Flow Lab

Syllabus / Module 14: Applications: Physics & Engineering

Lesson 14.4

Engineering Heat Transfer

Conduction, convection and radiation; cooling fins as BVPs; lumped capacitance; transient conduction; Biot number.

Written byClaude Sonnet

🌊
Why this matters for simulation: Convective heat transfer couples temperature to flow — buoyant smoke is the same physics.

By the end of this lesson you can…

  • Name the three modes of heat transfer, write Newton's cooling law with a coefficient $h$, and linearise radiation into $h_{rad}$.
  • Form the Biot number $\mathrm{Bi} = hL/k$ and decide when the lumped-capacitance ODE is a valid model.
  • Find the transient-slab eigenvalues from $\zeta\tan\zeta = \mathrm{Bi}$ by bisection and compare the series with the lumped model for $\mathrm{Bi} = 0.1, 1, 10$.
  • Derive the cooling-fin BVP $\theta'' = m^2\theta$ and compute tip temperature, heat rate and efficiency $\tanh(mL)/(mL)$.
  • Combine thermal resistances like circuit resistors, and relax 2D steady conduction to get a shape factor.
On this page
·
  1. Three ways heat moves
  2. Newton’s cooling, lumped capacitance and the Biot number
  3. The slab with a convective boundary
  4. Lumped against slab: what Bi buys
  5. Cooling fins: a boundary-value problem
  6. The sink as a network of resistances
  7. 2D steady conduction
  8. Heat exchangers: a glimpse
  9. The lab
  10. The exercise: fin and Biot eigenvalue
  11. Concept checks
  12. Problems
  13. What comes next
  14. Key Takeaways

14.3 Pendulums: Linear, Nonlinear, Driven & Double ended on the nonlinear term that broke superposition. Heat transfer is the opposite case: inside a solid the physics is the linear heat equation of 9.3 Solving the Heat Equation: Separation of Variables. What is new is the boundary, where a hot body loses heat to a fluid through a Robin condition with a coefficient hh. One dimensionless number, the Biot number, says whether the PDE is needed or a first-order ODE like the cooling coffee of 3.5 Modeling with First-Order ODEs will do. The rest uses the course so far: the fin is a boundary-value problem (8.1 Two-Point Boundary Value Problems), the sink a resistor network (14.2 Electrical Circuits), and 2D conduction a relaxation (10.8 Elliptic Solvers: Jacobi, Gauss–Seidel & SOR).

