The first networks have been developed in the 1960s and '70s by large electrical machines companies such as Siemens Dynamowerk in
Berlin as an analogy to the electrical circuit. The network is also called an analytical lumped circuit or cluster; the thermal
network is also called heat source network, system-level thermo-fluid tool and
in German Wärmequellennetz. The technique involves discretizing the model
domain into sections; each section is represented by a flow or a temperature
node. These nodes are interconnected in a circuit made of several branches in
which the flow or heat transfer is moving one-dimensionally in a defined
direction. Flow and thermal network are fast tools which neither require a 3D
CAD model nor 2D drawings. Flow network calculations should be performed already in the tender phase in
order to sketch the cooling concept and select the fan design. They can be
developed separately; in this case, the flow parameters are inputs of the
thermal network; they can also be combined in one program.
We recommend developing the tool either by programming in Fortran or by using the commercial tool FloMASTER, Amesim, SimulationX ...
Developing a flow network requires a high level of experience and measured values on flow models and machines in order to input the pressure loss coefficients and the active pressure generating coefficients. The main tasks of the flow network are the calculation of the pressure drops and average flow velocities in order to design the fan system, calculate the windage losses as well as the detection of abnormal operations like reverse flows.
Developing a thermal network requires a high level of experience and measured values, especially for the input of the heat transfer coefficients α, heat conduction values of the laminated iron core and coil insulation. Electrical losses, ventilation losses and flow distribution are further inputs.
The flow and thermal processes in a large motor or generator are highly complicated and they cannot be described in details by flow and heat source networks. Measurements inside a fast rotating rotor are very complicated due to the high circumferential speed. 3D Computational Fluid Dynamics is the correct alternative for the analysis, concept and for the detailed design of electrical machines. It allows to predict the flow structure around the stator coil ends, the distribution of the cooling medium, pressure losses as well as the temperature distribution. Using these results, the optimization potentials of the initial design are visible. The optimum cooling should be ensured and realized with additional CFD calculations.
Depending on the ventilation concept, the model can be reduced to a smaller
extent by using the symmetry. The air enters from both sides of
the machine for dual ventilation, the model can there be reduced to half as the definition of symmetry
conditions extends the machine to its full size.
For cooling medium working under normal pressures, the calculation will be equally difficult regardless of whether the medium is air, hydrogen, water or oil. The electrical losses are usually known as calculated or measured values; they can be allocated homogeneously in the places of production. Losses due to friction and turbulent dissipation are calculated in the place of production by the CFD-tool.
A conjugated thermal and flow calculation of a
stator coiler head without an extreme simplification of the
geometry is only possible with a cartesian
mesh tool like FloEFD. Thereby up to 40 Million cells might be needed. FloEFD allows the definition of the inlet conditions
through the input of the fan characteristics; the operating point will be
calculated by the program.
New page available: