Distortion – Types and causes
This article covers several key issues on distortion in arc welded fabrications, especially basic types of and factors affecting the degree of distortion.
What causes distortion?
Because welding involves highly localised heating of joint edges to fuse the material, non-uniform stresses are set up in the component because of expansion and contraction of the heated material. Initially, compressive stresses are created in the surrounding cold parent metal when the weld pool is formed due to the thermal expansion of the hot metal (heat affected zone) adjacent to the weld pool. However, tensile stresses occur on cooling when the contraction of the weld metal and the immediate heat affected zone is resisted by the bulk of the cold parent metal.
The magnitude of thermal stresses induced into the material can be seen by the volume change in the weld area on solidification and subsequent cooling to room temperature. For example, when welding C-Mn steel, the molten weld metal volume will be reduced by approximately 3% on solidification and the volume of the solidified weld metal/heat affected zone (HAZ) will be reduced by a further 7% as its temperature falls from the melting point of steel to room temperature.
If the stresses generated from thermal expansion/contraction exceed the yield strength of the parent metal, localised plastic deformation of the metal occurs. Plastic deformation causes a permanent reduction in the component dimensions and distorts the structure.
What are the main types of distortion?
Distortion occurs in six main forms:
- Longitudinal shrinkage
- Transverse shrinkage
- Angular distortion
- Bowing and dishing
The principal features of the more common forms of distortion for butt and fillet welds are shown below:
Contraction of the weld area on cooling results in both transverse and longitudinal shrinkage.
Non-uniform contraction (through thickness) produces angular distortion in addition to longitudinal and transverse shrinkage.
For example, in a single V butt weld, the first weld run produces longitudinal and transverse shrinkage and rotation. The second run causes the plates to rotate using the first weld deposit as a fulcrum. Hence, balanced welding in a double side V butt joint can be used to produce uniform contraction and prevent angular distortion.
Similarly, in a single side fillet weld, non-uniform contraction produces angular distortion of the upstanding leg. Double side fillet welds can therefore be used to control distortion in the upstanding fillet but because the weld is only deposited on one side of the base plate, angular distortion will now be produced in the plate.
Longitudinal bowing in welded plates happens when the weld centre is not coincident with the neutral axis of the section so that longitudinal shrinkage in the welds bends the section into a curved shape. Clad plate tends to bow in two directions due to longitudinal and transverse shrinkage of the cladding; this produces a dished shape. Dishing is also produced in stiffened plating. Plates usually dish inwards between the stiffeners, because of angular distortion at the stiffener attachment welds (see main photograph).
In plating, long range compressive stresses can cause elastic buckling in thin plates, resulting in dishing, bowing or rippling.
Distortion due to elastic buckling is unstable: if you attempt to flatten a buckled plate, it will probably ‘snap’ through and dish out in the opposite direction.
Twisting in a box section is caused by shear deformation at the corner joints. This is caused by unequal longitudinal thermal expansion of the abutting edges. Increasing the number of tack welds to prevent shear deformation often reduces the amount of twisting.
How much shall I allow for weld shrinkage?
It is almost impossible to predict accurately the amount of shrinking. Nevertheless, a ‘rule of thumb’ has been composed based on the size of the weld deposit. When welding steel, the following allowances should be made to cover shrinkage at the assembly stage.
Fillet Welds 0.8mm per weld where the leg length does not exceed 3/4 plate thickness
Butt weld 1.5 to 3mm per weld for 60° V joint, depending on number of runs
Fillet Welds 0.8mm per 3m of weld
Butt Welds 3mm per 3m of weld
Increasing the leg length of fillet welds, in particular, increases shrinkage.
What are the factors affecting distortion?
If a metal is uniformly heated and cooled there would be almost no distortion. However, because the material is locally heated and restrained by the surrounding cold metal, stresses are generated higher than the material yield stress causing permanent distortion. The principal factors affecting the type and degree of distortion, are:
- Parent material properties
- Amount of restraint
- Joint design
- Part fit-up
- Welding procedure
- Parent material properties
Parent material properties which influence distortion are coefficient of thermal expansion and specific heat per unit volume.
