- 积分
- 75
- 注册时间
- 2002-10-7
- 仿真币
-
- 最后登录
- 1970-1-1
|
楼主 |
发表于 2006-1-19 12:07:18
|
显示全部楼层
来自 上海浦东新区
Re:defining periodical changing boundary conditions
Hi,
At the risk offending all users again by plugging my book, I did
something similar with Jordan MacInnes in Chapter 9 of Process
Modelling and Simulation with Finite Element Methods...
We wanted a "square wave" input to each of two voltages controlling
an electrokinetic, microfluidic flow in a Y-junction. Each leg of
the Y has to alternate between "idle" and "on" voltages so that flow
switches downstream with source from the other leg. You might think
that this can be easily done
by altering your Dirichlet boundary conditions to include sinusoids
in time with the appropriate logical operator to turn it into a
square wave (sine > 0 for instance) and out-of-phase behavior for
each leg. I tried this, and it is not straightforward because of the
elliptic nature of my PDE system and because the FEMLAB
implementation involves weak boundary constraints. My system
was "stiff" as it involves an infinitely fast time scale, so I tried
approximating the square wave with a few terms of its Fourier series
to smooth it, and then I simply did the periodic switching in a
Matlab driver file that calls the Femlab subroutines with the
appropriate changes to the contants and the solution in the last time
interval as the initial condition. If your system is not "stiff", I
think the weak boundary constraint that is time varying ought to work
for you.
Regards,
Will Zimmerman |
|