Doubts in implementing User element

[pradeep]

Hi everyone,

As I am new to User Elements in FEAP, My questions might be naive but please help me out  .
I am trying to write a user element with 1 DOF which is concentration(Using Ficks Law). I am using implicit scheme to solve the time derivative.So I have a few queries regarding how it should be implemented-
1.How do I extract time step from the FEAP Input file?
2.I would need concentration values of previous time step to solve for the next time step.How can I extract the solution vector of previous step.
3.How can I prescribe flux to a boundary using FEAP Input file?(This question would also apply to prescription of Thermal Flux.)

Thanks in Advance.

[Prof. R.L. Taylor]

The programmer manual describes how displacements are transmitted to the element program.  Basically, the array ul(ndf,nen,*) gives all the information you need.

ul(ndf,nen,1) are values for the current time, which we usually denote as t_n+1 -- for ndf dof's, nen nodes.

ul(ndf,nen,2) are values of the difference between solutions at t_n+1 and t_n, the previous step.

thus the solution at t_n is obtained as ul(ndf,nen,1) - ul(ndf,nen,2)


For flux boundary conditions on your element you will need to  perform the integrations and compute the loading.  For thermal problems there are some convec1d.f convec2d.f convec3d.f routines in the thermal subdirectory you can consult for how these are implemented in feap.

[pradeep]

Thanks for your earlier reply,It was helpful.
I am encountering a problem in prescription of concentration to the surface.
I am working with brick elements(attached geometry screenshot)
It is like placing a brick over the other and the top surface of top brick has been prescribed with concentration of 10 units.
The ul vector doesnot reflect the initial boundary contions before the first time step,  which is leading to negative concentration values in the bottom element.Can I know why exactly is the intial boundary condition not being prescribed to the ul vector .
.I am attaching input file ,user element and also text file with printed ul vector.
Thank you.

[FEAP_Admin]

I'm not sure that the negative values are not 'correct'.  With Fourier's law (Fick's law) and traditional time stepping schemes oscillations are a known artifact of step loadings.  I believe that Isaac Harari has written a paper about this -- I. Harari, “Stability of semidiscrete formulations for parabolic problems at small time steps,” Computer Methods in Applied Mechanics and Engineering 193(15-16), 1491–1516 (2004).

[pradeep]

Thank You for replying.
I implemented implicit scheme for the time derivative which should be unconditionally stable.
KU=R is what I am solving for where R(Integral of(shpfn*shpfn*ul)) aso contains my solution from previous time step.This means when Time is zero(at t=0) ul vector should reflect my initial boundary conditions(ul = [ 0 0 0 0 20 20 20 20 ]) for element 2 and for element 1 it should be ( 0 0 0 0 0 0 0 0 ]).But I cannot see it when I print out my ul vector.I also attached the printed values above.