$f''+2 \eta f-4mf=0$ where $f=1$ at $\eta=0$, $f \to 0$ as $\eta \to \infty$. The case of $m=0$ arises in the analysis of motion of fluid above a suddenly accelerated flat plate.
Asked
Active
Viewed 946 times
1
-
Are you interested in HOW to solve this equation numerically? or you just wanna look at what the numerically solution looks like. – Shuhao Cao Mar 25 '13 at 18:45
-
yeah, I want to know the details procedure of numerical solution – hellomecha Mar 25 '13 at 18:54
-
Non-dimensionalize around $\eta$ and let $\infty$ be approximately 5. – Emily Mar 25 '13 at 18:57
-
How can I non-dimensionalize the equation around η ? I want to solve it by shooting method ? Is it at all possible preferably using MATLAB? – hellomecha Mar 25 '13 at 19:03
-
Sure, you can do shooting, and you can implement shooting in MATLAB. I assume you know how to do this. The boundary layer flow profile above a flat plate starts at zero at the boundary, and as you go upwards, it pretty rapidly approaches the steady-state flow velocity. Mathematically, we treat this as the limit to infinity. However, if you non-dimensionalize your units (make $\eta$ be on the scale of unity), then you don't need to go out to infinity. It's sufficient to set $f(\eta=5) = f(\eta = \infty)$ as your boundary condition. – Emily Mar 25 '13 at 19:16
-
You got the point- "Implementation of shooting method in MATLAB" Can you help me with that ? – hellomecha Mar 25 '13 at 19:24
-
I can easily convert the above BVP into an IVP with the following set of differential equations. f'=g (let) z'=f"= 4mf-2ηf Here, for the solution we need two conditions.One is already there i.e., f=1 at η=0.Now I need to guess another condition for f '(η=0). How can i find my initial guess and how to confirm the boundary condition f(η=∞/5)=0 is satisfied by that wild guess ? – hellomecha Mar 25 '13 at 19:39
1 Answers
1
The general solution of your differential equation is $$ f(\eta) = a \text{Ai}(2^{1/3}(2m-\eta)) + b \text{Bi}(2^{1/3}(2m-\eta))$$ where Ai and Bi are Airy functions, and $a$ and $b$ are arbitrary constants. Now both $\text{Ai}(t)$ and $\text{Bi}(t)$ go to $0$ as $t \to -\infty$: according to Wolfram Alpha, $$\eqalign{\text{Ai}(-t) &= \frac{t^{-1/4}}{\sqrt{\pi}} \sin(2 t^{3/2}/3+\pi/4) + O(t^{-7/4})\cr \text{Bi}(-t) &= \frac{t^{-1/4}}{\sqrt{\pi}} \cos(2 t^{3/2}/3+\pi/4) + O(t^{-7/4})}$$ So the boundary condition at $\infty$ does not restrict the solution. You need another boundary condition to determine a solution.
Robert Israel
- 448,999
-
It's a second order differential equation and we have two boundary conditions already. Isn't that sufficient enough ? – hellomecha Mar 25 '13 at 19:09
-
Not if all the solutions satisfy one of the boundary conditions. – Robert Israel Mar 25 '13 at 19:14
-
I think you have provided the analytical solution but I am interested in numerical solution.Okay, after the completion of numerical solution we can compare the the two results. – hellomecha Mar 25 '13 at 19:17
-
Are you sure you have the sign of the $2\eta f$ term right? It would work for the equation $f'' - 2 \eta f - 4 m f = 0$ (or equivalently if you wanted $f \to 0$ as $\eta \to -\infty$). – Robert Israel Mar 25 '13 at 19:26
-
My prof provided me the above equation & I think it's correct. Ref- Boundary Layer theory by H Schlichting seventh edition (page -90) – hellomecha Mar 25 '13 at 19:44