FEA Forums » Forum: COMSOL Forums

FEA Forums » Forum: COMSOL Forums - Recent Posts http://www.FEA-Forums.com/ FEA Forums » Forum: COMSOL Forums - Recent Posts en Tue, 07 Feb 2012 22:56:28 +0000 lm_0202004 on "help with the coomand asseminit" http://www.FEA-Forums.com/topic/298#post-395 Wed, 30 Mar 2011 19:04:09 +0000 lm_0202004 395@http://www.FEA-Forums.com/ I have the same problem. why? michele on "help with the coomand asseminit" http://www.FEA-Forums.com/topic/298#post-391 Sun, 20 Mar 2011 13:52:50 +0000 michele 391@http://www.FEA-Forums.com/ hi everybody..I'm triying to use a file initially written on FEMLAB,on COmsol 3.5a. Comsol show me error in the line 57-58..and I can't understand why..(i took this code from an article that has been sent to me from a professor). I have to figure out this trouble cause I need this code for my thesis.. I attach the code. Thank anyone will help me % FEMLAB CODE FOR THE 4-TERMINAL DEVICE EXAMPLE OF SECTION 4.2 clear fem femadj % DEFINE REYNOLDS NUMBER, DARCY NUMBER, LENGTH OF DESIGN DOMAIN, AND VOLUME FRACTION Re = 50; Da = 1e-4; L0 = 3.0; beta = 0.4; % DEFINE GEOMETRY, MESH, AND SUBDOMAIN/BOUNDARY GROUPS [SEE FIGURE 6] fem.geom = rect2(0,L0,0,5) + rect2(-2,0,1,2) + rect2(-2,0,3,4) + rect2(L0,L0+2,1,2) ... + rect2(L0,L0+2,3,4); fem.mesh = meshinit(fem,'Hmaxsub',[3 0.125]); % subdomain groups 1:design domain 2:inlet/outlet leads fem.equ.ind = {[3] [1 2 4 5]}; % boundary groups 1:walls 2:inlets 3:outlets 4:interior fem.bnd.ind = {[2:3 5:8 10 12:14 16:18 20:22] [4 23] [1 24] [9 11 15 19]}; % DEFINE SPACE CO-ORDINATES, DEPENDENT VARIABLES, AND SHAPE FUNCTIONS fem.sdim = {'x' 'y'}; fem.dim = {'u' 'v' 'p' 'gamma'}; fem.shape = [2 2 1 1]; % DEFINE CONSTANTS fem.const.rho = 1; fem.const.eta = 1; fem.const.umax = Re; fem.const.alphamin = 0; fem.const.alphamax = 1/Da; fem.const.q = 0.1; Phi0 = 96*(fem.const.eta)*(L0+4)*(fem.const.umax).^2/9; % DEFINE EXPRESSIONS ON SUBDOMAIN AND BOUNDARY GROUPS fem.equ.expr = {'A' 'eta*(2*ux*ux+2*vy*vy+(uy+vx)*(uy+vx))+alpha*(u*u+v*v)' ... 'alpha' {'alphamin+(alphamax-alphamin)*q*(1-gamma)/(q+gamma)' '0'}}; fem.bnd.expr = {'B' '0'}; % DEFINE GOVERNING EQUATIONS AND INITIAL CONDITIONS [SEE EQUATIONS (8) AND (9)] fem.form = 'general'; fem.equ.shape = {[1:4] [1:3]}; % only define gamma on subdomain group 1 fem.equ.ga = {{{'-p+2*eta*ux' 'eta*(uy+vx)'} {'eta*(uy+vx)' '-p+2*eta*vy'} {0 0} {0 0}}}; fem.equ.f = {{'rho*(u*ux+v*uy)+alpha*u' 'rho*(u*vx+v*vy)+alpha*v' 'ux+vy' 1}}; fem.equ.init = {{0 0 0 beta}}; % DEFINE BOUNDARY CONDITIONS fem.bnd.shape = {[1:3]}; % do not define gamma on any boundaries fem.bnd.r = {{'u' 'v' 0 0} ... % walls: no-slip {'u*nx+4*umax*s*(1-s)' 'v' 0 0} ... % inlets: parabolic profile {0 'v' 0 0} ... % outlets: normal flow {0 0 0 0}}; % interior: nothing fem.bnd.g = {{0 0 0 0}}; % zero prescribed external forces everywhere % PERFORM LINEARIZATION, DEGREE-OF-FREEDOM ASSIGNMENT, AND ASSEMBLE INITIAL CONDITION fem = femdiff(fem); fem.xmesh = meshextend(fem); fem.sol = asseminit(fem); % DEFINE STRUCTURE FOR COMPUTING RIGHT-HAND-SIDE IN ADJOINT PROBLEM [SEE EQUATION (29)] femadj = fem; femadj.equ.ga = {{{'diff(A,ux)' 'diff(A,uy)'} {'diff(A,vx)' 'diff(A,vy)'} ... {'diff(A,px)' 'diff(A,py)'} {'diff(A,gammax)' 'diff(A,gammay)'}}}; femadj.equ.f = {{'diff(A,u)' 'diff(A,v)' 'diff(A,p)' 'diff(A,gamma)'}}; femadj.bnd.g = {{'diff(B,u)' 'diff(B,v)' 'diff(B,p)' 'diff(B,gamma)'}}; femadj.xmesh = meshextend(femadj); % GET INDICES OF DESIGN VARIABLE IN THE GLOBAL SOLUTION VECTOR (fem.sol.u) i4 = find(asseminit(fem,'Init',{'gamma' 1},'Out','U')); % COMPUTE VOLUME BELOW DESIGN VARIABLE BASIS FUNCTIONS L = assemble(fem,'Out',{'L'}); Vgamma = L(i4); Vdomain = sum(Vgamma); % GET INDICES OF VELOCITY-PRESSURE VARIABLES i123 = find(asseminit(fem,'Init',{'u' 1 'v' 1 'p' 1},'Out','U')); % DEFINE VARIABLES AND PARAMETERS FOR MMA OPTIMIZATION ALGORITHM [SEE REFERENCES [11,12,13]] a0 = 1; a = 0; c = 20; d = 0; xmin = 0; xmax = 1; xold = fem.sol.u(i4); xolder = xold; low = 0; upp = 1; % DESIGN LOOP FOR THE ACTUAL TOPOLOGY OPTIMIZATION for iter = 1:100 % SOLVE NAVIER-STOKES FLOW PROBLEM TO UPDATE VELOCITY AND PRESSURE fem.sol = femnlin(fem,'Solcomp',{'u' 'v' 'p'},'U',fem.sol.u); % SOLVE ADJOINT PROBLEM FOR LAGRANGE MULTIPLIERS [K N] = assemble(fem,'Out',{'K' 'N'},'U',fem.sol.u); [L M] = assemble(femadj,'Out',{'L' 'M'},'U',fem.sol.u); femadj.sol = femlin('In',{'K' K(i123,i123)' 'L' L(i123) 'M' zeros(size(M)) 'N' N(:,i123)}); % SENSITIVITY ANALYSIS gamma = fem.sol.u(i4); Phi = postint(fem,'A','Edim',2) + postint(fem,'B','Edim',1); dPhidgamma = L(i4) - K(i123,i4)'*femadj.sol.u; % PERFORM MMA STEP TO UPDATE DESIGN FIELD x = gamma; f = Phi/Phi0; g = gamma'*Vgamma/Vdomain - beta; dfdx = dPhidgamma/Phi0; dgdx = Vgamma'/Vdomain; d2fdx2 = zeros(size(gamma)); d2gdx2 = zeros(size(gamma')); [xnew,y,z,lambda,ksi,eta,mu,zeta,s,low,upp] = mmasub(1,length(gamma),iter, ... x,xmin,xmax,xold,xolder,f,dfdx,d2fdx2,g,dgdx,d2gdx2,low,upp,a0,a,c,d); xolder = xold; xold = x; gamma = xnew; % TEST CONVERGENCE if iter >= 100 | max(abs(gamma-xold)) < 0.01 break end % UPDATE DESIGN VARIABLE u0 = fem.sol.u; u0(i4) = gamma; fem.sol = femsol(u0); % DISPLAY RESULTS FOR EACH ITERATION STEP disp(sprintf('Iter.:%3d Obj.: %8.4f Vol.: %6.3f Change: %6.3f', ... iter,f,xold'*Vgamma,max(abs(xnew-xold)))) postplot(fem,'arrowdata',{'u' 'v'},'tridata','gamma','trimap','gray') axis equal; shg; pause(0.1) end qrie on "turbulence modelling--pls help" http://www.FEA-Forums.com/topic/186#post-241 Wed, 23 Jun 2010 23:54:26 +0000 qrie 241@http://www.FEA-Forums.com/ hi i'm simulating the flow over an airfoil with k-epsilon turbulence model in comsol. my prof says i have to start from low velocity input, then solve n find the pressure grad over the surface, determine velocity from that and find the range over which reynolds number varies. he says i should gradually increase velocity and iterate this way. does this make sense to anyone?? pls help!! thanks in advance! ziv.seemann on "I need help!!" http://www.FEA-Forums.com/topic/154#post-182 Wed, 10 Mar 2010 00:30:07 +0000 ziv.seemann 182@http://www.FEA-Forums.com/ Hello I'm a new user with comsol but i know my way around, I need to solve a heating problem in a closed space. In order to do so I'm using comsol multiphysics - incompressible naviar stokes and convection and conduction. After defining the geometry, boundary and subdomains I solved the heat only and then the whole problem, I got an error something like sqrt(-1). see file at <a href="http://rapidshare.com/files/361389700/didnt_work.mph" rel="nofollow">http://rapidshare.com/files/361389700/didnt_work.mph</a> Where did I go wrong? And another question: If I'm trying to use a fan and I want to set the boundary I have only velocity boundary that fits, But I need to enter a source (Q) to the equation so how do I do it? I'm attaching the file that i made Thanx!!!!!!! AlexZinchenko on ".dxf import" http://www.FEA-Forums.com/topic/141#post-165 Tue, 19 Jan 2010 06:00:55 +0000 AlexZinchenko 165@http://www.FEA-Forums.com/ hi all! I have problems with .dxf import in COMSOL.. even with the simplest 3d objects like sphere comsol import it as 2d image any suggestions? .dxf file I've created in Maple13 does anybody know any other CAD program that can