As distortion is determined by expansion and contraction of the material, the coefficient of thermal expansion of the material plays a significant role in determining the stresses generated during welding and, hence, the degree of distortion. For example, as stainless steel has a higher coefficient of expansion than plain carbon steel, it is more likely to suffer from distortion.
If a component is welded without any external restraint, it distorts to relieve the welding stresses. So, methods of restraint, such as ‘strong-backs’ in butt welds, can prevent movement and reduce distortion. As restraint produces higher levels of residual stress in the material, there is a greater risk of cracking in weld metal and HAZ especially in crack-sensitive materials.
Both butt and fillet joints are prone to distortion. It can be minimised in butt joints by adopting a joint type which balances the thermal stresses through the plate thickness. For example, a double-sided in preference to a single-sided weld. Double-sided fillet welds should eliminate angular distortion of the upstanding member, especially if the two welds are deposited at the same time.
Fit-up should be uniform to produce predictable and consistent shrinkage. Excessive joint gap can also increase the degree of distortion by increasing the amount of weld metal needed to fill the joint. The joints should be adequately tacked to prevent relative movement between the parts during welding.
This influences the degree of distortion mainly through its effect on the heat input. As welding procedure is usually selected for reasons of quality and productivity, the welder has limited scope for reducing distortion. As a general rule, weld volume should be kept to a minimum. Also, the welding sequence and technique should aim to balance the thermally induced stresses around the neutral axis of the component.
Distortion – prevention by design
General guidelines are given below as ‘best practice’ for limiting distortion when considering the design of arc welded structures.
At the design stage, welding distortion can often be prevented, or at least restricted, by considering:
- elimination of welding
- weld placement
- reducing the volume of weld metal
- reducing the number of runs
- use of balanced welding
Elimination of welding
As distortion and shrinkage are an inevitable result of welding, good design requires that not only the amount of welding is kept to a minimum, but also the smallest amount of weld metal is deposited. Welding can often be eliminated at the design stage by forming the plate or using a standard rolled section, as shown in Fig 1.
Fig. 1 Elimination of welds by:
a) Forming the plate;
b) Use of rolled or extruded section
If possible, the design should use intermittent welds rather than a continuous run, to reduce the amount of welding. For example, in attaching stiffening plates, a substantial reduction in the amount of welding can often be achieved whilst maintaining adequate strength.
Placing and balancing of welds are important in designing for minimum distortion. The closer a weld is positioned to the neutral axis of a fabrication, the lower the leverage effect of the shrinkage forces and the final distortion.
Examples of poor and good designs are shown in Fig 2.
As most welds are deposited away from the neutral axis, distortion can be minimised by designing the fabrication so the shrinkage forces of an individual weld are balanced by placing another weld on the opposite side of the neutral axis.
Whenever possible, welding should be carried out alternately on opposite sides, instead of completing one side first. In large structures, if distortion is occurring preferentially on one side, it may be possible to take corrective actions, for example, by increasing welding on the other side to control the overall distortion.
Reducing the volume of weld metal
To minimise distortion, as well as for economic reasons, the volume of weld metal should be limited to the design requirements.
For a single-sided joint, the cross-section of the weld should be kept as small as possible to reduce the level of angular distortion, as illustrated in Fig 3.
Fig. 3 Reducing the amount of angular distortion and lateral shrinkage by:
a) reducing the volume of weld metal;
b) using single pass weld
Joint preparation angle and root gap should be minimised providing the weld can be made satisfactorily.
To facilitate access, it may be possible to specify a larger root gap and smaller preparation angle. By cutting down the difference in the amount of weld metal at the root and the face of the weld, the degree of angular distortion will be correspondingly reduced. Butt joints made in a single pass using deep penetration have little angular distortion, especially if a closed butt joint can be welded (Fig 3).
For example, thin section material can be welded using plasma and laser welding processes and thick section can be welded, in the vertical position, using electrogas and electroslag processes.
Although angular distortion can be eliminated, there will still be longitudinal and transverse shrinkage.
