Moment-Resisting Timber Connections

Most timber connections just carry shear, maybe a little axial load. A moment-resisting connection has a harder job: it has to hold an angle. Think of a beam meeting a column at the eaves, or two members spliced mid-span where the frame is counting on the joint staying stiff. That connection has to carry bending moment across the joint — and the way a bolt group does that catches most people out the first time.

This note walks through how it works in EC5, why the spacing of the bolts matters more than where you put them, and the one mistake that trips engineers up most. There's a full worked example at the end, straight from a real design.

What makes a connection "moment-resisting"

It comes down to what your frame model assumes about the joint. You model it as either a pin (rotates freely, carries no moment) or rigid (holds the angle, carries the full moment). That choice is a promise — and the connection has to keep it.

A pinned lap splice just needs its bolts to carry the shear. A rigid eaves joint needs its bolt group to push back with an equal and opposite moment, and it does that by making some bolts work much harder than others. That's the whole trick: the moment is carried by a couple across the group — the outer bolts pulling hardest, the inner ones barely doing anything.

And if the model says rigid but the connection can't actually deliver that moment, the frame won't behave the way you analysed it. Nothing dramatic happens — no bang — but the loads redistribute in ways your model never accounted for. That's the one to watch.

Use the moment you already have

This is the big one, and it's where people slip up.

For a moment connection, you don't work out the moment — your frame analysis already did. The joint was modelled as rigid, so the analysis hands you the moment My,Ed the connection has to carry. You design the bolt group for that. You don't add anything to it.

The trap is something called "lap eccentricity." In a pure shear connection — no moment in the frame model — the shear doesn't line up exactly with the centre of the bolt group, and that small offset creates a little local moment (My,ad = −V·e) the frame never saw, because it treated the lap as a pin. That one is real, and you do have to add it.

But in a moment connection, adding a lap-eccentricity moment on top is just double-counting. The frame moment already is the full demand at that joint. The joint was modelled as rigid, the bolt group is the joint, and there's nothing left over to add. On top of that, "lap midpoint" barely means anything at an eaves joint where the column and beam cross at 90° — it's just the centre of an overlap, not where any load is applied.

So, simply:

• Pure shear (pinned lap): apply the shear, and add the small local moment the frame couldn't see.
• Moment connection (rigid joint): apply the frame moment as-is. Don't second-guess the analysis by adding an eccentricity it already covered.

Get this wrong and you can end up over-designing the connection by half, for nothing.

How a bolt group carries moment: the polar moment method

Once you've got the moment, you share it out across the bolts. The method is called the polar moment method — it's the same idea as a bolt group resisting twist.

The core of it: don't treat every bolt the same. Each one takes a force in proportion to how far it sits from the centre of the group. The outer bolts carry the most, the inner ones almost nothing. It's just leverage — the bolts furthest out have the longest arm, so they do the heavy lifting.

Here's the sequence:

1. Find the centre of the bolt group.
2. Measure how far each bolt is from it — call that rᵢ.
3. Add up Σrᵢ² across all the bolts.
4. Share out the moment: each bolt takes Fm,i = M · rᵢ / Σrᵢ², pushing at right angles to its radius — the direction it would move if the group spun.
5. Share out shear and axial equally — V and N don't care where a bolt sits.
6. Add up the forces at each bolt — moment, shear and axial as vectors — to get its resultant Fd.
7. The bolt with the biggest Fd is the critical one. Check that against EC5 capacity. If it passes, the whole group passes.

Quick note on "polar": the Σrᵢ² term is measured about an axis through the centre, pointing out of the plane of the bolts — the axis the group would spin around. That's different from the Ixx/Iyy you'd use for bending.

Spacing matters. Position doesn't.

This next bit changes how you actually detail the connection, and it drops straight out of the formula.

Look again at Fm,i = M · rᵢ / Σrᵢ². The force on the worst bolt depends on how the bolts sit relative to each other — not on where the group as a whole sits on the member.

