ENBIS-16 in Sheffield

11 – 15 September 2016; Sheffield Abstract submission: 20 March – 4 July 2016

Design and Models for the Prediction of In-Flight Particle Properties in Thermal Spraying with Additive Day-Effects

13 September 2016, 09:20 – 09:40

Abstract

Submitted by
Sonja Kuhnt
Authors
Sonja Kuhnt (Dortmund University of Applied Sciences and Arts)
Abstract
Thermal spraying technology can be employed to apply a particle coating on a surface. Due to uncontrollable day-effects thermal spraying processes are often lacking in reproducibility. This fact, combined with a time-consuming and not immediately available analysis of the quality of the coating, leads us to measure and use in-flight properties of the particles. We derive separate generalized linear models suitable for describing multiple in-flight properties based on the assumption of a Gamma distribution. These models are needed for a later optimization procedure to search for settings of the process parameters which return values of the particles in-flight known to ensure a good quality of the final coating. We show how these models can be extended to include additional day-effects. The models are to be updated on a limited number of
additional experiments on any day based on a suitable optimal experimental design. Our focus is on determinant type optimal design criteria maximizing the determinant of the Fisher information, which in case of generalized linear models may depend on the unknown parameter. As separate models are considered for the in-flight particles a design has to be determined which yields large determinants of all Fisher information matrices simultaneously. A classical D-optimal design for one component is not necessarily efficient for the other components. Therefore a new multi-objective optimality criterion is introduced which yields efficient designs for several models simultaneously. We demonstrate for the thermal spraying application that the constructed designs improve a reference design substantially.

Return to programme