Towards a federated approach to energy systems modeling

Reinhart, Zachary
Major Professor
Kenneth M. Bryden
Committee Member
Journal Title
Journal ISSN
Volume Title
Research Projects
Organizational Units
Mechanical Engineering
Organizational Unit
Journal Issue
Mechanical Engineering

This dissertation presents a novel extension of the federated model set concept proposed by Bryden (2014) to the challenge of integrated modeling of advanced energy systems. The growing complexity of such systems along with the desire to incorporate modeling earlier and more deeply into the the design and development process requires the creation of flexible integrated modeling capabilities. In this dissertation, these capabilities are implemented for energy systems modeling through the federation of stateless component model microservices. Specifically, the design and architecture of a federated modeling software environment for the development of integrated models of advanced energy systems is detailed. This software environment is designed as a microservice architecture, and includes support for execution of simulation workflows through a system of state-less component microservices, including support for coordinated differential equation solving and the creation and use of subassemblies of components. This environment is demonstrated with the creation of a systems model, or federated model set, of the Hyper facility, an advanced power plant testbed. The process by which an existing systems model can be disaggregated and made into stateless components is defined and demonstrated with an existing systems model developed by Tsai (2007) of the Hyper facility. The results and performance of this federated model set are compared to the original systems model by Tsai (2007), showing equivalent results to the original model with significantly improved flexibility and extensibility. This extensibility and the suitability of the federated modeling environment as a development platform are demonstrated with the development of a turbine modeling component based on machine learning techniques. This component shows compatibility with the other differential equation based components and can be interchanged with the original turbine model component freely, even while a simulation is running.