Compute the flux through an open spherical cap with the divergence theorem by closing it with a disk and tracking the normal.
§4.7Estimated reading time: 10 min
The divergence theorem computes flux through a closed boundary. A spherical cap is open: its circular edge is not itself a surface, so it must be closed by adding a disk. The normal of the added disk points out of the solid, which for an upper cap means downward. The cap's requested flux is then the closed total minus the disk contribution.
Describe the solid and its boundary
Let 0<a<1 and
Ea={(x,y,z):x2+y2+z2≤1,z≥a}.
Definition 4.7.1
(Outward orientation of the closed boundary)
The boundary of Ea consists of the upper spherical cap Sa and the disk Da at height z=a. The orientation is the unit normal n pointing out of Ea: on Sa it is n=(x,y,z), and on Da, which bounds the solid from below, it is (0,0,−1).
The radius of the disk follows from x2+y2=1−a2, and its area is π(1−a2).
Proposition 4.7.2
(Flux through a cap by closing it)
For a C1 field on a neighborhood of the closed solid, the divergence theorem gives
∬SaF⋅ndS=∭Ea∇⋅FdV−∬DaF⋅ndS.
Proof
(Subtracting the disk term)
The divergence theorem for Ea, with outward normals on both pieces of the boundary, states ∬SaF⋅ndS+∬DaF⋅ndS=∭Ea∇⋅FdV. Moving the disk term to the right-hand side gives the displayed identity.
The theorem's left-hand boundary includes both pieces. If the disk flux is negative, subtracting it increases the cap flux. A diagram is helpful, but writing this equation fixes the sign before calculating.
Compute the cap volume by slices
At height z, the cross-section is a disk of radius 1−z2, for a≤z≤1. Thus
The factored expression makes the shrinkage as a↑1 clear. This is a volume, not the surface area of the cap. The two quantities have different integrands even though both use the same height variable.
Quick check
(Compute vol(Ea) for a=21.)
Use the factored form of the volume.
Solution · Answer
vol(E1/2)=3π(1−21)2(2+21)=3π⋅41⋅25=245π.
Worked example: the position field
Worked example 4.7.3
(The position field through a cap)
Take F=(x,y,z). Its divergence is 3. The closed flux is
3vol(Ea)=π(1−a)2(2+a).
On the bottom disk F⋅n=−a, so the disk flux is −aπ(1−a2). Therefore
∬SaF⋅ndS=π(1−a)2(2+a)+aπ(1−a2)=2π(1−a).
This has an independent geometric check: on the unit sphere the outward normal is n=(x,y,z), so F⋅n=1. The answer must equal the cap area 2π(1−a).
To derive that area rather than memorize it, parametrize by
r(θ,z)=(1−z2cosθ,1−z2sinθ,z),0≤θ≤2π,a≤z<1.
For z<1 the cross-product magnitude is 1, so integration gives ∫a1∫02π1dθdz=2π(1−a). The coordinate formula is singular at the pole z=1; compute up to 1−ε and take the limit. The pole is one point and does not contribute extra area.
Common mistake
(Using the upward normal on the disk)
If n=(0,0,1) is used on Da for F=(x,y,z), the disk flux becomes +aπ(1−a2) and the cap flux becomes π(1−a)2(2+a)−aπ(1−a2)=2π(1−a)(1−a−a2). For every 0<a<1 this differs from the true value 2π(1−a), by 2π(1−a)(a+a2). The disk is part of the boundary of the solid, so its normal points out of the solid, downward.
Worked example 4.7.4
(A constant vertical field)
Take F=(0,0,1), so ∇⋅F=0 and the closed flux vanishes. On Da we have F⋅n=−1, so ∬DaF⋅ndS=−π(1−a2), and the cap flux is 0−(−π(1−a2))=π(1−a2). As a check, on the sphere F⋅n=z, and the parametrization above gives dS=dθdz, so ∬SazdS=2π∫a1zdz=π(1−a2). In the limit a→0 the flux is π, the area of the equatorial disk onto which the hemisphere projects; as a→1 it tends to 0 because the cap shrinks to a point.
Quick check
(For F=(0,0,z), find the flux through Sa.)
Find the closed flux, then subtract the disk flux.
Solution · Answer
∇⋅F=1, so the closed flux is vol(Ea)=3π(2−3a+a3). On Da, F⋅n=−a, so the disk flux is −aπ(1−a2). The cap flux is 3π(2−3a+a3)+aπ(1−a2)=32π(1−a3), which agrees with the direct value 2π∫a1z2dz.
What changes for another field?
A field need not be parallel to the surface normal. On the unit sphere, the flux integrand is F(x,y,z)⋅(x,y,z); replacing it by ∥F∥ is generally incorrect. For example, if F=(x,y,2z), then the sphere integrand is x2+y2+2z2=1+z2, not a constant. The closure method remains valid because the divergence and the disk flux can still be calculated independently.
Worked example 4.7.5
(A field that is not the normal)
Take F=(x,y,2z). Then ∇⋅F=4, so the closed flux is 4vol(Ea)=34π(1−a)2(2+a). On Da we have F⋅n=−2a, so the disk flux is −2aπ(1−a2) and
Direct check: on the unit sphere F⋅n=1+z2, so ∬Sa(1+z2)dS=2π∫a1(1+z2)dz=32π(4−3a−a3). At a=0 the hemisphere flux is 38π.
Common mistake
(Replacing the normal component by the length of the field)
For F=(x,y,2z) on the unit sphere, F⋅n=1+z2 while ∥F∥=1+3z2. At z=21 these are 45 and 27≈1.32, so the integrands differ. Only the component of F along n enters the flux.
The divergence theorem requires F to be C1 on an open neighborhood of the closed solid. If a field has a singularity inside Ea, first excise it and retain all newly created boundary terms; one cannot apply the formula above to a solid containing the singularity without further justification. Reversing the normal on only one part of the boundary also breaks the oriented identity.
Counterexample mode
A singular field inside the solid
Let F=(x,y,z)/ρ3 with ρ=x2+y2+z2. For ρ=0 one has ∇⋅F=0, so one might claim that the flux out of the unit ball is ∭0dV=0. But on the unit sphere n=(x,y,z) and F⋅n=1, so the flux equals the area 4π. The field is not C1 on a neighborhood of the ball, since it is undefined at the origin. For Ea with a>0 the origin is not in the solid and the theorem applies. Here F⋅n=1 on Sa gives flux 2π(1−a), while on Da we have F⋅n=−a/(r2+a2)3/2, so the flux is −2πa∫01−a2(r2+a2)3/2rdr=−2πa(a1−1)=−2π(1−a). The two cancel, as ∇⋅F=0 requires.
Quick check
(For the solid {x2+y2+z2≤1,z≤−a}, what is the outward normal on its flat face?)
Decide whether the flat face bounds the solid from above or from below.
Solution · Answer
The flat face lies at z=−a and bounds the solid from above, so the outward normal is (0,0,1). Its area is again π(1−a2).