Vessel Upending & Tailing Lift Calculator - methodology & sources

The subject is the manoeuvre nobody's spreadsheet models: a long body - a pressure vessel, a column, a pile, a monopile, any fabrication with a declared weight distribution - rotated from horizontal to vertical between a main crane and a tail support. The tool solves the statics at every inclination, finds the worst angle for each check separately, resolves the load direction on each attachment as it sweeps, and hands the governing load case to the shipped attachment calculators rather than re-deriving their mathematics. Every factor between the declared weight and any reported utilisation sits in a visible ledger.

The counts are read from the engine's own check register at this render. The hand-worked fixture is the colinear degenerate case - the product's thesis expressed as an assertion, re-run against the engine on every build: with every point on the axis and plumb lines, the engine's hook loads at 0°, 37° and 80° all equal the lever-arm answer 317.6 kN, agreeing to floating-point precision (the spread across the three evaluations is below a micronewton at this render) - the inclination plays no part, which is exactly why the textbook case misleads.

The body frame measures the axial station $s$ from the tail end and the radial offset $r$ from the axis. The world frame is horizontal $X$ in the lay direction and vertical $Z$ up, with the inclination $\phi$ the angle of the vessel axis above horizontal - 0° lying down, 90° vertical. A point at station $s_a$ with radial offset $r_a$ sits at:

The minus sign in $X_a$ is the entire subject. A lift-side attachment's horizontal lever arm about the tail shrinks as the vessel rises, and at 90° the attachment sits at $X=-r_a$ - offset to the side away from the lay direction, which is what drives the residual hang tilt after transfer. Flipping the sign is not a small error: on the worked example with the top lift point offset 300 mm to the lift side, the tail unloads in a transfer window of 87.9°–88.6° and the main hook peaks at 600 kN; with the same offset on the ground side there is no transfer at all (the suite pins this) - the tail is still loaded at vertical and the main hook never exceeds 500 kN. Both runs are the engine's, at this render. A sign-flip test in the suite asserts that flipping $r_a$ moves the governing inclination, so a silent regression cannot survive the build gate.

Two-crane upend (both lines plumb)

Moments about the tail lift point in world coordinates give the main hook load; vertical equilibrium gives the tail:

On the worked example the flat-position share is 355.1 kN on the main hook and 244.9 kN on the tail - the lever-arm answer, valid at exactly one inclination because the tail lug stands off the axis.

The colinear textbook case - everything on the axis - makes both lever arms proportional to $\cos \phi$, the cosines cancel, and the loads are independent of the inclination. The engine carries that limit as an algebraic branch rather than a numerical accident. When the two plumb lines coincide while the CoG is off that line, the load share has no statically defined value: the tool reports the degeneracy in words and refuses to print a number, because near that configuration the two hook loads genuinely diverge and any number would be an artefact of how close you looked.

Tail on the ground, tail on a dolly

With the tail skidding on the ground the unknowns are the hook force, the contact normal and the horizontal resistance at the contact. The resistance is Coulomb-type, $R=\mu N$, signed by the declared direction of motion, and reacted through the hook-line lean - which is how axial compression enters the vessel, invisibly to every hand calculation that assumes a plumb hook line:

Because no single resistance coefficient is defensible, the tool requires a declared band and evaluates its corners - never a midpoint. On the worked example skidded at 30° with μ ∈ [0.30, 0.50], the hook force envelope is 398–424 kN, the contact normal 187–207 kN, and the hook line leans 9.0°–12.7° from vertical. Every downstream quantity carries the envelope. The dolly variant is the same statics with the contact raised to the declared interface height and a rolling-resistance band in place of friction - the raised contact changes the moment arms, and with them the sign of the axial force the restraint induces.

Free-hanging

After the tail lets go, one hook carries everything and the body rotates until the lift point is plumb above the CoG. The tool solves that attitude from the equilibrium condition - it never imposes verticality - and reports the residual tilt, which is the number fabricators actually ask for. An attachment coincident with the CoG leaves the attitude with no defined value and is refused in words.

