Slicing parallel to the axis of revolution, a shell's volume as circumference times height times thickness, why the method needs no inversion of the boundary, shells about lines other than the coordinate axes, and choosing between shells and washers.
Subject: Calculus I · 63 slides · symbolic lesson
Open the interactive version of this deck
Title
Calculus I · Section 6.3
Slicing parallel to the axis instead of across it
Objectives
Section 6.2 sliced a solid of revolution perpendicular to its axis and got discs and washers. That forces the boundary to be written as a function of the variable along the axis, and sometimes it cannot be. This lesson slices the other way.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, pp. 569-581 — learning objectives 6.3.1 and 6.3.2
You already have two ways to find the volume of a solid of revolution, and both of them slice the solid across its axis: the cross-sections come out as discs or as washers. That works whenever you can say how wide the solid is at a given position along the axis, and the price is that the boundary of the region has to be written as a function of the variable running along the axis.
Sometimes it cannot be. Revolve the region under a cubic about the vertical axis and the washer method wants the cubic solved for x; revolve the region under an exponential with a squared exponent and it wants a logarithm inside a square root. This lesson slices the other way instead, parallel to the axis, and the thin piece that results is a tube rather than a disc.
By the end you will be able to derive the formula for the volume of that tube, use it in either direction, handle an axis that is not a coordinate axis, and, most usefully of all, decide before you start which of the three methods will give you the easier integral.
Warm-up
Take the region under the arch given below, between the curve and the horizontal axis, and revolve it about the vertical axis.
\[ y = 2x - x^2, \qquad 0 \le x \le 2 \]
Discussion prompt
A washer slice is horizontal, at height y. Write its outer and inner radii. What do you have to do to the boundary to get them?
Write your answer down before revealing it, because the work you do here is exactly the work the shell method is going to save you.
The slice is horizontal, at some height y, and it stretches from the left branch of the arch to the right branch. To name those two positions you have to solve the equation of the boundary for x, and because the boundary is a quadratic in x you get two roots, one on each side of the peak. They come from the quadratic formula, and the whole calculation of the washer's radii is that formula rearranged.
There is nothing wrong with any of this, and in Section 6.2 you did it many times. Notice only what it depended on: the boundary had to be invertible, and in a usable form. Keep the two roots in view, because you will meet this exact region again in a few slides, revolved about the same axis, with no inversion anywhere in the work.
Section
Part 1
Concept
Figure (svg): The arch y equals 2x minus x squared over the interval from 0 to 2, with a horizontal line drawn at height 0.64 cutting the arch at x equals 0.4 and x equals 1.6; the two crossing points are marked.
The arch is a function of x, so its height at any x is immediate. The washer method never asks for that. It asks for the two x values at a given height y, which means solving the boundary equation for x.
\[ x = 1 - \sqrt{1-y} \quad\text{and}\quad x = 1 + \sqrt{1-y} \]
Here the quadratic formula supplies them. For a cubic, or for an exponential with a squared exponent, there may be no usable formula at all.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 573 — Example 6.13, the region of Figure 6.30
Look at where the horizontal line meets the arch. Two crossings, at two different x values, both of which depend on the height. That is the information a washer needs: the outer radius is the further crossing and the inner radius is the nearer one.
Contrast that with what the picture hands you for free. At any x, the arch's height is the function's value there, no solving required. The function was defined in that direction, so questions asked in that direction are easy and questions asked across it are hard.
Here the quadratic formula rescues the situation. The point of the next slides is that it very often does not: replace the arch by a cubic that mixes powers, or by an exponential of a square, and there is no formula to reach for. The washer method is not wrong for those regions, it is simply unusable, and something else has to slice them.
Intuition
Figure (svg): The same arch on the right of the y-axis with a narrow vertical strip at x equals 1.4, and its mirror image on the left; dashed ellipses join the two strips through the top and the bottom, showing the thin tube they sweep out.
Take a thin vertical strip instead, parallel to the axis of revolution. Revolved, it does not become a disc: it becomes a thin tube, open at top and bottom, wrapped around the axis.
Nothing about the strip needed inverting. Its distance from the axis is x and its height is the function's value at that same x, both read straight off the picture.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 570 — Figure 6.26, the representative rectangle
This is the whole idea of the section, and it is worth staring at the picture until the tube is obvious. The strip stands upright, parallel to the vertical axis of revolution. As the region spins, the strip does not sweep out a flat disc; it sweeps out a thin cylindrical wall, like a tin can with no lid and no base, and the dashed ellipses show where its rim travels.
Every quantity you need is measured in the picture without solving anything. The distance from the strip to the axis is just the coordinate x. The height of the strip is the function's value at that same x, because the strip runs up to the curve.
One more thing to notice, and it surprises people: the strips are laid out along the horizontal direction even though the axis of revolution is vertical. So this method integrates with respect to x when it revolves about the vertical axis, which is the opposite of what discs and washers do.
Section
Part 2
Concept
Figure (svg): A thick-walled cylinder drawn in perspective: the top is a ring with inner radius x sub i minus 1 and outer radius x sub i, the wall has thickness delta x, and the height of the cylinder is f of x star.
The strip runs from one partition point to the next, so the tube it sweeps has an inner radius, an outer radius and a single height. Its cross-section is a ring, and its volume is the ring's area times the height.
\[ V_{\text{shell}} = f(x_i^*)\left(\pi x_i^2 - \pi x_{i-1}^2\right) \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 571 — Figure 6.27
Give the thin tube its proper name: a cylindrical shell. It is the solid caught between two cylinders of the same height, one just inside the other, and the picture shows it cut open at the top so you can see the ring.
The volume of any cylinder is the area of its cross-section times its height, and the cross-section here is a ring, an annulus. The ring's area is the big circle minus the little one. So the shell's volume is the height, which is the function's value at the sample point, multiplied by the difference of the two circle areas.
Written out it looks worse than the disc formula rather than better, and if you stopped here you would have gained nothing. The next slide is the whole reason the method is useful: that difference of two squares collapses into something with no squares in it at all.
Concept
That expression hides a much friendlier one. Take the common factor out and use the difference of two squares.
\[ \pi f(x_i^*)\left(x_i^2 - x_{i-1}^2\right) = \pi f(x_i^*)(x_i + x_{i-1})(x_i - x_{i-1}) \]
Write the sum as twice the average
Why: Half the sum of the two radii is the radius half way through the wall.
\[ x_i + x_{i-1} = 2\cdot\frac{x_i + x_{i-1}}{2} \]
Name the thickness
Why: Consecutive partition points differ by the width of one subinterval.
\[ x_i - x_{i-1} = \Delta x \]
\[ V_{\text{shell}} = 2\pi f(x_i^*)\left(\frac{x_i + x_{i-1}}{2}\right)\Delta x \]
That average is the midpoint of the subinterval, so choosing the sample point to be the midpoint makes the line exact, not approximate.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 571 — the shell volume derivation
Follow the algebra one line at a time. Pull out the common factor of pi and the height, and what is left is the difference of the squares of the two radii. That factors into the sum of the radii times their difference, which is a standard move you have made since algebra.
Now look at what the two factors mean geometrically. The difference of the radii is the thickness of the wall: the width of the strip you started with. Half the sum of the radii is the radius half way through the wall, which is the midpoint of the subinterval. Writing the sum as twice the average is what produces the factor of two in front of pi.
So the shell's volume is twice pi, times the average radius, times the height, times the thickness. Every factor is a length, and none of them is squared. The last line matters for honesty: if you take the sample point to be the midpoint, this expression is not an approximation at all. It is the shell's exact volume.
Picture it
Figure (svg): On the left a thin cylinder with a dashed cut line down its front. An arrow points right to a flat rectangular plate whose width is labelled two pi x, whose height is labelled f of x, and whose depth is labelled delta x.
The same formula arrives with no algebra at all. Cut the tube down its side and flatten it into a plate. The plate is as wide as the tube's circumference, as tall as the tube, and as thick as the strip.
\[ V_{\text{shell}} \approx \underbrace{2\pi x_i^*}_{\text{width}}\cdot\underbrace{f(x_i^*)}_{\text{height}}\cdot\underbrace{\Delta x}_{\text{thickness}} \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, pp. 571-572 — Figure 6.28
Here is the same result with no algebra, and it is the version to remember. Imagine slitting the tube down its side and flattening it out on a table. You get a flat plate, and the volume of a plate is width times height times thickness.
The width is how far it is around the tube, which is the circumference: twice pi times the radius. The height is the height of the tube. The thickness is the thickness of the wall, the width of the original strip.
Say it out loud a few times, because it is the sentence that protects you from every sign and squaring error in this lesson: circumference times height times thickness. The circumference contains the radius exactly once, which is why no shell formula ever squares anything. A disc squares its radius because a disc is a circle; a shell does not, because a shell is a rectangle rolled up.
Prediction
Unrolled, the outer edge of the wall is longer than the inner edge, so the flattened shape is not really a rectangle.
Predict first
Does that make the shell formula an approximation that never goes away?
Correct: No: taking the width at the average radius makes each shell exact
Why: The algebra on the previous slide gave the shell's volume exactly, with the radius equal to the midpoint of the subinterval. The long outer edge and the short inner edge average out precisely. The only approximation left in the whole derivation is choosing a sample point, and taking the limit of the Riemann sum removes it.
Commit to an option before reading on. The objection in the question is a real one and the textbook raises it too: the outside of the wall really is longer than the inside, so the flattened piece is slightly trapezoidal rather than rectangular.
The answer is that the discrepancy cancels exactly, not approximately. The algebra two slides back produced the average of the two radii, and the average of a longer edge and a shorter edge is precisely what a rectangle of the same area would use. Every shell's volume is exact once you place the sample point at the midpoint.
What remains approximate is something else entirely, and it is the same approximation every Riemann sum makes: the height of the shell is taken to be the function's value at one sample point, although the function varies a little across the subinterval. Letting the number of shells grow without bound removes that too, and the limit is the integral.
Concept
Figure (svg): The arch and its mirror image with five pairs of rectangles standing on the axis, each pair representing one cylindrical shell; the rectangles are shaded and appear in sequence.
Fill the solid with nested tubes, one for each subinterval, and add their volumes. What appears is a Riemann sum for one particular function.
\[ V \approx \sum_{i=1}^{n}2\pi x_i^* f(x_i^*)\,\Delta x \]
\[ V = \lim_{n\to\infty}\sum_{i=1}^{n}2\pi x_i^* f(x_i^*)\,\Delta x = \int_a^b 2\pi x f(x)\,dx \]
Notice which variable the sum runs over. The strips are laid out along x, which is perpendicular to the axis of revolution, so the integration is in x even though the axis is the vertical one.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 572 — Rule: the method of cylindrical shells, equation 6.6
Watch the shells appear one at a time. Each pair of rectangles in the picture is one tube seen side on, and stacking them inside one another fills the solid the way a set of nested tins fills a cupboard.
Add the volumes and you have a sum of terms, each of the form twice pi times a sample x, times the function's value there, times the common width. That is a Riemann sum, and not for the original function: for twice pi times x times the function. Take the limit as the number of shells grows and you have the definite integral of that product.
Read the last line of the slide carefully, because it is where most of the confusion in this section lives. The axis of revolution is vertical, and yet the integration is with respect to x. The variable of integration is always the one that runs across the axis, because that is the direction in which the strips are laid out.
Notation
Annotate
On: \( V = \int_a^b \left(2\pi x f(x)\right) dx \)
The rule as printed assumes the function is continuous and never negative on the interval, and that the region runs down to the horizontal axis.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 572 — equation 6.6
Step through the four annotations and connect each to the unrolled plate, not to the formula as a string of symbols. The first factor is the width of the plate, the second is its height, and the differential is its thickness. A volume built as a product of three lengths cannot contain a square, so the annotation on the first factor is the one to keep.
The last annotation answers the question students most often get wrong on a test. The limits are the ends of the region measured across the axis of revolution. For the vertical axis that means the smallest and largest x values the region reaches, and it does not matter how tall the region is.
The conditions printed with the rule are also worth respecting: the function should be continuous and not negative on the interval, and the region should run down to the horizontal axis. When it does not, only the height changes, and you will make that change yourself in Part 5.
Matching
Match the pairs
Why: The shell integrand is the volume of a box: width times height times thickness. Reading it that way is the quickest defence against squaring the radius, because a box has no square in it anywhere.
Make the four pairings before revealing them. If you can do this without hesitating, you can rebuild the shell formula from scratch even if you forget it completely, which is worth more than memorising it.
The one that catches people is the last pair. The interval is not where the region is tall; it is where the strips are, measured across the axis. When you later meet a region described by its top and bottom heights rather than its left and right edges, this is the pairing that tells you the description has to be converted before a shell integral can be written.
Estimation
Revolve the region under one over x, from x equal to 1 to x equal to 3, about the vertical axis. Before setting up anything, look at what one shell contributes.
Predict first
A shell at x has radius x and height one over x. What is its volume per unit of thickness?
Correct: It is the same for every x: two pi
Why: Radius times height is x times one over x, which is 1, for every strip in the region. So each shell contributes two pi per unit of thickness, the strips cover an interval of length 2, and the volume is four pi before any antiderivative is taken.
Figure (svg): A flat horizontal line at height two pi over the interval from x equals 1 to x equals 3, with the region under it shaded; the shaded rectangle has area four pi.
Guess before you check. This is the section's first example and the book does it with a full integral, but the interesting part happens before any integration.
The strip at position x is x units from the axis and one over x units tall. Multiply those together and the x cancels: the product is one, for every strip in the region. That means each shell delivers the same volume per unit of thickness, twice pi, whether it is near the axis or far from it. The narrow tall shells near the left edge and the wide short shells near the right edge contribute equally.
So the volume is twice pi times the length of the interval, which is 2, giving four pi, and the figure is the whole computation: a rectangle. You will do it properly on the next slide, but noticing structure like this before integrating is exactly the habit that makes the rest of this chapter quick.
Worked example
Revolve the region below the curve and above the horizontal axis about the vertical axis.
\[ f(x) = \frac1x, \qquad 1 \le x \le 3 \]
Figure (svg): The curve one over x with the region under it from x equals 1 to x equals 3 shaded, mirrored on the left of the y-axis, and a highlighted strip at x equals 2 joined to its mirror by dashed ellipses.
Name the radius and the height
Why: A strip at x is x units from the vertical axis and reaches up to the curve.
\[ r(x) = x, \qquad h(x) = \frac1x \]
Write the shell integral
Why: Rule 6.6 with a equal to 1 and b equal to 3.
\[ V = \int_1^3 2\pi x\left(\frac1x\right)dx \]
Simplify the integrand
Why: The radius cancels the height.
\[ 2\pi x\cdot\frac1x = 2\pi \]
Integrate the constant
Why: The antiderivative of a constant is that constant times x.
\[ V = \int_1^3 2\pi\,dx = 2\pi x\Big|_1^3 = 6\pi - 2\pi = 4\pi \]
Check with washers
Why: Slicing the other way must give the same number. Below y equal to one third the slice is a washer of radii 3 and 1; above it the outer radius is one over y.
\[ \pi(3^2-1^2)\cdot\tfrac13 + \pi\int_{1/3}^{1}\left(\frac{1}{y^2}-1\right)dy = \frac{8\pi}{3} + \frac{4\pi}{3} = 4\pi \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, pp. 572-573 — Example 6.12
This is the model solution for the whole lesson, so copy its shape. Name the radius, name the height, put them into the rule, simplify, integrate, then check. Four of those six steps happen before any calculus at all.
The picture is worth a moment. The region sits between one and three on the right of the axis, so the solid has a cylindrical hole of radius one down the middle, and the dashed ellipses show the rim of the shell at x equal to two sweeping round through the mirror strip on the left.
The check at the end is the habit to build. Slicing the other way must give the same number, and doing so here takes two washer integrals and a split at height one third, against one line of work by shells. That contrast is the argument for the method, and the agreement of the two answers is your evidence that the set-up was right.
Worked example
\[ f(x) = x^2, \qquad 1 \le x \le 2, \quad \text{about the } y\text{-axis} \]
Figure (svg): The parabola y equals x squared with the region under it between x equals 1 and x equals 2 shaded, mirrored on the left, and a highlighted strip at x equals 1.5 joined to its mirror by dashed ellipses.
Name the radius and the height
Why: Same axis, so the radius is still x; the height is the parabola.
\[ r(x) = x, \qquad h(x) = x^2 \]
Write and simplify the integrand
Why: Radius times height is a cube.
\[ 2\pi x\cdot x^2 = 2\pi x^3 \]
Integrate
Why: Power rule on the cube.
\[ V = \int_1^2 2\pi x^3\,dx = 2\pi\left[\frac{x^4}{4}\right]_1^2 \]
Evaluate
Why: Sixteen quarters minus one quarter is fifteen quarters.
\[ V = 2\pi\left(4 - \frac14\right) = \frac{15\pi}{2} \]
Check with washers
Why: The solid is a ring with a cylindrical hole of radius 1. Below y equal to 1 the slice is a washer of radii 2 and 1; above it the inner radius is the square root of y.
\[ \pi(4-1)(1) + \pi\int_1^4(4-y)\,dy = 3\pi + \frac{9\pi}{2} = \frac{15\pi}{2} \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 573 — Checkpoint 6.12
Try this one on paper before reading the solution; it is the same shape of problem with a different curve. The radius is again x because the axis is again the vertical one, and the height is the parabola, so the integrand is a cube.
Notice the hole again. The region starts at x equal to one, not at the axis, so the solid is a ring with a cylindrical hole of radius one bored through it. Shells handle that without any special treatment: the strips simply start one unit out, and the lower limit of the integral records it.
The washer check is instructive about effort. It needs a split at height one, because below that height the slice runs from the inner cylinder to the outer one, and above it the inner boundary becomes the curve. Two integrals against one, and the same answer, fifteen halves of pi.
Worked example
This is the region whose horizontal slices needed the quadratic formula. By shells it needs nothing of the kind.
\[ f(x) = 2x - x^2, \qquad 0 \le x \le 2, \quad\text{about the } y\text{-axis} \]
Write the integrand
Why: Radius x, height the arch, exactly as given.
\[ 2\pi x\left(2x-x^2\right) = 2\pi\left(2x^2 - x^3\right) \]
Integrate term by term
Why: Power rule twice.
\[ V = 2\pi\left[\frac{2x^3}{3} - \frac{x^4}{4}\right]_0^2 \]
Evaluate at the ends
Why: Sixteen thirds minus four is four thirds.
\[ V = 2\pi\left(\frac{16}{3} - 4\right) = \frac{8\pi}{3} \]
Figure (svg): Two curves over the interval from 0 to 2: the arch 2x minus x squared, low and flat, and the integrand two pi x times that arch, peaking near 7.4 at x equals four thirds, with the region under the taller curve shaded.
Check with the washer set-up from Part 1
Why: The two roots differ by twice the square root, and the difference of their squares collapses.
\[ \pi\int_0^1\left[(1+\sqrt{1-y})^2 - (1-\sqrt{1-y})^2\right]dy = \pi\int_0^1 4\sqrt{1-y}\,dy = \frac{8\pi}{3} \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 573 — Example 6.13
Here is the payoff for the warm-up. This is the region whose horizontal slices needed the quadratic formula, and by shells the set-up is a single line: radius x, height the arch as written, multiply, integrate.
The figure shows something worth understanding about every shell integral. The lower curve is the region's height; the taller curve is the integrand, the height multiplied by twice pi times x. Multiplying by the radius tips the graph to the right, because a strip far from the axis travels further as it spins and so sweeps out more volume than an equally tall strip near the axis.
The check completes the story that Part 1 started. Take the two roots from the quadratic formula, square them, subtract, and almost everything cancels, leaving four times a square root. That integral comes out as eight thirds of pi, exactly what shells gave you, which confirms both the set-up and the inversion.
Fill the middle
The same axis, one coefficient changed: revolve the region under the curve given below about the vertical axis.
\[ f(x) = 3x - x^2, \qquad 0 \le x \le 2 \]
Fill in the blanks
V = \int_0^2 2\pi x\left(3x-x^2\right)dx = 2\pi\left[x^3 - \frac8\pi___\right]_0^2 = ___
Why: The height is the function itself, so the integrand is two pi times x times 3x minus x squared, which is two pi times 3x squared minus x cubed. Its antiderivative is x cubed minus a quarter of x to the fourth; at x equal to 2 that is 8 minus 4, so the volume is two pi times 4, namely eight pi.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 573 — Checkpoint 6.13
Fill in both blanks before checking. The first is the height of the strip, and the only thing you have to notice is that the height is the function exactly as it is given, with no rearranging and no squaring.
The second blank is the arithmetic at the end. The antiderivative is already written for you, so the work is evaluating it at 2 and at 0: eight minus four is four, and twice pi times four is eight pi.
If you wrote the height as a square root, or divided by anything, look back at the unrolled plate. The height of the plate is the height of the strip, and the strip's top is on the curve.
Trap
Discs and washers square their radius, so the hand writes it again:
\[ V = \int_1^2 \pi x^2\left(x^2\right)dx \]
\[ = \frac{31\pi}{5} \approx 19.5 \]
Wrong. That is neither method.
A shell's volume is a plate: width times height times thickness. The circumference contributes the radius once.
\[ V = \int_1^2 2\pi x\left(x^2\right)dx \]
\[ = \frac{15\pi}{2} \approx 23.6 \]
A quick dimension check catches it too: length times length times length, three factors, never four.
This is the most common error in the section, and it comes from good intentions: you have spent Section 6.2 writing pi times a radius squared, and the hand carries on writing it.
The defence is dimensional and takes a second. A volume is three lengths multiplied together. The shell integrand is circumference times height times thickness, which is exactly three. The wrong version has a radius squared, a height and a thickness, which is four lengths, so it cannot be a volume at all.
The second defence is the picture of the unrolled plate. Nothing in a flat plate is squared. If you ever find a square in a shell integrand, it came either from squaring the radius, which is this mistake, or from a genuine square inside the function itself, which is fine.
Section
Part 3
Concept
Figure (svg): The curve y equals x cubed plus x on the interval from 0 to 1 with the region under it shaded, a highlighted vertical strip at x equals 0.45, and a dashed horizontal line at height 1.2 meeting the curve at one unmarked point.
A vertical strip's height is the value of the function at that x. A horizontal slice's width is the set of x values at a given height, which is a different question entirely: it needs the inverse function.
\[ \text{shells: } h = f(x) \qquad \text{washers: } r = f^{-1}(y) \]
For the cubic drawn here the inverse exists, because the curve is increasing, but writing it down takes the cubic formula. The shell set-up does not care.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 569 — the ability to choose the variable of integration
This slide states the advantage in one sentence, and the picture makes it concrete. The highlighted vertical strip has a height you can write down immediately, because the curve is given as a function of x and the strip stands at a definite x.
The dashed horizontal line is the other question. At that height, where is the curve? The marked point has an x value, and to write it you would have to solve a cubic. It can be done, since the curve is increasing and so has an inverse, but the inverse takes the cubic formula, and no sensible course asks for it.
So the choice of method is not about which one is cleverer. It is about which question the boundary is prepared to answer. A boundary given as y in terms of x answers height questions cheaply and position questions expensively, and shells about a vertical axis only ever ask for heights.
Worked example
Revolve the region under the cubic below, over the interval from 0 to 1, about the vertical axis.
\[ f(x) = x^3 + x, \qquad 0 \le x \le 1 \]
Write the integrand
Why: Radius x times height, exactly as the curve is given.
\[ 2\pi x\left(x^3+x\right) = 2\pi\left(x^4 + x^2\right) \]
Integrate term by term
Why: Power rule twice.
\[ V = 2\pi\left[\frac{x^5}{5} + \frac{x^3}{3}\right]_0^1 \]
Evaluate
Why: One fifth plus one third is eight fifteenths.
\[ V = 2\pi\left(\frac15 + \frac13\right) = \frac{16\pi}{15} \approx 3.351 \]
Check with four shells by hand
Why: Take four strips of width one quarter and use the midpoints as radii; the sum should sit just under the exact value, since the integrand is concave up over most of the interval.
\[ 2\pi\sum_{i=1}^{4} m_i\left(m_i^3+m_i\right)\tfrac14 \approx 3.254 \quad \text{against} \quad 3.351 \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 569 — the introduction's promise about complicated functions
The whole solution is four lines because the multiplication by the radius keeps everything polynomial: x times the cubic is a quartic plus a square, and the power rule finishes it.
The check is a numerical one, and it is worth doing at least once in this lesson so that the formula stops being abstract. Four shells, each of thickness one quarter, with radii at the midpoints of the four subintervals, give about 3.254, against the exact value of about 3.351. Four crude tubes already land within three per cent.
You should also be able to say why the estimate is a little low. The integrand is increasing and concave up over most of the interval, so the midpoint rectangles fall slightly short of the area. If your own hand computation had come out above the exact value, that would have been a sign to look for an arithmetic slip.
Intuition
A solid has one volume. If shells and washers both apply, they must agree, and that is the cheapest check available to you: set the problem up the other way and see whether the numbers match.
\[ \int_a^b 2\pi x f(x)\,dx = \pi\int_c^d \left[R(y)^2 - r(y)^2\right]dy \]
The two integrals rarely look alike. Example 6.12 turned a constant into four pi; the washer route to the same number needed a split at one third and an improperish-looking fraction. Same solid, same volume, very different work.
This is not a new technique, it is a working habit, and it is the cheapest quality control you have. A solid has exactly one volume. If both methods apply, both must produce it, so setting up the second one is a complete check on the first.
Do not expect the two integrals to look related. In Example 6.12 the shell integrand collapsed to a constant, while the washer route needed a split and a reciprocal square. They agreed only in their value, four pi.
In an exam you usually have time only for a partial check, and there are cheaper ones: compare your answer with a cylinder that obviously contains the solid, or estimate with two or three shells, or check that the answer grows when the axis moves further away. Any of these will catch a factor of two or a squared radius.
Worked example
Revolve the region under the curve below about the vertical axis. The region's AREA cannot be found in closed form.
\[ f(x) = e^{x^2}, \qquad 0 \le x \le 1 \]
Figure (svg): The curve y equals e to the x squared over the interval 0 to 1, rising from 1 to about 2.7, with the region under it shaded and mirrored across the y-axis, and a highlighted strip at x equals 0.62.
Write the shell integrand
Why: Radius x times height.
\[ 2\pi x e^{x^2} \]
Substitute
Why: Let u be the exponent; its differential is twice x dx, and an x is exactly what the radius supplied.
\[ u = x^2, \quad du = 2x\,dx \]
Rewrite the integral in u
Why: The limits move with the substitution: x equal to 0 gives u equal to 0, and x equal to 1 gives u equal to 1.
\[ V = \int_0^1 2\pi x e^{x^2}dx = \pi\int_0^1 e^{u}\,du \]
Evaluate
Why: The exponential is its own antiderivative.
\[ V = \pi\left[e^{u}\right]_0^1 = \pi(e-1) \approx 5.398 \]
Check with washers
Why: Below y equal to 1 the slice is a full disc of radius 1; above it the inner radius is the square root of the logarithm of y, and the integral of the logarithm is known.
\[ \pi + \pi\int_1^{e}\left(1 - \ln y\right)dy = \pi + \pi(e-2) = \pi(e-1) \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 569 — choosing the variable of integration is an advantage
This example is the strongest argument for the method, so read the two integrals on the slide side by side. The area of this region cannot be written in closed form at all: the antiderivative of the exponential of a square is not an elementary function, which is why statisticians have tables for it.
The volume, on the other hand, takes four lines. The shell radius multiplies the integrand by x, and x is exactly the factor a substitution needs, because the derivative of the exponent is twice x. So the volume of the solid is easier than the area of the region that generates it, which is not something you would predict.
The check is worth following too. Washers do work here, since the curve can be inverted, but the route is longer: a full disc below the height one, then washers whose inner radius is the square root of a logarithm, and an integral of a logarithm to evaluate. Both come to pi times e minus one.
Socratic
Discussion prompt
The area of that region is the integral of e to the x squared, which has no elementary antiderivative. The volume came out in four lines. Say precisely what changed.
Write your answer before revealing, in a single sentence if you can. The aim is to name the mechanism, not just to agree that the example was nice.
The mechanism is this: the shell method multiplies the height by the radius, and when the axis is the vertical one the radius is x. Any time the function's awkwardness lies in a square of x inside it, the extra factor of x is precisely the inner derivative a substitution wants, and the integral becomes elementary.
That is a pattern you can use deliberately. If you meet a region whose boundary involves the exponential, sine or logarithm of a square and you are revolving about the vertical axis, set it up with shells and expect the substitution to work. If you are revolving the same region about the horizontal axis, expect trouble instead, because then the radius is y and it will not help.
Sorting
In every case the boundary of the region is given to you in the form y equal to f of x.
Sort into buckets
Sort each set-up by whether it can use that boundary as written.
The pattern is exactly complementary. Each method fails to invert for one orientation of the axis and needs the inverse for the other, which is why a course needs both.
Sort all six before checking, and for each one ask a single question: which direction does the strip run in, and what has to be measured along it?
Three of them use the boundary as given. Shells about a vertical line use vertical strips, whose heights are values of the function. Discs and washers about a horizontal line use vertical slices too, so their radii are also values of the function. That includes the axis y equal to minus one, since shifting a horizontal axis changes the radius but not the direction of slicing.
The other three need the inverse. Washers about a vertical line need radii measured horizontally, and shells about a horizontal line need lengths measured horizontally, and both of those are x values at a given height.
The pattern is perfectly complementary, and it is the reason a calculus course teaches both methods rather than picking a favourite. For any one region, whichever way the boundary happens to be written, one of the two will avoid the inversion.
Section
Part 4
Concept
Turn every word of the derivation ninety degrees. A horizontal axis of revolution wants strips parallel to it, so the strips are horizontal, the radius of a shell is y, and the strips are spread along y.
\[ V = \int_c^d \left(2\pi y\,g(y)\right)dy \]
the rule for the x-axis — If the region is bounded on the right by the graph of g(y), on the left by the vertical axis, and between the horizontal lines y equal to c and y equal to d, then revolving it about the horizontal axis gives the volume above.
| axis of revolution | strip direction | radius | integrate in |
|---|---|---|---|
| the y-axis (vertical) | vertical | x | x |
| the x-axis (horizontal) | horizontal | y | y |
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 574 — Rule: the method of cylindrical shells for solids of revolution around the x-axis
Nothing new has to be proved here. Turn every sentence of the derivation ninety degrees and it still holds: strips parallel to the axis, tubes lying on their sides, circumference times length times thickness.
What changes is the bookkeeping, and the table on the slide is the whole of it. With a horizontal axis the strips are horizontal, so the radius of a shell is its height above the axis, which is y, and the strips are spread out vertically, so the integration is with respect to y. The boundary therefore has to be available as x in terms of y, which is why the rule is stated with a function of y.
Keep the rule of thumb from Part 2 in front of you: the variable of integration is always the one that runs across the axis of revolution. It is the single sentence that keeps discs, washers and shells straight in all four configurations.
Worked example
The region sits to the left of the curve given below and to the right of the vertical axis, between the heights 0 and 4.
\[ g(y) = 2\sqrt{y}, \qquad 0 \le y \le 4, \quad\text{about the } x\text{-axis} \]
Figure (svg): The region to the left of the curve x equals two root y, between y equals 0 and y equals 4, shaded, with its mirror image below the x-axis, and a horizontal highlighted strip at y equals 2.5 running from the y-axis out to the curve.
Name the radius and the length
Why: The strip is horizontal, y units above the axis, and runs from the vertical axis out to the curve.
\[ r(y) = y, \qquad h(y) = 2\sqrt{y} \]
Write the integral
Why: The rule for the horizontal axis, with c equal to 0 and d equal to 4.
\[ V = \int_0^4 2\pi y\left(2\sqrt y\right)dy = 4\pi\int_0^4 y^{3/2}\,dy \]
Integrate
Why: Power rule with exponent three halves.
\[ V = 4\pi\left[\frac{2y^{5/2}}{5}\right]_0^4 \]
Evaluate
Why: Four to the five halves is 32.
\[ V = 4\pi\cdot\frac{64}{5} = \frac{256\pi}{5} \approx 160.85 \]
Check with washers
Why: Read the same region the other way: for x between 0 and 4 it runs from the parabola y equal to x squared over 4 up to the line y equal to 4.
\[ \pi\int_0^4\left(4^2 - \left(\frac{x^2}{4}\right)^2\right)dx = \pi\left(64 - \frac{64}{5}\right) = \frac{256\pi}{5} \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 574 — Example 6.14
The region here is described sideways, as everything to the left of a curve between two heights, and that is a hint in itself: a region described by its left and right boundaries is asking to be sliced horizontally.
Follow the three quantities in the picture. The strip sits at height y, so its radius, the distance to the horizontal axis, is y. Its length runs from the vertical axis out to the curve, so it is the curve's x value, which is twice the square root of y. Thickness is a small change in y, and the strips run from height zero to height four.
The check reads the same region the other way round, as everything between a parabola and the horizontal line four, for x from zero to four, and computes it with washers. Both give 256 fifths of pi, about 160.85. That washer integral involves a fourth power, which is a fair sign that the shell version was the friendlier one.
Worked example
\[ g(y) = \frac3y, \qquad 1 \le y \le 3, \quad\text{about the } x\text{-axis} \]
Figure (svg): The region left of the curve x equals three over y between y equals 1 and y equals 3, shaded, with a horizontal highlighted strip at y equals 2 reaching from the y-axis to x equals 1.5.
Write the integrand
Why: Radius y times length three over y.
\[ 2\pi y\cdot\frac3y = 6\pi \]
Integrate the constant
Why: Every shell contributes the same amount, as in Example 6.12.
\[ V = \int_1^3 6\pi\,dy = 6\pi y\Big|_1^3 = 12\pi \]
Check with washers
Why: Cut the region at x equal to 1: to its left every slice runs from height 1 to height 3, to its right the top boundary is three over x.
\[ \pi(3^2-1^2)(1) + \pi\int_1^3\left(\frac{9}{x^2}-1\right)dx = 8\pi + 4\pi = 12\pi \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 574 — Checkpoint 6.14
This is the horizontal twin of Example 6.12, and the same cancellation happens: the radius y multiplies the length three over y and the y disappears, leaving a constant integrand.
So every shell here has the same volume, no matter how far it is from the axis. The near shells are long and thin, the far shells are short and wide, and the two effects cancel exactly. The picture shows why: the boundary is a hyperbola, and a hyperbola is precisely the curve on which length times height is constant.
The washer check needs a split at x equal to one, because to the left of that line the slices run the full distance from height one to height three, while to the right the top boundary is the curve. Eight pi plus four pi is twelve pi, which agrees.
Concept
Suppose the axis is the vertical line a distance k to the left of the origin. Every radius in the derivation grows by k, and nothing else in it changes.
\[ \text{outer radius } x_i + k, \qquad \text{inner radius } x_{i-1} + k \]
Put the new radii into the shell volume
Why: The same difference of squares as before, now shifted.
\[ V_{\text{shell}} = 2\pi f(x_i^*)\left(\frac{(x_i+k)+(x_{i-1}+k)}{2}\right)\left(x_i - x_{i-1}\right) \]
Collect the average and the thickness
Why: The average of the shifted radii is the old average plus k.
\[ V_{\text{shell}} \approx 2\pi\left(x_i^* + k\right)f(x_i^*)\,\Delta x \]
\[ V = \int_a^b 2\pi(x+k)f(x)\,dx \]
So the working rule is: the factor x in the integrand was never really x. It was always the distance from the strip to the axis.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 575 — revolving around the line x equal to minus k
This is the derivation the textbook does, and it is short because almost nothing in it is new. Put the axis a distance k to the left of the origin. Each radius in the old derivation grows by k, and the difference of the radii, the thickness, is unaffected because both ends shifted by the same amount.
Run the same factoring as before and you get twice pi, times the average shifted radius, times the height, times the thickness. The average shifted radius is the old midpoint plus k, so the integrand is twice pi times x plus k, times the function.
