Vessel upending, done properly - the guide

A vessel arrives on the trailer horizontal and has to stand vertical. Between those two states is a continuous manoeuvre in which the main crane's share of the weight, the tail load, the bending in the shell and - above all - the direction of the force on every attachment all change with the inclination. An upending calculation answers, for every angle on the way up: what does each hook carry, what does the tailing attachment see and from which direction, what is the worst angle for each of those, and when does the tail hand its load over entirely.

The same statics covers far more than pressure vessels. The tool takes a declared weight distribution along an axis - so a column, a pile, a monopile upending calculation, a transition piece or any long fabrication is the same calculation with different numbers, and the bending checks work from the distribution rather than a point mass, which is what governs a long thin body.

Take moments about the tail lift point with everything on the axis. Both lever arms carry the same $\cos \phi$, the cosines cancel, and:

On the colinear benchmark the engine runs on every build -600 kN with the CoG 9 m from the tail lug and the pick point 17 m from it - the lever-arm answer is 317.6 kN, and the engine's hook loads at 0°, 37° and 80° are all exactly that number. The angle genuinely does not matter. That is why everyone believes upending is a one-line calculation: in the textbook case, it is. Every free calculator and most spreadsheets stop here.

Real lift points are not on the axis: a trunnion or lug stands on the shell, radially offset. A real CoG carries an uncertainty box, and its transverse component is rarely zero. The moment a point leaves the axis, its horizontal lever arm becomes

and the cancellation is gone: the load share now depends on $\phi$, and it does not vary monotonically - the governing angle can land anywhere between horizontal and vertical. The sign of the radial offset decides the endgame of the lift: on the worked example, offsetting the top lift point 300 mm toward the lift side makes the tail unload at 87.9° with the main hook peaking at 600 kN; the same offset toward the ground side means the tail never unloads and the main hook tops out at 500 kN. Same magnitude, opposite sign, different lift. The methodology page freezes the convention and draws it.

For a two-crane upend the load share follows from moments in world coordinates, per inclination. On the worked example - 600 kN (61.2 t) over 30 m - the flat-position share is 355.1 kN on the main hook and 244.9 kN on the tail crane. By the top of the lift the main hook carries 600 kN - the full weight - and the tail crane's governing load is 252.2 kN, found at the pick on a CoG-box corner: 3% above what the nominal CoG predicts, before any factor is applied.

Multi-crane load share is exactly this calculation with the de-rate applied to each declared capacity. The governing angle differs per check - the main hook governs at 90°, the tail at 0°, the tailing lug at 70.4° - which is why the tool reports each check at its own worst angle with the search resolution stated, instead of one table at one angle. The methodology page records how the sweep works and why its governing values are finite by construction.

A transverse tailing lug stands on the shell with its plate across the vessel axis. While the vessel is horizontal, the plumb tail load lies in the lug's plane - the strong direction. As the vessel rises, the load stays plumb but the lug rotates with the shell, so the angle between them is the inclination itself: ${\theta }_{oop}=\phi$. On the worked example the direction sweeps from 0° to 90° out of plane while the magnitude holds between 192 and 245 kN until the final degrees before transfer - at φ = 70° the lug still carries 216 kN, 88% of its flat-position load, pointed 70° off the plate.

A tailing lug calculation therefore has to check the lug with direction: out-of-plane bending on the base section, the weld group under the transverse component, and the pin-region checks under the resultant. This tool does not re-derive any of that - it resolves the solved force into the lug's own frame at every swept state and hands the vector to the shipped lifting lug calculator, whose ASME BTH-1-2020 base-section, pin and weld checks run on the factored demand and report clause by clause. On the worked example the delegated lug governs at φ = 70.4° with utilisation 0.76 across 22 rows - in the middle of the lift, where the direction is far out of plane and the load has barely shed. Neither end of the manoeuvre governs, and a one-angle check misses it entirely. The methodology page names every delegated clause.

On the worked example the transfer is not even universal across the CoG box: 2 of the 5 realisations reach zero tail load inside the sweep, between φ = 89.70° and 89.72°, while the others still carry up to 25 kN of tail load at vertical. That is the CoG box doing its job: depending on where the CoG really is, the tail either goes light just before vertical or has to be actively unloaded at the top. The tool locates each realisation's transfer by bisection on the re-evaluated statics and reports the envelope - plan the hand-off around BOTH outcomes, and expect the vessel to want to swing to its free-hanging attitude the moment the tail lets go. Past a transfer the tool refuses to print a negative tail load - a negative reaction is an arrangement change, not a result.

After release, the body rotates until the pick point is plumb above the CoG. The tool solves that hang attitude - it never assumes vertical - and reports the residual tilt, which is the number the fabricator asks about when the vessel hangs two degrees off plumb over the foundation.

