Drexel dragonThe Math ForumDonate to the Math Forum



Search All of the Math Forum:

Views expressed in these public forums are not endorsed by Drexel University or The Math Forum.


Math Forum » Discussions » Software » comp.soft-sys.matlab

Topic: Matrix is singular and is not supposed to be.
Replies: 4   Last Post: Feb 5, 2013 10:34 AM

Advanced Search

Back to Topic List Back to Topic List Jump to Tree View Jump to Tree View   Messages: [ Previous | Next ]
Hugo

Posts: 6
Registered: 5/14/10
Re: Matrix is singular and is not supposed to be.
Posted: Feb 5, 2013 9:09 AM
  Click to see the message monospaced in plain text Plain Text   Click to reply to this topic Reply

Hi,
Thanks for the reply.

There is a code from a similar problem that has already been published. I am just trying to adapt the code to my problem. It would be of great help if you could debug this code:

The original code is:


%----------------------------------------------------------------------------
% Example 10.10.2
% A cylinder is pinched across its diagonal direction with a force of
% 100 lb at center lengthwise. It has the radius of 5 in., length of
% 10.35 in., and thickness of 0.094 in.
% The material has elastic modulus of 10.5 msi and Poisson's ratio 0.3125.
% A one-eigths of the cylinder is modeled due to symmetry using
% 25 four-node shell elements.
% (see Fig. 10.10.2 for the finite element mesh)
%
% Variable descriptions
% k = element matrix in terms of local axes
% kt = element matrix in terms of global axes
% f = element vector
% kk = system matrix
% ff = system vector
% disp = system nodal displacement vector
% gcoord = coordinate values of each node
% nodes = nodal connectivity of each element
% index = a vector containing system dofs associated with each element
% point1 = matrix containing sampling points for inplne axes
% weight1 = matrix containing weighting coefficients for inplane axes
% pointz = matrix containing sampling points for transverse axis
% weightz = matrix containing weighting coefficients for transverse axis
% bcdof = a vector containing dofs associated with boundary conditions
% bcval = a vector containing boundary condition values associated with
% the dofs in 'bcdof'
% bmtx = matrix for kinematic equation
% dmtx = matrix for material property
% trsh = nodal variable transformation matrix
% rot6 = strain transformation matrix
% aj = 3-D Jacobian matrix
%
%----------------------------------------------------------------------------

%------------------------------------
% input data for control parameters
%------------------------------------

clear
nel=25; % number of elements
nnel=4; % number of nodes per element
ndof=6; % number of dofs per node
nnode=36; % total number of nodes in system
sdof=nnode*ndof; % total system dofs
edof=nnel*ndof; % degrees of freedom per element
emodule=10.5e6; % elastic modulus
poisson=0.3125; % Poisson's ratio
thick=0.094; % shell thickness
nglx=1; ngly=1; nglz=2; % 1x1x2 Gauss-Legendre quadrature
ngl=nglx*ngly*nglz; % number of sampling points per element

%---------------------------------------------
% input data for nodal coordinate values
% gcoord(i,j) where i->node no. and j->x,y,z
%---------------------------------------------

gcoord=[
0.0000 0.0000 5.0000
0.0000 1.0350 5.0000
0.0000 2.0700 5.0000
0.0000 3.1050 5.0000
0.0000 4.1400 5.0000
0.0000 5.1750 5.0000
1.5451 0.0000 4.7553
1.5451 1.0350 4.7553
1.5451 2.0700 4.7553
1.5451 3.1050 4.7553
1.5451 4.1400 4.7553
1.5451 5.1750 4.7553
2.9389 0.0000 4.0451
2.9389 1.0350 4.0451
2.9389 2.0700 4.0451
2.9389 3.1050 4.0451
2.9389 4.1400 4.0451
2.9389 5.1750 4.0451
4.0451 0.0000 2.9389
4.0451 1.0350 2.9389
4.0451 2.0700 2.9389
4.0451 3.1050 2.9389
4.0451 4.1400 2.9389
4.0451 5.1750 2.9389
4.7553 0.0000 1.5451
4.7553 1.0350 1.5451
4.7553 2.0700 1.5451
4.7553 3.1050 1.5451
4.7553 4.1400 1.5451
4.7553 5.1750 1.5451
5.0000 0.0000 0.0000
5.0000 1.0350 0.0000
5.0000 2.0700 0.0000
5.0000 3.1050 0.0000
5.0000 4.1400 0.0000
5.0000 5.1750 0.0000
];

