Methods for Hybrid Modeling and Simulation-Based Optimization in Energy-Aware Production Planning

Abstract

Production Planning (PPC) optimization in the operation of industrial production plants enables to achieve energy savings and improve energy efficiency. To provide decision support in production planning, simulation-based methods are particularly well-suited for complex production system to assess different planning scenarios and evaluate them with regard to their energy demand. To this end, a hybrid discrete/continuous modeling approach based on the Discrete-Event System Specification (DEVS) is investigated in this thesis. Hybrid modeling allows to accurately capture material flows as discrete entities on one hand and continuous energy flows on the other hand, in the production as well as for other domains, such as heating or cooling, up to the thermal building envelope, including their dynamic interactions. A meta-heuristic optimization procedure for operative production planning evaluates energy efficiency together with other production goals (delivery tardiness, storage costs, etc.) as part of a multi-objective optimization. The procedure follows a hybrid Variable Neighborhood Search (VNS) meta-heuristic and uses the hybrid simulation for the energetic evaluation of production programs. The VNS uses special energy-related operators, for example to optimize the setup times of heating ovens, in order to save energy. Furthermore, to support the modeling process of such DEVS-based simulations in practical applications, a domain-specific model abstraction is designed, which allows to describe hybrid simulation models in an intuitive way. The method is based on a Model-Driven Engineering (MDE) process and defines a component-based meta-model for production systems. Concrete implementations can be derived from the abstract specification by means of model transformations. Proof-of-concept implementations are realized to evaluate the individual contributions and case studies from industry demonstrate the practical applicability. The hybrid simulation approach was implemented in its entirety as a simulator prototype, together with reusable model components, which can be used by practitioners to design new application models with reduced efforts. Altogether, these methods contribute to the integration of energy efficiency aspects into modern PPC systems and to increase energy efficiency in production. i Kurzfassung Im operativen Betrieb von industriellen Anlagen lassen sich durch optimierte Produktionsplanung und-steuerung (PPS) gezielt Energieeinsparungen erreichen und die Energieeffizienz verbessern. Zur Entscheidungsunterstützung bieten sich gerade bei komplexen Produktionssystemen simulationsbasierte Methoden an, die es ermöglichen, unterschiedliche Planungsszenarien zu vergleichen und hinsichtlich ihres Energieeinsatzes zu bewerten. Dafür wird in der vorliegenden Arbeit ein hybrider diskret/kontinuierlicher Modellierungsansatz untersucht, der auf der Discrete-Event System Specification (DEVS) basiert. Die hybride Modellierung erlaubt es einerseits die Materialflüsse als diskrete Entitäten, andererseits auch die zei...

An Architecture for Model Behavior Generation for Multiple Simulators

“…Durak et al [45] relate the development of an ontology for trajectory simulation to MDA. The use of MDE principles according to the steps of the life cycle of a model in the M&S domain is considered for example by Cetinkaya et al [30] as well as by Heinzl [68]. In both papers, a prototype of a software framework according to the MDA specification is described.…”

“…Cetinkaya et al [30] analyze several other works dealing with the transformation from UML to DEVS. The approach of Heinzl [68] is similar in principle to that of Cetinkaya et al [30]. For conceptual modeling, Heinzl uses a specially developed modeling language called Cubes.…”

Simheuristics: An Introductory Tutorial

“…Other researchers have utilized simheuristics to reduce costs or energy consumption in manufacturing processes. Heinzl and Kastner (2020) reduced energy costs and other production costs in an industrial bakery. They utilized a general variable neighborhood search metaheuristic and simulation to optimize the costs.…”