-
Notifications
You must be signed in to change notification settings - Fork 19
Expand file tree
/
Copy pathbasis_QH8_2D.tex
More file actions
76 lines (69 loc) · 2.82 KB
/
basis_QH8_2D.tex
File metadata and controls
76 lines (69 loc) · 2.82 KB
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
\begin{flushright} {\tiny {\color{gray} \tt basis\_QH8\_2D.tex}} \end{flushright}
%~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
This element is proposed in \textcite{zhxi20} (2020). Two remarks
must be made: 1) Eq.~(29) of their publication which is the definition
of the basis functions contains an error\footnote{
Answer from the author: ``N5 to N8 is missing an A in the denominator and
the calculation program does not have this problem''}.
2) The authors use a rather
uncommon and annoying rotated numbering:
\begin{verbatim}
y
|
2=====5=====1 3=====6=====2
| | | | (r_0,s_0)=(-1,-1) (r_4,s_4)=( 0,-1)
| | | | (r_1,s_1)=(+1,-1) (r_5,s_5)=(+1, 0)
6 8--x 7 + 5 (r_2,s_2)=(+1,+1) (r_6,s_6)=( 0,+1)
| | | | (r_3,s_3)=(-1,+1) (r_7,s_7)=(-1, 0)
| | | |
3=====7=====4 0=====4=====1
Zhang & Xiang our numbering
\end{verbatim}
For each element they define (their numbering):
\begin{eqnarray}
A &=& \frac{1}{2} [ (x_1-x_3)(y_2-y_4)-(x_2-x_4)(y_1-y_3) ] \nn\\
m_x &=& (x_1-x_4)(y_2-y_3)-(x_2-x_3)(y_1-y_4) \nn\\
m_y &=& (x_3-x_4)(y_1-y_2)-(x_1-x_2)(y_3-y_4) \nn
\end{eqnarray}
Note that $A$ is the area of the element, and that in the case when
the element is a rectangle then $m_x=m_y=0$.
\begin{eqnarray}
\bN_1(r,s)&=& n_1(r,s) +(m_x^2 - m_xm_y + m_y^2)\frac{E(r,s)}{D} \nn\\
\bN_2(r,s)&=& n_2(r,s) +(m_x^2 + m_xm_y + m_y^2)\frac{E(r,s)}{D} \nn\\
\bN_3(r,s)&=& n_3(r,s) +(m_x^2 - m_xm_y + m_y^2)\frac{E(r,s)}{D} \nn\\
\bN_4(r,s)&=& n_4(r,s) +(m_x^2 + m_xm_y + m_y^2)\frac{E(r,s)}{D} \nn\\
\bN_5(r,s)&=& n_5(r,s) -m_x(2Am_x+m_y^2)\frac{E(r,s)}{AD} \nn\\
\bN_6(r,s)&=& n_6(r,s) -m_y(2Am_y+m_x^2)\frac{E(r,s)}{AD} \nn\\
\bN_7(r,s)&=& n_7(r,s) +m_x(-2Am_x+m_y^2)\frac{E(r,s)}{AD} \nn\\
\bN_8(r,s)&=& n_8(r,s) +m_y(-2Am_y+m_x^2)\frac{E(r,s)}{AD} \nn
\end{eqnarray}
with
\[
E(r,s)=(1-r^2)(1-s^2)
\qquad
D=4(4A^2+m_x^2+m_y^2)
\]
and where the $n_i$ functions are the basis functions of the 'regular'
8-node element (see Section~\ref{sec:serendipity2D}).
This is implemented in \stone 52.
\todo[inline]{not finished. SHOW CONSISTENCY !! like in paper
email sent to author about mistake. }
Let us verify consistency:
\begin{eqnarray}
\sum_{i=1}^8 \bN_i(r,s)
&=& \underbrace{\sum_{i=1}^8 n_i(r,s)}_{=0} + \frac{E(r,s)}{D}
\left[
(m_x^2 - m_xm_y + m_y^2)
+(m_x^2 + m_xm_y + m_y^2)
+(m_x^2 - m_xm_y + m_y^2)
+(m_x^2 + m_xm_y + m_y^2) \right. \nn\\
&& \left.
-m_x(2Am_x+m_y^2)\frac{1}{A}
-m_y(2Am_y+m_x^2)\frac{1}{A}
+m_x(-2Am_x+m_y^2)\frac{1}{A}
+m_y(-2Am_y+m_x^2)\frac{1}{A}
\right] \nn\\
&=& \frac{E(r,s)}{D}
\left[(4m_x^2 + 4m_y^2) + \frac{1}{A} (-4Am_x^2 - 4A m_y^2) \right] \\
&=& 0
\end{eqnarray}