Non-intrusive instrumentation and estimation -Applications for control of an additive manufacturing process

Doctoral thesis, 2015

For integration of additive manufacturing into industrial production, there is a need for capable yet robust automation solutions. Such solutions are to ensure consistent process outputs, both with regard to deposit geometry and material properties. In this thesis, instrumentation and control solutions have been investigated for the laser metal wire deposition additive manufacturing process. This particular process is promising with regard to e.g. high deposition rates and negligible material waste. However, due to its inherent dynamics, it requires automatic control in order to prove competitive. A large number of process parameters affect the resulting quality of the deposit. Successful control of these parameters is crucial for turning laser metal wire deposition into an industrially tractable process. This requires relevant and reliable process information such as the temperature of the deposit and the positioning of the tool relative to the workpiece. Due to the particular requirements of instrumenting the process, only non-intrusive measurement methods are viable. In this thesis, such measurement solutions are presented that advance automatic control of the laser metal wire deposition. In response to the need for accurate temperature measurements for the process, a new temperature measurement method has been developed. By adopting the novel concept of temporal, rather than spectral, constraints for solving the multispectral pyrometry problem, it opens up for temperature measurements which compensates for e.g. an oxidising deposit. For maintaining a good deposition process in laser metal wire deposition, control of tool position and wire feed rate is required. Based on measurements of resistance through the weld pool, a simple yet well performing control system is presented in this thesis. The control system obtains geometrical input information from resistance measurements made in-situ, and feeds this information to an iterative learning controller. This results in a robust, cheap and practical control solution for laser metal wire deposition, which is suitable for industrial use and that can easily be retrofitted to existing equipment.