Moment of Inertia of an Equilateral Triangle with Pivot at one Vertex

There are two ways (at least) to derive the expression for an equilateral triangle that is rotated about
one vertex, and I’ll show you both here. One way is direct: divide the triangle into horizontal strips,
write the expression for the moment of inertia of one strip, and integrate the expression to get the sum
of moment of inertia of all the strips (i.e. the entire triangle.)

The second way is very clever and is much shorter. Sometimes a bit of thinking can help us see a
shorter way to solve a problem. The only danger is that a shorter route to a solution typically involves
some logic, and if one step in the logic is faulty the entire solution ends up incorrect. Fortunately, the
logic we need for this problem is pretty simple.

And now, without further ado...

The long, direct way (which works for any situation):

Consider the thin horizontal strip in the triangle below. The strip has width “w” and height “dy”. I’ve
defined the y-axis to go from y = 0 at the top of the triangle (at the pivot) to y = H at the bottom of
the triangle. So the distance from the pivot to the center of the strip is simply “y”. Note that each
horizontal strip of the triangle will have a different “y” and a different width “w”.
I’ve also defined the mass of the strip as “dm”. We can connect dy and dm by recognizing that:

mass of strip / mass of triangle = area of strip / area of triangle

or

dm / M = w dy / (1/2) s H (Note that the base of the triangle is “s”, the length of the side.)

or
2 M ⋅ w ⋅ dy
dm =
s⋅H

We will use this below when we write the expression for the moment of inertia of the strip. We can
also make two more connections (which we will need later):

Connect H to s
This is a simple matter of using either the left half or right half of the big triangle. The lower angle in
each corner is 60 degrees, so:
3
sin60O = H / s or H= s
2

Connect w to y
This one is also simple: use a triangle created by y and either the left or right half of the strip. The
angle in the upper corner is 30 degrees, so:
2
tan30O = (w/2) / y or w= y
3

Notice that if you use these two expressions together, when y = H, w equals s... just as it should (i.e.
the width of the triangle at the bottom is the length of the side.)

Now we can write the moment of inertia of the strip as it is rotated about the pivot at the top. Using the
parallel axis theorem (the same thing we did to derive the moment of inertia for the rectangle in lab):

1
dI = dm ⋅ w 2 + dm ⋅ y 2
12

If we replace dm with the appropriate expression:

1  2 M ⋅ w ⋅ dy  2  2 M ⋅ w ⋅ dy  2 M 2M
dI =  ⋅w + ⋅ y = ⋅ w 3 dy + ⋅ w ⋅ y 2 dy
12  s ⋅ H   s ⋅ H  6 s ⋅ H s ⋅ H
And then replace w with its expression:
3
M  2  2M  2  2 4M 4M
dI = ⋅  y  dy + ⋅  y  ⋅ y dy = ⋅ y 3 dy + ⋅ y 3 dy
6s ⋅ H  3  s ⋅ H  3  9 3⋅s⋅H 3⋅s⋅H

Note that both terms include a constant with y3 dy. We can simplify the expression with a bit of
factoring:
4M 1  4M  10  40 M
dI = ⋅  + 1 ⋅ y 3 dy = ⋅   ⋅ y 3 dy = ⋅ y 3 dy
3⋅s⋅H 9  3⋅s⋅H   9 9 3⋅s⋅H

Now we can add the dI for all the horizontal strips that make up the triangle. Note that the strips start
at y = 0 and end at y = H, so:
10 ⋅ M ⋅ H 3
H
40M 40M 1 4 H 40M 1
I= ⋅ ∫ y 3 dy = y ⋅ = ⋅ H4 =
9 3⋅s⋅H 0 9 3⋅s⋅H 4 0
9 3⋅s⋅H 4 9 3⋅s

We’re almost done! We want to express the moment of inertia using M, the mass of the triangle, and
s, the length of the side of the triangle. Which means we need to replace H in our expression using
the connection we wrote earlier for H and s :
3
 3  3 3 3
10 ⋅ M ⋅  s  10 ⋅ M ⋅  s 
10 ⋅ M ⋅ H 3
 2   8  30 ⋅ M ⋅ s 2
I= = = =
9 3⋅s 9 3⋅s 9 3⋅s 72

Or:
Moment of inertia for an equilateral 5
triangle rotated about one corner I = M ⋅ s2
12
And that’s it! Notice how the final expression came out very similar to the expression that we used for
the square in lab. Is this coincidence? Yes and no: obviously triangles and squares share some of the
same geometry, and the definitions we used for both are similar; but remember that the pivot we used
for the squares in lab was arbitrary (i.e. I chose it to be a distance of s/2 from the center.) So there is
some coincidence involved here.

And now the shorter, clever way...

The big equilateral triangle can be divided into four smaller equilateral triangles, as shown in the figure
below.
The center of mass of each triangle is shown by the blue dot. Note that the center of mass of the
central triangle is also the center of mass of the big triangle. The side length of each small triangle is
s/2, as shown. And the distance from the center of mass of the big triangle to its pivot is d.

We can first write d in terms of s. We can use the triangle shaded in green on the picture to write
that:
s
cos30O = (s/4) / x or x=
2 3
3
Recall that, from the earlier derivation, the height of the triangle is H= s
2
Then we can write that d = H - x

3 1  3 1   3 −1 1
So: d= s− s =  −  ⋅ s = 
 
 ⋅ s = s
2 2 3  2 2 3   2 3  3

Now for the clever part: the mass of each small triangle is only one-fourth the mass of the big triangle.
And the side length of the small triangle is only half the side length of the big triangle. We know that
the moment of inertia of the big triangle, if it was rotated about its center of mass, must be the same
expression as that of each small triangle... but appropriately scaled for its mass and size dimensions.
Since moment of inertia is proportionate to the mass of an object and proportionate to the square of
the linear dimensions, we know that

Due to the mass, I for the big triangle must be four times as much as the I for the small triangle

and

due to the length of side, I for the big triangle must be four times as much as the I for the small
triangle.

This means that due to both factors,

Ibig = 16 Ismall

This expression simply recognizes how the size of the triangle affects moment of inertia. But now we
can also write how the four small triangles together must equal the big triangle! But we have to be
careful: our Ismall is the moment of inertia for a small triangle if it is rotated about its own center of
mass. Three of the four small triangles are not rotated about their own center of mass, so we must
use the parallel axis theorem for these three:

Ibig = four Ismall plus additional for three triangles using parallel axis theorem

Recall that the additional for the parallel axis theorem is the mass of the object times the distance
from its center of mass to the center of rotation squared. Note that this distance in the picture is two
times z (the pink line). We can find z in terms of s, then double it and then use this distance in our
expression for the moment of inertia:

2 2
s  s  1 1 1 1
z = x −   = 
2
 − s 2 = − ⋅s = ⋅s
4  2 3  16 12 16 4 3

1
So the distance we want is twice this, or: ⋅s
2 3

And now:
2
 1  M 1 1
I big = 4 ⋅ I small + 3 ⋅ m ⋅  ⋅ s  = 4 ⋅ I small + 3 ⋅ ⋅ ⋅ s 2 = 4 ⋅ I small + M ⋅ s 2
2 3  4 12 16
Next, we can replace Ismall with the connection we developed above:

1 I big 1 1 1
I big = 4 ⋅ I small + M ⋅ s2 = 4 ⋅ + M ⋅ s 2 = I big + M ⋅ s 2
16 16 16 4 16

Now subtract ¼ Ibig from each side, and then solve for Ibig :

3 1
I big = M ⋅ s 2
4 16

1
I big = M ⋅ s2
12

Note that this is the moment of inertia of the big triangle if it is rotated about its center. Also note that
we didn’t have to use an integral to get it... we simply used a geometry argument to relate the small
and big triangles. Also note... this only worked because we could construct the large triangle from
identical smaller triangles!

We have one more step: since we want to place the pivot for our big triangle at the top corner, we
need to use the parallel axis theorem to get the expression for the moment of inertia about that pivot.

Ipivot = Ibig + M d2

or
2
1  1  1 1
I pivot = M ⋅ s2 + M ⋅  ⋅ s  = M ⋅ s 2 + M ⋅ s 2
12  3  12 3

or:
Moment of inertia for an equilateral 5
triangle rotated about one corner I= M ⋅ s2
12
Of course we arrive at the same expression as we did using the first method. This second method is
useful for two reasons: its helps to confirm that our expression is probably correct, since we derived it
two completely different ways and got the same result; and it demonstrates a different way (other than
the direct way) of deriving a moment of inertia expression.