Measurement and analysis of thermal cycles in submerged arc welding

TWI Bulletin, January 1982

by J A Barlow, BSc

James Barlow is a Research Engineer with the Production Research Section, Arc Welding Department.

The thermal cycle of a weld directly affects its properties by controlling the development of the microstructure. Measurement of the thermal cycle should therefore allow weld microstructure and subsequently properties to be estimated without the need for lengthy sectioning and microexaminations. To do this, measurement and analysis techniques have been developed which are described here. It is intended that subsequent articles will describe the use of these techniques to examine practical production problems.

The submerged arc (SA) welding process offers considerable scope for the development of procedure selection systems based on procedural welding information. The link between welding parameters and weld bead geometry has already been the subject of detailed studies at The Welding Institute,^[1,2] but there remains a requirement to obtain better control over weld quality, and it is necessary to extend these links to include mechanical properties and other metallurgical quantities. Work in this field has so far been limited to specific material compositions^[3] or to simulated thermal cycle specimens.^[4] Mathematical solutions have mainly concentrated on weld bead shape,^[5] because of the volume of data required to relate welding procedures to mechanical properties or microstructure.

To obtain an accurate indication of the metallurgical effects of different welding procedures, it is necessary to have detailed information on the composition of the parent plate and weld metal and their thermal history. As the weld metal composition may be estimated from the dilution level and element transfer characteristics generated from flux/metal studies,^[6] weld metal and heat affected zone (HAZ) thermal history remains a vital link between metallurgical and procedural areas.

This field warrants increased study, as several gaps in our knowledge exist, namely:

• The relationships between welding procedure and weld thermal cycle are not fully understood and it is not certain what potential is available for thermal cycle control within acceptable tolerances for a given welding technique;
• The relationships between thermal cycle and metallurgical transformation need further clarification before metallurgical predictions are possible on a large scale.

A predictive system relating metallurgical and procedural areas is a long term objective of the programme, but before this can be achieved, it is necessary to have reliable tools for measurement of thermal cycles, and a means of predicting thermal cycles from given welding parameters in both HAZ and weld metal. Outline techniques to do these jobs have been reported.^[7,8] The present article serves to update them and to bring together techniques for measuring thermal cycles both in and around the weld.

Development of thermal cycle measurement techniques

For the HAZ

The thermal cycle induced by the welding operation determines the effect on the metallurgical characteristics of the parent material, producing an HAZ around the fusion line. To be able to define this effect it is necessary to measure the thermal cycle. A logical experimental method is to introduce a suitable thermal sensor into the HAZ and to record the resulting thermal cycle. For this, the position of interest relative to the weld fusion line must be known, and the sensors must be positioned accurately at this point before the welding operation. Significant disturbance to heat flow must be avoided and for this reason the sensors are placed normal to the fusion line, requiring a drill hole to be made at an angle to the plate surface. The precise position is obtained from a trial weld.

Methods of thermal cycle measurement considered in this work included use of optical infrared pyrometry and electronic photosensors viewing down a small flat ended drill hole, and conventional or unconventional thermocouple arrangements. In the last category the most promising method involved the use of a single insulated wire in a drill hole running right through the plate. On welding, the thermocouple wire was destroyed down to the fusion line and the wire end then became one half of the thermocouple hot junction, the parent plate itself acting as the other half. This method was eventually rejected because of the need to recalibrate the system for every plate composition and, further, each sensor position would give a different output for the same weld because of different emf path lengths between the output wires of the hot junctions.

The use of infrared pyrometry was unsatisfactory because of the need to drill deep flat ended holes of small diameter with extreme accuracy. Unfortunately, these requirements could not easily be met with portable tools, use of which was made necessary by the need to study multipass SA welds, with on-site drilling operations.

The only really effective sensor system available for HAZ thermal cycle measurements in these experimental circumstances was the conventional two wire thermocouple assembly. The further requirement of a mobile on-site tool for drilling sensor channels, restricted the choice to a standard electric drill assembly mounted on a mobile stand. In practice, a drill hole diameter of 3.2mm was selected, to allow adequate clearance for the thermocouples and attachment of the hot junction to the base of the holes by means of capacitor discharge welding. Figure 1 shows a drill hole for locating a thermocouple at the weld fusion line bay position.

The technique used involved drilling a hole completely through the plate, subsequently sealing this with a suitably dimensioned plug. Thermocouples were capacitor discharge welded on to the plugs before the whole assembly was inserted.

These large diameter holes may restrict the use of the technique, and much smaller diameters are possible, but these are limited by the availability of suitable thermocouple wires and insulation, although coating the wires with a refractory mixture may eliminate the need for ceramic insulators. Holes of 1.5mm diameter have been used in some work^[9] (Fig.2).

The most suitable thermocouple for measurement around the fusion line is platinum/platinum 13% rhodium which can measure temperatures from below those of molten steel to room temperature. However, Chromel-Alumel is more sensitive to temperatures below 1100°C, and may be preferred for measurements at a distance from the fusion line. In either case, a fabricated hot junction is required. This is kept as small as possible to improve sensitivity, and is made in a carbon arc.

For the weld pool

The presence of the slag and unused flux cover prohibits use of optical systems for SA weld pool thermal studies, and thermocouple implanting techniques are commonly used. The technique is simple in concept and involves injecting, or 'harpooning', the thermocouple into the weld pool behind the welding electrode.