%---------------------------------------------------------
% input data for nodal connectivity for each element
% nodes(i,j) where i-> element no. and j-> connected nodes
%---------------------------------------------------------

nodes=[
7 8 2 1 ;
8 9 3 2 ;
9 10 4 3 ;
10 11 5 4 ;
11 12 6 5 ;
13 14 8 7 ;
14 15 9 8 ;
15 16 10 9 ;
16 17 11 10 ;
17 18 12 11 ;
19 20 14 13 ;
20 21 15 14 ;
21 22 16 15 ;
22 23 17 16 ;
23 24 18 17 ;
25 26 20 19 ;
26 27 21 20 ;
27 28 22 21 ;
28 29 23 22 ;
29 30 24 23 ;
31 32 26 25 ;
32 33 27 26 ;
33 34 28 27 ;
34 35 29 28 ;
35 36 30 29 ];

%-------------------------------------
% input data for boundary conditions
%-------------------------------------

bcdof=[1 5 6 ...
7 11 12 ...
13 17 18 ...
19 23 24 ...
25 29 30 ...
31 32 34 35 36 ...
38 ...
68 70 72 ...
104 106 108 ...
140 142 144 ...
176 178 180 ...
183 184 185 ...
189 190 191 ...
195 196 197 ...
201 202 203 ...
207 208 209 ...
212 213 214 215 216 ]; % constrained dofs
bcval=zeros(size(bcdof)); % whose described values are 0

%----------------------------------------------
% initialization of matrices and vectors
%----------------------------------------------

ff=zeros(sdof,1); % system force vector
kk=zeros(sdof,sdof); % system matrix
disp=zeros(sdof,1); % system displacement vector
index=zeros(edof,1); % index vector
dmtx=zeros(6,6); % constitutive matrix for shear

%----------------------------
% force vector
%----------------------------

ff(33)=-25; % transverse force at node 3

%-----------------------------------------------------------------
% compute material property matrix
%-----------------------------------------------------------------

dmtx(1,1)=emodule/(1.0-poisson^2);
dmtx(1,2)=poisson*dmtx(1,1);
dmtx(2,1)=dmtx(1,2);
dmtx(2,2)=dmtx(1,1);
dmtx(4,4)=emodule/(2.0*(1.0+poisson));
dmtx(5,5)=(5/6)*dmtx(4,4);
dmtx(6,6)=dmtx(5,5);

%-----------------------------------------------------------------
% computation of element matrices and vectors and their assembly
%-----------------------------------------------------------------
%
% for inplane integration
%
[point1,weight1]=feglqd2(nglx,ngly); % sampling points & weights
%
% for transverse integration
%
[pointz,weightz]=feglqd1(nglz); % sampling points & weights

for iel=1:nel % loop for the total number of elements

for i=1:nnel
nd(i)=nodes(iel,i); % extract connected node for (iel)-th element
xcoord(i)=gcoord(nd(i),1); % extract x value of the node
ycoord(i)=gcoord(nd(i),2); % extract y value of the node
zcoord(i)=gcoord(nd(i),3); % extract z value of the node
end

k=zeros(edof,edof); % initialization of element matrix
kt=zeros(edof,edof); % initialization of transformed element matrix
trsh=zeros(edof,edof); % initialization of coordinate transformation matrix

%------------------------------------------------------
% compute direction cosine vectors
% v1 and v2: tangent to the element
% v3: normal to the element
%------------------------------------------------------

d1(1)=xcoord(2)-xcoord(1); % define the vector from node 1 to node 2
d1(2)=ycoord(2)-ycoord(1);
d1(3)=zcoord(2)-zcoord(1);

d2(1)=xcoord(4)-xcoord(1); % define the vector from node 4 to node 1
d2(2)=ycoord(4)-ycoord(1);
d2(3)=zcoord(4)-zcoord(1);

v3(1)=d1(2)*d2(3)-d1(3)*d2(2); % vector v3 normal to vectors d1 and d2
v3(2)=d1(3)*d2(1)-d1(1)*d2(3);
v3(3)=d1(1)*d2(2)-d1(2)*d2(1);
sum=sqrt(v3(1)^2+v3(2)^2+v3(3)^2);
v3(1)=v3(1)/sum; % make vector v3 a unit vector
v3(2)=v3(2)/sum;
v3(3)=v3(3)/sum;

