Acquisition setup
Direct path · emitter → reflector → receiver
Bevel faces reflect directionally; crack tips diffract more broadly; pores scatter over more directions. These are educational patterns, not calibrated amplitudes.
Grid density adjusts automatically with array size. 64 elements use 4,096 FMC pairs at 0.2 mm and can still be slow on phones; finer pixels do not increase acoustic resolution.
Mode travel times use 28 direct, half-skip and full-skip paths in an ideal 25 mm flat homogeneous plate. Reflections are modeled at the backwall; converted paths use L/T velocities and least-time ray geometry. Waveform strengths are illustrative, not validated scattering, probe directivity or mode-conversion amplitudes. No wedge, real FMC acquisition or code-qualified sizing.
TFM reconstruction Ready
Position the probe
Drag the sliders or the scan map to position the probe manually. Moving the probe reacquires synthetic FMC data.
Probe raster map — top-down X/Y position. Tap or drag to move the array; gold line shows the weld axis. Use “Recenter selected training case” to recover a lost indication.
Tap or click the image to read out a location. Image is an educational delay-and-sum reconstruction, not an inspection-grade instrument display.
Interpretation challenge
Reconstruct the data to begin the exercise.
Teaching model and limitations: One shared synthetic FMC waveform set contains idealized contributions from multiple direct, half-skip and full-skip routes, using mode-dependent travel times and deliberately illustrative mode strengths. Backwall reflection time follows a flat-wall least-time model for longitudinal and shear segments. Re-selecting a mode reprocesses the same synthetic FMC data unless acquisition settings change. A-scans are synthesized from ideal point reflectors, Gaussian-modulated pulses and deterministic noise, then reconstructed by delay-and-sum. Peak image intensity depends on sampling, aperture, gain and noise. The simulated flaws use point/microfacet waveforms with qualitative orientation-sensitive reflection and tip-diffraction weighting, not validated elastic-wave scattering coefficients or amplitude calibration. The defect response weights include approximate specular orientation, broad tip diffraction and diffuse volumetric scatter. Fixed amplitude comparisons use an arbitrary teaching reference, not dB FSH; an auto-normalized image is useful for inspecting faint detail but is not suitable for comparing mode strengths. The contact-array abstraction assumes both L and T contributions can be excited and received; actual probe/wedge and polarization constraints differ. Single-V defects are simplified point/microfacet responses mapped from the 14 PAUT training cases; raster motion changes element position and simulated along-weld intersection, but this is not full 3D beam scattering or qualified defect sizing. These images cannot establish probability of detection, real flaw size or code acceptance. Consult approved inspection procedures and actual FMC data for field work. No copyrighted course diagrams or acquired inspection images are reproduced.