Intuition What this deep-dive is for
The parent note taught you the one formula d V = 2 π r h d x and showed three clean examples. But real problems hide in the corners : the axis is on the wrong side, the region straddles the axis, you're forced to switch to d y , a curve is given as x in terms of y , or the numbers come dressed as a word problem.
This page builds a matrix of every case class and then walks one example through each cell — so that when an exam hands you a weird one, you have already met its twin.
c — the axis you rotate around
Throughout this page, c is just a number naming where the axis of rotation sits . If you spin the region around the vertical line "x = 5 ," then c = 5 ; around "x = − 1 ," then c = − 1 ; around the y -axis itself, that line is x = 0 , so c = 0 . For a horizontal axis "y = 3 " we'd have c = 3 , and the x -axis is y = 0 , so c = 0 again.
You read c straight off the problem — it's never something to solve for. Every radius on this page is measured from the axis at c to the strip , which is why c shows up inside ∣ x − c ∣ or ∣ y − c ∣ .
f ( x ) — "the upper curve"
When I write f ( x ) I do not mean a mysterious new object. It is simply shorthand for "the height of the top boundary curve at horizontal position x ." If the region's top edge is the curve y = 4 − x 2 , then f ( x ) = 4 − x 2 . Likewise g ( y ) later will mean "the right boundary curve written as an x -value at height y ." Read them as "top curve" and "side curve," nothing more.
Before anything, meet the three pieces of every shell. In the figure below, the radius r (cyan, along the bottom) is the horizontal distance from the axis to the strip; the height h (amber, on the right) is how tall the cyan strip is; and the thickness — which for a vertical strip is exactly the symbol d x (a horizontal axis would make it d y instead) — is how wide the strip is (white, at the top). The amber vertical line marked "axis" is the y -axis (c = 0 ) — notice the radius arrow starts exactly there. Nothing on this page uses a symbol beyond these three plus ∫ , which just means "add up all the strips."
Figure s01 — one cylindrical shell dissected: cyan radius r = x from the amber axis, amber height h , white thickness labelled d x .
Every shell-method problem is one (or a blend) of these cells. The whole difficulty is always the radius and the height — get those right and the integral is routine. (Recall c = the number naming the axis, defined just above; f ( x ) = the top curve's height.)
Cell
What changes
Radius r
Height h
Thickness
Example
A
Axis = y -axis (c = 0 ), region to its right
x
f ( x )
d x
Ex 1
B
Axis = vertical line x = c , region left of it
c − x
top − bottom
d x
Ex 2
C
Axis = vertical line x = c , region right of it
x − c
top − bottom
d x
Ex 3
D
Region straddles the axis x = c (radius sign flips)
$
x-c
$
f ( x )
E
Axis = horizontal line y = c → must use d y
$
y-c
$
right − left
F
Curve given as x = g ( y ) (no inversion needed)
y or $
y-c
$
g ( y )
G
Degenerate : axis touches the region (radius = 0 at an edge)
x
f ( x )
d x
Ex 7
H
Word problem with units
x (distance in cm)
f ( x ) (length in cm)
d x
Ex 8
I
Exam twist : region between two curves + shifted axis
$
x-c
$
top − bottom
Mnemonic The two questions that solve every cell
For every shell, ask only:
"How far is my strip from the axis (at c )?" → that is r .
"How long is my strip?" → that is h .
Everything else is bookkeeping.
Worked example Example 1 (Cell A) — region to the right of the axis
Region under y = 4 − x 2 , above y = 0 , for 0 ≤ x ≤ 2 , rotated about the y -axis (c = 0 ).
Forecast: the region is a "hill" from x = 0 to x = 2 . Guess: is the volume closer to 8 π or 30 π ? Write your guess down.
Step 1 — strips parallel to the axis. The axis is vertical, so I use vertical strips (thickness d x ).