if v3(2) > 0.999999 % if v3 is along the y-axis, v1 is set along x-axis
v1(1)=1.0;
v1(2)=0.0;
v1(3)=0.0;
else % otherwise, v1 is the cross product of y-axis and v3
v1(1)=v3(3);
v1(2)=0.0;
v1(3)=-v3(1);
sum=sqrt(v1(1)^2+v1(2)^2+v1(3)^2);
v1(1)=v1(1)/sum; % make vector v1 a unit vector
v1(2)=v1(2)/sum;
v1(3)=v1(3)/sum;
end

v2(1)=v3(2)*v1(3)-v3(3)*v1(2); % v2 is cross product of v3 and v1
v2(2)=v3(3)*v1(1)-v3(1)*v1(3);
v2(3)=v3(1)*v1(2)-v3(2)*v1(1);
sum=sqrt(v2(1)^2+v2(2)^2+v2(3)^2);
v2(1)=v2(1)/sum; % make vector v2 a unit vector
v2(2)=v2(2)/sum;
v2(3)=v2(3)/sum;

%---------------------------------------------------------
% construct nodal variable transformation matrix
%---------------------------------------------------------

a(1,1)=v1(1);
a(2,1)=v1(2);
a(3,1)=v1(3);
a(1,2)=v2(1);
a(2,2)=v2(2);
a(3,2)=v2(3);
a(1,3)=v3(1);
a(2,3)=v3(2);
a(3,3)=v3(3);

ainv=inv(a);

for i=1:nnel
i1=(i-1)*ndof+1;
i2=i1+1;
i3=i2+1;
i4=i3+1;
i5=i4+1;
i6=i5+1;

trsh(i1,i1)=1.0;
trsh(i2,i2)=1.0;
trsh(i3,i3)=1.0;
trsh(i4,i4)=ainv(1,1);
trsh(i4,i5)=ainv(1,2);
trsh(i4,i6)=ainv(1,3);
trsh(i5,i4)=ainv(2,1);
trsh(i5,i5)=ainv(2,2);
trsh(i5,i6)=ainv(2,3);
trsh(i6,i4)=ainv(3,1);
trsh(i6,i5)=ainv(3,2);
trsh(i6,i6)=ainv(3,3);
end

%------------------------------------------------------
% strain transformation matrix
%------------------------------------------------------

rot6(1,1)=v1(1)^2;
rot6(1,2)=v1(2)^2;
rot6(1,3)=v1(3)^2;
rot6(1,4)=v1(1)*v1(2);
rot6(1,5)=v1(2)*v1(3);
rot6(1,6)=v1(1)*v1(3);
rot6(2,1)=v2(1)^2;
rot6(2,2)=v2(2)^2;
rot6(2,3)=v2(3)^2;
rot6(2,4)=v2(1)*v2(2);
rot6(2,5)=v2(2)*v2(3);
rot6(2,6)=v2(1)*v2(3);
rot6(3,1)=v3(1)^2;
rot6(3,2)=v3(2)^2;
rot6(3,3)=v3(3)^2;
rot6(3,4)=v3(1)*v3(2);
rot6(3,5)=v3(2)*v3(3);
rot6(3,6)=v3(1)*v3(3);
rot6(4,1)=2.0*v1(1)*v2(1);
rot6(4,2)=2.0*v1(2)*v2(2);
rot6(4,3)=2.0*v1(3)*v2(3);
rot6(4,4)=v1(1)*v2(2)+v2(1)*v1(2);
rot6(4,5)=v1(2)*v2(3)+v2(2)*v1(3);
rot6(4,6)=v1(3)*v2(1)+v2(3)*v1(1);
rot6(5,1)=2.0*v2(1)*v3(1);
rot6(5,2)=2.0*v2(2)*v3(2);
rot6(5,3)=2.0*v2(3)*v3(3);
rot6(5,4)=v2(1)*v3(2)+v3(1)*v2(2);
rot6(5,5)=v2(2)*v3(3)+v3(2)*v2(3);
rot6(5,6)=v2(3)*v3(1)+v3(3)*v2(1);
rot6(6,1)=2.0*v3(1)*v1(1);
rot6(6,2)=2.0*v3(2)*v1(2);
rot6(6,3)=2.0*v3(3)*v1(3);
rot6(6,4)=v3(1)*v1(2)+v1(1)*v3(2);
rot6(6,5)=v3(2)*v1(3)+v1(2)*v3(3);
rot6(6,6)=v3(3)*v1(1)+v1(3)*v3(1);

%------------------------------------------------------
% material property matrix transformation
%------------------------------------------------------

dmtxt=rot6'*dmtx*rot6;

%------------------------------------------------------
% numerical integration loop
%------------------------------------------------------

for intx=1:nglx
r=point1(intx); % sampling point in x-axis
wtx=weight1(intx,1); % weight in x-axis
for inty=1:ngly
s=point1(inty,2); % sampling point in y-axis
wty=weight1(inty,2) ; % weight in y-axis
for intz=1:nglz
t=pointz(intz); % sampling point in z-axis
wtz=weightz(intz) ; % weight in z-axis

[shape2,dhdr,dhds]=feisoq4(r,s); % compute 2-D shape functions and
% derivatives at sampling point

%[shape1,dhdz]=feisol2(t); % compute 1-D shape functions and
% derivatives at sampling point
shape1(1)=0.5*(1-t);
shape1(2)=0.5*(1+t);
dhdt(1)=-0.5;
dhdt(2)=0.5;

%--------------------------------------------------
% compute jacobian matrix and its inverse
%--------------------------------------------------

hz=thick*0.5*t;
hzdt=thick*0.5;

aj(1,1)=dhdr(1)*xcoord(1)+dhdr(2)*xcoord(2)+dhdr(3)*xcoord(3)+ ...
dhdr(4)*xcoord(4)+dhdr(1)*hz*v3(1)+dhdr(2)*hz*v3(1)+ ...
dhdr(3)*hz*v3(1)+dhdr(4)*hz*v3(1);
aj(2,1)=dhds(1)*xcoord(1)+dhds(2)*xcoord(2)+dhds(3)*xcoord(3)+ ...
dhds(4)*xcoord(4)+dhds(1)*hz*v3(1)+dhds(2)*hz*v3(1)+ ...
dhds(3)*hz*v3(1)+dhds(4)*hz*v3(1);
aj(3,1)=hzdt*v3(1);
aj(1,2)=dhdr(1)*ycoord(1)+dhdr(2)*ycoord(2)+dhdr(3)*ycoord(3)+ ...
dhdr(4)*ycoord(4)+dhdr(1)*hz*v3(2)+dhdr(2)*hz*v3(2)+ ...
dhdr(3)*hz*v3(2)+dhdr(4)*hz*v3(2);
aj(2,2)=dhds(1)*ycoord(1)+dhds(2)*ycoord(2)+dhds(3)*ycoord(3)+ ...
dhds(4)*ycoord(4)+dhds(1)*hz*v3(2)+dhds(2)*hz*v3(2)+ ...
dhds(3)*hz*v3(2)+dhds(4)*hz*v3(2);
aj(3,2)=hzdt*v3(2);
aj(1,3)=dhdr(1)*zcoord(1)+dhdr(2)*zcoord(2)+dhdr(3)*zcoord(3)+ ...
dhdr(4)*zcoord(4)+dhdr(1)*hz*v3(3)+dhdr(2)*hz*v3(3)+ ...
dhdr(3)*hz*v3(3)+dhdr(4)*hz*v3(3);
aj(2,3)=dhds(1)*zcoord(1)+dhds(2)*zcoord(2)+dhds(3)*zcoord(3)+ ...
dhds(4)*zcoord(4)+dhds(1)*hz*v3(3)+dhds(2)*hz*v3(3)+ ...
dhds(3)*hz*v3(3)+dhds(4)*hz*v3(3);
aj(3,3)=hzdt*v3(3);

ajinv=inv(aj);
det3=det(aj);

%-----------------------------------------------------------
% compute global derivatives
%-----------------------------------------------------------

for i=1:nnel
derivg(1,i)=ajinv(1,1)*dhdr(i)+ajinv(1,2)*dhds(i);
derivg(2,i)=ajinv(2,1)*dhdr(i)+ajinv(2,2)*dhds(i);
derivg(3,i)=ajinv(3,1)*dhdr(i)+ajinv(3,2)*dhds(i);
dhdz(1,i)=ajinv(1,3)*hzdt;
dhdz(2,i)=ajinv(2,3)*hzdt;
dhdz(3,i)=ajinv(3,3)*hzdt;
end

