Contents

1. Why the Root Pass Is Welded from the Inside

Every pipeline welding sequence obeys the same order: root, hot pass, fill, cap. The root pass accounts for only 15 to 20 percent of the total arc time on a joint, yet it decides the fate of the entire weld. The reason is structural rather than procedural: the root is the only pass that is made once and cannot be repaired in place. A missing fill pass can be added. A poorly shaped cap can be ground back and re-run. A root defect means cutting the joint out, re-bevelling, re-line-up and starting over. A 40-minute joint becomes a four-hour operation, and the cut-out introduces its own risks — out-of-roundness, residual stress redistribution, and a second chance to get the alignment wrong.

The difficulty of root welding comes down to one physical fact: the molten pool is unsupported. On a 12 to 25 mm wall, the bevel root offers perhaps 1.5 to 2.5 mm of land and 2 to 3 mm of gap. Molten steel in that space wants to fall. Because the pipe is fixed horizontally on skids, a welder walking around the circumference sees gravity rotate relative to the pool: flat position pushes the pool down, overhead position pushes it up. Manual root welders hold the pool with feel — arc force, travel speed, keyhole size — balancing surface tension against arc pressure. That skill takes three to five years to become reliable, and it still drifts with fatigue, night shifts, rain and heat.

Welding from the inside changes the mechanics of the problem rather than automating the manual solution. An internal welder clamps and centres itself inside the pipe, presents the torch at the bevel root, and welds outward. Below the pool there is now a solid copper shoe instead of air. Gravity stops being an adversary and becomes an ally: the pool settles against the backup and freezes into a controlled profile. On a 48-inch line, an internal welder completes a root in 60 to 110 seconds. A two-man manual root crew needs 8 to 15 minutes for the same joint. More importantly, the machine does not shake after eight hours.

This is the real justification for the pipeline automatic internal welder — also sold as an internal root welding machine, internal line-up welder or internal welding system. It does not make manual welding slightly faster. It removes the mechanical boundary condition that made manual root welding hard in the first place. Everything downstream — clamping design, copper shoe geometry, procedure windows, defect signatures — follows from that single change.

2. What an Internal Welder Actually Is

Newcomers often assume an internal welder is a welding power source with a travel carriage bolted on. Its ancestry is different. The machine descends from the internal line-up clamp, introduced in the 1960s and 1970s to round and align pipe ends from the inside using hydraulic expansion shoes. Engineers quickly realised that if a device was already inside the pipe and already centred to a fraction of a millimetre, it could carry a torch as well. The first internal welding heads appeared in the late 1970s, limited to small diameters and a single torch. Inverter power supplies and closed-loop seam tracking in the 1990s produced the architecture we recognise today.

The correct mental model is therefore a precision positioning platform that happens to weld. Its three core capabilities, in order of importance, are centring accuracy, travel synchronisation accuracy and arc stability. Buyers tend to focus on the last one — how many torches, how many amps, how fast — while the first two determine roughly 80 percent of field root defects. If the machine is not centred, the gap is not uniform. If the gap is not uniform, no power source setting can save the joint.

Three mechanical families cover most of the market. The monobloc design integrates gas bottles, power supply, controls and torches into a single body supported by its own clamping shoes; it suits diameters from about 20 inches upward. The split design keeps the power and control unit outside the pipe, connected by an umbilical, which shortens and lightens the in-pipe head for small diameters and bend-dense sections. The articulated or flexible design links three to five short modules with universal joints, allowing passage through bends down to 1.5D; it exists specifically for mountainous and station piping where a rigid body simply will not fit.

The first selection question is therefore not “what are the specifications” but “how curved is my pipe, how tight is my bevel tolerance and how hard is my schedule”. Answer those three and the family chooses itself.

3. Five Subsystems and How They Couple

A typical internal welder breaks into five subsystems. Understanding their coupling matters far more than memorising any single rating.