Take the general lesson rather than the specific formula. The x in the standard rule was never really the coordinate x; it was the distance from the strip to the axis, which happens to equal x when the axis is the vertical coordinate axis. Replace it by the correct distance and the rule works for any parallel line.
Notation
Annotate
On: \( V = \int_a^b 2\pi\,\underbrace{r(x)}_{\text{to the axis}}\,\underbrace{h(x)}_{\text{along the strip}}\,dx \)
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 575 — the x-term must be replaced by an expression for the radius
This is the form to carry into every problem, including the ones from Section 6.2 you may want to redo. Two lengths, measured in perpendicular directions, multiplied by twice pi and integrated.
The third annotation is the one worth testing yourself on. The radius runs across to the axis and the height runs along the strip, so they are measured in perpendicular directions and can never be the same expression. If your radius and your height came out identical, you have almost certainly measured one of them in the wrong direction.
The first annotation is where signs live. A radius is a distance, so write it in the order that makes it positive over the whole region: x plus one when the axis is one unit to the left, two minus x when the axis is at two and the region is on its left. If the region straddles the axis, split it at the axis and keep each piece positive.
Worked example
Revolve the region under the line below about the vertical line one unit to the left of the origin.
\[ f(x) = x, \quad 1 \le x \le 2, \quad\text{about } x = -1 \]
Figure (svg): The line y equals x with the region under it from x equals 1 to x equals 2 shaded, the vertical line x equals minus 1 drawn dashed, and the mirror image of the region on the far side of that line, from x equals minus 4 to minus 3.
Write the radius
Why: The strip at x is one unit further from this axis than from the vertical axis.
\[ r(x) = x + 1 \]
Write the integral
Why: Radius times height, with the height still x.
\[ V = \int_1^2 2\pi(x+1)x\,dx = 2\pi\int_1^2\left(x^2+x\right)dx \]
Integrate
Why: Power rule on each term.
\[ V = 2\pi\left[\frac{x^3}{3} + \frac{x^2}{2}\right]_1^2 \]
Evaluate
Why: Fourteen thirds minus five sixths is twenty-three sixths.
\[ V = 2\pi\left(\frac{14}{3} - \frac56\right) = \frac{23\pi}{3} \approx 24.09 \]
Check by splitting the radius
Why: Write the radius as x plus 1 and integrate the two pieces separately: the first is the volume about the vertical axis, the second is the extra.
\[ 2\pi\int_1^2 x^2dx + 2\pi\int_1^2 x\,dx = \frac{14\pi}{3} + 3\pi = \frac{23\pi}{3} \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, pp. 575-576 — Example 6.15
Work this one before reading it, because the only new thing is one bracket. The height of the strip is unchanged; the radius is the distance from the strip to the new axis, which is one unit more than its distance from the vertical axis.
The picture is drawn so that you can see the mirror image on the far side of the new axis, and so that you can see the hole the solid has in the middle. The region never gets closer than two units to the axis, so the solid is a ring with a substantial cylindrical hole.
The check is the one you should reach for whenever an axis has moved. Split the radius into its two pieces and integrate them separately: the first piece is the volume about the vertical axis, fourteen thirds of pi, and the second is three pi. Together they give twenty-three thirds of pi, which is the printed answer.
Intuition
That check is worth keeping. Splitting the radius splits the volume into the volume about the coordinate axis plus a second term, and the second term is the same for every region.
\[ \int_a^b 2\pi(x+k)f(x)\,dx = \int_a^b 2\pi x f(x)\,dx + 2\pi k\int_a^b f(x)\,dx \]
The last integral is the area of the region, so moving the axis k units further away adds twice pi times k times that area, and nothing else.
\[ V(k) = V(0) + 2\pi k A \]
Two consequences. The volume grows linearly in the distance, and you can check any shifted answer against the unshifted one in one line.
The check on the previous slide generalises, and the result is useful enough to remember. Split the radius, and the volume splits into the volume about the coordinate axis plus twice pi times the shift times the area of the region.
Two things follow. The first is that the volume grows linearly with the distance to the axis, which is not obvious from the geometry: the solid becomes much longer as the axis retreats, but it also becomes much thinner in the middle, and the two effects balance into a straight line.
The second is practical. Any shifted volume can be checked in one line against the unshifted one, provided you know the region's area, and you always do, because that is Section 6.1. This is also a special case of Pappus's theorem, which says that the volume equals the region's area times the distance travelled by its centroid.
Tweak it
Parameter explorer
The curve plots the shell integrand for the region under y = x on [1, 2] when the axis is the line x = −k. The area under the plotted curve is the volume. Slide k: how does that area grow?
\[ V(k) = \int_1^2 2\pi(x + {k})x\,dx \]
\[ V(k) = \frac{14\pi}{3} + 3\pi k \]
Move the slider slowly from zero and watch the block. The curve plotted is the shell integrand over the interval from one to two, and the area of the block under it is the volume, so you are watching the volume grow in real time.
The block rises but keeps its shape: at every x its height goes up by twice pi times x times the shift. The added area is therefore twice pi times the shift times the region's area, which is three halves here, so the volume goes up by three pi for every unit the axis moves.
Set the slider to one and the area is the volume of Example 6.15, twenty-three thirds of pi, about 24.1. Set it to zero and you are back to fourteen thirds of pi, about 14.7, the volume about the vertical axis. The relationship between them is a straight line, not a curve, which is the point made on the previous slide.
Worked example
\[ f(x) = x^2, \quad 0 \le x \le 1, \quad\text{about } x = -2 \]
Write the radius
Why: The distance from the strip at x to the line two units left of the origin.
\[ r(x) = x + 2 \]
Write the integral
Why: Radius times height, the height being the parabola.
\[ V = \int_0^1 2\pi(x+2)x^2dx = 2\pi\int_0^1\left(x^3 + 2x^2\right)dx \]
Integrate
Why: Power rule on each term.
\[ V = 2\pi\left[\frac{x^4}{4} + \frac{2x^3}{3}\right]_0^1 \]
Evaluate
Why: A quarter plus two thirds is eleven twelfths.
\[ V = 2\pi\cdot\frac{11}{12} = \frac{11\pi}{6} \approx 5.76 \]
Check with the shift rule
Why: About the vertical axis the volume is two pi times one quarter; the region's area is one third, so the shift of 2 adds twice pi times 2 times one third.
\[ \frac{\pi}{2} + 2\pi(2)\left(\frac13\right) = \frac{\pi}{2} + \frac{4\pi}{3} = \frac{11\pi}{6} \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 576 — Checkpoint 6.15
Do this one on paper first. The radius is the distance to a line two units left of the origin, so it is x plus two, and the height is the parabola.
The integrand expands to a cube plus twice a square, both handled by the power rule, and the evaluation is at x equal to one, so the arithmetic is a quarter plus two thirds.
Check it with the shift rule rather than with washers, because the shift rule is faster. About the vertical axis this region gives a half of pi. Its area is one third. Moving the axis two units away adds twice pi times two times one third, which is four thirds of pi, and a half plus four thirds is eleven sixths. Being able to run that check in your head is why the previous slide is worth remembering.
Worked example
Now the region lies to the left of the axis of revolution, which changes which way round the subtraction goes.
\[ y = \sqrt x, \quad y = 0, \quad x = 1, \quad\text{about } x = 2 \]
Figure (svg): The curve y equals root x from 0 to 1 with the region under it shaded, the vertical line x equals 2 dashed, and the mirror region from x equals 3 to 4; an arrow from the strip at x equals 0.5 points right to the axis.
Write the radius as a positive length
Why: The strip at x is to the left of the line, so the distance is the line minus x.
\[ r(x) = 2 - x, \qquad 0 \le x \le 1 \]
Write the integral
Why: Radius times height, the height being the square root.
\[ V = \int_0^1 2\pi(2-x)\sqrt x\,dx = 2\pi\int_0^1\left(2x^{1/2} - x^{3/2}\right)dx \]
Integrate
Why: Power rule on each half-integer power.
\[ V = 2\pi\left[\frac{4x^{3/2}}{3} - \frac{2x^{5/2}}{5}\right]_0^1 \]
Evaluate
Why: Four thirds minus two fifths is fourteen fifteenths.
\[ V = 2\pi\left(\frac43 - \frac25\right) = \frac{28\pi}{15} \approx 5.86 \]
Check with washers
Why: Horizontal slices run from the curve out to the line x equal to 1, so the outer radius is 2 minus y squared and the inner radius is 1.
\[ \pi\int_0^1\left[(2-y^2)^2 - 1^2\right]dy = \pi\left(3 - \frac43 + \frac15\right) = \frac{28\pi}{15} \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 580 — Exercise 143
Everything so far has had the region to the right of the axis. Here it is to the left, and exactly one thing changes: the order of the subtraction in the radius.
The distance from a strip at x to the line at two is two minus x, and on this region that is a number between one and two, comfortably positive. Written the other way round it would be negative everywhere, which is the subject of the next slide.
The integration is routine once the radius is right: two half-integer powers, both handled by the power rule. The washer check is a good one to follow through, because the roles of the boundaries swap: horizontal slices now run from the curve out to the vertical line at one, so the outer radius is two minus y squared and the inner radius is the constant one. Both routes give twenty-eight fifteenths of pi.
