Final Report Summary - PHASEMASTER (Advanced spatial phase shifting techniques and applications to non-destructive testing of large engineering components on-site)
This project aims to develop an advanced shearography system that can be used to inspect wind turbine blades (WTB) on-site. Spatial phase shift techniques and phase extracting algorithms from a single specklegram were designed so that the shearography system can work on a wind turbine tower to carry out on-site inspection. Both laboratory tests and mock-up field trials have been conducted to validate the developed technique. During the two year period, extensive work has been carried out, with the following achievements:
A spatial phase shifting shearography system has been developed. The system consists of laser source, diffuser, one slit, a Wollaston prism integrated with a linear polariser, an imaging lens and a digital camera. Blades with different sizes and different artificial defects have been tested using the shearography system when the blades are stationary. Heating gun is used to produce thermal excitation. Experiments show that the spatial phase shifting shearography system can identify subsurface defects within the blade samples efficiently.
However due to the use of slit in the setup to produce carrier fringe on the specklegram, only a small proportion of the laser illumination is utilised. To measure larger area in an open field such as on a wind tower, a temporal phase shifting shearography system has also been designed and developed. Two methods have been investigated to achieve a large field of view. One is doubling the size of the beam splitter in the Michelson-interferometer-based shearography. The other one is embedding a 4F (F stands for focus length) system in the shearography, where the imaging lens is placed in front of the Michelson interferometer rather than behind it as in traditional digital shearography. Thus, the angle of view is no longer limited by the Michelson interferometer. A three channel Piezo Phase Shifter has been employed for temporal phase shifting. This new temporal phase shifting shearography system has been applied in defect detection of a composite test board as well as metal weld components.
A number of de-noise algorithms for the shearogram have been implemented and demonstrated. The low-pass filters are based on Fourier transform, window Fourier transform and wavelet transform. Other de-noise methods for the shearogram studied include partial differential equations and principal component analysis. Algorithms for phase extraction from single shearograms with carrier fringes have been developed, i.e. Fourier transform method, window Fourier transform method, spatial phase shifting method, and a specific method based on optical flow (OL) and spiral phase transform (SPT). Numerical simulations and experimental test results demonstrated that the method based on OL and SPT is more robust and effective than those methods based on Fourier transform.
Further algorithms for phase extraction from multiple dynamic shearography fringe images have also been developed. One is the advanced iterative algorithm (AIA) based on least-squares, the other is a non-iterative algorithm based on principal component analysis (PCA). Both algorithms can extract the phase from randomly phase-shifted fringes. The AIA algorithm is simple but time-consuming. The PCA algorithm is fast and fully automated. The PCA algorithm has been successfully applied to extract a video of phase information from a video of shearograms and thus potential defects within the test piece can be identified accurately and efficiently.
Two methods of phase-changes calculation, the phase-of-differences method and the difference-of-phases method have been studied. Direct compensation of in-plane rigid body motion by rearrangement of the pixel intensities has been demonstrated by experiment. In-plane rigid body motion compensation algorithms based on the rearrangement of phase maps have been demonstrated via numerical simulation. Out-of-plane rigid body motion has also been demonstrated. Experimental result shows that the maximum out-of-plane rigid body motion tolerance is only 0.5mm. This will restrict the application of the shearography system in dynamic measurement.
A new phase compensation algorithm has been developed to compensate the rigid body rotation of a WTB on-site. In this algorithm, two spatial carrier algorithms, Fourier transfer method (FTM) and spatial carrier phase shifting (SCPS) method are developed to extract the phase from the spatial carrier shearogram. The displacement field between two phase patterns are calculated by DIC. Then the second phase pattern is rearranged, such that the phase value of the n-th pixel is replaced by the phase value of the m-th pixel if the displacement between the two positions is (n-m) pixels. Some techniques are used in the phase compensation algorithm, such as the 5x5 windowed phase filtering and spline interpolation. The phase map simulations performed have indicated that the maximum in-plane rigid body motion that can be compensated is still in the range of 0.2-0.5mm.
A variety of test samples have been used in testing: two WTB samples, a test board with six artificial defects embedded in the middle, and several metal weld plates. The spatial phase shifting shearography system worked well in detecting subsurface defects in composite samples. For smaller defects, the temporal phase shifting shearography system has been shown to work better than the spatial phase shifting shearography by analysing the video of speckle images. For the test board and weld samples, it is difficult for the spatial phase shifting shearography to detect weak defect signals from only one image, but it is relatively easy to identify the defects by looking at the dynamic fringe pattern through video playing backward and forward. For the temporal phase shifting shearography system, by applying the new phase extraction algorithm based on correlation and PCA method, it is easy to detect the defects from the phase video.