▶ Video Conduction, convection and radiation; the Biot number decides between the lumped ODE and the slab series; a cooling fin as a boundary-value problem; a heat sink as a resistor network; 2D conduction by relaxationcaptions only
Transcript
  1. Heat moves three ways. Conduction is diffusion through a solid, Fourier's law. Convection is heat carried off by a moving fluid, which engineers squeeze into Newton's cooling law with a coefficient h. Radiation needs no medium: a surface emits epsilon sigma T to the fourth, with T absolute. Factor T to the fourth minus T infinity to the fourth, and radiation also becomes Newton cooling, with h radiation equal to epsilon sigma times T plus T infinity times T squared plus T infinity squared. For an anodised surface at 80 degrees in 25 degree air, with epsilon 0.9, h radiation is 7.10 watts per metre squared kelvin. That is the size of natural convection in air, so it cannot be ignored.
  2. A uniform body of volume V loses heat only through its surface, so rho c V theta dot equals minus h A theta. The time constant tau is rho c V over h A, which is R th times C th: the RC circuit again, with no thermal inductor. The assumption is a uniform body. Heat crosses conduction and convection resistances in series, and their ratio is the Biot number, h L over k. Take a steel plate, L equal to 10 millimetres. In still air, h is 25, Bi is 0.0056 and tau is 1529 seconds: lumped is fine. Quenched in water, h is 1000, Bi is 0.222, tau is 38.2 seconds, and the surface is cold while the centre is hot.
  3. When Bi is not small, solve the slab. Separation of variables and symmetry give a cosine, and the Robin condition becomes zeta tan zeta equals Bi, a transcendental eigenvalue problem. Zeta tan zeta rises from zero to infinity on each interval, so there is exactly one root per interval and bisection finds it. For small Bi the leading mode is exactly the lumped result. At t over tau equal to 1, with Bi equal to 0.1, the centre is 0.3861 and the surface 0.3676, and the lumped curve is 0.3679. At Bi equal to 1, the centre has only dropped to 0.5339. At 10 it is 0.9684, while the surface has nearly reached the fluid. The lumped error against the centre is minus 4.7, minus 31.1 and minus 62.0 percent.
  4. A fin adds surface. A slice of the rod has conduction in, conduction out, and convection from its sides, so k A theta double prime equals h P theta, which is theta double prime equals m squared theta. With the base fixed and an insulated tip, the solution is cosh of m times L minus x, over cosh of m L, and the heat rate is M theta b times tanh of m L. Take an aluminium fin, k equal to 200, h equal to 50, 2 millimetres thick and 30 long. Then m is 16.12 per metre and m L is 0.4837. The tip is at 74.14 degrees, the fin passes 7.968 watts, and its efficiency, tanh of m L over m L, is 92.87 percent.
  5. Now a sink with 8 of these fins on a base. Each fin passes 7.968 watts, and the unfinned base passes 6.05. Conductances add, like parallel resistors, so the total is 8 times 7.968 plus 6.05, which is 69.79 watts. The bare base alone passes 8.25 watts, so the fins multiply the heat by 8.46, and the sink resistance, theta b over q, is 0.788 kelvin per watt. Add a fan, h equal to 100, and the heat rises to 131.3 watts, the resistance falls to 0.419. This is Kirchhoff's law from lesson 14.2, with temperature as voltage and watts as current.
  6. Without time dependence, the heat equation becomes Laplace's equation: each node is the average of its four neighbours, and sweeping that rule until nothing changes is relaxation. The test problem is a hot inner square in a cold outer square. The heat per metre is the shape factor S prime, times k, times the temperature difference, and S prime depends only on the ratio b over a. For b over a equal to 2 it is about 10.23, and for 3, about 6.22. Counting sweeps with 60 cells across, Gauss-Seidel takes 715 and optimal SOR 199. At 120, 2568 against 394. At 240, 9121 against 786. Doubling the grid multiplies Gauss-Seidel by about 3.6, and SOR by 2.
  7. Fins turned a boundary condition into the whole problem: theta double prime equals m squared theta, with cosh and sinh solutions mixed by the tip condition. The next lesson is vibrating beams. The syllabus lists Euler-Bernoulli theory, the static deflection problem, clamped, pinned and free ends, mode shapes and natural frequencies, numerical beam solvers, and a glimpse of wing bending and flutter. The beam equation is fourth order, and, from memory, to be shown there, its mode shapes mix cosh and sinh with cos and sin. So this lesson's boundary-condition bookkeeping returns, with two conditions at each end.

Three ways heat moves

Conduction is diffusion through a solid, Fourier’s law q′′=−k ∂T/∂xq'' = -k\,\partial T/\partial x (W/m2^2) from 9.2 The Heat (Diffusion) Equation. Convection is heat carried off by a moving fluid; engineers lump the flow into Newton’s cooling law

q′′=h (Ts−T∞),q'' = h\,(T_s - T_\infty),

with hh in W/(m2^2 K). Typical ranges (from memory, orders of magnitude): natural convection in air 22 to 2525, forced air 2525 to 250250, forced water 100100 to 15 00015\,000. Radiation needs no medium: a surface emits εσT4\varepsilon\sigma T^4 with TT absolute, emissivity ε\varepsilon and σ=5.670×10−8\sigma = 5.670\times10^{-8} W/(m2^2 K4^4) (from memory). Factoring T4−T∞4T^4 - T_\infty^4 gives Newton cooling with

hrad=εσ (T+T∞)(T2+T∞2).h_{rad} = \varepsilon\sigma\,(T + T_\infty)(T^2 + T_\infty^2).

For 80∘80^\circC (353.15353.15 K) in 25∘25^\circC air (298.15298.15 K) with ε=0.9\varepsilon = 0.9 (anodised aluminium, from memory), hrad=7.10h_{rad} = 7.10 W/(m2^2 K): the size of natural convection, so it cannot be ignored there.

Newton’s cooling, lumped capacitance and the Biot number

3.5 Modeling with First-Order ODEs fitted a cooling constant and warned that Newton’s law needs a uniform temperature. Now we derive it and quantify the warning. A body of volume VV, density ρ\rho, specific heat cc and surface area AA, at uniform T(t)T(t), loses heat only through its surface. With θ=T−T∞\theta = T - T_\infty,

ρcV θ˙=−hA θ⟹θ=θ0 e−t/τ,τ=ρcVhA=Rth Cth.\rho c V\,\dot\theta = -hA\,\theta \quad\Longrightarrow\quad \theta = \theta_0\,e^{-t/\tau}, \qquad \tau = \frac{\rho c V}{hA} = R_{th}\,C_{th}.

