Calculus › Section P · Vector analysis › Topic P.6
Surface integrals
The line integral's big sibling: one type weighs a curved shell, the other counts a field's flow through a surface. The package r_u × r_v du dv is the whole trick — normal direction and tilted area in one bundle.
Weighing a curved shell
A line integral weighed a wire by adding density times arc length; a surface integral weighs a shell by adding density times surface area. If \(f(x, y, z)\) is mass per unit area on a surface \(S\) parametrised by \(\mathbf{r}(u, v)\) over a domain \(D\), then
The parametric-surfaces topic built the engine: \(|\mathbf{r}_u \times \mathbf{r}_v|\,du\,dv\) is the tilted-parallelogram element \(dS\), and it does all the geometry while \(f\) rides along. As always, only the values of \(f\) on \(S\) itself matter.
Two types, one warning
Like line integrals, surface integrals come in two types. The scalar type \(\iint_S f\,dS\) adds a number per unit area and does not care which side of the surface is which — a shell weighs the same from either side. The flux type \(\iint_S \mathbf{F} \cdot \mathbf{n}\,dS\), also written \(\iint_S \mathbf{F} \cdot d\mathbf{S}\), counts a field's flow through the surface and demands an orientation: \(\mathbf{n}\) must be chosen, and reversing it negates the integral. Flux is signed — positive means net flow in the direction of \(\mathbf{n}\) (outward, for a closed surface), negative means net flow inward, zero means as much crosses one way as the other.
Flux by parametrisation
For \(S\) parametrised by \(\mathbf{r}(u, v)\) over \(D\), the flux of \(\mathbf{F}\) is
Notice there is no absolute value. The package \((\mathbf{r}_u \times \mathbf{r}_v)\,du\,dv\) is \(\mathbf{n}\,dS\) in one bundle: it carries the normal direction and the area magnitude together, so dotting with \(\mathbf{F}\) extracts the component crossing the surface, weighted by area. If the parametrisation's cross product points against the intended orientation, negate the integral or swap \(u\) and \(v\). For a closed surface, outward is the standard; a parametrisation that produces inward normals must be flipped before quoting an outward flux.
The graph shortcut
For a graph \(z = g(x, y)\) with upward orientation, the parametric-surfaces topic computed \(\mathbf{r}_x \times \mathbf{r}_y = \langle -g_x, -g_y, 1\rangle\), so
No square root, no separate \(dS\): the normal vector absorbs all the stretching. Sanity check on a flat roof, \(g\) constant: the formula reduces to \(\iint_R P\,dA\) — the vertical component of the field times floor area, which is exactly what rain through a horizontal roof should measure.
Worked example: a cone's shell
A shell lies on the cone \(z = \sqrt{x^2 + y^2}\) from the tip to height 1, with density \(f = z\) mass per unit area — heavier toward the rim. Parametrise \(\mathbf{r}(u, v) = \langle v\cos u, v\sin u, v\rangle\) with \(0 \le v \le 1\): the surface area element is \(dS = v\sqrt{2}\,du\,dv\) from the parametric-surfaces topic, and \(z = v\) on the surface, so
Second route, over the shadow: for the same cone, \(\sqrt{1 + g_x^2 + g_y^2} = \sqrt{2}\) everywhere, so \(dS = \sqrt{2}\,dA\) and, in polar coordinates, \(\text{mass} = \int_0^{2\pi}\int_0^1 r \cdot \sqrt{2} \cdot r\,dr\,d\theta = \sqrt{2} \cdot 2\pi \cdot \tfrac{1}{3}\), the same value from a different parametrisation.
