Licentiate thesis on efficient modelling and simulation of district energy systems

Johan Simonsson, from Flexi-Sync partner Luleå University of Technology, has worked on modelling and simulation within the project and he will present his licentiate thesis Towards Efficient Modeling and Simulation of District Energy Systems online on May 6. Welcome to listen!

Towards Efficient Modeling and Simulation of District Energy Systems

Dynamic simulation of district energy systems has an increased importance as an aid in the transition towards renewable energy sources, lower temperature district heating grids, and utilization of waste heat from e.g industrial plants and data centers. Physics-based models using equation-based modeling languages are commonly used for use cases such as grid design and validation. These models are in general too complex and computationally expensive for long term simulation runs, or for optimization and control.

Using specialized models for each use case on the other hand, causes redundant work, and the models become difficult to update and maintain. The aim of the thesis is to reduce the gap between these model paradigms, towards computationally efficient models that can be adapted for various use cases.

In the first of the three research papers composing the thesis, the experiences, challenges and lessons learned from city-scale simulation of district heating grids are presented. On the basis of a case study and literature, research gaps are identified, and relevant research directions are suggested.

In the second paper, a robust and computationally efficient method for prediction of heat load in residential buildings is proposed. The prediction uses a nominal model for the prediction of outdoor temperature dependent space heating load, combined with a latent variable model for the residual load, mainly for generating hot tap water. The validity of the prediction is shown on a multi-dwelling building located in Luleå, Sweden.

In the third paper, a probabilistic Gaussian process model is trained to emulate the temperature dynamics of a district heating pipe. A state-of-the-art physics-based model is used as a reference for training and validation, and the kernel is derived from the underlying physics. It is shown that model can both emulate the thermal dynamics of the physics-based model and propagate the uncertainty of the inputs.

The results indicate that these approaches are two viable ways towards efficient modeling adaptable for a wide range of use cases. Based on the results from the research papers, future research is suggested.

Date/Time: 2021-05-06, 13:00 CET

Online link: https://ltu-se.zoom.us/j/67886093079

Link to the thesis: http://ltu.diva-portal.org/smash/record.jsf?pid=diva2%3A1536765&dswid=3711 External link.