Closure, asserted every run

Every solved state is closed by an equilibrium check whose two residuals are normalised dimensionless - $|\Sigma F|/W$ and $|\Sigma M|/(WL)$ - so a force is never compared against a moment. Both must sit at solver precision or the run fails.

Found, never interpolated

The sweep evaluates the full statics on a grid (default 1°, declared, 0.05°–15°), then refines around each realisation's best grid candidate by re-evaluating the physics at each probe - on the worked example the tailing-lug maximum is locally pinned to ±0.0010°. The value reported at the governing angle is always an evaluated sample, never a fit between two grid points, and the report states both figures honestly: the ± bracket is local to the best candidate, and global localisation is only as fine as the grid - which is why the chart draws the whole band rather than one point. For context, DNV-ST-N001 (2018, amended 2020-01) §16.3.2.7 expects upending manoeuvres to be analysed in steps normally no coarser than 15°, finer near critical stages; the default grid is more than an order of magnitude finer, and that clause is cited in the report as documentation, not as the basis of any factor.

Finite by construction

Near the configuration where both plumb lines coincide, the two-crane loads genuinely diverge - there is a mathematical pole. The sweep does not chase it. A swept state counts as feasible only while the arrangement physically holds: both hook loads non-negative in the two-crane arrangement, a valid band corner with a non-negative contact normal and a non-negative hook load in the tail-supported ones. The pole can only be approached through infeasible states - one load goes negative first - so every governing value reported over the feasible domain is finite by construction, not by numerical luck. Where a governing point sits against the edge of the feasible range, the check says so explicitly.

The transfer point is an event, not a failure

The angle where the tail reaction reaches zero is found by bisection on the re-evaluated statics, per CoG realisation - and it is a per-realisation event, not a universal one. On the worked example 2 of 5 realisations transfer inside the sweep, between 89.70° and 89.72°; the others still carry up to 25 kN of tail load at vertical, where the tail must be actively unloaded rather than waited out. Where a realisation does pass its transfer, its two-crane feasible domain ends there - the vessel wants to hang - and the tool reports that as an arrangement change to plan for; a negative tail load is never printed as if it were a result.

Boxes are swept, never averaged

The CoG uncertainty box translates the whole declared weight distribution rigidly - attachments stay put - so the statics and the internal forces stay mutually consistent within each realisation. Every reported quantity is an envelope over the realisations, and the suite asserts monotonicity: widening the declared box or the resistance band can never narrow a reported envelope.

For each attachment the engine resolves the solved rigging force into the attachment's own frame as $(F,{\theta }_{ip},{\theta }_{oop})$ - magnitude, in-plane angle, out-of-plane angle - at every swept state, runs the lifting-lug pack's full check set on that vector, and governs it over the sweep with the same non-interpolating search as everything else - scored by the lifting-lug page's own verdict rule (the selected route's family, mechanics as fallback), so the hand-off reproduces the verdict. One lug implementation serves both the lifting-lug calculator and this pack's delegation: the delegated rows carry the lifting-lug registry's own clause citations - the governing clauses among them being ASME BTH-1-2020 §3-3.3 pin-region checks, §3-2.3–§3-2.5 base-section member checks with the §3-1.3 design factor, §3-3.4.3 weld allowables, EN 1993-1-8:2005 §3.13 pinned-connection and §4.5.3 weld methods, and the named mechanics identities; the cross-check rows additionally cite EN 1993-1-1:2005 §6.2.1(5) (the base yield criterion) and AISC 360-22 §J2.4 (the weld cross-check) - and this pack re-derives none of them. The weld checks ride along inside the same delegation, so this pack adds no weld mathematics of its own.

