Gforge Advanced Server - SDT Support - Answers and discussion http://support.sdtools.com/gf/ Gforge Advanced Server RSS How to compute geometric stiffness for volumesEtienne Balmes <balmes@sdtools.com> Question : I would like to account for stiffening effects due to pretension in computing modes of a volume structure. Answer : stiffnening effects are due to the existence of geometrically non-linear effects. You must thus Perform a static computation to your loads (calling fe_simul static for example) and store the response in the stack as mdl=stack_set(mdl,'curve','StaticState',def). Assemble matrices that account for prestiffening effects (note the MatDes=5 and not 1). Typically a call of the form  [SE,Case]=fe_case(mdl,'assemble -matdes 2 5 -SE NoT') Compute modes  def=fe_eig({SE.K,Case.T},eigopt) Re: white noise excitationSarehati Umar <sarehati@gmail.com>   I want to simulate a structure with white noise excitations to get the raw time-series output data. How can I do that using fe_time?   Thank you. Global/local matricesEtienne Balmes <balmes@sdtools.com> Question : I have created a model and I use these functions:          mo4=fe_case(mo4,'FixDof','clamped dofs','z==0');           [mo4,Case]=fe_case('assemble -MatDes 2 1 -NoT -SE',mo4);  The output we obtain are here Stiffness and Mass matrices. Are these Global or Local Stiffness/Mass matrices??   Answer : I don’t know what you call local mass matrices. So the answer is global. If you want matrices in local coordinates, then you need to look up the basis trans comman. When importing local coordinate definitions (DID in column 3 of nodes) you will be able to build the local to global coordinate transformation matrix (cGL) which is used as q_global = cGL q_local     Windows / MATLAB > 2014b bug : zoom corrupts screenEtienne Balmes <balmes@sdtools.com> A recent windows update has generated a major bug where calling rbbox for zooming corrupts the screen and shows black squares (or previous screens). This is being investigated by The MathWorks but you can use the following work-around. add the attached rbbox2.m file to the $matlabroot/toolbox/sdt/private directory run rehash toolboxreset to refresh your toolbox path edit iimouse.m too call rbbox2 : in iimouse replace the string rbbox by rbbox2. You can open at the proper line using sdtweb('iimouse','rbbox')   Re: white noise excitationEtienne Balmes <balmes@sdtools.com> Could you be more precise. How do you want the solution to be computed ? time/frequency iterations see script in : sdtweb d_signal shaker time integration in state-space/simulink time integration with SDT fe_time with that info, I will generate an example. white noise excitationSarehati Umar <sarehati_umar09@yahoo.com>  Do you have a demo for a structure which are under white noise excitation? Re: Base excitationEtienne Balmes <balmes@sdtools.com> For harmonic enforced displacement you should use a DofSet entry in fe_simul dfrf. Obviously the node with enforced motion must be part of the model. Base excitationMarwa Mallouli <marwa_mel@yahoo.fr> Hi, I have done a program that analyse the behavior of a three node cantilever beam but i want to add the harmonic base excitation at the clamped end. If i will make difference in the node matrix (especially the node1) : %    node#  unused         x             y          z node=[ 1    0 0 0              0             0          0;               2    0 0 0       2.5e-03        0          0;               3    0 0 0       5e-03           0          0;               4    0 0 0         1                  1         1]; What should i do? Please help me     Re: Piezo layer Etienne Balmes <balmes@sdtools.com> elt is a Matlab matrix, you must keep the number of colums constant. You can also use the utilities :feutil addelt elt=[ %declaration of element groupe     Inf abs('beam1') 0  %node1   node2    pl     il     refnode    1       2       1      1          22       0    0%longerons    2       3       1      1          22       0    0    3       4       1      1          22       0    0]; elt=feutil('addelt',elt,'mass1', ...    [4     0.0012   0.0012   0.0012    0        0    0]); elt=feutil('addelt',elt,'piez1', ...   [1       2       1      1          22       0    0%longerons    2       3       1      1          22       0    0    3       4       1      1          22       0    0]);   Piezo layer Marwa Mallouli <marwa_mel@yahoo.fr>  Hi I have done a program that analyse the behavior of a cantilever bema with tip. Now i want to add two piezo layers (bimorph) ,having the same lenght and widht of the beam, and analyse its behavior. Can you correct me the elt vector? elt=[ %declaration of element groupe     Inf abs('beam1') 0  %node1   node2    pl     il     refnode    1       2       1      1          22       0    0%longerons    2       3       1      1          22       0    0    3       4       1      1          22       0    0   elt= [elt ; Inf abs('mass1') 0    4     0.0012   0.0012   0.0012    0        0    0]; elt=[elt; Inf abs('piez1') 0   1       2       1      1          22       0    0%longerons    2       3       1      1          22       0    0    3       4       1      1          22       0    0];   and eltp=[ Inf abs('piez1') 0   1       2       1      1          22       0    0%longerons    2       3       1      1          22       0    0    3       4       1      1          22       0    0]; piez1 is the same function of beam1 but the pl, il, elt are changed function [m,k,idof]=piez1(node,eltp,plp,ilp)       Please help me. Re: Natural frequencyEtienne Balmes <balmes@sdtools.com> Sorry but this questions is not related to SDT. And I doubt it would be exact (0.23 migth be an approximation). Natural frequencyMarwa Mallouli <marwa_mel@yahoo.fr>  Hi The natural frequency of at twenty-element beam with tip mass is : W=(beta*l^2)*sqrt((3EI/(Mtip+0.23Mbeam)*l^3) Is it true ? Re: Input / output responseMarwa Mallouli <marwa_mel@yahoo.fr>  size(mode)=40*40 size(b)=126*1   Re: Input / output responseMarwa Mallouli <marwa_mel@yahoo.fr>  The error is : Error using  *  Inner matrix dimensions must agree. bp=mode'*b; Stress computation/display for platesEtienne Balmes <balmes@sdtools.com> I have used fe_stress to calculate the stress of beam, bar and quad4 element. then I really need to plot the stress of those elements because i want to compare/ check the stress elements before and after applying the zero-thickness interface element. so I would like to ask you two questions:   1). Could you give me an example displaying stresses for a plate. [model,def]=quad4('TestEig'); C1=fe_stress('Stress -gstate',model,def,'group1');C1=C1.GroupInfo{5}; % Display mean on element as element color icomp=4; % Beware that fe_stress computes stress but returns strain labels fprintf('Comp %i %s\n',icomp,C1.X{1}{icomp})% kx axial (moment) C2=struct('EltId',C1.X{3}, ...    'data',squeeze(mean(squeeze(C1.Y(icomp,:,:,:)),1))); feplot(model,def);fecom('colordataelt',C2)    I could plot and see the stress of quad4 elements. But when I have tried the same way for bar1, and beam1t, I cannnot plot them. could you provide me some comment on this matter.  The example above will also work to display stresses in a beam/bar.   2). In case that I set for quad4 element          model.il =[1    fe_mat('p_shell','SI',1)       0         1        0        0.1];       The results from stress calculation are  e_x, e_y, e_xy, k_x, k_y, k_xy, s_x, s_y and d_z.       I would like to you ask what is d_z  ?   d_z is the drilling DOF moment (not really physical but useful).       Re: Input / output responseEtienne Balmes <balmes@sdtools.com> The calling format seems very old fashioned (recent versions of nor2xf use structures) but correct. You should give the error that MATLAB actually shows. Re: fe_cMarwa Mallouli <marwa_mel@yahoo.fr>  Thanks Input / output responseMarwa Mallouli <marwa_mel@yahoo.fr>Hi, My program describe a twenty -element -cantilever beam with attached mass and i want to plot the FRF amplitude in function of the frequency But The dimension of bp and cp (The input/output responses) are inadequate. the input is a force associated to the extremity of the beam. size(b)=126 *1 size(cd)=1*126 size(mdof)=126*1 size(adof)=40*40 Please can you help me.   b=fe_c(mdof,[21.02],1) ' ; cd=fe_c(mdof,[20.02],1) ;   bp=mode' * b ; cp= [cd*mode     zeros(1,size(mode,2))] ;      % disp. sensor IIw = linspace(80,250,40) ' ; IIxf=nor2xf(freq,0.01,bp,cp,IIw*2*pi); plot(IIw*2*pi,IIxf)   Re: fe_cEtienne Balmes <balmes@sdtools.com> 1/ you should use a column of active DOFs  [1.01;2.01] 2/ [1 -1] is the components associated with each of the dofs in adof. Thus 1 on 1.01 and -1 on 2.01. Which is the meaning of a relative input. b=fe_c ( mdof ,  [1.01;2.01] ,  [1 -1] )'; fe_cMarwa Mallouli <marwa_mel@yahoo.fr>  Bojour, For the input shape matrix b A relative force between node 1 and 3 in the u direction is written like this  : b=fe_c ( mdof ,  [1.01 2.01] ,  [1 -1] ); Please can you explain me the meaning of  [1 -1]. Best regards.