Estimating a Beam's Safe Load: Span, Deflection and the Load Case

Which power of the span: measured, not assumed

The tool ships seven load cases, and the exponent is not the same in all of them. For a uniformly distributed load the moment is wL squared over 8 and the deflection 5wL to the fourth over 384EI; for a point load at midspan the moment is PL over 4 and the deflection PL cubed over 48EI. So the moment grows with the square of the span in the distributed cases and only linearly in the point-load cases, and the deflection grows with the fourth power in one family and the third in the other. Doubling the span on a real section and reading the ratios back confirmed it exactly: 4.000 and 16.000 for the distributed case, 2.000 and 8.000 for the point load.

The tool's own "good to know" note says, without qualification, that the moment grows with the square of the span and the deflection with its fourth power. That is right for three of its seven cases and wrong for four, and its own formula panel prints the contradicting expression a few inches further up the page. Read the panel, not the note.

There is an honest refinement that matters when you are comparing two designs. The fourth power holds the distributed load constant per unit of length, so widening the span also adds load: a 6 m beam at 8 kN/m carries twice the total weight of a 3 m beam at 8 kN/m. Spread the same total load over twice the span and the deflection rises eightfold, not sixteen. Both statements are true; they answer different questions, and the tool answers the first one because its distributed-load field is per metre, or per foot, and not a total.

Which loading case the tool assumes, and why it changes the answer

It opens on a simply supported beam with a uniformly distributed load, and the load field is then read as force per unit length. Switch the case to a point load and the same field is read as a total force. That is the single most consequential setting on the page, and nothing warns you when it changes meaning under the same box. Take one section, one span and one total load and put it on both ways: an IPE 240 over 5 m carrying 40 kN gives 25.00 kN-m and 7.97 mm as a distributed load, and 50.00 kN-m and 12.74 mm as a single load at midspan. The moment doubles; the deflection rises by 60 %; the demand/capacity ratio goes from 0.498 to 0.996, which is the difference between comfortable and finished.

The comparison table at the bottom of the tool, headed with the seven load cases, quietly reuses the number you typed rather than the load. With a distributed load of 8 kN/m entered, the point-load rows are computed from a point load of 8 kN — not from the 40 kN total the distributed case actually represents. The rows are internally consistent and mutually misleading: what they show is what would happen if the same figure meant something else, which is rarely the question anybody is asking.

What L/360 is in fractions of an inch

A deflection limit is a ratio of the span, so it means nothing until you divide. On a 12 ft span L/360 is 0.400 in; on 16 ft it is 0.533 in; on 20 ft it is 0.667 in. The looser limits are simply bigger fractions of the same span: at 16 ft, L/240 is 0.800 in and L/180 is 1.067 in. Write the number down before you look at the tool's answer, because a deflection quoted in thousandths of an inch is easy to accept without noticing that the beam will visibly sag.

Two things the tool does with that ratio deserve a note. It offers only L/120, L/180, L/240, L/360 and L/480 in the menu, yet one of its own examples sets the limit to L/300 — a value the menu cannot display, so the control shows a limit the calculation is not using. That example passes at L/300 with 15.96 mm against 16.67 allowed, and fails at the L/360 the menu will snap you to, where the allowance drops to 13.89 mm. And for a cantilever it divides the cantilever's own length by the ratio, which is stricter than the usual convention of taking twice the projection.

The bigger gap is which load the limit is applied to. A code deflection table does not name one ratio; it names several, one per load case. The International Building Code's deflection table gives floor members l/360 under live load alone and l/240 under dead plus live, and adds a footnote instructing that for steel members the dead load be taken as zero, and another that for cantilevers l be taken as twice the length of the cantilever. This calculator has one load field and one ratio, so whichever you choose is applied to the single number you typed. If you want the real serviceability check you have to run it twice, with different loads and different ratios.

Deflection usually governs — but not always, and the tool's own note is wrong about it

Strength and stiffness run out at different rates because they depend on different section properties: bending stress on the section modulus, deflection on the moment of inertia and on the elastic modulus of the material. On steel, stiffness usually runs out first, and the gap widens with the span. Take the tool's own W12x26 in A992 at L/360 and raise the load until something gives. At 12 ft the two checks arrive almost together: the deflection reaches its 0.400 in limit at about 5,075 plf, where the demand/capacity ratio is already 0.995. At 24 ft the deflection fails at about 635 plf while the bending check would still have carried 1,280 — the beam runs out of stiffness at half the load that would have exhausted its strength.