On the worked example the delegated tailing lug governs at φ = 70.4° with ${\theta }_{oop}$ = 70.4° and F = 224.7 kN - utilisation 0.76 across 22 delegated rows. Neither the flat position nor the vertical position governs; the worst case is in the middle of the lift, where the out-of-plane bending on the lug base is near its peak while the load has barely shed. That combination is what hand calculations checking one angle miss, and it is why tailing lugs - not lifting lugs - are the detail that fails. DNV-ST-N001 §16.9.2.2 states the same doctrine: design lift points for the most onerous possible sling force direction.

Honesty at the interface

• Domain holes, never clamped. A swept state whose resolved vector leaves the lifting-lug pack's validated domain (for example an in-plane angle past 90° under a leaned hook line) is excluded from the governing search as infeasible and reported - the input is never clamped to make the delegate answer.
• Declared spec or no numbers. An attachment without a declared geometry, material and route reports “not evaluated” - the pack never invents an attachment capacity.
• The hand-off is the same code path. Every delegated row set carries a link that opens the governing load case in the lifting-lug calculator with the solved vector pre-filled, built by the same input builder the delegation itself uses - a round-trip test asserts the lug page reproduces the delegated utilisation exactly.
• The local shell is computed where its source reaches. A TRUNNION-kind attachment's delegated row set includes the lifting-trunnion pack's WRC 537 local shell assessment, within that kernel's validated β/γ envelope - outside it the row refuses rather than extrapolates. Local shell stress under a lug-type attachment is not computed and arrives with the nozzle pack's shared kernel work; the assumption register says so rather than leaving silence where a check should be.

At each swept state the solved support forces and the inclined self-weight are integrated piecewise-exactly along the axis - stations at every breakpoint, both sides of every point action - including the applied moments $m=-r{f}_a$ that radially offset axial components put into the beam. Closure past the far end is asserted. Fibre stresses use the exact annulus, $\sigma =N/A\pm M/S$, peak shear the thin-annulus identity ${\tau }_{max}=2V/A$, and the closed-vessel longitudinal pressure stress ${\sigma }_L=p{D}_i^2/(D_o^2-D_i^2)$ joins the combination only when pressurised-during-lift is explicitly opted into - the erection condition defaults to unpressurised and the report states it.

Local buckling - the one coded acceptance on the body. A thin shell in longitudinal compression does not fail by yielding, and that is exactly the population this tool serves: long, thin vessels picked near horizontal. ASME BPVC Section VIII, Division 1 (2025) UG-23(b) gives the limit, and the buckling row implements Steps 1–5 of it. Factor $A=0.125/(R_o/t)$ from the shell's own proportions; factor $B$ read off the Section II, Part D, Subpart 3 material/temperature chart at that $A$ (or $B=AE/2$ where $A$ falls to the left of the line); and the allowable is the smaller of $B$ and your declared allowable tensile stress, exactly as the clause writes it. The chart, the erection metal temperature and the modulus are declarations - the tool derives none of them, and without them the row reports what it needs and computes nothing rather than assuming a material. It is the same chart machinery the pressure-vessel calculator reads for external pressure, shared rather than reimplemented; the two packs enter the charts by different doors and read $B$ off the same curve.

A stability limit does not replace a strength check. The buckling row and the bending row report the same governing compressive stress against different acceptances - the code's limit and your declared basis - and the tool shows both rather than quietly substituting one. What UG-23(b) does not give is a separate bending-buckling allowable or an axial-plus-bending interaction rule: Step 5 compares $B$ against “the computed longitudinal compressive stress” without qualifying which action produced it. So the total extreme-fibre compression - lifting axial force and bending together - goes into that comparison. That is a direct reading of the step, and a conservative one, since the classical knockdown for bending-induced buckling is less severe than for uniform axial compression. A less conservative interaction treatment exists in other documents; it is not in VIII-1 and is not invented here. Two neighbouring provisions are named on the row so you know they were read: UG-23(c) permits membrane plus bending to reach 1½ times the allowable tensile value, which this tool does not apply because its allowables are your declarations rather than code-derived values; and UG-23(d)'s 1.2 factor belongs to earthquake or wind combinations, which an erection lift is not.