The harpoon device used in the initial study by Kohno and Jones^[10] was hand triggered, making accurate positioning within the weld pool difficult. A redesigned Mark II version on a modified more versatile mounting was manufactured and used in later tests (Fig.3). This system incorporated an automatic release mechanism triggered from a remote proximity sensor. A further modification was made to allow the harpoon device to be removed, leaving an intact thermocouple implanted in the weld to obtain subsequent measurements in multipass welds.

A harpooning technique was then developed for positioning a thermocouple hot junction to within ±1mm in the three dimensional space of a weld pool zone. The technique finally chosen involved manufacture of a brass template of the weld fusion line, which was positioned along the weld path, to allow positioning of the harpoon thermocouple in three dimensions (Fig.4). The jig was then replaced by a test plate and the harpoon thermocouple released automatically by the welding head as it passed the proximity release sensor.

The injection angle was precisely measured before each test and was designed to place the thermocouple tip as close to the arc cavity as possible, without hitting the region of intense heat which extends back towards the weld tail at the slag/weld pool interface (Fig.5). This region was assumed to contain ionised gases and the temperature, combined with slag activity, was high enough to destroy the recrystallised alumina insulators around the wires of the thermocouple. The optimum release position was selected after trials to be 10mm behind the rear of the electrode.

Due to the high temperatures encountered in the weld pool and the length of time that the thermocouple may have to spend there, platinum/platinum 13% rhodium thermocouples are best replaced by tungsten/tungsten 26% rhenium, which suffers less deterioration in the pool. Micro-plasma welding was used to form hot junctions in these materials.

The positioning accuracy of the system is shown by the weld cross sections in Fig.6, while Fig.7 shows thermal cycle traces from these welds.

Development of analytical technique

Early thermal cycle work used chart recorders to provide hard copy traces of the thermocouple outputs, as in Fig.7. These were then converted to temperature/time plots using a thermocouple calibration table. In many experimental set ups, especially those made using HAZ thermal cycle measurements, this involved 15 chart traces or more per weld, and analysis of these took a considerable time.

In view of the limitations of conventional analysis techniques, computer data acquisition and analysis systems were investigated. The possibility of maintaining a complete record of a number of simultaneously recorded thermal cycles in a computer file was also attractive for future requirements in the field of weld information data bases.

The present studies used an Intercole systems Compulog IV unit with dual floppy disks as a storage medium. As installed it was connected to a remote cabinet containing 20 inputs for direct attachment of thermocouples. It contained facilities for room temperature and signal drift correction and the system was capable of acquiring twenty channels of data at frequencies of up to 300Hz, storing the data on disk, and subsequently analysing them to give the required thermal cycle information. Cooling rate and other related data may then be printed out while thermal cycles may be permanently recorded on to disk for future use. A typical output is shown in Fig.8.

These tools now provide a facility for acquiring accurate thermal cycle information and for building up a data bank to enable statistically based correlations to be made with both welding procedure and metallurgical properties. Some of the initial work already carried out has been on the effects of parent plate dimensions and measurement position on thermal cycle, and this will be described in future articles.

Summary

1. Methods are described for examining thermal cycles in submerged arc welding, firstly in the heat affected zone by using pre-positioned platinum/platinum 13% rhodium thermocouples and, secondly, for examining weld pool thermal cycle with a harpooning technique.
2. Weld pool thermal cycle is best measured using tungsten/tungsten 26% rhenium thermocouples, and these can be positioned to within ±1mm in the weld pool by using fully automatic release units.
3. Thermocouple signal processing is achieved by using a computer data logger to log at frequencies up to 300 Hz, and to process and output data in numerical or graphical form.

References

1. Shinoda T and Doherty J: 'The relationships between arc welding parameters and weld bead geometry - A literature survey'. Welding Institute Report 74/1978/PE.
2. McGlone J C: 'The submerged arc butt welding of mild steel - A decade of procedure optimisation'. Welding Institute Report 133/1980.
3. Alberry P J, Coleman M C and Jones W K C: 'The effect of weld thermal cycles on the structure and hardness of ½ CrMoV heat affected zones'. CEGB Report R/M/N803, December 1974.
4. Glover A G et al: 'The influence of cooling rate and composition on weld metal microstructure in a C/Mn and a HSLA steel'. Weld J 1977 56 (9) 267s-273s.
5. Krivosheya V E: 'Mathematical methods of calculating the parameter of the welding condition for the submerged arc welding of butt joints with no edge bevel'. Automatic Welding 1978 31(2) 3-7.
6. Davis M L E and Bailey N: 'The influence of flux on element transfer during submerged arc welding'. Welding Institute Report 66/1978/M.
7. Pedder C: 'Harpoon thermocouples'. Welding Institute Research Bulletin 1973 14 (11) 333-334.
8. Dawes M G: 'Thermal cycles and straining effects in a multipass butt weld'. Brit Weld J 1968 15 (11) 563-570.
9. Jacobsen P J: 'The use of harpoon thermocouples to record thermal cycles in submerged arc welding of steels'. Welding Institute, unpublished work, July 1980.
10. Kohno R and Jones S B: 'An initial study of arc energy and thermal cycles in the submerged arc welding of steel'. Welding Institute Report 8l/1978/PE.