A mock-up trial was performed in which the WTB sample was suspended on a test frame such that it can move slightly due to environmental disturbances and/or air flow. It allowed us to test the robustness of the shearography system and the associated algorithms. This led to a new approach to addressing the rigid body motion, with a tolerance well beyond the existing limit of 0.2-0.5mm.
A spatial phase shifting shearography system has been developed. The system consists of laser source, diffuser, one slit, a Wollaston prism integrated with a linear polariser, an imaging lens and a digital camera. Blades with different sizes and different artificial defects have been tested using the shearography system when the blades are stationary. Heating gun is used to produce thermal excitation. Experiments show that the spatial phase shifting shearography system can identify subsurface defects within the blade samples efficiently.
However due to the use of slit in the setup to produce carrier fringe on the specklegram, only a small proportion of the laser illumination is utilised. To measure larger area in an open field such as on a wind tower, a temporal phase shifting shearography system has also been designed and developed. Two methods have been investigated to achieve a large field of view. One is doubling the size of the beam splitter in the Michelson-interferometer-based shearography. The other one is embedding a 4F (F stands for focus length) system in the shearography, where the imaging lens is placed in front of the Michelson interferometer rather than behind it as in traditional digital shearography. Thus, the angle of view is no longer limited by the Michelson interferometer. A three channel Piezo Phase Shifter has been employed for temporal phase shifting. This new temporal phase shifting shearography system has been applied in defect detection of a composite test board as well as metal weld components.
A number of de-noise algorithms for the shearogram have been implemented and demonstrated. The low-pass filters are based on Fourier transform, window Fourier transform and wavelet transform. Other de-noise methods for the shearogram studied include partial differential equations and principal component analysis. Algorithms for phase extraction from single shearograms with carrier fringes have been developed, i.e. Fourier transform method, window Fourier transform method, spatial phase shifting method, and a specific method based on optical flow (OL) and spiral phase transform (SPT). Numerical simulations and experimental test results demonstrated that the method based on OL and SPT is more robust and effective than those methods based on Fourier transform.
Further algorithms for phase extraction from multiple dynamic shearography fringe images have also been developed. One is the advanced iterative algorithm (AIA) based on least-squares, the other is a non-iterative algorithm based on principal component analysis (PCA). Both algorithms can extract the phase from randomly phase-shifted fringes. The AIA algorithm is simple but time-consuming. The PCA algorithm is fast and fully automated. The PCA algorithm has been successfully applied to extract a video of phase information from a video of shearograms and thus potential defects within the test piece can be identified accurately and efficiently.
Two methods of phase-changes calculation, the phase-of-differences method and the difference-of-phases method have been studied. Direct compensation of in-plane rigid body motion by rearrangement of the pixel intensities has been demonstrated by experiment. In-plane rigid body motion compensation algorithms based on the rearrangement of phase maps have been demonstrated via numerical simulation. Out-of-plane rigid body motion has also been demonstrated. Experimental result shows that the maximum out-of-plane rigid body motion tolerance is only 0.5mm. This will restrict the application of the shearography system in dynamic measurement.
A new phase compensation algorithm has been developed to compensate the rigid body rotation of a WTB on-site. In this algorithm, two spatial carrier algorithms, Fourier transfer method (FTM) and spatial carrier phase shifting (SCPS) method are developed to extract the phase from the spatial carrier shearogram. The displacement field between two phase patterns are calculated by DIC. Then the second phase pattern is rearranged, such that the phase value of the n-th pixel is replaced by the phase value of the m-th pixel if the displacement between the two positions is (n-m) pixels. Some techniques are used in the phase compensation algorithm, such as the 5x5 windowed phase filtering and spline interpolation. The phase map simulations performed have indicated that the maximum in-plane rigid body motion that can be compensated is still in the range of 0.2-0.5mm.
A variety of test samples have been used in testing: two WTB samples, a test board with six artificial defects embedded in the middle, and several metal weld plates. The spatial phase shifting shearography system worked well in detecting subsurface defects in composite samples. For smaller defects, the temporal phase shifting shearography system has been shown to work better than the spatial phase shifting shearography by analysing the video of speckle images. For the test board and weld samples, it is difficult for the spatial phase shifting shearography to detect weak defect signals from only one image, but it is relatively easy to identify the defects by looking at the dynamic fringe pattern through video playing backward and forward. For the temporal phase shifting shearography system, by applying the new phase extraction algorithm based on correlation and PCA method, it is easy to detect the defects from the phase video.
A mock-up trial was performed in which the WTB sample was suspended on a test frame such that it can move slightly due to environmental disturbances and/or air flow. It allowed us to test the robustness of the shearography system and the associated algorithms. This led to a new approach to addressing the rigid body motion, with a tolerance well beyond the existing limit of 0.2-0.5mm.