This paper deals with the simulation of rotary heat exchangers made of polyethylene terephthalates using CFD software. First steps for the simulation of simplified flow processes within a heat exchanger are presented. The aim of the project is the substitution of commercially available materials such as aluminum to plastic. The use of plastic as a storage mass material will create new possibilities which will make the heat exchangers even more efficient than their predecessors made of aluminum. For this purpose, the two materials are compared with each other using simulation models with ANSYS Fluent© and the resulting temperature curve is examined. The results show that PET offers a more suitable temperature profile than aluminum for heat exchanger.
Rotary heat exchangers are used with permitted circulating air. The main task is the
supply and removal of heat energy in computer centers, office buildings, factory halls
and cruise ships. Due to the large surface area and the associated large air volume,
rotary heat exchangers are considered to be the most effective heat exchangers. During
operation, two air streams flow through the storage mass, the supply air stream from
the outside and the exhaust air stream from the inside. Each of these two flows transfers
its heat energy to the storage mass. The energy absorbed is released again and the heat
is transferred by the rotation in the respective other stream. [
In counterflow heat exchangers, the air flows are guided through a series of parallel plates. Figure 2 shows the simplified principle of the heat exchanger as a volume body in CATIA V5R20®. The arrows indicate the direction of flow of the two air streams. The warm air flow directs the energy through the plate into the cold air flow, resulting in heat exchange. The shown model was calculated under the theoretical basis of singlelayer walls. The proportion of the heat flow absorbed depends on the design of the heat exchanger, the size of the effective heat transfer surface and the material of the transfer surface. The cooperation company Klingenburg GmbH has already carried out the first practical tests in this field, in which the material of the heat exchanger mass was changed from aluminum to PET. It has been shown that PET leads to a better efficiency as soon as the layer thickness becomes very thin.
If the stationary heat conduction is considered in a single-layer wall, the following
relationship applies to the heat flow for a plate [
− ·
) = = = = = = = · thermal resistance temperature warm/cold time wall thickness thermal conductivity trea flowed through
3 (1) (2) with the thermal resistance: /
If the layer thickness in equation 2 is continuously smaller with the same denominator, the thermal resistance also becomes arbitrarily low. This means that the significantly higher thermal conductivity of aluminum (approx. factor 1000 better than PET) continuously has a considerable influence.
ANSYS Fluent© The first simulations were created with ANSYS 18.2 Fluent© CFD software. The geometry used for this was 2.1 mm wide, 0.5 mm high and 5 mm long. Two channels were provided with a height of 0.15 mm and a width of 2 mm over the entire length of this geometry. The dividing wall between them had a continuous thickness of 0.05 mm (50 µm).
In Fluent©, a stationary state was simulated with an air velocity of 1 m/s flowing through both channels. The temperature of the supply air flow duct is 0°C and the temperature of the exhaust air flow is 25°C. The thermal boundary conditions on the outer walls were defined as adiabatic and the material properties varied between aluminum and PET.
Looking at the results in Figure 3, it can be seen that PET has a more favourable temperature profile for heat exchangers. In the case of aluminum, a constant temperature has been set over the entire cross-section, while the PET volume body has a higher temperature difference in the air flow channels.
This simulation is a first comparison of the two materials, aluminum and PET, with a simplified model. The results obtained are similar to the practical tests carried out by Klingenburg GmbH. In order to verify the simulation results, the model will be simulated in practice in the next phase of the project. 6
The GFTN research team would like to thank the BMWi and the cooperation partner Klingenburg GmbH for their support and cooperation in this research project.