3.1 Travel and Support

Front and rear clamping shoe groups, drive wheels and motors. The front group centres and brakes; the rear group transmits thrust and stabilises the body; drive wheels roll on the inner wall and generate traction through friction. Shoe facings are usually polyurethane or copper-based composite — enough friction to avoid slip, soft enough not to score internal coating. Travel speed ranges from 0 to 3 m/min, but welding uses only the bottom slice, roughly 0.15 to 0.45 m/min.

3.2 Centring and Clamping

This is where precision originates. Six to twelve shoe groups are distributed around the circumference, independently or group-controlled. Clamping force is not a “more is better” parameter. Too little and the body creeps under welding reaction force, producing torch drift. Too much and the pipe end is expanded into an ellipse, which destroys the fit-up gap. A practical target is the pressure that restores roundness to within 1 percent of nominal without changing the circumference by more than 2 mm. High-end machines close the loop with displacement sensors that report actual ovalisation in real time.

3.3 Welding Execution

Rotating ring, torches, wire feed and gas path. Ring rotational stability directly determines line-energy uniformity. Torch count runs from one to four; twin opposing torches are the most common configuration because simultaneous symmetric welding from the twelve o’clock position produces symmetric heat input and minimal ovalisation. Wire feed is push-pull, with spool capacity of 5 to 15 kg — enough for 8 to 20 joints.

3.4 Power and Control

Inverter DC sources dominate, 200 to 400 A, with constant-current, constant-voltage and pulsed modes. The control system is the real barrier to entry: it must process ring position, travel position, arc voltage, current and gas flow simultaneously and close the loop within roughly 10 ms. Premium systems add laser profile scanning of the bevel before the arc starts, generating a per-segment welding path automatically. That single feature removes most of the influence of fit-up quality from the outcome.

3.5 Gas and Shielding

Internal root welding typically uses a mixed shield: 80% argon with 20% CO2, or 85/15 with a small helium addition. Helium raises arc energy density and improves penetration, but it is expensive and demands better gas-tightness. Bottles ride on the machine (10 to 20 L) or connect to an external manifold. The dead volume between solenoid and nozzle is an underrated detail: the longer the run, the later the shield arrives, and the first 0.3 s of every arc is where porosity is born.

4. Clamping and Centering: The Half-Millimetre War

Root gap windows are typically 2 to 3 mm with a 1.5 to 2 mm land. A centring error of 0.5 mm therefore produces a 3.5 mm gap on one side and 1.5 mm on the other. The wide side is prone to burn-through or concavity; the narrow side is prone to lack of fusion. That 0.5 mm can originate in four places: pipe ovality, bevel concentricity, shoe facing wear, and body attitude drift.

Field diagnosis should follow that order. Measure ovality first with an internal micrometer at 0, 90, 180 and 270 degrees; a spread above 3 mm means the pipe should be re-rounded or rejected. Check the bevel next by placing a square against the pipe end; more than 1 mm of deviation indicates tool runout or a mis-clamped bevelling machine. Inspect the shoes third; facing wear beyond 1.5 mm, or a polished “mirror” appearance that signals glazing, means replacement. Verify attitude last by running the body through an empty pipe and tracking the head centre with a laser or taut wire; more than 1 mm of drift points to drive-wheel synchronisation.

One coupling is routinely missed: clamping force changes the pipe’s actual shape, and the shape changes contact stress distribution. On wall thickness above 25 mm and grade X80, springback is pronounced — after the shoes reach set pressure, the pipe relaxes 0.2 to 0.4 mm, so the effective centring radius is smaller than nominal. The practical fix is a two-step clamp: overshoot to 1.2 times set pressure for three seconds, then release to set pressure. This lets the wall complete its plastic micro-adjustment and typically improves centring repeatability from ±0.4 mm to ±0.15 mm. On heavy-wall, high-grade pipe this is not optional.

Centring is also not a one-time event. It is a state maintained throughout the pass. Centrifugal force from the rotating ring, reciprocating inertia from the wire feeder and drag from torch cables all apply small disturbances. Heavier bodies, lower centres of gravity and longer shoe spans resist disturbance better. This is why large-diameter internal welders are long and heavy — not through over-engineering, but because stiffness is the mechanism.

