15.7 Surface and Flux Integrals

1027

OBJECTIVES

When you finish this section, you should be able to:

  1. Find a surface integral using a double integral (p. 1027)
  2. Determine the orientation of a surface (p. 1030)
  3. Find the flux of a vector field across a surface (p. 1033)
    Application: Electric Flux

Surface integrals are similar to line integrals, but the integration takes place on a surface instead of along a curve.

Suppose the smooth surface \(S\) has the parametrization \[ \begin{equation*} \mathbf{r}(u,v) =x(u,v) \mathbf{i}+y(u,v) \mathbf{j}+z(u,v) \mathbf{k} \end{equation*} \]

and its parameter domain \(R\) is a closed, bounded region of the \(uv\)-plane. Then following our usual practice, we partition a rectangular region containing \(R\) into subrectangles using lines parallel to the \(u\)-axis and the \(v\)-axis and discard any subrectangles not lying entirely in \(R\). Suppose we are left with \(n\) subrectangles \(R_{k},\) \(k=1,2,\ldots ,n\), with width \(\Delta u_{k}\) and height \(\Delta v_{k}.\) The area of the subrectangle \(R_{k}\) is \(\Delta A_{k}=\Delta u_{k}\Delta v_{k}.\)

The \(k\)th subrectangle \(R_{k}\) is formed by the lines \(u=u_{k},\) \(u=u_{k}+\Delta u_{k},\) \(v=v_{k},\) and \(v=v_{k}+\Delta v_{k}.\) The portions of the lines forming the edges of the subrectangle \(R_{k}\) map to portions of their corresponding coordinate curves on the surface \(S.\) That is, each subrectangle \(R_{k}\) maps into a closed curved patch \(S_{k}\) on \(S\) whose area is \(\Delta S_{k}.\) We define the norm \(\left\Vert \Delta \right\Vert \) to be the largest area \(\Delta S_{k}.\) We now choose a point \(P_{k}=(u_{k}^{\ast },v_{k}^{\ast })\) in each subrectangle.

Let \(w=F(x,y,z)\) be a function that is defined over a solid in space containing the surface \(S\). We evalute \(F\) at the point \((x(u^*_k,v^*_k), y(u^*_k,v^*_k), z(u^*_k,v^*_k) )\) on \(S_{k}\) and form the product \(F(x(u^*_k,v^*_k), y(u^*_k,v^*_k), z(u^*_k,v^*_k) )\Delta S_{k}\). Now we add obtaining the sums \[ \sum_{k=1}^{n}F(x(u^*_k,v^*_k), y(u^*_k,v^*_k), z(u^*_k,v^*_k))\,\Delta S_{k} \]

Then we have the following definition.

spanDEFINITIONspan Surface Integral

Let \(S\) be a surface defined on a closed, bounded region \(R\) with a smooth parametrization \(\mathbf{r}(u,v) =x(u,v) \mathbf{i}+y(u,v) \mathbf{j} +z(u,v) \mathbf{k}\). Also, let \(w= F(x,y,z)\) be defined over a solid containing the surface \(S.\) Then the surface integral of \(F\) over \(S\) is defined as \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ \iint\limits_{\kern-3ptS}F(x,y,z)\,dS=\lim\limits_{\Vert \Delta \Vert \rightarrow 0}\sum\limits_{k=1}^{n} F\left(x\left( u_{k}^{\ast },v_{k}^{\ast }\right)\, y\left(u_{k}^{\ast },v_{k}^{\ast }\right)\,z\left( u_{k}^{\ast },v_{k}^{\ast }\right) \right) \Delta S_{k} }} \]

provided this limit exists.

1 Find a Surface Integral Using a Double Integral

When the function \(w=F( x,y,z) \) is continuous on some solid that contains \(S\), the surface integral of \(F\) over \(S\) can be expressed as a double integral.

THEOREM A Surface Integral Expressed as a Double Integral

Let \(S\) be a surface defined on a closed, bounded region \(R\) with a smooth parametrization \(\mathbf{r}(u,v) =x(u,v) \mathbf{i}+y(u,v) \mathbf{j} +z(u,v) \mathbf{k}\). Also, suppose that \(w=F(x,y,z)\) is continuous on a solid containing the surface \(S\). Then the surface integral of \(F\) over \(S\) is given by \[ \begin{equation*} \bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{\iint\limits_{\kern-3ptS}F(x,y,z)\,dS=\iint\limits_{\kern-3ptR}F(x(u,v) ,y(u,v) ,\,z(u,v))\left\Vert \mathbf{r} _{u}\times \mathbf{r}_{v}\right\Vert \,du\,dv } } \tag{1} \end{equation*} \]

1028

It is important to note that just as a line integral does not depend on the parame- trization used, surface integrals do not depend on the smooth parametrization used in the integration. Also, there are occasions when we want to find a surface integral over a piecewise-smooth surface. A piecewise-smooth surface is a finite collection of smooth surfaces that intersect only along piecewise-smooth curves. In these cases, just as with line integrals, we integrate over each smooth surface separately and then sum the results.

Figure 52 The surface \(x^{2}+z^{2} = 9,\) \(0\leq y\leq 2\).

Finding a Surface Integral

Find \(\iint\limits_{\kern-3ptS}ye^{x^{2}+z^{2}}dS,\) where \(S\) is the part of the surface of the cylinder \(x^{2}+z^{2}=9\) that lies between \(y=0\) and \(y=2.\)

Solution The surface \(S\) is shown in Figure 52. Since the surface is a piece of a cylinder, we use a variation of cylindrical coordinates to parametrize the surface. Here, we use \(x=r\cos \theta ,\) \(y=y,\) and \(z=r\sin \theta .\) (Do you see why?) Then \(r^{2}=x^{2}+z^{2}=9,\) so \(r=3.\) We parametrize the surface \(S\) using \[ \begin{equation*} \mathbf{r}( \theta ,y) =3\cos \theta\ \mathbf{i}+y\mathbf{j}+3\sin \theta\ \mathbf{k} \end{equation*} \]

where the parameter domain \({R}\) is \(0\leq \theta \leq 2\pi \) and \(0\leq y\leq 2.\)

The partial derivatives of \(\mathbf{r}\) are \[ \mathbf{r}_{\theta }( \theta ,y) =-3\sin \theta\ \mathbf{i}+3\cos \theta\ \mathbf{k}\qquad\hbox{and}\qquad \mathbf{r}_{y}( \theta ,y) =\mathbf{j} \]

and their cross product is \[ \begin{equation*} \mathbf{r}_{\theta }\times \mathbf{r}_{y}= \left|\begin{array}{c@{\quad}c@{\quad}c} \mathbf{i} & \mathbf{j} & \mathbf{k} \\ -3\sin \theta & 0 & 3\cos \theta \\ 0 & 1 & 0 \end{array}\right| =-3\cos \theta\ \mathbf{i}-3\sin \theta\ \mathbf{k} \end{equation*} \]

Then \(\left\Vert \mathbf{r}_{\theta }\times \mathbf{r}_{y}\right\Vert =\sqrt{9\cos ^{2}\theta +9\sin ^{2}\theta }=3.\) So, we have \[ \begin{eqnarray*} \iint\limits_{\kern-3ptS}ye^{x^{2}+z^{2}}dS &=&\iint\limits_{\kern-3ptR}ye^{9}\left\Vert \mathbf{r}_{\theta }\times \mathbf{r}_{y}\right\Vert d\theta dy=\int_{0}^{2}\int_{0}^{2\pi }( ye^{9}) 3\,d\theta\, dy \notag \\[4pt] &=&3e^{9}\int_{0}^{2}y\big[ \theta \big] _{0}^{2\pi }dy=3e^{9}\int_{0}^{2}2\pi y\,dy=6\pi e^{9}\left[ \dfrac{y^{2}}{2}\right] _{0}^{2}=12\pi e^{9} \end{eqnarray*} \]

RECALL

A lamina is a thin, flat sheet of material.

NOW WORK

Problem 3.

The formulas for finding the center of mass of a lamina can be generalized. The following formulas for the center of mass of a general lamina with variable mass density \(\rho =\rho ( x,y,z) \) in the shape of a surface \(S\) result from a derivation similar to that found in Section 14.4 for a flat sheet.

NEED TO REVIEW?

The center of mass of a lamina with variable density is discussed in Section 14.4, pp. 929-932.

\[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ \begin{array}{r@{\qquad}rcl} \hbox{Mass of \(S\):} & M&=&\iint\limits_{\kern-3ptS}\rho ( x,y,z) \,dS \\ \hbox{Moment about the \(yz\)-plane:} & M_{yz}&=&\iint\limits_{\kern-3ptS}x\rho ( x,y,z) \,dS\\ \hbox{Moment about the \(xz\)-plane:} & M_{xz}&=&\iint\limits_{\kern-3ptS}y\rho ( x,y,z) \,dS\\ \hbox{Moment about the \(xy\)-plane:} & M_{xy}&=&\iint\limits_{\kern-3ptS}z\rho ( x,y,z) \,dS\\ \hbox{Center of mass:} & ( \overline{x},\overline{y},\overline{z}) &=&\left( \dfrac{M_{yz}}{M},\dfrac{M_{xz}}{M},\dfrac{M_{xy}}{M}\right) \end{array} }} \]

1029

Finding the Mass of a Lamina

A lamina in the shape of the cone \(z=6-\sqrt{x^{2}+y^{2}}\) lies between the planes \(z=2\) and \(z=5\). If the mass density of the lamina is \(\rho (x,y,z)=\sqrt{x^{2}+y^{2}}\), find the mass \(M\) of the lamina.

Figure 53

SolutionFigure 53 illustrates the lamina. Using \(x=r\cos \theta , \) \(y=r\sin \theta ,\) and \(z=6-\sqrt{x^{2}+y^{2}}=6-r,\) we parametrize the surface as \[ \mathbf{r}=\mathbf{r}\left( r,\theta \right) =r\cos \theta \,\mathbf{i} +r\sin \theta \,\mathbf{j}+\left( 6-r\right) \mathbf{k}\qquad 0\leq \theta \leq 2\pi\quad 1\leq r\leq 4 \]

First we find \(\mathbf{r}_{r}\times \mathbf{r}_{\theta }.\) \[ \begin{eqnarray*} \mathbf{r}_{r}\times \mathbf{r}_{\theta } &=& \left|\begin{array}{c@{\quad}c@{\quad}c} \mathbf{i} & \mathbf{j} & \mathbf{k} \\ \cos \theta & \sin \theta & -1 \\ -r\sin \theta & r\cos \theta & 0 \end{array}\right| =r\cos \theta \,\mathbf{i} + r\sin \theta \,\mathbf{j}+\left( r\cos ^{2}\theta +r\sin ^{2}\theta \right) \mathbf{k}\\ &=&r\cos \theta \,\mathbf{i}+r\sin \theta \,\mathbf{j}+r\,\mathbf{k} \\[4pt] \hspace{-2pt}\left\Vert \mathbf{r}_{r}\times \mathbf{r}_{\theta }\right\Vert &=&\sqrt{r^{2}\cos ^{2}\theta +r^{2}\sin ^{2}\theta +r^{2}}=r\sqrt{\cos ^{2}\theta +\sin ^{2}\theta +1}=r\sqrt{1+1}=\sqrt{2}\,r \end{eqnarray*} \]

Since the mass density of the lamina is \(\rho (x,y,z)=\sqrt{x^{2}+y^{2}}=r\) and \(dS=\left\Vert \mathbf{r}_{r}\times \mathbf{r}_{\theta }\right\Vert dr\,d\theta =\sqrt{2}r\,dr\,d\theta ,\) the mass \(M\) of the lamina is given by \[ \begin{eqnarray*} M &=&\iint\limits_{\kern-3ptS}\rho (x,y,z)\,dS=\iint\limits_{\kern-3ptR}\rho \left\Vert \mathbf{r}_{r}\times \mathbf{r}_{\theta }\right\Vert dr\,d\theta =\iint\limits_{\kern-3ptR}r(\sqrt{2}r) dr\,d\theta \\ &=&\sqrt{2}\int_{0}^{2\pi }\int_{1}^{4}r^{2}\,dr\,d\theta =\sqrt{2} \int_{0}^{2\pi }\left[ \dfrac{r^{3}}{3}\right] _{1}^{4}\,d\theta =21\sqrt{2} \int_{0}^{2\pi }d\theta =42\sqrt{2}\,\pi \end{eqnarray*} \]

NOW WORK

Problem 33.

