On-line weld penetration detection and control in automated gas tungsten arc welding. Ph.D. Thesis

Thesis/Dissertation · Sat Jan 01 04:00:00 EST 1994

OSTI ID:237289

Banerjee, P

The present work was undertaken to study the feasibility of monitoring and controlling weld penetration variations in real-time using an infrared detector. Weld penetration variations induced on mild steel plates were examined with an infrared detector and the acquired data analyzed to detect weld penetration variations using a mathematical analysis. Selected weld penetration indicators were developed and used to demonstrate on-line weld penetration control. A three-dimensional solid-state, transient heat transfer model was also developed to help identify key changes in thermal distributions which could be used as weld penetration indicators. A transient three-dimensional heat transfer model was used to solve the differential energy balance for the GTA welding process. The computed temperatures obtained from the model were used to determine the variation of the isothermal map, surface temperature profile and temperature gradient with weld penetration. Good agreement between experimental and computed indicators was obtained. Gas tungsten arc welds were performed on steel plates with intentionally introduced defects such as sudden thickness changes and minor element content changes. Extensive mathematical analysis helped correlate weld penetration variations to parameters such as the peak infrared intensity, intensity gradient over a fixed distance, intensity gradient at the inflection points in the first differential of the intensity profile (linescan), computed width of the weld pool, intensity gradient at a point in the weld pool, area enclosed by the linescan and area enclosed by the linescan after compensating for background illumination. The shape of the weld pool front was also determined from a differential analysis of the infrared data. Inconsistent behavior of some weld penetration indicators led to their elimination from the selection process for the best error signal.

• Research Organization: Auburn Univ., AL (United States)
• OSTI ID: 237289
• Report Number(s): N--96-22434; NIPS--96-34128
• Country of Publication: United States
• Language: English

Related Subjects

36 MATERIALS SCIENCE
AUTOMATION
GAS TUNGSTEN-ARC WELDING
HEAT TRANSFER
MATHEMATICAL MODELS
ON-LINE MEASUREMENT SYSTEMS
STEELS
TEMPERATURE GRADIENTS
WELDED JOINTS