Verification

Every change to Elmer Studio runs a test suite that builds models through the same code the GUI uses, solves them with real Elmer, and compares the results with closed-form solutions or published benchmarks. The tests run on Linux (with the official Elmer PPA), Windows and macOS on every push (the solver tests need Elmer, so they run on Linux).

Closed-form solutions

Case	Result
3D conduction between fixed end temperatures (linear profile)	max error 1.7e-11 K
Slab with a uniform heat source, T_max = QL²/8k	relative error 9e-16
Copper bar of the Quick start, T(L) = T₀ + q₀L/k	45 °C exactly
Cantilever tip deflection vs Euler–Bernoulli	0.3 % (3D and shear effects)
Cantilever first eigenfrequency vs 1.875²/(2π)·√(EI/ρAL⁴)	40.80 vs 40.77 Hz
2D parallel-plate capacitor potential	max error 1.6e-13 V
Poiseuille channel, centreline velocity vs 1.5·Q/H	0.13 %; mass balance exact
1D diffusion, linear profile	4e-14
Acoustic duct, p = p₀ sin k(L−x)/sin kL	3e-8 Pa
Rotating annulus vs Lamé’s plane-stress solution	hoop stress within 0.7 %, rim displacement within 0.01 %
Pulsating sphere: in-mesh pressure and exterior field	0.2 % in the mesh; 1.5 % (axisymmetric) and 2 % (3D) five radii beyond it
Spherical scatterer vs the analytic series (ka = 2.3 and 6.4)	within 5 %; SPL at 3 m within 0.8 dB

Case

Result

3D conduction between fixed end temperatures (linear profile)

max error 1.7e-11 K

Slab with a uniform heat source, T_max = QL²/8k

relative error 9e-16

Copper bar of the Quick start, T(L) = T₀ + q₀L/k

45 °C exactly

Cantilever tip deflection vs Euler–Bernoulli

0.3 % (3D and shear effects)

Cantilever first eigenfrequency vs 1.875²/(2π)·√(EI/ρAL⁴)

40.80 vs 40.77 Hz

2D parallel-plate capacitor potential

max error 1.6e-13 V

Poiseuille channel, centreline velocity vs 1.5·Q/H

0.13 %; mass balance exact

1D diffusion, linear profile

4e-14

Acoustic duct, p = p₀ sin k(L−x)/sin kL

3e-8 Pa

Rotating annulus vs Lamé’s plane-stress solution

hoop stress within 0.7 %, rim displacement within 0.01 %

Pulsating sphere: in-mesh pressure and exterior field

0.2 % in the mesh; 1.5 % (axisymmetric) and 2 % (3D) five radii beyond it

Spherical scatterer vs the analytic series (ka = 2.3 and 6.4)

within 5 %; SPL at 3 m within 0.8 dB

Electrochemistry

Case	Result
1D cell, every kinetics type, vs i = i_loc(V − iL/σ)	exact to 1e-9
Tabulated i(η) through an Interpolation function	1e-6
Polarization sweep (V–I curve)	exact at every voltage
Newman’s disc electrode, R = 1/(4σa)	about 0.06 %
Sinusoidal anode, current modulation a·k·coth(kL)	0.02–0.06 %
Tertiary: Nernst diffusion layer, i = σV/(L + Λ + βδ)	1e-8
Face currents of a 3D electrode add up to the total current	exact

Case

Result

1D cell, every kinetics type, vs i = i_loc(V − iL/σ)

exact to 1e-9

Tabulated i(η) through an Interpolation function

1e-6

Polarization sweep (V–I curve)

exact at every voltage

Newman’s disc electrode, R = 1/(4σa)

about 0.06 %

Sinusoidal anode, current modulation a·k·coth(kL)

0.02–0.06 %

Tertiary: Nernst diffusion layer, i = σV/(L + Λ + βδ)

1e-8

Face currents of a 3D electrode add up to the total current

exact

Tutorials

Each tutorial’s Check the results section is a test as well: the busbar (energy balance, bolt temperatures, sweep trend), the wrench (deflection convergence against beam theory), the postprocessing and acoustics walkthroughs, and the electrochemistry tutorials.

Run the tests yourself

From a source checkout, with Elmer installed:

pip install pytest
pytest tests -v
python tests/test_gui_smoke.py      # drives the real main window through nearly every command

Tests that need Elmer are skipped when it isn’t found.