－－－欢迎您发表自己的心得和建议，共建ANSYS电磁FAQ－－－暂时关闭，整理中

北极熊甲

占位

北极熊甲

占位

北极熊甲

占位

北极熊甲

转自lilu版主的帖子，感谢!

（一）当线圈通过1A的电流能够产生1WB的自感磁链,则线圈的电感量就是1H.由于H这个单位太大,常用的单位有mH,uH

电感定律之一:电流周围将形成闭合磁力线圈
电感定律之二:电感是导体上流过单位安培电流时,导体周围磁力线圈的韦伯值
电感定律之三:当导体周围的磁力线圈匝数变化时,导体两端将产生感应电压
（二）计算
1.lmatrix计算线圈电感，Incremental energy method
适用：Conductors may be either current fed PLANE13, PLANE53, SOLID97, and SOLID117 or SOURC36 in SOLID96 (MSP) models
条件：
Nonlinear materials
Permanent magnets
Multiple coils (returns differential self and mutual coil inductance matrix)
例子：
vm221
http://www.simwe.com/forum/viewthread.php?tid=778580&extra=page%3D1%26amp%3Bfilter%3D0%26amp%3Borderby%3Ddateline
电磁力、磁链、电感、磁通求解及结果分析的问题。此例说明了各量的关系。
2.Summing ETABLE items (PLANE53: NMISC, 9 /SOLID97: NMISC, 17)
适用：Self inductance of voltage fed stranded conductors
条件：
–No material nonlinearities exist
–No permanent magnets in model
–Only one coil in the model
例子：
http://www.simwe.com/forum/viewthread.php?tid=620643帖中指出
用plane53单元的话，可以直接求出电感，如下：
/Post1
!Part 7.1: Resistance and inductance;
Esel,S,Mat,,2
Etable,R,NMISC,8
Etable,L,NMISC,9
SSum
Allsel,All

3.复数阻抗
z = v/i
Re{Vphysical} = -2pf*Im{VANSYS}
Im{Vphysical} = 2pf*Re{VANSYS}
VANSYS= ANSYS time integrated electric potential (“VOLT”)
（三）关于能否计算导体电感：
http://www.simwe.com/forum/viewthread.php?tid=235062

直线法结合有限差分法计算多层多导体传输线电容和电感矩阵
一篇论文，可以在学校数据库里搜搜

北极熊甲

非常好的一些提问！

1。menyueqi http://www.simwe.com/forum/thread-798941-1-1.html

在ansys的电容计算中，有两种方法可以求取电容：
一是采用CMATRIX的宏命令，这是一种电荷法的求解，即C=Q/V的方法，分别可以求得接地的自电容矩阵和集中的电容矩阵。这个矩阵里面包含了自电容和互电容，这可以求解多导体之间的关系。CMATRIX默认了施加电势为100V。同时，要想获得准确的电容值，接地导体必须准确定义，在近似的情况下可以将远场设定为地。若是有回路的话，需将另一类导体设为地。当然还有Trefftz domain求解方法。最关键的是如何有效的设置地的单元。
另一种求解的方法是采用能量法求解，电容值=2*W/电势降，通过建立ETABLE\ENERGY，进行能量的提取，但是只能获得单个导体的电容值，无法获取互电容。此时，无法采用CMATRIX求解。
因此，这两种方法的求解，都有优缺点，电荷法采用CMATRIX矩阵获得，需要有准确的地导体；二能量法，则是无法获得互电容，除非进行理论上的推导演化。

在ansys的电感计算中，也是存在同样的问题：
一是LMATRIX只适应于线圈的电感计算，可以得出互电感；
二是采用plane53可以获得单根导体的计算，无法获得互电感；
三是采用电感和电容矩阵的关系，可以得出一定的矩阵：L=介电常数*磁导率*（电容矩阵的导数），但是不如采用电荷法，或能量求解的准确。

所以，在计算这两个参数上，如何获取准确的求解值还是个问题，希望大家有好的建议提出来。

北极熊甲

解决超大结果文件的方案:
  
1. 将不同时间段内的结果分别写入一序列的结果记录文件；
  
2. 使用/assign命令和重启动技术；
  
3. ANSYS采用向指定结果记录文件追加当前计算结果数据方式使用/assign指
  
定的文件，所以要求指定的结果记录文件都是新创建的文件，否则造成结果文
  
件记录内容重复或混乱。特别是，反复运行相同分析命令流时，在重复运行命
  
令流文件之前一定要删除以前生成的结果文件序列。

引自雨人版主的帖子

http://www.simwe.com/forum/viewthread.php?tid=321349

[ 本帖最后由 北极熊甲 于 2007-9-24 21:38 编辑 ]

北极熊甲

我有200个实常数用来定义分段线性电流（电流源circu124），怎么办？

1）RMORE, R7, R8, R9, R10, R11, R12
Adds six more real constants to the most recently defined set. Repeat the RMORE command for
constants 13 to 18, again for 19-24, etc.

