Moment of inertia feeds the deflection calculation, the Euler critical load, and the lateral-torsional buckling check. An error in this one value propagates down the entire chain of checks. Getting to it takes one of three routes, depending on the section: a formula for simple shapes, decomposition with axis transfer for built-up ones, or a published table for rolled profiles. This article walks through all three routes and the failure mode each one carries.
What the quantity describes
The definition is the integral I = ∫y²dA over the cross-sectional area, with y measured from the axis under consideration. Dimension L⁴, in engineering practice usually mm⁴ or cm⁴.
The square inside that integral drives the main property: the contribution of an area element grows with the square of its distance from the axis. Material near the neutral axis contributes little. The flanges of an I-beam contribute most of it.
Geometry alone determines the value, independent of material and load. The practical consequence shows up in the structure of every formula: the dimension perpendicular to the axis is always cubed. For a rectangle, I = bh³/12, so doubling the height gives an eightfold gain in stiffness, while doubling the width gives only double.
Basic shapes
For basic shapes, the formulas are tabulated and taken as given. Reference tables from Penn State cover the full set about centroidal axes.
Rectangle: I_x = bh³/12. Circle of diameter d: I = πd⁴/64. Annular section with outer diameter D and inner diameter d: I = (π/64)(D⁴ − d⁴), a solid circle minus the hole. The same subtraction principle applies to a rectangular tube, outer outline minus inner, both taken about the common centroidal axis.
A triangle shows how much the choice of axis matters. About the centroidal axis I = bh³/36, about the base I = bh³/12. Same shape, factor of three apart.
Transferring axes
Moving between parallel axes runs on the parallel axis theorem: I = I_c + A·d², where I_c refers to the centroidal axis of the shape, A is the area, and d is the distance between axes. Course ME 101 at IIT Guwahati derives it from the definition, then applies it in reverse to obtain bh³/36 for a triangle by subtracting A·d² from the value about the base.
Two constraints matter in practice. The A·d² term is always positive, so the centroidal moment of inertia is the minimum across all parallel axes. And the starting point has to be a centroidal axis: the theorem cannot be applied directly between two arbitrary non-centroidal axes.
Built-up sections
A built-up section gets broken into rectangles, each transferred to a common axis, and the values are summed. Cutouts and holes are subtracted.
An I-beam is the simplest case here. Three rectangles, two flanges, and a web, both geometric axes coincide with the principal axes, and the centroid position is known in advance from symmetry. The transfer applies only to the flanges, over half the section depth.
A channel needs one more step. It is symmetric about the strong axis but not about the weak one, so the centroid position across the width has to be found first through the first moments of area of the three components. Only then does the transfer work. Most errors in hand calculation of a channel land on exactly this step.
For a welded box, an I-beam with cover plates, or a stack of several rolled profiles, no tabulated value exists. The only way to calculate the moment of inertia here is decomposition into elementary shapes plus axis transfer, and the arithmetic grows with the number of components.
Advertisement
Asymmetric profiles
An angle does not fit the scheme above. Neither of its geometric axes is an axis of symmetry, which means the product of inertia I_xy = ∫xy dA does not vanish.
A nonzero I_xy changes the physics of the problem. Bending about the horizontal geometric axis produces curvature outside the vertical plane: displacements out of the loading plane appear, along with twisting of the section. Computing stresses with the ordinary formula σ = M·y/I on geometric axes gives the wrong answer.
The correct procedure runs in four steps: locate the principal axes, resolve the moment into components along them, compute stresses separately, and superimpose. The rotation angle follows from tan 2α = −2I_xy/(I_x − I_y), the values themselves from I_max,min = (I_x + I_y)/2 ± √[((I_x − I_y)/2)² + I_xy²].
In the worked example from that same ME 101 course, a section with I_x = 7.24×10⁶ mm⁴, I_y = 2.61×10⁶ mm⁴ and I_xy = −2.54×10⁶ mm⁴ has principal axes rotated 23.8° from the geometric ones. Designing that section on geometric axes means working with the wrong stiffness.
Where tabulated values come from
For rolled profiles, the moment of inertia normally comes ready-made from tables. The European standard EN 10365 specifies nominal dimensions and mass per meter for I-sections, H-sections, and channels. It does not fix the root radii, which vary between rolling mills, and it does not publish moments of inertia. The I values for an IPE 300 or an HEA 200 come from separate section property tables aligned with Eurocode 3, or from a producer’s catalog.
Those tables account for the fillet at the web-to-flange junction. Explanatory notes to the UK Blue Book state that the moment of inertia is calculated taking into account all tapers, radii, and fillets. North American practice follows the same rule with one distinction of its own: section properties are computed on the smallest theoretical fillet radius, while the detailing dimension k uses the largest. Angles were historically the exception, with fillets excluded from area and moments of inertia.
Analysis in the AISC Engineering Journal quantifies what that inclusion costs. Moving from idealized sharp-corner geometry to the real geometry with fillets, rolled angles gain 0.2 to 1.2% in area and lose 0.1 to 2.7% in maximum moment of inertia. The torsion constant J moves the other way, up by 4.5 to 15.4%.
Idealizing a profile as three rectangles without fillets works for preliminary sizing on bending stiffness, where the discrepancy stays within a couple of percent. For torsion checks, that simplification already introduces noticeable error. EN 1993-1-1 confirms the root radius as a design parameter in its own right: the shear area of a rolled I-section is defined there as A_v,z = A − 2b·t_f + (t_w + 2r)·t_f, with r the root radius.
Where moment of inertia stops being enough
The quantity governs stiffness: deflection through δ = 5wL⁴/384EI, critical load through N_cr = π²EI/(KL)². Bending strength is governed by the section modulus W_el = I/y, where y is the distance to the extreme fiber. Two sections with identical I give different W if the extreme fiber sits at a different distance, and substituting one for the other happens regularly.
Cross-section class sets another limit. For thin-walled Class 4 profiles the gross section overestimates capacity: part of the compression zone loses local stability before yield is reached, and effective section properties per EN 1993-1-5 enter the calculation instead. The moment of inertia is then recomputed on the reduced section.
The torsion constant sits apart. For open profiles, J is computed as the sum of b·t³/3 over the component plates and does not match the polar moment of inertia I_x + I_y. The equality J = 2I holds only for circular sections.
The overall sequence comes down to three operations: locate the centroid, compute the individual moments of inertia of the component shapes, and transfer them to a common axis through A·d². Asymmetric sections add a fourth, rotation to the principal axes. For rolled profiles, these calculations are already done and tabulated, but knowing which assumptions went into them stays useful for any check outside the standard catalog.
This is a sponsored post by Promodo, Inc. All reviews and opinions expressed in this post are not based on the views and opinions of Tomorrow’s World Today.
Read Exclusive Stories
Bookmark & Save Articles For Later
