Ensuring critical welded joint quality and repeatability is largely dependent on robust, well-designed Welding Procedure Specifications (WPS). Highly skilled manual welding engineers automatically recognise many imperfections, adjusting their responses according to inputs from vision, smell and sounds made during the welding process.
Unfortunately, exceptional human ability does not guarantee performance when less predictable influences occur during welding processes. Human error and materials imperfections can result in defective welds for critical applications, commonly attributed to material surface impurities and contamination.
Fault detection is problematic; the only finite method of weld testing is destructive testing which is not applicable to final product verification. Quality assurance and control is used to guarantee the welding process repeatability by production of a Procedure Qualification Record. This often-lengthy approval process restricts welding technology and materials application advancement.
An alternative method of testing is the detection of flaws and defects in real-time to allow immediate process corrections. Development of real time welding evaluation instrumentation requires welding process parameters measurements combined with high-speed data processing. This real time monitoring and evaluation produces a weld defect fingerprint used to determine quality.
We aim to highlight variations found in welding process quality using real-time monitoring and assess if it is within the acceptable standards for nuclear applications. To achieve this, we first must understand the human welding engineer using data taken from a series of manual weld trials. The trials use a common welding operation found in nuclear reactor pressure vessels. Reference data comparisons are made using identical trials with robotic welding equipment. Trial comparison results indicate that real time evaluation of welding processes detects flaws in weld quality. We then demonstrate how applications of welding process parameters are exceptionally effective methods for the control of robotic welding applications.
This paper can be accessed via the Springer website. [Paywall]
Part of the Advances in Intelligent Systems and Computing book series (AISC, volume 606)
Presented at the International Conference on Applied Human Factors and Ergonomics (AHFE), 2017