2）When copying real constants to new sets, ANSYS recommends that you use the command input.
If you do use the GUI, restrict the real constant copy to only the first six real constants (real constants seven and greater will be incorrect for both the master and copy set).

但是如果需要定义6个以上的点的话就不能用RMORE，只能用RMODIF，在何在步间改实常数来实现任意电流函数的加载

[ 本帖最后由 北极熊甲 于 2007-10-8 09:19 编辑 ]

北极熊甲

关于读入数据 *VREAD等，这是一篇由luckier大侠发表的高质量帖子，整理到这里以方便查询。

mread bete1 for test


因为要用到读入矩阵，ansys自带了*vread 和*tread 可是都不好用
*vread要带格式 我的矩阵格式很不统一，有整型 有小数等，所以没法用
*tread读入第一行和第一列会独到表头上去，我不知道怎么再把它搞到array内
所以就写了一个宏，很简单 仅供测试 如果有什么问题希望反馈
工作目录或宏目录下需要名字为mread的两个文件


可以读入自由格式的矩阵，包括 1，1e2， 1000，1.1，1.1111 这类混合的
可以读入任何字符割断的 比如 1 1 2  或 1,1,2,  以及 1 ] 1 ]   3(包括割断之间的空格不等)
预先不需要行数和列数 不需要定义数组

原帖出处：http://www.simwe.com/forum/viewthread.php?tid=789196

再次感谢luckier

北极熊甲

如何改变contour legend中的字体的大小，样式？

Utility Menu>PlotCtrls>Font Controls>Entity Font

Utility Menu>PlotCtrls>Font Controls>Legend Font

北极熊甲

贴图 ：感应电流是怎么产生的（磁场－－静电场？电路？）？ 速度效应（产生的涡流）是怎么考虑的？AZ和AZ＋Volt的区别（短路与开路的区别）？ 开路？开在哪个方向上？AZ＋V比AZ灵活，AZ能作的东西，AZ＋V都能做。但是AZ＋V能作的东西，AZ未必能。涡流要构成回路，必须得用短路条件，可以用AZ自由度。如果用AZ＋V自由度你需要使得两个终端的Volt耦合，即两个终端之间是短路的。加个Volt自由度，可以处理复杂的3D模型中的短路，但是一般的轴对称中的短路用AZ就行了[code]3.4.1. Using DOFs to Manage Terminal Conditions on Conductors
The ANSYS program gives you several options to handle terminal conditions on conductors. These options offer enormous flexibility in modeling, for example, stranded and massive conductors, short circuit and open circuit conditions, and circuit-fed devices. To model each of these entities, you perform these tasks:



Add extra degrees of freedom (DOFs) to the conducting region.

Assign required real constants, material properties, and special treatments to the DOFs. Element types and options, material properties, real constants and element coordinate systems are defined as "attributes" of the solid model; you assign them using the AATT and VATT commands or equivalent GUI paths.




3.4.2. The AZ Option
Conductors modeled with the AZ DOF simulate short-circuit conditions, due to the absence of electric scalar potential which implies zero voltage drop along the length of the conductor.



3.4.3. The AZ-VOLT Option
The AZ-VOLT option allows you to model massive conductors with various terminal conditions by including an electric potential in the overall electric field calculation:

E = δA/δt -V

Note
In ANSYS, V is replaced by ν =  Vdt (time-integrated potential)
This gives you additional flexibility to consider open circuit conditions, current-fed massive conductors, and multiple conductors with end connections, by allowing control over the electric field (VOLT).
The potential, ν, has units of volt-seconds and uses the ANSYS DOF VOLT. In an axisymmetric analysis, ν is r * VOLT. In a planar or axisymmetric analysis, ν is constant over the conductor cross-section (that is, the voltage drop is in the out-of-plane direction only). To enforce this requirement, you must couple nodes in each conducting region using one of the following:



Command(s):
CP

GUI:
Main Menu> Preprocessor> Coupling/Ceqn> Couple DOFs
The coupling essentially reduces the unknowns to a single potential drop unknown in the conducting region.



