Page 347 - Process Modelling and Simulation With Finite Element Methods
P. 347

334         Process Modelling and Simulation with Finite Element Methods

         Now select the Restart button on the toolbar.  The initial condition now has the
         pseudosteady  velocity and electric fields set up.  After the solution, which takes
         about  six  minutes  on  a  PI11  866MHz  Intel  machine,  use  Post  mode/Plot
         parameters  to  set up  the  surface color plot  of  species Y,  the  contour plot  for
         potential  lines,  and  the  arrow  plot  for  velocity  vectors.  This  should  resemble
         Figure 9.8. Then animate and watch the streams merge and flow out.  We can see
         that the final velocity and potential fields are unchanged from those set up in the
         initial  “fast  elliptic  solver”  step.  This  would  not  be  true  if  the  conductive
         properties  of  the  fluid  or  viscosity  were  concentration  dependent.  The  local
         concentration  profile  would  then  globally  affect  the  potential  field,  thus
         modifying the velocity field, which in turn disperses the fluid differently.  Such
         highly  coupled  electrokinetic flows are well  suited to  numerical  analysis  by  a
         modest  change  to  this  coding  - adding  the  concentration  coupling  to  the
         conductivity  according  to  a  mixture  rule.  Zirnmerman  and  Homsy  treat  the
         mixture  rule  for  concentration  dependency  of  viscosity,  for  instance,  in  the
         instability of viscous fingering in porous media [2].

         Y-Junction Switcher: An Application of Linkages Through Coupling Variables
         As  described  in  Figure  9.6,  by  including  relatively  long  channels  to  the
         reservoirs  of  the  component  species where  electric  potentials  are applied, the
         flow  and  the  electric potential  can  be  described  by  algebraic relations  for the
         inlet and outlet dependencies.  We implement these physical linkages between
         nodes a  and A,  for instance, using the appropriately named coupling variables.
         Originally, we  set up  a pseudo 0-D geometry  (geom2 as in  Chapters four and
         seven), but on advice from COMSOL consultants, realized that the conceptual 0-
         D domain was not needed - it can all be done in coupling variables.  Using the
         model  MAT-file  of  the  last  section,  set  up  the  following  twelve  coupling
         variables:

         Select AddEdit Coupling Variables from the Options Menu.


          scalar add Qa.  Source Geom 1, subdomain 1, boundary 3.
          Integrand: zetal *(PHIA-volta)/dsa-Re*(PA-bara)/f/dsa
          Integration order: 2
          Destination Geom 1 bnd 3 Check “Active in this domain” box.
          scalar add Qb.  Source Geom 1, subdomain 1, boundary 1.
          Integrand: zeta 1 *(PHIB-voltb)/dsb-Re*(PB-barb)/f/dsb
          Integration order: 2
          Destination Geom 1 bnd 1 Check “Active in this domain” box.
          scalar add Qc.  Source Geom 1, subdomain 1, boundary 9.
          Integrand: zetal “(voltc-PHIC)/dsc-Re*(PC-barc)/f/dsc
   342   343   344   345   346   347   348   349   350   351   352