Patent Application
20130003152
This invention relates generally to manufacturing, and specifically to non-destructive techniques for inspecting high-precision parts including gas turbine engine components. In particular, the invention concerns interferometry techniques for surface quality distributions and stress analysis, as applicable to turbine airfoils and other precision components with low-tolerance surface features and protective coatings.
This invention concerns a non-contact testing method and related systems. The method includes illuminating a sample with a coherent source, generating a first interference image of the sample, inducing a phase shift in the coherent source, generating a second interference image, inducing a load on the sample, generating a third interference image.
The first interference image represents surface stress in the sample. The second interference image includes carrier fringes based on the phase shift. The third interference image represents a change in the surface stress, due to the load. A phase distribution is generated based on the interference images, where the phase distribution represents the change in the surface stress.
Laser source 11 includes a coherent light emitter for generating laser beam (coherent light) L, in either pulsed or continuous-wave mode. In one configuration, laser source 11 includes a 658 nm laser diode operating at 50 mW. Alternatively, the wavelength and power output vary, for example with wavelength from 635 nm to 650 nm, or between about 400 nm and about 800 nm, and with power output from 10 to 100 mW, or up to 250 mW or more.
Optical splitter 12 includes a beam splitter for splitting source beam L from laser 11 into two separate laser beams (coherent sources) B1 and B2, with coherent light propagation along different directions toward sample part 14 and phase plate 13, respectively. Depending on configuration, splitter 12 may include a polarizing beam splitter such as a Wollaston prism, so that light has different linear polarizations in coherent beams B1 and B2, or an achiral layered structure, so that light has different circular polarizations in coherent beams B1 and B2. Alternatively, splitter 12 includes a half-silvered mirror or a pair of non-polarizing triangular prisms, and light has substantially the same polarization in coherent beams B1 and B2, or no particular polarization.
Phase plate 13 includes an optical element to provide a phase delay between two different optical components, for example a birefringent crystal with fast and slow optical axes oriented to delay one plane polarization in beam B1 with respect to a perpendicular (second) plane polarization in beam B2. In quarter-wave and half-wave configurations, phase plate 13 provides a phase shift of one quarter or one half of a wavelength (or cycle), respectively. In general, however, the magnitude of the phase shift depends on the optical thickness of phase plate 13, and the relative phase shift (or delay) can be adjusted by changing angle θ of phase plate 13 with respect to beam B2.
Sample part (or sample) 14 can be either a reference (test) component or a precision part, for example a blade or vane airfoil for a gas turbine engine. Sample part 14 typically includes a protective coating or other low-tolerance surface S, for example a thermal barrier coating (TBC), a metallic (e.g., NiCrAlY) coating, or another low-tolerance coating or precision surface.
Optical medium 15 includes a holographic medium illuminated by reflected beam B1′ and phase-delayed reference beam B2′, which combine to form an interference pattern. Depending on configuration, holographic medium 15 may include a reversible holographic film or a polymer-based read/write/erase optical memory medium for recording the interference pattern as a holographic image. Alternately, holographic medium 15 includes a registration chamber with a thermoplastic top layer for recording (or registering) the interference pattern, or a thermoplastic registration chamber with a photorefractive crystal medium or a photosensitive polymeric medium for recording a holographic image.
Detector 16 includes a charge-coupled device (CCD) or other photosensitive detector configured to record an interference pattern or holographic image formed in holographic medium 15, or other optical information. Depending on configuration, detector 16 may utilize one or more lenses 20 to focus a holographic image from holographic medium 15, or detector 16 may collect optical data by transmission or reflection. In general, system 10 also includes additional lenses, beam spreaders, mirrors and other optical components to control the width, direction and intensity of beams L, B1 and B2, and to increase or decrease illumination across sample part 14, holographic medium 15 and detector 16.
Controller 17 includes actuators and electronic switches configured to control various other components of interferometer system 10, including laser source 11, phase plate 13 and heater 19. In particular, controller 17 directs laser source 11 to turn on and off, or to pulse source beam L, and controller 17 directs phase plate 13 to rotate within reference beam B2, changing setting angle θ to adjust the phase delay of beam B2 with respect to beam B1. Controller 17 also turns heater 19 on and off, in order to control thermal loading on sample part 14.
Microprocessor (μp) 18 analyzes the data from CCD 16, in order to reproduce surface stress patterns based on the interference images generated by reflected beams B1′ and phase-delayed beam B2′ in holographic medium 15. Depending on configuration, microprocessor (or processor) 18 also includes software and hardware to direct operation of controller 17.