5. Copper Backup and Backing Gas: Moulding the Molten Pool

The copper backup shoe is the soul of the machine. Mounted behind the torch and pressed against the bevel root, it presents a grooved copper block that supports the pool and forms a uniform internal bead. Copper conducts heat about eight times faster than steel, so deposited metal freezes on contact and does not stick — which is why one shoe survives thousands of joints.

Groove geometry governs bead profile. Three shapes dominate. The radius groove gives the smoothest profile and the lowest stress concentration, and is the default for transmission pipelines. The rectangular groove produces a fuller bead but tends to undercut at the toes. The trapezoidal groove sits between them and suits sections with varying wall thickness. Typical groove depth is 1.5 to 2.5 mm and width 4 to 6 mm, fixed by wall thickness and procedure qualification rather than by preference.

Contact pressure matters more than most operators assume. Too little and the pool leaks through the seam, producing concavity or an internal bead; too much and the shoe deforms the bevel root, widening the gap it was meant to control. A practical check is a feeler gauge between shoe and pipe wall: a 0.05 mm blade should not enter more than 5 mm. Cooling governs shoe life as well — after 20 consecutive joints, copper temperature can exceed 200 °C, and softened copper develops depressions within a few more passes. Water cooling or forced air is standard on production machines.

Backing gas is the other half. The copper shoe supports the pool but does not protect it: the root face is still exposed to internal air at high temperature, oxidising and picking up nitrogen to form a brittle layer. Better machines drill shielding holes around the shoe to create a local gas pocket. Whole-pipe argon purging is reserved for stainless, duplex and sour-service lines because of cost. Carbon steel transmission work generally relies on shoe ports with a short pre- and post-flow: two seconds before the arc, three seconds after. Get the timing wrong and you will find oxidation colour, or outright porosity, in the first and last 20 mm of every joint.

6. Torch Geometry: Work Angle, Travel Angle, Stick-Out

Three adjustable dimensions define torch attitude, and together they set arc force direction, pool shape and penetration distribution. This is where commissioning takes the most patience.

Work angle is the angle between torch axis and bevel face, normally 0 to 15 degrees. Too small and the arc favours the sidewall, inviting lack of sidewall fusion. Too large and the arc drives straight into the root, increasing the risk of the pool dropping. With twin torches, the two work angles must be strictly symmetric; otherwise penetration differs side to side and AUT returns the strange result of “acceptable on one side, rejectable on the other”.

Travel angle is the tilt relative to welding direction — push or drag. Internal welders generally run 5 to 10 degrees of push, which drives the pool backward over already-deposited metal and flattens the internal profile. Drag increases penetration but is far more sensitive to gap variation. On sites where gap scatter is high, push is the safer default.

Stick-out is contact-tip-to-work distance, usually 12 to 18 mm. Increasing it raises arc voltage, reduces penetration and degrades shielding; decreasing it deepens penetration but raises spatter build-up on the tip. The complication inside a pipe is that the bevel is a V: as the torch climbs the bevel, stick-out changes by geometry alone. Servo compensation must therefore be calibrated per angular position. A poorly calibrated table produces the classic signature of deep penetration at twelve o’clock and shallow penetration at three o’clock. Calibration is simple: tape a marker band inside the pipe end, run the body through without welding, log the height at each angle, and generate the compensation table from that log.

Weave width and frequency are the fourth and fifth variables. Root passes usually run with no weave, or 1 to 2 mm of micro-weave, because the groove is narrow and the pool small; weaving invites toe undercut. Wider oscillation is used only when the gap is out of tolerance and the operator needs to bridge it, and then only at reduced travel speed.

7. Root Penetration Control: From “It Penetrated” to “Every Pass Is the Same”

The minimum acceptance criterion is a continuous internal bead, 0 to 2 mm reinforcement, no concavity, no burn-through. What a project actually needs, however, is not that one joint penetrated but that three hundred joints behaved identically. The gap between those two statements is process capability.