%-----------------------------------------------------------
% compute strain-displacement matrix called bmtx
%-----------------------------------------------------------

bmtx=zeros(ndof,edof);

for i=1:nnel
i1=(i-1)*ndof+1;
i2=i1+1;
i3=i2+1;
i4=i3+1;
i5=i4+1;
i6=i5+1;

gk1=derivg(1,i)*hz+shape2(i)*dhdz(1,i);
gk2=derivg(2,i)*hz+shape2(i)*dhdz(2,i);
gk3=derivg(3,i)*hz+shape2(i)*dhdz(3,i);

bmtx(1,i1)=derivg(1,i); % elements associated with epsilon_x
bmtx(1,i4)=gk1*(-v2(1));
bmtx(1,i5)=gk1*v1(1);
bmtx(2,i2)=derivg(2,i); % elements related to epsilon_y
bmtx(2,i4)=gk2*(-v2(2));
bmtx(2,i5)=gk2*v1(2);
bmtx(3,i3)=derivg(3,i); % elements related to epsilon_z
bmtx(3,i4)=gk3*(-v2(3));
bmtx(3,i5)=gk3*v1(3);
bmtx(4,i1)=derivg(2,i); % elements related to gamma_xy
bmtx(4,i2)=derivg(1,i);
bmtx(4,i4)=gk2*(-v2(1))+gk1*(-v2(2));
bmtx(4,i5)=gk2*v1(1)+gk1*v1(2);
bmtx(5,i2)=derivg(3,i); % elements related to gamma_yz
bmtx(5,i3)=derivg(2,i);
bmtx(5,i4)=gk3*(-v2(2))+gk2*(-v2(3));
bmtx(5,i5)=gk3*v1(2)+gk2*v1(3);
bmtx(6,i1)=derivg(3,i); % elements related to gamma_xz
bmtx(6,i3)=derivg(1,i);
bmtx(6,i4)=gk3*(-v2(1))+gk1*(-v2(3));
bmtx(6,i5)=gk3*v1(1)+gk1*v1(3);
end

detwt=det3*wtx*wty*wtz;

k=k+bmtx'*dmtxt*bmtx*detwt; % element matrix in local axes

end
end
end % end of numerical integration loop

kt=trsh'*k*trsh; % element matrix in global axis


index=feeldof(nd,nnel,ndof);% extract system dofs associated with element

kk=feasmbl1(kk,kt,index); % assemble element matrices

end

%-----------------------------
% apply boundary conditions
%-----------------------------

[kk,ff]=feaplyc2(kk,ff,bcdof,bcval);

%----------------------------
% solve the matrix equation
%----------------------------

disp=kk\ff;

num=1:1:sdof;
displace=[num' disp] % print nodal displacements

%------------------------------------------------------------------------



After running the code, the result is:



>> ori
Warning: The value of local variables may have been changed to match the
globals. Future versions of MATLAB will require that you declare
a variable to be global before you use that variable.
> In feeldof at 3
In ori at 455
Warning: The value of local variables may have been changed to match the
globals. Future versions of MATLAB will require that you declare
a variable to be global before you use that variable.
> In feeldof at 3
In ori at 455
Warning: The value of local variables may have been changed to match the
globals. Future versions of MATLAB will require that you declare
a variable to be global before you use that variable.
> In feeldof at 3
In ori at 455

r =

1

Attempted to access nd(1); index out of bounds because numel(nd)=0.

Error in feeldof (line 25)
start=(nd(r)-1)*ndof;

Error in ori (line 455)
index=feeldof(nd,nnel,ndof);% extract system dofs associated with element

>>





I would be glag if you could debug the code. This code is about obtaining deflection results by the Finite Element Method using a method called "Matrix Displacement method"

regards,
Hugo



"Steven_Lord" <slord@mathworks.com> wrote in message <kenc6p$m5r$1@newscl01ah.mathworks.com>...
>
>
> "Hugo " <hugolopessilva@gmail.com> wrote in message
> news:kejmtf$igm$1@newscl01ah.mathworks.com...

> > Hello, I am trying to obtain deflection results on a beam made with shell
> > elements
> >
> > I wrote a MATLAB code. The problem is that kk is a matrix with 3000*3000
> > full of zeros (its a singular matrix), and it is not supposed to be.
> > The code runs well until I call the function feasmbl1.m, which is shown
> > next

>
> *snip*
>
> No one will be able to debug this without your data. That means that you're
> going to have to be the one to step through your code (ideally on a SMALLER
> problem, one that you can work out on paper) to determine why it doesn't
> return the results you expect (or determine why your expectations are
> incorrect.)
>
> I will suggest that you may want to use the ACCUMARRAY or SPARSE functions
> to perform your accumulation instead of building your own function.
>
> --
> Steve Lord
> slord@mathworks.com
> To contact Technical Support use the Contact Us link on
> http://www.mathworks.com




Point your RSS reader here for a feed of the latest messages in this topic.

[Privacy Policy] [Terms of Use]

© Drexel University 1994-2014. All Rights Reserved.
The Math Forum is a research and educational enterprise of the Drexel University School of Education.