problem with constraints

[mgs]

In my model I have nodes, constraint and elements like below;

# ID X Y Z
node 1 0.0 0.0 0.0
node 2 0.0 0.0 0.0
node 3 0.0 0.0 0.0

# TAG X Y Z MX MY MZ
fix 1 1 1 1 1 1 1
fix 2 0 1 0 1 1 1
fix 3 0 1 0 0 0 0

# tag i-node j-node material X Z
element zeroLength 1 1 2 -mat 10 26 -dir 1 3
element zeroLength 2 2 3 -mat 7 7 -dir 1 3

My problem is that with the said definition I will get the below error in static analysis and complex eigenvalues in eigen analysis which means the models in not fixed.

WARNING SuperLU::solve(void)- Error 1 returned in factorization dgstrf()
WARNING NewtonRaphson::solveCurrentStep() -the LinearSysOfEqn failed in solve()

But when I change the 2nd element‘s definition to the below it works properly;

element zeroLength 2 2 3 -mat 7 20 7 -dir 1 2 3

The Mat tag 20 is rigid i.e. E=1e10 ksi.

Why it should be like this? Can you give me an idea?

I am using the OpenSeenNavigator.

[fmk]

the problem really needs either 4,5,and6 dof at node 3 fixed or materials in those direction defined for element 2. what constraint handler are you using? if Penalty it is possibly working because of round-off error.

[mgs]

Thanks Franks,

I think I have found the problem: the zerolengthelement orientation, however, I could not fix it yet.

Element 2 actually represents a fixed type bearing and I think assigning more constraints and/or material could not be write.

The Navigator write the orientation at the end of each zerolength element, as default, like below:

# Element "IMP_i": eleTag NodeI NodeJ -mat matTag1 matTag2 ... -dir dir1 dir2 ... <-orient x1 x2 x3 y1 y2 y3> <-doRayleigh flag>
element zeroLength 456 107 123 -mat 21 -dir 1 -orient +1.000000E+00 +0.000000E+00 +0.000000E+00 +0.000000E+00 +0.000000E+00 +1.000000E+00

Does the above mean that the local element axes coincide with the global axes? If not what should I specify then? Is it better to remove the orientations?