Master's Thesis Lisa Karber


Investigation and comparison of different modelich approaches for thermal networks

Heat loss Copyright: EBC Heat loss of different pipe simulations under consideration of the mass flow

Thermal grids have been in use for more than 120 years and are currently experiencing a new boom, especially due to the EU's climate goals. By means of modern simulations, the use of new insulation materials and high-quality pipe materials, currently planned heating and cooling networks should be designed for the future. The networks are becoming more and more complicated due to the large number of heat feeders and the branched pipe system. New, low-computational and thus fast simulations must work exactly enough to map the thermal network as well as possible.

In this thesis the different modeling approaches of thermal networks are investigated. For this purpose, the modeling approaches summarized in the prior art are described in detail and simulated on the basis of an example network. Particular attention is paid to the simulation of the pipes, which have a high computational effort and a great influence on the energy storage in the network.

The basic equation of the pipe simulation is a simplified balance of the internal energy equation. By neglecting axial heat conduction and assuming equal cross-sectional temperature, the energy balance is reduced to the pressure change and heat losses. The heat losses are calculated differently depending on whether a static or dynamic modeling approach is used. In this thesis the following pipe models are considered: AixLib StaticPipe, AixLib PlugFlow Pipe, AixLib EmbeddedPipe, MSL Static Pipe and MSL Dynamic Pipe.

A network is simulated with one producer, 16 customers and a total of 408 m pipe length. All customers have the same heat demand, which varies over the year depending on the outside temperature.

The results of the individual simulations are evaluated and discussed. Results show that the StaticPipe from AixLib uses the shortest CPU time of 66.5 seconds and the DynamicPipe of the ModelicaStandardLibary with 315.6 seconds the longest CPU time to simulate the heating network. The various pipe simulations are discussed for different cases of consideration, such as design of the heat generator, flow temperature at the customer and pressure drop. Through this master thesis, it is possible to select the suitable pipe approach for simulations of heat networks in the future. In addition, the models of the pipes can be specifically developed and improved.