High resolution X-ray radiography:
X-ray dynamic defectoscopy

Daniel Vavřík

The principle of the XRDD is to illuminate the sample object by X-rays during the loading process (if loading is presented). The unique relationship between the object thickness and the intensity of radiograph element is determined thanks to a precisely ‘measured signal to material thickness’ calibration by the Direct thickness calibration (DTC). Consequently, measured changes in transmission represent alterations of effective thickness of the specimen. The effective thickness changes are understood as weakening of the material by damage volume fraction (voids and cracks) and by the contraction.


Inspection of fatigue internal crack in multilayer composites joint with titanium rivets

The composite is compounded from Al-alloy and glass-epoxy layers. Two composite parts are joint by an array of  titanium rivets.

The fatigue crack is in the glass-epoxy layer and is shielded by a rivet head and Al-alloy layer, in this way it is impossible to observe this crack optically on the surface. However this internal crack is radiographically visible thanks to the high dynamic range of pixel detector. A sequence of six hundred snapshots with one-second exposition were measured (tungsten tube at 70kV).

Geometry of the specimen

Fatigue crack observed in the glass epoxy interlayer


Simultaneous radiographic observation of damaging and optical measurement of 3d surface displacement evolution in loaded specimen

Material damage is investigated by means of the “X ray Dynamic Defectoscopy” technique. To understand related physical processes, the surface plastic strain field is investigated by the optical Method of Interpolated Ellipses while the out-of plane displacement field is inspected simultaneously by the optical Coded photometric stereo method. Related experiments have been realized using unique experimental setup which consists of microfocus X-ray source, compact loading device, CCD camera, LED red-green-blue lights and single X-ray photon counting device Medipix-2.

The Method of Interpolated Ellipses (MIE) is a technique based on the optical monitoring of deformations during loading processes of hexagonal grids of dots deposited on the surface of the monitored specimen. We assume that the surface deformation of a continuous material is approximately homogeneous inside a circle with initial radius r. Loading of the specimen will deform a circle on the surface into an ellipse. Each ellipse is interpolated by six neighboring dots of a hexagonal grid. Knowledge of the ellipse parameters yields the magnitude and direction of the principal in plane strains.

The Photometric Stereo (PS) method makes use of the close relation between the relative lightness of uniformly illuminated surface and the angle of direction of light. This situation means that knowing the illumination geometry we can determine the slope (normal) of the surface at every point. Surface topography is obtained as an integration of this normal field. This fact can be used for the purpose of out-of plane measurements of loaded object deformations. Three images required by PS using three light beams illuminating the given surface from different directions are sufficient for the determination of both x and y slopes at any studied point of the surface regardless of its color variations. Knowledge of both slopes is advantageous for the reconstruction integration especially in the case of locally irregular surfaces. For this reason three light beams have been used in our experiment. The Coded Photometric Stereo (CPS) is the enhanced PS method utilizing Red-Green-Blue (RGB) light coding. Monochromatic RGB lights positioned around the observed area produce directional illumination. Three different scenes coded in one compose image can be separated using standard RGB color channels of the digital camera we use. Out of plane displacement is measured by CPS using these scenes. The composed RGB image serves for the measurement of the in-plane displacement field by the MIE. As the 3D displacement fields can be determined from one image, the simultaneous combination of both methods is suitable also for dynamic experiments.

For the purpose of radiographic measurements, a new transferable 25 kg and highly stiff loading device was developed. This device is equipped with four stepper engines ensuring the symmetrical loading of specimens with relatively stable position of the observed area in the X-ray beam. Grips displacement is realized by the screws rotation using stepper motors with harmonic transmission. The resultant transmission ratio is extremely high and allows precise and very slow loading. The loading force is recorded by two load cells while grip displacement is measured by two extensometers. The loading force capacity of the device is 100 kN and weights only 25 kg with dimensions 377x343x190 mm. These parameters allow to fix the loading equipment onto a PC controlled motorized stage during measurements.

Simultaneous optical and X-ray imaging of sample during loading.

The specimen was loaded in uni axial tension by grips displacement with velocity 0.4 m/sec until specimen failure occurred. The complete experiment took 1:20 h time. Data from extensometers provided grips displacement and from load cells, measuring loading force, were scanned each 200 ms; see Fig. 3 for recording of “loading force – grips displacement” dependence (loading record). X-ray images were taken every second. Optical images were acquired every 20 sec. 3D strain field and X ray images were analyzed in post processing phase.




Experimental setup including the X‑ray tube (left), loading equipment fixed on the loading stage (middle) and the X‑ray detector Medipix (behind the carousel with calibration filters on right). Optical camera is behind X‑ray tube on left, RGB lights are fixed on the supporting frame of the loading equipment.

Loading records

Optical record of the loaded specimen

XRDD record of the loaded specimen


First principal strain Second principal strain Out of plane displacement

Radiographic observation of 3D displacement field using X-ray Digital Image Correlation technique

3D displacement field and consequent damage evolution during loading of flat specimen was observed by recognition of specimen grain induced structure. Grains of the observed material have elongated shape with typical dimensions of 30x120 microns. Recognibility of details with 15 microns size arises from geometrical magnification of 3.6 which has been used in our experimental setup. Grain specimen structure is exhibited by attenuation variations which are inducted by non homogenous chemical concentrations in material observed. These variations are in proportion of 0.5 % to average attenuation which corresponds to geometrical thickness of observed 5 mm thick flat specimen. Thanks to natural poisson distribution and required signal-to-noise ratio minimally 100.000 of X-ray photons has to be detected per X ray imager pixel. This extremely high dynamic range is possible thanks to zero dark current, noiseless and absolute linearity of X ray imager used. Practically unlimited dynamic range is provided by this imager thanks to these properties where required dynamic range is acquired by summing of sequence of 13 bit radiographs.

Assuming flat geometry of the loaded specimen, the in plane deformation is evaluated from radiographs using the image correlation technique. This technique is quite known for materials with distinct inner structure such as reinforced composites. However, this approach is applicable also for metals where the grain structure becomes recognizable in high quality radiographs. Inner material structure can be visualized thanks to an extremely high dynamic range of used pixelated X ray imager and proper data processing.

The grid of control points is generated based on the first image. Tracks of the grid points are calculated over the all taken radiographs with a subpixel accuracy. Submicron resolution of the control points position is reachable. The related out-of plane displacement field is measured thanks to an accurate ‘radiograph intensity to material thickness’ calibration, which is done for each detector pixel separately. The resultant 3D displacement field is determined using actual and reference radiograph.


Tracing paths at the maximal loading force level.

Dz isolines at the maximal loading force level


x displacement at the maximal loading force level Dy displacement field

Other Applications:

- Biomedical imaging


- X-ray radiography
- Tomographic reconstruction
- Phase sensitive imaging
- Neutron radiography
- Radiography with heavy charged particles


- Pixelman: Beam hardening correction
- Tomographic reconstruction
- High precision positioning