Which steel, how low the silicon, how thick the wall. Those are the right questions, and they are incomplete — because a kettle is not made of plate. It is made of plate and weld metal, and weld metal is a different steel with a different chemistry, deposited by a different process, from a different consumable.
When a kettle goes through at a seam while the surrounding wall is still comfortably thick, this is usually why.
Why silicon attacks a kettle
Molten zinc dissolves iron. At the interface, iron and zinc form an Fe-Zn alloy layer, and the fate of the kettle depends on what that layer is like.
In a low-silicon steel, the alloy layer is dense and adherent. It sits on the wall and slows further attack. Raise the silicon and the layer becomes loose and non-adherent — it detaches, exposes fresh iron underneath, and dissolution continues at close to its initial rate. Phosphorus and aluminium do the same thing by the same mechanism.
This is why BG/T 1501-2015 specifies plate at Si ≤ 0.03 % in its baseline requirement (Table 1), tightens it to Si ≤ 0.01 % in its recommended specification (Table 2), and adds a combined limit: Si + P + Al ≤ 0.045 %.
It is also why stainless and heat-resisting steels — which beat plain carbon steel almost everywhere else — resist molten zinc worse. The standard says so directly in §3: alloying and impurity elements mostly increase the corrosion rate of steel in a zinc bath.
Now apply the same logic to the weld
If silicon in the plate accelerates attack, silicon in the weld metal does exactly the same thing. And weld metal silicon is not automatically low. It depends entirely on the welding process and consumable.
| Process | Consumable | Typical deposited Si | Note |
|---|---|---|---|
| Manual metal arc | E4303 (J422) | 0.13 % | Slow, but silicon is manageable |
| Manual metal arc | CJ427A | 0.04 % | Low silicon |
| Semi-automatic gas shielded | ER50-6 + CO₂ | 0.86 % | Very high silicon. The standard restricts this to flange work |
| Submerged arc | H08A + HJ431 flux | High | The standard flags this explicitly |
| Submerged arc | H08A + low-silicon kettle flux | ≤ 0.03 % | The one you want |
Look at the ER50-6 line. 0.86 % silicon — roughly thirty times the 0.03 % limit specified for the plate.
A kettle built from carefully specified low-silicon plate and then welded with a high-silicon consumable has a set of seams that will dissolve far faster than the wall around them. The plate certificate will be perfect. The kettle will still go through at the seam.
What the standard requires
BG/T 1501-2015 Table 10 grades finished kettles on six criteria. Two of them are about welds.
| Grade 1 | Grade 2 | |
|---|---|---|
| Weld quality class | I, II | III |
| Weld metal silicon | ≤ 0.03 % | ≤ 0.12 % |
Grade 1 holds weld metal to the same silicon limit as the plate. Grade 2 allows four times as much.
Note also what the standard does not require of a kettle weld: there is no impact energy requirement anywhere in Table 10. A galvanizing kettle runs at a constant 445–460 °C under hydrostatic load. It is not an impact-loaded structure, and low-temperature toughness is not what decides its life. Silicon and internal soundness are.
How the process choices follow
Two consequences fall out of this.
First, the flux matters as much as the wire. The difference between the last two rows of the table above is not the electrode — both are H08A. It is the flux. Standard HJ431 puts silicon into the deposit; a low-silicon kettle flux holds it under 0.03 %. Same wire, same machine, order-of-magnitude difference in the number that decides kettle life.
Second, gas shielded welding gets restricted to where it cannot hurt you. BG/T 1501-2015 Table 5 permits ER50-6 with CO₂ specifically for flange work — the top edge, above the bath, where the weld never sees molten zinc. Below the zinc line, it does not belong.
The standard’s default arrangement (§5.5.3) is manual metal arc on the inner wall and submerged arc on the outer wall — with the note that if a low-silicon flux is used, submerged arc can be used on the inner wall too.
The questions worth asking your supplier
- What is your weld metal silicon limit, and how do you verify it? If the answer is a number, ask to see it on a report. If the answer is a description of their quality culture, that is not an answer.
- Which flux, on the inner wall? Standard submerged arc flux below the zinc line is a red flag.
- Are you building to Grade 1 or Grade 2 of Table 10? The difference is 0.03 % versus 0.12 % on the seams.
- What is the NDT standard and acceptance level for the structural welds? Silicon governs how fast a sound weld dissolves. It says nothing about a weld that had a defect in it to begin with.
Where we stand
We build to Grade 1: weld metal Si ≤ 0.03 %, using low-silicon flux. Every structural weld is ultrasonically tested to GB/T 11345-2023, which is identical (IDT) to ISO 17640:2018, with acceptance to GB/T 29712.
We publish this because we would want to see it if we were buying, and because as far as we can tell nobody else does.
Source: BG/T 1501-2015, Steel Hot-Dip Galvanizing Kettles, China Surface Engineering Association — §3.2, §5.5.3, Tables 1, 2, 5, 6, 10.