3.4.4. The AZ-CURR Option
You use the AZ-CURR option to model a voltage-fed stranded coil. The CURR DOF represents the current per turn in the coil. You can apply a voltage drop to the coil using one of the methods shown below (this applies to all coil elements):



Command(s):
BFE,VLTG

GUI:
Main Menu> Solution> Define Loads> Apply> Magnetic> Excitation> Voltage drop> On Elements
Alternatively, you can apply voltage-drop loadings to areas of the solid model by using the BFA command. You can then transfer the specified voltage-drop loadings from the solid model to the finite element model by using either the BFTRAN command or the SBCTRAN command.
No eddy currents are calculated in a stranded coil source, because it is assumed that the strands are insulated. The coil is described by "real" constants that the ANSYS program uses to calculate the coil resistance and inductance.[/code]In ANSYS, V is replaced by ν =  Vdt (time-integrated potential)
This gives you additional flexibility to consider open circuit conditions, current-fed massive conductors, and multiple conductors with end connections, by allowing control over the electric field (VOLT).
The potential, ν, has units of volt-seconds and uses the ANSYS DOF VOLT. In an axisymmetric analysis, ν is r * VOLT. In a planar or axisymmetric analysis, ν is constant over the conductor cross-section (that is, the voltage drop is in the out-of-plane direction only). To enforce this requirement, you must couple nodes in each conducting region using one of the following

[ 本帖最后由 北极熊甲 于 2007-10-15 22:29 编辑 ]

北极熊甲

AZ和AZ+V的区别：就是开路于短路(涡流导体别短接)的区别＝开路没有净电流，短路有净电流

AZ：在导磁体中有净涡流产生，产生一个大的loop（见上面的帖子中的虚线loop，是一个虚拟的loop）,有净电流。而AZ+V模拟的开路在导磁体中虽有涡流但是涡流在导体自身内产生loop(看贴图)，而不是象短路那样的loop，所以没有净电流。一般涡流模拟用AZ+V。在3D结构中，如果可以找到垂直与电流流向的2个截面，那么就可以用AZ+V来做AZ的事情，但是2D中是不行的，所以AZ＋V只能在3D中实现AZ的功能。

看下面图中，你就可以知道ANSYS如何计算加载实导体中的涡流了，如果不是加载导体，那么其中的与加载电流F，，A的那一项（电压降V的那一项）产生的电流就没有了，只剩下前面那个磁场产生 的涡流项了。

[ 本帖最后由 北极熊甲 于 2007-10-18 13:37 编辑 ]

北极熊甲

3D 涡流模拟应该注意：

1. In the 3D model more care needs to be exercised in the modeling sufficient number of elements in the skin depth. In the case of 3D you need to have sufficient elements along all sides of the magnet, since you must allow the eddy currents to circulate in the magnet. When you plot the power in the magnets, you will see the sides of the magnet with the power. by Mike Yaksh

复制代码

The element is solid97 Tex(Hex for air gap).for eddy region, 97's DOF is AZ Volt. One plane of magnet is set with 0 volt, to model the eddy current perpendicular to it. The analysis is done with transient analysis,to investigate the eddy current in the magnets due to the space hamonic and slot teeth. The rotor is rotated mannually and certain nodes is couple with the stator,per step.

I use CE to couple AZ DOF for Period condition and rotor rotation,but CEs are just on the midside nodes of 117.It work well.I didnot couple the corner node, because only the AZ DOF is needed(kopt(1)=0) for these region. I want to ask: if 117 only have AZ DOF(kopt(0)=1), its corner nodes do not work?-- by GuoWei

SOLID117 models 3-D magnetic fields. The element is defined by 20 nodes. It has 12 edge-flux DOFs (AZ), one at each midside node. The eight corner nodes carry the time-integrated electric potential DOF, VOLT (classical formulation) or the electric potential DOF, VOLT (solenoidal formulation).  

The 97 allows CEs, but I do not think the 117 does. This is not a real limitation, it would just mean that you have to step the rotor to line up with the nodes on the stator side, and it would mean that both sides of the interface between the rotor and the stator have to have similar node patterns.

I did it in the way like you mentioned:mesh the slide surface between rotor and stator with equaled spaced nodes,and always make the node to meet each other during the rotation.But I find
that Hex and Prism element model gives better result than Tex element model.In Tex model result,
the air gap's Br is wrong on several points.If I use Hex and Prism element, this doesnot happened,and it meets the results of 97.

