TWI Industrial Member Report Summary 887/2007
By B Blakeley
Conventional film radiography is typically used in manufacturing environments in order to inspect welds on a wide range of structures. At present, the radiographic film is manually developed and inspected for indications by an operator. There is often very little time allowed for sentencing a weld and either rejecting it for repair or accepting it, and this puts a great deal of pressure on the operator to take the shot and inspect it quickly. Large numbers of radiographs and the rapid nature of the work inevitably lead to subjective errors and the missing of indications and defects. The use of powerful x-ray generators needed for film radiography requires radiological protection measures to manage the health and safety risk.
Digital radiography is emerging as a replacement for film radiography and offers greater opportunity for analysis using automated defect recognition (ADR) software. Automated defect recognition of digital radiographs of castings is commonly used, as each casting is dimensionally identical to the last. Any differences between the radiographs are only due to flaws, making the image analysis both rapid and relatively simple. In addition, a lower intensity of x-rays may be used to generate a digital image than that used for film, thereby reducing the level of radiological protection measures required.
Images from film and digital radiography of unground welds both show the uneven thickness profile of the weld cap and root as a variation in density and contrast. This can affect the interpretation of the radiograph and the ability to use ADR. In order to enhance the image sufficiently to use ADR with digital radiographs of unground welds, allowance must be made for the weld cap and root thickness profile in analysing the digital image. This work has been undertaken to address this problem.
The objective of this project was to demonstrate the feasibility of using laser profiling to digitally remove the effect of the weld-cap thickness profile from a digital radiograph of a weld. The outcome was aimed to assist radiographic interpretation and improve the applicability of ADR software.