RECALL

The centroid is the geometric center of a lamina, and it is numerically equivalent to the center of mass of a homogeneous lamina that has mass density \(\rho =1.\)

Finding the Centroid of a Surface

Find the centroid of the part of the unit sphere that lies in the first octant. See Figure 54.

Figure 54

Solution In spherical coordinates, the equation of the unit sphere \(S\): \(x^{2}+y^{2}+z^{2}=1\) is \(\rho =1.\) So using \(x=\cos \theta \sin \phi \), \(y=\sin \theta \sin \phi \), and \(z=\cos \phi\), we obtain the parametrization \[ \mathbf{r}=\mathbf{r}\left( \theta ,\phi \right) =\cos \theta \sin \phi \, \mathbf{i}+\sin \theta \sin \phi \,\mathbf{j}+\cos \phi\, \mathbf{k};\qquad 0\leq \theta \leq \dfrac{\pi }{2},\quad 0\leq \phi \leq \dfrac{\pi }{2} \]

We seek \(\left\Vert \mathbf{r}_{\theta }\times \mathbf{r}_{\phi }\right\Vert \) and \(dS.\) The partial derivatives of \(\mathbf{r}=\mathbf{r}\left( \theta ,\phi \right) \) are \[ \begin{eqnarray*} \mathbf{r}_{\theta }&=&-\sin \theta \sin \phi \,\mathbf{i}+\cos \theta \sin \phi \,\mathbf{j} \,\quad \mathbf{r}_{\phi }=\cos \theta \cos \phi \,\mathbf{i}+\sin \theta \cos \phi \,\mathbf{j}-\sin \phi\, \mathbf{k}\\[4pt] \mathbf{r}_{\theta }\times \mathbf{r}_{\phi } &=&\left|\begin{array}{c@{\quad}c@{\quad}c} \mathbf{i} & \mathbf{j} & \mathbf{k} \\[3pt] -\sin \theta \sin \phi & \cos \theta \sin \phi & 0 \\[3pt] \cos \theta \cos \phi & \sin \theta \cos \phi & -\sin \phi \end{array}\right|\\[4pt] &=&-\cos \theta \sin ^{2}\phi \,\mathbf{i}-\sin \theta \sin ^{2}\phi \,\mathbf{ j}-\sin \phi \cos \phi\, \mathbf{k} \\[4pt] \left\Vert \mathbf{r}_{\theta }\times \mathbf{r}_{\phi }\right\Vert &=&\sqrt{ \cos ^{2}\theta \sin ^{4}\phi +\sin ^{2}\theta \sin ^{4}\phi +\sin ^{2}\phi \cos ^{2}\phi }\\[4pt] &=&\sqrt{( \cos ^{2}\theta +\sin ^{2}\theta ) \sin ^{4}\phi +\sin ^{2}\phi \cos ^{2}\phi } \\[4pt] &=&\sqrt{\sin ^{2}\phi (\sin ^{2}\phi +\cos ^{2}\phi) }=\sin \phi\\[4pt] dS&=&\left\Vert \mathbf{r}_{\theta }\times \mathbf{r}_{\phi }\right\Vert d\theta\, d\phi =\sin \phi \,d\theta \,d\phi \end{eqnarray*} \]

1030

Since the surface area \(S\) of a sphere is \(S=4\pi r^{2},\) the surface area of the part of the unit sphere \(r=1\) in the first octant is \(\dfrac{1}{8}S=\dfrac{\pi }{2},\) so the mass \(M\) is \(\dfrac{\pi }{2}.\) Because of symmetry, \(\overline{x}=\overline{y}=\overline{z},\) so we need to find only one moment, say \(M_{xy}.\) \[ \begin{eqnarray*} M_{xy}&=&\iint\limits_{\kern-3ptS}z\,dS=\iint\limits_{\kern-3ptS}\cos \phi \,dS=\int_{0}^{\pi /2}\int_{0}^{\pi /2}\cos \phi\, \sin \phi\, d\theta\, d\phi\\[4pt] &=&\int_{0}^{\pi /2}\dfrac{\pi }{2}\cos \phi\, \sin \phi\, d\phi =\dfrac{\pi }{2}\left[ \dfrac{\sin ^{2}\phi }{2}\right] _{0}^{\pi /2}=\dfrac{\pi }{4} \end{eqnarray*} \]

The centroid of the part of the unit sphere in the first octant is \[ \left( \overline{x},\overline{y},\overline{z}\right) =\left( \dfrac{M_{yz}}{M },\dfrac{M_{xz}}{M},\dfrac{M_{xy}}{M}\right) =\left( \dfrac{1}{2},\dfrac{1}{2},\dfrac{1}{2} \right) \]

2 Determine the Orientation of a Surface

NEED TO REVIEW?

Unit normal vectors are discussed in Section 11.2, pp. 769-770.

Suppose the surface \(S\) has a tangent plane at each point, except possibly at its boundary points. Then there are two unit normal vectors, \(\mathbf{n}_{1}\) and \(\mathbf{n}_{2}\mathbf{,}\) where \(\mathbf{n}_{1}=-\mathbf{n}_{2},\) that can be drawn at each point of \(S.\) See Figure 55.

Figure 55

Suppose it is possible to choose a unit normal vector \(\mathbf{n}\) at each point of the surface \(S\) so that \(\mathbf{n}\) varies continuously over the surface, and as \(\mathbf{n}\) moves around any closed curve on \(S\), \(\mathbf{ n}\) returns to its original direction, as shown in Figure 56. Such a surface is said to be orientable, with the vector \(\mathbf{n}\) providing its orientation. As a result, a surface \(S\) that is orientable has two possible orientations, one for \(\mathbf{n}\) and one for \(-\mathbf{n}\).

Figure 56 Two orientations for a surface \(S.\)

Suppose \(S\) has a smooth parametrization \(\mathbf{r}=\mathbf{r}(u,v) \), and \((u_{0},v_{0}) \) is in the parameter domain \(R. \) Then the vector \(\mathbf{r}_{u}(u_{0},v_{0}) \times \mathbf{r} _{v}(u_{0},v_{0})\) is normal to the tangent plane at \(\mathbf{r}(u_{0},v_{0}) ,\) and the two unit vectors normal to the surface \(S\) are \[ \begin{equation*} \mathbf{n}=\dfrac{\mathbf{r}_{u}\times \mathbf{r}_{v}}{\left\Vert \mathbf{r} _{u}\times \mathbf{r}_{v}\right\Vert }\qquad\hbox{and}\qquad -\mathbf{n}=-\dfrac{ \mathbf{r}_{u}\times \mathbf{r}_{v}}{\left\Vert \mathbf{r}_{u}\times \mathbf{ r}_{v}\right\Vert } \tag{2} \end{equation*} \]

When \(S\) is defined by the graph of a function \(z=f(x,y) ,\) then the two orientations can be called upward and downward depending on the sign of the vector component \(\mathbf{k}\). For general surfaces, there is no natural choice for up and down.

Suppose \(F( x,y,z) =z-f(x,y) \). Then the two unit vectors normal to \(S\) are \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ \mathbf{n}=\dfrac{{\mathbf\nabla }F}{\Vert {\mathbf\nabla } F\Vert }=\dfrac{-f_{x}(x,y)\mathbf{i}-f_{y}(x,y)\mathbf{j}+\mathbf{k}}{\sqrt{ [f_{x}(x,y)]^{2}+[f_{y}(x,y)]^{2}+1}} } } \tag{3} \]

and \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ -\mathbf{n}=\dfrac{-{\mathbf\nabla }F}{\Vert {\mathbf\nabla }F\Vert }=\dfrac{f_{x}(x,y)\mathbf{i}+f_{y}(x,y)\mathbf{j}-\mathbf{k} }{\sqrt{[f_{x}(x,y)]^{2}+[f_{y}(x,y)]^{2}+1}} }} \tag{4} \]

NOTE

When \(F( x,y,z) =z-f(x,y)\), then \(S\) is the level surface \(F( x,y,z) =0\). Level surfaces are discussed in Section 12.1, pp. 812-815.

Since the \(\mathbf{k}\) component of \(\mathbf{n}\) is positive, it points in the upward direction. For this reason, \(\mathbf{n}\) is called the upward-pointing unit normal vector of \(S\). The \(\mathbf{k}\) component of \(-\mathbf{n}\) is negative. So, \(-\mathbf{n}\) is called the downward-pointing unit normal vector of \(S\). Using the two unit vectors, we define the positive side or outside of \(S\) as the side associated with the upward-pointing unit vector, and the negative side or inside of \(S\) as the side associated with the downward-pointing unit vector. See Figure 57.

1031

Figure 57 The outside of the open surface is the side on which the upward-pointing unit normal vectors \(\mathbf{n}\) originate.

Finding the Orientations of a Surface

Find both orientations for the paraboloid defined by \(z=9-x^{2}-y^{2},\) \(z\geq 0.\)

  1. Use rectangular coordinates to find the unit normal vectors.
  2. Parametrize the surface, and then find the unit normal vectors.

Solution (a) Since the paraboloid is given by \(f(x,y) = 9-x^{2} - y^{2}, z \ge 0\), we can use the equations (3) and (4) to find the unit normal vectors. We begin by finding the partial derivatives \[ f_{x}(x,y) =-2x \qquad f_{y}(x,y) =-2y \]

Then \[ \begin{eqnarray*} \mathbf{n}&=&\dfrac{2x\,\mathbf{i}+2y\,\mathbf{j}+\mathbf{k}}{\sqrt{[-2x]^{2}+[-2y]^{2}+1}}=\dfrac{2x\,\mathbf{i}+2y\,\mathbf{j}+\mathbf{k}}{\sqrt{4x^{2}+4y^{2}+1}} \\[7pt] -\mathbf{n}&=&\dfrac{-2x\,\mathbf{i-}2y\,\mathbf{j}-\mathbf{k}}{\sqrt{[-2x]^{2}+[-2y]^{2}+1}}=\dfrac{ -2x\,\mathbf{i}-2y\,\mathbf{j}-\mathbf{k}}{\sqrt{4x^{2}+4y^{2}+1}} \end{eqnarray*} \]

Notice that the \(\mathbf{k}\) component of \(\mathbf{n}\) is positive, indicating \(\mathbf{n}\) is the upward-pointing unit normal vector.

(b) We parametrize \(z=9-x^{2}-y^{2},\) \(z\geq 0\), using cylindrical coordinates, by defining the parametric equations \[ \begin{equation*} x=r\cos \theta \qquad y=r\sin \theta \qquad z=9-(x^{2}+y^{2}) =9-r^{2} \end{equation*} \]

Then the parametrization is \[ \begin{equation*} \mathbf{r}=\mathbf{r}( r,\theta ) =r\cos \theta \,\mathbf{i} +r\sin \theta \,\mathbf{j}+( 9-r^{2}) \mathbf{k} \end{equation*} \]

We seek \(\left\Vert \mathbf{r}_{r}\times \mathbf{r}_{\theta }\right\Vert \). The partial derivatives of \(\mathbf{r}\) are \[ \begin{eqnarray*} \mathbf{r}_{r}&=&\cos \theta \,\mathbf{i}+\sin \theta \,\mathbf{j}-2r\mathbf{k} \qquad \mathbf{r}_{\theta }=-r\sin \theta \,\mathbf{i}+r\cos \theta \,\mathbf{j}\\[4pt] \mathbf{r}_{r}\times \mathbf{r}_{\theta } &=& \left|\begin{array}{c@{\quad}c@{\quad}c} \mathbf{i} & \mathbf{j} & \mathbf{k} \\[3pt] \cos \theta & \sin \theta & -2r \\[3pt] -r\sin \theta & r\cos \theta & 0 \end{array}\right| =2r^{2}\cos \theta\, \mathbf{i}+2r^{2}\sin \theta\, \mathbf{j}+r\mathbf{k} \\[4pt] \left\Vert \mathbf{r}_{r}\times \mathbf{r}_{\theta }\right\Vert &=&\sqrt{ 4r^{4}\cos ^{2}\theta +4r^{4}\sin ^{2}\theta +r^{2}}=\sqrt{4r^{4}+r^{2}}=r \sqrt{4r^{2}+1} \end{eqnarray*} \]

The unit normal vectors are \[ \begin{eqnarray*} \mathbf{n}&=&\dfrac{\mathbf{r}_{r}\times \mathbf{r}_{\theta }}{\left\Vert \mathbf{r}_{r}\times \mathbf{r}_{\theta }\right\Vert }=\dfrac{2r^{2}\cos \theta\, \mathbf{i}+2r^{2}\sin \theta\, \mathbf{j}+r\mathbf{k}}{r\sqrt{4r^{2}+1}} =\dfrac{2r\cos \theta\, \mathbf{i}+2r\sin \theta\, \mathbf{j}+\mathbf{k}}{\sqrt{ 4r^{2}+1}}\\[4pt] -\mathbf{n}&=&\dfrac{-2r\cos \theta\, \mathbf{i} -2r\sin \theta\, \mathbf{j}-\mathbf{k}}{\sqrt{4r^{2}+1}} \end{eqnarray*} \]

Figure 58 The outer unit normal vectors of \(z=9-x^{2}-y^{2},\) \(z\geq 0.\)

Notice that the unit vectors found in (a) and (b) are equivalent.

Figure 58 shows the surface and some of the outer (upward-pointing) unit normal vectors of the paraboloid.

NOW WORK

Problem 17.

For a closed surface \(S,\) that is, a surface that forms the boundary of a closed, bounded solid \(E\) in space, such as a sphere, we agree that the positive orientation of \(S\) has the unit normal vectors \(\mathbf{n}\) to \(S\) pointing away from \(E\) and that the negative orientation of \(S\) has the unit normal vectors \(-\mathbf{n}\) to \(S\) pointing inward. For closed surfaces \(S\) with a positive orientation, the unit normal vectors are called outer unit normal vectors; for closed surfaces \(S\) with a negative orientation, the unit normal vectors are called inner unit normal vectors.* See Figure 59.

* Although at first these definitions seem easy, they contain some hidden difficulties. The discussion here and in the next section is limited to simple types of oriented solids \(E\). A discussion of more general solids is found in metric topology.

1032

All curves are orientable, but not all surfaces are orientable. A famous example of a surface that is not orientable is a Möbius strip. See Figure 60. Although it is not immediately apparent, a Möbius strip has only one side and one boundary component. If you were to begin tracing a line on one side of a Möbius strip and continue to follow it once around, you would end on the other side of the strip, that is, the unit normal vector \(\mathbf{n}\) now points in the opposite direction.

Figure 59 \(S\) is a closed surface. The outer unit normal vectors point away from the surface.
Figure 60 Möbius strip: a nonorientable surface.

NOTE

A Möbius strip is formed by taking a long rectangular strip of paper, giving one end a half-twist, and attaching the two ends. The Möbius strip is named after the German mathematician August Möbius (1790–1868).

Finding the \(\mathbf{k}\) Component of the Outer Unit Normal Vectors of an \(xy\,\)-Simple Surface

Consider the surface \(S\) with positive orientation that forms the boundary of the closed, bounded solid \(E\) shown in Figure 61. Notice that the surface \(S\) can be decomposed into three surfaces, \(S_{1},\) \(S_{2},\) and \(S_{3}\), and that each of the outer unit normal vectors \(\mathbf{n}_{1},\) \(\mathbf{n}_{2}\), and \(\mathbf{n}_{3}\) of the three surfaces points away from the solid \(E\) enclosed by \(S\).

Figure 61

Assume the surfaces \(S_{1}\) and \(S_{2}\) are defined by the equations \[ \begin{equation*} S_{1}{:}\quad z=f_{1}(x,y)\qquad \hbox{and}\qquad S_{2}{:}\quad z=f_{2}(x,y) \end{equation*} \]

where \(f_{1}(x,y)<f_{2}(x,y)\) on \(D,\) the functions \(f_{1}\) and \(f_{2}\) are continuous on \(D\), and both \(f_{1}\) and \(\ f_{2}\) have continuous partial derivatives on \(D\). The surface \(S_{3}\) is the portion of a cylindrical surface between \(S_{1}\) and \(S_{2}\) formed by lines parallel to the \(z\)-axis and along the boundary of \(D\). The solid \(E\) is enclosed by \(S_{2}\) on the top, \(S_{1}\) on the bottom, and the lateral surface \(S_{3}\) on the sides. A surface \(S\) with these properties is called \(\boldsymbol x\)\(\boldsymbol y\)-simple.

Find the \(\mathbf{k}\) component of the outer unit normal vectors of \(S\).

Solution For the top surface \(S_{2}\): \(z=f_{2}(x,y)\), the outer unit normal vector \(\mathbf{n}_{2}\) is the same as the upward-pointing unit normal vector to \(S_{2}\). \[ \begin{equation*} \mathbf{n}_{2}=\dfrac{-(f_{2})_x(x,y)\mathbf{i}-(f_{2})_ y(x,y)\mathbf{j}+\mathbf{k} }{\sqrt{[(f_{2})_x(x,y)]^{2}+[(f_{2})_y(x,y)]^{2}+1}} \end{equation*} \]

NEED TO REVIEW?

Direction cosines of a vector in space are discussed in Section 10.4, pp. 718-719.

