SUBJECT: Ph.D. Proposal Presentation
BY: Matthew Rogge
TIME: Thursday, January 31, 2008, 3:00 p.m.
PLACE: MARC Building, 401
TITLE: In-Process Sensing of Weld Penetration Depth Using Non-Contact Laser Ultrasound System
COMMITTEE: Dr. Charles Ume, Chair (ME)
Dr. Nader Sadegh (ME)
Dr. Ye-Hwa Chen (ME)
Dr. George Vachtsevanos (ECE)
Dr. Jennifer Michaels (ECE)


Gas Metal Arc Welding (GMAW), a widely used method for joining structural members, generates large time and material costs associated with the disposal and remanufacture of defective parts. These high costs prove the importance of weld quality and control. The development of non-contact, non-destructive, in-process weld quality sensors seeks to reduce defective weld production and realize closed-loop GMAW process control. Previous research has produced an ultrasonic weld penetration depth measurement method based on the time of flight (TOF) of the Rayleigh Generation Longitudinal to Shear (RGLS) wave path. This RGLS TOF method performs very well when the temperature of the material is constant and unifrom. When non-uniform and uncertain temperature fields are present, the RGLS TOF technique’s performance diminishes. The objective of this research is to develop an in-process extension to this method. The mechanism through which the Rayleigh wave is generated on the far surface of the specimen will be investigated analytically and validated experimentally. The applicability of the underlying assumptions for the RGLS TOF method will be determined and the method updated if necessary. An automated measurement system will be developed and optimized for in-process measurements. A nonlinear dynamic model will be developed to estimate the penetration depth based on process inputs and online measurement. Successful completion of the research outlined in this proposal will yield a new in-process weld penetration depth measurement method. The method will be directly applicable towards real-time GMAW control. Using this method, weld quality can be ensured and improved in industrial settings, resulting in reduced material waste and associated costs. In addition, the neuro-fuzzy dynamic compensation technique can be extended towards sensing of other processes (such as chemical batch reactions) in which process dynamics affect the performance of monitoring sensors.