I am not sure why the 2D has 50% more power. I have compared 2d and
3D models in the past and the agreement has been much smaller, maybe
10%. (like for example, using a 2D axisymmetric ring and compared it to
a 3D ring ). If anything, the 3D should show more power loss since you
are accounting for the loss in the end where as the 2D model does not do
this. Lower power in the 3D would mean the eddy current is less in the
3D. This might suggest that you are introducing some artificial
resistance. While more resistance would increase power, the increased
resistance reduced I and the power is based on I squared. The power in a
3D eddy current problem is much more sensitive to the mesh than in a 2D
model. The 3D mesh has to have enough elements all along the path of
the eddy current and where it changes direction including on the end
face of the magnet.

[ 本帖最后由 北极熊甲 于 2007-10-18 20:47 编辑 ]

北极熊甲

Unknown parameter name= _Z3

_Z1,_Z2,_Z3 是某个量的3个分量？

[ 本帖最后由 北极熊甲 于 2007-10-19 12:25 编辑 ]

北极熊甲

关于电机模拟中的旋转问题 (3种方法)

以下来自XANSYS的精彩对话

================================ 1.1

Some thought on machines:

1) To allow for ease of sliding build the machine in two parts: 1)
the rotor which includes half of the air gap, and then 2)the stator
which includes the other half of the air gap. Both parts share the same
radial boundary, and the nodes do not have to be coincident, but have
the same radius.

2) Use CEINTF to generate constraint equations between the two. The
node component should have the nodes on the part which have the larger
number of nodes. The elements would be the other side

3) Use ESYS of the elements in the magnets to model the
magnetization. For radial, the alternating magnets have to use different
materials to point in or out. For parallel, define a LOCAL coordinate
system to orient the magnet.

4) "Typically" it is not required to model infinite boundary
elements for 2D since the stator captures the field

5) To rotate, delete the existing CEs, rotate the rotor (NGEN, AGEN
with the IMOVE option), then regenerate the CEs, If using parallel
magnets, update the coordinate system with the LOCAL command (radial
magnets would not require this). A full 360 model is easier to work
with.

6) This works for 3D, but you have to use the 96 since CEs are not
available with 117, and the 97 needs some extra CPs for the iron air
interface. Start with a 2D model if possible.

7) You can solve this as a static or as a dynamic problem, if you
need eddy currents. If you have laminates this is probably not an issue.



Mike Yaksh

NAC International

Norcross GA



Ioannis Anastasiadis wrote:



I have to simulate a rotating wheel which modifies the magnetic field
coming from a permanent magnet.



M problem is that i cannot find anywhere in the help of ANSYS a problem
having to do with rotating structure and changing of the elements.



Can anyone recommend me where I can find more informations in this
problem, or which APDL commands I have to use for managing the rotation
movement or how I procee with this problem

============================= 1.1

[ 本帖最后由 北极熊甲 于 2007-11-3 09:14 编辑 ]

北极熊甲

============================= 1.2

1. Beside what Mike said, I make some complements:
   
2. 
   
3. The machine can be model with 1/2 or 1 electrical period,instead of full model.This can be
   
4. done with CEs.It works in static and dynamics analysis. It saves time and element, but the
   
5. code is more difficult.
   
6. 
   
7. When doing CEs, I suggest you to NSEL node by coordinate.
   
8. 
   
9. Phd student Guo Wei
   
10. Electrical Deparment,Tsinghua Univ,
    
11. Beijing China
    
12. guow_at_mails.tsinghua.edu.cn

复制代码

============================= 1.2

[ 本帖最后由 北极熊甲 于 2007-10-22 16:19 编辑 ]

北极熊甲

============================= 2

Author: Mike Yaksh
Date: 2007-03-14 13:08:34

Hi Ioannis



I would use the two model approach of having a rotor and a stator, and
eliminate the problem of the meshing. You would need to remesh if you do
not have an air gap which can have a common boundary with a constant
radius. Using the CEINTF is straight forward.



