You are on page 1of 4

Page 1 of 4

Ten Day ISTE Workshop on Computational Fluid Dynamics


Course Coordinators: Prof. Bhalchandra Puranik Prof. Atul Sharma Department of Mech. Engg. Project Coordinator: Prof. D. B. Phatak Dr. Mukta Atrey Project Manager: Department of Comp. Sc.

Indian Institute of Technology Bombay


Lab # 1: Finite Difference solutions Date: 16/06/2012
I. Diffusion Equation. Consider the problem of transient flow between infinite parallel plates, whose analytical solution was worked out. Recall that the governing equation for the unknown u(y, t) in this case is given by

Time: 2 PM - 3:30 PM

u 1 dp 2u 2 , t dx y
where we used

s being a constant.

1 dp s, dx

The boundary conditions of the problem are u(0, t) = 0 and u(h, t) = U, while the initial condition is u(y, 0) = 0. Additionally, the following data has been incorporated in the codes: = 0.000217 m2/s, = 800 kg/m3, U = 40 m/s, h = 4 cm. 1. Use the FTCS (Explicit) scheme (code Lab1_1_1D Unsteady Diffusion_FTCS Solver.sci) with tmax=3 s to compute the velocity within the domain for (i) dp/dx = 0, (ii) dp/dx = 20000 N/m2/m, and (iii) dp/dx = -30000 N/m2/m. Select the option of computing the time step using the stability criterion. 2. Use the BTCS (Implicit) scheme (code Lab1_2_1D Unsteady Diffusion_Implicit Solver.sci) with tmax=3 s to compute the velocity within the domain for a. t = 0.01 s, dp/dx = 0, b. t = 0.01 s, dp/dx = 20000 N/m2/m, c. t = 0.002 s, dp/dx = 20000 N/m2/m.

BEST OF LUCK Keep Playing with the OPEN-SOURCE Codes (developed by Mr. Vishesh Aggarwal, under the Supervision of Dr. Bhalchandra Puranik, IIT Bombay) in future also.

Page 2 of 4

(a)

(b)

(c) Fig. 1.1: Numerically and analytically calculated velocity profiles using fully explicit scheme for (a) dp/dx = 0, (b) dp/dx = 20000 and (c) dp/dx = -30000 N/m2/m

Discuss Fig. 1.1 here, limited inside this text box only

Page 3 of 4

(a)

(b)

(c) Fig. 1.2: Numerically and analytically calculated velocity profiles using fully implicit scheme for (a) t = 0.01 s, dp/dx = 0, (b) t = 0.01 s, dp/dx = 20000 N/m2/m, (c) t = 0.002 s, dp/dx = 20000 N/m2/m.

Discuss Fig. 1.2 here, limited inside this text box only

Page 4 of 4

II. Laplace Equation. Consider the problem of steady-state conduction in a rectangular domain, whose analytical solution was worked out. Recall that the governing equation for the unknown T(x, y) in this case is given by

2T 2T 0. x 2 y 2
The boundary conditions of the problem are T(0, y) = 0, T(a, y) = 0, T(x, 0) = 0 and T(x, b) = T1. Use the following data: a = 1, b= 3, T1 = 50, x-grid points = 20, y-grid points = 50. These values will be required to be provided when you execute the code Lab1_3_2DLaplacian.sci. Additionally, use three different values for the relaxation parameter: 0.5, 1, 1.5, and for each of these note the number of iterations (printed on the Scilab console as Step No) required to reach convergence. Comment on the use of relaxation parameter. Copy and paste the output plots (temperature contour plots) for one case:

(a)

(b) Fig. 2.1: Temperature contours: (a) Numerical solution, (b) Analytical solution.

You might also like