AUTONEST VCCASE · CROSS-CHECK

A 3-ton jacketed tank: drawing checked against the report — a real design run from a few lines of input

2026-08-30 · 8-min read

The value of automation comes down to one thing — are the results correct? So we ran one 3-ton jacketed tank all the way through and measured the report's numbers directly on the drawing. Below is that record. The figures are exactly as produced.

Report·Shell length 1700·Jacket length 1524·Thk 12/8/16/5·Ext. L 1622.57Measured on drawing✓1700✓1524✓12/8/16/5✓1622.57
Report numbers measured directly on the drawing — all match

Input — that's all of it

ItemValue
Vessel type · target capacityJacketed (JACKET) · 3 tons
Material (inner / jacket)A240-316L / A240-304
Design pressure (inner / jacket)0.3 / 0.35 MPa · vacuum −0.1
Design temperature (inner / jacket)130 / 140 ℃
Manhole · nozzleManhole 1 · Nozzle N1 (3S, top head)
OptionsSWEEP AUTO · SNAP ON · pad avoidance ON

No dimensions, no thicknesses entered. The thickness fields are just starting values; the rest is decided automatically. The run was VC → Jacket → 3 ton → A (Auto), 2 min 15 s.

What the automation decided

No dimensions were entered — so how are the diameter and length decided?

Neither diameter nor length was entered. Give the target capacity and the rest is worked back automatically. In this run it came out as inner diameter 1505 · length 1700, with working 3.00 m³ of a total 3.81 m³ (79%) to meet the 3-ton target. All a person did was fill in the input table above.

Why are the top and bottom heads different thicknesses?

The result shows top head 8T, bottom head 16T — different top and bottom. The automatic optimisation picks a thickness for each part to suit its own conditions instead of lumping top and bottom into one value. The top head was raised from 6T to 8T to remove the nozzle reinforcing pad, and SWEEP brought the bottom head down from 18T to 16T.

Report summary

Shell 12 · top head 8 · bottom head 16 · jacket 3 · jacket head 5 · closure 15 — all PASS. Nozzles, manhole flange, sight glass, safety device, and nozzle closure — the judgments all gather in one table. The same file contains the material list, the strength-report cover, the weld map (seam lengths and counts), the plate layout, and the hydrostatic test pressure (shell 0.51 / jacket 0.59 MPa).

What we measured on the drawing

ItemReportMeasured on drawing
Shell length17001700
Jacket length15241524
Thickness (shell / top / bottom / jacket head)12 / 8 / 16 / 512 / 8 / 16 / 5
External-pressure design length L1622.571525.35 + 291.67 ÷ 3 = 1622.57

The last row is the key. The design length for external-pressure calculation is the cylinder section plus one third of the head depth; measuring 1525.35 on the drawing and a head depth of 291.67 gives exactly the report's 1622.57. That means the external-pressure judgment comes from the drawn geometry. With that L, L/Do 1.06, Do/t 130 → A 0.00083, and on the 316L 140 ℃ curve B 49.8 → allowable external pressure 0.51 > design 0.45: judged PASS.

The red and blue lines on the chart are not fixed pictures. They are redrawn every run at the A/B values the engine computed, and the judgment text is re-decided from those values. The weld map's seam lengths and counts are the same — change the input and they change with it.

How do you filter out a "false pass"?

The result does not come out as values alone — it comes with the external-pressure judgment attached. To catch a "false pass", where the thickness field is filled but the part actually falls short, you need to be able to follow the numbers used in the judgment all the way to the drawing. The design length L above is exactly that case: the value used in the judgment can be measured on the drawn geometry, so a reviewer can follow the basis step by step.

After the report, what comes out for ordering?

It does not stop at design. The same run also produces the material list, the plate layout, the weld map with seam lengths and counts and the radiography (RT) count, the working liquid level, and the cutting list for ordering. Change the input and these numbers change with it.

Doing this one tank by hand — strength calculation, sizing, head, jacket and closure design, external-pressure check — takes hours. This run took 2 min 15 s.

※ The figures in this post are actual run results (as of 2026-08). Different input conditions give different results.

Frequently asked questions

Q. What do you enter to design a 3-ton jacketed tank?
Vessel type and target capacity, inner vessel and jacket materials, design pressure and temperature, and the manhole and nozzles. In this run no dimensions or thicknesses were entered; the thickness fields are only starting values.

Q. If you only enter the capacity, how are the diameter and length decided?
They are worked back automatically from the target capacity. In this run: inner diameter 1505 · length 1700, working 3.00 m³ of a total 3.81 m³ (79%).

Q. How did you check that the report numbers match the drawing?
We measured the shell length, jacket length and thicknesses directly on the drawing and compared them with the report. The external-pressure design length L also matches: 1525.35 measured on the drawing plus one third of the 291.67 head depth equals the report's 1622.57.

Q. Why are the top and bottom heads different thicknesses?
The automatic optimisation picks a thickness for each part to suit its conditions. In this run the top head was raised from 6T to 8T to remove the nozzle reinforcing pad, and the bottom head was brought down from 18T to 16T, ending at 8T and 16T.

Q. How long does one run take?
This 3-ton jacketed tank took 2 min 15 s. The same run produces the report, drawing, material list, weld map and plate layout together.

Q. Are the results the same if the input conditions differ?
No. The figures here are actual run results as of 2026-08; different input conditions give different results.


From simple input to the drawing, the report, the material list and the weld map — and they all agree with each other. That is what a real demonstration looks like.

We filmed the run and the measuring for this case.
From input to report-vs-drawing cross-check, 5 min 23 s. Watch: the 3-tonne jacketed tank run →