INTRODUCTION: With the age of the average welder approaching fifty-five, the welding industry is facing a shortage of experienced workers, with estimates on the scale of four hundred thousand men and women by 2024. This workforce gap has created an ever-growing need for automated processes that can perform complex welding applications for industries like power generation. For many of these welding applications, gas tungsten arc welding (GTAW) is the preferred method for its precise heat control, cleanliness, and consistent weld quality. The University of Tennessee, Knoxville (UTK) and the Electric Power Research Institute (EPRI) have collaborated on the development of an adaptive welding system specifically designed to address the implementation of GTAW process for advance welding research and future application in the nuclear power industry. The end goal for this machine is to develop a control system and accompanying algorithms to perform multi-pass welds without the need for interaction from experienced craft once the initial system is setup.
MOTION PLATFORM: The current research platform is designed with five degrees of freedom, with three axes dedicated to tungsten position and two axes dedicated to wire filler material position. The tungsten electrode position axes are orientated such that each axis is independently translated in the parallel, perpendicular, and vertically position relative to the weld groove. The vertical axis which controls tungsten height relative to the groove also functions as an automatic voltage control (AVC) system. It compares the actual voltage to a target value and adjusts the tungsten position to minimize the error between the two. The two wire feed axes control the vertical height and the rotational angle at which the wire is introduced into the molten weld puddle. The system also includes manually adjustable stages for both work and travel angles which could be modified to include servo drives as needed. All five stages are driven by servomotors with built in encoders that give positional feedback for each axis during the welding process.
SENSING and DATA ACQUISITION: Monitoring of the real-time welding condition is performed using various sensors. LabVIEW performs the data acquisition to both record and manage the collected data. The system includes two different optical sensor types that are used for process control and monitoring. One type, two Cavitar C300 weld cameras are positioned to view the leading and trailing edges of the molten weld puddle and the orientation of the welding wire entry. The cameras subsequently record high resolution videos of the puddle size and motion of the welding wire. The C300 utilizes laser illumination and filtering to improve weld puddle imaging in the presence of the GTAW arc. A Keyence LJ-7080 series laser profilometer is also positioned behind the torch and oriented perpendicular to the weld direction. The profilometer is used to measure the groove geometry and to produce a 3D topology image of both the groove and the solidified weld deposit. Real time values of the welding arc current and voltage, and wire feed speed are monitored and collected directly from the Liburdi P300 power supply using LabVIEW.
System motion is currently controlled using Beckhoff’s TwinCAT software while process control algorithms, data acquisition and the user interface have been developed using LabVIEW. This combination has worked for initial development and proof of concept, but a portion of the control algorithms are being transferred into TwinCAT. The TwinCAT software provides greater flexibility with control strategies, deterministic control of process sequences and timing, faster computational cycles and has a higher system bandwidth. The benefit of TwinCAT is demonstrated in recent improvements to the arc initiation process which has increased the reliability of system arc initiation from 30-40% to over 90%.
SYSTEM OPERATION: On-going work is focused on upgrading the TwinCAT and LabVIEW control systems. A demonstration of these systems will be discussed based on attempts to fill a groove weld geometry through a multi-pass welding process. Initial experiments are focused on pre-planned, open loop weld paths to identify control and process typical weld challenges. Single bead weld tests will be performed to assist in software debugging and parameter tuning of the system. This will be followed by a series of manually positioned and monitored multi-pass welds to determine and validate weld bead geometry relationships based on weld bead location, weld speeds and power supply parameters, i.e., voltage and current throughout the process of filling the weld groove cavity.
FUTURE DEVELOPMENT: Future work will focus on developing the path planning and control algorithms necessary to improve weld groove filling sequence and to reduce or eliminate typical weld defects. Work will rely on groove scanning performed on each weld pass with the Keyence profilometer. The system will evaluate bead positions and geometry, and determine the optimal tungsten position and wire entry necessary to completely fill the groove while maintaining a consistent layer height for each deposited layer.