Variables affecting penetration fall into four groups: energy input (current, voltage, travel speed), geometry (gap, land, bevel angle), material (grade, wall thickness, surface condition) and environment (preheat, wind, internal temperature). A manual welder reads keyhole size in real time and adjusts. A machine can only rely on preset parameters plus feedback compensation. The machine’s penetration capability is therefore, precisely, its ability to sense deviation and correct it.

The first sensing layer is arc voltage feedback. As penetration deepens, arc length changes and voltage shifts; the controller trims wire feed accordingly. This layer is fast — milliseconds — but handles only small disturbances. The second layer is bevel profile scanning. A laser maps the actual geometry of the whole circumference before the arc starts, including gap distribution, land height and hi-lo, then generates segment-wise parameters: lower current and higher speed where the gap is wide, the reverse where it is tight. This layer handles variation in the pipe ends themselves and is now standard on premium machines. The third layer is weld pool vision: a high-speed camera observes pool geometry and an image algorithm infers penetration state, closing the loop in real time. Still frontier technology, still expensive, but valuable on heavy-wall, high-grade, low-temperature work.

Above all three sits data traceability. Logging the parameter curve, scan result, operator and ambient temperature for every joint means that when AUT flags an indication you can trace it to the exact segment where a parameter drifted. Traceability does not improve penetration directly, but it converts “random” into “analysable”, which is the precondition for any real improvement. Several major pipeline projects in China now write this requirement into the technical specification.

8. Developing the Procedure Window: A Four-Factor Design

An internal welder cannot be qualified with a single parameter set. The same machine on 32-inch by 12.7 mm and 48-inch by 22 mm may need settings that differ by a factor of two. Practice is to map a window using a four-factor, three-level orthogonal array: current, travel speed, gap and hi-lo.

The method: fix gap at 2.5 mm and hi-lo at 0.5 mm on a standard coupon, run nine current-speed combinations, section each for macrography, measure penetration, bead width, internal reinforcement and HAZ hardness, and identify the region that gives full penetration with 0.5 to 1.5 mm reinforcement. Then repeat at gaps of 1.5 and 3.5 mm and hi-lo of 1.0 and 1.5 mm. The output is a three-dimensional envelope, not a single line. In the field the machine interpolates inside that envelope using its scan results — which is what “adaptive” actually means.

The most common mistake during development is optimising for metallurgy and ignoring throughput. Some combinations produce beautiful macrographs at 0.15 m/min, which means four minutes per joint and a dead spread. A sensible target is “the highest speed that meets ISO 9606 or API 1104 acceptance, derated by 15 percent”. That 15 percent comes from accumulated project experience with parameter sets pushed to the limit, which then produce batch rework the moment field conditions shift.

One more point is routinely skipped: qualification coupons are clean, field pipe is not. Coupons should include deliberately “dirty” specimens — light surface rust, a 0.3 mm internal misalignment step, 0.2 mm bevel burrs — to confirm the window still holds. A parameter library that survives the dirty-pipe test is a library that can go to site.

9. Bevel and Pipe-End Preparation: Upstream Tolerance Sets Downstream Yield

Internal welders are far more sensitive to upstream quality than manual welders. A skilled hand can absorb 2 mm of hi-lo with technique; a machine cannot. On a fully automatic line, the real quality control station is the bevelling machine, not the welder.

Four bevel metrics matter: angle deviation within ±1.5 degrees; land height deviation within ±0.5 mm; face roughness Ra not exceeding 12.5 µm; end squareness within 1 mm. Of these, land height scatter is the most punishing. A 0.5 mm increase in land raises the energy required for penetration by roughly 8 percent; if land varies from 1.5 to 2.2 mm around one circumference, a single parameter set will half-penetrate and half-fail.

Tool condition needs monitoring. As inserts wear, cutting force rises and the bevel face develops chatter marks — visually bright, microscopically 30 µm of waviness. The arc then follows the waviness and produces a snake-like bead. A practical rule is to index inserts every 50 joints and to check roughness with a comparator twice per shift. Cleaning matters equally: 25 mm either side of the bevel must be free of oil, rust and moisture; above roughly 0.1 mL per square centimetre of residual water, porosity rises measurably.

