The Ends Take All the Load: Shear Stress in Bonded Lap Joints

The problem with dividing load by area

The naive way to check a bonded lap joint is to take the applied load P, divide by the bonded area — joint width w times overlap length L — and compare that mean shear stress against the adhesive's allowable. It's a two-line calculation, it's wrong, and it's wrong in a specific and dangerous direction: it always underestimates the peak stress the joint actually experiences.

The reason is that a bonded lap joint doesn't share load uniformly across its overlap. The two adherends being joined have finite stiffness, and as load transfers from one adherend to the other through the adhesive layer, the strain in each adherend changes along the length of the joint. That differential straining concentrates shear stress at the two ends of the overlap and leaves the middle of the bond line doing comparatively little work. E.O. Volkersen worked out the closed-form solution for this back in 1938, and it remains the standard first-pass hand calculation for lap joint design.

Setting up the Volkersen shear-lag model

The key insight is that load transfer through the adhesive depends on the relative stiffness of everything in the joint: the adhesive's shear modulus Ga, the two adherend moduli E1 and E2, and their respective thicknesses t1 and t2, plus the overlap length L over which all this happens.

Volkersen defines a shear-lag parameter, often written omega, that combines these:

omega^2 = (Ga / ta) x [ 1/(E1 * t1) + 1/(E2 * t2) ]

where ta is the adhesive layer thickness. This parameter has units of 1/length^2, and physically it describes how quickly load transfer concentrates near the joint ends versus spreading evenly. A stiff adhesive relative to thin, compliant adherends produces a sharply peaked distribution; a soft adhesive relative to stiff, thick adherends spreads load more evenly.

The mean shear stress across the bond is the naive value everyone starts with:

Mean tau = P / (w x L)

The actual peak shear stress at the ends of the overlap is higher by a stress concentration factor K that depends on omega and L:

Peak tau (end) = K x Mean tau

where K grows with omega*L — longer overlaps or stiffer adhesive-to-adherend stiffness ratios push more of the concentration toward the very ends of the bond, which is exactly why simply making the overlap longer gives diminishing returns. Past a certain length, the middle of the joint carries almost no additional load, and you're just adding dead weight and cost without proportionally increasing strength.

A worked lap joint

Take a single-lap joint bonding two aluminum adherends (E1 = E2 = 70 GPa) with matched thickness t1 = t2 = 2 mm, using a structural epoxy with adhesive shear modulus Ga = 1.2 GPa and a nominal adhesive layer thickness of 0.2 mm. The joint width w is 25 mm, overlap length L is 20 mm, and it's carrying an applied load P of 8,000 N.

First, the mean shear stress, the number a naive calculation would stop at:

Mean tau = P / (w x L) = 8000 / (25 x 20) = 8000 / 500 = 16.0 MPa

If the adhesive's rated shear strength is, say, 25 MPa, that mean stress alone would suggest a safety factor of about 1.56 — seemingly fine.

Now compute the shear-lag parameter. Using consistent N/mm units:

1/(E1*t1) + 1/(E2*t2) = 2 x [1 / (70,000 MPa x 2 mm)] = 2 x [1/140,000] ~= 1.4286e-5
omega^2 = (Ga/ta) x 1.4286e-5 = (1200/0.2) x 1.4286e-5 = 6000 x 1.4286e-5 ~= 0.0857 (1/mm^2)
omega ~= 0.293 mm^-1

With overlap length L = 20 mm, the dimensionless group omega*L ~= 5.86 — well into the regime where load transfer is sharply concentrated at the ends rather than spread across the bond. For omega*L in this range, the Volkersen stress concentration factor K works out to roughly 3, meaning:

Peak tau (end) ~= 3 x 16.0 MPa = 48.0 MPa

That's nearly double the adhesive's 25 MPa rated shear strength. The joint that looked like it had a comfortable 1.56 safety factor under the naive mean-stress calculation is actually running with a safety factor under 0.6 at the overlap ends, meaning it's expected to yield or crack right at the joint edges well before the applied load reaches its nominal design value. This gap between mean and peak stress is exactly why a proper joint calculation reports safety factor based on peak tau, not mean tau — using the mean-stress figure for a go/no-go decision here would sign off on a joint that fails.

Design implications and where linear theory runs out

The practical lesson from Volkersen's model is that overlap length has diminishing structural value past a certain point, roughly where omega*L exceeds about 4 to 5, while adherend and adhesive stiffness matter enormously. Two adherends with very different stiffness, say a stiff steel bracket bonded to a compliant composite panel, will always show worse stress concentration than two matched, symmetric adherends, because the stiffness mismatch is exactly what forces rapid load transfer near the joint ends.

The Volkersen model itself is linear-elastic and ignores adhesive plasticity, which is where Hart-Smith's later extension comes in: real structural adhesives yield locally near the high-stress ends before the joint actually fails, which redistributes some load back toward the center of the bond and makes the joint somewhat more forgiving than the pure Volkersen peak stress suggests. It also ignores peel stress entirely — bending-induced tension perpendicular to the bond plane, which for a single-lap (as opposed to double-lap) joint geometry is often the actual failure mode, not shear. A full joint design check needs both the shear analysis above and a separate peel-stress check.

Finally, this whole model assumes the adhesive layer thickness ta and properties are uniform and well controlled. In practice, bond-line thickness variation, voids, and surface preparation quality often dominate real-world joint strength more than the idealized stiffness numbers feeding into omega. The calculation gives you the mechanics-driven upper bound on performance, not a guarantee that a poorly prepared bond will reach it.

Why the fix is rarely "make the adhesive stronger"

Faced with a peak stress that exceeds allowable, the instinctive fix is to switch to a higher-strength adhesive. That can help, but it's often the wrong lever to pull first, because a stiffer adhesive (higher Ga) actually increases omega, which increases the stress concentration factor K and pushes the peak even higher relative to the mean, partially offsetting the benefit of the higher raw strength. A softer, more compliant adhesive with a lower shear modulus can sometimes outperform a "stronger" one in this specific joint geometry, because it spreads load more evenly across the overlap even though its nominal strength rating is lower.

The more reliable levers are on the geometry and adherend side. Increasing joint width w reduces both mean and peak stress linearly and doesn't change the stress concentration factor at all, which makes it the most predictable way to add margin. Tapering or stepping the adherend thickness near the ends of the overlap, a common move in aerospace bonded joints, reduces the local stiffness mismatch exactly where the peak stress occurs, lowering K directly rather than fighting it with a stronger adhesive. And matching adherend stiffness on both sides of the joint, when the application allows it, keeps the load transfer symmetric and avoids the worse concentration that comes from bonding a stiff material to a compliant one.

None of this replaces testing. Volkersen and Hart-Smith give you a defensible first-pass number and, more importantly, tell you which direction to push a design before you commit to coupon testing or a full qualification program. Treat the calculated safety factor as a screening tool: a joint that fails this hand calculation by a wide margin needs a redesign before it's worth the cost of physical testing, while one that passes comfortably still deserves validation against the specific surface preparation and cure process actually used in production.

Try it yourself

If you're sizing a bonded joint and want to see how far the peak stress diverges from the naive mean-stress estimate for your actual geometry and materials, you can try the adhesive joint stress calculator here. It runs the Volkersen and Hart-Smith models against your applied load, adherend properties, and adhesive shear modulus, and returns peak and mean stress, the stress concentration factor, and a safety factor. If your joint also carries significant bending, the beam-column tool is a useful companion check for the combined loading case.