Why TRADA's connection tools disagree with EC5:2004 - a worked comparison

Many UK engineers still reach for the TRADA Nail, Screw, Bolt & Dowel Joint Designer (v2.04, 2004-08) - or its online successor, timberconnectionsPro - when checking a single fastener. For timber-to-timber connections the desktop tool still produces numbers that match the current standard almost exactly. For steel-to-timber connections it can over-predict design capacity by roughly 50%.

This article shows both cases with full working, explains the three changes to the standard that cause the difference, and sets out what it means if the tool is still part of your workflow.

The short version

TRADA's desktop Joint Designer implements the ENV-era draft Eurocode (its own reports print the design basis as DD ENV, 1994). The current standard is EN 1995-1-1:2004+A2:2014, and the 2004 conversion from draft to full standard changed the steel-to-timber rules in three specific ways: it introduced interpolation for intermediate plate thicknesses, added a bearing cap to the thick-plate failure modes, and reduced the fastener yield moment formula. Timber-to-timber equations survived the conversion essentially intact - which is why the tool still agrees there and disagrees on steel plates.

Case 1 - timber to timber: agreement to 0.02%

A single M12 grade 8.8 bolt (fu = 800 N/mm²), two C24 members 30 mm thick each, service class 2, medium-term duration, load parallel to grain.

Working to EN 1995-1-1:2004, embedment strength (Eq 8.32) gives fh,k = 0.082 × (1 − 0.01 × 12) × 350 = 25.26 N/mm². The governing failure is mode (c) of Table 8.2 - rigid rotation of the bolt without a plastic hinge:

Fv,Rk = (fh,k · t1 · d)/(1+β) · (√(β + 2β²(1+r+r²) + β³r²) − β(1+r)) = 3766 N

Note that mode (c) carries no rope-effect contribution - the Fax,Rk/4 term in Table 8.2 appears only in the plastic-hinge modes (d), (e) and (f).

Design value: Fv,Rd = kmod × Fv,Rk / γM = 0.8 × 3766 / 1.30 = 2318 N.

TRADA v2.04 for the identical inputs: 2318 N. ConnForge: 2317.6 N. Agreement to 0.02% - the timber-to-timber equations are the same in ENV and EN, and both tools implement them correctly.

Case 2 - steel plate to timber: a 49% disagreement

Same bolt, one C24 member 30 mm thick, and a 10 mm steel plate. Same service class and duration.

With t = 10 mm and d = 12 mm, the plate sits between the thin-plate limit (t ≤ 0.5d = 6 mm) and the thick-plate limit (t ≥ d = 12 mm).

EN 1995-1-1:2004, §8.2.3(1) requires linear interpolation between the thin- and thick-plate mode sets for this case. The thin set governs at 3637 N, the thick set at 9092 N; interpolating at (10 − 6)/(12 − 6) = 0.667 gives a characteristic capacity of 7274 N, and a medium-term design value of

Fv,Rd = 0.8 × 7274 / 1.30 = 4476 N.

TRADA v2.04 for the identical inputs: 6688 N - 49% higher.

Where the difference comes from

Back-calculating TRADA's characteristic value: 6688 × 1.30 / 0.8 = 10,868 N. Three observations follow, each traceable to a change made when the ENV draft became the EN standard. (The reconstruction below is inferred from the tool's outputs - the source code isn't published - but the numbers line up to within about 2%.)

1. No intermediate-plate interpolation. The ENV draft had no provision equivalent to §8.2.3(1). A 10 mm plate with an M12 bolt is treated as thick, full stop. That alone moves the answer from the interpolated 7274 N toward the thick-plate set.

2. No bearing cap in the thick-plate modes. TRADA's implied characteristic of 10,868 N exceeds the pure bearing capacity fh,k × t1 × d = 25.26 × 30 × 12 = 9092 N. Under EN 1995-1-1:2004 Table 8.2 that is impossible - the thick-plate set includes a bearing mode that caps capacity at exactly 9092 N for these inputs. The ENV thick-plate set contained only the hinge modes; the bearing cap was added in 2004.

3. A larger yield moment. The ENV draft used My,k = 0.8 × fu × d³/6 = 184,320 N·mm for an M12 grade 8.8 bolt. EN 1995-1-1:2004 Eq 8.30 gives My,Rk = 0.3 × fu × d^2.6 = 153,432 N·mm - about 17% lower. The exponent change was a deliberate recalibration that reduced yield moments for larger-diameter fasteners. Running the one-hinge thick-plate mode with the ENV yield moment reproduces roughly 10,600 N - within about 2% of TRADA's implied value, with the residual attributable to rounding and national-application details.

Stacked together: no interpolation, no bearing cap, and a fatter yield moment turn 4476 N into 6688 N.

What this means in practice

If you use the v2.04 desktop tool for timber-to-timber single-fastener checks, its answers still match the current standard closely. If you use it for steel-to-timber connections - particularly with plate thicknesses between 0.5d and d, which covers the 8-10 mm plates common in brackets and flitch details - its design values can sit far on the unsafe side of EN 1995-1-1:2004. The 2004 revisions specifically tightened these cases; a tool implementing the withdrawn draft cannot reflect that.

It is worth being precise about the status: the ENV documents were draft standards intended for provisional use, formally superseded when EN 1995-1-1 was published in 2004 and adopted with the UK National Annex. The desktop tool was right for its era. The era ended two decades ago.

The online successor

TRADA's browser-based replacement, timberconnectionsPro (EC5 edition), is stated to be based on the current Eurocodes with the UK National Annexes. It sits behind a TRADA login; the free tier is limited to strength classes C16 and C24, with the full material range on the paid/member tier. Following the 2022 separation of Timber Development UK and BM TRADA, the tool remains on the trada.co.uk estate. We have not been able to run the comparison cases above through it, so we make no claim about its results either way - if you have access, the two cases in this article take five minutes to check, and we would genuinely like to hear the outcome.

How ConnForge verifies its numbers

ConnForge's single-fastener calculators implement EN 1995-1-1:2004+A2:2014 and are verified three ways: against published worked examples from Porteous & Kermani's Structural Timber Design to Eurocode 5 (the bolt, screw and nail calculators each carry a book-anchored regression test, matching to within 0.2%); against the TRADA tool where the standards coincide (Case 1 above); and against the printed standards themselves - the material database is pinned test-by-test to the tabulated values of EN 338:2016 and EN 14080:2013. When we find and fix an error in our own engine, it is published in the changelog with the direction and magnitude of the change.

All four calculators - bolt, screw and coach screw, nail and dowel - are free to use, with PDF export and no login.