What Mezzanine Floor Beam? Sizing for a Real Live Load Where Deflection Controls

You have a mezzanine to frame, a raised steel floor tucked inside a warehouse, a shop, or an office, and the practical question is blunt: what beam? Which section carries the floor, and how do you know it is the right one? The instinct is to reach for a bending check, find the section whose moment capacity beats the applied moment, and call it sized. For a mezzanine floor beam that instinct picks the wrong section almost every time.

The reason is that a floor is judged twice. It has to be strong enough not to fail, and it has to be stiff enough not to sag, bounce, or crack the finishes below it. These are two different limit states with two different governing quantities, and for the short, heavily loaded spans that mezzanines live in, the stiffness check wins. A beam can pass bending with room to spare and still be unacceptable because it deflects too far. This article works one real mezzanine beam all the way through both checks so you can see exactly where, and by how much, deflection takes over the decision.

Keep it concrete. The mezzanine is a light storage and general commercial floor, framed with simply supported steel beams spanning L = 6.0 m, spaced at 2.5 m centres, in grade S355 steel. A steel deck with a thin topping spans between the beams and, once fixed, provides continuous lateral restraint to the top (compression) flange, so lateral-torsional buckling does not reduce the bending capacity. That is a normal, defensible mezzanine build.

Each beam carries a 2.5 m wide tributary strip of floor, so the area loads convert straight to line loads on the beam:

• Superimposed dead load (deck, topping, finishes, services): g_k = 2.0 kN/m², giving 5.0 kN/m on the beam, plus the beam's own self-weight.
• Live (imposed) load: q_k = 5.0 kN/m², giving 12.5 kN/m on the beam.

Those two numbers, 5.0 and 12.5 kN/m, drive everything that follows. The live load in particular deserves its own step, because it is not something you invent.

The single most important input to a mezzanine floor beam is the live load, and it is fixed by occupancy in a code table. Get it wrong and every downstream number is wrong. The values are not small, and they climb fast with use:

• Residential floors sit at roughly 1.5 to 2.0 kN/m² (about 40 psf in the IBC). This is the light end, and it is what our companion piece on the residential mezzanine works through.
• Offices read about 2.4 kN/m² (50 psf), and the IBC adds a 15 psf partition allowance when the specified live load is under 80 psf, whether or not walls are built.
• Light storage, the most common industrial mezzanine use, jumps to 125 psf, about 6.0 kN/m², and Eurocode 1 puts storage areas (Category E) at 7.5 kN/m² and up.

Our worked value, 5.0 kN/m², is a representative real commercial and light-storage load, several times a home floor. This matters for the punchline: deflection is proportional to load, so a mezzanine's high live load is exactly what pushes the serviceability check ahead of strength. The same beam under a residential 2.0 kN/m² might pass deflection comfortably; under a storage 5.0 kN/m² it does not.

Every floor beam has to clear two independent gates, and they use different loads and ask different questions.

Strength (ultimate limit state). Can the section carry the factored load in bending and shear without yielding or buckling? Loads are amplified by partial factors (1.35 dead and 1.5 live in Eurocode, 1.2 dead and 1.6 live in AISC LRFD), and the section's moment resistance must beat the resulting design moment. This is the check people mean by 'will it hold?'.

Serviceability (deflection). Under the unfactored (characteristic) load, does the beam stay within a deflection limit set so that finishes do not crack, doors do not stick, and the floor does not look or feel wrong? The classic limits, used by the IBC's Table 1604.3 and mirrored across codes, are L/360 under live load and L/240 total. For our 6 m span:

• Live-load limit: L/360 = 6000/360 = 16.7 mm
• Total-load limit: L/240 = 6000/240 = 25.0 mm

The trap is that these two gates do not move together. Strength depends on the section modulus; deflection depends on the second moment of area and, critically, on the span to the fourth power. On a short mezzanine span with a big live load, a section can stroll through strength and slam into the deflection wall. The only way to know which gate governs is to check both, so that is what we do next, on the same beam.