Two things follow from that:

Spread the bolts out and each one works less. A wider group means a bigger Σrᵢ², so the same moment gives smaller forces all round. Bolts far apart share the moment easily; bolts crammed together fight over it. Roughly double the spread, roughly halve the peak force.

Slide the whole group around and nothing changes. Move every bolt together, keeping the spacing, and Σrᵢ² stays exactly the same — each rᵢ is measured from the group's own centre, which moved with it. For a pure moment, shifting the bolt group up or down the member doesn't change a single bolt force. (It does change the plate length and its EC3 check — but not the EC5 bolt capacity.)

So if a moment connection is overstressed, the lever to pull is spacing — spread the outer bolts. Adding bolts near the middle, or shuffling the group along, barely helps.

The circular pattern

A circular pattern — you'll see these a lot in moment connections and exposed frames — makes the maths especially clean. Every bolt sits at the same radius R, so Σrᵢ² = n · R², and every bolt takes the same size of moment force, just pointing a different way round the circle. It's the most efficient way to spread n bolts against a pure moment, which is exactly why architectural timber portals so often use a ring of dowels.

Angle to the grain, and why the worst bolt isn't always the busiest one

Here's where timber parts ways with steel. Timber is strong along the grain and weaker across it, so its bearing (embedment) strength depends on the angle between the load and the grain. And since every bolt in a moment group pushes in a slightly different direction, every bolt meets the grain at a different angle — so every bolt has a different embedment strength.

EC5 deals with this using the Hankinson formula (Eq. 8.31), which blends the along-grain and across-grain strengths for whatever angle a given bolt is actually loaded at. In a timber-to-timber joint each member has its own grain direction, so one bolt has two angles — one per member — feeding into the calculation.

The upshot is worth holding onto: the critical bolt isn't always the one carrying the most force. It's the one with the worst mix of force and angle. A bolt taking slightly less load, but at a nasty angle to the grain, can be the one that governs. Sorting through that bolt by bolt is exactly the kind of fiddly bookkeeping that's painful by hand and instant by machine — which is why moment connections are a natural thing to hand to a tool.

Bolts or dowels?

Dowels tend to win for moment connections, for two reasons. They sit in a tight pre-drilled hole with no slack, so the joint stays properly stiff instead of rotating a touch as bolts take up their clearance — and that little bit of slip is exactly what eats into the rigid-joint assumption your frame relied on. They also look cleaner on exposed timber, with no heads or nuts on show. Bolts are the pick when you want the bit of extra capacity from the rope effect (up to 25% under EC5 §8.2.2(2) — dowels don't get it), or when clearance holes and easy assembly matter more than a perfectly tight fit. There's more on the trade-off in the notes on dowels and choosing between the two.

A worked example: eaves moment connection

To make it concrete, here is a real moment connection designed to EN 1995-1-1 — a timber-to-timber (T|T) eaves joint where a beam is slotted between the two outer plies of a column.

The connection:

• Both members GL24h (ρk = 380 kg/m³)
• Applied at the joint from frame analysis: My,Ed = 5.0 kNm, Vz,Ed = −50 kN, NEd = 0
• Medium-term load, service class 2 → kmod = 0.80
• 8 no. M12 grade 8.8 bolts in a circular pattern, 230 mm diameter (so radius R = 115 mm)

Distributing the actions.

Because the pattern is circular, every bolt is at R = 115 mm, so the polar second moment is:

Σrᵢ² = n · R² = 8 × 115² = 105,800 mm²

Each bolt carries the same moment force magnitude, Fm = M · R / Σrᵢ² = 5,000,000 × 115 / 105,800 ≈ 5.43 kN, acting tangentially. The shear Vz = 50 kN splits equally, 50 / 8 = 6.25 kN per bolt, all pointing the same way. Vector-summing the tangential moment component with the shared shear at each of the eight positions gives eight different resultants — ranging from 0.82 kN on the bolt where they cancel to 11.68 kN on the bolt where they add.

The critical bolt.