Why this step? A strip parallel to the axis sweeps out a clean tube; a perpendicular strip would give disks, which here would force me to solve x = 4 − y .
Step 2 — radius. r = x (here c = 0 , so ∣ x − c ∣ = ∣ x − 0∣ = x ).
Why? Distance from the y -axis (the line x = 0 ) to a strip at position x is just x .
Step 3 — height. h = f ( x ) = 4 − x 2 (recall f ( x ) = the top curve's height).
Why? The strip runs from the floor y = 0 up to the curve y = 4 − x 2 ; height is top minus bottom.
Step 4 — why these limits, then integrate. The region is defined to run from x = 0 (the left edge, the axis) to x = 2 (the right edge, given in the problem), so x sweeps 0 → 2 and those are the integral bounds:
V = ∫ 0 2 2 π x ( 4 − x 2 ) d x = 2 π ∫ 0 2 ( 4 x − x 3 ) d x = 2 π [ 2 x 2 − 4 x 4 ] 0 2 .
= 2 π ( 8 − 4 ) = 8 π .
Verify: at x = 1 the shell has r = 1 , h = 3 , so d V / d x = 2 π ( 3 ) = 6 π — a plausible mid-value; integrated over width 2 that lands near 8 π . ✓ Units: if x , y are cm, volume is cm³.
The single subtle thing here is the sign inside the radius . Look at the figure: the cyan strip sits at some x between 0 and 2 , and the same strip is measured to two different axes. To the right axis (c = 5 ) the distance is 5 − x ; to the left axis (c = − 1 ) it is x + 1 . Both cyan arrows come out positive — that is the whole lesson of this pair.
Figure s02 — the same cyan region measured to two amber axes; cyan arrows show r = 5 − x (right axis) and r = x + 1 (left axis), both positive.
Worked example Example 2 (Cell B) — region LEFT of the axis
Region under y = x 2 , 0 ≤ x ≤ 2 , rotated about the vertical line x = 5 (so c = 5 ).
Forecast: the axis is far to the right of the region. Bigger radius everywhere ⇒ do you expect the volume to be much larger than the 8 π we got about x = 0 ? Guess a number.
Step 1 — radius. The axis x = 5 is to the right of every strip (0 ≤ x ≤ 2 ), so the strip is at distance r = ∣ x − 5∣ = 5 − x .
Why? 5 − x is positive on [ 0 , 2 ] (it ranges from 5 down to 3 ). Writing x − 5 would give a negative number — wrong sign for a distance.
Step 2 — height. h = x 2 (unchanged; height doesn't care about the axis).
Step 3 — why these limits, then integrate. The region's own edges are x = 0 and x = 2 (both given in the statement), so those are the bounds:
V = ∫ 0 2 2 π ( 5 − x ) x 2 d x = 2 π ∫ 0 2 ( 5 x 2 − x 3 ) d x = 2 π [ 3 5 x 3 − 4 x 4 ] 0 2 .
= 2 π ( 3 40 − 4 ) = 2 π ⋅ 3 28 = 3 56 π .
Verify: 3 56 π ≈ 58.6 , indeed far bigger than 8 π ≈ 25.1 — matches the "far axis, big radius" forecast. ✓
Worked example Example 3 (Cell C) — region RIGHT of the axis
Same region y = x 2 , 0 ≤ x ≤ 2 , but rotated about the vertical line x = − 1 (so c = − 1 ).
Forecast: now the axis sits to the left of the region. Bigger or smaller than the x = 5 case? (Hint: x = − 1 is only 1 to 3 units away; x = 5 was 3 to 5 away.)
Step 1 — radius. The axis x = − 1 is to the left of every strip, so r = ∣ x − ( − 1 ) ∣ = x + 1 .
Why? On [ 0 , 2 ] , x + 1 runs from 1 to 3 , positive throughout. The strip is farther from a left axis the larger x is.
Step 2 — height. h = x 2 .