The tool's note goes one step further and says that for floor beams it is almost always deflection, not strength, that picks the section. Run its own second example and that turns out to be false for timber. A 2x10 in Douglas fir-larch No. 2, at the tool's unadjusted 900 psi allowable and L/360: over 12 ft the bending check fails at about 90 plf while the deflection check would still have taken 140; over 16 ft it is 55 against 60; over 18 ft, 40 against 45. Bending is the check that runs out first at every span in that range, and at the load where it does, the deflection is still only L/362.

That is not a mistake in the arithmetic; it is a consequence of the numbers the tool offers. Its timber allowables are bare reference values with no adjustment factors applied — nothing for load duration, moisture, member depth or repetitive members — so bending capacity is left artificially low relative to stiffness. A designer would apply those factors before comparing the two checks, and in real floor design the deflection check is also usually run on live load alone while the bending check carries everything. Neither of those moves is available on this page, which is precisely why its answers are an order of magnitude and not a verdict.

What the tool never gives you

There is no allowable-load output anywhere in it. It checks a load you propose; it does not solve for the load a beam can take. If you want a safe load you have to search by hand — raise the load until the demand/capacity ratio reaches 1 or the deflection reaches its limit, and take the smaller of the two. That is not a flaw so much as a design choice, but it is the question most people arrive with, and the page never says it cannot answer it.

The self weight of the rolled section is displayed and never added to the load. Type 8 kN/m for an IPE 200 and the calculation uses 8 kN/m, while a line underneath reports the profile's 22.4 kg/m as information. That is roughly 0.22 kN/m you are meant to add yourself, and the same figure is missing entirely for a rectangle, a circle, a hollow section or a custom I and S, where no self weight is shown at all because the tool has no density for them.

Frequently asked questions

Can I size a beam for my house with this?

No. You can find out whether the size you are imagining is roughly right, which is worth doing before a conversation with an engineer, and that is the whole of it. Everything that actually turns a beam calculation into a design is missing here: load combinations, lateral-torsional buckling, shear and bearing, the connections at each end, continuity with the rest of the frame, fire, and the adjustment factors that turn a reference stress into an allowable one. Removing or replacing a load-bearing member is also usually a permitted alteration where you live, which means someone has to sign for it.

Why does a deeper beam help so much more than a wider one?

Because depth is cubed and width is not. For a rectangle the moment of inertia is width times depth cubed, over twelve, and the section modulus is width times depth squared, over six. Double the width and both double. Double the depth and the section modulus quadruples while the moment of inertia goes up eightfold, so bending capacity is four times better and stiffness eight times better for the same amount of material rotated. That is why joists are laid on edge and why a beam on its side is a very expensive plank.

The tool says my beam passes. Does that mean it is safe?

It means the two checks it performs came out under their limits for the load you typed. Its own wording is careful about this: the green banner reads that the member passes in this model, not that it passes. The load you typed is the load it believes, and if you forgot the self weight, underestimated the live load or used an unfactored figure where your code wants a factored one, the model was answered correctly and the question was wrong. An I-beam also only reaches its allowable stress if its compression flange is held sideways by joists, decking or bridging; unbraced, it buckles laterally well before that, and the tool does not check it.

How much does fixing both ends really help?

On paper, a great deal: under a distributed load the maximum moment drops from wL²/8 to wL²/12 and the deflection from 5wL⁴/384EI to wL⁴/384EI, which is a moment 1.5 times smaller and a deflection five times smaller. In practice it is the assumption most often taken for free and most often wrong. Fixity means the support genuinely prevents the beam end from rotating, which requires a moment connection designed for it. A beam resting on two walls is not fixed; a beam bolted through a web-only cleat is not fixed either. If you choose that case in the tool, you are asserting something about the connection that the tool has no way to check.

Why does the section list differ between the imperial and metric modes?

It does not, and that is worth knowing. The same three catalogues are offered in both unit systems: American W-shapes, and the European IPE and HEA series. The tool converts, so you can put a W12x26 into a metric calculation or an IPE 200 into an imperial one, and the section properties come back in the units you chose. What you should not do is take that as licence to specify one where the other is sold. Ask a supplier here for a W-shape and you will be quoted an IPE or an HEA; ask abroad for an IPE and you may be quoted the nearest W. The section properties are the calculation; the designation is the purchase order.