Given a triangle: $\{(x_1,y_1,z_1),(x_2,y_2,z_2),(x_3,y_3,z_3)\}$.

How do I calculate the surface integral: $ I=\int\int \vec{F}\vec{n}dS$ with $ \vec{F} = [0,0,z] $, and where $ \vec{n} $ is the plane normal?

Introducing the angle $ \alpha $ between the normal and the vertical I guess the integral can be written as: $ I=cos(\alpha)\int\int zdS$?

If I project the triangle down into the xy plane, am I correct in assuming that the area element becomes $ dS'=cos(\alpha)dS $, in which case the integral becomes:

$ I=\int\int zdS'$?

Should I then transform the problem to barycentric cordinates and how do I do this?


$ x = ux_1 + vx_2 + (1-u-v)x_3 $

$ y = uy_1 + vy_2 + (1-u-v)y_3 $

$ x_u = x_1 - x_3 $

$ y_v = y_2 - y_3 $

$ x_v = x_2 - x_3 $

$ y_u = y_1 - y_3 $

$ J = (x_1 - x_3)*(y_2 - y_3) - (x_2 - x_3)*(y_1 - y_3)$

$ I=\int_{u=0}^1\int_{v=0}^1 z(u,v)Jdudv = J\int_{u=0}^1\int_{v=0}^1 uz_1 + vz_2 + (1-u-v)z_3dudv$

$ I=J\int_{u=0}^1\int_{v=0}^1 (z_1 - z_3)u + (z_2 - z_3)v + z_3dudv$

$ I=J[(z_1 - z_3)\int_{u=0}^1udu+(z_2 - z_3)\int_{v=0}^1vdv + z_3]$

$ I=J[(z_1 - z_3)*0.5+(z_2 - z_3)*0.5 + z_3] $

Is this correct?


Note that $\nabla\cdot \mathbf{F}=1$, so the divergence is simple which suggests an appeal to the Divergence Theorem may save us some labor.

Suppose we take the triangle $S$ and project it into the $xy$-plane as $S_0$. (For the present answer I'll assume that $z>0$ for all $\mathbf{x}\in S$; I'll try to remedy this defect later.) We can then construct a vertical triangular prism $D$ whose base is $S_0$ and whose whose top is cut off by a plane to make $S_0$ a slanted triangle face. We denote the remaining three side walls as the region $A$. Thus $\partial D=S+S_0+A$, and the divergence theorem implies that

$$\int_D (\nabla\cdot \mathbf{F})\,dV=\int_{\partial D} \mathbf{F}\cdot d\mathbf{S}=\int_{S} \mathbf{F}\cdot d\mathbf{S}+\int_{S_0} \mathbf{F}\cdot d\mathbf{S}+\int_{A} \mathbf{F}\cdot d\mathbf{S}$$ For the RHS, note that $\mathbf{F}=0$ in the $xy$-plane and $\mathbf{F}\cdot d\mathbf{S}=0$ for $x\in A$ since the local normal vector of $A$ is always in the $xy$-plane; consequently only the first flux integral is non-vanishing. But the divergence of $\mathbf{F}$ is zero, so $\int_{S}\mathbf{F}\cdot d\mathbf{S}$ must be identical to the volume of $D$. This is then purely a geometry problem which I'll leave to the reader (for now, at least.)

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.