Step 3 — why these limits, then integrate. Same region, same edges x = 0 to x = 2 :
V = ∫ 0 2 2 π ( x + 1 ) x 2 d x = 2 π ∫ 0 2 ( x 3 + x 2 ) d x = 2 π [ 4 x 4 + 3 x 3 ] 0 2 .
= 2 π ( 4 + 3 8 ) = 2 π ⋅ 3 20 = 3 40 π .
Verify: 3 40 π ≈ 41.9 — smaller than the x = 5 case (58.6 ), because this axis is closer. Sign of radius came out positive as required. ✓
The danger here is a radius that changes sign as you cross the axis. The figure shows a region running from x = 1 to x = 4 split by the axis x = 2 : strips to the left of the axis have radius 2 − x , strips to the right have x − 2 , and the single formula ∣ x − 2∣ captures both. Watch the cyan radius shrink to zero exactly at the axis and grow again on the far side.
Figure s04 — rectangle straddling the amber axis x = 2 ; cyan arrows show radius 2 − x on the left, x − 2 on the right, both vanishing at the axis.
Worked example Example 4 (Cell D) — region crossing the line
x = 2
Region under y = 3 , above y = 0 , for 1 ≤ x ≤ 4 (a rectangle), rotated about the vertical line x = 2 (so c = 2 ). The axis cuts through the region.
Forecast: part of the rectangle is left of x = 2 (width 1 ), part is right (width 2 ). Both parts sweep real, positive volume. Will they add, or partly cancel? Guess.
Step 1 — spot the sign flip. For strips left of the axis (1 ≤ x < 2 ) the signed quantity x − 2 is negative ; for strips right of it (2 < x ≤ 4 ) it is positive. A distance can't be negative, so the true radius is r = ∣ x − 2∣ .
Why this step? If you used x − 2 raw, the left part would contribute negative volume and wrongly cancel the right part.
Step 2 — split at the axis. Break the region where the radius formula changes, i.e. at x = 2 , and drop the absolute value on each piece:
V = ∫ 1 2 2 π ( 2 − x ) ⋅ 3 d x + ∫ 2 4 2 π ( x − 2 ) ⋅ 3 d x .
Why? On [ 1 , 2 ] , ∣ x − 2∣ = 2 − x ; on [ 2 , 4 ] , ∣ x − 2∣ = x − 2 . The bounds 1 , 2 , 4 come from the region's edges and the axis crossing.
Step 3 — integrate each piece. Height is h = 3 throughout (top y = 3 minus bottom y = 0 ).
∫ 1 2 2 π ( 2 − x ) 3 d x = 6 π [ 2 x − 2 x 2 ] 1 2 = 6 π ( ( 4 − 2 ) − ( 2 − 2 1 ) ) = 6 π ⋅ 2 1 = 3 π .
∫ 2 4 2 π ( x − 2 ) 3 d x = 6 π [ 2 x 2 − 2 x ] 2 4 = 6 π ( ( 8 − 8 ) − ( 2 − 4 ) ) = 6 π ⋅ 2 = 12 π .
V = 3 π + 12 π = 15 π .
Verify: each half of the rectangle sweeps out a solid cylinder (no hole, because the strip reaches the axis), so use the disk method to check. Left half → solid cylinder of radius 1 , height 3 : π ( 1 ) 2 ( 3 ) = 3 π ✓. Right half → solid cylinder of radius 2 , height 3 : π ( 2 ) 2 ( 3 ) = 12 π ✓. Total 15 π . If you had forgotten the absolute value, ∫ 1 4 2 π ( x − 2 ) 3 d x = 6 π [ 2 x 2 − 2 x ] 1 4 = 6 π ( 0 − ( − 2 3 )) = 9 π — wrong, confirming the split was essential. ✓
Common mistake Straddling the axis with a signed radius
Why it feels right: the boxed formula ∫ 2 π x f ( x ) d x literally has "x ."
