|
1.3
Sept 4, 2015 18:06:45 GMT
Post by matthorr on Sept 4, 2015 18:06:45 GMT
GIVEN: The diffusion equation on the domain 0 ≤ x ≤ π with Dirichlet boundary conditions (\(u_a(x = 0, t) = u_b(x = \pi, t) = 0\), has an initial condition given by: \begin{equation} u(x,0) = \sum_{m=2}^5 \frac{1}{m}\sin(mx) \end{equation} REQUIRED: Plot the initial condition on the domain 0 ≤ x ≤ π and -0.5 ≤ u ≤ 1.5 . Derive the exact analytic solution as a function of time. Plot the solution at time t = 2, with ν = 0.05 on the same domain. How did the solution change shape with time? Why? (Problem 1.3 is based on Lomax et al Problem 2.7 –that you can get some guidance from: en.wikipedia.org/wiki/Heat_equation#Solving_the_heat_equation_using_Fourier_seriesIf you are having difficulty you should review separation of variables form any book on PDEs or engineering mathematics.)
|
|
|
1.3
Sept 8, 2015 23:33:17 GMT
Post by matthorr on Sept 8, 2015 23:33:17 GMT
Analytic Solution.
The diffusion equation is given by \begin{equation} \frac{\partial{u}}{\partial{t}} = \nu \frac{\partial^2{u}}{\partial{x^2}} \end{equation} Using separation of variables, let \begin{equation} u(x,t) = T(t)X(x) \end{equation} then \begin{equation} \frac{\partial{(XT)}}{\partial{t}} = \nu\frac{\partial^2{(XT)}}{\partial{x^2}} \end{equation} \begin{equation} \frac{1}{\nu T} \frac{\mathrm{d}T}{\mathrm{d}t} = \frac{1}{X} \frac{\mathrm{d^2}X}{\mathrm{d}x} = -\frac{1}{\lambda^2} \end{equation} Separating the two solutions, \begin{equation} T(t) = A e^{\frac{-\nu t}{\lambda^2}} \end{equation} \begin{equation} X(x) = B\cos(\frac{x}{\lambda}) + C\sin(\frac{x}{\lambda}) \end{equation} The general solution is then of the form (after multiplying through by \(A$\)) \begin{equation} \label{eq:general} u(x,t) = e^{\frac{-\nu t}{\lambda^2}}\left[D\cos(\frac{x}{\lambda}) + E\sin(\frac{x}{\lambda})\right] \end{equation} Next, the boundary conditions are applied to determine the coefficients. \begin{equation} u(x=0,t) = e^{\frac{-\nu t}{\lambda^2}}\left[D\right] = 0 \end{equation} Therefore, $D = 0$. \begin{equation} u(x = \pi,t) = e^{\frac{-\nu t}{\lambda^2}}\left[E\sin(\frac{\pi}{\lambda})\right] = 0 \end{equation} which gives the non-trivial solution \(\lambda = \frac{1}{n}\). The general solution for a given \(n\) is \begin{equation} u_n(x,t) = E_n e^{-\nu n^2t}\sin(nx) \end{equation} and for all \(n\) \begin{equation} u(x,t) = \sum_{n=1}^\infty E_n e^{-\nu n^2t}\sin(nx) \end{equation} Now, the initial condition is applied \begin{equation} u(x,t = 0) = \sum_{m=2}^5 \frac{1}{m} \sin(mx) = \sum_{n=1}^\infty E_n \sin(nx) \end{equation} giving \(E_n = \frac{1}{m}\) and \(n = m = [2,3,4,5]\). The final form of the solution with all boundary and initial conditions satisfied over the domain \(0 \leq x \leq \pi\) is \begin{equation} u(x,t) = \sum_{m=2}^5 \frac{1}{m} e^{-\nu m^2t}\sin(mx) \end{equation} The solution at \(t=2\) and \(\nu = 0.05\) is \begin{equation} u(x=[0,\pi],t=2)|_{\nu=0.05} = \sum_{m=2}^5 \frac{1}{m} e^{-0.1m^2}\sin(mx) \end{equation}
|
|
|
1.3
Sept 8, 2015 23:38:57 GMT
Post by matthorr on Sept 8, 2015 23:38:57 GMT
|
|