Heater 19 includes a resistive or radiant heating element such as a quartz heater, which is configured to regulate thermal loading on sample part 14. In particular, heater 19 is configured to heat sample part 14 from an ambient temperature of about 20° C. (or 68° F.) to a thermal loading temperature of about 30-50° C. (86-122° F.). Alternatively, the ambient temperature range is about 0-25° C. (32-77° F.), and the thermal loading temperature ranges up to 50-70° C. (122-158° F.).
Thermal loading is accomplished by positioning heater 19 relatively close to sample part 14, and operating heater 19 for a period of one to ten minutes, more or less, for example about six minutes. The power output of heater 19 is relatively low, and the resulting temperature and thermal loading on sample part 14 are relatively moderate. This contrasts with other systems, in which testing requires extensive thermal or mechanical loading and heating to temperatures above 100° C. (or 212° F.), which can damage sensitive components of sample part 14.
In operation of system 10, processor 18 directs controller 17 to pulse (or turn on) laser source 11, generating source beam L. Beam splitter 12 splits source beam L into first (sampling) beam B1, which illuminates surface (or coating) S of sample part 14, and second (reference) beam B2, which passes through phase plate 13. Reflected illuminating beam B1′ is incident from surface S of sample part 14 onto holographic medium 15, where it combines with phase-shifted reference beam B2′ from phase plate 13 to produce a holographic image or other interference pattern.
Detector 16 records the interference pattern as a function of intensity and surface (x,y) position, based on the photon count in each pixel in the CCD array. In the configuration of
A second image is formed by rotating phase plate 13 to angle θ with respect to sampling beam B2, altering the phase shift of sampling beam B2′ to introduce carrier fringes into the interference pattern. A third image is formed after activating heater 19 to place sample part 14 under a thermal load.
Microprocessor (or computer processor) 18 analyses the optical data stored on detector 16, in order to locate stress defects and other surface features on sample part 14. Processor 18 also directs controller 17 to actuate the various components of holographic interferometer system 10, including laser source 11, wave plate 13 and heater 19, as described above.
System 10 compares holographic fringe patterns for two different conditions of sample part 14 with coating S, before heating (initial condition; cool) and after heating (final condition; hot). The mechanical stress (σs) on coating S is:
σS=σT+σI, [1]
where σI is the internal stress caused by deposition of the coating, and σT is the thermal stress produced by differences in the coefficient of thermal-expansion (CTE) for the substrate and coating materials.
Thermal stress σT can be written in terms of the coefficients of thermal expansion of the coating (αF) and substrate (αs), respectively:
σT=(αF−αS)·ΔT·E, [2]
where ΔT is temperature gradient or difference in temperature of the coating with respect to the substrate, as defined during deposition of the coating, and E is elastic modulus of the coating material.
Upon (isothermal) heating of sample part 14, mechanical stresses σS result in elastic deformation of the surface, and a surface relief pattern forms. Interference system 10 obtains fringe patterns that represent the surface stresses and corresponding relief (displacement) pattern on the surface of sample part 14, using a carrier interference fringe technique.
The advantages of this technique lie in the fact that carrier fringes reduce or eliminate zero fringe effects, allowing processor 16 to determine a working fringe number and reconstruct 3-dimensional phase portraits across surface S of sample part 14. The phase portraits generate more precise displacements, because phase measurements have higher accuracy than direct determinations based on the order and location of the interference fringes.
System 10 also produces a monotonic phase distribution over the surface of sample part 14, allowing processor 16 to perform an automated computer analysis of the resulting fringe pattern. In addition, system 10 provides for reconstruction and analysis of the phase relief pattern (phase distribution) in the surface of the part being inspected, introducing the standard deviation and z-score distribution analysis of the width and magnitude of the phase distribution as a criterion for inspecting part surface quality.
Phase characteristics are determined by transforming real intensity values reconstructed from the interference image of sample part 14 into a complex-valued function. The transformation is realized by means of direct and inverse (or reverse) Fourier and Hilbert transformations of the intensity distributions on detector 16, as described below.
When stress defects occur in sample part 14, the interference fringe pattern is shifted or altered, making it possible to locate stress-critical regions of surface S. System 10 determines the relevant characteristics of the stress defects by analysis of the interference fringe shift, using processor 18. The cycle time is a few seconds or less, allowing for repeated real-time interference analysis, using both holographic (
Interferometer system 10 also provides highly accurate, easily interpreted data based on contactless measurements and non-destructive testing procedures, useful to a wide range of different sample parts 14 and coating materials S. In particular, system 10 provides local stress and defect (crack or blister) locations on coated airfoils and other precision parts, as well as efficient and reliable detection of weak coating zones, stress ridges and other precursors to stress-related failure modes.