This is the RC charging of 14.2 Electrical Circuits with thermal capacitance Cth=ρcVC_{th} = \rho cV (J/K) and convective resistance Rth=1/(hA)R_{th} = 1/(hA) (K/W). Heat has no inertia, so there is no thermal inductor and no ringing.

The model assumes a uniform body. Heat actually crosses two resistances in series: conduction through the solid, about L/(kA)L/(kA), and convection at the surface, 1/(hA)1/(hA). Their ratio is the Biot number

Bi=L/(kA)1/(hA)=hLk,L=VA.\mathrm{Bi} = \frac{L/(kA)}{1/(hA)} = \frac{hL}{k}, \qquad L = \frac{V}{A}.

If Bi≪1\mathrm{Bi} \ll 1 the solid conducts easily compared with the surface carrying heat away, the interior stays uniform, and the lumped model holds. A common rule from memory is Bi<0.1\mathrm{Bi} \lt 0.1; we measure below what it buys. The kk is the solid’s.

Worked example: a steel plate. Take steel (from memory: k=45k = 45 W/(m K), ρ=7800\rho = 7800 kg/m3^3, c=490c = 490 J/(kg K)), 2020 mm thick, so L=10L = 10 mm. In still air, h=25h = 25: Bi=0.0056\mathrm{Bi} = 0.0056 and τ=ρcL/h=1529\tau = \rho cL/h = 1529 s, about 2525 min. Quenched in water, h=1000h = 1000 (an order of magnitude): Bi=0.222\mathrm{Bi} = 0.222, τ=38.2\tau = 38.2 s, and the lumped model is doubtful, since the surface is cold while the centre is hot.

The slab with a convective boundary

Take a slab −L<x<L-L \lt x \lt L, initially θ=1\theta = 1, with both faces cooled by a fluid:

θt=α θxx,−k θx(L,t)=h θ(L,t),θx(0,t)=0.\theta_t = \alpha\,\theta_{xx}, \qquad -k\,\theta_x(L, t) = h\,\theta(L, t), \qquad \theta_x(0, t) = 0 .

Separation as in 9.3 Solving the Heat Equation: Separation of Variables and symmetry give X=cos⁡(ζx/L)X = \cos(\zeta x/L) with λ=(ζ/L)2\lambda = (\zeta/L)^2. The Robin condition k(ζ/L)sin⁡ζ=hcos⁡ζk(\zeta/L)\sin\zeta = h\cos\zeta is

ζtan⁡ζ=Bi.\zeta\tan\zeta = \mathrm{Bi}.

This is the Sturm-Liouville eigenvalue problem of 8.2 Eigenvalue Problems & Sturm–Liouville Theory with a Robin end, and it is transcendental. But ζtan⁡ζ\zeta\tan\zeta rises from 00 to infinity on each interval (nπ,(n+12)π)(n\pi, (n + \tfrac12)\pi), so there is exactly one root per interval; bisection on g(ζ)=ζsin⁡ζ−Bicos⁡ζg(\zeta) = \zeta\sin\zeta - \mathrm{Bi}\cos\zeta finds it with no pole to cross. Time dependence is e−ζ2 Foe^{-\zeta^2\,\mathrm{Fo}} with the Fourier number Fo=αt/L2\mathrm{Fo} = \alpha t/L^2, and orthogonality of the cosines gives

θ(x,t)=∑n=1∞Cn e−ζn2 Focos⁡ζnxL,Cn=4sin⁡ζn2ζn+sin⁡2ζn.\theta(x, t) = \sum_{n=1}^\infty C_n\,e^{-\zeta_n^2\,\mathrm{Fo}}\cos\frac{\zeta_n x}{L}, \qquad C_n = \frac{4\sin\zeta_n}{2\zeta_n + \sin 2\zeta_n}.

