In electromagnetic field simulation, what are the advantages and disadvantages of FDTD and FEM algorithms?
[Copy link]
Computational electromagnetics can be divided into two categories: full-wave simulation algorithm and high-frequency algorithm. Full-wave simulation is an accurate algorithm, but it consumes a lot of computing resources. A simple estimation method is: Usually we need to divide the object, and the division should be at least 0.1 wavelength. That is to say, if the electrical size of the object is 10 wavelengths, there are 100*100*100=1 million grids. For each grid, you also need to store a large number of electromagnetic parameters, which are generally single-precision floating point types. So it is easy to need hundreds of megabytes of memory. If the electrical size is 20 wavelengths, then you need more than G memory. If the geometric features of the object are relatively abnormal and there are many subtle structures, more intensive division is required, which can easily exceed the computing power of ordinary computers. For example, the wavelength of 1GHZ is 0.3 meters, and the frequency of GSM is about the same, so it is possible to simulate objects of one or two meters. If it is 3G communication, the frequency is about 2GHz, so we can only calculate objects that do not exceed one meter, and there cannot be any strange structures. [p=null, 2, High-frequency algorithms are designed to solve this problem. For military systems, we need to analyze the electromagnetic performance of aircraft and ships. According to the previous discussion, full-wave simulation is obviously not enough. Therefore, high-frequency algorithms use a lot of approximations, such as physical optics, geometric ray method, etc., to perform approximate calculations. In this field, there is currently no open commercial software. FDTD (Finite Difference Time Domain) FEM (Finite Element) MOM (Method of Moments) FIT (Finite Volume Integration) are all full-wave simulation algorithms. Among them, FDTD and FIT are time domain algorithms, represented by CST software, while FEM and MOM are frequency domain algorithms, represented by HFSS and Feko software. Time domain algorithms are suitable for broadband signal analysis, because as long as a pulse input is simulated, very wide bandwidth information can be obtained. Frequency domain algorithms can only calculate one frequency point at a time, which is suitable for narrowband signals. [font=-apple-system-font, BlinkMacSystemFont, Compared with FDTD, FEM is mainly more accurate in meshing because FEM uses triangular meshes while FDTD uses quadrilateral meshes.
[attach] 62)]Now let's talk about the difference between FDTD and FIT. FDT2000D directly discretizes the differential equations. As we all know, for each grid, we consider it to be uniform.
[font=-apple-system-font, BlinkMacSystemFont,FIT is a discretization of the integral equation. It takes many electromagnetic parameters along the integral curve. It is suitable for dealing with non-uniform media and interfaces. It is more accurate than FDTD. This is also the calculation method of CST.
| 
提升卡
变色卡
千斤顶