The \(\mathbf{k}\) component of \(\mathbf{n}_{2}\) is given by the direction cosine \(\cos \gamma _{2}\), where \[ \cos \gamma _{2}=\mathbf{n}_{2}\,{\cdot}\, \mathbf{k}=\dfrac{1}{\sqrt{ [(f_{2})_x(x,y)]^{2}+[(f_{2})_y(x,y)]^{2}+1}} \]

1033

For the bottom surface \(S_{1}\): \(z=f_{1}(x,y)\), the outer unit normal vector \(\mathbf{n}_{1}\) equals the downward-pointing unit normal vector \(-\mathbf{n}\) to \(S_{1}\). \[ \begin{equation*} \mathbf{n}_{1}=-\mathbf{n}=\dfrac{(f_{1})_{x}(x,y)\mathbf{i}+(f_{1})_y(x,y)\mathbf{j}- \mathbf{k}}{\sqrt{[(f_{1})_x(x,y)]^{2}+[(f_{1})_y(x,y)]^{2}+1}} \end{equation*} \]

The \(\mathbf{k}\) component of \(\mathbf{n}_{1}\) is given by the direction cosine \(\cos \gamma _{1}\), where \[ \cos \gamma _{1}=\mathbf{n}_{1}\,{\cdot}\, \mathbf{k}=\dfrac{-1}{\sqrt{[(f_{1})_x(x,y)]^{2}+[(f_{1})_y(x,y)]^{2}+1}} \]

For the lateral surface \(S_{3}\), the outer unit normal vector \(\mathbf{n} _{3} \) is orthogonal to \(\mathbf{k}\) at each point on \(S_{3}\). The \(\mathbf{k }\) component of \(\mathbf{n}_{3},\) given by the direction cosine, \(\cos \gamma _{3}\), is therefore \(0\).

Finding the Outer Unit Normal Vectors to a Surface \(S\)

Find the outer unit normal vectors to the solid \(E\) enclosed by \[ z=f(x,y)=\sqrt{R^{2}-x^{2}-y^{2}}\quad\hbox{and}\quad z=0\qquad 0\leq x^{2}+y^{2}\leq R^{2} \]

Figure 62

Solution The solid \(E\) is the interior of a hemisphere with center at \((0,0,0)\) and radius \(R\) as shown in Figure 62. The surface \(S\) consists of two surfaces, \(S_{1}\) and \(S_{2}\). The bottom surface \(S_{1}\) and top surface \(S_{2}\) are defined by \[ \begin{equation*} S_{1}{:}\quad z=0 \quad \hbox{and}\quad S_{2}{:}\quad z=f(x,y)=\sqrt{R^{2}-x^{2}-y^{2}}\qquad 0\leq x^{2}+y^{2}\leq R^{2} \end{equation*} \]

The outer unit normal vector \(\mathbf{n}_{1}\) of \(S_{1}\) is \(-\mathbf{k}\).

To find the outer unit normal vector \(\mathbf{n}_{2}\) of \(S_{2}\), we find \(f_{x}(x,y) \) and \(f_{y}(x,y) .\) \[ \begin{equation*} f_{x}(x,y)=\dfrac{-x}{\sqrt{R^{2}-x^{2}-y^{2}}}=\dfrac{-x}{z}\qquad \hbox{and} \qquad f_{y}(x,y)=\dfrac{-y}{\sqrt{R^{2}-x^{2}-y^{2}}}=\dfrac{-y}{z} \end{equation*} \]

Then \[ \begin{eqnarray*} \mathbf{n}_{2}&=&\dfrac{-f_{x}(x,y)\mathbf{i}-f_{y}(x,y)\mathbf{j}+\mathbf{k}}{ \sqrt{[f_{x}(x,y)]^{2}+[f_{y}(x,y)]^{2}+1}}=\dfrac{\dfrac{x}{z}\mathbf{i}+ \dfrac{y}{z}\mathbf{j}+\mathbf{k}}{\sqrt{\dfrac{x^{2}}{z^{2}}+\dfrac{y^{2}}{ z^{2}}+1}}\\ &=&\dfrac{x\mathbf{i}+y\mathbf{j}+z\,\mathbf{k}}{\sqrt{ x^{2}+y^{2}+z^{2}}}=\dfrac{x\mathbf{i}+y\mathbf{i}+z\mathbf{k}}{R} \end{eqnarray*} \]

NOW WORK

Problem 13.

3 Find the Flux of a Vector Field Across a Surface

An important application of a surface integral measures the rate of flow of a fluid across a surface \(S,\) also called the flux across \(S.\) Consider solar winds (protons and electrons flowing out of the Sun) as they strike Earth (a spherical surface). The rate of flow of the solar winds across the surface of Earth is an example of flux.

Figure 63

Suppose \(\mathbf{F}=\mathbf{F}(x,y,z)\) is a vector field that represents the velocity of a fluid, such as air or water, and suppose \(\rho =\rho (x,y,z)\) denotes the variable mass density of the fluid. Then \(\rho \mathbf{F}\) equals the rate of flow (mass per unit time) per unit area of the fluid. Let \(S\) be an orientable surface immersed in the fluid that does not impede its flow. If \(\mathbf{n}\) is a unit normal vector of \(S\), then \(\rho \mathbf{F}\,{\cdot}\, \mathbf{n}\) equals the rate of flow of fluid in the direction of \(\mathbf{n}\). See Figure 63.

We seek a formula for the flux across \(S,\) that is, the mass of fluid that flows across \(S\) in a unit of time.

We begin by partitioning \(S\) into \(n\) patches \(S_{i},\) \(i=1,\,2,\,\ldots ,\,n,\) and concentrate on patch \(S_{i}\) whose area is \(\Delta S_{i}.\) The mass of fluid per unit time crossing patch \(S_{i}\) in the direction of the unit normal \(\mathbf{n}_{i}\) is approximately \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ \rho \mathbf{F}\,{\cdot}\, \mathbf{n}_{i}\Delta S_{i} } }\]

1034

where \(\mathbf{F}\) is calculated at some point on the patch \(S_{i}\). We define the norm \(\left\Vert \Delta \right\Vert \) of the partition to be the largest area \(\Delta S_{i}.\) See Figure 64.