Two limits check it. As Bi→∞\mathrm{Bi} \to \infty, ζn→(n−12)π\zeta_n \to (n - \tfrac12)\pi, the Dirichlet eigenvalues. As Bi→0\mathrm{Bi} \to 0, tan⁡ζ≈ζ\tan\zeta \approx \zeta gives ζ12≈Bi\zeta_1^2 \approx \mathrm{Bi}, and the leading mode decays as e−Bi Fo=e−t/τe^{-\mathrm{Bi}\,\mathrm{Fo}} = e^{-t/\tau}: exactly the lumped result. The lab’s roots, checked with scipy.optimize.brentq:

Bi\mathrm{Bi}ζ1\zeta_1ζ2\zeta_2ζ3\zeta_3C1C_1C2C_2C3C_3
0.10.10.31110.31113.17313.17316.29916.29911.01611.0161−0.0197-0.01970.00500.0050
110.86030.86033.42563.42566.43736.43731.11911.1191−0.1517-0.15170.04660.0466
10101.42891.42894.30584.30587.22817.22811.26201.2620−0.3934-0.39340.21040.2104

The series converges slowly at small times (Fo=0.001\mathrm{Fo} = 0.001 needs about 6060 terms; the lab uses 300300), while implicit finite differences (10.4 Implicit Methods & Crank–Nicolson) handle the Robin end as a ghost-point row: use the series to validate such a code.

Lumped against slab: what Bi buys

The lab plots four curves against t/τt/\tau, so the lumped curve e−t/τe^{-t/\tau} is the same for every Bi\mathrm{Bi} (and Fo=(t/τ)/Bi\mathrm{Fo} = (t/\tau)/\mathrm{Bi}). At t/τ=1t/\tau = 1 the lumped value is e−1=0.3679e^{-1} = 0.3679; the slab values, matching an independent 400400-term script, are:

Bi\mathrm{Bi}centresurfaceaveragelumpedlumped against centre
0.10.10.38610.38610.36760.36760.37990.37990.36790.3679−4.7%-4.7\%
110.53390.53390.34820.34820.47040.47040.36790.3679−31.1%-31.1\%
10100.96840.96840.17060.17060.72610.72610.36790.3679−62.0%-62.0\%

At Bi=0.1\mathrm{Bi} = 0.1 the centre is 5%5\% above the surface and lumped is within 3.2%3.2\% of the volume average, which is what the lumped model really computes. At Bi=1\mathrm{Bi} = 1 the centre has only dropped to 53%53\% while the surface is at 35%35\%. At Bi=10\mathrm{Bi} = 10 the surface has nearly reached T∞T_\infty while the centre has barely noticed. The error also grows with time, because the rate is wrong: the late-time slope of the average is ζ12/Bi\zeta_1^2/\mathrm{Bi} of the lumped slope, 0.96750.9675, 0.74020.7402 and 0.20420.2042. At Bi=10\mathrm{Bi} = 10 the lumped time constant is about 4.94.9 times too short.

At late times centre over surface is 1/cos⁡ζ11/\cos\zeta_1: 1.05041.0504, 1.53331.5333, 7.06967.0696. Since cos⁡ζ1≈1−ζ12/2\cos\zeta_1 \approx 1 - \zeta_1^2/2, this is about 1+Bi/21 + \mathrm{Bi}/2, so Bi<0.1\mathrm{Bi} \lt 0.1 means a spread near 5%5\%. It is a tolerance, not a law.

Cooling fins: a boundary-value problem

A fin adds surface for convection. A rod of cross-section AA, perimeter PP and conductivity kk conducts along xx and loses heat through its sides. A slice of width dxdx has conduction in, conduction out and convection hP dx θhP\,dx\,\theta:

kA θ′′=hP θ⟹θ′′=m2θ,m2=hPkA,kA\,\theta'' = hP\,\theta \quad\Longrightarrow\quad \theta'' = m^2\theta, \qquad m^2 = \frac{hP}{kA},

with 1/m1/m a decay length. Using one coordinate is the Biot idea again: it needs the transverse Biot number h(t/2)/kh(t/2)/k small, which is 2.5×10−42.5\times10^{-4} for the fin below. With θ(0)=θb\theta(0) = \theta_b and an insulated tip θ′(L)=0\theta'(L) = 0,

θ(x)θb=cosh⁡m(L−x)cosh⁡mL,q=−kAθ′(0)=M θbtanh⁡mL,M=hPkA.\frac{\theta(x)}{\theta_b} = \frac{\cosh m(L - x)}{\cosh mL}, \qquad q = -kA\theta'(0) = M\,\theta_b\tanh mL, \qquad M = \sqrt{hPkA}.

A long fin has tanh⁡→1\tanh \to 1 and q→Mθbq \to M\theta_b, the infinite-fin rate. For a convective tip, −kθ′(L)=hθ(L)-k\theta'(L) = h\theta(L) with B=h/(mk)B = h/(mk) gives q=Mθb(sinh⁡mL+Bcosh⁡mL)/(cosh⁡mL+Bsinh⁡mL)q = M\theta_b(\sinh mL + B\cosh mL)/(\cosh mL + B\sinh mL). The efficiency is actual heat over the heat if the whole fin sat at θb\theta_b; for an insulated tip

η=qhPL θb=tanh⁡mLmL.\eta = \frac{q}{hPL\,\theta_b} = \frac{\tanh mL}{mL}.

