Composite Sections
Break the section into rectangles, move each one to the axis, and add.
Before this Lesson
A W18X50 steel beam has a moment of inertia of 800 in⁴. Weld a plate 8 in wide and 1 in thick to each flange. Each plate's own is about 0.67 in⁴, so you might expect to gain a little over 1 in⁴. You gain about 1,445. The steel in those plates is ordinary. What matters is where it sits.
You already own the tools: the part-by-part table from Centroids of Areas and from Moment of Inertia of Areas. A built-up section is those two run back to back. After that, two new quantities come straight out of : the radius of gyration and the polar moment of inertia.
Two more columns
Take the T from Centroids of Areas: a 12 in by 2 in flange on a 2 in by 8 in stem, with its centroid 7 in above the bottom edge. Each part brings its own , about its own centroid, plus for being away from the section's centroid. Here is that part's area, and runs from the part's centroid to the section's.
| Part | ||||||
|---|---|---|---|---|---|---|
| Stem | 16 | 4 | 64 | 3 | 85.33 | 144 |
| Flange | 24 | 9 | 216 | 2 | 8 | 96 |
| Sum | 40 | 280 | 93.33 | 240 |
Both columns count. Keep only , and you land 28% low. And for a centroidal , ends at the section's centroid, never at an edge.
Look at how each part earns its share. The flange is thin, so only 8 of its 104 in⁴ is its own , and the rest is . The stem is deep, and depth gets cubed, so its supplies over a third of its 229 in⁴. A thin part contributes almost entirely through distance, so the farther out you push it, the more it does. That's the job of a beam's flanges.
A named axis skips the centroid
Sometimes the question names the axis: the base, or the line where two parts meet. Then there's no composite centroid to find, because every runs to that line. Each part still starts at its own centroid, with a triangle's or a semicircle's from the area table.
There's one shortcut. If the line happens to be a row's own -axis, the table's has already done the transfer. Take the plate from Centroids of Areas, an 8 in by 6 in rectangle capped by a 4 in radius semicircle, about the line where they meet. That line is the rectangle's top edge, which works just like its base, and the semicircle's flat side:
A hole subtracts its transfer too
A hole is a part with a minus sign, as it was for centroids. Its share about the axis is still , and all of it comes off. In Moment of Inertia of Areas, the holes were concentric, so was zero. Move the hole off center, and the transfer term is usually the bigger of the two.
The hole's is squared, so the side of the axis it's on doesn't change the sign. For the section's own centroidal axis, locate that centroid first with the hole as a negative area, then transfer both the solid and the hole to it.
Rolled steel comes with its numbers
A rolled shape is one part whose homework has been done. The Handbook's Civil Engineering chapter, under Design of Steel Components, has Table 1-1: W Shapes Dimensions and Properties. W18X50 means about 18 in deep and 50 lb per foot.
| Shape | (in²) | (in) | (in⁴) | (in) | (in⁴) | (in) |
|---|---|---|---|---|---|---|
| W18X50 | 14.7 | 18.0 | 800 | 7.38 | 40.1 | 1.65 |
| W16X36 | 10.6 | 15.9 | 448 | 6.51 | 24.5 | 1.52 |
| W14X48 | 14.1 | 13.8 | 484 | 5.85 | 51.4 | 1.91 |
Adapted from the FE Reference Handbook, Table 1-1: W Shapes Dimensions and Properties
The table groups its columns under Axis X-X and Axis Y-Y and writes plain and , so and are my labels. X-X runs horizontally through mid-depth, and Y-Y runs along the web. A W mirrors across both, so both axes pass through its centroid. That makes the table's the shape's , ready for the theorem.
The Handbook doesn't mention cover plates, and it doesn't need to. The W goes in as one row of the table, and each plate is a rectangle. One thing to watch: the table's is the beam's depth, not a transfer distance, so I'll call it . A plate of thickness sits on the outside of the flange, so the distance from the W's centroid to the plate's is:
With equal plates top and bottom, the section's centroid stays at the W's mid-depth, and the W needs no transfer.
One length for the whole area
Here's a thought experiment. Squeeze all of a section's area into thin strips, keeping the total the same. How far from the axis must the strips sit to give the same ? With every bit at one distance , the integral collapses to . That distance is the radius of gyration. The Handbook's Statics chapter defines it just this way, under Radius of Gyration:
It's a measure of spread, not a place where material sits. For a rectangle about its centroidal axis, the width cancels: , so . That's more than , the plain average distance, because far bits count more.
For anything else, the recipe is the same: find about the stated axis and divide by the area. The T, for example, has in. A W shape's table lists outright: a W18X50 has in, and agrees.
The least r marks the weak axis
A section has a different about each axis, and the smallest is often the one that counts. A slender column buckles the easy way, about the axis its area hugs most closely, as Euler Buckling will show. That axis is the section's weak axis, a term the Handbook doesn't use. For a W shape it's Y-Y: a W18X50 has 1.65 in against 7.38 in.
Don't guess for other sections. Compute about both centroidal axes and report the smaller. Remember that the area is the same for both, and only changes.
Radii don't add, but their squares do
Move the axis to a rectangle's base, and it's tempting to add the offset to the centroidal . That doesn't work. What transfers is , so divide through by :
The Handbook doesn't print that form, but it's only one line of algebra. Read it as a right triangle: and are the legs, and is the hypotenuse. For a rectangle about its base, that's . Simply adding gives , 37% too high.
The area table saves you the step for each row's own axes. Its column headed (Radius of Gyration)² prints , so remember the square root.
| Shape | Its -axis | ||
|---|---|---|---|
| Rectangle | the base | ||
| Triangle | the base | ||
| Circle | tangent at the bottom |
Adapted from the FE Reference Handbook, Statics area table
From a line to a point
So far, every distance has run to a line. A twisting shaft, though, turns about a point, and each bit of area counts by its squared distance from that point. A right triangle splits that distance into and , so the integral splits too. That gives the polar moment of inertia, printed in the Handbook's Statics chapter under Moment of Inertia:
The catch is that both axes must be perpendicular and pass through the point. So about a new point means transferring both values to axes through it, then adding. For a composite, that means tabulating and separately.
The area table's rectangle row lists without naming the point. It's , so it's about the centroid. The same row's and are about two edges that meet at a corner.
Round sections, from their diameters
A circle looks the same from every direction, so and is twice either one. The area table has and, with the outer radius and the inner, . Those are radii, while shafts usually come with diameters, and the Handbook prints no diameter form. So substitute :
Put a diameter into , and you land 16 times too high. For a hollow section, take the fourth powers first and then subtract.
You'll use these in Torsion of Circular Shafts.