Moment of Inertia of Areas
Area far from the axis counts for much more than area near it.
Before this Lesson
Lay a plank 2 in thick and 10 in wide flat across a ditch and walk out to the middle. It dips under you like a diving board. Now stand the same plank on edge, brace it so it can't roll, and try again. It hardly moves. It's the same wood with the same area. All you changed is how far that area sits from the line the plank bends about.
The number that tells those two planks apart is the cross section's moment of inertia. Like the centroid, it's pure geometry.
Distance counts twice
Cut through a sagging beam and look at the cut face. The top is squashed, the bottom is stretched, and a line between them, through the section's centroid, does neither. That line is the axis.
Compare two equal patches of material, one a distance from the axis and one . The far patch stretches twice as much, so it pulls twice as hard, and its pull has twice the lever arm. That's four times the resistance to bending. Distance gets counted twice, so each bit of area counts by its distance squared.
Add up every bit, and you have what the Handbook's Statics chapter, under Moment of Inertia, calls the moment of inertia, or second moment of area:
The in is the distance from the -axis. In Centroids of Areas, was the first moment, area times distance, and this has one more factor of distance. So bits on opposite sides of the axis add instead of canceling, and the units are length to the fourth power.
Despite the name, nothing here has mass or spins. The Handbook's Dynamics chapter uses the same letters for Mass Moment of Inertia, a different quantity.
Why depth is cubed
Take a rectangle of width and depth , with the axis through its centroid. Double the width, and you've set two copies side by side: twice the . Double the depth, and there's twice the area with every bit twice as far out, which counts four times over: eight times the . That's the Handbook's area table, for a horizontal axis through the centroid:
The dimension that's cubed is the one perpendicular to the axis, however the page is turned. About a vertical axis, cubes the width. For the planks:
That's the same area with 25 times the , and 25 is . Under the same load, the plank on edge sags a twenty-fifth as far.
Converting units takes all four powers: 1 ft⁴ is in⁴, and 1 m⁴ is mm⁴. The Handbook prints neither factor.
Read the row's axis first
The area table from Centroids of Areas has a column headed Area Moment of Inertia. Each row gives about the centroid and about the -axis in that row's own sketch, and that -axis moves from row to row. In the circular rows, is the radius. The Handbook's sketches mark only the centroid, so the centroidal axes are drawn in here.
Adapted from the FE Reference Handbook, Statics area table
| Shape | Its -axis | |||
|---|---|---|---|---|
| Rectangle | the base | |||
| Triangle | the base | |||
| Circle | tangent at the bottom | |||
| Semicircle | the flat side |
Adapted from the FE Reference Handbook, Statics area table
A triangle's looks like a rectangle's centroidal formula, but it's about the base. Its centroidal value is .
A semicircle's is about the flat side, which is from its centroid. Its centroidal value is the awkward one, about .
Later chapters mostly want the centroidal value: the bending-stress formula writes it plain , and Fluid Mechanics writes it .
Moving the axis
For an axis the table doesn't list, such as one through a triangle's tip, follow one bit of area. If it's above the centroidal axis and the new axis is below that, the bit is above the new axis.
Add up every bit:
The first integral is the centroidal value, , and the last term is . The middle integral is the first moment about the centroidal axis, which is zero: that's what a centroid is. What survives is the parallel-axis theorem:
The Handbook, under Moment of Inertia Parallel Axis Theorem, prints it as , with the vertical distance between the axes, and likewise for . Test it on a rectangle, whose base is from its centroid: , the table's .
Two things follow:
- can't be negative, so of all parallel axes, the one through the centroid has the smallest .
- The middle term vanished only because the starting axis ran through the centroid, so one end of every transfer has to be centroidal. Going from a base to another axis takes two steps, through the centroid.
A hollow section is a solid minus its hole
is a sum over the area, so as with centroids, a hole subtracts, as long as both values are about the same axis. In a concentric section, the outer shape and the hole share a centroid, so no transfer is needed. The Handbook doesn't spell this out, and it has no hollow-rectangle row, so with capitals for the outer rectangle:
Subtract the values, not the dimensions, and count a wall on each side. The table does have a hollow-circle row, with the outer radius and, in this row, the inner radius:
Hollowing costs little. Bore a 4 in hole through a 6 in round bar, and you remove 44% of the area but only 20% of the , because that material sat near the axis, where it counted least.
Judging without computing
Many questions ask only which section has more , or how changes. Three rules cover them:
- A rectangle scales as width times depth cubed, so flat to on edge gains .
- Scale every dimension of a shape by one factor, and scales by its fourth power, which covers a circle's diameter.
- With the area fixed, push it away from the axis, and climbs. That's why steel beams keep so much of their area in two flanges.
Other questions ask which parallel-axis expression is set up correctly for one part of a section. The full calculation waits for Composite Sections, but the setup is just the theorem applied to that part. Take the T from Centroids of Areas.
The flange's own centroid is 9 in above the bottom edge, and the T's is 7 in, so the flange's share of about the T's axis is:
Each part uses its own centroidal value, with its own depth cubed, and runs from the part's centroid to the new axis, not from an edge.