Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

General Method of Characteristics Problem

  1. Sep 25, 2010 #1

    I am trying to solve...

    u*ux + uy -u = 0 with i.c. u(x,0) = x +10

    Determine: a.) characteristic equation
    b.) compatibility equation
    c.) value at u(5,10)

    I've tried the general method of determining the characteristic equation dy/dx=b/a and attempted a parametrization but I don't think I am coming-up with the correct answer. I keep getting Us or Xs in the result - I am assuming there is a clean answer but I could be wrong.
    Last edited: Sep 25, 2010
  2. jcsd
  3. Sep 26, 2010 #2
    You're using the method of Lagrange right? If so, you should be solving:


    with characteristic system:

    [tex]\frac{dy}{dx}=\frac{1}{z},\quad\quad \frac{dz}{dx}=1[/tex]

    Now, I haven't worked it out manually yet, but we should get:


    using the method of characteristics. Then the solution is:

    [tex]C(u-x,y-\log(u))=0[/tex]. And an interesting particular case involving the Lambert W function is:


    So we got the start, the end, and now just need to fill-in.
    Last edited: Sep 26, 2010
  4. Sep 26, 2010 #3
    Well, I'm learning this from a section in Gas Dynamics by Zucrow. It only presents one method...

    It says to find the characteristic from [tex]\frac{dy}{dx}[/tex] = [tex]\frac{b}{a}[/tex]. In this case a = u, b = 1, c = -u.

    So, [tex]\frac{dy}{dx}[/tex] = [tex]\frac{1}{u}[/tex], integrating yields uy=x+c1 or y=[tex]\frac{x+c_{1}}{u}[/tex].

    The compatibility equation would be adu +cdx = 0.

    But maybe I am supposed to use one of the other methods as you suggested, it just didn't go over any of that. I didn't know there were multiple methods.
    Last edited: Sep 26, 2010
  5. Sep 26, 2010 #4
    I'm not familiar with that approach. Also, it's quasi-linear so seems to me, need two characteristic equations but I'm not sure. Never seen "compatibility equation" also. The method I use, which ultimately yields the equation C(r,s)=r-10 for you initial conditions, is summarized in "Basic Partial Differential Equations" by Bleecker and Csordas.
  6. Sep 26, 2010 #5
    Yeah, you're correct, quasi-linear hyperbolic PDE.

    The reason the "compatibility equation" is needed is that the governing relations for the particular flow-field are PDEs, they need to be reduced to total differential equations. The characteristics are the paths of a physical disturbance, in this case, Mach waves. The total differential equation (i.e., compatibility equation) is valid only along the particular characteristic specified by dy/dx = b/a.

    It notes that the compatibility and characteristic equations must, in general, be solved simultaneously. This may be why I am getting 'u's in the results... but I wouldn't imagine that it would have given me such as a problem in the book, at least without noting that it required a numerical method.

    IOW, I need to know 'u' in order to solve for 'u'.

    Here is the book example...

    ux + 2xuy - 3x2 = 0

    i.c., u(o,y) = 5y +10

    Determine: a) the equation of the characteristic passing through (2,4)
    b) compatibility equation valid along that characteristic
    c) the value u(2,4)

    so, coefficients are a = 1, b = 2x, and c = - 3x2

    Characteristic Equation: [tex]\frac{dy}{dx}[/tex]=[tex]\frac{b}{a}[/tex]=[tex]\frac{2x}{1}[/tex]= 2x

    Integrating: y = x2 +C1

    Characteristic through point (2,4): C1 = 0, y = x2

    Comaptibility Equation is obtained from adu + cdx = 0,

    substituting and integrating gives: u = x3 + C2

    The value of C2 valid along the characteristic that we found above is determined from the i.c. And the i.c. requires that y at x = 0 be determined on the characteristic.

    so, y = (0)2 = 0

    Hence the characteristic passing through (2,4) crosses the initial-value line at (0,0).

    u(0,0) = 5(0) +10 = 10 (Value of u along the characteristic at (0,0))


    10 - (0)3 = C2, C2 = 10

    giving, u = x3 + 10 (Compatibility equation valid along the characteristic passing through (2,4))

    so, u(2,4) = (2)3 + 10 = 18

    I don't know if that helps or not.


    I just reworked everything and got u = -5, so that might be correct.
    Last edited: Sep 26, 2010
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook