| Author | Message | |||
     Bob Gardner (bgardner) New member Username: bgardner Post Number: 1 Registered: 10-2003 |
Dear Dr Nelson, I am just beginning to learn how to use FlexPDE and have a few questions. I would like to use FlexPDE to simulate the effect of tides on the watertable of beaches and salt marshes. On the beach face I need to be able to change boundary conditions from value to natural depending on the elevation of the tide and the exit point of the watertable at the top of the seepage face, if any. For the subtidal portion of the beach face the pressure (or head) can be specified as a function of the tidal elevation. For the seepage face portion (if any) of the beach face the pressure can be set equal to zero at each point along the seepage face (or the head can be set equal to the elevation of each point along the seepage face). At points along the beach face above the seepage face there is no flow out of the aquifer domain so a natural condition is apropriate. Thus as the tide rises and falls I need to be able to switch back and forth between value and natural boundary conditions on the beach face. Can this be done in FlexPDE? My next question concerns overriding the dynamic time step feature of FlexPDE in transient problems. In the attached pde file I made a stab at the beach problem by specifying heads all along the beach face using Turners equations for predicting the time of first appearance of the seepage face and the position of the exit point through time. The issue here is not the validity of the formulation but rather the fact that during the run the time step appears to exceed the value of fixdt specified in the select section. Have I not specified fixdt properly? Finally in the plots is there anyway to introduce vertical exaggeration? In groundwater problems it is commom for the horizontal dimension of an aquifer to exceed the vertical dimension by one or more orders of magnitude. Without vertical exaggeration it is often difficult to see features in the contour plots. Can one zoom in on certain portions of such plots? Thank you for your help. Bob Gardner
| |||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 43 Registered: 06-2003 |
1. FlexPDE will not let you dynamically switch the class of a boundary condition (from VALUE to NATURAL, etc). But you can write a NATURAL BC that simulates a VALUE, and switch it to a "natural" NATURAL at will. (The meaning of the natural will depend on your equation, so what I say here must be adapted to the actual conditions.) For the equation div(k*grad(u))+s = 0, for example, the natural BC means the incoming flux of u (since the argument of the div is the negative flux). If the value of u at the boundary is less than the target value U0, we want an incoming flux to drive it upward. So if I write NATURAL(u)=Large*(U0-u), the value of u will be clamped very near to U0, simulating a VALUE BC. I can further switch the Natural, writing something like NATURAL(u)=IF(a<b) THEN Large*(U0-u) ELSE influx. Hard IF..THEN switches sometimes cause oscillations in an application like this, so you might want to use a SWAGE or RAMP function (q.v.) instead, to put a little fuzz on the transition. 2. FIXDT in the SELECT section is merely a TRUE/FALSE flag, and the value of DT is taken from the TIME statement. When you write "FIXDT=number", the parser sets the flag to the integer value of the number, which in your case is 0 or OFF. Take out the =0.01 and it should work. (I know this syntax is clumsy - I will look into changing it). 3. If you ZOOM a plot, the zoom area is remapped to a square. So ZOOM(0,0, 10,1) will exaggerate the Y dimension by a factor of 10. |
beach2.pde