In contrast to x-ray, sonic, magnetoelastic, and eddy current methods, interferometer system 10 also provides more accurate measurements, which are applicable to a wider range of substrates and coating materials. In contrast to gel-based silver halogenide emulsions and other “wet treatment” systems, moreover, interference images can be rapidly recorded, erased and re-recorded on reversible optical (holographic) medium 15, using digital imaging system (detector) 16 to increase speed and repeatability. System 10 also avoids the need to form indentations or drill holes into surface S of sample part 14, eliminating contact operations while providing a combination of holographic and speckle interference techniques to measure both normal and tangential components of the surface displacement, as described below.
Interference fringes appear as horizontal (cross) stripes in
In
Fringes are counted from top to bottom in
Because the angle of the phase plate is known, the introduction of carrier fringes (and taking of multiple interference images) allows a three-dimensional image of the deformed part surface to be reconstructed, based on the fringe patterns in
The carrier fringe frequency is determined by the value of the phase plate slope angle. That is, the carrier fringe frequency is determined by the phase shift, and the phase shift depends on angle θ of phase plate 13 with respect to reference (or carrier) beam B2, as shown in
In general, the minimum carrier fringe frequency should be higher than the maximum fringe frequency appearing on the strained (heated) part, as shown
The process of holographic interference analysis based on carrier fringes includes the following steps. The steps are not presented in any particular sequence, and not all are necessarily required. In particular, results can also be achieved by performing some or all of these steps in a different order.
1. Recording a hologram of an unloaded (unheated) part. In one application the part is at a room or laboratory temperature of about 20° C. or 68° F. (
2. Introducing carrier fringes by shifting the phase of a reference beam used to record the hologram, with respect to the illuminating beam. In one application, the phase is shifted by rotating a phase plate located along the reference beam. The rotation can be about 20°, or more or less, either clockwise or counter-clockwise with respect to the reference beam direction.
3. Recording an interference pattern in the form of generally parallel (e.g., horizontal), artificially-introduced carrier fringes, which cover the surface of the unloaded part. In general, an initial-state interference image is obtained, with artificially introduced carrier fringes having higher spatial frequency than in the original image (compare, e.g.,
4. Loading the part to introduce mechanical stress, for example by heating. In one application, the part is heated for about six minutes using a quartz heater. The heating is typically moderate, with an increase of about 10° C. (or 18° F.), or up to 20-30° C. (36-54° F.), resulting in a final (loaded) temperature of 30-50° C. (86-122° F.). Alternatively, the part is cooled, for example to a temperature of about 0° C. (32° F.), or colder.
5. Recording an interference pattern showing the surface of the part, as displaced by the load (
6. Recording additional interference patterns as the part returns to the initial (unloaded) state. In one example, four additional interference images are recorded over the first two minutes (i.e, one every 30 seconds), with four more over the next four minutes (every 60 seconds), three more over the next six minutes (every 120 seconds), and six more over the next eighteen minutes (every 180 seconds).
This technique produces twenty different images over a total of 30 minutes, each of which is recorded by the CCD in real time and downloaded to the processor for analysis. Alternatively, the imaging rate and analysis time vary, and a different number of interference patterns are recorded.
In each of
As shown in
In
The phase reconstructions of
I
0(x,y)=A0(x,y)+B0(x,y)×cos [ω(x,y)+(φ0(x,y)], [3]
where A0(x,y) is the background illumination at point (x,y) on the surface of the part, and B0(x,y) is the fringe intensity. Fringe intensity B0(x,y) is modulated by a cosine function based on spatial frequency ω(x,y) of the carrier fringes and phase distribution φ0(x,y). The spatial frequency of the carrier fringes is defined in the vertical (y) direction, corresponding to the vertical axis in
Loading deforms the surface of the sample part, giving:
I
m(x,y)=Am(x,y)+Bm(x,y)×cos [ω(x,y)+φ0(x,y)+ΔT(x,y)]. [4]
In this expression, Δφ(x,y) is the phase contribution due to the stress (or strain) distribution along the surface of the part. This is the component that is relevant to deformation and displacement, allowing three-dimensional surfaces to be reconstructed by determining Δφ(x,y).
