Computational Electromagnetism in Transformer and Switchgear Design.
Current Trends
Computation of electromagnetic fields
has a long tradition in design of high voltage equipment like transformers
and switchgear. In particular 2D analysis became an important part
of the industrial design process. Since mid 90s a new design technology
based on 3D solid modeling has been widely introduced. Designers develop
virtual prototypes using efficient 3D modeling systems. These prototypes
are used for creation of technical drawings and in some cases are directly
applied in manufacturing. A designer experienced in 3D CAD modeling is
able to create a virtual prototype of new components within few hours.
Even a complex assembly representing the whole product can be accomplished
within few days. Since the models are now precisely defined in a computer
the designers naturally ask for simulation. The efficiency of 3D simulation
is however strongly limited. The main barriers we would like to discuss
are as follows:
(1) Complexity of simulated phenomena. Three basic
analysis types including dielectric, electromechanical and thermal design
require the computation of electrostatic fields, electromagnetic forces
and losses respectively. But this is only the first step in the analysis.
For example, the goal of dielectric design is prediction of the withstand
voltage that cannot be obtained directly from the standard analysis. Usually
empirical rules specific for oil and gases are applied in evaluation of
electrostatic results. In case of the electromechanical and thermal analyses
the material stresses and temperature rise are of interest. They can be
obtained by employing mechanical and thermal solvers that use the previously
calculated electromagnetic forces and losses as loads.
(2) Discretization of complex 3D geometries. In
particular the generation of volume mesh in the space outside of solid
parts (usually required by finite element method) seems to be a significant
barrier in industrial applications. Therefore the use of the boundary element
method is preferable. It requires mainly discretization of surfaces that
is much easier for standard mesh generators.
(3) Access to resources. The expensive simulation
tools are used in many medium size companies just few times a year. Furthermore,
the education of people and maintenance of the expertise are also expensive.
Consequently, very attractive is access to centrally installed and maintained
resources over Intranet. Included is access to: floating software licenses,
compute servers with installed solvers, Web servers with updated description
of simulation procedures and technical standards for dimensioning criteria,
interactive hot-line support. In particular large but decentralized companies
can efficiently implement this kind of access using Web/Java technology.
Based on selected examples from the transformer
and switchgear design we will show our approach in organizing an efficient
simulation environment that removes the described barriers.
|