7.4 Arc Length and Surfaces of Revolution 7.4.1 Arc length In this sub-section we consider the arc length of a curve which is the graph of a function on a closed interval. Let f : [a, b] Ñ R be a continuous function, and P = ta = x0 ă x1 ă ¨ ¨ ¨ ă xn = bu be a partition of [a, b]. Then the arc length of the graph of f on [a, b] can be approximated by n b ÿ ( )2 (xk ´ xk´1 )2 + f (xk ) ´ f (xk´1 ) , k=1 b ( )2 where for each k, the number (xk ´ xk´1 )2 + f (xk ) ´ f (xk´1 ) is the length of the line ( ) ( ) segment joining points xk , f (xk ) and xk´1 , f (xk´1 ) .
Figure 7.1: The length of the polygonal path P0 P1 P2 ¨ ¨ ¨ Pn approximates the arc length of the graph of f on [a, b]
With ∆xk denoting xk ´ xk´1 , then
( f (x ) ´ f (x ) )2 d n b ÿ ( )2 ÿ n k k´1 2 (xk ´ xk´1 ) + f (xk ) ´ f (xk´1 ) = 1+ ∆xk . k=1 k=1 xk ´ xk´1
If f is differentiable on (a, b), then the Mean Value Theorem implies that for each 1 ď k ď n there exists ck P (xk´1 , xk ) such that f (xk ) ´ f (xk´1 ) = f 1 (ck ) ; xk ´ xk´1 thus n b ÿ ( )2 ÿ n a 2 (xk ´ xk´1 ) + f (xk ) ´ f (xk´1 ) = 1 + f 1 (ck )2 ∆xk k=1 k=1
1 + f 1 (x)2 for partition P. Therefore, if f
a which is a Riemann sum of the function y = a is continuously differentiable on [a, b]; that is, f 1 is continuous on [a, b], then 1 + f 1 (x)2 is Riemann integrable on [a, b]. Therefore, using the arguments in Section ??, we find that the arc length of the graph of a continuously differentiable function f on [a, b] is żba 1 + f 1 (x)2 dx . a Example 7.1. Compute the perimeter of a circle with radius r. ? Let f (x) = r2 ´ x2 . Then the perimeter of a circle with radius r is the same as twice the arc length of the graph of f on [´r, r]. Therefore, the perimeter of a circle with radius r is żr a żr c żr 1 2 x2 1 2 1 + f (x) dx = 2 1+ 2 2 dx = 2r ? dx ´r ´r r ´x ´r r ´ x2 2 żπ 2 1 = 2r ? r cos u du ´ π2 r2 ´ r2 sin2 u żπ żπ 2 r cos u 2 = 2r du = 2r du = 2πr . ´ π2 r cos u ´ π2
x3 1 [1 ] Example 7.2. The arc length of the graph of y = + on the interval , 2 is 6 2x 2 ż2c [ d ( x3 ] 1) 2 [ ] c 2 x2 ż 1 2 1+ + dx = 1+ ´ 2 dx 1 dx 6 2x 1 2 2x 2 2 ż2c ż 2 c( ) x4 1 1 x2 1 2 = 1+ ´ + 4 dx = + 2 dx 1 4 2 4x 1 2 2x ż 22 ( 2 2
x 1 ) ( x3 1 )ˇˇx=2 33 = + 2 dx = ´ ˇ 1 = . 1 2 2 2x 6 2x x= 2 16 π Example 7.3. The arc length of the graph of y = ln(cos x) from x = 0 to x = is 4 (d )2 żπc żπc π sin2 x ż 4 4 4 a 1+ ln(cos x) dx = 1+ dx = 1 + tan2 x dx 0 dx 0 cos2x 0 żπ ˇ π 4 ˇx= 4 = sec x dx = ln | sec x + tan x|ˇ 0 x=0 ? ? = ln( 2 + 1) ´ ln 1 = ln( 2 + 1) . Let f be continuously differentiable on [a, b]. Then the arc length of the graph of f on [a, x], where x P [a, b], is given by żxa s(x) = 1 + f 1 (t)2 dx . a
The fundamental theorem of Calculus then shows that
ds a s 1 (x) = (x) = 1 + f 1 (x)2 dx or equivalently, ds c ( dy )2 = 1+ . dx dx a ż Symbolically, ds = dx2 + dy 2 ; thus the arc length of the graph of a function is ds. This variable s is usually called the arc length parameter.
