To develop the workflow a simple dish and feed model was initially developed, Figure 3.2 shows this model. This model had all the parameters required to perform the EM simulation defined, such as the operating frequency and the waveguide source that powers the feed. The simple model was chosen so that the focus could be on the development of the workflow process. The model was used to develop the methods for "integrating" the structural analyses and EM analyses together, allowing for a robust method to perform a structural analysis and then an EM analysis on the deformed structure. This antenna has a main reflector diameter of 12 m and all EM simulations were performed at a frequency of 1 GHz. The main dish was assumed to be made of steel.
To be able to investigate the effects of deformation on electromagnetic performance, that deformation needs to be represented in the EM model. By
using the same mesh between the structural model and the EM model there is greater certainty that deformation calculated by the structural FEA solver is accurately captured in the EM model.
First, the methods for exporting/importing the meshes and geometries be- tween the two software packages were investigated. In HyperMesh the model was loaded with two equal forces along the periphery of the dish pointing in- wards, as shown in Figure 3.3. This was to cause some deformation in the dish so that a deformed mesh could be obtained. This is an unlikely load case, but it is known in what manner the dish would deform and the deformations would be obvious. It was found that the most reliable formats for the export- ing/importing of the meshes and geometries were the STL and STEP formats respectively. In this phase of the development only the dish was allowed to deform. This was for the sake of simplicity as only the exporting/importing methods were of interest.
Figure 3.3: Load case applied to aid in developement of workflow process. The arrows are two equal forces applied at the periphery of the dish as shown
In reality, the feed will deform in some way due to it being mounted to the same structure as the dish surface. It was assumed that the feed would only translate and rotate as a rigid body. This was achieved by giving the feed model an artificially larger Young’s modulus and shear modulus in the structural model. This assumption was made because the feeds used in reflector antennas are often made of significantly thicker material than the rest of the antenna and thus it is expected to deform significantly less, it is also much smaller which will also lead to there being less deformation. This assumption made the updating of the feed’s position and orientation significantly easier in the EM model as its rotation and translation could easily be calculated. A problem that was encountered here was that if the feed was not an ideal cylindrical shape the definition of its source would fail and the EM simulation would not run. Thus it was decided to have a predefined FEKO model of the feed and update that
CHAPTER 3. COMBINED STRUCTURAL AND ELECTROMAGNETIC
SIMULATION APPROACH 20
as opposed to attempting to create a new feed model to match the deformed mesh in the structural model.
To start automating the feed update procedure four support struts were added to the model to connect the feed to the dish, thus causing the feed to displace, in the z-direction only, when the dish was deformed. Two scripts were written, one in Python and the other in EDITFEKO, to transform the feed and its waveguide port in the FEKO model by the same amount determined in the structural simulations. The functionality of these scripts will be explained in the next section.
Once the displacement of the feed in the z-direction had been automated, the model was changed slightly to allow for new translations and rotations to be included with this. This was implemented incrementally until all six degrees of movement could be automated. This was implemented as follows;
• Only one strut was used to connect the feed to the dish resulting in the feed displacing in the y and z-directions and rotating about the x-axis. • A second strut was added so that the feed would displace in all three
directions and rotate about the x and y-axes.
• Lastly four struts were used again; however, this time a torque was added to the feed so that it would rotate about the z-axis. This also resulted in a displacement in the z-direction.
Figure 3.4 shows the models used to aid in the creation of the feed update automation. After these steps had been performed the feed could be updated for all translations and rotations it may experience during deformation of the dish.
Figure 3.4: Antenna models used to aid in feed update automation. The different strut configurations will cause the feed to deform in predictable ways so that the focus could be on a few degrees of freedom at a time.