Fab Mistakes

Costly Mistakes in Tank and Pressure Vessel Fabrication and How to Catch Them

The recurring errors that wreck vessel and tank fabrication estimates, from mixing corrosion allowance into thickness to underbooking nozzle hours, each with a symptom and a fix.

The most expensive early mistake is buying plate off the nominal thickness instead of the required thickness plus allowances. A shell computed at 0.412 inch under ASME VIII Div 1 is not a 7/16 inch plate order. You add corrosion allowance, usually 0.0625 or 0.125 inch, then round up to the next stocked gauge. Skip that and you either fail the design review or scrap a heat. Symptom: the calculated wall passes but the stamped MDR does not. Root cause: treating minimum required thickness as purchase thickness. Fix: run the Shell Thickness Allowance figure, then round up, never down.

Unit slips on internal pressure are silent and brutal. Someone enters 150 as psi when the datasheet meant 150 psig above a full vacuum jacket, or mixes bar and psi (1 bar equals 14.5 psi), and the wall comes out 20 to 40 percent thin. Symptom: your thickness looks suspiciously light against a similar past job. Root cause: pressure and stress values pulled in mismatched units, or design pressure confused with MAWP. Fix: force everything to psi and psi-based allowable stress before the calc, and sanity check against a known vessel of the same diameter and class.

Volume errors come from ignoring heads. A 10 foot diameter by 20 foot tangent-to-tangent vessel is not just the cylinder. Two 2:1 elliptical heads add roughly 4 percent to 8 percent of shell volume each, and hemispherical heads add far more. Symptom: hydrotest water volume or fill weight is off by several hundred gallons. Root cause: modeling the vessel as a plain cylinder. Fix: use Vessel Volume with the correct head type, and cross the water figure with Hydrotest Capacity so your test pump and blocking loads match the real filled weight, not the cylinder-only number.

Weld footage is chronically underestimated because people count only the long seams. A vessel has longitudinal seams, one circ seam per shell course, head-to-shell joints, and every nozzle attachment weld. Missing the circ and nozzle welds can hide 30 to 50 percent of arc length. Symptom: consumable and labor budgets run out at 70 percent completion. Root cause: counting main seams only. Fix: build total inches from Weld Length Estimate including all courses and attachments, then feed that into Weld Consumable Cost so wire and flux are booked against real deposited metal, not a guess.

Nozzle labor is where quotes quietly bleed. Estimators price a vessel by weight and forget that a 6 inch RFWN nozzle with reinforcing pad, layout, fit-up, welding, and PWHT can carry 6 to 14 shop hours each. A vessel with 18 nozzles hides 100-plus hours. Symptom: actual hours beat the estimate by 15 to 25 percent on nozzle-heavy jobs. Root cause: lumping nozzles into a flat per-pound rate. Fix: count openings explicitly through Nozzle Labor and Fit-Up Labor, and separate set-on from set-through details because they are not the same hours.

Coating quantity errors trace back to using flat plate area instead of developed surface area. Heads, nozzles, saddles, and lugs add area a simple pi times D times L never captures, and internal lining doubles the count. Symptom: you run short one pail of two-part epoxy at 6 to 8 mils DFT. Root cause: shell-only area with no heads or appurtenances. Fix: pull developed area from Paint/Coating Area, then apply your real material loss factor. Airless spray on a vessel often wastes 30 to 45 percent, so a 90 percent transfer assumption underbuys paint every time.

Data-entry drift on material grade quietly changes everything downstream. Assuming SA-516-70 allowable stress when the job is SA-240-304L, or forgetting the joint efficiency drops from 1.0 to 0.85 without full radiography, shifts wall thickness, weight, and plate cost together. Symptom: the whole estimate is internally consistent but wrong versus the spec. Root cause: default material and E carried over from the last job. Fix: reset grade, allowable stress, and joint efficiency at the start of every estimate, and confirm plate weight through Plate Yield before it flows into Fabricated Vessel Cost.

The final trap is treating the estimate as static after design changes. A bumped design pressure, an added inspection port, or a switch from spot to full RT ripples through thickness, weld length, consumables, and test volume at once, and people update only one field. Symptom: the shop build does not match the priced build. Root cause: partial re-estimation after a revision. Fix: on any change to pressure, diameter, nozzle count, or radiography level, rerun the full chain from Shell Thickness Allowance through Fabricated Vessel Cost so every dependent number moves together.

Published 2026-07-01.