One uniform shell section (declared outside diameter and wall) runs the full length; stepped walls, cones, stiffening rings and local reinforcement are not sectioned in this release - declare the thinnest governing wall. The buckling row inherits that limitation: a ring-stiffened shell has a shorter effective length than the row assumes.

The overlay resolves three factors from DNV-ST-N001 (2018, amended January 2020 - the 2023-12 edition is not held, and every citation states the edition used):

The application map is tested factor by factor, per hook: each hook × its own DAF; each sling × its own hook's DAF·SKL; each delegated attachment × its own hook's DAF·SKL·γ_c - 1.43 on the worked tailing lug - applied to the load magnitude before delegation so the delegated rows check the factored demand with the direction untouched, and disclosed on the row beside the solved force; body × the larger of the two DAFs, because the beam carries both hooks at once. The overlay also applies §16.9.3.1's default lateral load: each attachment's factored out-of-plane demand is floored at 3% of its side's factored maximum sling force, simultaneous with the in-plane load - an in-plane-loaded lift point is never checked with zero out-of-plane demand. The pressure term is never amplified - internal pressure is not a crane dynamic. On the worked example the overlay moves the main hook from utilisation 0.80 to 0.88 and the tailing lug from 0.76 to 1.09 - past unity, which is the overlay doing its job: the factored design case is a harder case, and the tool shows you exactly where the margin went.

Admissibility is cited, not assumed: §16.2.5.6 is the clause that extends Table 16-1 to lifts by two cranes on one vessel and to onshore lifts by two or more cranes. Offshore lifts by two or more vessels need operation-specific analysis per §16.2.5.3, and the overlay points there rather than defaulting a number. Weight and CoG contingency (§16.2.2) remain the project's weight-control responsibility and enter through the declared weights; tilt and yaw tolerances (§16.2.3–16.2.4) are the lift plan's and are not modelled.

EN 13001-2 §4.2.2.2 is the clause that matches an upend directly. When an unrestrained grounded load is hoisted, the dynamic effect of transferring it off the ground onto the crane is taken into account by multiplying the gravitational force due to the mass of the hoist load by a factor ${\varphi }_2$, and the clause defines that mass to include the payload, the lifting attachments and a portion of the suspended hoist ropes. A vessel coming off its stands, off the ground or off a dolly is that load. By equilibrium the amplified force passes through the rigging into the vessel, so nothing in the load path escapes it.

Unlike DNV's dynamic amplification factor, which is banded on the static hook load, ${\varphi }_2$ is a property of the machine and the system: the stiffness class (Table 2, from the characteristic vertical load displacement), the hoist drive class (from how the drive picks the load up), and the hoisting speed the drive actually runs at. On the worked example at the shipped starting classes - HC2 stiffness, HD1 drive, load combination A1 - ${\varphi }_2$ resolves as 1.10 + 0.34 s/m × 0.050 m/s = 1.117. The stiffness class comes from a displacement this tool does not compute and the drive class from the drive's own control behaviour, so both are your declarations about the crane, not something the vessel geometry can imply.

Where the factor lands is not uniform, and the difference is stated on every row that carries it. The hook and sling rows take ${\varphi }_2$ alone - 1.117 here. The body and the delegated lift points additionally carry the partial safety factor ${\gamma }_p$ (Table 12a) and the risk coefficient ${\gamma }_n$ (§4.3.2), giving 1.497. ${\varphi }_2$ amplifies a force that is really there, so it reaches everything; ${\gamma }_p$ and ${\gamma }_n$ are proof-of-competence factors that pair with the resistance coefficient ${\gamma }_m$ on the other side of a structural proof, and a hook row is not a structural proof - it compares against a capacity read off a crane load chart, which already carries its own margin. If a project wants a factor there, it is declared as the multi-crane de-rate; the tool will not invent one.

