Periodical BC on all segments of the ...

Log Out | Topics | Search Moderators | Register | Edit Profile

Author Message   Adrian Rozanski (mendoza81) New member Username: mendoza81 Post Number: 1 Registered: 04-2006 Posted on Tuesday, April 04, 2006 - 10:04 am:    Hi. I`m using the student version of FlexPDE. I have a problem with setting the periodic boundary conditions on all segments of the figure. While i`m trying to solve the problem, FlexPDE shows an error: "Multiple Periodic Images of Point(...,...). Click 'Domain'". I`m sending to you script file. Thanks for your help. 0.9 K periodic.txt "script file"   Adrian Rozanski (mendoza81) New member Username: mendoza81 Post Number: 2 Registered: 04-2006 Posted on Tuesday, April 04, 2006 - 02:16 pm:    Here is an input file: VARIABLES u v SELECT painted DEFINITIONS R=0.25 { parameter definitions } A=30 G=30 Ex=1 Ey=0 Exy=0 Sx=A*(Ex+Ey)+2*G*Ex+A*(dx(u)+dy(v))+2*G*dx(u) Sy=A*(Ex+Ey)+2*G*Ey+A*(dx(u)+dy(v))+2*G*dy(v) Txy=2*G*(Exy+0.5*(dx(v)+dy(u))) ! INITIAL VALUES EQUATIONS u: dx(Sx)+dy(Txy)=0 v: dx(Txy)+dy(Sy)=0 BOUNDARIES REGION 1 START(-0.5,-0.5) VALUE(u)=0 VALUE(v)=0 LINE TO (0.5,-0.5) VALUE(u)=0 VALUE(v)=0 LINE TO (0.5,0.5) VALUE(u)=0 VALUE(v)=0 LINE TO(-0.5,0.5) VALUE(u)=0 VALUE(v)=0 LINE TO CLOSE START(R,0) NATURAL(U)=0 NATURAL(V)=0 ARC(CENTER=0,0) ANGLE=360 PLOTS CONTOUR(Sx) CONTOUR(Sy) CONTOUR(Txy) SUMMARY REPORT(Area_Integral(Sx)) REPORT(Area_Integral(Sy)) REPORT(Area_Integral(Txy)) END   Adrian Rozanski (mendoza81) Junior Member Username: mendoza81 Post Number: 3 Registered: 04-2006 Posted on Tuesday, April 04, 2006 - 02:22 pm:    I would like the BCs to be like this: BOUNDARIES REGION 1 START(-0.5,-0.5) PERIODIC(x,y+1) LINE TO (0.5,-0.5) PERIODIC(x-1,y) LINE TO (0.5,0.5) PERIODIC(x,y-1) LINE TO(-0.5,0.5) PERIODIC(x+1,y) LINE TO CLOSE Is it possible to set this kind of BCs?   Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 575 Registered: 06-2003 Posted on Tuesday, April 04, 2006 - 03:17 pm:    Firstly, a periodic boundary implies periodicity of two sides, since the relationship is reciprocal. So at most, you need two, not four, periodic specifications. However, as currently implemented, each mesh node can have only one periodic image node. Imaging a corner node in two directions requires two image nodes, and is not supported. You can fudge around this restriction by creating two nodes near the corner (one periodic in X and the other in Y) and leaving a small hole in the periodicity. Use a Natural()=0 boundary on the stub connecting the two nodes. This falsifies the periodicity somewhat, but it is the best we can do at the moment.