In thick section material, as the cross sectional area of a double-V joint preparation is often only half that of a single-V preparation, the volume of weld metal to be deposited can be substantially reduced. The double-V joint preparation also permits balanced welding about the middle of the joint to eliminate angular distortion.
As weld shrinkage is proportional to the amount of weld metal, both poor joint fit-up and over-welding will increase the amount of distortion. Angular distortion in fillet welds is particularly affected by over-welding. As design strength is based on throat thickness, over-welding to produce a convex weld bead does not increase the allowable design strength but it will increase the shrinkage and distortion.
Reducing the number of runs
There are conflicting opinions on whether it is better to deposit a given volume of weld metal using a small number of large weld passes or a large number of small passes.
Experience shows that for a single-sided butt joint, or a single-side fillet weld, a large single weld deposit gives less angular distortion than if the weld is made with a number of small runs.
Generally, in an unrestrained joint, the degree of angular distortion is approximately proportional to the number of passes.
Completing the joint with a small number of large weld deposits results in more longitudinal and transverse shrinkage than a weld completed in a larger number of small passes. In a multi-pass weld, previously deposited weld metal provides restraint, so the angular distortion per pass decreases as the weld is built up. Large deposits also increase the risk of elastic buckling particularly in thin section plate.
Use of balanced welding
Balanced welding is an effective means of controlling angular distortion in a multi-pass butt weld by arranging the welding sequence to ensure that angular distortion is continually being corrected and not allowed to accumulate during welding.
Comparative amounts of angular distortion from balanced welding and welding one side of the joint first are shown schematically in Fig 4. The balanced welding technique can also be applied to fillet joints.
If welding alternately on either side of the joint is not possible, or if one side has to be completed first, an asymmetrical joint preparation may be used with more weld metal being deposited on the second side.
The greater contraction resulting from depositing the weld metal on the second side will help counteract the distortion on the first side.
The following design principles can control distortion:
- eliminate welding by forming the plate and using rolled or extruded sections
- minimise the amount of weld metal
- do not over weld
- use intermittent welding in preference to a continuous weld pass
- place welds about the neutral axis
- balance the welding about the middle of the joint by using a double-V joint in preference to a single-V joint
Adopting best practice principles can have surprising cost benefits. For example, for a design fillet leg length of 6mm, depositing an 8mm leg length will result in the deposition of 57% additional weld metal.
Besides the extra cost of depositing weld metal and the increase risk of distortion, it is costly to remove this extra weld metal later.
However, designing for distortion control may incur additional fabrication costs. For example, the use of a double-V joint preparation is an excellent way to reduce weld volume and control distortion, but extra costs may be incurred in production through manipulation of the workpiece for the welder to access the reverse side.
Distortion Control – Prevention by fabrication techniques
In general, the welder has little influence on the choice of welding procedure but assembly techniques can often be crucial in minimising distortion. The principal assembly techniques are:
- tack welding
- back-to-back assembly
Tack welds are ideal for setting and maintaining the joint gap but can also be used to resist transverse shrinkage. To be effective, thought should be given to the number of tack welds, their length and the distance between them. With too few, there is the risk of the joint progressively closing up as welding proceeds. In a long seam, using MMA or MIG, the joint edges may even overlap. It should be noted that when using the submerged arc process, the joint might open up if not adequately tacked.
The tack welding sequence is important to maintain a uniform root gap along the length of the joint. Three alternative tack welding sequences are shown in Fig. 1:
a) tack weld straight through to the end of the joint (Fig 1a). It is necessary to clamp the plates or to use wedges to maintain the joint gap during tacking
b) tack weld one end and then use a back stepping technique for tacking the rest of the joint (Fig 1b)
c) tack weld the centre and complete the tack welding by back stepping (Fig 1c).
a) tack weld straight through to end of joint
b) tack weld one end, then use back-step technique for tacking the rest of the joint
c) tack weld the centre, then complete the tack welding by the back-step technique
Directional tacking is a useful technique for controlling the joint gap, for example closing a joint gap which is (or has become) too wide.
