David Wasserman MEAE 6630 Conduction Heat Transfer HW # 6 In class exercise 3: 1)a) The heat conduction equation governing this steady state problem with heat generation is given by equations 3-187a-c in the text, with the addition of the Insulated boundary condition at r = 0. The steady state portion of this problem is given by equations 3-188a-b in the text. To solve this problem we can use a U substitution along with an integrating factor. If we let U = dT/dr then dU/dr = d2T/dr2. Then the steady state problem becomes a linear first order ODE like such: 1 1 U U g r k This is of the form: U P( r )U Q ( r ) We can solve for U using 1 U ( r )Q ( r )dr (r) P ( r ) dr (r) e 1 r g Q(r ) k We can solve for U and then integrate once more with respect to r in order to get T(r) : 1 gr 2 T (r) C 4 k Using the boundary conditions of T = 0 at r = b we find C to be gb2/4k and thus T(r) is g 2 b r 2 T (r) 4k b) We can also solve the problem numerically using three control volumes: two boundary volumes and an internal volume. The finite difference equations for the r = 0 boundary node and the interior node are, respectively: 2g T0 T1 4k 1 3 1 2g T1 Ti 1 Ti 1 2 2 2 k We do not need an equation for the r = b node because that temperature is held at zero for all times. Using the values for the k, g, and using a mesh spacing of 0.05 for delta, we can iterate for the final solution. Using an initial guess of zero temperature for all three nodes, after 10 iterations the temperatures at the nodes are T = 9.9902 K for r = 0m T = 7.4926 K for r = 0.05m P( r ) T = 0 K for r = 0.1m The analytic solution gives T = 10, 7.5, 0 for r = 0, 0.5, 0.1 respectively. As we can see, the results agree very well. c) To agree within 1% of the analytical results we would require temperatures of 9.9 K and 7.425 K. Since our numerical solution has given even better results than that, our mesh size is sufficient. 2)a) If we divide the slab into three volumes we get a node on the x = -L surface, a node at x = 0, and a node at x = L. We know that initially the slab is at 1000 K. For times thereafter, the nodes at the surfaces are maintained at 0 K. The implicit equation is: Ti n1 Ti n T n1 2Ti n1 Ti n11 i 1 t ( x ) 2 If we let n = 0 we know that the slab is at 1000 K, however the n+1 terms at the i-1 and i+1 nodes are equal to zero when the exponent is greater than 0. Therefore, they drop out of the equation, which can be rearranged as: ( x ) 2 Ti n 1 t Ti ( x ) 2 2 t If we use a time step of 1 second, a mesh spacing of 0.1 meter, and the alpha for pure copper, after 90 seconds we get a center temperature of 132.3961 K. The analytical solution can be obtained from the Heisler charts or by summing the first few terms of the series solution. The Heisler charts yield a center temperature of 100 K after 90 seconds. The more accurate method of summing the first four terms (m=1,2,3,4) of the series solution yields a center temperature of 105 K. Therefore, the numerical solution for the implicit method differs from the analytical solution by about 30%. b) c) The equation for the explicit method is: Ti n 1 rTi n1 (1 2r )Ti n rTi n1 r t ( x ) 2 Again the slab is initially at 1000 K and the boundary nodes are held at 0 K for times thereafter. For n > 0 the i-1 and i+1 nodes have temperatures of 0 K and thus drop out of the equation. If we use a time step of 1 second and a mesh spacing of 0.1 meters, and initial temperatures of 1000 K all nodes at n = 0, we get a center temperature of 129.3566 K after 90 seconds. Again, this differs from the analytic solution by about 30%. d) If I use a time step of 0.1 seconds and seven nodes, the temperature at the center of the plate is 105.9509 K, which agrees with the analytical solution within 1%.