Quantitative in-process temperature measurement in laser metal deposition using multispectral emissivity prediction

In the additive manufacturing of metals, the solidification temperature intervals, phase transformation temperatures, thermal gradients and solidification and cooling rates are the key variables that determine the microstructural characteristics, internal stresses and possible cracks in the workpiece after the process. For the vision of a future, possibly graduated, targeted adjustment of the mechanical and surface properties of additively manufactured components, precise knowledge of the relationships between the control variables of the manufacturing process as well as the component geometry and the temperature fields during production is therefore required. A quantitatively traceable in-process measurement of surface temperature fields is therefore required to validate models and simulations of the processes with regard to the above-mentioned properties. In particular, the simultaneous temperature determination of the melt pool and the solid body, even at temperatures well below the solidification temperature, is considered to be unsolved and is to be addressed within the scope of the project. To this end, two camera-based measurement techniques for determining actual temperature developments in the process despite varying emissivity are to be researched, compared and finally merged using the example of laser powder cladding with the material AISI 316L. The project partner BAM is using the method of temperature-emissivity separation based on multispectral thermography in the medium-wave infrared spectral range. For this purpose, suitable parameterized analytical spectral emissivity functions must be developed for the different material states, which describe the actual spectral emissivity curve sufficiently well with a few degrees of freedom. In order to be able to use these in the various image areas for temperature-emissivity separation, the image areas must also be classified according to material state. In the MRT part of the project, the data from a visual RGB camera (including external lighting) will first be pre-processed to reduce disturbance variables and then segmented into regions of mathematically describable emissivity curves and mapped to the emissivity using data-driven learned prediction. The emissivity information obtained is merged with the images from a broadband medium-wave infrared camera with a wide temperature measurement range for quantitative temperature measurement.

Lars Sommerlade, M.Sc.

October 2024 - September 2027

Bundesanstalt für Materialforschung und -prüfung (BAM)

The measurements were carried out using the LMD²- Laser Metal Deposition with Integrated X-Ray Analysis system..