Square vs round column: same concrete, same steel, different capacity

Two tied columns with the same gross area and the same eight #8 bars. Their P-M interaction curves meet at the top and the bottom, but in bending the square one reaches about 12% further. Here is where that 12% comes from, which is not quite where most people would look first.

PM Curve · 7 min read

The two columns

Both were run through PM Curve, which builds the ACI 318-19 interaction diagram by strain compatibility, and the square was bent about one of its main axes.

0 20 40 60 80 100 -300 -150 0 150 300 450 Design moment φMn (kip-ft) Design axial load φPn (kip) 9686 φPn,max = 441 kip, both Pure tension: 341 kip, both 12 × 12 in square 13.54 in round
Design interaction curves (φPn, φMn) from PM Curve. The curves share the compression cap and the pure tension point, but the square one is wider all the way through the bending range. The dots mark pure bending: 96 against 86 kip-ft.
Point on the design curveSquareRoundSquare vs round
Compression cap, φPn,max441 kip441 kipsame
Pure tension, φPnt341 kip341 kipsame
Balanced point, φMn88.2 kip-ft78.4 kip-ft+12.5%
Pure bending, φMn95.9 kip-ft85.8 kip-ft+12%
Largest design moment97.0 kip-ft86.0 kip-ft+13%

Why the ends of the curves match

Pure compression and pure tension do not depend on the shape at all, only on how much concrete and steel there is:

Pure compression. P0 = 0.85 f′c (Ag − Ast) + fy Ast = 0.85 × 4 × (144 − 6.32) + 60 × 6.32 = 468.1 + 379.2 = 847.3 kip. For tied columns ACI 318-19 caps it at 0.80 P0, and φ = 0.65, so φPn,max = 0.65 × 0.80 × 847.3 = 441 kip for both.

Pure tension. Only the steel works: φPnt = 0.90 × 60 × 6.32 = 341 kip for both.

Everywhere in between, the section has to bend, and then the shape decides how far from the centre the concrete and steel can work.

Where the bending difference comes from

The usual explanation is that a round column has bars near the neutral axis that do little. That is not what happens here: both columns have exactly two bars on the centre line. The difference comes from three other things.

1. The square puts more of its steel at the full lever arm

In the square, six of the eight bars sit in the two faces, 3.625 in from the centre. In the round column only two bars are at the extreme, 4.395 in from the centre, and four sit at 45°, just 3.108 in from it. Those four bars are strained less, and at the balanced point they have not reached yield: 38.8 ksi in compression and 43.1 ksi in tension, against 50.7 and 60 ksi in the square's faces.

The balanced point is the fairest place to compare, because φ = 0.65 for both. Splitting the nominal moment there into its parts:

Nominal moment at the balanced pointSquareRoundDifference
From the steel76.9 kip-ft64.6 kip-ft12.3 kip-ft
From the concrete58.9 kip-ft56.1 kip-ft2.8 kip-ft
Total, Mn135.8 kip-ft120.7 kip-ft15.1 kip-ft

About four fifths of the gap is the steel arrangement.

2. A circle is narrow where the compression block starts

12 × 12 in, 8-#8 block a = 3.38 in 13.54 in round, 8-#8 block a = 4.20 in
The two columns to the same scale, with the compression block at pure bending shaded. The top of a circle is narrow, so the round column's block has to reach deeper to carry about the same force: 4.20 in against 3.38 in.

The square's compression block is the full 12 in wide from the very top. A circle starts at a point and widens slowly, so to collect about the same concrete force (130 kip against 138 kip at pure bending) the round column's block has to go deeper, and the neutral axis with it.

3. A deeper neutral axis lowers φ at pure bending

At pure bending the nominal moments are only 6% apart: 115.0 against 108.3 kip-ft. The design moments are 12% apart because φ differs. ACI 318-19 Table 21.2.2 sets φ from the net tensile strain in the deepest bar, rising from 0.65 at yield (εty = 0.00207) to 0.90 at 0.00507:

Square. c = 3.97 in, deepest bar at 9.625 in: εt = 0.00427, φ = 0.65 + 0.25 × (0.00427 − 0.00207) / 0.003 = 0.834.

Round. c = 4.95 in, deepest bar at 11.165 in: εt = 0.00377, φ = 0.792.

The round column's deeper compression block pushes the neutral axis down, so the tension steel strains less and the section is further from tension-controlled. Half of the 12% at pure bending is this φ effect.

Rotating the round column's bars by 22.5°, so that two bars sit at the top instead of one, changes its design moment by less than 1 kip-ft. The result does not hinge on the bar orientation.

When the round column is still the better choice

For tied columns governed by bending about a known axis, the rectangular section gets more out of the same material.

Try it yourself

Open PM Curve, enter the square column, then switch the shape to circular with D = 13.54 in and eight #8 bars. Each curve comes with its seven key points, the φ factors and a step-by-step PDF report. Sign in, free, to overlay the two sections on one chart with the compare mode.

FAQ

Frequently asked questions

Is a square column stronger than a round column of the same area?

In bending about a main axis, often yes. In this example a 12 x 12 in square and a 13.54 in round tied column with the same 8-#8 bars have the same axial strength, but the square carries about 12% more design moment. A spiral round column, or bending in an unknown or diagonal direction, can change that.

Why are pure compression and pure tension the same for both?

They depend only on the areas of concrete and steel, not their arrangement: P0 = 0.85 f'c (Ag - Ast) + fy Ast, and pure tension is fy Ast. With the same Ag and Ast, both columns have P0 = 847 kip and a pure tension strength of 379 kip.

Why does the round column get a lower phi at pure bending?

ACI 318-19 sets phi from the net tensile strain in the deepest bar. The narrow top of a circle makes the compression block deeper, which pushes the neutral axis down and lowers that strain: 0.00377 against 0.00427 for the square, so phi is 0.792 against 0.834.

Does a spiral round column change the comparison?

Yes. Spiral columns use phi = 0.75 instead of 0.65 in the compression-controlled range and a higher axial cap of 0.85 P0, so a spiral round column needs its own comparison.

Keep reading

More from the blog

For use by qualified engineers. All results must be checked independently before use in any design. See the Terms of Use.