For butt joints above 20 mm, long straight seams and the longitudinal or circumferential seams of pressure vessels, submerged arc welding (SAW) is still the process with the highest deposition efficiency and the most consistent weld quality available: wire feeds continuously, the arc burns entirely beneath a blanket of granular flux, there is no arc glare and almost no spatter, and a single pass can replace three or four manual layers. This article explains how SAW actually works, the logic behind wire and flux matching, four mistakes we see repeatedly, and a selection procedure you can apply straight away.
1. How submerged arc welding works
The defining feature is that the arc burns underneath a layer of granular flux. Flux flows from a hopper and is laid along the joint ahead of the wire. The wire feeds through a contact tip and strikes an arc against the base metal beneath that blanket. Arc heat melts the wire, the base metal and some of the flux simultaneously; the molten pool stays covered by flux and slag, which solidifies into a removable slag crust. Unfused flux is recovered and reused.
That structure produces four direct consequences:
- Very high thermal efficiency — with the arc shielded, heat loss is low and penetration is far greater than manual or gas-metal arc welding at the same current.
- No arc glare, little fume — the flux blanket traps the arc light and most of the fume. Working conditions are far better than open-arc processes and operators do not watch the arc through a helmet.
- Almost no spatter — no metal is thrown away as spatter, post-weld cleaning is minimal, and wire utilisation approaches 100 percent.
- Restricted welding positions — flux has to sit on the joint under gravity, so SAW is essentially limited to flat position and shallow horizontal work. It cannot do vertical, overhead or all-position welding.
One sentence summarises the sweet spot: long, straight, thick joints that can be rotated into the flat position are where SAW belongs.
2. The parameter triangle: current, voltage and travel speed
| Parameter | Effect of increasing it | Typical value (carbon steel, 4.0 mm wire) |
|---|---|---|
| Welding current | Deeper penetration, higher deposition, faster wire melt-off; too high causes burn-through, undercut and coarse grain | 500–800 A |
| Arc voltage | Wider bead, shallower penetration, lower reinforcement; too high hurts slag release and causes porosity | 28–36 V |
| Travel speed | Lower heat input, less penetration and width; too fast risks lack of fusion and poor bead shape | 25–55 m/h |
| Stick-out | Longer stick-out adds resistive preheat: deposition rises, penetration falls, arc wanders | 30–40 mm |
| Flux layer depth | Too thin exposes the arc and brings spatter and porosity; too thick traps gas and gives a narrow, high bead | 25–40 mm |
Rule of thumb: current sets penetration, voltage sets bead width, speed sets heat input. Tune in that order — current, then voltage, then speed — and change only one variable at a time.
3. Matching wire to flux
In SAW, weld metal composition is determined by the wire and the flux together. That is fundamentally different from MIG/MAG, where the wire alone decides, and it is the single biggest trap for newcomers.
| Base metal | Typical wire | Typical flux | Notes |
|---|---|---|---|
| Q235 / Q345 carbon and low-alloy steel | H08A, H08MnA | High-manganese high-silica (e.g. HJ431) | The most universal combination; works on AC or DC with good slag release |
| Q390 and higher-strength low-alloy steel | H08MnMoA, H10Mn2 | Low-manganese, medium-silica, medium-fluoride (e.g. HJ350) | Used where low-temperature impact toughness must be met |
| Creep-resistant / pearlitic heat-resistant steel | Matching wire such as H08CrMoA | Basic agglomerated flux | Select to base metal chemistry; requires a procedure qualification |
| Stainless steel | ER308, ER316 and equivalents | Dedicated stainless agglomerated flux | Control carbide precipitation and ferrite content |
Three practical reminders: always bake the flux (typically 300–350 °C for one to two hours — moisture is the number one cause of porosity); screen recovered flux and blend it with fresh flux in a controlled ratio; re-qualify the procedure whenever the wire batch changes.
4. Four mistakes we see repeatedly
- Mistake 1: more current is always faster. Excessive current coarsens the grain and degrades heat-affected-zone toughness, and in multi-pass thick plate it undermines the normalising effect of later passes. Size the current from the groove dimensions, not from the machine's maximum.
- Mistake 2: SAW never needs preheat. Steel with a carbon equivalent above 0.45 percent, plate thicker than 30 mm, or ambient temperatures below 5 °C all still require preheat and interpass control, or cold-cracking risk rises sharply.
- Mistake 3: one machine covers every current range. A 1000 A class and a 2000 A class differ completely in wire drive, contact tip cooling and power source characteristics. Running outside the designed range gives an unstable arc and premature failures.
- Mistake 4: DC is always better than AC. DC reverse polarity gives deep penetration and suits root passes and thick plate; AC suffers far less arc blow at high current and deposits more, which is why tandem wire setups almost always use AC. Choose for your dominant application instead of defaulting to DC.
5. Machine selection in five steps
- Fix the maximum thickness and groove design — thickness sets the current class: 20–40 mm calls for 1000 A, 40–80 mm for 1250 A, and above 80 mm or high-productivity tandem work for 2000 A.
- Define the joint geometry — longitudinal and circumferential seams suit a welding column and rotator; long straight plate seams suit a tractor or gantry.
- Choose the power source characteristic — thick wire at high current wants a drooping (constant-current) source with variable-speed feed; fine wire at high speed wants a flat (constant-voltage) source with fixed-speed feed.
- Decide the automation level — tractor travel, oscillation, seam tracking, tandem wire, or integration with a CNC or robot cell.
- Specify the peripherals — flux delivery and recovery, fume extraction, water cooling, rotators and positioners, and groove preparation equipment.
The Icyrobot ZD7 submerged arc range spans 630 A to 2000 A, in four classes: SAW welder, 1000 A, 1250 A and 2000 A. For vertical seams over 30 mm that cannot be rotated, consider consumable-nozzle electroslag welding, which completes the joint in a single pass.
6. Frequently asked questions
How thick can SAW weld?
Single-wire SAW routinely covers 8–100 mm and can reach 150 mm or more with multi-pass technique and a suitable groove. The real limit is usually groove preparation volume and interpass cleaning time, not the power source. Beyond 100 mm in a vertical seam, electroslag welding is usually more economical.
Why is my SAW weld porous?
In order of probability: damp flux (by far the most common), rust or oil in the groove, too thin a flux layer or recovered flux overloaded with fines, and excessive arc voltage. Bake the flux, clean the groove, then re-check voltage and layer depth.
Can SAW weld stainless and non-ferrous metals?
Stainless, yes — with a dedicated wire and agglomerated flux. Aluminium and copper are not suitable for SAW; use TIG or MIG instead. See our MIG/MAG vs TIG comparison for the details.
Is tandem-wire SAW worth it?
Yes when seams are long, plate is 20–60 mm and annual volume is stable. Tandem or parallel twin wire raises deposition by 60–100 percent, but capital cost, manipulator stiffness and flux consumption all rise too — for high-mix, low-volume work it reduces flexibility.
Conclusion
The value of SAW is that it makes thick-plate, long-seam work both fast and repeatable. Selection is not about chasing the biggest current; it is about matching current class, groove design and automation to one another. For a configuration built around your specific thicknesses and joint geometry, contact the Icyrobot engineering team for a proposal and procedure qualification support. Welding procedure and qualification references are published by the American Welding Society.
