Heat Conduction in Cylindrical Coordinates
Heat Conduction in Cylindrical Coordinates: Theoretical Foundations
Heat Conduction Equation in Cylindrical Coordinates
Professor, doesn't thermal analysis for pipes and shafts work well with Cartesian coordinates?
For cylindrical shapes, Cartesian coordinates lead to inefficient shape representation and complex meshing. Using cylindrical coordinates $(r, \phi, z)$ allows leveraging symmetry to reduce dimensions. The one-dimensional radial steady-state heat conduction equation is
Compared to Fourier's law in Cartesian coordinates, there's an extra $1/r$ term.
This term reflects the change in area because the volume element is proportional to $r$. This is also the physical cause of the critical insulation radius.
Analytical Solution (No Internal Heat Generation)
The solution for the case $\dot{q}_v = 0$ is as follows.
The temperature distribution is logarithmic, different from the linear distribution in a flat plate. The heat flow rate is
The $\ln$ appears because the area changes, right?
Correct. Since the cross-sectional area $A = 2\pi r L$ is proportional to $r$, for a constant heat flow rate $q$, the heat flux $q'' = q/A$ is inversely proportional to $r$. The heat flux is larger on the inner surface.
Case with Internal Heat Generation
The solution for the case with uniform internal heat generation $\dot{q}_v$ is
The maximum temperature at the center $r=0$ is $T_{\max} = T_s + \dot{q}_v R^2/(4k)$.
This is the case for Joule heating in electrical wires, right?
Exactly. For an AWG18 copper wire (diameter 1.02mm) carrying 10A, the heat generation is about 10 W/m. $\dot{q}_v \approx 1.2 \times 10^7$ W/m$^3$, and the center temperature rise can be estimated from this formula.
Origin of the Cylindrical Coordinate Laplacian
The heat conduction equation in cylindrical coordinates is expressed as โยฒT/โrยฒ + (1/r)โT/โr + (1/rยฒ)โยฒT/โฮธยฒ + โยฒT/โzยฒ + qฬ/ฮป = 0. The (1/r)โT/โr term in this Laplacian originates from the Jacobian of the coordinate transformation. Lame systematized the cylindrical coordinate Laplacian in elasticity theory in 1833, and it was applied to the Fourier heat equation about 20 years later, which is the historical background.
Computational Methods for Heat Conduction in Cylindrical Coordinates
Axisymmetric FEM Model
When solving cylindrical heat conduction with FEM, is a 3D model necessary?
If symmetric in the circumferential direction, a 2D axisymmetric model is sufficient. Computational cost becomes less than 1/100.
| Model Type | Degrees of Freedom | Computation Time | Accuracy |
|---|---|---|---|
| 3D Full | ~1 million | Minute order | Baseline |
| 3D 1/4 Symmetry | ~250k | Second order | Equivalent |
| 2D Axisymmetric | ~1000 | Instant | Equivalent |
Running a 3D model when 2D axisymmetric is sufficient is wasteful, isn't it?
Leveraging symmetry is fundamental for computational efficiency. In Ansys Mechanical, use PLANE55 (KEYOPT(3)=1); in Abaqus, use DCAX4 (axisymmetric 4-node element).
Discretization with Finite Difference Method
Central differencing in cylindrical coordinates requires caution.
At $r = 0$ (central axis), $1/r$ becomes a singularity, so apply L'Hรดpital's rule to get
.
The singularity at the center needs special treatment, huh.
Similarly, handling the central axis is important in FEM as well. Axisymmetric elements automatically perform correct processing at nodes where $r = 0$, but caution is needed when using wedge elements on the axis in a 3D model.
Solution for Multilayer Cylinders
For multilayer cylinders (pipe + insulation + cladding, etc.), connect the analytical solutions for each layer.
To include contact thermal resistance between layers, add $1/(2\pi r_i h_{c,i} L)$ to the denominator.
It's the same addition as series resistances in an electrical circuit, right?
Exactly, it's the concept of a thermal resistance network. The strength of cylindrical coordinates is that even for 5 or more layers, hand calculations are possible.
Solution Steps for Logarithmic Temperature Distribution
The steady-state temperature distribution of an infinitely long hollow cylinder (inner radius ri, outer radius ro) is T(r)=Ti + (TiโTo)ยทln(r/ro)/ln(ri/ro), which is logarithmic. This solution is completed in three steps: separation of variables โ ODE integration โ determining C with two boundary conditions. Similar problems appear in Japanese university entrance exams, and domestic heat engineering textbooks always treat this as a chapter-opening example problem.
Heat Conduction in Cylindrical Coordinates in Practice
Pipe Heat Loss Calculation
Please tell me typical practical situations where cylindrical heat conduction is used.
The most common is heat loss calculation for plant piping. Here's an example for a steam pipe (outer diameter 114.3mm, SUS304, insulation 50mm).
| Layer | $r$ [mm] | $k$ [W/(m K)] | Thermal Resistance [m K/W] |
|---|---|---|---|
| Pipe Wall | 48.6โ57.15 | 16.3 | 0.00160 |
| Insulation | 57.15โ107.15 | 0.05 | 2.016 |
| Internal Convection | โ | โ | 0.00033 |
| External Convection | โ | โ | 0.0149 |
Total thermal resistance $R_{\text{total}} = 2.033$ m K/W. The insulation accounts for 99% of the total.
The thermal resistance of the pipe wall is almost negligible, huh.
Yes. The thermal resistance of metal pipes is less than 1/1000th of the insulation, so there's virtually no temperature drop. Therefore, whether the pipe material is copper or SUS has almost no effect on heat loss.
Log Mean Radius
The thermal resistance per unit length for a cylinder $R = \ln(r_2/r_1)/(2\pi k)$ can be expressed as an equivalent flat plate's $R = t/(k \cdot A_{lm})$. $A_{lm}$ is the log mean area:
So you can reuse the flat plate formula by using the logarithmic mean, right?
If $r_2/r_1 < 2$, the arithmetic mean $(r_1+r_2)/2$ gives an error below 4%. For simplified calculations, the arithmetic mean is often sufficient.
Mesh Generation Notes
Here are points to note for axisymmetric mesh generation for cylinders.
- Use biased mesh in the radial direction (finer on the inner side)
- Minimum of 3 elements in the radial direction for thin-walled pipes
- For multilayer structures, share nodes between layers or use Tied Contact
Biased mesh is because the heat flux is larger on the inner surface, right?
Exactly. The heat flux is inversely proportional to the radius, so it's larger on the inner surface. Using a finer mesh there improves accuracy for heat flux evaluation.
Related Topics
Experience the theory firsthand with the interactive simulator for this field
All Simulators