How do I find the neutral axis for an unsymmetrical beam section?

[Addison91]

I’m trying to design a simple beam support and I keep running into confusion about where the true neutral axis lies when the cross-section isn’t symmetrical. My textbook’s explanation about the centroid and moment of area isn’t clicking when I try to apply it to an uneven I-beam.

[BenjaminRR]

About the neutral axis: For a homogeneous cross-section, the neutral axis is the horizontal line through the section's centroid. In an uneven I beam that centroid sits off the middle, so the line of zero strain isn’t mid-height anymore. When you bend, the stress at a point is proportional to its distance from that axis, so you still use sigma = M y / I with y measured from the line through the centroid. You just have to compute I about that axis, which means using the parallel-axis theorem to move from the centroid to your chosen axis if you ever switch. The math finally clicks when you start by finding the centroid of the whole cross-section, then build I for that axis; everything else follows, even with asymmetry.

[StellaET]

I once banged my head on this when the top flange was heavier. I drew the cross section, split into parts, computed each area moment, and the line of zero strain ended up sitting higher toward the heavy flange. Then plugging that into the M y / I formula gave a stress pattern that matched what I saw on a couple of strain tests. It isn’t magical, just bookkeeping, but it did clear up the confusion for me.

[Anthony_T]

Could it be you are chasing the wrong culprit? In that shape the math hinges not just on centroids but also on whether you’re really in pure bending and whether there’s shear lag or warping with the unsymmetrical parts. It’s easy to slip if the load isn’t in the plane you expect?

[StephenW]

Another thing I learned the hard way: I once used the mid-height as the axis by habit, then found the numbers were off by a noticeable amount. The fix was to actually compute the centroid of the whole cross section and build the I about that axis. Then the stress distribution with distance from that line lined up with what I saw in the lab. If you’re stuck, try redrawing with the correct reference and check your sign conventions.