Ilaria Corni and Nicola Symonds look at the application and use of X-ray computed tomography as a non-destructive volume imaging technique to support engineering failure analysis.
Microfocus X-ray computed tomography (μ-CT) is a non-destructive volume imaging technique that uses X-rays to generate high-resolution 3D datasets from solid objects observing their internal and external structure. Many single radiographs require large computing power and time, with some scans taking more than 24 hours to complete. Once produced, the volume data can be manipulated – cropped, sliced, reoriented – for inspection in a virtual space.
This technique works in a similar manner to a medical CT scanner – the same physics and mathematics principles are applied to both modalities. A typical μ-CT scanner comprises an X-ray source, a rotation stage and an X-ray detector. To acquire a tomogram, the specimen is illuminated using X-rays in a full 360° rotation, during which radiographs are captured at multiple orientations – typically in the order of thousands. Unlike a conventional clinical X-ray CT system, where the source and the detector rotate around the specimen to capture the radiographs, in μ-CT the illumination of the object is achieved by rotating it around a set centre of rotation keeping the source and the detector static.
Each radiograph expresses the decrease in X-ray intensity along a linear path, which is dependent upon a multitude of factors, including the X-ray energy, material electron density and optical path length. Upon completion of the acquisition, reconstruction algorithms are employed to collate the information captured by the radiographs into the final volume representing the object. A typical CT volume comprises a number of equidistant virtual ‘slices’, which correspond to the object’s structure along a certain plane. 3D pixels, called voxels, which depending on the acquisition protocol may or may not be isotropic (for example x = y = z) compose each CT slice. When isotropic voxels are used, arbitrary ‘virtual’ serial sectioning of the object becomes possible. In broad terms, the grey levels of each voxel in a CT slice correspond to the material density (electron density) of corresponding unit of volume, therefore, a dense object will appear bright and a less dense object will appear darker.
A multitude of laboratory CT scanning systems support a wide range of sample sizes with imaged volumes from a few millimetres in cross-section up to 1.5m x 1m x 1m, and spatial resolutions down to around 200nm. The larger and denser materials result in lower resolution scans. The μ-VIS X-ray imaging centre, muVIS, based at the University of Southampton, UK, houses Europe’s biggest high-energy μ-CT scanner. Computed tomography is an invaluable tool for failure investigations as it can provide non-destructive information about the condition of the inside of the failed object.
Two injectors from a V12 marine diesel engine used in yachts or sport fishing boats were investigated using μ-CT. Injector A2 showed a significant amount of contamination on its tip and would open only when a pressure of 280 bar was applied, while injector B2 would not open after applying a pressure of 400 bar. The tip area of the contaminated A2 injector was scanned with a 5µm resolution, producing a 3D volume, which was ‘virtually’ sliced and examined to determine the extent of the contamination. From the scan it was clear that the density of the contaminant was significantly lower than that of the metallic nozzle, tip and needle. A single slice extracted from the volume unmistakably displayed the level of contamination present across the tip – four out of eight nozzles were significantly blocked by a deposit that had solidified across the surface. This deposit would hinder the release of fuel and it was thought to explain the malfunction of this injector. µ-CT confidently resolved this problem – if the tip was cut and embedded, part of the deposit could have broken down, partially destroying crucial evidence and compromising the investigation.
The B2 injector was inoperable under a pressure of 400 bar. A fast scan with a 70µm resolution was used to determine the internal components and understand the operating mechanism of the injector. A single radiograph revealed a spring-loaded needle valve that closed or opened the nozzle holes in the tip of the injector.
Based on this, it was hypothesised that a spring failure of some form could prevent the correct operation of the injector. The scan of the B2 injector was scrutinized for unusual features that could indicate problems. A shadow was noted adjacent to one of the active coils of the spring. Based on the observed shadow, a higher resolution isotropic-voxel scan was performed consisting of 15 slices every 0.6mm with a resolution of 56µm. This scan was reconstructed into a 3D volume revealing the suspected foreign object. It was possible to determine that the object was flake shaped, with its shortest side (through thickness <0.5mm) in the radial direction of the spring and the longest (>2mm) following the circumference of the coil preventing the operation of the needle valve.
A destructive strip and examination of the injector subassemblies was directed from the CT data. For example, the best location to cut open the injector was selected to cause as little damage as possible. Upon opening the injector, a flake of debris at 3mm x 1.8mm x 0.4mm was found, as expected. Further results indicated that this was the same material that had coated the A2 injector, and had travelled back through the common line into the overflow of the B2 injector.
μ-CT was also applied during the investigation of a failed bronze centrifugal impeller from the seawater pump of a military ship. The impeller was from one of three main high-pressure seawater pumps that maintain a constant supply of high-pressure seawater piped around the ship. The impeller is driven in rotation by a shaft and the liquid between the blades is subjected to a centrifugal force – flowing rapidly from the centre of the impeller to the outer edge, if it is not opposed by a pressure at the outlet. The impeller comprises seven blades and five balancing holes. The balancing holes showed evidence of severe wear. The leading edges of each of the blades displayed thinning and wear and the blades and the back sloping surface that blended into the central key way collar were rough with pitting holes.
µ-CT was employed to examine the internal surfaces of the impeller without the need to cut through the damage. A scan was conducted to visualise the damage on the blades, as it was deemed they constituted the part more heavily affected by wear. The whole impeller was too large to allow for sufficient X-ray transmission and therefore some regions of the structure were excluded from the scan. Moreover, a detector designed to eliminate detection of scattered radiation that typically occurs when ‘hard’ X-rays interact with metals causing imaging artefacts was used.
The impeller was scanned taking radiographs every 0.3o, obtaining a resolution of 122.2μm. A 3D volume was created showing the impeller without the suction and discharge flanges. By circling around in the virtual space all the surfaces could be examined and assessed without the need for destructive cutting. The 3D volume showed the central inlet area of the hub and clearly displayed the extent of pitting damage on the blades. Virtual cropping of the blades provided further access to the inside of the impeller to analyse the worn surfaces. The high-detailed view of the morphology of the damaged impeller revealed clear evidence of severe localised cavitation attack, with numerous steep-sided pits that in places coalesced to form a honeycomb-like structure. Using this non-destructive technique, it was possible to determine that vaporisation cavitation was the dominant mechanism occurring in the failure of this impeller. Computed tomography could also be employed to scan a new impeller identical to the inspected unit to determine the true loss of form of this impeller due to cavitation and particle erosion observed during use.
X-ray computed tomography is a powerful non-destructive tool able to provide invaluable information not only on the surface but also on the entire volume of an object. This technique is very useful in failure analysis investigations as it preserves evidence, supplies information about internal conditions and allows virtual inspections.Dr Ilaria Corni is a practicing trobological researcher, specialising in micro CT fractography. Dr Nicola Symonds has 20 years experience in the field of forensic materials examinations and tribology. Ilaria and Nicola are consultants within nC2 - a business unit providing services for industry at the University of Southampton, UK.