A furnace comes back from relining on a Friday afternoon. First heat Monday morning, and someone opens the argon valve on the porous plug because a first heat without purging feels wrong, like pouring without degassing.
The steel goes into the brick joints. The joints are not sintered yet. The plug is finished, and so is the lining, and the reline that was supposed to last a hundred and fifty heats is being redone in week one.
The purge system is the cheapest piece of equipment on the melting floor and the most expensive one to lose. Almost everything that keeps it working is a rule about when not to use it.
What bottom purging actually does
Argon bubbling up through the melt from the bottom does two jobs, and neither of them is “removing gas” in the simple sense.
First, it lowers the partial pressure of hydrogen and nitrogen above the melt. Dissolved gas leaves the metal because the atmosphere it is equilibrating against no longer contains much of it. Second, the bubble stream sweeps small inclusions upward, where they can be taken off with the slag.
Both effects depend on something happening after the bubbles stop. Gas that has left the metal still has to escape the bath surface, and inclusions that have been lifted still have to be removed rather than stirred back in. A furnace that purges continuously while the bath is being stirred, tapped and poured is doing very little for cleanliness. The purging is only worth anything when it is followed by a period where the melt is not being disturbed.
That is why purge time and settling time belong in the same sentence, and it is the most common way the system gets used correctly for years and then quietly stops working: somebody moves the argon from “before settling” to “throughout”, the gas readings look fine, and the inclusion rate climbs.
The rules, in the order they get broken
1. No purging on the first heat of a new lining
The brick joints have not sintered. Argon pressure at the plug face is enough to push liquid steel into those joints, and steel that enters a joint does not come back out. The plug is scrap, and often so is the lining around it.
This is the one rule that costs a reline to learn, which is why it should be written on the furnace, not in a manual.
2. Verify flow at the first heat of every shift
Not “check the pressure gauge”. Verify that gas actually comes through. Blocked plugs do not announce themselves; the regulator reads normal because the line is pressurised and the plug is not passing anything.
The rule in our procedure is that a plug that does not pass is not melted on. That sounds harsh until you price a heat of contaminated steel against a fifteen-minute delay.
3. A blocked plug gets 1750 °C and a hard purge — twice
Blockage is steel or slag frozen in the capillaries of the brick. Getting it out means getting it liquid, and 1750 °C is well above normal superheat. Purge hard at that temperature and it usually clears.
If it does not clear on the second attempt, the furnace comes down and the brick gets changed. No third attempt, no “we’ll run it lean this week”. Running a dead plug means running without degassing and without inclusion float-out while believing you have both, which is worse than knowing you have neither.
The 1750 °C treatment is also hard on the lining. That is precisely why it is capped at two attempts rather than repeated until it works.
4. Clean the bottom on the last heat
100 g of furnace cleaner on the final heat of the campaign. This is the maintenance that prevents rule 3 from ever being needed. It costs almost nothing and it is skipped almost always, because the last heat of a campaign is the one everybody wants finished.
5. Shut down in sequence, not in whatever order is convenient
The shutdown procedure is specific and every step has a reason:
| Step | Action | Why |
|---|---|---|
| 1 | Leave a little steel in the furnace after pouring | A dry furnace cools unevenly and the residual slag sets hard |
| 2 | Clean the furnace bottom within 3 minutes | Slag is still soft; after that it is a chipping job or it stays |
| 3 | Keep purging for a further 6 minutes | Clears steel and slag out of the plug while it is still fluid |
| 4 | Tilt 30° and fit the lid | Drains residue away from the plug face |
Step 3 is the one that protects the plug, and it is the one that gets dropped because the shift is over. A plug that is left with steel sitting in its face overnight starts the next campaign partially blocked.
Stainless: take the human out of the loop
On stainless heats, purge timing is interlocked with a timer rather than left to the operator. Not because operators are careless, but because purge duration on stainless is a specification, and a specification that depends on someone watching a clock while also watching a bath is not a specification.
The cost nobody puts in the routing: three minutes of cooling
Purging has to be followed by a still bath, and here the arithmetic starts to matter. The plant rule is a zero-power hold of 3 minutes — zero power, not just “stop tapping”. If the power is still on, the bath is still being stirred by the induced current, and there is no quiescent liquid for anything to float through. Counting the minutes without cutting the power means the hold did not happen.
Those three minutes are not free. The furnace loses heat the whole time, at a rate that depends on furnace size:
| Furnace | Cooling rate, power off | Loss over a 3-minute hold |
|---|---|---|
| Small (~160 kg) | 20 °C/min | about 60 °C |
| Large (~360 kg) | 15 °C/min | about 45 °C |
Which gives the working formula: melting temperature = tapping temperature + hold time × cooling rate.
If the target tapping temperature is 1,600 °C on the small furnace, the bath has to come off power at about 1,660 °C, because the hold will take 60 °C out of it. Skip that correction and the pour happens cold, and a cold pour produces exactly the defects that get blamed on the purge system: misruns, cold shuts, rough surface, and a bath that never gave up its inclusions because it was too viscous for them to rise.
Note also that the general handbook figure for the refining hold is 1–2 minutes, while the plant procedure runs 3. Longer hold, more temperature loss. Whichever figure your shop uses, the formula has to be applied to your number, not to a textbook one.
A note on furnace capacity
Our melting procedure gives every addition as a percentage — 0.1% silicon-calcium-manganese, 0.125% aluminium wire — but the document itself does not state furnace capacity. Back-calculating from the absolute figures quoted for the two furnaces (0.36 kg and 0.16 kg of silicon-calcium-manganese) puts them at roughly 360 kg and 160 kg.
That is a derived number, not a nameplate reading, and it should be verified against the equipment record before it goes onto a process card. The reason it matters: percentages are portable between shops, absolute grams are not. A procedure that quotes only percentages is only complete if the capacity appears in the header. Otherwise every gram figure derived from it is off by whatever the capacity ratio happens to be.
What the plug cannot do for you
Bottom purging gets oversold. It will not compensate for a wet charge, oily scrap or damp alloy additions — those bring in hydrogen faster than any argon flow can take it out. It will not substitute for deoxidation. And it will not help at all if the bath is never allowed to stand still, because the whole mechanism terminates in a quiescent period.
The rest of the sequence it depends on:
- Slag former covering the bath throughout, not just at the start.
- Pre-deoxidation at or below 1,550 °C, manganese before silicon, at roughly 2:1.
- Refining at 1,580–1,600 °C.
- Final deoxidation with aluminium at tapping temperature, added below the surface rather than on it.
- Slag removal at two points: after the quiescent hold, and after final deoxidation. Two to three passes after the final deoxidation, not one.
Then the pour itself inside four minutes per heat, with a ladle sample taken past the halfway point. That four-minute limit comes from shell gas pickup and steel cooling, not from heat weight — a full furnace still gets four minutes. Scaling the pour time to the tonnage is how a full furnace ends up pouring the last third cold.
If you only check one thing
At the start of the next shift, before the first charge goes in, confirm that gas comes out of the plug. Not that the valve is open. Not that the regulator reads pressure. That gas passes.
Everything else on this page is about not destroying the thing you just verified.