A DN50 sanitary line on 316L takes about ninety seconds of arc time per joint on an automatic orbital welding machine. If that joint comes out with a grey, sugared root, the ninety seconds is the cheapest part of the job. Cutting out the weld, re-machining the bevel, re-fitting, re-purging and re-inspecting typically costs twenty to forty times the original weld, and it holds up release of the whole batch while the paperwork catches up.
That asymmetry is the reason to be precise about setup. Automation does not make a bad procedure good — it makes a procedure repeatable. If the window is wrong, you now produce wrong welds faster and with better documentation.
This guide covers the four things that actually determine whether a joint passes: the process window, the purge, the fit-up, and the verification chain. It assumes closed-head orbital TIG on stainless steel, which is where most of the failure modes in sanitary and high-purity work show up.
The four variables move together
Operators new to orbital welding tend to treat current, travel speed, wire feed and gas flow as independent dials. They are not. Every adjustment you make to one shifts the others.
Raise current to cure lack of fusion and you add heat input, which pushes the root toward suck-back and drives oxygen pick-up on the inside surface. Raise travel speed to cut heat input and you may lose sidewall fusion at the six o’clock position, where gravity works against the puddle. Add filler wire to bridge a loose fit and you thicken the bead, which changes how much shielding you need.
The practical consequence: change one variable at a time, weld a test coupon, cut it, look at the macro section. Three coupons and two hours will tell you more than a week of guessing on production pipe.
A starting window you can tune from
The table below is a set of starting points for 316L tube, autogenous where the wall allows it. Treat these as a first coupon, not a recipe. Wall thickness, joint design, bevel geometry and the machine’s own arc characteristics all shift the numbers, and every qualified procedure has to be proven by destructive testing anyway.
| Wall thickness | Current (A) | Travel speed (rpm) | Filler wire | Shield gas (L/min) |
|---|---|---|---|---|
| 1.0 mm | 35–55 | 4.0–6.0 | None | 10–12 |
| 1.5 mm | 55–75 | 3.0–4.5 | 0.8 mm, optional | 10–14 |
| 2.0 mm | 75–100 | 2.0–3.5 | 0.8 mm @ 300–600 mm/min | 12–15 |
| 3.0 mm | 100–130 | 1.5–2.5 | 0.8–1.0 mm @ 500–900 mm/min | 12–15 |
Two notes on reading this table. First, travel speed is quoted in rpm of the torch around the joint, which is how most orbital power supplies program it — convert to mm/min if your procedure is written that way. Second, the jump from 2.0 mm to 3.0 mm is not linear. Above roughly 2.5 mm, single-pass autogenous welding stops being reliable and you should be planning a wire-fed, and often multi-pass, procedure.
Arc voltage on these machines typically lands between 8 and 12 V. If you find yourself pushing voltage to hold arc length, the torch is probably not centred on the joint — fix the setup, not the voltage.
Head choice shapes the same window. A closed head gives you the most rigid arc path and the best gas seal, but it needs a clean run of pipe on both sides to seat. On short spools, fittings and tight skids, an open-head orbital welding system is the practical alternative, at some cost in arc stability.
Purging is where most failures are born
The inside of the pipe is the part nobody sees and the part that fails audits.
For austenitic stainless, target below 50 ppm oxygen in the purge gas at the joint before the arc starts. Duplex grades sit around 100 ppm. Titanium wants tighter still, under 20 ppm. Those numbers correspond roughly to a heat-tint chart where silver or straw is acceptable, blue or purple is marginal, and grey-black oxide means cut it out.
Reaching those numbers is a volume problem, not a flow problem. A sealed length of pipe has to be displaced several times over before residual oxygen drops to single-digit parts per million. The arithmetic is straightforward:
- Ø50 mm tube, 3 m long: cross-section 19.6 cm² × 300 cm ≈ 5.9 litres of internal volume
- At six volume changes: roughly 35 litres of argon
- At 10 L/min through the purge line: about 3.5 minutes of flowing before you weld, plus time to stabilise
What kills purge quality in practice is not the flow meter setting. It is leaks at the tape and plug, argon escaping through a gap you did not seal, and starting the arc while the reading is still falling. Measure at the joint, not at the regulator. A cheap oxygen analyser downstream of the joint, reading continuously, is the single highest-value accessory on the machine.
One more thing: purge gas has to keep flowing during the weld and for long enough after the arc stops for the weld to drop below the oxidation temperature. Cut the argon the moment the arc extinguishes and you will see the root discolour inside a joint that looked perfect from outside.
Fit-up decides more than the machine does
The best power supply in the world cannot close a 1 mm gap. In sanitary and high-purity work, fit-up tolerance usually governs acceptance more tightly than the welding parameters do.
What to control before the head goes on:
- Internal misalignment. Most specifications limit it to 10–20% of wall thickness, with an absolute cap around 0.5 mm. Check with a dedicated alignment gauge, not by eye.
- Joint gap. Tight fit-up with essentially zero gap is the goal; anything approaching 0.3 mm on thin wall will produce either burn-through or lack of fusion depending on which way the arc wanders.
- Squareness of the cut. Mechanically cut and deburred, with the face perpendicular enough that the two ends meet across the full wall. A slight bevel on the cut face shows up as uneven penetration on the macro.
- Cleanliness. Oil, cutting fluid and fingerprints are hydrocarbon and sulphur sources that turn into porosity. Wipe the joint faces with acetone or isopropanol immediately before closing the head, and handle with clean gloves.
The reason fit-up matters so much in orbital work is that the arc travels a fixed path at a fixed speed. Unlike a manual welder, the machine cannot back off when it hits a wide gap. It commits.
The verification chain: five layers
Acceptance of an orbital weld is not a single test. It is a chain, and each link catches a different failure mode.
- Visual, outside. Bead profile, undercut, colour of the outer surface, spatter. Fast, and it catches gross parameter errors.
- Borescope, inside. Root colour, penetration, suck-back, lack of fusion at the root. This is the inspection that matters most for hygienic service and the one most often skipped when production is behind.
- Non-destructive testing. Dye penetrant on tube-to-tubesheet fillet welds; radiography on butt joints where the specification calls for it. Ultrasonic is rarely useful on 1–3 mm wall.
- Procedure qualification and destructive sampling. Per ASME IX or ISO 15614-1, depending on the contract. Macro section, bend test, sometimes tensile. This is what qualifies the procedure in the first place; production sampling then confirms it is holding.
- System-level testing. Hydrostatic or pneumatic pressure test, helium leak test on high-purity gas lines, and for pharmaceutical work the passivation and rinse-water checks that follow.
A point that trips up project teams: with automatic orbital welding, the “welder qualification” is really a procedure qualification plus operator training on the specific machine. The machine executes; the operator sets up, and setup errors are the dominant variable. Document who set up each joint.
Defect to root cause, without guesswork
| What you see | Most likely cause | First thing to check |
|---|---|---|
| Grey or black root | Purge oxygen too high, or gas cut too early | Oxygen reading at the joint, post-flow timer |
| Lack of fusion at root | Current low, travel too fast, gap too wide | Macro on a coupon at the same settings |
| Burn-through or suck-back | Current high, travel too slow, excessive gap | Joint gap, then current |
| Porosity | Oil or moisture on the faces, gas leak, low flow | Wipe-down discipline, hose and fitting leaks |
| Undercut | Arc voltage high, torch off-centre | Torch centring on the joint |
| Crater crack | Inadequate downslope or post-flow | Program end sequence |
| Weld wandering off centre | Head not aligned, cut not square | Alignment and squareness |
Work down the table from the outside in: check the setup before you touch the program. Nine times out of ten the parameters were fine and the pipe was not.
Record what you will be asked about later
Traceability is not bureaucracy in this business. When a client auditor or a regulator asks how you know a specific joint is sound, “the machine said so” is not an answer.
For every joint, log the machine and power supply ID, program number, head model, diameter, wall thickness, heat number, current, voltage, travel speed and wire feed, shield and purge flow, the oxygen reading at arc start, the operator, the timestamp, and the inspection result including borescope images. Most modern orbital power supplies export this as CSV per weld; link the export to a joint number and the audit becomes a lookup instead of an investigation.
For pharmaceutical and semiconductor clients, this log is often a contractual deliverable, not an internal nicety.
When the machine is the wrong answer
Being honest about the limits is part of specifying well.
Single-piece or one-off work in an odd diameter rarely justifies the setup time. Field locations with no clearance for the head — a closed head needs a clean run of pipe on both sides to seat and rotate — are better served by an open head or by manual TIG. Very thin wall below about 0.5 mm is a burn-through exercise that micro-TIG or laser handles better. And on large-diameter heavy wall, above roughly DN300 and 6 mm, the single-pass advantage disappears and you are back to multi-pass work where the machine’s benefit is consistency rather than speed.
Where the machine earns its cost is repeatable production: the same joint, hundreds of times, in a material where the inside surface has to be provably clean. That is a narrow description, and it covers most of the hygienic and high-purity piping built today.
Questions that come up at specification stage
Can an automatic orbital welding machine weld without filler wire?
Yes, on thin wall. Autogenous welding is normal up to about 1.5 mm on stainless. Thicker wall, or a joint with real gap, needs wire feed to avoid underfill.
What oxygen level do I need for a stainless purge?
Below 50 ppm is the working target for austenitic stainless, verified at the joint with an analyser rather than assumed from the flow rate.
Do I need to requalify the procedure for every diameter?
Not always. ASME IX groups diameters and thicknesses into ranges, so a qualified procedure covers a band. Check the essential variables in the code edition your contract cites; the ranges are wider than most people assume.
How fast is it really?
Arc time on a DN50 joint is typically 60–120 seconds. The joint cycle is longer — fit-up, purge, weld, cool-down, inspection — and that cycle, not the arc time, sets your production rate.
Can it weld tube to tubesheet?
Yes, with a dedicated tube-to-tubesheet welding head and an orbital program built for the geometry; the fillet weld profile and heat sink are different from a butt joint.
Next step
If you are specifying a system, the useful conversation is about your joint list: diameters, wall thickness, material, and the acceptance standard you have to meet. That determines head size, whether you need wire feed, and what the purge and logging setup has to look like.
KEPUNI builds closed-head and open-head orbital welding systems for sanitary, high-purity and process piping, along with tube-to-tubesheet heads for heat exchanger work. For the wider picture on machine types and components, see our complete guide to automatic orbital welding machines. If budget is the current question, the price guide covers what drives cost in these systems.
Send us your joint list and we will tell you which head and program structure fits it.