Trap
Same problem, radius written the other way round:
\[ V = \int_0^1 2\pi(x-2)\sqrt x\,dx \]
\[ = -\frac{28\pi}{15} < 0 \]
Wrong. A volume cannot be negative.
A radius is a distance, and a distance has no sign. With the region to the left of the axis, subtract the other way.
\[ r = 2 - x > 0 \]
The sign of the answer is the alarm. If a shell integral comes out negative, the radius was written backwards; if part of the region is on each side of the axis, split it there and keep both radii positive.
A negative volume is impossible, so this error announces itself if you look at the sign of your answer, and that is the habit to build: glance at the sign before you write the final line.
The cause is always the same. The radius was written as the coordinate minus the axis when the region lies on the other side of it. A radius is a distance and distances have no sign, so subtract in whichever order makes it positive throughout the region.
There is one case where you cannot fix it by swapping, and that is when part of the region lies on each side of the axis. Then split the region at the axis, use the correct positive radius on each piece, and add the two volumes. If you try to run one integral through with a signed radius, the two halves will partly cancel and you will get an answer that is too small, sometimes even zero.
Error analysis
A solution to Example 6.15, with the axis at the line one unit left of the origin.
Annotate
On: \( V = \int_1^2 2\pi(x-1)x\,dx = \frac{5\pi}{3} \)
Read the line and try to find the problem before revealing the annotations, because this error is subtle enough that it often survives a re-reading of the algebra: the integration itself is flawless.
The fault is one sign in the radius. The axis is one unit to the left of the origin, so the distance from a strip at x is x plus one, not x minus one. Writing the radius as x minus the axis coordinate, and then simplifying, produces the correct expression automatically, since subtracting a negative one adds one.
The third annotation shows how to catch such things without checking the algebra at all. The volume about the vertical axis is fourteen thirds of pi. Moving the axis away can only make the solid bigger, so any shifted answer smaller than that is wrong before you look at a single step. Sanity checks of that kind cost seconds and save whole questions.
Section
Part 5
Concept
Nothing changes except the strip. It now starts on the lower boundary instead of on the horizontal axis, so its length is the difference of the two functions.
\[ V = \int_a^b 2\pi x\left(f(x) - g(x)\right)dx \]
The radius is untouched: it depends on where the strip is, not on how long it is. Only the height term takes the correction.
the two lengths — A shell has one distance measured across to the axis, the radius, and one length measured along the strip, the height. The second is a difference whenever the region does not reach the axis line of the strip.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 576 — Example 6.16, adjusting the f(x) term
This extension is smaller than it looks, and the reason is worth saying explicitly: the radius depends on where the strip is, and the height depends on how long it is, so changing the region's lower boundary can only affect the height.
So the recipe is unchanged except in one factor: measure the strip from the boundary it starts on to the boundary it ends on, and use that difference as the height. It is the same adjustment you made in Section 6.1 when you moved from areas under a curve to areas between two curves.
The term on the slide is the one to hold on to. A shell always involves two lengths, measured in perpendicular directions: one across to the axis, and one along the strip. Confusing them is the other main error in this section, and naming them separately every time is the cure.
Worked example
The region is under the first curve and above the second, over the interval from 1 to 4, revolved about the vertical axis.
\[ f(x) = \sqrt x, \quad g(x) = \frac1x, \quad 1 \le x \le 4 \]
Figure (svg): The curves y equals root x and y equals one over x between x equals 1 and x equals 4, with the region between them shaded and mirrored across the y-axis, and a highlighted strip at x equals 2.5 running from the lower curve to the upper one.
Name the radius and the height
Why: The axis is the vertical one, so the radius is x; the height is the difference of the curves.
\[ r(x) = x, \qquad h(x) = \sqrt x - \frac1x \]
Write the integrand and expand
Why: Multiplying by the radius clears the fraction.
\[ 2\pi x\left(\sqrt x - \frac1x\right) = 2\pi\left(x^{3/2} - 1\right) \]
Integrate
Why: Power rule on the three-halves power, then on the constant.
\[ V = 2\pi\left[\frac{2x^{5/2}}{5} - x\right]_1^4 \]
Evaluate at the ends
Why: Four to the five halves is 32, so the bracket is 64 fifths minus 4, less 2 fifths minus 1.
\[ V = 2\pi\left(\frac{44}{5} + \frac35\right) = \frac{94\pi}{5} \approx 59.06 \]
Check by differentiating back
Why: The antiderivative must return the integrand, and the height must be positive throughout, which it is since the square root exceeds the reciprocal for every x at least 1.
\[ \frac{d}{dx}\left(\frac{2x^{5/2}}{5} - x\right) = x^{3/2} - 1 \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, pp. 576-577 — Example 6.16
Notice what the multiplication by the radius does here, because it is the same gift as in the exponential example. The height contains a reciprocal, but x times a reciprocal of x is one, so the integrand has no fraction left in it at all: just a three-halves power and a constant.
The evaluation is the fiddliest arithmetic in the lesson, so take it slowly. Four to the five halves is thirty-two, so the bracket at the upper limit is sixty-four fifths minus four. At the lower limit it is two fifths minus one, which is negative, and subtracting it adds three fifths.
The check here is not a second method but a differentiation. Differentiate the antiderivative and you must get the integrand back, which catches a mis-applied power rule immediately. It is also worth confirming that the height really is positive across the interval, because if the two curves crossed inside it, the integral would silently subtract one part of the solid from another.
Trap
The lower boundary gets forgotten:
\[ V = \int_1^4 2\pi x\sqrt x\,dx \]
\[ = \frac{124\pi}{5} \approx 77.9 \]
Wrong. That solid has no hollow below.
The strip stops at the lower curve. Subtract it, then multiply by the radius.
\[ h = \sqrt x - \frac1x \]
\[ V = \frac{94\pi}{5} \approx 59.06 \]
The wrong answer is always the larger one, because it fills in a region that is not part of the solid. A quick glance at the sketch, and at which boundary the strip rests on, prevents it.
This one hides well, because the resulting integral is perfectly respectable and gives a clean answer. Nothing about the arithmetic looks wrong; it is the solid that is wrong.
What the mistake computes is the volume of the solid you would get if the region reached all the way down to the horizontal axis. That solid contains the real one, so the wrong answer is always too big, which is a check you can apply in your head.
The cure is in the sketch. Draw the strip, and draw it starting on the boundary it actually starts on. If you never draw a strip, there is nothing to stop your pen from writing the top curve alone, and on a longer problem you may not notice for several lines.
Worked example
\[ f(x) = x, \quad g(x) = x^2, \quad 0 \le x \le 1, \quad\text{about the } y\text{-axis} \]
Name the height
Why: On this interval the line is above the parabola.
\[ h(x) = x - x^2 \]
Write the integrand
Why: Radius x times that difference.
\[ 2\pi x\left(x - x^2\right) = 2\pi\left(x^2 - x^3\right) \]
Integrate and evaluate
Why: A third minus a quarter is a twelfth.
\[ V = 2\pi\left[\frac{x^3}{3} - \frac{x^4}{4}\right]_0^1 = 2\pi\cdot\frac1{12} = \frac{\pi}{6} \]
Check with washers
Why: Horizontal slices run from the line out to the parabola, so the outer radius is the square root of y and the inner radius is y.
\[ \pi\int_0^1\left(y - y^2\right)dy = \pi\left(\frac12 - \frac13\right) = \frac{\pi}{6} \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 577 — Checkpoint 6.16
On the interval from zero to one the line lies above the parabola, so the height is the line minus the parabola. Check that ordering on a sketch rather than by instinct: outside this interval the parabola is the upper one, and the ordering would flip.
The integrand is a square minus a cube, and the evaluation is a third minus a quarter, which is a twelfth. The volume is pi over six, which is small because the region is small and close to the axis.
The washer check shows how neatly the two methods swap roles here. Horizontal slices run from the line out to the parabola, so the outer radius is the square root of y and the inner one is y, and squaring them gives y minus y squared. Both routes are easy for this region, which is exactly why it is a good one to practise the comparison on.
Picture it
Figure (svg): Two panels of the same region between y equals x and y equals x squared. On the left a horizontal slice runs from the parabola to the line, labelled washer. On the right a vertical strip stands on the parabola, labelled shell.
The region of Checkpoint 6.16, cut both ways. On the left the slice crosses the axis, so both boundaries must be rewritten with y as the variable. On the right the strip lies along the axis and both boundaries stay as they are.
Whichever way you cut, notice that the strip's length is always measured along the strip, and the radius always across it to the axis.
Two panels, one region, one axis: the only difference is the direction of the cut. Look at the highlighted pieces and say which one is a washer and which is a shell before reading the captions.
On the left the slice crosses the axis, so its ends are positions, and positions at a given height mean both boundaries have to be rewritten with y as the variable. On the right the strip runs alongside the axis, so its ends are heights, and heights at a given x are what the boundaries already give you.
Keep this pair of pictures as the mental image for the whole lesson. Perpendicular slicing asks position questions; parallel slicing asks height questions. Which one is easier is decided entirely by how the boundary happens to be written.
Concept
There is no rule that always wins. The question to ask is which integral you would rather evaluate, and there are three things to look at before deciding.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 577 — Figure 6.34 and the discussion of which method to use
There is no method that always wins, and the textbook is honest about it: the choice usually comes down to which integral you would rather evaluate. The three checks on this slide are the ones to run before committing.
The first is about the form of the boundary and is the one you have spent this lesson on. The second is about splitting: if the region's top boundary changes part way across, discs need two integrals, and a strip running the other way may need only one. The third is the case the book raises in Example 6.17, where a horizontal strip has both ends on the same curve, so shells need two branches of it and are best abandoned.
One more piece of practical advice: make the decision on a sketch and take thirty seconds over it. Choosing the awkward method rarely makes a problem impossible, but it routinely doubles the work and the number of chances to slip.
Comparison
Comparison matrix
| about the x-axis | disc | washer | shell |
|---|---|---|---|
| strip direction | vertical | vertical | horizontal |
| integrate in | x | x | y |
| radius squared? | yes | yes | no |
| cavity in the middle | none | yes | with or without |
| interval partitioned | [a, b] on the x-axis | [a, b] on the x-axis | [c, d] on the y-axis |
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 577 — Figure 6.34
Fill in every blank before revealing, and do it from the pictures rather than from memory. The table is for revolution about the horizontal axis; the book asks you to build the vertical-axis version yourself, and that is a good exercise for later.
Two rows carry the most information. The strip direction row says that discs and washers slice perpendicular to the axis while shells run parallel to it, and the row underneath follows from it: the variable of integration is always the one that runs across the axis.
The squared row is the one that shows up in marking. Discs and washers square their radii because their cross-sections are circles; shells never square anything, because their contribution is an unrolled plate.
Worked example
Revolve the triangle bounded by the two lines below and the horizontal axis about that horizontal axis, and set the volume up both ways.
\[ y = x, \quad y = 2 - x, \quad y = 0 \]
Figure (svg): A triangle with vertices at the origin, at 2 on the x-axis and at the point 1 comma 1, bounded by the lines y equals x and y equals 2 minus x; a horizontal highlighted strip is drawn at height 0.45 and a dashed vertical line marks x equals 1.
Try discs first
Why: A vertical slice's top boundary changes at x equal to 1, so the disc set-up needs two integrals.
\[ V = \int_0^1\pi x^2dx + \int_1^2\pi(2-x)^2dx \]
Try shells
Why: A horizontal strip runs from the first line to the second in one piece, at height y, for y from 0 to 1.
\[ V = \int_0^1 2\pi y\left[(2-y) - y\right]dy = \int_0^1 2\pi y\left(2-2y\right)dy \]
Evaluate the shell integral
Why: Expand and use the power rule.
\[ V = 4\pi\left[\frac{y^2}{2} - \frac{y^3}{3}\right]_0^1 = 4\pi\cdot\frac16 = \frac{2\pi}{3} \]
Check with the disc integrals
Why: Each cone has volume one third of pi, and the solid is two cones base to base. Both routes are easy here, but the shell route is one integral instead of two and needs no simplification, so it is the better choice.
\[ \frac{\pi}{3} + \frac{\pi}{3} = \frac{2\pi}{3} \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 578 — Example 6.17a
This is a choosing problem, so read both set-ups before evaluating either. The region is a triangle standing on the horizontal axis with its apex at the point one comma one, and the axis of revolution is the horizontal axis it stands on.
The disc route slices vertically, and the top boundary changes at x equal to one, so it needs two integrals. The shell route slices horizontally, and every horizontal strip runs from one line to the other in one piece, so it needs one. Neither is hard, but the shell version is shorter and needs no simplification, which is the book's verdict too.
The check is geometric and satisfying. The solid is two right circular cones, base to base, each with radius one and height one, so its volume is twice a third of pi. It matches the integrals, and it is the kind of check worth hunting for whenever the region is made of straight lines.
Worked example
The same question for a region bounded above by a parabola and below by the horizontal axis, revolved about that axis.
\[ y = 4x - x^2, \quad y = 0 \]
Figure (svg): The arch y equals 4x minus x squared from 0 to 4 with the region under it shaded, a vertical highlighted strip at x equals 3.2, and a dashed horizontal line at height 2.5 meeting the arch at two points marked with dots.
Test the horizontal strip
Why: Both ends of a horizontal strip lie on the same parabola, so a shell set-up needs the curve solved for x on both branches.
\[ x = 2 \pm \sqrt{4-y} \]
Dismiss shells and slice vertically
Why: The solid has no cavity, so every vertical slice is a full disc of radius equal to the curve's height.
\[ V = \int_0^4 \pi\left(4x-x^2\right)^2dx \]
Expand the square
Why: Multiply out before integrating.
\[ \left(4x-x^2\right)^2 = 16x^2 - 8x^3 + x^4 \]
Integrate and evaluate
Why: Power rule on each term, at x equal to 4.
\[ V = \pi\left(\frac{1024}{3} - 512 + \frac{1024}{5}\right) = \frac{512\pi}{15} \approx 107.2 \]
Check with four midpoint discs
Why: Slices of width 1 at x equal to 0.5, 1.5, 2.5 and 3.5 give heights 1.75, 3.75, 3.75 and 1.75, and the sum of their squares times pi should be close to the exact value.
\[ \pi(1.75^2 + 3.75^2 + 3.75^2 + 1.75^2) \approx 34.25\pi \quad\text{against}\quad \tfrac{512}{15}\pi \approx 34.13\pi \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 579 — Example 6.17b
Here the choice goes the other way, and the reason is new. The region is bounded on the left and on the right by the same parabola, so a horizontal strip has both ends on that one curve and you would have to solve it for x twice, once for each branch.
The vertical slice has no such problem. The solid has no cavity down the middle, because the region rests on the axis of revolution, so every slice is a full disc whose radius is the curve's height. Expand the square first and the integration is three power-rule terms.
The check with four midpoint discs is a good habit for an answer with awkward fractions in it. Widths of one, heights of 1.75, 3.75, 3.75 and 1.75, squares added and multiplied by pi, give about 34.25 pi against the exact 34.13 pi. Close enough to confirm the set-up, and the slight overestimate is what you expect from midpoints on a curve of this shape.
Step zero
\[ y = 2 - x^2 \quad\text{and}\quad y = x^2, \quad\text{about the } x\text{-axis} \]
Discussion prompt
Sketch it in your head: the curves meet where x is 1 and where x is minus 1. Which method takes one integral here, and what is the set-up?
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 579 — Checkpoint 6.17
Answer before revealing, and answer the choosing question first: it is worth more marks than the arithmetic in most exams.
The two parabolas cross where x is one and where x is minus one, one opening up and one opening down, enclosing a lens-shaped region symmetric about the vertical axis. Revolved about the horizontal axis, vertical slices are washers running from the lower curve to the upper one over the whole interval, so one integral does it.
Shells would need horizontal strips, and the description of a strip changes at the height one, where the boundary switches from one parabola to the other, so that route takes two integrals. There is a second bonus in the washer version: the fourth-power terms cancel when you subtract the squares, leaving a quadratic and a very short evaluation.
Sorting
Sort into buckets
For each region and axis, which family of slices gives the easier single integral?
Sort all six, and for each one run the three checks from the guidelines slide: what form is the boundary in, how many integrals would each method need, and can a strip in each direction even be described in one piece.
The three shell cases are shell cases for different reasons. The cubic will not invert usefully; the exponential of a square would invert into a square root of a logarithm and give a horrible washer integrand; and the triangle needs two disc integrals but only one shell integral.
The three disc or washer cases share a reason: the region is bounded left and right by the same curve, or it is exactly the standard set-up of a function of x revolved about the horizontal axis. In those cases horizontal strips need two branches of a curve, and vertical slices need nothing at all.
Worked example
Revolve the right half of a disc of radius r, centred at the origin, about the vertical axis. The solid is the whole ball.
Figure (svg): A circle of radius r centred at the origin with the right half shaded, a highlighted vertical strip at x equals 0.6 running from the lower semicircle to the upper one, and its mirror strip on the left.
Name the radius and the height
Why: A strip at x runs from the lower semicircle up to the upper one, so its length is twice the semicircle's height.
\[ r(x) = x, \qquad h(x) = 2\sqrt{r^2 - x^2} \]
Write the integral
Why: Radius times height, for x from 0 to r.
\[ V = \int_0^r 2\pi x\left(2\sqrt{r^2-x^2}\right)dx = 4\pi\int_0^r x\sqrt{r^2-x^2}\,dx \]
Substitute
Why: Let u be the expression under the root; then x dx is minus a half du, and the limits swap.
\[ u = r^2 - x^2, \quad du = -2x\,dx \]
Integrate in u
Why: The limits become u equal to r squared at the bottom and u equal to 0 at the top, and the minus sign turns them back around.
\[ V = 2\pi\int_0^{r^2}\sqrt u\,du = 2\pi\left[\frac{2u^{3/2}}{3}\right]_0^{r^2} \]
Evaluate
Why: The three-halves power of r squared is r cubed.
\[ V = \frac{4\pi r^3}{3} \]
Check against the known formula
Why: This is the classical volume of a sphere; with r equal to 1 both the integral and the formula give 4.18879.
\[ \frac{4\pi(1)^3}{3} = 4.18879\,\ldots \]
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 581 — Exercise 158
It is worth deriving a formula you have known since school, because a derivation you can produce is a formula you cannot misremember.
