View Full Version : General Method of Characteristics Problem
Hello!
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.
jackmell
Sep26-10, 07:12 AM
You're using the method of Lagrange right? If so, you should be solving:
z\phi_x+\phi_y+z\phi_z=0
with characteristic system:
\frac{dy}{dx}=\frac{1}{z},\quad\quad \frac{dz}{dx}=1
Now, I haven't worked it out manually yet, but we should get:
\phi(x,y,z)=C(z-x,y-\log(z))
using the method of characteristics. Then the solution is:
C(u-x,y-\log(u))=0. And an interesting particular case involving the Lambert W function is:
u-x+y-\log(u)=0
So we got the start, the end, and now just need to fill-in.
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 \frac{dy}{dx} = \frac{b}{a}. In this case a = u, b = 1, c = -u.
So, \frac{dy}{dx} = \frac{1}{u}, integrating yields uy=x+c1 or y=\frac{x+c_{1}}{u}.
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.
jackmell
Sep26-10, 10:05 AM
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 \frac{dy}{dx} = \frac{b}{a}. In this case a = u, b = 1, c = -u.
So, \frac{dy}{dx} = \frac{1}{u}, integrating yields uy=x+c1 or y=\frac{x+c_{1}}{u}.
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.
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.
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: \frac{dy}{dx}=\frac{b}{a}=\frac{2x}{1}= 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))
substituting...
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.
Edit:
I just reworked everything and got u = -5, so that might be correct.
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.