Welding sequence and balance

The same welds, made in a different order, give a differently shaped part. Chapter 3 showed that a girder welded one flange at a time bows, while the same girder welded with its flanges in alternation comes out nearly straight. Chapter 4 showed that a double-V joint completed one side at a time kinks, while the same joint welded side by side stays flat. Sequence costs almost nothing in materials, sometimes a little in time, and it is one of the most effective distortion controls available.

This chapter sets out the principles of welding sequence: balance about the neutral axis, the back-step and skip methods along seams, working from the middle outwards and from the restrained parts to the free ones, and building assemblies from sub-assemblies. It shows how the shop wrote sequences for its girders and base frames.

Balance about the axis

The first principle is balance. Welds on opposite sides of a member’s neutral axis, or of a plate’s mid-thickness, pull in opposite directions; if they are made so that neither side gets far ahead of the other, their pulls cancel as the member stiffens. This means:

• welding both flanges of a girder in alternation, rather than one completely first;
• welding both sides of a double-V joint in alternation, a few passes at a time;
• welding pairs of fillets on opposite sides of a web together, or alternately;
• welding attachments on opposite faces of a member in matching pairs.

Balance is easiest with two welders working on opposite sides at once, or with a twin-head mechanised carriage, and with fixtures that let the part be turned quickly. Where only one welder is available, short alternating lengths achieve much of the benefit.

Along a seam: back-step welding

A long weld made continuously from one end to the other pushes its shrinkage ahead of itself: each length cools and pulls on the unwelded remainder, accumulating gap movement and distortion towards the far end, as Whittemore’s creep showed in Chapter 2. In back-step welding, the seam is divided into short lengths, and each length is welded in the direction opposite to the overall progress: the welder starts a short distance from the beginning, welds back to the start, moves on, and welds back to the end of the previous length. Each length’s shrinkage is then balanced by the length before it, and the gap movement does not accumulate. Weld Repairs That Hold describes back-step welding in plate repairs, with the lengths used in a 1919 booklet, in its chapter on welding sequence: back-step and balance.

Skip welding

Skip welding spreads heat along a seam by welding short lengths at intervals, skipping the spaces between, then returning to fill them. The heat is distributed rather than concentrated, each length cools before its neighbour is welded, and the local expansion and buckling of thin sheet are reduced. It is the principle behind the stitch sequence for the shop’s stainless panels in Chapter 5. On thicker plate, skip welding’s benefit is smaller and its extra stops and starts add defects, so it is used where heat concentration is the main problem.

Figure 1Sequences along a seam

	How	Best for	Watch for
Continuous, end to end	one run from start to finish	short seams, thick plate	gap movement and distortion accumulate towards the far end
Back-step	short lengths welded back towards the start	long seams; plate and repairs	more stops and starts, each to be sound
Skip	spaced lengths, gaps filled later	thin sheet; heat-sensitive work	many stops; cooling time between lengths
Middle outwards	from the centre towards each end	seams with both ends free or restrained alike	needs two directions of travel

Middle outwards and restrained to free

Two further principles guide the order of welds in a part. Welding from the middle of a seam or an assembly outwards lets the shrinkage move towards free ends, rather than building up against a closed end; Dunham’s advice to weld a fender crack from the centre outwards, leaving the end untacked, followed this principle. And welding the most restrained joints first, while the rest of the assembly is still free to move, and the least restrained last, lets the assembly adjust as it is built up, rather than locking the last joints into a rigid frame where they crack. A frame whose last joint closes a ring of members is the classic case of a joint welded under maximum restraint; where possible, such closing joints are designed out, or made with extra care, preheat and a balanced sequence.

Sub-assemblies

Large assemblies distort less if they are built from sub-assemblies, each welded and, if necessary, straightened before being joined to the others. A sub-assembly is small enough to handle, turn and fixture, and its distortion can be corrected while it is still accessible. The final joining welds are then fewer, and their distortion is easier to control. The shop’s base frames were redesigned to be built from two welded side frames and a set of cross members, each side frame welded flat in a fixture and checked before the frame was assembled, rather than welding all the members together on the floor at once.

The girder sequence

This is an illustration. The shop’s written sequence for its plate girders became:

Figure 2The shop's welding sequence for an 8 m plate girder

1. 1Tackflanges to web at 300 mm spacing, top and bottom, both sides, working from mid-length outwards
2. 2Root runstop flange welds, 1 m each side of mid-length, then the same on the bottom flange
3. 3Alternatecontinue outwards in 1 m lengths, top then bottom, both sides together with the twin-head carriage
4. 4Stiffenersweld in pairs on opposite sides of the web, from mid-length outwards
5. 5Checkmeasure camber, sweep and flange tilt when cool; correct if needed (Chapter 9)

The sequence takes longer than welding one flange end to end, because the girder is turned and the carriage moved more often, but the girders now come out within tolerance without straightening, which saves far more time than the sequence costs.

Interpass temperature and waiting

Sequence interacts with heat. Making a second pass on a joint before the first has cooled adds heat to an already hot joint, widening the heated band and increasing distortion; the welding procedure often sets a maximum interpass temperature for metallurgical reasons, which also limits distortion. On thin and stainless steel work, where heat build-up causes buckling, the sequence is planned so that the welder moves to a different part of the work while each weld cools, rather than waiting idle, which keeps the work moving and the heat spread.

Tack welds as part of the sequence

The sequence begins with the tacks. Tacks placed from the middle of a joint outwards, at even spacing, fix the parts in position and set the gap before any heavy welding; tacks placed from one end let the gap drift as each tack is made. Tacks also restrain the joint, so their size and number are part of the distortion plan: too few, and the parts move; too many and too large, and the joint is restrained heavily from the start. Tacks are made with the same procedure and care as the main weld, inspected before they are welded over, and any cracked tack ground out, because a crack under a weld is a crack in the weld.

Long seams in several sections

Very long seams, such as the full length of a large girder or a ship’s deck plate, are welded in sections by several welders at once, each working outwards from a starting point, or by welders starting at the middle and working towards the ends together. The sections are arranged so that each welder’s shrinkage is balanced by a neighbour’s, and so that no section is welded against a part already locked by its neighbours on both sides. The planning of such sequences is a skilled job, and on large structures it is done by the welding engineer and drawn up as a sequence plan before work starts.

Sequence on thin sheet

On thin sheet, the sequence is mostly about heat. Short, spaced welds, made so that each cools before its neighbour is welded, keep the sheet from building up the heated region that causes buckling; welds made from the middle of a panel’s edge outwards, or from the corners inwards, avoid driving the shrinkage into a corner. The shop’s panel sequence, described in Chapter 5, numbers every stitch on a laminated card at the welding station, so that every welder follows the same order.

Checking as you go

A sequence is a prediction, and checking during welding catches the occasions when the prediction is wrong. On critical assemblies, the shop’s welders measure a key dimension, such as a girder’s camber or a frame’s diagonal, after each stage of the sequence, and the supervisor can change the order of the remaining welds to correct a trend before it becomes a problem: welding more on the side that will pull the part back, or pausing to let it cool. A part checked only at the end can only be straightened; one checked as it is built can be steered.

Writing the sequence down

For simple parts, the welder chooses the sequence. For critical parts, where distortion has caused trouble or tolerances are tight, the sequence is written into the welding procedure or a sequence drawing, with each weld numbered in order and the direction of welding shown. A written sequence makes the result repeatable from welder to welder and shift to shift, and it lets the shop learn: when a part comes out of tolerance, the sequence actually used can be compared with the written one. The shop now issues a sequence sheet with every girder and base frame drawing.

Sequence and productivity

A good sequence is not always slower. Mechanised carriages with twin heads make balanced flange welding faster than one welder working one side at a time. Fixtures that turn the part let welders work in the flat position. Sub-assemblies let several welders work in parallel. The shop found that its new girder sequence, with the rotating fixture and twin-head carriage, was about as fast overall as its old method, once the straightening it had needed was counted.

Training welders in sequence

A written sequence works only if welders understand why it matters. Welders used to finishing one side before starting the next find alternating sides slow and awkward at first, and under time pressure they drift back to the old habit. The shop’s production manager showed the welders the measurements: girders welded one flange at a time against girders welded in alternation, panels welded continuously against panels stitched in sequence. Once the welders could see the difference in numbers, and the straightening they no longer had to do, the sequences stuck. Welder Training by the Numbers discusses how procedures are taught and kept in practice; for distortion, the lesson is that the reason travels with the rule.

When sequence is not enough

Sequence cannot cancel distortion that is built into the design, such as heavy welds on one side of a member with nothing to balance them, or a thin panel with a heat input far above what it can absorb. In those cases, the sequence reduces the problem but does not remove it, and presetting, fixtures, a change of weld size or a change of design are needed as well. The next chapter deals with presetting and fixtures, which work alongside sequence rather than instead of it.

A failure: the frame closed last

This is an illustration. A rectangular machine frame of heavy hollow sections had been welded corner by corner round the rectangle, the last corner closing the ring. The last corner weld cracked as it cooled, and the frame, when measured, was out of square by several millimetres.

By the time the last corner was welded, the other three had made the frame a rigid ring, and the last weld’s shrinkage had nowhere to go but into stress, which cracked it, and into pulling the frame out of square. The revised sequence tacked all four corners, then welded opposite corners in pairs, part of each weld at a time, so that no corner was ever the last to close a rigid ring. The next frame was square and uncracked.

Exercise: plan a sequence

A long box beam is made of two channels joined by two continuous longitudinal fillet welds, one on the top face and one on the bottom face, each 6 m long. Sketch a welding sequence that will keep the beam straight, and say what would happen if the top weld were made completely before the bottom.

A balanced sequence would tack both seams at intervals from mid-length outwards, then weld the top and bottom seams in alternation, in short lengths of perhaps 500 mm, working from mid-length towards each end, turning the beam between lengths, or using two welders or a carriage on top and bottom at once. Back-step lengths within each section would limit gap movement.

If the top weld were made completely first, its shrinkage would bow the beam upwards at the ends (towards the weld) while the bottom was unwelded and the section was less stiff; the bottom weld, made afterwards on a stiffer section, would pull it back only partly, leaving the beam bowed towards the top. The beam would need straightening, as the shop’s girders did before their sequence was changed.