Figure 64

Now we form the sums \[ \sum\limits_{i=1}^{n}\rho \mathbf{F}\,{\cdot}\, \mathbf{n}_{i}\Delta S_{i} \]

The flux of \(\mathbf{F}\) across \(S,\) that is, the mass of fluid flowing across the surface \(S\) in a unit of time in the direction of \(\mathbf{n}\), is \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ \dfrac{\hbox{Mass of fluid}}{\hbox{Unit time}} =\lim\limits_{\Vert \Delta \Vert \rightarrow 0}\sum\limits_{i=1}^{n}\rho \mathbf{F}\,{\cdot}\, \mathbf{n}_{i}\Delta S_{i}=\iint\limits_{\kern-3ptS}\rho \mathbf{F} \,{\cdot}\, \mathbf{n}\,dS }} \]

provided the limit exists.

Finding the Flux of \(\mathbf{F}\) Across a Cylinder

A fluid has a constant mass density \(\rho .\) Find the mass of fluid flowing across the cylinder \(x^{2}+y^{2}=4,\) where \(0\leq z\leq 3,\) in a unit of time in the direction outward from the \(z\)-axis, if the velocity of the fluid at any point on the cylinder is \(\mathbf{F}=\mathbf{F}(x,y,z)=x\mathbf{i} +y\mathbf{j}+2z\mathbf{k}\). That is, find the flux of \(\mathbf{F}\) across the cylinder.

Solution In cylindrical coordinates \(x^{2}+y^{2}=r^{2}=4,\) or equivalently, \(r=2,\) which leads to the parametrization \(\mathbf{r}\left( \theta ,z\right) =2\cos \theta \,\mathbf{i}+2\sin \theta \,\mathbf{j}+z\,\mathbf{k}\), \(0\leq \theta \leq 2\pi ,\) \(0\leq z\leq 3.\)

We seek \(\left\Vert \mathbf{r}_{\theta}\times \mathbf{r}_{z }\right\Vert \) and \(dS\).

The partial derivatives of \(\mathbf{r}\) are \[ \begin{eqnarray*} \mathbf{r}_{\theta }&=&-2\sin \theta \,\mathbf{i}+2\cos \theta \,\mathbf{j}\quad \quad \ \mathbf{r}_{z}=\mathbf{k}\\[4pt] \mathbf{r}_{\theta }\times \mathbf{r}_{z} &=& \left|\begin{array}{c@{\quad}c@{\quad}c} \mathbf{i} & \mathbf{j} & \mathbf{k} \\[3pt] -2\sin \theta & 2\cos \theta & 0 \\[3pt] 0 & 0 & 1 \end{array}\right| =2\cos \theta\, \mathbf{i}+2\sin \theta\, \mathbf{j} \\[4pt] \left\Vert \mathbf{r}_{\theta }\times \mathbf{r}_{z}\right\Vert &=&\sqrt{ 4\cos ^{2}\theta +4\sin ^{2}\theta }=2 \\[4pt] dS &=&\left\Vert \mathbf{r}_{\theta }\times \mathbf{r}_{z}\right\Vert d\theta \,dz=2\,d\theta \,dz \end{eqnarray*} \]

Figure 65 \(x^{2}+y^{2}=4,\) \(0\leq z\leq 3,\) and \(\mathbf{n}=\cos \theta\, \mathbf{i}+\sin \theta\, \mathbf{j}\)

Then the unit normal vector \(\mathbf{n}\) is \[ \begin{equation*} \mathbf{n}=\dfrac{\mathbf{r}_{\theta}\times \mathbf{r}_{z}}{\left\Vert \mathbf{r}_{\theta}\times \mathbf{r}_{z}\right\Vert }=\dfrac{2\cos \theta\, \mathbf{i}+2\sin \theta\, \mathbf{j}}{2}=\cos \theta\, \mathbf{i}+\sin \theta\, \mathbf{j} \end{equation*} \]

These vectors point outward from the surface of the cylinder, so we have the correct orientation. Figure 65 shows the cylinder \(x^{2}+y^{2}=4,\) where \(0\leq z\leq 3,\) and the outer unit normal vectors to the cylinder.

We now write \(\mathbf{F}\) using the parametrization. \[ \mathbf{F}=\mathbf{F}(x,y,z)=\mathbf{F}(\theta ,z)=2\cos \theta \,\mathbf{i} +2\sin \theta \,\mathbf{j}+2z\,\mathbf{k} \]

The velocity of the fluid in the direction of \(\mathbf{n}\) is \[ \begin{equation*} \mathbf{F\,{\cdot}\, n}=\left( 2\cos \theta \,\mathbf{i}+2\sin \theta \,\mathbf{j} +2z\,\mathbf{k}\right) \,{\cdot}\, \left( \cos \theta\, \mathbf{i}+\sin \theta\, \mathbf{j}\right) =2\cos ^{2}\theta \,+2\sin ^{2}\theta =2 \end{equation*} \]

The mass of fluid flowing across the cylinder \(x^{2}+y^{2}=4\) in a unit of time in the direction of \(\mathbf{n}\) is given by \[ \iint\limits_{\kern-3ptS}\rho \,\mathbf{F}\,{\cdot}\, \mathbf{n}\,dS=\int_{0}^{3} \int_{0}^{2\pi }\left( \rho \,2\right) 2\,d\theta \,dz=4\rho \int_{0}^{3}2\pi \,dz=\left( 8\rho \pi \right) 3=24\rho \pi \]

The flux of \(\mathbf{F}\) across \(S\) is \(24\rho \pi.\)

NOW WORK

Problem 39.

1035

Finding the Flux of \(\mathbf{F}\) Across a Cube

A fluid has constant mass density \(\rho .\) Find the mass per unit time of fluid flowing across the cube enclosed by the planes \(x=0\), \(x=1\), \(y=0\), \(y=1\), \(z=0\), and \(z=1\) in the direction of the outer unit normal vectors if the velocity of the fluid at any point on the cube is \(\mathbf{F}=\mathbf{F} (x,y,z)=4xz\mathbf{i}-y^{2}\mathbf{j}+yz\mathbf{k}\). That is, find the flux of \(\mathbf{F}\) across the cube.