Worked example. An aluminium fin (from memory, k=200k = 200 W/(m K)), t=2t = 2 mm thick, L=30L = 30 mm long, w=50w = 50 mm deep, in air with h=50h = 50 W/(m2^2 K); base 80∘80^\circC, air 25∘25^\circC, so θb=55\theta_b = 55 K. P=2(w+t)=104P = 2(w + t) = 104 mm, A=100A = 100 mm2^2, so m=16.12m = 16.12 m−1^{-1} (the thin-fin 2h/kt=15.81\sqrt{2h/kt} = 15.81 is 2%2\% low) and mL=0.4837mL = 0.4837. The tip is at 0.8934 θb0.8934\,\theta_b, that is 74.14∘74.14^\circC; q=7.968q = 7.968 W; η=92.9%\eta = 92.9\%. The convective tip gives 8.1868.186 W, only 2.7%2.7\% more, because B=0.0155B = 0.0155. Both match a solve_bvp run.

How long should it be? The infinite-fin limit is Mθb=17.74M\theta_b = 17.74 W:

LL (mm)mLmLqq (W)efficiency
10100.1610.1612.8352.83599.1%99.1\%
30300.4840.4847.9687.96892.9%92.9\%
60600.9670.96713.2613.2677.3%77.3\%
1001001.6121.61216.3816.3857.3%57.3\%
1641642.652.6517.5617.5637.3%37.3\%

Going from 3030 to 100100 mm multiplies the metal by 3.33.3 but the heat by only 2.062.06. The last row has tanh⁡mL=0.99\tanh mL = 0.99; beyond it extra metal is wasted.

The fin BVP also discretises: θi−1−(2+m2Δx2)θi+θi+1=0\theta_{i-1} - (2 + m^2\Delta x^2)\theta_i + \theta_{i+1} = 0 with a ghost point at the tip and a tridiagonal solve, as in 8.1 Two-Point Boundary Value Problems. The exercise does this against the closed form.

The sink as a network of resistances

A sink has NN fins on a base. Each passes qfinq_{fin} and the unfinned base passes h(W−Nt) w θbh(W - Nt)\,w\,\theta_b. Conductances (W/K) add like parallel resistors, so Rsink=θb/qtotalR_{sink} = \theta_b/q_{total}. For N=8N = 8 of the fins above on a 6060 mm base (4444 mm unfinned, 5050 mm deep, insulated tips):

qtotal=8(7.968)+50 (0.044)(0.050)(55)=63.74+6.05=69.79 W.q_{total} = 8(7.968) + 50\,(0.044)(0.050)(55) = 63.74 + 6.05 = 69.79\ \text{W}.

The bare base passes 8.258.25 W, so the fins give ×8.46\times8.46, and Rsink=55/69.79=0.788R_{sink} = 55/69.79 = 0.788 K/W. With a fan, h=100h = 100, the total is 131.3131.3 W and Rsink=0.419R_{sink} = 0.419 K/W (fin efficiency 86.9%86.9\%): doubling hh cut the resistance by 1.881.88, not 22.

Series resistances add too: for a 4040 W chip with (hypothetical) 0.30.3 K/W junction-to-case and 0.20.2 K/W interface, Tj=25+40(0.3+0.2+0.788)=76.5∘T_j = 25 + 40(0.3 + 0.2 + 0.788) = 76.5^\circC, or 61.8∘61.8^\circC with the fan. This is Kirchhoff’s law of lesson 14.2 with temperature as voltage and watts as current.

2D steady conduction

With θt=0\theta_t = 0 the heat equation becomes Laplace’s equation, θxx+θyy=0\theta_{xx} + \theta_{yy} = 0, with the neighbour-average property of 9.6 Laplace's Equation: Rectangles & Disks. On a grid each node is the mean of its four neighbours, and sweeping that rule until nothing changes is relaxation, the solvers of 10.8 Elliptic Solvers: Jacobi, Gauss–Seidel & SOR.