Start where the instinct sends you, at strength. Factor the loads (Eurocode 1.35 dead, 1.5 live), including a first guess at self-weight:

w_Ed = 1.35 · (5.0 + 0.30) + 1.5 · 12.5 = 25.9 kN/m

For a simply supported beam the design moment and shear are the textbook results, and because they come only from statics they do not depend on the section at all:

M_Ed = w_EdL²/8 = 25.9 · 6²/8 = 116.6 kN·m
V_Ed = w_EdL/2 = 77.7 kN

Now find a section that carries it. An IPE 240 in S355 has a plastic modulus W_pl,y = 366.6 cm³, so with the deck restraining the compression flange its bending resistance is M_pl,Rd = f_y·W_pl,y = 355 · 366.6/1000 = 130.1 kN·m. That beats the 116.6 kN·m demand at a utilisation of 0.90. Shear is nowhere close: the IPE 240 shear resistance is about 391 kN against a 77.7 kN demand. On strength, the IPE 240 passes, and at 90% it even looks efficient. A designer who stops here writes down IPE 240 and moves on. That is the mistake.

Now the second gate. Deflection uses the characteristic loads (no factors), and the live-load deflection of a simply supported beam under a uniform load is:

δ = 5·q·L⁴ / (384·E·I)

with q = 12.5 kN/m (the live line load), L = 6000 mm, E = 200 000 MPa (200 GPa, the value CalcSteel's engine uses, equal to the AISC 29 000 ksi), and I the beam's second moment of area. For the IPE 240, I = 3892 cm⁴, and the arithmetic gives:

δ_live = 27.1 mm against a limit of 16.7 mm

The IPE 240 deflects 62% past its live-load limit. The total-load deflection is worse still at 38.6 mm against the 25 mm cap. The section that passed bending at 90% utilisation fails serviceability outright, and it is not close. This is the whole story of the mezzanine floor beam in one line: strength said yes, deflection said no.

So go up a size. An IPE 270 (I = 5790 cm⁴) is a much stiffer beam, and its bending utilisation drops to a comfortable 0.68. Its live deflection? 18.2 mm, still over the 16.7 mm limit by 9%. The tempting middle section, the one that feels like it should be plenty, misses too. You have to reach an IPE 300 (I = 8356 cm⁴) before the live deflection lands at 12.6 mm, inside the limit, with the total deflection at 18.1 mm, comfortably under 25 mm. The IPE 300 is the answer, and its bending utilisation is a mere 0.53.

Line the three candidates up and the cost of serviceability is plain. Strength alone would have bought an IPE 240. Deflection forced an IPE 300, two sizes deeper, at 42.2 kg/m against 30.7 kg/m, about 37% more steel for the same beam. None of that extra material is doing strength work: the final section is barely half utilised in bending.

Reading a table like this is the actual work of sizing a floor beam. The bending-utilisation column tells you the beam is over-designed for strength; the deflection column tells you why you cannot use anything lighter. If you only ever looked at the first, you would ship an IPE 240 and get a floor that sags visibly and feels alive underfoot.

The reason deflection governs a mezzanine floor beam is built into the two formulas. Bending strength is proportional to the section modulus and to the yield strength f_y. Deflection is proportional to 1/I, the reciprocal of the second moment of area, and to L⁴. Those are different levers.

The cleanest way to see it is to solve the deflection limit for the stiffness you need. Setting δ = L/360 and rearranging:

I_req = 5·q·L⁴ / (384·E·(L/360)) = 6328 cm⁴

That single number decides the beam. The IPE 270 offers 5790 cm⁴ and is short; the IPE 300 offers 8356 cm⁴ and clears it. No strength calculation was involved.

This also explains the most common wrong instinct: upgrading the steel grade does nothing for deflection. Going from S275 to S355, or S355 to S460, raises f_y and so raises bending capacity, but the modulus of elasticity E is essentially the same for all structural steels (about 200 GPa), and deflection depends on E and I, not on f_y. A stronger steel lets you use a lighter section for strength, which makes the deflection problem worse. When stiffness governs, the only real levers are a deeper section (bigger I), a shorter span or an extra support (L⁴ is brutal), composite action with the concrete topping (which can lift the effective I sharply), or pre-cambering the beam to offset the dead-load portion. Reaching for a higher grade is the one move that cannot help.

Everything above was hand calculation. To make it auditable we rebuilt the identical beam in CalcSteel's finite-element engine, the real shipping solver, and let it compute the demand from first principles by assembling and solving K·d = F. Under the 12.5 kN/m characteristic live load it returns support reactions of 37.500 kN each, a maximum shear of 37.500 kN, and a mid-span moment of 56.250 kN·m, matching the closed-form wL/2 and wL²/8 to three decimals. Scale the same model to the factored load and the design moment lands on the 116.6 kN·m used above.