The proof method is not a preference. EN 13001-1 §4.2.7 gives two proofs, and they are the same proof with the factor on different sides: the limit state method puts ${\gamma }_p$ on the demand and divides the resistance by ${\gamma }_m=1.10$; the allowable stress method folds the partial safety factors into an overall ${\gamma }_f=1.48$ that divides the resistance instead. The two are close without being equal, and that is the standard's choice rather than rounding: Table 12a states ${\gamma }_f$ as its own value - 1.48 here, where 1.34 × 1.10 would be 1.474. The stated cell is what this tool reports; deriving it from the other two would invent a coefficient the table already gives. This tool checks against allowables you declare and cannot tell which side of that identity a declared number already sits on, so you say which, and the row states what your declared allowable then has to be. Picking the allowable stress method does not apply ${\gamma }_p$ to the demand at all.

On the worked example the route moves the main hook from utilisation 0.800 to 0.894, the body bending row from 0.158 to 0.237, and the tailing lug from 0.759 to 1.136.

Two edition gaps, stated rather than implied. EN 13001-1:2004 is held without its A1:2009 amendment, and the held EN 13001-2 is the 2014 edition, which is superseded by EN 13001-2:2021 - not held. Every row that reaches a report repeats both, because a reader who meets one row and not the others still has to know which editions produced the number. Check them against the editions your project is contracted to.

Why this route is not built on EN 13155. That standard's scope (§1) is non-fixed load lifting attachments - plate clamps, vacuum lifters, magnets, lifting beams, C-hooks, forks, clamps and concrete lifting inserts - and it states that slings are not covered. A lifted vessel and its own welded lugs or trunnions are not on that list, so its 2× and 3× design-load coefficients (§5.1.2.1) and the risk coefficient 1.4 it fixes for attachments above 16 000 load cycles (§5.1.2.2) are named here and deliberately not inherited by the vessel. What EN 13155 does contribute is the route itself: §5.1.2.2 is the clause that sends an attachment's proof into the EN 13001 chain implemented above. A top pick that genuinely runs through a lifting beam or spreader is in EN 13155's scope - that item belongs to the spreader-beam calculator, not this page.

Two more boundaries, so the routes are not mistaken for interchangeable. The EN route applies no out-of-plane minimum: EN 13001-2 carries no provision equivalent to DNV §16.9.3.1's default lateral load on lift points, and this pack does not invent one - so an attachment moved from the DNV overlay to the EN route loses that floor, and the row says so. And the B load combinations are not offered: B1 is A1 with in-service wind and other environmental actions added, carrying the lower partial safety factor 1.22 precisely because that wind action is present. This pack models no wind action, so offering B1 would hand out the reduced factor without the load that earns it.

• Crane charts and crane selection. Declared hook capacities only, permanently. A transcribed chart would invite trust in the transcription over the chart in the cab. Ground bearing under the cranes is the crane ground bearing tool's subject.
• Wind during the lift. Real and sometimes governing; the projected area of an inclined cylinder plus a shielding judgement is a modelling exercise this release does not pretend to contain.
• Dynamics beyond the declared factors. The sweep is quasi-static; dynamics enter as the DNV DAF or a declared factor, visibly.
• Sling and shackle capacity design. Declared WLLs are checked against; hardware capacity is never invented. The sling-angle check itself (bridle geometry, D/d) belongs to the multi-leg sling-geometry work - with plumb single-fall lines there is no angle to check, and the tension row states that the line tension equals the hook load until bridles are modelled.
• Buckling beyond a uniform unstiffened shell. UG-23(b) is implemented on the declared uniform section (see §06), but stiffening rings, cones, stepped walls and local reinforcement are not sectioned - a ring-stiffened shell is stiffer than the row assumes. Nor is there an axial-plus-bending interaction rule: VIII-1 gives none, so the total compression goes into the one comparison the clause writes, which is the conservative reading.
• Local shell stress under lug-type attachments. Arrives with the nozzle pack's shared WRC 537 kernel work; stated in the assumption register. (Trunnion-kind attachments already carry the local shell assessment inside their delegated row set.)
• Boom and hook-block clearance. The clearance tab answers the two questions the declared geometry can answer - ground interference of the tail rim for the tail-supported arrangements, and the swing envelope (far-end reach, tail-rim drop below the pin). Rows needing the crane's own geometry say they need it.
• Vessel transport, lashing and saddle design. Different problems, different tools.

