies up to 50MHz a good
numerical match could be obtained.

FIGURE 7. Coating degradation investigation. The actual coating structure (left) is simplified into a 3-layer model (right) used
for experimental data inversion.

110022-6
 ECI
100%
Destructive

80%

Residual ȕ-phase [%] 60%

40%

20%

0%
0 3 6 9 12 15
Time – [arb.]
FIGURE 8. Coating degradation investigation. Residual ȕ-phase percentage of aged coupons as measured with FSECT eddy
current instrument against destructive tests results.

FIELD EXPERIENCE AND DISCUSSION

The eddy current methodology described in this work has been applied to more than 10 buckets sets form
different engines and plants, yielding valuable information. Even if it is not possible to accurately measure the ȕ-
phase rich layer thickness, the AECC spectra can qualitatively indicate when an area of coating has undergone
oxidation phenomena. For instance Fig. 9 shows the pressure side trailing edge of two buckets from the same
engine. From visual inspection, bucket Fig. 9(a) is in perfect conditions, while bucket Fig. 9(b) shows some
localized damage. Eddy current FSECT measurements were performed in these areas yielding the spectra of Fig.
9(c).There is a significant difference between the two spectra, both in terms of magnitude and of shape. For
increasing excitation frequency, bucket (a) spectrum shows a typical transition between a lower conductivity
substrate and a higher conductivity aluminum rich coating. On the other hand, bucket (b) shows anomalous
impedance spectrum with low levels of AECC throughout the measured frequency range. Subsequent destructive
tests and micrographic examination confirmed that bucket (a) had indeed a coating with substantial ȕ-phase rich
layer while bucket (b) had undergone local oxidation, as could be guessed by eddy current spectrum.
This example shows that while the way forward to obtain accurate ȕ-phase layer thickness is clear – i.e.
increasing the frequency range of spectrum acquisition, it is not clear if this would be required for practical
applications. A qualitative evaluation of AECC spectra together with knowledge of engine operating conditions
normally yields enough information to decide whether a coating is fit for service or if it needs to be refurbished.
Moreover, field experience demonstrated that apart from coating deterioration other failure modes are at work:
domestic or foreign object damages, bucket tip rubbing and other phenomena can be decisive for part rejection or
acceptance. This means that measuring with high precision coating properties may not be necessary for operative
decisions.

110022-7
 2.0
Bucket (a)
1.5 Bucket (b)
1.0
0.5

AECC [arb.]
0.0
-0.5
-1.0
-1.5
-2.0
5.E+05 5.E+06
Frequency [Hz]
(a) (b)
(c)
FIGURE 9. (a) and (b) Surface conditions of two first stage buckets following operation. The trailing edge, pressure side area
near the tip is shown. (c) AECC spectra for the buckets shown in (a) and (b).

CONCLUSIONS
A commercial eddy current instrument has been employed to measure coating thickness of heavy duty gas
turbine buckets. The study was focused on MCrAlY overlay coatings only. The measurement accuracy has been
validated against destructive tests on both coupons and real components; overall almost 200 experimental data
points have been collected. The validation demonstrated good agreement between eddy current and destructive
result with standard deviation error lower than 15% of thickness.
In-service coating conditions were also evaluated on artificially aged coupons. Apparent eddy current
conductivity spectra successfully evaluated coating conditions qualitatively, i.e. were able to discriminate between
coatings with large ȕ-phase rich layer against heavily depleted coating. On the other hand, a precise quantitative
measurement of ȕ-phase rich layer was only successful for lightly depleted coating, i.e. below 30% thickness
depletion. In order to obtain better accuracy even for thin ȕ-phase layer, higher frequency electronics shall be used
such that the standard penetration depth is finer.

ACKNOWLEDGMENTS
The authors would like to acknowledge the valuable discussions with G. Antonelli (CESI), Prof. P.B. Nagy
(University of Cincinnati) and Y. A. Plotnikov (GE Global Research). One of the authors, R. Ribichini,
acknowledges V. Depau (GE Oil & Gas) for his support in the publication of this work and G. Di Marzo (Opus
Automazione) for performing most of the experimental tests presented.

REFERENCES
1. GE Oil & Gas website: https://www.geoilandgas.com/sites/geog.dev.local/files/ms5002e-heavy-duty-gas-
turbine-product-brochure.pdf
2. P.W. Schilke, A.D. Foster., and J.J. Pepe. Advanced gas turbine materials and coatings. General Electric
Company, 1991.

110022-8
3. V. Zilberstein, I. Shay, R. Lyons, N. Goldfine, T. Malow and R. Reiche. “Validation of multi-frequency eddy
current MWM sensors and MWM-Arrays for coating production quality and refurbishment assessment” in
ASME Turbo Expo 2003, collocated with the 2003 International Joint Power Generation Conference.(American
Society of Mechanical Engineers, New York, NY), pp. 581-590.
4. H.A. Sabbagh, E. Sabbagh, R.K. Murphy, and J. Nyenhuis. “Assessing thermal barrier coatings by eddy-current
inversion” in Review of Progress in Quantitative Nondestructive Evaluation, edV D. Chimenti and D.
Thompson (AIP Publishing0HOYLOOH1<, 722-727
5. H.A. Sabbagh, J.S. Knopp, J.C. Aldrin, R.K. Murphy, E. Sabbagh and J. Nyenhuis. “Eddy Current Inversion
and Estimation Metrics for Evaluating Thermal Barrier Coatings” in Review of Progress in Quantitative
Nondestructive Evaluation, edV D. Chimenti and D. Thompson (AIP Publishing  0HOYLOOH 1<), 

6. C. Rinaldi, B. Valerio and P. Colombo. J. Eng. Gas Turb. Power 128, 73-80 (2006).
7. Y. Feng, and P.B.. Nagy. J. Appl. Phys. 96, 1257-1266 (2004)
8. Y. Feng, and P.B.. Nagy. " J. Nondestruct. Eval. 24 143-151 (2005)
9. G. Antonelli, M. Ruzzier, and F. Necci. "Thickness measurement of MCrAlY high-temperature coatings by
frequency scanning eddy current technique" in ASME 1997 International Gas Turbine and Aeroengine
Congress and Exhibition. (American Society of Mechanical Engineers, New York, NY, 1997)
10. A. Schnell, K. Germerdonk, H. Mohlig, B. Fehrmann and P. Lee. “An Innovative Non-Destructive Method for
the In-Service Metal Surface Temperature Estimation of Coated GT Parts”. in ASME Turbo Expo 2007: Power
for Land, Sea, and Air (American Society of Mechanical Engineers, New York, NY) pp 53-60
11. E. Pietrangeli, F. Cappuccini, M. Bellacci, F. Iozzelli. “Characterization and oxidation damage modelling of
MCrAlY” in ASME Turbo Expo 2006: Power for Land, Sea, and Air (American Society of Mechanical
Engineers, New York, NY) pp 925-930
12. D. C. Jiles, Introduction to Magnetism and Magnetic Materials (Chapman and Hall, London, 1998).
13. C.V. Dodd and W.E. Deeds. J. Appl. Phys. 39 2829-2838 (1968)
14. FSECT System v. 3.00 - User’s Manual, (Milan, 2011)

110022-9