When tack welding, it is important that tacks which are to be fused into the main weld are produced to an approved procedure using appropriately qualified welders. The procedure may require preheat and an approved consumable as specified for the main weld. Removal of the tacks also needs careful control to avoid causing defects in the component surface.
By tack welding or clamping two identical components back-to-back, welding of both components can be balanced around the neutral axis of the combined assembly (Fig. 2a). It is recommended that the assembly is stress relieved before separating the components. If stress relieving is not done, it may be necessary to insert wedges between the components (Fig. 2b) so when the wedges are removed, the parts will move back to the correct shape or alignment.
a) assemblies tacked together before welding
b) use of wedges for components that distort on separation after welding
Longitudinal shrinkage in butt welded seams often results in bowing, especially when fabricating thin plate structures. Longitudinal stiffeners in the form of flats or angles, welded along each side of the seam (Fig. 3) are effective in preventing longitudinal bowing. Stiffener location is important: they must be placed at a sufficient distance from the joint so they do not interfere with welding, unless located on the reverse side of a joint welded from one side.
A suitable welding procedure is usually determined by productivity and quality requirements rather than the need to control distortion. Nevertheless, the welding process, technique and sequence do influence the distortion level.
General rules for selecting a welding process to prevent angular distortion are:
- deposit the weld metal as quickly as possible
- use the least number of runs to fill the joint
Unfortunately, selecting a suitable welding process based on these rules may increase longitudinal shrinkage resulting in bowing and buckling.
In manual welding, MIG, a high deposition rate process, is preferred to MMA. Weld metal should be deposited using the largest diameter electrode (MMA), or the highest current level (MIG), without causing lack-of-fusion imperfections. As heating is much slower and more diffuse, gas welding normally produces more angular distortion than the arc processes.
Mechanised techniques combining high deposition rates and high welding speeds have the greatest potential for preventing distortion.
As the distortion is more consistent, simple techniques such as presetting are more effective in controlling angular distortion.
General rules for preventing distortion are:
- keep the weld (fillet) to the minimum specified size
- use balanced welding about the neutral axis
- keep the time between runs to a minimum
In the absence of restraint, angular distortion in both fillet and butt joints will be a function of the joint geometry, weld size and the number of runs for a given cross section. Angular distortion (measured in degrees) as a function of the number of runs for a 10mm leg length fillet weld is shown in Fig. 4.
If possible, balanced welding around the neutral axis should be done, for example on double sided fillet joints, by two people welding simultaneously. In butt joints, the run order may be crucial in that balanced welding can be used to correct angular distortion as it develops.
a) Back-step welding
b) Skip welding
The sequence, or direction, of welding is important and should be towards the free end of the joint. For long welds, the whole of the weld is not completed in one direction. Short runs, for example using the back-step or skip welding technique, are very effective in distortion control (Fig. 5).
- Back-step welding involves depositing short adjacent weld lengths in the opposite direction to the general progression (Fig. 5a).
- Skip welding is laying short weld lengths in a predetermined, evenly spaced, sequence along the seam (Fig. 5b). Weld lengths and the spaces between them are generally equal to the natural run-out length of one electrode. The direction of deposit for each electrode is the same, but it is not necessary for the welding direction to be opposite to the direction of general progression.
The following fabrication techniques are used to control distortion:
- using tack welds to set up and maintain the joint gap
- identical components welded back to back so welding can be balanced about the neutral axis
- attachment of longitudinal stiffeners to prevent longitudinal bowing in butt welds of thin plate structures
- where there is choice of welding procedure, process and technique should aim to deposit the weld metal as quickly as possible; MIG in preference to MMA or gas welding and mechanised rather than manual welding
- in long runs, the whole weld should not be completed in one direction; back-step or skip welding techniques should be used.
These articles were prepared by Bill Lucas in collaboration with Geert Verhaeghe, Rick Leggatt and Gene Mathers.
Copyright © TWI Ltd 2014
The content of this article was correct at the time of publication. For more information visit www.twi-global.com