【CFD理论】梯度项-01

Gradient schemes
$\nabla \varphi ={\mathbf{e}}_1\frac{\partial }{\partial x_1}\varphi +{\mathbf{e}}_2\frac{\partial }{\partial x_2}\varphi +{\mathbf{e}}_3\frac{\partial }{\partial x_3}\varphi$

fvSchemes

gradSchemes
{
	default none;
	grad(p) <optional limiter> <gradient scheme> <interpolation scheme>;
}

Gauss gradient scheme

gradSchemes
{
    default         none;
    grad(U)         Gauss <interpolation scheme>;
}

Interpolation schemes

1. linear: cell-based linear
2. pointLinear: point-based linear
3. leastSquares: Least squares

Least-squares gradient scheme

gradSchemes
{
    default         none;
    grad(U)         leastSquares;
}

理论

正交 computing gradient in cartesian grids

$$(\frac{\partial \varphi }{\partial x}{)}_C=\frac{{\varphi }_E-{\varphi }_W}{X_E-X_W}=\frac{{\varphi }_N-{\varphi }_S}{X_N-X_S}$$

非正交 green-gauss gradient

两种计算方法${\varphi }_f$

• face-based
• vertex-based

体心处梯度：
散度定理$\Rightarrow$
$(\nabla \varphi {)}_C=\frac{1}{V_C}{\sum }_f{\varphi }_f\cdot {\mathbf{S}}_f$
${\varphi }_f?$

fig. 9.3a，no skewness,$f=f^{\prime }$
${\varphi }_f=\gamma {\varphi }_C+(1-\gamma ){\varphi }_F$
$\gamma =\frac{||{\mathbf{r}}_F-r_f||}{||{\mathbf{r}}_F-r_C||}=\frac{d_{Ff}}{d_{FC}}$
二阶精度：$\gamma =0.5,{\varphi }_f=\frac{{\varphi }_P+{\varphi }_N}{2}$

fig. 9.3c，skewness,$f\ne f^{\prime }$
泰勒公式：
$f(x)=f(x_0)+f^{\prime }(x_0)(x-x_0)+f^{\prime \prime }(x-x_0{)}^2+...$
$$\begin{array}{rl}{\varphi }_f& ={\varphi }_{f^{\prime }}+correction\\ & ={\varphi }_{f^{\prime }}+(\nabla \varphi {)}_{f^{\prime }}({\mathbf{r}}_f-{\mathbf{r}}_{f^{\prime }})\\ & \approx f(x_0)+f^{\prime }(x_0)(x-x_0)\end{array}$$

• 基于体心的需要修正，基于节点的不需要修正，因为需要求${\varphi }_f$
• 用Gauss求体心的梯度，就要知道面心值，而面心值要修正就要知道体心的梯度

1 face-based graient computation

$$\begin{array}{rl}& 1.计算{\varphi }_{f^{\prime }}=\gamma {\varphi }_P+(1-\gamma ){\varphi }_N\\ & 2.计算(\nabla \varphi {)}_P=\frac{1}{V_P}\sum _f{\varphi }_{f^{\prime }}\cdot {\mathbf{S}}_f\\ & 3.计算更新的{\varphi }_f={\varphi }_{f^{\prime }}+\gamma (\nabla \varphi {)}_P\cdot ({\mathbf{r}}_f-{\mathbf{r}}_P)+(1-\gamma )(\nabla \varphi {)}_N\cdot ({\mathbf{r}}_f-{\mathbf{r}}_N)\\ & 4.更新(\nabla \varphi {)}_P=\frac{1}{V_P}\sum _f{\varphi }_f\cdot {\mathbf{S}}_f\\ & 5.重复3\end{array}$$

example1

a.Green-Gauss method with no correction

$(\nabla \varphi {)}_C=\frac{1}{V_C}{\sum }_f{\varphi }_f\cdot {\mathbf{S}}_f$

face centrolids:
$f_1(8.5,11),f_2(10,6.5),f_3(14.5,7),f_4(17.25,11.5),f_5(\mathrm{14.75.15.5}),f_6(10.5,15.5)$

surface vectors:
$S_{f1}=-6\mathbf{i}+\mathbf{j}$
$S_{f2}=-3\mathbf{i}-4\mathbf{j}$
$S_{f3}=4\mathbf{i}-5\mathbf{j}$
$S_{f4}=5\mathbf{i}-0.5\mathbf{j}$
$S_{f5}=3\mathbf{i}+5.5\mathbf{j}$
$S_{f6}=-3\mathbf{i}+3\mathbf{j}$

interpolation factor $(gc{)}_n$:
$(gc{)}_1=0.487$
$(gc{)}_2=0.427$
$(gc{)}_3=0.502$
$(gc{)}_1=0.538$
$(gc{)}_1=0.492$
$(gc{)}_1=0.455$

${\varphi }_f$ value:
${\varphi }_{f1}=(gc{)}_1{\varphi }_C+(1-(gc{)}_1){\varphi }_F=110.442$
${\varphi }_{f2}=91.364$
${\varphi }_{f3}=123.674$
${\varphi }_{f4}=206.27$
${\varphi }_{f5}=263.012$
${\varphi }_{f5}=158.28$

$(\nabla \varphi {)}_C=\frac{1}{V_C}{\sum }_f{\varphi }_f\cdot {\mathbf{S}}_f=11.889\mathbf{i}+12.433\mathbf{j}$

b.Green-Gauss method with correction

1. ${\varphi }_{f^{\prime }}$ value:
   ${\varphi }_{f{1}^{\prime }}=\frac{{\varphi }_C+{\varphi }_{F1}}{2}=111.875$
   ${\varphi }_{f{2}^{\prime }}=101$
   ${\varphi }_{f{3}^{\prime }}=123.5$
   ${\varphi }_{f{4}^{\prime }}=209.5$
   ${\varphi }_{f{5}^{\prime }}=261.5$
   ${\varphi }_{f{6}^{\prime }}=158.5$

2. $(\nabla \varphi {)}_C=\frac{1}{V_C}{\sum }_f{\varphi }_{f^{\prime }}\cdot {\mathbf{S}}_f=11.53\mathbf{i}+11.826\mathbf{j}$

3. ${\varphi }_f$ value:
   ${\varphi }_{f1}={\varphi }_{f{1}^{\prime }}+0.5\ast [(\nabla \varphi {)}_C+(\nabla \varphi {)}_{F1}]\cdot [{\mathbf{r}}_f-0.5\ast ({\mathbf{r}}_C+{\mathbf{r}}_F)]=111.875+3.7435=115.6185$
   ${\varphi }_{f2}=91.911$
   ${\varphi }_{f3}=115.764$
   ${\varphi }_{f4}=224.097$
   ${\varphi }_{f5}=265.566$
   ${\varphi }_{f6}=176.046$

4. $(\nabla \varphi {)}_C=\frac{1}{V_C}{\sum }_f{\varphi }_f\cdot {\mathbf{S}}_f=11.614\mathbf{i}+13.76\mathbf{j}$

2.vertex-based gradient computation

${\varphi }_n=\frac{{\sum }_{k=1}^{NB(n)}\frac{{\varphi }_{F_k}}{||{\mathbf{r}}_n-{\mathbf{r}}_{F_k}||}}{{\sum }_{k=1}^{NB(n)}\frac{1}{||{\mathbf{r}}_n-{\mathbf{r}}_{F_k}||}}$
${\varphi }_f=\frac{{\varphi }_{n1}+{\varphi }_{n2}}{2}$

补充学习

fluid mechanics 101 26-28

1. GreenGaussGradient
2. LeastSquaresGradient
3. NodeBasedGradient