Solving coupled PDEs

  1. Hi,

    I am trying to simplify the following equations to get a relationship involving just [itex] \eta [/itex]:

    1) [itex] \nabla^2 \phi(x,z,t) = 0 [/itex]

    for [itex] x\in [-\infty,\infty][/itex] and [itex] z\in [-\infty,0] [/itex], [itex] t \in [0,\infty] [/itex]

    subject to the boundary conditions

    2) [itex] \phi_t+g \eta(x,t) = f(x,z,t)[/itex] at z=0

    3) [itex] \eta_t = \phi_z [/itex] at z=0


    4) [itex] \phi \to 0 \ as \ z \to -\infty [/itex]

    Here, g is a constant, [itex] \eta, \phi [/itex] are the dependent variables of the system and f represents a forcing function. Another important constraint is that for systems I'm interested in, f is non zero only for a small time interval.

    For the case where f=0, one can find that

    [itex] \eta_{tt}-\frac{g}{k} \eta_{xx} =0 [/itex]

    where k is the wavenumber of the system.

    I want to find an analogous relation when forcing is present.

    Any help is appreciated,

  2. jcsd
  3. kai_sikorski

    kai_sikorski 162
    Gold Member

    Seems to me like you need some conditions on the t=0 face of the domain.
  4. In the case where [itex]f=0 [/itex] it is clear, from the fact that [itex] \eta(x,t) [/itex] is governed by a wave equation, that we will need to know [itex]\eta (x,0)[/itex] and [itex]\eta_t(x,0)[/itex] to completely describe the (d'alambert) solution. Let us say that before the forcing occurs, we know both [itex] \eta(x,0) , \eta_t(x,0) [/itex].

    I do not see how this helps me find the 'particular' solution to this system of equations.

    This problem comes from physics - namely, it's the solution to (conservatively) forced, inviscid, irrotational surface gravity waves. The forcing that I'm interested acts in a 'spatially compact' region over a short time, say from [itex] [t_o,t_o+\Delta t] [/itex]. As a first step, i'm trying to solve this in the limit that the forcing is all concentrated at a particular point in space and time [itex](x_o,z_o,t_o) =(0,0,0) [/itex] but have not made any headway.
  5. Also, an alternative way of looking at this problem is the following: The form of the Bernoulli equation in post 1 (condition 2) comes from

    [itex] \vec{u}_t=-\frac{1}{\rho} \nabla p + \vec{F} [/itex]

    Where [itex] \vec{F} = \vec{\nabla} f [/itex].

    The reason I took the route I did in post 1 was to avoid discussion of the pressure field, but an alternative way to look at this problem is by resolving this field. By taking the divergence of the Navier Stokes equation, we find

    [itex] \nabla^2 p = \nabla \cdot \vec{F} [/itex]

    such that p=0 at z=0 and [itex] \nabla p \to 0 [/itex] as [itex] x \to \pm \infty [/itex]

    If I can solve for the pressure field, then I can find the vertical velocity, [itex] \phi_z [/itex] at z=0 and then from there resolve the form of [itex] \eta(x,t) [/itex]

    I am trying to solve this for a very simple form of the forcing - namely [itex] F = C_o \delta(x_o,z_,t_o) \ \hat{x} [/itex] but have not made much progress.
Know someone interested in this topic? Share a link to this question via email, Google+, Twitter, or Facebook