The mid-span deflection comes out at 13.0 mm for the IPE 300, right beside the 12.6 mm hand value. The small gap is honest and worth naming: the engine derives the second moment of area from the profile's actual geometry, which sits a few percent below the rounded catalogue table value we used by hand, so the engine reads marginally more flexible. It moves the deflection the safe way, and it does not change a single verdict: IPE 240 and IPE 270 still fail L/360, IPE 300 still passes. That is the point of checking both by hand and by machine, the two agree, and where they differ you understand exactly why.

The calculator below is the same FEM engine, running in your browser. Set a 6 m simply supported beam, apply a 12.5 kN/m uniform load, and read the bending moment and the deflection together, then swap the section and watch the deflection move while the moment stays put. Push the span from 6 m toward 7 or 8 m and see how fast the L⁴ term drives the deflection, and therefore the required depth, up. It is free, and there is no login for the analysis.

Sizing a beam is a loop, load, section, check, repeat, and a live model collapses that loop to a single edit. When you want to see the second moment of area a section actually gives you, the moment of inertia calculator pairs naturally with this one.

Strength and deflection are the two gates that size the beam here, but a slender steel floor has a third check that occasionally overrides both: walking vibration. A beam that passes strength and even deflection can still feel like a trampoline, because the floor's natural frequency can sit near the rhythm of footsteps and resonate. That is a comfort (serviceability) issue, governed by AISC Design Guide 11, and for a long, lightly damped mezzanine it can dictate a still deeper section than L/360 alone.

It does not usually govern a 6 m storage floor loaded to 5 kN/m² like this one, where deflection is already forcing an IPE 300, but it is the reason you cannot fully automate a mezzanine to a single moment check. We work the vibration check in detail, alongside the load tables and deflection, in the companion residential mezzanine article, and the broader picture of what actually limits a beam is in the five checks that decide a steel beam's capacity.

Sizing on bending alone. The headline error, and the whole subject of this article. A moment check picks the lightest section that holds; for a floor beam that section is routinely too flexible. Always run deflection, on characteristic loads, before you commit.

Checking total deflection but not live. The L/240 total limit and the L/360 live limit are separate, and for imposed-load-dominated floors the L/360 live check is usually the tighter one. Do both.

Guessing the live load. It is a code table entry keyed to occupancy, and a storage mezzanine is not an office is not a home. Missing the partition allowance, or reading a residential number for a commercial floor, quietly under-loads the whole design.

Reaching for a higher steel grade to fix a deflection failure. It cannot work: E is the same for all structural steels, so a stronger grade changes strength, not stiffness. Add depth, add a support, go composite, or camber.

Does the deck really give lateral restraint? Only once it is fixed to the beam and able to act as a diaphragm. If the compression flange is unrestrained during construction or by a bare deck, lateral-torsional buckling can cut the bending capacity and must be checked separately.

Can composite action save the section? Often, yes. Connecting the beam to the concrete topping with shear studs can raise the effective second moment of area enough to bring the deflection back inside the limit without going deeper, which is exactly why composite mezzanine beams are so common.

Is IPE the only choice? No. The same logic sizes a UB, a W-section, or a built-up member; the profile family changes the exact numbers, not the conclusion that stiffness governs. Search your own catalogue for the shallowest section that clears I_req.

A mezzanine floor beam is the textbook case where the strength check is the easy one and the deflection check does the real sizing.

• Take the live load from the code table, not intuition: a storage or commercial mezzanine carries 4 to 6 kN/m² or more, several times a home floor, which is what pushes deflection ahead of strength.
• Check two serviceability limits on characteristic loads: L/360 under live (16.7 mm here) and L/240 total (25 mm), separately from the factored strength check.
• In the worked case the IPE 240 passes bending at 90% utilisation but deflects 27.1 mm, 62% past L/360; even the IPE 270 misses by 9%; the IPE 300 is the section deflection demands.
• The controlling number is the required stiffness, I_req = 6328 cm⁴ for L/360, verified against the FEM engine's M = 56.25 kN·m and V = 37.5 kN to three decimals.
• When stiffness governs, add depth, span, support, composite action, or camber. A higher steel grade raises strength but not E, so it cannot fix a deflection failure, and often makes it worse.