In
The spatial spectra of
A one-sided spectrum is obtained by nulling (zeroing or suppressing) spatial spectrum frequencies that lie outside the spectrum peaks at carrier frequencies ω(x,y). The one-sided spatial spectrum is the spectrum of an analytic function of a complex variable formed by adding the real intensity, as defined along a particular x or y coordinate, to an imaginary component based on the Hilbert transform of the intensity. For example:
γ(y)=Ix(Y)+i×Ĩx(y), [5]
where Ĩx(y) is the Hilbert transform of real intensity Ix(y). Real intensity Ix(y) is defined along a (vertical) row at horizontal value x, and “i” is the square root of negative one. A corresponding function γ(x) can also be defined for a horizontal intensity distribution, switching horizontal coordinate x and vertical coordinate y in the definition of real intensity Iy(x) and its Hilbert transform Ĩy(x).
The phase of the carrier fringes is determined by the arc tangent of the ratio of the imaginary and real parts:
φx(y)=arctan {Im[γ(y)]/Re[γ(y)]}, [6]
where φx(y) is the instantaneous phase of the carrier fringes along the selected (vertical) row defined by horizontal value x. Analytic function γ(y) is obtained by applying a reverse Fourier transform to modified spatial spectra Fd(x,iω) of
Phase distribution φx(y) is indefinite to within a factor of 2π. This is due to the periodicity of the arc tangent function, as characterized by a phase jump from −π to +π. The data analysis is designed to automatically find and eliminate these phase jumps, both along and in between vertical rows of the corresponding interference patterns.
Phase plots (
Quantitative values Δr representing the vector displacement of points on the surface of a loaded part are thus found from phase distribution Δφ(x,y):
Δφ(x,y)=(2π/λ)×Δr(r0−r), [7]
where λ is the wavelength and Δr is the (vector) displacement between initial (unloaded) and final (loaded) state. Vector displacement Δr depends on wave vector r0 of the illuminating radiation, and base vector r in the direction of observation.
In operation of system 10, controller 17 directs laser source 11 to generate source beam L, in either continuous or pulsed beam mode. Beam splitter 12 splits source beam L into first (illuminating) beam B1 and second (reference) beam B2. Minor 21 comprises a silvered surface or other specular reflector to reflect first beam B1 onto sample part 14, and phase plate 13 shifts the phase of second beam B2 with respect to first beam B1.
Reflected illuminating beam B1′ and phase-shifted reference beam B2′ combine to form a speckle interference pattern on surface (or coating) S of sample part 14. The interference pattern can be focused onto CCD 16 with one or more lenses 20, as shown in
Phase plate 13 is then rotated to angle θ, for example by about 20°, more or less, in order to shift the phase of reference beam B2′ with respect to illuminating beam B1′. The phase delay introduces carrier fringes into the interference pattern, and a second image is generated with the carrier fringes.
Controller 17 activates heater 19 to place sample part 14 under a thermal load, and a third image is generated. Detector 16 records each of the interference patterns, and microprocessor (or computer processor) 18 analyses the corresponding optical data, mapping the phase distribution across surface S of sample part 14 in order to quantify thermal deformations and locate stress defects.
In the double-beam speckle interferometry configuration of
Intensity I of the speckle pattern formed at sample part 14 can be written:
I
0
=a
1
2
+a
2
2+2a1a2×cos(Δφ), [8]
where a1 and a2 are the (real) light wave amplitudes from beams B1′ and B2′, and Δφ is the phase difference between the two beams. After loading sample part 14, the surface experiences stress displacement and the new speckle pattern intensity is:
I
m
=a
1
2
+a
2
2+2a1a2×cos(Δφ+δ), [9]
where phase shift δ depends on the displacement of surface S on sample part 14, independent of the observation direction.
As with the holographic configuration of system 10, above, the speckle interferometer configuration utilizes a double-exposure method to analyze part deformations under relatively low temperature thermal stress. The initial part state is recorded as a speckle interferometry image with an initial phase difference Δφ1 between illuminating beams B1′ and B2′, and a second reference image is obtained with a second phase difference Δφ2.
A third (comparison) image is recorded after loading the part under thermal or mechanical stress. The third image is typically obtained without changing the relative phase, but it is also possible to use original phase Δφ1, in which case the identities of the first and second reference images are reversed.
Speckle images are generated based on the intensity registered by the CCD, and converted to sets of matrix elements corresponding to the three different speckle pattern intensities. In this analysis, the matrix corresponding to the first reference image is subtracted from the matrices corresponding to the second reference image and the (third) comparison image, and the differential signal is squared.
In regions where speckle contrast does not change substantially between the two different images, the differential signal is close to zero and the region appears dark. In regions where the change is substantial, inverse contrast regions appear as bright speckle-modulated features (see
Dark fringes are thus fringes of constant (normal) displacement, as described by:
l
y
=Nλ/2 sin(Δθ). [10]
In this expression, Δθ is the (average) angle between illuminating beams B1′ and B2′, and ly is the normal displacement. Using an optical scheme with two illuminating waveforms at different angles, system 10 can determine both normal and tangential displacements; that is, either along the direction of observation, or perpendicular to the direction of observation.
