FEAP User Forum

FEAP => Input File Issues => Topic started by: lredmond on April 13, 2013, 05:32:17 AM

Title: Time value changing during solution step
Post by: lredmond on April 13, 2013, 05:32:17 AM
I have the following issue with a recent input file:


The residual norm after the 1st time step is either NaN or inifite, and the time step appears to change. Rather than t=0.02 for all three values, it reads as follows:

                                                           t=   135.78     0.00

   Solution tolerance =  1.00000E-07

                                                           t=   135.81     0.00
   End Triangular Decomposition                            t=   222.62     0.00
  Updating history after #         8056  iterations

   Computing solution at time  1.0000E-02: Total proportional load  1.0100E+00
   Individual factors:
      1 =  1.0000E-02   2 =  1.0000E+00
  *ERROR* Residual norm is NaN or Inf

When I re-run the file, the values of t for the first step change each time (138.5, 120.8, etc.)

I have the following for the input of the load and the solution commands:

BATCh
  PROP,,1,2
END
   1,0,0,100,0,1,0,0,0
   1,0,0,100,1,0,0,0,0

BATCh
  TPLOt,,1
END
REACs and DISPs...


BATCh
  NOPRint
  DT,,0.01
  TOL,,1e-7
  TANG
  LOOP,print,400
   LOOP,time,1
    TIME
      LOOP,iter,3000
   FORM
   SOLV
      NEXT,iter
    NEXT,time
   DISP,all
    STRE, all 
   REAC,all
  NEXT,print
plot mesh
plot fill
plot node
plot cont 1

END
INTER

Can anyone think of anything that I could have done to cause the time step to drift like this?

Thanks
Laura
Title: Re: Time value changing during solution step
Post by: Prof. R.L. Taylor on April 13, 2013, 08:40:12 AM
The time values you see are the CPU time and (when it is active) SYSTEM time.  They do not have anything to do with your set of DT or failure of the analysis. 

You should check your input file carefully before trying to do so many time steps.  Use plot to check that the mesh is correct (plot mesh, plot bound, plot load, etc.)  also enter solution command CHECk to make sure you do not have badly numbered elements.  Also check the output file that what you think you input is what the program read.

NaN's are failure of the analysis.  They can be caused by a variety of issues --many times just a bad mesh.  But if you have inelastic materials then they can fail due to yielding, etc.