Two routes remain, and both are blocked on acquisition, not on engineering time. The Australian route is blocked because the held AS 4991-2004 file is an 8-page sample missing the operative sections. The Canadian route is blocked twice over: CSA S16-09 is held but superseded by S16:24, and the load-factor source (the National Building Code of Canada) is not held - a further trap is recorded against it, because the file in the library named as S16-14 is a vendor software manual, not the standard. Neither will compute so much as a coefficient until the standards themselves are in hand. Explaining a placeholder is this page's job; computing one is forbidden.

The European route used to sit in this table. It no longer does: it is implemented against the held EN 13001 chain and is documented in section 08 above, with both of its edition gaps stated. That is the only honest way for a route to leave this list.

The lifted body is treated as rigid for load distribution - the industry-standard treatment, normally conservative (DNV-ST-N001 §16.8.6.1 records the assumption and its exceptions). Slender or torsionally soft bodies and lifts with more than four lift points can redistribute loads through deflection and need a stiffness model, which this version does not provide.

The analysis is quasi-static: the manoeuvre is a sequence of equilibrium states. Dynamics enter only as declared factors (a DAF overlay arrives with the DNV route in a later phase); no dynamic simulation is performed.

Hook lines are plumb unless the arrangement itself implies a lean: in the tail-supported arrangements the hook-line lean is derived from the tail resistance ($\tan \beta =P_x/P_z$) and reported. Declared offlead/sidelead inputs arrive with the sling-geometry phase.

The model is planar: the vessel axis, the CoG offset, the attachments and all forces lie in the vertical lift plane. Out-of-lift-plane CoG offsets and sling components (yaw, sidelead) are not resolved in this version.

Crane capacities are declared hook capacities at the relevant radius, taken from the manufacturer's load chart by the user. The tool never replaces the load chart, and the multi-crane de-rate is a declared project rule (visible in the check, default 1.0) - many owners require 0.75–0.80 for multi-crane lifts.

The tail resistance coefficient (sliding friction or rolling resistance) is never a single defensible number: a declared band is evaluated at its corners and every derived quantity reports the envelope. The band's sign follows the declared direction of motion (upending drags the tail toward the hook plumb line).

The DNV-ST-N001 §16 overlay applies demand-side factors per hook: each hook × its own DAF, each sling and delegated attachment × its own hook's DAF·SKL(·γ_c), and the body × the larger of the two DAFs (the beam carries both hooks at once; γ_c = 1.00 for other members). Table 16-1 is a function of the static hook load, so each hook's DAF is read at the load that hook is checked at: its own governing static hook load, and for the delegated attachments the hook load at the state being evaluated. The table's note 1 still raises the band entry to the 3 t row wherever a checked load is genuinely below 3 t. The §16.9.3.1 default lateral load is applied: each delegated attachment's factored out-of-plane demand is floored at 3 % of its side's factored maximum sling force, simultaneous with the in-plane load, so an in-plane-loaded lift point is never checked with zero out-of-plane demand. Weight/CoG contingency factors (§16.2.2) are the project's weight-control responsibility and enter through the declared weights; tilt and yaw effects (§16.2.3/16.2.4) are not modelled and their tolerances remain the lift plan's.