With the tail on the ground, the contact resists sliding, and that resistance has to be reacted somewhere: the hook line leans, and an axial force enters the vessel that a plumb-line hand calculation never sees. Because no single friction value is defensible, the tool requires a declared band and evaluates its corners. On the worked example skidded at 30° with μ between 0.30 and 0.50, the hook force runs 398–424 kN and the hook line leans 9.0°–12.7° from vertical - every derived quantity carries the whole envelope, never the midpoint. A dolly raises the contact to its declared interface height, which changes the moment arms and can reverse the sign of the induced axial force; the calculation is the same with a rolling-resistance band.

A declared CoG carries a tolerance - from the weight take-off, from as-built deviations, from internals that moved between revisions. The tool takes that tolerance as a box (±300 mm axially, ±50 mm radially on the worked example) and runs the nominal plus all four corners through the entire sweep - 5 realisations of the whole manoeuvre. Every band on the chart, every governing value and the transfer window are envelopes over those realisations. The tail crane's governing load lands on a box corner, not the nominal - which is typical, and exactly what a point-CoG calculation cannot show.

Enable the overlay and the resolved factors appear in the ledger: DAF 1.10 on the main hook and 1.10 on the tail (Table 16-1, onshore column, each hook banded at the static hook load that hook is checked at), SKL 1.00 for the determinate single-fall arrangement (valid only with sling lengths within ±0.5% of nominal - the condition is printed with the factor), and a consequence factor of 1.30 on the attachment demands. On the worked example that moves the main hook from utilisation 0.80 to 0.88 and pushes the tailing lug from 0.76 to 1.09 - past unity. That is the overlay doing its job: the factored design case is a harder case, and a lug with comfortable static margin can still need resizing once the contractual factors are on. The methodology page records the full application map, clause by clause, with the 2018/2020-01 edition stated.

For a European job the alternative is the EN 13001 route, and it asks a different question. DNV bands its dynamic factor on how heavy the lift is; EN 13001-2 §4.2.2.2 derives ${\varphi }_2$ from how the crane picks the load up - the stiffness class of the whole load-supporting system, the hoist drive's control behaviour, and the speed it actually runs at. On the worked example at the shipped starting classes (HC2 stiffness, HD1 drive, load combination A1) that gives 1.117, and the main hook moves from 0.80 to 0.89. The body and the delegated lift points carry more than the hooks do - 1.497 rather than 1.117 - because they additionally take the partial safety factor and the risk coefficient, which belong to a structural proof of competence and not to a comparison against a crane load chart.

Two things to know before switching it on. The route asks which EN 13001-1 §4.2.7 proof your declared allowables belong to, and that is not a preference: under the limit state method the demand carries ${\gamma }_p$ and your allowable must be a limit design stress; under the allowable stress method the same margin is already inside the allowable through ${\gamma }_f$, so it is not applied twice. And the two routes are mutually exclusive - each factors the same demands, so the tool refuses the combination rather than quietly picking one. The methodology page carries the full application map, both held-edition gaps, and why EN 13155 is cited for scope but not used as the basis.

The setup: a 61.2 t (600 kN) vessel, 30 m long, two-crane upend. Top pick through a trunnion pair 25 m from the tail end; tailing lug 80 mm × 350 mm on the shell at r = +1200 mm, checked to ASME BTH-1-2020 Design Category B with an all-around 28 mm fillet weld; CoG box ±300 × ±50 mm; sweep step 1°.

One honest caveat about this setup before the numbers: the top lift point is declared as a position and an orientation only - no geometry, material or route - so the tool reports not evaluated for it and checks nothing there. That is deliberate: the pack never invents an attachment capacity. Declare the top attachment's spec and its capacity checks delegate exactly as the tailing lug's do (to the lifting-lug pack for a plate lug, to the lifting-trunnion pack - WRC 537 local shell assessment included - for a radial trunnion). Every verdict below is therefore about the hooks, the shell and the tailing lug.

1. The pick (φ = 0°). Load share 355.1 kN main / 244.9 kN tail - the lever-arm answer, valid at this one angle.
2. The sweep. The main hook climbs to 600 kN - the full weight - at φ = 90°, utilisation 0.80 against the declared capacity × the declared de-rate. The tail crane governs at 252.2 kN (φ = 0°, a CoG corner), utilisation 0.63.
3. The tailing lug. Governs at φ = 70.4° with the load 224.7 kN at 70.4° out of plane - utilisation 0.76 over 22 delegated rows, the governing angle pinned to ±0.0010° by re-evaluating the statics. The worst case is mid-lift; no single-angle check would have found it.
4. The transfer. The tail goes light between 89.70° and 89.72° depending on where in its box the CoG really is - the hand-off window to plan around.
5. The verdict. Overall pass at utilisation 0.80, governed by “Main hook load vs declared capacity (governing over the sweep)”. Switch the DNV overlay on and the factored tailing-lug demand goes to 1.09 - the honest next conversation with whoever owns the contract.

