Page 164 - Process Modelling and Simulation With Finite Element Methods
P. 164

Extended Multiphysics                151

          Simply,  the  mass  transfer  flux  and  surface  reaction  constraints  are  not  time
         evolution  equations.  FEMLAB  naturally  gives  these  equations  in  the  weak
          form  a  ddt  accumulation  term  in  the  time  dependent  solver.  Without  the
          time  evolution  terms,  these  algebraic  constraints  make  the  system  rather
          stiff.  The  additional  coupling  variables  also  subtract  from  the  "sparsity"  of
          the system, thereby  making the sparse matrix solvers strain harder to converge.
          So  don't  be  surprised  if  the  solver  steps  in  extended  multiphysics  problems
         take longer.

               Time=O.l,  concentration of U and V)   Time=O 1,  concentration of US and VS












                0'5  ;   1'5  ;  7'5  4   3'5  ;  4'5  ;   01   " " " " "
             03A                              0   05   1   15   2   25   3   35   4   45   5
                       z coordinate                    z coordinate
           Figure 4.8  Bulk concentrations decay SIOwlY.   Figure 4.9  US is still depleted throughout the
                                           reactor.

         The results are shown in Figures 4.8 and 4.9.  One would typically expect that in
         a  nonlinear  reactive-diffusion  problem  at  low  Peclet  number  (about  OS), the
          steady state would be well on its way at t=0.1.  But the influence of the imperfect
         mixing  in the buffer  tank  is  apparent.  The bulk  and  surface concentrations  are
          slowly developing towards their steady state profiles, but the de-charging of the
         capacitor  is  a  slow process,  which  will  hamper  the  development  of  the  steady
          state.  If  further  system  time  lags  were  modelled,  say  the  piping  and  control
          systems, then it is likely that  this  buffering resistance  to developing  the  steady
          state from start up would be exacerbated.

         Exercise 4.1
         In  53.3,  we  added  an  oscillatory  disturbance  to  the  steady  state  solution  of
          a  tubular  reactor  and  observed  the  transient  response.   Store  the  steady
          state solution to this heterogeneous reactor with buffer tank process, then create
          m-file  functions  for  the  initial  conditions,  adding  an  oscillatory  disturbance
         to  U  only.  How  many  m-file  functions  do  you  need  to  specify  the  steady
          state condition  with  oscillation?  Comment  about  the  transient  response  of  the
          system.
   159   160   161   162   163   164   165   166   167   168   169