Vector-summing gives a different resultant at each of the eight positions. Bolt 7 governs, and it is worth seeing why, because it is the worst of two things at once. Its moment component (tangential, pointing straight down at that position) and its shear share both act downward, so they add directly: it carries Fx ≈ 0, Fz = −11.68 kN, a resultant of

Fd = 11.68 kN

— the largest force in the group. And it sits at grain angles of 0° to the outer plies and 90° to the central member — the least favourable embedment combination, load running across the grain of the central member. So bolt 7 has both the highest force and the weakest embedment. (For contrast, bolt 3 on the opposite side carries just 0.82 kN, because there the moment and shear components nearly cancel — a vivid reminder that "eight bolts" does not mean "eight equal bolts.")

Capacity.

Running the critical bolt through the EC5 double-shear failure modes with its per-bolt Hankinson embedment, the joint is governed by bearing in the central member (mode h) — the weak perpendicular-to-grain embedment of the central member is what limits it, at Fv,Rk = 9.68 kN per shear plane, so 19.36 kN across the two planes. Applying the modification and material factors:

Fv,Rd = kmod · Fv,Rk / γM = 0.80 × 19.36 / 1.30 ≈ 11.91 kN

The check.

Fd / Fv,Rd = 11.68 / 11.91 = 0.981 → 98.1% utilisation. Pass — but only just.

That thin margin is the interesting part. The lever from earlier tells you exactly what to do about it: because the pattern is circular and the demand is moment-dominated, widening the bolt circle is the direct fix — a larger R gives a larger Σrᵢ², which reduces Fm per bolt, which drops the resultant on the critical bolt. You bring the utilisation down without adding a single fastener. Adding bolts near the centre, or sliding the group along the member, would do almost nothing — the maths from the last section says so.

Don't stop at the bolts: the splitting check.

The bolt check passing isn't the end of it. When fasteners drag load across the grain near a member edge, the timber can split along the grain before the bolts themselves reach capacity — a brittle failure the Johansen bolt calculation never sees. EC5 §8.1.4 (Eq. 8.4) checks it separately, and it's geometry-only — no grade, no density, just the shape of the member and where the load sits.

For the beam in this connection, with width b = 90 mm, depth h = 330 mm and loaded-edge distance he = 280 mm:

F90,Rk = 14 · b · √(he / (1 − he/h)) = 14 × 90 × √(280 / (1 − 280/330)) ≈ 54.17 kN

F90,Rd = kmod · F90,Rk / γM = 0.80 × 54.17 / 1.25 ≈ 34.67 kN

Note the γM = 1.25 — this is a member check on the timber, so it uses the timber material factor, not the 1.30 you'd use for the connection capacity itself. Against a perpendicular-to-grain demand of F90,Ed = 31.25 kN, that gives:

F90,Ed / F90,Rd = 31.25 / 34.67 = 0.901 → 90.1%. Pass — but close.

That's the point worth taking away. The bolts came out at 98.1% and the splitting at 90.1% — two completely different failure modes, both near the limit, and if you'd only checked the bolts you'd have missed a brittle one sitting right behind it. On a moment connection near a loaded edge, the splitting check earns its place.

Where this leaves you

The mechanics of a moment-resisting timber connection are not complicated, but they are unforgiving of two mistakes: double-counting a moment the frame analysis already reported, and under-spreading the bolt group. Get the demand right (use the frame moment, do not add lap eccentricity), spread the outer fasteners to build lever arm, and let the polar moment method with per-bolt Hankinson embedment find the critical fastener.

That last part — the per-bolt vector sum, the angle-dependent embedment, the search for the governing fastener — is exactly the arithmetic that is slow and error-prone by hand and instant by machine.

ConnForge designs moment-resisting timber connections to EC5 directly — eaves and splice moment configurations (STM, STSM, TSTM, TTM), the full polar moment analysis, per-bolt Hankinson embedment, a labelled critical bolt, and a complete calculation report and drawing you can put in front of building control. It is free to use. Try it →