1. Weight from the weight take-off, as a distribution - not a guessed point mass.
2. The compression side checked for stability, not just strength. A long thin shell picked near horizontal buckles before it yields, and ASME VIII-1 UG-23(b) is the limit: factor B off the Section II-D chart at $A=0.125/(R_o/t)$, capped by the allowable tensile stress. Declare the chart, the erection temperature and the modulus so the tool can read it - an absent buckling check is not a passing one.
3. CoG declared with its uncertainty box, and the box swept, not averaged.
4. Radial offsets signed by one drawn convention - lift side positive - on every attachment and the CoG.
5. Load share evaluated through the whole sweep, each check at its own governing angle.
6. The tailing attachment checked with direction, at its governing angle - not at 0° and 90° only.
7. The transfer window located and the hand-off planned inside it.
8. Tail friction or rolling resistance declared as a band; every result read as an envelope.
9. Hook capacities from the chart at the working radius; the multi-crane de-rate declared, not assumed.
10. Sling WLLs declared and checked; bridle angle effects added before anyone relies on them.
11. Every factor between weight and verdict visible in the ledger - nothing applied silently.

How do you calculate a two-crane lift load share?

Moments about one lift point in world coordinates: the main hook load is the weight times the ratio of horizontal distances from the tail lift point to the CoG and to the top lift point, and the tail load is the remainder. The catch is that with real (off-axis) lift points those distances change with inclination, so the share is a curve, not a number - evaluate it through the whole manoeuvre and take each check at its own worst angle.

How do you calculate the tail crane load?

At the pick it is the lever-arm result - on this page's worked example, 245 kN of a 600 kN vessel. Through the lift it drifts with the inclination and the CoG position, and it reaches zero at the transfer point. The governing value on the worked example is 252 kN, found at the pick on a corner of the declared CoG box - above the nominal-CoG answer before any factor is applied, which is why the box is swept rather than averaged.

What is a tailing lug calculation?

A lug check with direction. During an upend the load on the tailing lug rotates through roughly ninety degrees relative to the lug's plane while its magnitude barely drops, so the governing case is out-of-plane bending on the base section and the weld group, somewhere mid-lift. Check the solved load vector - magnitude, in-plane angle, out-of-plane angle - at the governing inclination against a traceable lug methodology, not the flat-position force against the strong axis.

What angle governs an upending lift?

A different one for every check. On the worked example the main hook governs at transfer near vertical, the tail crane at the pick on a CoG corner, and the tailing lug at about 70° - mid-lift, where the direction is far out of plane but the load has barely shed. The honest method is to search the sweep per check and state the resolution of the search.

Does the same calculation work for a monopile or a column?

Yes. The engine takes a weight distribution along an axis, so a monopile, pile, column, transition piece or any long fabrication is the same statics, the same sweep and the same attachment delegation - the vessel vocabulary is just the most common case.

Do I need to de-rate the cranes for a multi-crane lift?

Many owners and marine warranty surveyors require the cranes be taken at 75–80% of chart capacity for a multi-crane lift. It is a project rule, not physics, so this tool ships it as a declared factor defaulting to 1.0 with the practice noted beside it - visible in the ledger, never applied silently.

What is the transfer point of an upending lift?

The inclination at which the tail reaction reaches zero and the whole weight rides the main crane. With a declared CoG box it is a window rather than a point - and it may not happen at all: on this page's worked example some CoG realisations still carry tail load at vertical, where the tail must be actively unloaded. Plan the hand-off around both outcomes, and expect the body to swing to its solved free-hanging attitude - not necessarily plumb - the moment the tail releases.

The methodology page documents every check, the frozen sign convention, the feasible-domain doctrine behind the sweep, the delegated clauses, the UG-23(b) buckling acceptance and both demand-side routes' application maps - with every edition stated. The calculator opens on the worked example above, with the inclination scrubber, the sweep chart and the direction dials live.

An upend is one calculation in a lift plan, and the rest of the cluster carries the same traceability posture: the lifting lug guide and lifting trunnion guide for the attachments this tool loads, the spreader beam guide for beam-assisted picks, and the crane ground bearing guide for the machines under it all. Trunnion-type attachments delegate to the lifting trunnion calculator, whose row set includes the WRC 537 local shell assessment; a pressure vessel nozzle tool - sharing that same shell-stress kernel - is planned but not yet published.