The fix: that formula silently assumes the strip stays on one side of the axis. When the region crosses x = c , use r = ∣ x − c ∣ and split the integral at x = c (or use symmetry if the region is symmetric, as in the y -axis case).
When the axis is horizontal , strips must be horizontal too (parallel to the axis), so their thickness is d y and their length is measured left-to-right . In the figure, the amber length is now a width (x right − x left ) and the cyan radius is measured vertically from the axis y = c up to the strip. Here the axis is drawn as the line y = 1 (so c = 1 ), not the x -axis — watch the radius arrow start at y = 1 , giving r = ∣ y − 1∣ .
Figure s03 — horizontal strip (thickness d y ) with cyan vertical radius r = ∣ y − 1∣ from the amber axis y = 1 and amber width h = right − left.
Worked example Example 5 (Cell E) — rotation about the shifted line
y = 1 , using d y
Region bounded by x = y , x = 0 , and y = 3 (a triangle with vertices ( 0 , 0 ) , ( 0 , 3 ) , ( 3 , 3 ) ), rotated about the horizontal line y = 1 (so c = 1 ).
Forecast: the axis is inside the vertical extent of the triangle (y runs 0 to 3 , and 1 is between them). So near y = 1 the radius is tiny. Bigger or smaller than rotating about y = 0 (which gave 18 π )? Guess.
Step 1 — strips horizontal. Axis is horizontal ⇒ thickness d y . The region's vertical extent is y = 0 (bottom vertex) to y = 3 (given top edge), so those are the bounds.
Why these limits? y = 0 is the lowest point of the triangle, y = 3 its top edge; the strips sweep the region as y goes 0 → 3 .
Step 2 — radius. r = ∣ y − 1∣ (vertical distance from the axis y = 1 to the strip).
Why the absolute value? Below the axis (0 ≤ y < 1 ) the strip is 1 − y away; above it (1 < y ≤ 3 ) it is y − 1 away. The region crosses the axis, so we split at y = 1 .
Step 3 — height (a width!). At height y , the strip runs from x = 0 on the left to x = y on the right, so h = y − 0 = y .
Why? Height is right minus left when strips are horizontal — not a y -value of a curve.
Step 4 — split at the axis and integrate.
V = ∫ 0 1 2 π ( 1 − y ) y d y + ∫ 1 3 2 π ( y − 1 ) y d y .
∫ 0 1 2 π ( y − y 2 ) d y = 2 π [ 2 y 2 − 3 y 3 ] 0 1 = 2 π ( 2 1 − 3 1 ) = 2 π ⋅ 6 1 = 3 π .
∫ 1 3 2 π ( y 2 − y ) d y = 2 π [ 3 y 3 − 2 y 2 ] 1 3 = 2 π ( ( 9 − 2 9 ) − ( 3 1 − 2 1 ) ) = 2 π ( 2 9 + 6 1 ) = 2 π ⋅ 3 14 = 3 28 π .
V = 3 π + 3 28 π = 3 29 π .
Verify: 3 29 π ≈ 30.4 , larger than the y = 0 answer 18 π ≈ 56.5 ? No — 18 π ≈ 56.5 is bigger, which is right: an axis through the region gives smaller radii on average than an axis 1 unit below it. Forecast confirmed (smaller than 18 π ). ✓
Worked example Example 6 (Cell F) — no inversion needed, rotate about
x -axis
Region bounded by x = y 2 , x = 0 , and y = 2 (with y ≥ 0 ), rotated about the x -axis (c = 0 ).
Forecast: the curve x = y 2 opens rightward. Guess whether V beats 10 π .
Step 1 — horizontal strips. Axis horizontal ⇒ d y . The region runs vertically from y = 0 (where x = y 2 meets x = 0 ) up to y = 2 (the given cap), so bounds are 0 → 2 .
Why these limits? x = y 2 and x = 0 intersect at y = 0 ; the line y = 2 closes the top.