Worked example: flux out of a sphere
Find the flux of \(\mathbf{F} = \langle x, y, z\rangle\) outward through the sphere of radius \(a\). Geometry route: on the sphere \(\mathbf{F} = a\,\hat{\mathbf{r}}\) while the outward normal is \(\hat{\mathbf{r}}\), so \(\mathbf{F} \cdot \mathbf{n} = a\) at every point and the flux is \(a \times\) (area \(4\pi a^2\)) = \(4\pi a^3\). Parametrised route: the sphere's \(\mathbf{r}_\phi \times \mathbf{r}_\theta = a^2\sin\phi\,\hat{\mathbf{r}}\), so the integrand is \(\mathbf{F} \cdot (\mathbf{r}_\phi \times \mathbf{r}_\theta) = a^3\sin\phi\) and \(\int_0^{2\pi}\int_0^{\pi} a^3\sin\phi\,d\phi\,d\theta = a^3(2)(2\pi) = 4\pi a^3\). Both routes agree — and the answer deserves a second look: it is exactly 3 times the ball's volume \(\tfrac{4}{3}\pi a^3\). That is no coincidence; it is the divergence theorem, one topic away.
The volume behind the flux
The divergence of \(\mathbf{F} = \langle x, y, z\rangle\) is 3 — a uniform source density, everywhere the same. And the sphere's flux came out as \(4\pi a^3\), which is 3 multiplied by the ball's volume, as if each cubic unit of interior contributed exactly 3 units of outflow. That is the divergence theorem's promise: for a closed surface, total outward flux equals \(\iiint \operatorname{div}\mathbf{F}\,dV\), the source density integrated over the interior. The next topic proves it. This topic's job is fluency in flux itself — through graphs, through parametrised patches, outward through spheres — so that the theorem has something to add up.
Common mistakes to avoid
First, using \(|\mathbf{r}_u \times \mathbf{r}_v|\) in a flux integral: the dot product wants the oriented cross product; the absolute value erases the sign and the orientation with it. Second, double-counting the area: the flux formula already carries \(dS\) inside the cross product; multiplying by \(|\mathbf{r}_u \times \mathbf{r}_v|\) again scales the answer by the surface's stretch factor. Third, wrong-side normals: a parametrisation whose cross product points inward turns 'outward flux' negative — test one sample point before trusting the sign. Fourth, inventing orientation where none belongs: the scalar integral \(\iint f\,dS\) needs no normal and none should be invented. Fifth, mixing up domains: for a graph the shadow \(R\) is the domain because \(\langle -g_x, -g_y, 1\rangle\,dA\) already contains the stretch, but for a sphere the \((\phi, \theta)\) rectangle is the domain — the settings differ, and the surface element knows the difference.
Key takeaways
The scalar surface integral \(\iint_S f\,dS = \iint_D f(\mathbf{r}(u, v))\,|\mathbf{r}_u \times \mathbf{r}_v|\,du\,dv\) adds \(f\) per unit area — a shell's mass — and ignores orientation. The flux \(\iint_S \mathbf{F} \cdot d\mathbf{S} = \iint_D \mathbf{F}(\mathbf{r}(u, v)) \cdot (\mathbf{r}_u \times \mathbf{r}_v)\,du\,dv\) counts flow through \(S\) with orientation: the bundle \((\mathbf{r}_u \times \mathbf{r}_v)\,du\,dv\) is \(\mathbf{n}\,dS\), direction and magnitude in one. For an upward graph, flux is \(\iint_R \langle M, N, P\rangle \cdot \langle -g_x, -g_y, 1\rangle\,dA\) with no square root. The sphere's flux of \(\langle x, y, z\rangle\) is \(4\pi a^3\) — three times the ball's volume, one divergence per unit volume — and that arithmetic is the divergence theorem, glimpsed from below.
Flashcards
Flashcard
Add f per unit area over the shell
Test your understanding
Question 1 of 5
A shell lies on the cone z = √(x² + y²) from the tip to height 1 with density f = z per unit area. What is its mass?
Question 2 of 5
What is the flux of F = ⟨x, y, z⟩ outward through a sphere of radius a?
Question 3 of 5
The flux of F through a parametrised surface r(u,v) is computed as which integral?
Question 4 of 5
The upward flux of F = ⟨M, N, P⟩ through the graph z = g(x,y) over region R is...
Question 5 of 5
If the orientation of S is reversed, what happens to the flux ∬_S F · dS?
Standards: Beyond AP