The local-buckling acceptance is ASME BPVC Section VIII, Division 1 (2025) UG-23(b): the maximum allowable longitudinal compressive stress is the smaller of the allowable tensile stress and factor B, with $A=0.125/(R_o/t)$ and B read from the Section II, Part D, Subpart 3 material/temperature chart (or $B=AE/2$ where A falls left of that line). The chart, the erection metal temperature and the elastic modulus at that temperature are DECLARED - the app derives none of them, and without them the row computes nothing rather than assuming a material. The stress compared is the total extreme-fibre longitudinal compression from the lifting axial force and bending together, because Step 5 compares B against the computed longitudinal compressive stress without qualifying which action produced it; that is conservative, since the classical knockdown for bending-induced buckling is less severe than for uniform axial compression, and the held edition offers no interaction rule to use instead. The check is a STABILITY limit and stands beside the strength check against your declared compressive basis - neither supersedes the other. UG-23(c)'s $1\frac{1}{2}$ membrane-plus-bending allowance is not applied (the allowables here are declarations, not code-derived values, so the bending row is conservative against it), and UG-23(d)'s $1.2$ factor is not applied because it belongs to earthquake or wind combinations rather than to an erection lift. One uniform shell section is assumed, as everywhere else in this pack: stiffening rings, cones, stepped walls and local reinforcement are not sectioned, and the buckling row inherits that limitation - declare the thinnest governing wall.

The EN 13001 route applies demand-side factors from the crane-design chain, off by default and mutually exclusive with the DNV overlay (both factor the same demand; the combination is refused, never resolved silently). The hoist-load dynamic factor ${\varphi }_2$ for hoisting an unrestrained grounded load (EN 13001-2:2014 §4.2.2.2, Formula (3) with Tables 2, 3 and 4) multiplies EVERY demand - hook loads, sling tensions, delegated lift-point loads and body stresses - because the clause amplifies the gravitational force due to the mass of the hoist load, and by equilibrium that amplified force passes through the whole load path. The partial safety factor ${\gamma }_p$ (Table 12a) and the risk coefficient ${\gamma }_n$ (§4.3.2) additionally multiply the STRUCTURAL proofs only: they pair with the resistance coefficient ${\gamma }_m$ on the other side of a proof of competence, and the hook and sling rows compare against a declared crane chart capacity and a declared sling WLL, each already carrying its own margin - a project that wants a factor there declares it as the multi-crane de-rate. Whether ${\gamma }_p$ is applied at all follows the declared EN 13001-1 §4.2.7 proof method, because the allowable stress method folds it into the overall safety factor ${\gamma }_f$ on the resistance side instead, and this pack checks against allowables the user declares and cannot tell which side of that identity a declared number already sits on. Editions: EN 13001-1:2004 held WITHOUT A1:2009, and BS EN 13001-2:2014, superseded by :2021. EN 13155:2020 is cited for scope only - its own scope is non-fixed load lifting attachments and it does not reach a lifted vessel or its welded lift points, so its 2×/3× coefficients and its ${\gamma }_n=1.4$ are named and not inherited. Wind during the lift, and the EN 13001-3-1 resistance side, are not part of this route.

Attachment capacities are computed by the shipped owning pack through the shared interface - lifting-lug for plate lugs, lifting-trunnion for radial trunnions - with the solved load vector resolved into the attachment's own frame at each swept state. The delegated rows carry their own clause citations and assumptions, and each verdict follows the owning page's own governing rule so the hand-off reproduces it. A trunnion-kind spec's row set INCLUDES the local shell assessment (WRC 537, within its validated β/γ envelope); for lug-type attachments the local shell stress arrives with the nozzle pack's shared kernel work and is not computed.

The body checks use one uniform cylindrical shell section (declared OD and wall) along the full length. Stepped walls, cones and local reinforcement are not sectioned in this version - declare the thinnest governing wall.

Body allowable stresses are user declarations: the tensile value is the ASME Section II-D allowable the user reads for their material at the erection temperature; the compressive value is the user's declared basis (for example from the VIII-1 external-pressure charts) until the buckling check lands in a later phase; the shear value is the user's basis. The app never derives a code allowable itself.

The weight distribution (segments and point masses) and the CoG uncertainty box are declared inputs. Weighing, weight-control class and contingency factors are the project's responsibility; the CoG box is swept, never averaged.