Circumference measurement is the other key action. Within one batch, pipes differing by more than 4 mm in circumference should not be paired directly. Projects that skip this step to protect the schedule end up with excessive hi-lo, centring difficulty and AUT rejection rates below 85 percent — and the rework cost dwarfs the time saved.

10. Defect Map: Six Root Defects and Their Causal Chains

Internal welder defects are relatively concentrated because the variables are constrained. In field statistics, more than 90 percent of rejections fall into six categories.

10.1 Lack of Root Fusion

The most typical and the most dangerous. A discontinuity remains between bevel face and deposited metal; AUT shows a linear indication along the bevel. Causes, in order of frequency: gap below 1.5 mm, low current, excessive travel speed, work angle deviation, oxide film on the bevel face. Diagnose by checking the current log first, then the scanned gap distribution, then torch attitude.

10.2 Burn-Through and Concavity

The combined result of excessive gap, thin land, high current and slow travel. Concavity is fundamentally a shortage of filler: the pool is drawn inward as it solidifies. Remedies are higher wire feed speed, lower heat input, weaving to bridge, or — in extreme cases — a manual low-hydrogen root repair.

10.3 Porosity

Internal welder porosity concentrates at arc start and arc stop; the middle of the pass is usually clean. Causes are shielding timing, gas-path dead volume and through-draught inside the pipe. Pre-flow under 1.5 s or post-flow under 2.5 s guarantees porosity in the first and last 20 to 30 mm. A subtler cause is bottle pressure falling in cold weather: the flowmeter reads normal while actual flow is short, so a heated regulator is required.

10.4 Poor Internal Bead Profile

Insufficient reinforcement means low heat input or an over-deep groove. Excessive reinforcement means high heat input or poor shoe contact. A groove-depth comparison test on the same coupon set separates process problems from shoe problems quickly.

10.5 Incomplete Penetration from Hi-Lo

Above 1.5 mm of hi-lo, the thin side is hard to penetrate and the thick side burns through easily. The machine centres on the pipe axis, so hi-lo translates directly into a gap difference. Countermeasures: stricter pipe pairing, clamping shoes with hi-lo compensation, or a manual backing run on the thick side.

10.6 Crater Cracking

Common on high grade, heavy wall and low ambient temperature. The pool solidifies rapidly at arc stop; sulphur and phosphorus segregate and hydrogen assists cracking. Countermeasures: a current down-slope of two to three seconds to 40 percent of working current, additional filler at the crater, and preheat raised by 20 to 30 °C.

11. Mountain Sections, Bends and Unequal Wall Thickness

Flat terrain is the easy case. Mountain sections are the real test. Flexible mountain pipelines routinely run at 6D to 20D bend radii with dense fittings, steep gradients and narrow working platforms. Three additional requirements appear: passability, modularity and vibration resistance.

Passability comes from articulation. Splitting the body into three to five modules with universal joints, each under 600 mm long, brings the minimum passable radius down to about 1.5D. The price is stiffness, and therefore centring accuracy; articulated machines compensate with more shoe groups — twelve or more — using multi-point support to recover rigidity.

Modularity means separating the power unit from the head. Mountain sites often cannot crane a complete machine into position; everything goes in by hand or small winch. Total mass should stay under 180 kg with no single piece above 60 kg, and assembly should be fast. Assembly time feeds directly into cycle time; the best machines in the market assemble in eight minutes.

Vibration resistance is where mountain work most often fails. Pipe rests on skids, and vibration from the travelling body is amplified by the wall, producing torch chatter and arc instability. Countermeasures include a 30 percent reduction in travel speed, 15 percent higher clamping pressure, dampers on the torch mount and pulsed current to stabilise the pool. In cold weather, hydraulic oil viscosity rises and shoe actuation slows, so low-temperature oil and an oil pre-heater are needed.