7.4.2 Surface of Revolution
In this section we consider the surface area of a surface formed by revolving a curve about a line (again, this line is called the axis of revolution and usually is a line parallel to the x-axis or y-axis). Let f : [a, b] Ñ R be a continuous function, and S be the surface formed by revolving the graph of f on [a, b] about the x-axis. The general procedures shown in the previous sections is first finding a way to compute an approximated value of the surface area and then see what is the limit of this approximation as }P} approaches 0. We first try the following idea: let P = ta = x0 ă x1 ă ¨ ¨ ¨ ă xn = bu be a partition of [a, b] and ∆xk = xk ´ xk´1 . Consider the sum n ÿ 2πf (ck )∆xk , ck P [xk´1 , xk ] k=1
which is the sum of the area of cylinders formed by revolving the graph of the constant function y = f (ck ) on [xk´1 , xk ] about the x-axis. Since the sum above is a Riemann sum of the function y = 2πf (x) for partition P, we expect that if f : [a, b] Ñ R is continuous, then as }P} Ñ 0 the sum approaches żb 2π f (x) dx . a If this is true, then the surface of the sphere with radius r is given by żr ? π π 1 + cos(2u) ż ż 2 2 2 2 2 2π r2 ´ x2 dx = 2π r cos u du = 2πr du = π 2 r2 ´r ´ π2 ´ π2 2
which is definitely not the correct area of the sphere with radius r. What is wrong with this idea? The mistake is due to that the area of surface of revolution has to be approximated by the sum of the lateral surface area of frustum of right circular cones rather than sum of lateral surface area of cylinders. The lateral area of the frustum in Figure 7.2 below
Figure 7.2 r1 + r2 is given by 2πrL, where r = ; thus the surface area of S can be approximated by 2 n ÿ f (xk ) + f (xk´1 ) b ( )2 2π (xk ´ xk´1 )2 + f (xk ) ´ f (xk´1 ) k=1 2 n ÿ f (xk ) + f (xk´1 ) a = 2π 1 + f 1 (ck )2 ∆xk k=1 2 żb a It can be shown that the sum above approaches 2πf (x) 1 + f 1 (x)2 dx as }P} approaches a 0. Therefore, the area of the surface formed by revolving the graph of f on [a, b] about the x-axis is given by żb ˇ ˇa 2π ˇf (x)ˇ 1 + f 1 (x)2 dx . a
In general, the area of the surface formed by revolving the graph of f on [a, b] about y = L is given by żb ˇ ˇa 2π ˇf (x) ´ Lˇ 1 + f 1 (x)2 dx . a Example 7.4. The surface area of a sphere with radius r is given by ( ) żr ? c żr ? x2 2π 2 2 r ´x ´0 1+ 2 2 dx = 2π r2 dx = 4πr2 , ´r r ´ x ´r ? where we treat the sphere as a surface formed by revolving the graph of y = r2 ´ x2 about the x-axis. Example 7.5. In this example we consider the area of the surface formed by revolving the x2 y2 b? 2 ) (upper part) ellipse 2 + 2 = 1 (or the graph of y = a ´ x2 on [´a, a] about the a b a x-axis. Using the formula of computing the area of surfaces of revolution above, we find that the surface area is given by ża ? c 2 2 ża c b 2 b x 2πb b2 2 2π a ´ x2 1 + 2 2 dx = a 2 ´ x2 + x dx ´a a a a ´ x2 a ´a a2 ża c 2πb a a 4 a2 ´ b2 2 ż = 2 a ´ (a2 ´ b2 )x2 dx = 2πb 1´ x dx . a ´a ´a a4 ? a2 ´ b2 1. Suppose that a ą b; that is, x-axis is the major axis. Let c = . Then a2 ża ? c ża ? b 2 2 b2 x 2 2π a ´x 1+ 2 2 dx = 2πb 1 ´ c2 x2 dx . ´a a a a ´ x2 ´a 1 Making use of the substitution x = sin u, we find that c ża ? c ż arcsin(ac) a b 2 b2 x 2 1 2π 2 a ´x 1+ 2 2 2 dx = 2πb 1 ´ sin2 u ¨ cos u du ´a a a a ´x ´ arcsin(ac) c ż arcsin(ac) ż arcsin(ac) 2πb 2πb 1 + cos(2u) = cos2 u du = du c ´ arcsin(ac) c ´ arcsin(ac) 2 2πb ( u sin(2u) )ˇˇu=arcsin(ac) = + c 2 4 ˇ u=´ arcsin(ac) [ ? ? 2πa b2 a2 ´ b 2 b a2 ´ b 2 ] =? 2 arcsin + a ´ b2 a a2 ? 2πa2 b a2 ´ b 2 =? 2 arcsin + 2πb2 . a ´ b2 a ? b2 ´ a2 2. Suppose that a ă b; that is, x-axis is the minor axis. Let c = . Then a2 ża ? c ża ? b 2 2 b2 x 2 2π a ´x 1+ 2 2 dx = 2πb 1 + c2 x2 dx . ´a a a a ´ x2 ´a 1 Similar to the previous case, the substitution x = sinh u implies that c ża c ż sinh´1 (ac) a b? 2 b 2 x 2 1 2π a ´ x2 1 + 2 2 2 dx == 2πb 1 + sinh2 u ¨ cosh u du ´a a a a ´x ´1 ´ sinh (ac) c ´1 ´1 2πb sinh (ac) 2πb sinh (ac) 1 + cosh(2u) ż ż 2 = cosh u du = du c ´ sinh´1 (ac) c ´ sinh(ac) 2 2πb ( u sinh(2u) )ˇˇu=sinh´1 (ac) = + c 2 4 ˇ u=´ sinh´1 (ac) [ ? ? ( ? 2πa b2 ´1 a2 ´ b 2 a2 ´ b 2 ´1 b2 ´ a2 )] =? 2 sinh + cosh sinh b ´ a2 a a a 2 ? 2πa b 2 a ´b 2 =? 2 sinh´1 + 2πb2 . b ´a 2 a