The test problem is a square flue: a hot inner square (θ=1\theta = 1, side aa) in a cold outer square (00, side bb). The heat per metre is Q′=S′k ΔTQ' = S'k\,\Delta T, where the shape factor S′S' depends only on b/ab/a, because Laplace’s equation has no length scale. By Green’s identity S′=∬∣∇θ∣2 dAS' = \iint|\nabla\theta|^2\,dA, so the lab sums squared differences over all grid edges. For N=60,120,240N = 60, 120, 240 cells across the outer square, converged to a largest change of 10−810^{-8} and checked against a sparse direct solve:

b/ab/aN=60N = 60N=120N = 120N=240N = 240extrapolated
2210.271510.271510.248810.248810.239910.2399about 10.2310.23
336.24156.24156.22586.22586.21966.2196about 6.226.22

The differences shrink by 2.552.55 per halving (order 1.351.35, not 22: the inner corners are singular). A fit I remember from heat-transfer texts, S′=2π/(0.93ln⁡(b/a)−0.0502)S' = 2\pi/(0.93\ln(b/a) - 0.0502) for b/a>1.4b/a \gt 1.4, gives 10.5710.57 and 6.476.47, about 3%3\% and 4%4\% high; I could not check the coefficients against a source. For an assumed k=1k = 1 W/(m K) and ΔT=30\Delta T = 30 K, b/a=2b/a = 2 gives Q′=307Q' = 307 W per metre.

Sweeps to converge for b/a=2b/a = 2 (with b/a=3b/a = 3 in brackets), using ω=2/(1+sin⁡(π/N))\omega = 2/(1 + \sin(\pi/N)), optimal for the plain square:

NNGauss-SeidelSOR
6060715715 (12381238)199199 (200200)
12012025682568 (44214421)394394 (388388)
24024091219121 (1558515585)786786 (801801)

Doubling NN multiplies Gauss-Seidel by about 3.63.6 (toward the 1/h21/h^2 limit of 44) and SOR by 22, the 1/h1/h law. Multigrid (10.9 Multigrid) removes the growth.

Heat exchangers: a glimpse

In a heat exchanger two streams swap heat across a wall. With capacity rates ChC_h, CcC_c (W/K) and overall coefficient UU, parallel flow along area AA obeys

Ch dThdA=−U(Th−Tc),Cc dTcdA=+U(Th−Tc).C_h\,\frac{dT_h}{dA} = -U(T_h - T_c), \qquad C_c\,\frac{dT_c}{dA} = +U(T_h - T_c).

This linear system (7.2 Linear Systems & the Eigenvalue Method) has eigenvalue 00 (energy conservation) and −U(1/Ch+1/Cc)-U(1/C_h + 1/C_c), so the difference decays as ΔTout=ΔTin e−NTU(1+Cr)\Delta T_{out} = \Delta T_{in}\,e^{-\mathrm{NTU}(1 + C_r)}, with NTU=UA/Cmin\mathrm{NTU} = UA/C_{min}, Cr=Cmin/CmaxC_r = C_{min}/C_{max}. The effectiveness is

ε=QCmin ΔTin=1−e−NTU(1+Cr)1+Cr,\varepsilon = \frac{Q}{C_{min}\,\Delta T_{in}} = \frac{1 - e^{-\mathrm{NTU}(1 + C_r)}}{1 + C_r},

and counterflow has (from memory) ε=(1−e−NTU(1−Cr))/(1−Cre−NTU(1−Cr))\varepsilon = (1 - e^{-\mathrm{NTU}(1 - C_r)})/(1 - C_re^{-\mathrm{NTU}(1 - C_r)}). With UA=500UA = 500 W/K, Ch=1000C_h = 1000, Cc=2000C_c = 2000 W/K (NTU=0.5\mathrm{NTU} = 0.5, Cr=0.5C_r = 0.5), inlets 100∘100^\circC and 20∘20^\circC: parallel ε=0.3518\varepsilon = 0.3518 (hot outlet 71.86∘71.86^\circC), counterflow 0.36230.3623 (71.02∘71.02^\circC), both matching a numerical ODE solve.

The lab

◆ Interactive Heat sink fin designer, transient slab against lumped capacitance, and 2D conduction by relaxation

Loading simulation…

The view menu switches between three experiments.

