The Blankenbach convection benchmark

The Blankenbach convection benchmark#

This folder allows running the benchmark cases 1a, 1b, 1c, 2a, and 2b from Blankenbach et al. [1989]:

Blankenbach, B., et al. “A benchmark comparison for mantle convection codes.” Geophysical Journal International 98.1 (1989): 23-38.

The reference values (given in reference_statistics.txt) are from that paper (see Table 9). After building the plugin in the plugin/ subfolder and running the benchmarks by executing the script run_all_models.sh, you can use the python script get_statistics.py to analyze and plot the results. Please note that these computations take a long time to reach steady state, especially on finer meshes, consider adjusting the number of parallel processes and the choice of refinement levels in run_all_models.sh and use ASPECT’s optimized mode (see Debug vs. optimized mode).

The reference output for this benchmark is stored in statistics.txt and shown in Fig. 164.

Screenshot

Fig. 164 Convergence of the Blankenbach benchmark cases. Left panel: Nusselt number over cell size h. Right panel: Relative error over cell size. Gray lines indicate the reference values from Blankenbach et al. [1989] (left panel) and theoretical convergence predicted for different convergence orders (right panel). Dashed lines show a simple heat flux computed as the gradient of temperature, solid lines show the heat flux computed using a consistent boundary flux method.#

Table 8 For reference the number of timesteps needed to compute these results:#

Case

Resolution

#Timesteps

1a

3

72

1a

4

128

1a

5

232

1a

6

470

1a

7

939

1b

3

210

1b

4

347

1b

5

690

1b

6

1368

1b

7

2745

1c

3

830

1c

4

1916

1c

5

3397

1c

6

6728

1c

7

13445

2a

3

1857

2a

4

3088

2a

5

6638

2a

6

13593

2a

7

27244

2b

3

4076

2b

4

8404

2b

5

17034

2b

6

34164

2b

7

68363