-
Notifications
You must be signed in to change notification settings - Fork 19
/
benchmark_blbc89.tex
132 lines (116 loc) · 5.85 KB
/
benchmark_blbc89.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
\begin{flushright} {\tiny {\color{gray} benchmark\_blbc89.tex}} \end{flushright}
%~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The abstract of the original publication by Blankenbach \etal (1989) \cite{blbc89} reads:
\begin{center}
{\it
We have carried out a comparison study for a set of benchmark problems
which are relevant for convection in the Earth's mantle. The cases comprise
steady isoviscous convection, variable viscosity convection and time-dependent
convection with internal heating. We compare Nusselt numbers, velocity,
temperature, heat-flow , topography and geoid data. Among the applied codes
are finite-difference, finite-element and spectral methods. In a synthesis
we give best estimates of the `true' solutions and ranges of uncertainty. We
recommend these data for the validation of convection codes in the future.
}
\end{center}
The temperature is fixed to zero on top and to $\Delta T$ at the bottom,
with reflecting symmetry at the sidewalls (i.e. $\partial_x T=0$)
and there are no internal heat sources.
Free-slip conditions are implemented on all boundaries.
The Rayleigh number is given by
\[
\Ranb = \frac{\alpha g_y \Delta T h^3 }{\kappa \nu}
=\frac{\alpha g_y \Delta T h^3 \rho^2 c_p}{k \mu}
\]
The initial temperature field is given by
\[
T(x,y)=(1-y) - 0.01\cos(\pi x) \sin(\pi x)
\]
The perturbation in the initial temperature fields leads to
a perturbation of the density field and sets the fluid in motion.
Depending on the initial Rayleigh number, the system ultimately reaches a
steady state after some time.
\begin{center}
a)\includegraphics[width=4.5cm]{images/benchmark_blbc89/temp1a}
b)\includegraphics[width=4.5cm]{images/benchmark_blbc89/temp1b}
c)\includegraphics[width=4.5cm]{images/benchmark_blbc89/temp1c}\\
{\captionfont Temperature fields at steady-state for
$\Ranb=10^4$ (a), $\Ranb=10^5$ (b), $\Ranb=10^6$ (c).
Obtained with \elefant code \cite{thie14}.}
\end{center}
\begin{center}
a)\includegraphics[width=13cm]{images/benchmark_blbc89/krhb12a}
b)\includegraphics[width=13cm]{images/benchmark_blbc89/krhb12b}\\
{\captionfont
a) Results for the 2-D benchmark problem with uniform mesh refinement.
\# DoFs indicates the number of degrees of freedom.
Reference results from Blankenbach \etal (1989).
b) Results with adaptive mesh refinement. The number of degrees of freedom (\# DoFs) for
each finest mesh size $h$ varies between time steps;
the indicated numbers provide a typical range.
Note that these are obtained for $\Ranb=216,000$.}
\end{center}
%\begin{center}
%\begin{tabular}{llcccc}
%\hline
% & & Blankenbach \etal (1989) & Tackley (1994) & King (2009) & \\
% & & \cite{blbc89} & \cite{tack94} & \cite{king09} & \elefant \\
%\hline
%\hline
%$\Ranb=10^4$ & $\upnu_{rms}$ & $42.864947 \pm 0.000020$ & 42.775 (0.2\%) & 42.867 (0.005\%) & 42.867 (0.01\%) \\
% & $\Nunb$ & $4.884409 \pm 0.000010$ & 4.878 (0.1\%) & 4.885 (0.02\%) & 4.882 (0.05\%)\\
%$\Ranb=10^5$ & $\upnu_{rms}$ & $193.21454 \pm 0.00010 $ & 193.11 (0.05\%)& 193.248 (0.02\%) & 193.255 (0.02\%)\\
% & $\Nunb$ & $10.534095 \pm 0.000010$ & 10.531 (0.03\%)& 10.536 (0.02\%) & 10.507 (0.26\%) \\
%$\Ranb=10^6$ & $\upnu_{rms}$ & $833.98977 \pm 0.00020 $ & 833.55 (0.05\%)& 834.353 (0.04\%) & 834.712 (0.08\%)\\
% & $\Nunb$ & $21.972465 \pm 0.000020$ & 21.998 (0.1\%)& 21.981 (0.04\%) & 21.695 (1.2\%)\\
%\hline
%\end{tabular} \\
%{\captionfont Steady state Nusselt number and Vrms measurements as reported in the literature and
%obtained with \elefant on a $200\times 200$ grid.}
%\end{center}
\begin{center}
\begin{tabular}{llll}
\hline
& & $\upnu_{rms}$ & $\Nunb$ \\
\hline
Blankenbach \etal (1989) \cite{blbc89} & $\Ranb=10^4$ & $42.864947 \pm 0.000020$ & $4.884409 \pm 0.000010$ \\
& $\Ranb=10^5$ & $193.21454 \pm 0.00010 $ & $10.534095 \pm 0.000010$ \\
& $\Ranb=10^6$ & $833.98977 \pm 0.00020 $ & $21.972465 \pm 0.000020$ \\
\hline
Tackley (1994) \cite{tack94} & $\Ranb=10^4$ & 42.775 & 4.878 \\
& $\Ranb=10^5$ & 193.11 & 10.531 \\
& $\Ranb=10^6$ & 833.55 & 21.998 \\
\hline
King (2009) \cite{king09} & $\Ranb=10^4$ & 42.867 & 4.885 \\
& $\Ranb=10^5$ & 193.248 & 10.536 \\
& $\Ranb=10^6$ & 834.353 & 21.981 \\
\hline
Thieulot (2014) \cite{thie14} & $\Ranb=10^4$ & 42.867 & 4.882 \\
& $\Ranb=10^5$ & 193.255 & 10.507 \\
& $\Ranb=10^6$ & 834.712 & 21.695 \\
\hline
\aspect \cite{aspectmanual} & $\Ranb=10^4$ & & \\
& $\Ranb=10^5$ & & \\
& $\Ranb=10^6$ & & \\
\hline
\end{tabular}\\
{\captionfont Steady state Nusselt number and Vrms measurements as reported in the literature and
obtained with \elefant on a $200\times 200$ grid. King (2009) results on 200x200 grid with ConMan.}
\end{center}
\Literature:
\textcite{trab90} (1990),
\textcite{ogaw93} (1993),
\textcite{trha98} (1998),
\textcite{auha99} (1999),
\textcite{chgs02} (2002),
\textcite{chhl08} (2008),
\textcite{kaks05} (2005),
\textcite{king09} (2009),
\textcite{bepo10} (2010),
\textcite{lezh11} (2011),
\textcite{dawk11} (2011),
\textcite{vyrc13} (2013),
\textcite{trbs21} (2021),
\textcite{dakg22} (2002),
\textcite{siwi20} (2020),
\stone 3.