  1. Fin view, default (k=200k = 200, h=50h = 50, t=2t = 2 mm, L=30L = 30 mm, N=8N = 8): m=16.12m = 16.12, mL=0.4837mL = 0.4837, tip 74.14∘74.14^\circC, q=7.968q = 7.968 W, efficiency 92.87%92.87\%; sink 69.7969.79 W against 8.258.25 W bare, 0.7880.788 K/W.
  2. Convective tip: q=8.186q = 8.186 W, tip 73.80∘73.80^\circC, sink 71.5471.54 W.
  3. Press the aluminium-and-fan button (h=100h = 100): 131.33131.33 W, 0.4190.419 K/W. Reset h=50h = 50 and set L=100L = 100 mm: one fin 16.3816.38 W at 57.27%57.27\%, the sink 137.09137.09 W (0.4010.401 K/W), with 8080 cm3^3 of metal against the fan’s 2424.
  4. Slab view, Bi=1\mathrm{Bi} = 1, cursor at t/τ=1t/\tau = 1: centre 0.53390.5339, surface 0.34820.3482, average 0.47040.4704, lumped 0.36790.3679. At Bi=0.1\mathrm{Bi} = 0.1 the curves nearly merge; at 1010 the lumped line falls away.
  5. 2D view, outer/inner 22, grid 6060: Gauss-Seidel 715715 sweeps, SOR 199199, S′=10.2715S' = 10.2715. At 240240, Gauss-Seidel takes 91219121 sweeps and S′=10.2399S' = 10.2399.

The exercise: fin and Biot eigenvalue

The playgrounds reproduce the fin-length table and the slab-against-lumped comparison. The exercise builds three pieces: the closed form for an insulated tip, a finite-difference solve of the same BVP (the tridiagonal system with a ghost point), and the first Biot eigenvalue by bisection.

🐍 PythonFin length against heat rate
⌘/Ctrl + Enter
🐍 PythonSlab series against the lumped model
⌘/Ctrl + Enter
🧪 Coding exerciseExercise: fin tip, heat rate and Biot eigenvalue
⌘/Ctrl + Enter
🐍 PythonRelaxation on the square flue
⌘/Ctrl + Enter

Concept checks

✅ Concept check

A $20$ mm steel plate (half-thickness $L = 10$ mm, $k = 45$ W/(m K)) is quenched in water with $h = 1000$ W/(m$^2$ K). What is the Biot number?

✅ Concept checkselect all that apply

Select all statements that are true about the lumped-capacitance model $\tau\dot\theta = -\theta$.

✅ Concept check

At $\mathrm{Bi} = 10$, $\zeta_1 = 1.4289$. By what factor is the lumped time constant too short compared with the late-time decay of the slab, $\mathrm{Bi}/\zeta_1^2$?

✅ Concept check

A fin has $mL = 0.5$ and an insulated tip. You double $L$. Which statement describes what happens?

✅ Concept check

You refine the grid in a square-flue computation by a factor of $2$ in each direction. Roughly how does the sweep count change for optimal SOR and for Gauss-Seidel?

Problems

✏️ Pen & paperCooling an aluminium plate●○○

An aluminium plate (from memory: k=200k = 200 W/(m K), ρ=2700\rho = 2700 kg/m3^3, c=900c = 900 J/(kg K)), 66 mm thick, so L=3L = 3 mm, at 120∘120^\circC in 25∘25^\circC air with h=40h = 40 W/(m2^2 K). Check that the lumped model applies, then find the time to reach 40∘40^\circC.

Show worked solution

Bi=40×0.003/200=6×10−4\mathrm{Bi} = 40\times0.003/200 = 6\times10^{-4}, far below 0.10.1. τ=ρcL/h=2700×900×0.003/40=182.25\tau = \rho cL/h = 2700\times900\times0.003/40 = 182.25 s. The excess falls from 9595 K to 1515 K, so t=τln⁡(95/15)=336t = \tau\ln(95/15) = 336 s, about 5.65.6 min.

✏️ Pen & paperA pin fin●●○

A circular pin fin, D=5D = 5 mm, L=40L = 40 mm, k=200k = 200 W/(m K), h=40h = 40 W/(m2^2 K), insulated tip, base 6060 K above the air. Find mm, the tip excess temperature, the heat rate and the efficiency.

Show worked solution

P=πDP = \pi D, A=πD2/4A = \pi D^2/4, so m2=4h/(kD)=160m^2 = 4h/(kD) = 160, m=12.649m = 12.649 m−1^{-1}, mL=0.5060mL = 0.5060. Tip excess 60/cosh⁡(0.5060)=53.0660/\cosh(0.5060) = 53.06 K. M=hPkA=0.049673M = \sqrt{hPkA} = 0.049673 W/K, so q=0.049673×60×tanh⁡(0.5060)=1.391q = 0.049673\times60\times\tanh(0.5060) = 1.391 W. Efficiency tanh⁡(mL)/(mL)=0.9226\tanh(mL)/(mL) = 0.9226.

