Smart-Digital-Twin-Framework zur Kostenoptimierung in der Recyclingindustrie
摘要
The global shift towards a circular economy requires the recycling industry to operate at peak efficiency. However, recycling processes are notorious for their complexity, characterized by highly variable input materials and multi-stage plants where settings in one stage have cascading, non-intuitive effects on all subsequent stages. Optimizing these processes currently relies heavily on operator experience and costly trial-and-error. This paper introduces a novel framework for operational profit optimization by leveraging a “Smart Twin” of the recycling process. This Smart Twin, detailed in a companion paper, is a probabilistic digital replica that models the entire plant stage-by-stage, accurately predicting output quality and yield under uncertainty. We demonstrate how this model functions as a surrogate model within a Differential Evolution optimization loop. Crucially, our framework optimizes both the process features (machine settings) on a cumulative, stage-by-stage basis and the active process depth, determining the most profitable path to maximize a defined set of objectives. We present a detailed use case from the copper recycling industry focused on maximizing the single objective of total profit, demonstrating the framework’s ability to provide distinct, optimal strategies based on changing market conditions.