draw 3d surface by giving equation? please help, Alex rr00053 on "COMSOL HEAT MODULE" http://www.FEA-Forums.com/topic/77#post-99 Fri, 20 Feb 2009 14:18:54 +0000 rr00053 99@http://www.FEA-Forums.com/ hi everyone.... i am doing a simulation using COMSOL ... does anyone in here have come across heating a solid with laser (pr any em wave source). I would like to model the interaction of em waves to metals.... can anyone halp me???? or tell me in which module should i try this? alienator on "comsol/capacitor simulation" http://www.FEA-Forums.com/topic/76#post-98 Fri, 20 Feb 2009 06:42:00 +0000 alienator 98@http://www.FEA-Forums.com/ hi all , am working with annular capacitors , made a simulated annular plates capacitor in Comsol but my experimental results are not matching with the experimental results experimental capacitance's are far smaller than the simulated ones ex: actual capacitance for 3 mm distance is 1.4 pf by simulation we am getting around 14pf in Comsol i used : electrostatics module in AC/DC -3D module i made 2 annular plates and 2 guard rings around them , and one more big box surrounding all of them (guard rings are just 2 grounded rings surrounding both the capacitor plates , one with +ve potential other with -ve potential to reduce the effect of external disturbances on capacitance ) in sub domain settings i have deactivated all except the big box in which i have put field and other stuff active , have put epsilon(relative) as 1 . now for boundary settings i gave zero charge or symmetry for the bigger box , ground for both guard rings , -1V for one plate , and set port as input for other plate , when i simulate this i am getting results which are not matching with experimental values so somebody plzz help me out mtabesh on "PDE module" http://www.FEA-Forums.com/topic/73#post-95 Fri, 13 Feb 2009 12:42:00 +0000 mtabesh 95@http://www.FEA-Forums.com/ Hello everybody I am using time dependant PDE module in FEMLAB(COMSOL). I was wondering if anybody can help me with this question. How can I save values of an expression at specific times to be used during solution process. Thanks in advance. jvdb on "Scripting in COMSOL Multiphysics 3.3a" http://www.FEA-Forums.com/topic/68#post-90 Wed, 24 Dec 2008 04:46:44 +0000 jvdb 90@http://www.FEA-Forums.com/ Dear all I am a newbie to the field of FEA, so I apologize if the answers to my questions are rather obvious. I am working with Comsol 3.3a. I would like to solve a structural problem on a volume mesh (tetrahedrals). I have created a mesh in Matlab and I am using the Comsol scripting language in the Matlab environment. Subsequently, I export the femmesh structure to Comsol. Yet, does anyone has an idea of how to attribute certain subdomains on self-created meshes with scripting? At this moment, I only have a femmesh. Code (with node and elem the tetrahedral coordinates and elements): el = cell(1,0); el{1} = struct('type','tet','elem',elem'); m = femmesh(node',el); m = meshenrich(m); fem.mesh=m; Furthermore, when looking at this mesh in Comsol, it looks exactly the same as the one in Matlab, yet the boundaries (faces) do not equal, i.e. they are less detailed. Thanks Jan. hesamilati on "summation of Eigen functions" http://www.FEA-Forums.com/topic/27#post-32 Tue, 01 Jan 2008 07:57:47 +0000 hesamilati 32@http://www.FEA-Forums.com/ Hi I have two equation in my problem ((1) Eigen Value Problem + (2) Poisson Equ.). I want to use summation of eigenfunctions of first equation in second equation as a source term. however, I dont know how to use eigen functions of first equation in second equation. Please help me! Sincerely yours!