Unequal wall transitions are a separate problem. Going from 22 mm to 26.4 mm changes the bevel to a compound form, and the machine must switch parameters by segment. Some projects weld in halves — thin side first, thick side second, with a 30 mm overlap. The critical detail is that the parameter transition across the overlap must be gradual; a step change in current guarantees lack of fusion at the overlap point.

12. Selection Checklist and Factory Acceptance Testing

An internal welder is a high-unit-price, long-lead, low-tolerance purchase. Work through the following list item by item rather than reading the datasheet headline.

  • Diameter and wall range: check accuracy at the minimum diameter, not the maximum. Many machines claim 20 to 56 inches but deliver ±1 mm centring at 20 inches.
  • Shoe count and control: twelve or more groups with independent or grouped control beats global synchronisation; look for clamping-force closed loop and ovalisation feedback.
  • Travel synchronisation: demand under 1 mm centre drift on an empty run, under 1.5 mm loaded.
  • Torch count and symmetry: twin opposing is mainstream; four torches suit large-diameter high-speed lines; single torch suits small diameters and confined spaces.
  • Bevel scanning: laser profiling, accuracy better than 0.1 mm, automatic segment parameter generation.
  • Parameter library and data interface: at least 200 stored procedures, CSV export or direct integration with the project quality platform, remote diagnostics.
  • Gas and power integration: bottle capacity, manifold support, tolerance of ±10 percent generator voltage swing — common on remote sites.
  • Handling design: piece mass, lifting points, dedicated rigging, assembly time.
  • Spares and training: annual consumption of copper shoes, shoe facings, contact tips and drive rolls; whether the supplier can be on site within 48 hours.

Factory acceptance should not stop at “it welds”. Insist on four tests. First, centring repeatability: clamp the same pipe ten times and record the spread of head centre offset. Second, empty travel trace: log the deviation curve over full length. Third, standard coupon weld with macrography and hardness. Fourth, a continuous run of 20 joints, watching copper shoe temperature, hydraulic oil temperature and parameter drift. Pass all four and the machine is genuinely ready.

One commercial clause deserves equal attention: days of on-site process support. An internal welder in a laboratory and an internal welder on a spread are two different species, and the first hundred joints need a manufacturer’s engineer present. The contract should fix the number of resident days, the response time, and who bears the cost of rework attributable to the equipment. This clause generates more disputes than any other, and writing it clearly in advance saves a great deal of argument later.

13. Closing Notes

The technical content of a pipeline automatic internal welder is not in the word “automatic”. It is in converting a skill that lived in a welder’s hands into a problem of geometry and energy control. Centring accuracy sets gap distribution. Gap distribution sets the penetration window. The penetration window sets whether parameters can run at the upper edge of productivity. And that edge sets the throughput of the whole spread. Loosen any link in this chain and it reappears in the cost sheet as cut-out joints.

For a construction company, buying an internal welder is really buying a reproducible root-pass capability. Equipment specifications age; three principles do not: upstream tolerance comes first, procedure windows need headroom, and defects should be diagnosed by causal chain rather than by guesswork. Writing those three into the work instructions raises acceptance rates more reliably than buying a more expensive machine.

Beijing Anjie Weida Technology Co., Ltd. (DENOH Group) focuses on mobile welding and mobile energy equipment. The product range covers gasoline and diesel engine-driven welding generators, engine-driven welders, pipeline automatic internal welders, single- and dual-torch external welders, flexible mountain pipeline welding systems, welding engineering vehicles, mobile energy storage welders and mobile industrial power supplies, together with integrated solar-storage-diesel microgrid solutions. The company supports projects from procedure qualification and parameter library development through equipment supply, resident technical service and operator training, sized to the actual diameter, wall thickness, terrain, climate and acceptance standard. Whether you need a single machine or a complete spread, we welcome a conversation about your specific conditions.

Contact Us

Beijing Anjie Weida Technology Co., Ltd. (DENOH Group)

Tel: 86-010-86468776

Mobile / WeChat: 13521628344

Email: sales@denohgroup.com

Website: https://www.denohgroup.com

新产品&新技术

相关文章