A skirt carries the vessel's weight and the wind or seismic overturning moment as axial compression, and a manway cuts a piece out of that load path. The wall beside the hole takes the load that the hole removed, the corners of the opening concentrate it, and the weakened wall buckles earlier on the compression side. The sleeve (frame) and an optional pad put metal back.
How the job models it
- A cylinder from the bottom of the base ring to the skirt-to-head joint. Both are lines of support, so the model spans the whole skirt.
- The opening is a nozzle with an open, free end: tombstone, obround or round, with the sleeve wall round the clear size (outside size = clear + 2 × sleeve; the tombstone's mid-surface is its outside curve offset by half the wall). Sleeve projections outside and inside, a pad, and vertical stiffener bars either side are inputs. The bars mesh as attachments of their own and need about 200 mm clear.
- The mesher follows the decay length λ = √(R·t): first element at the junction min(t, 0.2λ), refined zone 2.5λ, far field up to λ, an O-grid from the grid rectangle to the exact sleeve and shell intersection.
- Pressure, temperature and cycles are off. Buckling is always run.
Loads at the base
The inputs are a foundation load summary: weight W (down), base shear V, base moment M, optionally a vertical seismic part and a moment of eccentric weight. They are written into the vessel rows at the base with the shear toward the opening and the overturning moment leaning the skirt toward it, so the opening side is in compression whatever its angle. Wind or seismic from the other side is checked too (occasional reversal). A test confirms that above the opening the shell stress agrees with beam theory within 5 %.
Hand check 1: skirt wall for axial compression
Hand check 2: VIII-2 4.5.17.3, opening under axial compression
The area available is the sum of A1 (shell within 0.75√(R·t) each side, 2·LR·(t − t_r)), A2 (sleeve outside, 2·LH·tn with LH = min(0.5√((d/2)·tn), 2.5·tn, projection)), A3 (sleeve inside), A41 and A43 (fillet welds, leg squared) and A5 (pad). The check passes when the total is at least Ar. The code applies the 25 % diameter limit of 4.5.17.1 and flags openings beyond it.
Example skirt of the job (calculated by hand from the formulas above)
| Skirt 3000 ID × 16, sleeve 16, opening 600 wide | R = 1500, Dm = 3016, d/D = 0.20 (inside 25 %) |
| Loads at base | W = 1500 kN, V = 120 kN, M = 2400 kN·m, Sc = 95 MPa, factor 1 |
| Opening centre height | 450 + 1000/2 = 950 mm; M(z) = 2400 − 120 · 0.95 = 2286 kN·m |
| Line force at the opening | q = 1.5×10⁶/(π·3016) + 4 · 2.286×10⁹/(π · 3016²) = 478.3 N/mm |
| Stress and required wall | q/t = 29.9 MPa; t_r = 478.3/95 = 5.03 mm (base: 30.9 MPa, 5.20 mm) |
| √(R·t), γn, limit | 154.9 mm; γn = 600/309.8 = 1.936 > (93.75/291 + 0.22)² = 0.294, so Ar = d·t_r |
| Required area Ar | 600 × 5.035 = 3021 mm² |
| A1, A2, A41 | 2 · 116.2 · 10.97 = 2548; 2 · 34.64 · 16 = 1109; 10² = 100 mm² |
| Available AT | 3757 mm², ratio 1.24: passes |
No inside projection and no pad here, so A3, A43 and A5 are zero. The dev log (2026-10-10) states that the 4.5.17.3 routine reproduces a COMPRESS skirt-access page to 4 decimals; that page is not in the repo.
Limits
- The base ring, anchor bolt chairs, gussets and the skirt-to-head weld detail are not modelled. The ring and the head joint are lines of support.
- Sc is entered; there are no II-D external-pressure charts in the code yet.
- Stiffener bars tight against the sleeve, a stiffening ring on the sleeve end and gussets need joined meshing and are planned, not built.
- 4.5.17.1 also limits the opening to 80 % of the ring spacing; only the 25 % diameter limit is checked in the code read.
[UNVERIFIED] The engineer decisions list (2026-10-10) says the skirt opening, tombstone openings and buckling job are not yet validated against a hand check or a CalculiX run recorded in files/. The dev log quotes a buckling factor of 24.7 with the opening in compression and 43.1 in tension from a browser run; only the ordering is asserted by a test.
FE results are early access and under verification.