Robust Multi-Objective Optimization in Product Development With Respect to User-Scenario and Manufacturing Uncertainties

Abstract

In product development, user-scenarios are a way of tailoring requirements to defined customer groups. Furthermore, a product design often involves multiple conflicting objectives that are analyzed within an iterative process. The models typically used for the analysis often do not accurately reflect the real-world representation. This can be alleviated by finding robust product designs. While usually uncertainties due to manufacturing tolerances are investigated, we additionally consider uncertainties in the user-scenario. Therefore, we present a robustness evaluation in a multi-objective numerical optimization in product development. For this, we consider manufacturing tolerances using an adjusted Latin Hypercube Sampling as well as deviations in the user-scenario by means of a Gaussian distribution. In the case study, we present the robust development of a customer specific coffee machine, where we show the robustness evaluation and the impact of the proposed adjustments. The advantage of the presented process is a product design tailored to the customer's requirements under specified uncertainties. In addition, this enables a time benefit in the product development due to the automated analysis used in the optimization.

Systematic Optimisation Process for an Ebike Drive Unit in a Highly Variable Environment

“…Applying this method to obtain the objective function of a sub-system can, of course, lead to a large number of influencing parameters that can either only be evaluated with enormous computational effort or cannot be integrated into the model of the sub-system at all. Especially for the consideration of uncertainties due to statistically scattering influences and their necessity to consider them holistically for the robust design increases the parameter scope significantly (Wolniak et. al 2021).…”

“…Technical products must be developed in a way that the required properties are sufficiently fulfilled for the relevant applications and in the associated environments (Weber and Husung, 2016). For the optimization of demanded properties such as fatigue strength, it is crucial to consider the entire load spectrum resulting from the different applications in order to minimize uncertainties during the design process (Wolniak et. al 2021).…”