Figure 66

SolutionFigure 66 shows the cube and its six faces. The mass per unit time of fluid flowing across the cube in the direction of the outer unit normal vector \(\mathbf{n}\) is \[ \iint\limits_{\kern-3ptS}\rho \mathbf{F}\,{\cdot}\, \mathbf{n}\,dS=\rho \iint\limits_{\kern-3ptS} \mathbf{F}\,{\cdot}\, \mathbf{n}\,dS \]

We decompose \(S\) into its six faces and find the surface integral over each one, as shown in Table 1.

Table 1: TABLE 1
Face \(\mathbf{n}\) \(\mathbf{F}\) \({\bf F} \,{\cdot}\, {\bf n}\) \(\rho \iint\limits_{\kern-3ptS}\mathbf{F}\,{\cdot}\, \mathbf{n}\,dS\)
\(S_{1}={\it ABCO}\) \(z=0\) \(-\mathbf{k}\) \(-y^{2}\mathbf{j}\) 0 0
\(S_{2}={\it OAEG}\) \(y=0\) \(-\mathbf{j}\) \(4xz\mathbf{i}\) 0 0
\(S_{3}={\it OCDG}\) \(x=0\) \(-\mathbf{i}\) \(-y^{2}\mathbf{j}+yz\mathbf{k}\) 0 0
\(S_{4}={\it ABFE}\) \(x=1\) \(\mathbf{i}\) \(4z\mathbf{i}-y^{2}\mathbf{j}+yz\mathbf{k}\) \(4z\) \(\rho \int_{0}^{1}\int_{0}^{1}4z\,dy\,dz=2\rho\)
\(S_{5}={\it BCDF}\) \(y=1\) \(\mathbf{j}\) \(4xz\mathbf{i}-\mathbf{j}+z\mathbf{k}\) \(-1\) \(-\rho \int_{0}^{1}\int_{0}^{1}\,dx\,dz=-\rho\)
\(S_{6}={\it DFEG}\) \(z=1\) \(\mathbf{k}\) \(4x\mathbf{i}-y^{2}\mathbf{j}+y\mathbf{k}\) \(y\) \(\rho\int_{0}^{1}\int_{0}^{1}y\,dx\,dy=\dfrac{1}{2}\rho\)

The mass per unit time of fluid flowing across the cube is \[ 0+0+0+2\rho -\rho +\dfrac{1}{2}\rho = \dfrac{3}{2}\rho \]

NOW WORK

Problem 23.

The idea of flux, the rate of flow of a fluid across a surface \(S,\) describes many different phenomena. As an example, we consider the flow of electricity across a surface \(S.\)

Application: Electric Flux

Suppose \(\mathbf{E}\) is an electric field and \(S\) is a piecewise-smooth surface with orientation \(\mathbf{n}\). Then the electric flux \(\Phi _{\mathbf{E}}\), the rate at which the electric field passes through \(S,\) is given by \[ \begin{equation*} \Phi _{\mathbf{E}}=\iint\limits_{\kern-3ptS}\mathbf{E}\,{\cdot}\, \mathbf{n}\,dS \end{equation*} \]

Further, suppose we place a point charge of \(q\) coulombs (\(\rm {C})\) at the origin. Then the corresponding electric field is \[ \begin{equation*} \mathbf{E}=\left( \dfrac{q}{4\pi \epsilon _{0}}\right) \left( \dfrac{x\mathbf{ i}+y\mathbf{j}+z\mathbf{k}}{(x^{2}+y^{2}+z^{2})^{3/2}}\right) \end{equation*} \]

The field \(\mathbf{E}\) is measured in volts per meter (\(\rm {V}/\rm {\!m}\)) and \(\epsilon _{0}\approx 8.854\times 10^{-12}\rm {C}^{2}/\rm {N}\rm {\!m} ^{2}\) is the permittivity of free space}.

1036

Now suppose \(S_{R}\) is the sphere \(x^{2}+y^{2}+z^{2}=R^{2},\) \(R>0\). If \(F(x,y,z)=x^{2}+y^{2}+z^{2}\), then \(\mathbf{n}=\dfrac{\nabla \mathbf{F}}{ \left\Vert \nabla \mathbf{F}\right\Vert }=\dfrac{x\mathbf{i}+y\mathbf{j}+z \mathbf{k}}{R}\) is the outward-pointing unit normal for \(S_{R}\). Then \[ \begin{eqnarray*} \mathbf{E}\,{\cdot}\, \mathbf{n} &=&\left( \dfrac{q}{4\pi \epsilon _{0}}\right) \left( \dfrac{x\mathbf{i}+y\mathbf{j}+z\mathbf{k}}{(x^{2}+y^{2}+z^{2})^{3/2}} \right) \,{\cdot}\, \dfrac{x\mathbf{i}+y\mathbf{j}+z\mathbf{k}}{R} \\[4pt] &=&\left( \dfrac{q}{4\pi \epsilon _{0}}\right) \left( \dfrac{x^{2}+y^{2}+z^{2} }{R^{3}\,{\cdot}\, R}\right) =\dfrac{q}{4\pi \epsilon _{0}}\dfrac{R^{2}}{R^{4}}= \dfrac{q}{4\pi \epsilon _{0}R^{2}} \end{eqnarray*} \]

Using the fact that the surface area \(S_{R}\) of a sphere of radius \(R\) is \(4\pi R^{2}\), we get \[ \begin{equation*} \Phi _{\mathbf{E}}=\iint\limits_{\kern-3ptS_{R}}\mathbf{E}\,{\cdot}\, \mathbf{n} \,dS=\iint\limits_{\kern-3ptS_{R}}\dfrac{q}{4\pi \epsilon _{0}R^{2}}\,dS=\dfrac{q}{ 4\pi \epsilon _{0}R^{2}}\iint\limits_{\kern-3ptS_{R}}1\,dS=\dfrac{q}{4\pi \epsilon _{0}R^{2}}\,{\cdot}\, 4\pi R^{2}=\dfrac{q}{\epsilon _{0}} \end{equation*} \]

This means that the electric flux of a point charge \(q\) originating at the center of a sphere is independent of the size of the sphere.