Parallelization methods

Abstract

The complexity of modern energy systems poses significant challenges on how these systems are planned, designed and operated. The design of each part (subsystem, component, algorithm) is a challenge due to the interactions and dimensions of the problem. This challenge cannot be simplified via de-coupling without risking the loss of essential dynamic behaviors. The impact of individual elements on the system, and vice versa, the impact of the system on individual elements, may not be inferred analytically, and the traditional design spiral maybe inadequate. In this context the use of numerical simulation tools becomes a critical need. Simulation is already a fundamental tool that supports design and analysis in many engineering fields. At the same time, despite a long history of use, traditional approaches and commercial tools are severely limited when treating complex, temporally and spatially distributed systems such as modern energy systems, partially because still largely based on serial execution algorithms. With time scales that span over 10 orders of magnitude and with systems of very large size that cannot be easily partitioned, high parallelizable simulation methods that ensure a computationally effective and scalable solution are needed.

Authors 1

1. Andrea Benigni corresponding Aachen

   RWTH Aachen University · Forschungszentrum Jülich

   Affiliation as printed

   Department of Mechanical Engineering

   Institute of Energy and Climate Research: Energy Systems Engineering (IEK-10)

   Juelich Research Center

   RWTH Aachen University

   Institute of Energy and Climate Research: Energy Systems Engineering (IEK-10), Juelich Research Center, , Germany