1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Homework Help: Fluid dynamics, solving ODE to find particle path

  1. Feb 9, 2016 #1
    1. The problem statement, all variables and given/known data

    A time-dependent two-dimensional fluid flow is given, in a Cartesian coordinates system (x, y), by the velocity field:

    u = (y, t-x)

    Show that, at time t = 0, the streamlines of this flow are circles centred on the origin.
    Find equation of the streamline that passes through the point (0, 1)
    Find the equation of the path of a particle released at the point (0,1) at t=0.

    2. Relevant equations

    For streamlines in u = (u, v):
    dx(t)/u = dy(t)/w

    Particle path:
    dx/dt = u
    dy/dt = v

    3. The attempt at a solution

    I think I'm okay with the streamline part but here it is anyway:

    Using the given equation for finding streamlines, plug in the values and rearrange to get:

    (t-x) dx = y dy

    y2/2 = tx - x2/2 + C

    Rearrange to show x2+y2 = 2tx + C

    For t = 0 , it reduces to x2+y2 = + C,
    which is a circle.

    Streamline passing though (0, 1) is given in:
    x2+y2 = C = 02+12 + C
    C = 1

    Streamline is x2+y2 = 1

    Now, onto the particle path. This is the bit I'm stumped on.

    I've plotted the stream flow in mathematica and intuitively it seems the path should be:
    (x, y) = (sin[ t ] + t, cos[ t ])
    But my answer is coming out slightly different.
    Here goes...

    Velocity u = (y, t-x)
    Particle path:
    dx/dt = u
    dy/dt = v

    dx/dt = y
    dy/dt = t-x

    Taking the 2nd derivative of the first part, we get
    d2x/dt2 = dy/dt = t - x
    This gives a differential equation for x(t):

    d2x/dt2 + x = t

    Solving the left hand side:
    d2x/dt2 + x = 0

    Using characteristic method to find particular solution, setting xc(t) = Aeλt
    λ = ±i
    xc(t) = Aeit + Be-it

    Then to solve for the full differential equation, d2x/dt2 + x = t
    Try particular solution xp(t) = t
    This appears to work, giving finally:
    x(t) = Aeit + Be-it + t

    Then, from the information above, we have:
    dx/dt = y
    y(t) = Aieit - Bie-it + 1

    So, final solution for positions x(t) and y(t) are:

    x(t) = Aeit + Be-it + t
    y(t) = Aieit - Bie-it + 1

    The information states that the particle is released at (x, y) = (0, 1) at time t=0
    So this can be used to find constants A and B:

    x(0) = Ae0 + Be0 + 0 = 0
    A + B = 0
    A = -B

    y(0) = Aie0 - Bie0 + 1 = 1
    Substitute A = -B
    -Bi - Bi + 1 = 1
    -2Bi = 0
    B = 0

    A = -B = 0

    So both functions x(t) and y(t) are reduced to zero for the given boundary conditions.

    And this is where I am lost.


    Thanks for any help anyone can give, and thanks for taking the time to look at this because I know it's a large post!
  2. jcsd
  3. Feb 9, 2016 #2


    User Avatar
    Science Advisor
    Homework Helper
    2017 Award

    I read (x,y) = (t,1) , not (0,0) :smile: ?

    And you can check that it satisfies the equation ! And the initial conditions ! Well done :rolleyes: !
    Last edited: Feb 9, 2016
  4. Feb 9, 2016 #3
    I'm sorry but I don't understand what you mean?

    Where have you got (x,y) = (t,1) from? Where did you read that? And where did you see any mention of (0,0)?

  5. Feb 9, 2016 #4


    User Avatar
    Science Advisor
    Homework Helper
    2017 Award

    So if A and B are 0 then x = t and y = 1, in other words (x,y) = (t, 1)
    Which I interpret as x = 0 and y = 0, in other words (x,y) = (0, 0)​
  6. Feb 9, 2016 #5
    Ahh I see what you mean now.

    So it does satisfy the equation for the given boundary conditions (woohoo!)

    BUT, the problem is that I'm certain it's wrong.

    The original velocity vector, u = (y, t-x), describes a circular vector field for any fixed time t. Like this:
    http://student.ulb.ac.be/~lclaesse/phystricks-documentation/_build/html/_images/Picture_FIGLabelFigChampVecteursPICTChampVecteurs-for_eps.png [Broken]

    As time increases, the field drifts to the right.

    This would surely mean that the particle path describes something a little bit 'loopy'.

    As I said, intuitively I'd have though the particle path is (x, y) = (sin[ t ] + t, cos[ t ]) . This would satisfy the start point (0,1) at t=0, and would describe a path that nicely follows the rightward-moving circular vector field that the particle exists in.

    However the answer I've got is (t, 1), which is a basically a straight line at y=1 ! :oldfrown:

    Good to know I got the ODE right but it seems I'm still wrong somewhere. What are your thoughts?


    EDIT: Hang on a minute... Maybe I'm not wrong??? If the vector field moves to the right at the same rate of its circular rotation, maybe the particle will sit at that y=1 spot continually...?
    Last edited by a moderator: May 7, 2017
  7. Feb 9, 2016 #6
    Actually... it makes more sense for it to stay on that straight line y=1...

    For every moment of time that passes, the vector field moves a step to the right, meaning that there is never an 'opportunity' for the y-component of the velocity to change its value away from 0, because the component uy = t - x , with the initial condition x = 0, t = 0, so of course t - x = 0 for the whole time.

    Fantastic. I think I've done it.

    Happy days.

    Thanks for the help!
  8. Feb 9, 2016 #7


    User Avatar
    Science Advisor
    Homework Helper
    2017 Award

    You have indeed! Well done (again) :rolleyes: !

    And you'll note that this surprising trajectory comes out for this particular starting point only.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted