Effect of fatigue crack orientation on the sensitivity of eddy current inspection in martensitic stainless steels Hamid Habibzadeh Boukani, Ehsan Mohseni, Martin Viens Département de Génie Mécanique, École de Technologie Supérieure (ETS) June 6-8, 2017 Quebec City, Quebec, Canada
Outline Introduction Experimental procedure and data analysis FE modelling and analysis Results and discussion Conclusions 2/21
Introduction Risk assessment programs material properties loading conditions Damage tolerance models flaw characteristics Nondestructive testing (NDT) methods Reliability quantification by probability of detection (PoD) a versus a 3/21
Introduction Martensitic stainless steels High strength, hardness, and toughness Moderate resistance to corrosion Low cost Bearings Shafts Compressors gas generators valves Eddy current testing (ECT) Principal degradation FATIGUE widely used method for the detection of surfacebreaking cracks High S/N eddy current split-d probe 4/21
AISI 410 Objectives In the framework of a PoD study of automated ECT, a versus a at 3 different crack orientations with respect to the scan direction, as the influential parameters for the PoD evaluation, is studied Due to limitations in crack length interval in experiments finite element modelling (FEM) is employed to expand the extent of our study to larger sizes and thus to gain a better insight into this observation 5/21
Outline Introduction Experimental procedure and data analysis FE modelling and analysis Results and discussion Conclusions 6/21
Test unit and samples used in the study Nortec 500S along with a reflection differential split-d probe are used The probe s frequency range is 500kHz-3Mhz Starter flaws using electrical discharge machining (EDM) process on the surface of samples cyclically loaded in order to grow fatigue cracks out of the starter flaws. Samples containing fatigue cracks of 0.76 to 2.95 mm in length According to destructive tests, Their depth linearly increases with their length. Depth = 0.0006 + 0.3475 Length, r 2 = 0.9845 0.61 mm 1.89 mm 7/21
ECT automated scans and signal analysis Calibration on a reference flaw: device gain the impedance plane angle perpendicularity of the probe to the sample s surface Initial lift-off of 0.03 mm Raster scans at frequencies of 500 khz and 1 MHz Scan index of 0.5 mm Gains are compensated for each axis signals characteristics are extracted ECT signal length (V pp ) : main parameter under study 8/21
Outline Introduction Experimental procedure and data analysis FE modelling and analysis Results and discussion Conclusions 9/21
3D model and material properties used for the assembly of the probe and sample 3-D modeling in Comsol multyphysics: A half-scaled CAD model for orientations of 0 and 90 owing to their plane symmetry Full model for orientation of 45 Cracks represented by semi-elliptical notches having 0.02 mm opening Dimensions of the probe s Interior components according to X-ray tomography reconstruction Initial lift-off of 30 μm Shield Coils Cores Material properties: measurements and data sheets Component Relative permeability Electrical conductivity Cores and shield 2500 1(S/m) Sample 300 1.9e6(S/m) 10/21
Physics, mesh and solver Physics: MF physics within AC/DC module Multi turn domains for coils Magnetic insulation boundary condition for encompassing air domain ( ( A)) / 0 r ( j 0 r ) A Je j 1 Mesh: Z ( VR 2 VR 1) / ID Second order tetrahedral elements 8 boundary layer mesh on the surface of the sample Each layer has the thickness of first standard penetration depth Finer elements for the notch geometry Solver: Iterative stationary solver 2 11/21
Details of simulated scans 1.3 mm displacements of the probe along the scan path Probe s displacement increments of 0.1 mm 90 Start of the scan End of the scan 0 mm 1.3 mm Probe is centered by the notch at the beginning of the scan Orientation Notch length Variations(mm) Steps (mm) 90 2-6 0.5 45 1.5-4.5 0.5 0 0.5-3 0.5 45 0 0 mm 1.4 mm 0 mm Depends on the notch length 12/21
Outline Introduction Experimental procedure and data analysis FE modelling and analysis Results and discussion Conclusions 13/21
Normalized Signal Amplitude Measured signal amplitude versus crack length Normalized Signal Amplitude 1 0.8 0.6 90 deg 45 deg 0 deg 1 0.8 0.6 90 deg 45 deg 0 deg 0.4 0.2 0 0.5 1.5 2.5 3.5 Crack Length (mm) 0.4 0.2 500 khz 1000 khz 0 0.5 1.5 2.5 3.5 Crack Length (mm) The same behaviour at both frequencies Amplitude is independent of the orientation for crack length below 1.8 mm Amplitude changes versus length variation becomes plateau after 1.8 mm for 0 orientation 14/21
Im (ΔZ) (Ω) Verifying simulated signals by comparing them to measurements Im (ΔZ) (Ω) 0.0E+0-5.0E-2-1.0E-1-1.5E-1-2.0E-1 Simulation Measurement 0 orientation 0.0E+0-5.0E-2-1.0E-1-1.5E-1-2.0E-1-2.5E-1 L=2.94 mm -3.0E-1-4.0E-2 6.0E-2 1.6E-1 Re (ΔZ) (Ω) Good amplitude agreement -2.5E-1 L=1.50 mm -3.0E-1-4.0E-2 6.0E-2 1.6E-1 Re (ΔZ) (Ω) Shape discrepancies : Deviation of notch geometry from the fatigue crack Material properties Probe manufacturing imperfections 15/21
Normalized signal amplitude Normalized signal amplitude Simulated and measured signal amplitude versus crack length 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 deg 90 deg 45 deg Simulation 0 2 4 6 8 Notch length (mm) 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 deg-sim 45 deg-sim 90 deg-sim 0 deg-exp 45 deg-exp 90 deg-exp 0 2 4 6 8 Length (mm) For each orientation, Amplitude variations becomes less than 5 % after a certain notch length: 0-2.5 mm 45-3.5 mm 90-4 mm 16/21
Relative position of the probe and notches at which the amplitude is maximum 2 mm 1.5 mm 1.0 mm 0.5 mm Orientation 0 17/21
Induced current density distribution as the notch length varies - 0 Notch length = 0.5 mm Notch length = 1.5 mm Notch length = 3 mm 18/21
Induced current density distribution as the notch length varies - 90 Notch length = 2 mm Notch length = 4 mm Notch length = 6 mm 19/21
Outline Introduction Experimental procedure and data analysis FE modelling and analysis Results and discussion Conclusions 20/21
Conclusions Depending on the flaw orientation, the signal amplitude increases with the crack length up to a critical (flaw length)/(drive coil diameter) ratio (L/D) which is specific to the flaw orientation. This critical ratio grows as the orientation increases from 0 to 90 The variation of the signal amplitude as the crack length increases is almost independent of the crack orientation until a L/D value equal to the unity and then the slope of amplitude versus crack length variations becomes orientation dependent. Accordingly, the probability of detection of fatigue is almost independent of crack orientation once the L/D ratio is below the unity. The results of our FEM study show a good agreement with the measurements outcome in terms of amplitude. Therefore, this model is a reliable means to carry out model-based studies of probability of detection. 21/21
Thank you for your attention