Step 2 — radius. r = ∣ y − 0∣ = y (distance from the x -axis up to the strip; y ≥ 0 so no sign issue).
Step 3 — height (width). From x = 0 to x = y 2 , so h = g ( y ) = y 2 (here g ( y ) = the right-boundary curve as an x -value).
Why this cell is easy: the curve is already x as a function of y , so no algebra to invert — exactly the situation the shell method loves.
Step 4 — integrate.
V = ∫ 0 2 2 π y ⋅ y 2 d y = 2 π ∫ 0 2 y 3 d y = 2 π [ 4 y 4 ] 0 2 = 2 π ⋅ 4 = 8 π .
Verify: at y = 2 , the strip has r = 2 , h = 4 , so d V / d y = 16 π near the top and 0 at the bottom — averaging to something well under 16 π ⋅ 2 = 32 π ; 8 π is consistent. ✓
Worked example Example 7 (Cell G) — axis touches the region
Region under y = x , y = 0 , 0 ≤ x ≤ 4 , rotated about the y -axis (c = 0 ).
Forecast: at x = 0 the strip sits on the axis, so its radius is 0 — that shell has zero volume. Does the integral still behave? Guess whether the answer is finite.
Step 1 — check the degenerate edge. At x = 0 : r = 0 , so d V = 2 π ( 0 ) h d x = 0 . No blow-up, no problem — the integrand simply vanishes there.
Why this step matters? Whenever the axis grazes the region, students panic; here the shell of radius 0 is a genuine "line," contributing nothing, and the integral stays finite.
Step 2 — radius & height. r = x , h = f ( x ) = x = x 1/2 .
Step 3 — why these limits, then integrate. The region spans x = 0 (where y = x meets y = 0 ) to x = 4 (given), so bounds 0 → 4 :
V = ∫ 0 4 2 π x ⋅ x 1/2 d x = 2 π ∫ 0 4 x 3/2 d x = 2 π [ 5 2 x 5/2 ] 0 4 .
Since 4 5/2 = ( 4 ) 5 = 2 5 = 32 :
V = 2 π ⋅ 5 2 ⋅ 32 = 5 128 π .
Verify: 5 128 π ≈ 80.4 , finite as forecast; the r = 0 endpoint caused no trouble because the integrand x 3/2 → 0 smoothly. ✓
Worked example Example 8 (Cell H) — a machined bushing
An engineer spins the region between y = 6 − x and y = 0 , for 2 ≤ x ≤ 6 (all in cm ), about the y -axis (c = 0 ) to make a tapered ring. Find its volume in cm³.
Forecast: the region is a triangle sitting between x = 2 and x = 6 , with a hole down the middle (nothing reaches the axis). Guess: tens or hundreds of cm³?
Step 1 — radius. r = x cm (distance from the y -axis to the strip).
Why? Same as Cell A, but the region starts at x = 2 , so no shell has radius below 2 — that leaves a hollow core, which is correct for a "ring."
Step 2 — height. h = f ( x ) = 6 − x cm (from floor y = 0 up to the line).
Step 3 — why these limits, then integrate. The strip's left edge is x = 2 and right edge x = 6 (both given), and note y = 6 − x hits 0 exactly at x = 6 , so the region closes there:
V = ∫ 2 6 2 π x ( 6 − x ) d x = 2 π ∫ 2 6 ( 6 x − x 2 ) d x = 2 π [ 3 x 2 − 3 x 3 ] 2 6 .
At x = 6 : 3 ( 36 ) − 3 216 = 108 − 72 = 36 . At x = 2 : 3 ( 4 ) − 3 8 = 12 − 3 8 = 3 28 .
V = 2 π ( 36 − 3 28 ) = 2 π ⋅ 3 80 = 3 160 π cm 3 .