✏️ Pen & paperThe small-Biot expansion●●●

(a) Use tan⁡ζ≈ζ+ζ3/3\tan\zeta \approx \zeta + \zeta^3/3 in ζtan⁡ζ=Bi\zeta\tan\zeta = \mathrm{Bi} to show ζ12≈Bi/(1+Bi/3)\zeta_1^2 \approx \mathrm{Bi}/(1 + \mathrm{Bi}/3) and compare with ζ1=0.311053\zeta_1 = 0.311053 at Bi=0.1\mathrm{Bi} = 0.1. (b) Show the leading-mode centre-to-surface ratio is about 1+Bi/21 + \mathrm{Bi}/2.

Show worked solution

(a) ζ2(1+ζ2/3)≈Bi\zeta^2(1 + \zeta^2/3) \approx \mathrm{Bi}; replacing ζ2\zeta^2 in the bracket by Bi\mathrm{Bi} gives Bi/(1+Bi/3)=0.096774\mathrm{Bi}/(1 + \mathrm{Bi}/3) = 0.096774 at 0.10.1, against the exact ζ12=0.096754\zeta_1^2 = 0.096754. (b) The ratio is 1/cos⁡ζ1≈1+ζ12/2≈1+Bi/21/\cos\zeta_1 \approx 1 + \zeta_1^2/2 \approx 1 + \mathrm{Bi}/2; exactly 1.05041.0504 at Bi=0.1\mathrm{Bi} = 0.1.

✏️ Pen & paperA wall with two layers●●○

Per square metre: 0.20.2 m brick (k=0.7k = 0.7 W/(m K)), 0.050.05 m foam (k=0.03k = 0.03), indoor h=8h = 8, outdoor h=25h = 25 W/(m2^2 K), air at 20∘20^\circC inside and −5∘-5^\circC outside (all assumed). Find the heat flux and the brick-foam interface temperature.

Show worked solution

Resistances (m2^2K/W): 0.125+0.2857+1.6667+0.04=2.11740.125 + 0.2857 + 1.6667 + 0.04 = 2.1174. Flux 25/2.1174=11.8125/2.1174 = 11.81 W/m2^2. Interface: 20−11.81×(0.125+0.2857)=15.15∘20 - 11.81\times(0.125 + 0.2857) = 15.15^\circC. The foam takes 19.719.7 K of the 2525 K drop.

What comes next

Fins turned a boundary condition into the whole problem: θ′′=m2θ\theta'' = m^2\theta, with cosh⁡\cosh and sinh⁡\sinh solutions mixed by the tip condition. The next lesson, 14.5 Vibrating Beams & Structures, is vibrating beams. The syllabus lists Euler-Bernoulli theory, the static deflection BVP, clamped, pinned and free ends, mode shapes and natural frequencies, numerical beam solvers, and a glimpse of wing bending and flutter. The beam equation is fourth order, and (from memory, to be shown there) its mode shapes mix cosh⁡\cosh and sinh⁡\sinh with cos⁡\cos and sin⁡\sin, so this lesson’s boundary-condition bookkeeping returns with two conditions at each end.

Key Takeaways

  • Heat moves by conduction, convection (hh) and radiation (hradh_{rad}); the engineering sits in the boundary condition.
  • Lumped capacitance is τθ˙=−θ\tau\dot\theta = -\theta, τ=ρcV/(hA)=RthCth\tau = \rho cV/(hA) = R_{th}C_{th}: an RC circuit with no inductor.
  • Bi=hL/k\mathrm{Bi} = hL/k compares internal and surface resistance: at 0.10.1 the centre-surface spread is about 5%5\%, at 1010 the lumped time constant is 4.94.9 times too short.
  • The convective slab has eigenvalues from ζtan⁡ζ=Bi\zeta\tan\zeta = \mathrm{Bi}, one per interval, found by bisection; ζ12→Bi\zeta_1^2 \to \mathrm{Bi} recovers the lumped model.
  • The fin BVP gives cosh⁡m(L−x)/cosh⁡mL\cosh m(L - x)/\cosh mL, q=Mθbtanh⁡mLq = M\theta_b\tanh mL and η=tanh⁡(mL)/(mL)\eta = \tanh(mL)/(mL): diminishing returns in length.
  • Thermal resistances add like electrical ones, in series and in parallel.
  • 2D conduction is Laplace relaxation: Gauss-Seidel sweeps grow like N2N^2, SOR like NN.
Finished this lesson?

Mark it complete to track your progress (saved in this browser). Reading every section does this for you.