To improve visibility of the fringes, the interference pattern can be filtered to suppress high and low spatial frequencies. The results of such an interference filter are shown in
In general, normal deformations increase with stress, even for relatively modest temperature increases of 10-30° C., or 18-54° F. The deformations appear as parallel lines or stress ridges along the surface, indicating critical regions where the coating may ultimately crack or blister, depending on the magnitude of the stress and associated thermal deformation.
This algorithm provides speckle interference images and corresponding phase portraits for both normal and tangential surface displacements, making it possible to map three-dimensional surface stress distributions. The steps of the process are not presented in any particular sequence, however, and not all are necessarily required. In particular, results can also be achieved by performing some or all of these steps in a different order.
1. Illuminating a sample part with two laser beams at different angles, in order to generate a speckle interference pattern. In one application the sample part is coated, for example a turbine airfoil with a metal or ceramic thermal barrier coating. In other applications the sample part is uncoated, for example an uncoated airfoil substrate, or metal test cylinder.
2. Recording a first reference speckle interference image of the sample part in an unloaded (unheated) state. In one example the sample part is at a room or laboratory temperature of about 20° C. or 68° F. (
3. Shifting a phase of one of the laser beams. In one example, the phase is shifted by rotating a phase plate located along one of the two laser beams. The rotation can be about 20°, or more or less, either clockwise or counter-clockwise with respect to the beam direction. In other examples the phase is shifted by introducing a quarter wave or half wave plate, or via a delay.
4. Recording a second reference speckle interference image of the unloaded sample part. The speckle patterns are typically recorded by forming an image of the interference pattern on a photosensitive device such as a CCD, using a lens. Alternatively, the interference pattern is recorded by reflection or transmission onto the CCD.
5. Loading the sample part, for example by heating. In one application, the sample part is heated for six minutes using a quartz heater. The heating is typically moderate, yielding a temperature increase of about 10-30° C. (or 18-54° F.), with a loaded (final) temperature of about 30-50° C. (or 86-122° F.). Alternatively, the heating is more or less substantial, with a greater or lesser temperature difference, or loading is achieved by cooling or mechanical deformation.
6. Recording a comparison speckle interference pattern showing the surface of the sample part, as displaced by the load. In one example, the load is a thermal load and the surface of the sample part is a coating, allowing the interference analysis to detect cracks, blisters and other stress faults in the coating. Alternatively, stress-induced displacements are induced in the surface of the sample part itself, for example in the substrate of an airfoil, or in a test part.
7. Analyzing the images by converting the interference images into digital matrix form. The matrix values represent intensity for each of two reference images and the comparison image, as a function of pixel number. Matrix values for the first reference image can be subtracted from matrix values for the second reference image and the comparison image, and a spatial frequency filter can be applied to increase visibility of the resulting fringes.
8. Recording additional comparison speckle interference patterns as the sample part returns to the initial (unloaded) state. In one example, four additional speckle images are recorded over the first two minutes (i.e., every 30 seconds), four more over the next four minutes (every 60 seconds), three more over the next six minutes (every 120 seconds), and six more over the next eighteen minutes (every 180 seconds). This produces twenty different images over a total of 30 minutes, each of which is recorded by the CCD in real time and downloaded to the processor for analysis.
Alternatively, the imaging rate and analysis time vary, and a different number of interference patterns is recorded. In addition, the sample part can be thermally loaded by heating or cooling, or a mechanical load can be applied.
Phase portrait analysis also provides for additional statistical measures of the surface deformation and stress profiles, including quality indictors based on the standard deviation of the phase distribution and distribution widths based on absolute minimum/maximum ranges (in radians or in physical length) or z-score distributions, which can be used to set inspection thresholds and determine acceptable surface deformations and stress profiles at particular loads. Thus, the standard deviation and width of phase relief distribution are used as criteria for inspecting part surface quality. These methods can also be used to evaluate different coating compositions, thicknesses and application techniques, in order to increase service life and reduce maintenance costs for turbine airfoils and other precision parts with lifetime-limited applications.
While this invention has been described with reference to exemplary embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the spirit and scope of the invention. In addition, modifications may be made to adapt a particular situation or material to the teachings of the invention, without departing from the essential scope thereof. Therefore, the invention is not limited to the particular embodiments disclosed herein, but includes all embodiments falling within the scope of the appended claims.