1) Select the nodes on one side of the common surface (coincident,
but do not share nodes) between the two parts and generate a node
component of the nodes (on one side only) on the common surface
(CM,ce_n,node)

2) Select the elements on the other side and generate an element
component of the elements on this other suface, which needs to be only a
few elements deep (CM,ce_e,elem_

3) cmsel,,ce_n $ cmsel,,ce_e $ CEINTF



After this the CE symbol should display the CE in a plot, and the lines
show which nodes are used. When a CELIST is done, the first term is the
node from the node component, and the coefficients of the nodes on the
element side will be smaller for the nodes further away from the node
from the node component.



On the AGEN command, the last option is the IMOVE option, sorry for the
confusion.



Mike Yaksh

NAC International

Norcross GA



Ioannis wrote:

well my rotating wheel has 20 tooth's and there is a problem in meshing
of the elements while in rotation.

I don't know if in this case i have to use also CE-constraint equation-
and NGEN,AGEN.

By the way Mike when you are mention IMOVE option, what you mean because
I cannot find anything relavent in the ANSYS help.

============================= 2

附件是一个主题为 Sliding Conductor的XANSYS上的对话

另外在ANSYS中文那网站中的文库资料中也给出了在ANSYS中如何实现运动的具体方法

http://articles.e-works.net.cn/Articles/408/Article43147.htm

有了这个，再加上你对文件的处理---保证（这次的求解是在上次的结果上进行的）（物理环境文件+resume+restart），你就学会结构场+磁场了；转动，平动，你都应该知道如何下手了。再详细的东西就得你自己琢磨了，呵呵

[ 本帖最后由 北极熊甲 于 2007-12-19 16:51 编辑 ]

北极熊甲

各种材料的电阻率，比较全

北极熊甲

很好的一个网站

http://www.coilgun.eclipse.co.uk/links.html

一些试用版软件 的下载，一些关于coil gun的知识。

把这个也写上吧：

http://pera.e-works.net.cn/classlist.htm

[ 本帖最后由 北极熊甲 于 2007-10-27 16:35 编辑 ]

北极熊甲

已经是NTFS格式了，但是生成的文件（结果文件等）依然太大，怎么办

压缩？分割？

北极熊甲

一个简单的同轴电缆集肤效应的2D例子

1. /prep7
   
2. /pnum,kp,1
   
3. /pnum,line,1
   
4. /pnum,area,1
   
5. cyl4,0,0,r0,0,r1,ang
   
6. cyl4,0,0,r2,0,r3,ang
   
7. aatt,1
   
8. asel,none
   
9. cyl4,0,0,r1,0,r2,ang
   
10. aatt,2
    
11. /pnum,mat,1
    
12. allsel
    
13. nummrg,kp,1.E-6
    
14. aplot
    
15. csys,1
    
16. lsel,s,tan1,x,0
    
17. lesize,all,,,1
    
18. lsel,s,loc,x,r0,r1
    
19. lsel,u,tan1,x,0
    
20. lesize,all,,,100,300
    
21. lsel,s,loc,x,r2,r3
    
22. lsel,u,tan1,x,0
    
23. lesize,all,,,40,.02
    
24. lsel,s,loc,x,r1,r2
    
25. lsel,u,tan1,x,0
    
26. lesize,all,,,50,-30
    
27. allsel
    
28. et,1,53,1
    
29. asel,r,mat,,1
    
30. amesh,all
    
31. et,2,53,0
    
32. type,2
    
33. asel,s,mat,,2
    
34. amesh,all
    
35. mp,murx,1,1
    
36. mp,murx,2,1
    
37. mp,rsvx,1,.017E-6
    
38. mp,rsvx,2,1.0E6
    
39. nsel,s,loc,x,r0
    
40. d,all,Az,0
    
41. !nsel,s,loc,x,r3
    
42. !d,all,Az,0
    
43. asel,s,loc,x,r0,r1
    
44. nsla,s,1
    
45. cp,1,volt,all
    
46. *get,nmin,node,,num,min
    
47. f,nmin,amps,1
    
48. asel,s,loc,x,r2,r3
    
49. nsla,s,1
    
50. cp,2,volt,all
    
51. *get,nmax,node,,num,min
    
52. f,nmax,amps,-1
    
53. allsel
    
54. csys,0
    
55. fini
    
56. /solu
    
57. kbc,1
    
58. antype,harm
    
59. outres,all,all
    
60. nsubst,10
    
61. harfreq,0,200.E3
    
62. !harfreq,200.E3
    
63. solve
    
64. /post1
    
65. set,1,10,,0
    
66. PLVECT,H, , , ,VECT,ELEM,ON,0
    
67. !/rep

复制代码

BY--------------------------Johann Riedler (EPCOS)