| Author | Message | |||
     Britta Hagmeyer (britta) Member Username: britta Post Number: 11 Registered: 11-2006 |
Hello everyone! I've built a 3D diffusion model in which substance enters the model from a finite source (e.g. a droplet). This means, the particle source has got a starting volume which decreases over time, and if the volume reaches 0, the flow of particles will stop . The amount of substance entering the model in one timestep depends on the volume which enters the model out of the source. I've realised the decreasing volume of the particle source by implementing the global variable vol into the system. The decrease of volume is being calculated by the law of Hagen-Poiseuille (Volume flow V out of the droplet through a round pore) and realised in the EQUATIONS section by dt(vol)= V. I've monitored the development of vol in a history plot, and the curve is not as smooth as it should be. Also, in the system there are exp-functions implemented in whose exponents the variable vol is being used, e.g. exp((-vol)/6.13249e-16). At a point where vol comes close to 0, FlexPDE stops the simulation with an 'exp argument too big'- error. I've worked around this difficulty by using an IF THEN ELSE construct, but I fear that something in my system is not working properly. Also, I've tried different parameters (s. .pde-file), and sometimes the exp-error occurs, and sometimes it doesn't. Another question: Is it possible to take the inflow of volume from the source into the system into account when calculating the particle concentration? Considering an incompressible fluid, the volume entering the system pushes the particles in the system further away from the inlet. Is there a formula to calculate this? I hope you can help me out! Greetings, Britta
| |||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 903 Registered: 06-2003 |
1. You have set the Threshold for U at 1, while the actual values of U are significantly smaller than this. THRESHOLD is "the smallest variation for which you wish FlexPDE to maintain accuracy". In declaring a threshold of 1, you have essentially allowed 100 percent oscillations in some areas. I believe this is what is causing the ragged behavior of Vol. Restore the Threshold of 0.001 that you have commented out, or make it even smaller, so that accuracy will be maintained in the small wings of the U distribution. 2. Your input flux is clamped at zero when Vol falls below 20e-20, but the time rate of change of Vol is not. So vol continues to go negative. This may not matter, since the flux is zero, but it looks funny on the plot. 3. In principle, you could model the pressure increase and fluid motion due to incoming material, but my guess is the diffusion dominates the process. |
problem file