Alexander F. H. Kaplan and John Powell

Luleå University of Technology, Dept. Applied Physics and Mechanical Engineering, Sweden

Welding defects have to be suppressed to maintain the mechanical strength of a product under load. For suitable choice of the process parameters operating windows can be entered where a certain welding defect can be suppressed. However, the identification of such operating windows is empirical and moreover difficult to transfer, as the process depends on 20-30 parameters and each industrial application used to be a different, thus new situation.

Spatter is among the main welding defects, causing deposited droplets as well as underfill or craters. Despite several successful measures to suppress spatter and some discoveries on the mechanism, a systematic guideline as well as a closed theory on spatter is missing, but desired. The paper presents several approaches to categorize spatter experiences in a systematic, extendable manner. One of them was applied for the 36 literature entries found. While traditional database search is based on verbal entries, such map enables for the first time to track the knowledge graphically, e.g. by the aid of the index finger and to travel through different categories where comparison with similar discovered knowledge of interest can be made. For example a certain innovative method like a twin laser spot can be quickly identified and an explanation is given how this method suppresses spatter. The mechanism can then lead to other options working similarly, e.g enlarging the laser beam.

From high speed imaging quantitative categorization is possible, e.g. of the drop ejection location or the drop kinetics. The Bifurcation Flow Chart, BFC, enables mapping of the trends and mechanisms revealed, combining even very different cases like pulsed and continuous, conduction and keyhole mode welding, or Zn-coating and pollutions. It is hoped that the here presented methods and initial mappings will be applied and continuously extended to once complete the map of and control over spatter. Better mapping will avoid parallelism of empirical experiments in research and industry and much more efficient because systematic use of data and knowledge.