Failure analysis for laser beam welding

Abstract

Laser beam welding is a complex manufacturing procedure which is influenced by a multitude of parameters. Interrelationships withdraw from a deep understanding of the physical processes, since cause-and-effect relations of parameters are not clearly determined yet. Owing to a permanent rise in demands on quality and reliability of product and process, the process capability has to be assured to achieve the goal of error-free manufacturing.The processing quality is determined by parameters which have to be controlled by algorithmic methods. The number of these parameters is too large to build up control cycles. Hence, there is a need to perform a fault analysis to obtain a reduced set which can be handled. This parameter set is required to cover the major part of influence on the processing results.Dependencies, influences and correlations between input and output parameters are ascertained by interviews with experts as well as by experimental results. By these measures cause-and-effect relations can be determined. The results lead to a state model which simulates both, machining structures and process cycles. Several methods of fault observation are tested for their suitability to monitor laser beam welding. A combination of well suited methods is utilized to generate the state model. The reliability of the model will be discussed with regard to successful failure detection and avoidance.Laser beam welding is a complex manufacturing procedure which is influenced by a multitude of parameters. Interrelationships withdraw from a deep understanding of the physical processes, since cause-and-effect relations of parameters are not clearly determined yet. Owing to a permanent rise in demands on quality and reliability of product and process, the process capability has to be assured to achieve the goal of error-free manufacturing.The processing quality is determined by parameters which have to be controlled by algorithmic methods. The number of these parameters is too large to build up control cycles. Hence, there is a need to perform a fault analysis to obtain a reduced set which can be handled. This parameter set is required to cover the major part of influence on the processing results.Dependencies, influences and correlations between input and output parameters are ascertained by interviews with experts as well as by experimental results. By these measures cause-and-effect relations can be de...