FEAP User Forum

FEAP => Input File Issues => Topic started by: crepes on March 13, 2022, 01:52:23 AM

Title: How should an input file for a cable element look like?
Post by: crepes on March 13, 2022, 01:52:23 AM

Dear all,

I attached my attempt with a 1D nurbs-mesh for a straight horizontal cable with vertical body force. The cable is fixed on it's ends.

With ndf=2 (using the second line in the input file), the tangent matrix becomes singular, or at least it has zeroes on the main diagonal. It misses entries for the y-direction displacements.
With ndf=1, it calculates something and it converges, but I am not sure how to interpret the outcome.

According to

Code: [Select]

.../ver86/iga/elements/cable/elmt04.f

it should have ndf =  2.

Kind regards

Title: Re: How should an input file for a cable element look like?
Post by: Prof. R.L. Taylor on March 13, 2022, 08:56:25 AM

You will have to specify a non zero initial displacement. They are hard to start

Title: Re: How should an input file for a cable element look like?
Post by: crepes on March 13, 2022, 10:27:26 AM

Many thanks for your reply.

For different non zero INITial DISPlacements, e.g., using catenary function values, I get a zero diagonal

Code: [Select]

*Command   3 * tang                v:   1.00       0.00       0.00   
                                                           t=     0.01     0.00
   Residual 1 norm =    3.0000000E-03   1.0000000E+00      t=     0.01     0.00
      DOF   1   0.0000000E+00   0.0000000E+00
      DOF   2   3.3333333E-04   3.4482759E-04
 *D4TRIU WARNING* Reduced diagonal is zero in         9 equations.
                 Step:      0 Iteration:     0
 Condition check: D-max 0.2000E+00; D-min 0.1111E+00; Ratio 0.1800E+01
 Maximum no. diagonal digits lost:  1
   End Triangular Decomposition                            t=     0.01     0.00
   Number of operations =     194 plus        0 Mega-ops
  Time: CPU =         0.01 , System =         0.00
--> SOLVE AT     4.04 Mflops. Time=        0.00
   Energy convergence test
    Initial   =    0.000000000000000E+00 Current   =    0.000000000000000E+00
    Relative  =    1.000000000000000E+00 Tolerance =    1.000000000000000E-16
 *Command   4 * next                v:   10.0       3.00       1.00   



When I, instead, use non zero y-coordinates for the nurbs coordinates, I get huge values in the (factorized) tangent (via TANG,,1 and OUTP TANG)

Code: [Select]

         1         1    5.009358224664411D+00
         2         2    2.277545928469153D+04
         3         3    1.000059638950751D+01
         5         5    1.000007386716542D+01
         7         7    1.000000442404896D+01
         9         9    1.000000442404896D+01
        10        10   -3.022314549036573D+22
        11        11    1.000007386716540D+01
        12        12    4.722366482869645D+21
        13        13    1.000059638950753D+01
        14        14    1.180591620717411D+21
        [...]

and the unfactorized using TANG,,-1

Code: [Select]

         1         1    1.996263703155293D-01
         2         2    2.489637161042502D-04
         3         3    1.999494674280283D-01
         4         4    3.368613414150350D-05
         5         5    1.999932977943863D-01
         6         6    4.468096952924846D-06
         7         7    1.999992170933321D-01
         8         8    5.219371701989166D-07
         9         9    1.999999115190600D-01
        10        10    5.898728912613671D-08
        11        11    1.999992170933323D-01
        12        12    5.219371701989179D-07
        13        13    1.999932977943862D-01
        14        14    4.468096952924828D-06
        [...]


Title: Re: How should an input file for a cable element look like?
Post by: Prof. S. Govindjee on March 13, 2022, 11:32:44 AM

I wonder if it would be easier to start with zero body forces and ramp them up slowly?

Title: Re: How should an input file for a cable element look like?
Post by: crepes on March 14, 2022, 08:43:48 AM

Dear Professor Govindjee,

Unfortunately, ramping the body force slowly, starting from zero, did not help. I also played around with the other input parameters ( I also tried finite strains or higher degree nurbs).

Yesterday, I only tried either INIT DISP or non-zero y-coordinates. Now I also did both at the same time. It did not help very much. But I will keep on using both. Using zero y-coordinates seems to be no real option since this leads to zero columns and lines for even numbers in the tangent matrix.

(For FRAMe instead of CABLe, my input file works.)

I will try to find out why I get these extremely high entries (10^25) on the main diagonal for the cable problem. Changing Youngs modulus, Poisson ratio, mass density, body force do not alter these values. The cross section and changing the nurbs coordinates have an influence. I guess I will track some quantities in the element routines step by step.