Validation Case: Bimetallic Strip Under Thermal Load
This article is part of the series of FEA validation cases performed using OnScale Solve.
Problem Statement
Bimetallic strips are subject to bending due to heat and can be used to convert change in temperature to mechanical displacement finding application in thermostats for example. The associated physics coupling thermal and mechanical domains are amenable to analytical solution making them ideal for validation of the coupled thermal and mechanical solvers.
The resulting comparison of the simulation results against the analytical calculations validates the use of the following conditions for thermomechanical static analysis in Solve:
- Thermomechanical Solver
- Temperature Load
- Restraints
Geometry
Download the geometry here used for this analysis.

The goemetry is composed of two metal strips of equal size bonded to each other across their major surfaces. Each strip being 20~\text{mm} long, 0.1~\text{mm} thick and 0.3~\text{mm} wide.
Part | Material | Length | Width | Thickness |
---|---|---|---|---|
Part 1 | Iron | 20~\text{mm} | 0.3~\text{mm} | 0.1~\text{mm} |
Part 2 | Cast Iron | 20~\text{mm} | 0.3~\text{mm} | 0.1~\text{mm} |
Materials:
The materials assigned to the parts are taken directly from OnScale’s library of materials.
Part | Material | Young’s Modulus | Thermal Expansion | Thermal Conductivity |
---|---|---|---|---|
Part 1 | Cast Iron | 200~\text{GPa} | 1\times 10^{-5}~\text{K}^{-1} | 60~\text{Wm}^{-1}\text{K}^{-1} |
Part 2 | Iron | 200~\text{GPa} | 2\times 10^{-5}~\text{K}^{-1} | 60~\text{Wm}^{-1}\text{K}^{-1} |
The poisson’s ratio for both materials was set to zero so that the strain response was only in the direction of the applied stress, giving a more appropriate comparison to the 2D case of the analytical solution.
Boundary Conditions
A full displacement restraint was applied to the surface at one of the lengthwise ends of the bimetallic strip while a second restraint constraining motion in the direction perpendicular to the lengthwise axis and parallel to the plane interface between the two materials was applied to the opposite end. Since this was a coupled thermomechanical study, additionally, a temperature constraint of 100~^\circ\text{C} was applied to the exposed major face of part 2. The remaining surfaces defaulted to zero heat flux adiabatic condition with no explicitly applied loads or constraints.
Meshing
OnScale automatically generates 3D second order tetrahedral meshes at 3 tiered densitites. The meshing statistics are:
Mesh Tier | Number of Vertices | Number of Cells | Degrees of Freedom |
---|---|---|---|
Fine | 11,074 | 45,250 | 295,204 |
Medium | 4,392 | 14,066 | 105,900 |
Coarse | 3,250 | 9,754 | 76,340 |
Analytical Solution
The equations describing the deflection of the bimetallic strip presented here are found in [1]:
d_x = l\cdot\left(T-T_\text{0}\right)\frac{\gamma_a -\gamma_b}{2} d_z = \frac{l^2}{2}\cdot\frac{6\left(\gamma_b -\gamma_a\right)\cdot\left(T-T_\text{0}\right)\cdot\left(t_a+t_b\right)}{t_b^2\cdot K_1} \sigma_{xx} = \frac{\left(\gamma_b -\gamma_a\right)\cdot\left(T-T_\text{0}\right)\cdot E_b}{K_1}\left(3\cdot\frac{t_a}{t_b}+ 2 - \frac{E_a}{E_b}\left(\frac{t_a}{t_b}\right)^3\right)
where l is the length of the bimetallic strip, t_a and t_b are the thicknesses of the top and bottom strips respectively, E_a and E_b are the Young’s moduli of the top and bottom strips, \gamma_a and \gamma_b are the coefficients of thermal expansion respectively. The temperature of the strips is T and T_\text{0} is the temperature at which the strip is not bent, which for this simulation is 20~^{\circ}\text{C}, \sigma_{xx} is the stress component at the base of the strip and K_1 is evaluated from:
K_1 = 4 + 6\cdot\frac{t_a}{t_b} + 4\left(\frac{t_a}{t_b}\right)^2 + \frac{E_a}{E_b}\left(\frac{t_a}{t_b}\right)^3 + \frac{E_b}{E_a}\cdot\frac{t_b}{t_a}
The respective symbols for the parameters for the analytical solution are in tbl. 1.
Part | Thickness | Young’s Modulus | Thermal expansion |
---|---|---|---|
Part 1 | t_a | E_a | \gamma_a |
Part 2 | t_b | E_b | \gamma_b |
Results
The values for d_x and d_z are taken from the middle of the free end of the strip, at the contact point between the top and bottom plate.

Output | Analytical | FEA (Fine) | FEA (Medium) | FEA (Coarse) |
---|---|---|---|---|
d_x~[\mu\text{m}] | 24.000 | 23.977 | 23.982 | 23.982 |
d_z~[\text{mm}] | 1.200 | 1.200 | 1.200 | 1.200 |
\sigma_{xx}~[\text{M Pa}] | 40 | 39.989 | 40.010 | 40.060 |
Conclusions
Results show excellent correspondence between the solver solution and the analytical solution, in particular the difference between the solver solutions for d_z and the analytical sotion falls beyond the precision available. Additionally, the automatically determined mesh densities can all capture the physical system interactions in sufficient detail for almost exact agreement.