The set-up is the only step that needs thought. The strip is a vertical chord of the circle, so it runs from the lower semicircle up to the upper one, and its length is twice the semicircle's height, not once. Forgetting the factor of two here gives exactly half the right answer, which is the classic slip in this exercise.
The integration is a substitution: the expression under the square root has derivative minus twice x, and the radius has supplied an x, so the substitution closes. This is the same phenomenon you met with the exponential of a square, and it is one more reason shells suit regions built out of circles.
Real world
Figure (svg): A circle of radius 2 centred at the origin, the vertical line x equals 4 dashed, and a faint mirror circle centred at x equals 8; a highlighted vertical strip inside the left circle has an arrow measuring its distance to the line.
Discussion prompt
A circle of radius 2, centred at the origin, is revolved about the vertical line x = 4. Set up the shell integral and evaluate it, using the area of a semicircle for one piece and symmetry for the other.
OpenStax Calculus Volume 1, §6.3 Volumes of Revolution: Cylindrical Shells §6.3, p. 581 — Exercise 162
Set this up yourself before revealing. The circle is centred at the origin with radius two and the axis is the line four, so the circle never reaches the axis and the solid really is a doughnut with a hole in the middle.
The strip is a vertical chord again, of length twice the square root, and the radius is four minus x, which is positive over the whole disc since x never exceeds two. Multiply out and you get two integrals, and this is the point of the exercise: neither of them needs an antiderivative. The first is the area of a semicircle of radius two, which is twice pi, and the second has an odd integrand on a symmetric interval, so it is zero.
The answer, thirty-two pi squared, is about 315.8, and the cancellation is telling you something physical. The extra volume swept by the far half of the disc exactly balances the shortfall from the near half, so the doughnut's volume is its cross-sectional area times the distance travelled by its centre. That is Pappus's theorem, and you met a special case of it two parts ago when you shifted the axis.
Section
Part 6
Pattern
Figure (svg): A six-step flow diagram: sketch the region and the axis; draw a strip parallel to the axis; radius equals distance to the axis; height equals the far boundary minus the near one; integrate two pi times radius times height; check against the other method.
Work through the six steps in order on every shell problem until the order is automatic. Notice that the first two are drawings and the last is a check, so only three of the six involve any symbols.
The order is not arbitrary. The strip has to be drawn before the radius can be named, because the radius is measured from the strip; and the radius has to be settled before the height, because the direction of the strip decides which variable the height is a function of.
The last step is the one most often skipped and the cheapest to keep. Set the problem up the other way, or compare against the unshifted volume, or estimate with three or four shells. Any of these catches the two errors this lesson keeps warning about: a squared radius and a radius written backwards.
Ranking
Put in order
Order the steps of a complete shell solution.
Why: The drawing comes first because the radius and the height are both read from it, and they must be read in that order: the radius fixes which variable you are integrating in, and the height is then measured along the strip you have already drawn.
Drag the six into order before checking. The only ordering that is genuinely debatable is where the check belongs, and the answer is at the end, because a check is only worth anything once there is an answer to check.
If you put the height before the radius, nothing terrible happens, but you lose the reason the order exists: the radius fixes the direction of the strip and therefore the variable of integration, and the height must then be expressed in that same variable. Deciding the height first is how people end up with a radius in x and a height in y in the same integral.
Check
Check your understanding
The region under y = 1 − x² for x from 0 to 1 is revolved about the y-axis. Which integral gives its volume?
Answer: A
Why: The radius of a strip at x is x, its height is the curve, and the integrand is twice pi times the product. The volume is pi over 2.
Two of these options are set-ups for real solids and two are simply wrong, so identify which is which before choosing. The question is only about the region revolved about the vertical axis.
The correct integrand is radius times height times twice pi, with the radius appearing once. One distractor squares the radius, which is the trap from Part 2. One drops the radius entirely, leaving twice pi times the area. One is the disc set-up for the same region revolved about the horizontal axis, which is a different solid with a different volume.
If you chose the disc option, reread the question rather than the formula: the axis of revolution decides everything, and it is the first thing to underline in a volume problem.
Check
Check your understanding
The region under y = x from x = 0 to x = 1 is revolved about the line x = 3. What is the radius of a shell at x?
Answer: B
Why: The region lies to the left of the axis, so the distance from the strip at x to the line is the line's coordinate minus x. The volume works out as 7π/3.
This is the sign question, and the safest way to answer it is to picture the strip and the line and ask which is further right. Here the line at three is to the right of every strip, so the distance is three minus the strip's coordinate.
The option x plus three is the answer for an axis three units to the left, at minus three, and it is a good idea to know which is which rather than to guess. The option x minus three is negative everywhere on the region and would give a negative volume. The constant three is the distance from the axis to the origin, not to a strip, and it makes the radius the same for every shell, which the picture flatly contradicts.
Check
Check your understanding
The region bounded by y = √x, y = 0 and x = 4 is revolved about the x-axis. Which shell integral gives its volume?
Answer: A
Why: Strips are horizontal, so the radius is y and the strip runs from the curve, at x equal to y squared, out to the line x = 4: its length is 4 − y². The volume is 8π, which the disc method also gives.
Everything turns on which part of the horizontal line is inside the region. The strip at height y runs from the curve, at x equal to y squared, out to the vertical line at four, so its length is four minus y squared.
The most tempting wrong answer measures from the vertical axis to the curve instead, giving y squared. That is the part of the horizontal line that lies outside the region, to the left of it, so it computes the volume of a different solid altogether.
Check the winner against discs if you like: slicing vertically gives discs of radius the square root of x, and the integral of pi times x from zero to four is eight pi, exactly what the shell integral gives.
Explain it to yourself
Discussion prompt
In two or three sentences, explain to someone who has just learned washers why the shell method never needs the boundary solved for the other variable, and name one region where that matters.
Write your sentences out before revealing. Being able to say this clearly is the difference between knowing a third formula and understanding why the section exists.
The core of a good answer is the direction in which each method measures. A shell's strip lies parallel to the axis, so the length you measure along it is a value of the function, which the boundary hands you directly. A washer's slice is perpendicular to the axis, so what you measure are positions at a fixed height, and those are values of the inverse function.
For the example, any region whose boundary cannot be solved for the other variable will do, and the cubic from Part 3 is the cleanest: shells turn it into a quartic and washers stall completely. The exponential of a square is an even more striking case, because there the shell method is not just easier, it makes the integral elementary.
Exit ticket
\[ y = \sqrt x, \quad y = 0, \quad 0 \le x \le 4, \quad\text{about } x = -1 \]
Discussion prompt
Set up the volume by shells, evaluate it, and check your answer against the volume of the same region about the y-axis.
Do the whole thing before revealing, including the check. The region is the one under a square root out to x equal to four, and the axis is one unit to the left of the origin.
The radius is x plus one, positive everywhere on the region, and the height is the square root as given. Multiplying out gives a three-halves power plus a one-half power, and both are power-rule integrals. At x equal to four the two terms are sixty-four fifths and sixteen thirds.
The check is the shift rule from Part 4. About the vertical axis the volume is 128 fifths of pi; the region's area is sixteen thirds; and moving the axis one unit adds twice pi times that area. The two routes give the same 544 fifteenths of pi, about 113.9. If your answer disagreed, the most likely culprit is the expansion of the product, so redo that line first.
Recap
| method | strip | integrand | when it is the easy one |
|---|---|---|---|
| disc | perpendicular to the axis | pi times radius squared | solid with no cavity, boundary in the right variable |
| washer | perpendicular to the axis | pi times the difference of squares | cavity, and the boundary inverts easily |
| shell | parallel to the axis | twice pi times radius times height | the boundary will not invert, or one integral beats two |
\[ V = \int_a^b 2\pi\,r(x)\,h(x)\,dx \qquad V = \int_c^d 2\pi\,r(y)\,h(y)\,dy \]
The radius is a positive distance to the axis, the height is measured along the strip, and moving the axis further away by k adds twice pi times k times the region's area.
Next, Section 6.4 keeps the revolving but measures the curve itself: arc length, and the area of the surface a curve sweeps out.
Stewart, Calculus: Early Transcendentals 8e, §6.3 Volumes by Cylindrical Shells §6.3, pp. 449-454 — the same material in Stewart
Three methods, one idea. Slice the solid into pieces so thin that each piece has a volume you can write down, add them up, and take a limit. Discs and washers slice across the axis and square a radius; shells slice along the axis and multiply three lengths together.
The two habits to carry out of this lesson are choosing and checking. Choose by looking at the form of the boundary, at how many integrals each direction needs, and at whether a strip in each direction can be described in one piece. Check by setting the problem up the other way, or by comparing a shifted volume with its unshifted version, or by estimating with a few shells.
Two errors account for most lost marks here, and both are preventable from the sketch: squaring the radius, which the unrolled-plate picture rules out, and writing the radius backwards, which the sign of your answer exposes. Next, Section 6.4 keeps a curve spinning but measures the curve itself, its length and the area of the surface it sweeps.
Want this taught 1-on-1? Alexander tutors Calculus I — $55/session, free consultation.