Verify: 3 160 π ≈ 167.6 cm 3 . Units: (cm)(cm)(cm) → cm³ ✓. The empty core (x < 2 ) correctly contributes nothing. ✓
Worked example Example 9 (Cell I) — between curves, about
x = − 1
Region between y = x and y = x 2 , rotated about the vertical line x = − 1 (so c = − 1 ).
Forecast: in the parent note this region about the y -axis gave 6 π . The axis is now 1 unit farther left, so every radius grows by 1 . Bigger, obviously — but by how much? Guess.
Step 1 — find where the curves meet (the limits). Set x = x 2 ⇒ x 2 − x = 0 ⇒ x ( x − 1 ) = 0 , so x = 0 and x = 1 . These intersection points are the integral bounds — the region only exists between them.
Why this step? The two curves enclose the region exactly on [ 0 , 1 ] ; outside that they don't bound anything.
Step 2 — height (top − bottom). On ( 0 , 1 ) , test x = 2 1 : y = x gives 2 1 , y = x 2 gives 4 1 , so x > x 2 and h = x − x 2 .
Why? Two curves ⇒ height is the gap between them, top minus bottom.
Step 3 — radius about x = − 1 . r = ∣ x − ( − 1 ) ∣ = x + 1 , positive on [ 0 , 1 ] (runs from 1 to 2 ).
Why? Axis is to the left, so distance is x + 1 .
Step 4 — integrate. First expand the product carefully:
( x + 1 ) ( x − x 2 ) = x ⋅ x − x ⋅ x 2 + 1 ⋅ x − 1 ⋅ x 2 = x 2 − x 3 + x − x 2 = x − x 3 .
The two x 2 terms cancel, leaving x − x 3 . So
V = ∫ 0 1 2 π ( x − x 3 ) d x = 2 π [ 2 x 2 − 4 x 4 ] 0 1 = 2 π ( 2 1 − 4 1 ) = 2 π ⋅ 4 1 = 2 π .
Verify: 2 π = 3 ⋅ 6 π , exactly 3 × the y -axis answer. Sanity: the "+ 1 " shift added ∫ 0 1 2 π ( x − x 2 ) d x = 2 π ( 2 1 − 3 1 ) = 3 π to the original 6 π , giving 6 π + 3 π = 2 π . ✓
Recall Which cell is each situation?
Axis is x = 4 , region lies in 0 ≤ x ≤ 3 — which radius? ::: Region is left of the axis (Cell B): r = 4 − x ; here c = 4 .
Rotating about the horizontal line y = 2 — what is the strip thickness? ::: d y (Cell E); strips are horizontal, radius ∣ y − 2∣ .
Region symmetric across the y -axis, rotated about it — what safeguard? ::: Use r = ∣ x ∣ or exploit symmetry (Cell D); a signed x would wrongly cancel.
Region crosses the axis x = 2 — how do you set up the integral? ::: Use r = ∣ x − 2∣ and split at x = 2 into 2 − x (left) and x − 2 (right).
A strip sits exactly on the axis — is the integral broken? ::: No (Cell G); that shell has r = 0 , contributing 0 , integral stays finite.
Curve given as x = g ( y ) , axis horizontal — why is shell ideal? ::: No inversion needed (Cell F); integrate in y directly.
Mnemonic The universal check
Before integrating, plug the two endpoint values into your radius and confirm both come out positive . If one is negative — or if the region crosses the axis at c — split at c before you waste the integration.
shell method — the parent formula this page stress-tests.
Volume of revolution — disk and washer method — the perpendicular-slice alternative; Cell D's check used solid cylinders (the disk method).
Definite integral as a limit of Riemann sums — why summing infinitely many shells is exact.
Area between two curves — the "top − bottom" height reused in Cells D, E, I.
Choosing dx vs dy strips — decides between Cells A–D (d x ) and Cells E–F (d y ).
Arc length and surfaces of revolution — revolve the boundary instead of the region.
radius = distance x to axis at c
radius = distance y to axis at c
region crosses axis? split at c, use absolute value
region one side? plain x minus c
height = right minus left