Weld size and ferrite measurement

TWI Bulletin, February 1985

Mike Gittos, MSc, CEng, MIM, MWeldI is a Senior Research Metallurgist in the Materials Department.

The accuracy of measurement of δ-ferrite levels in weld metal can be affected by the size of the sample on which measurements are made. Simple experiments are described which determine the limits of accuracy for two standard magnetic measuring instruments.

Many aspects relating to the occurrence of ferrite in austenitic stainless steel weld metal have been covered in previous issues of the Research Bulletin. These include its significance with respect to service behaviour ^[1] and weld cracking ^[2] as well as consideration of ferrite measurements. ^[3-5] The present article is concerned with the effect of weld size on measured ferrite number for two commonly used instruments.

The amount of δ-ferrite present in 'austenitic' stainless steel weld metal has a profound influence on both cracking during welding and the service performance of the weld. In consequence, measurement of weld metal δ-ferrite has received a great deal of attention and both the International Institute of Welding and the American Welding Society have established standard measuring procedures. ^[6,7] The most widely used methods utilise instruments in which the force required to detach a small permanent magnet from the sample surface is calibrated in terms of 'ferrite number' (for practical purposes equivalent to vol % δ-ferrite).

For large multi-run welds in thick material such magnetic instruments yield an average ferrite number (FN) for the deposited metal. However, in small welds, such as those joining sheet material or constituting thin layers of cladding, the measurements may be influenced by adjacent dissimilar material. If ferrite measurements are made on such welds, it is important to know the minimum volume of weld metal or the thickness of a weld deposited cladding which is required to give a valid result when using a particular instrument. A series of simple measurements has therefore been made to investigate the effects of specimen diameter and thickness using two commercial instruments commonly used for determining weld ferrite levels at The Welding Institute - the 'Elcometer Inspector Gauge' and the 'Magnegage.'

Experimental

Specimen diameter

In practice, measurement of the ferrite content of small weld beads may be influenced by the surrounding material, which could be either of higher or lower ferrite content or may be empty space in the case of thin material or when working close to specimen edges. The extreme cases are represented by a weld bead surrounded by material of 0% ferrite ( i.e. non-ferritic material or empty space) and 100% ferrite ( e.g. mild or low alloy steel). These two possibilities were investigated by making measurements on cylinders of ferrite containing weld metal mounted in bakelite for the non-ferritic case and in mild steel blocks to simulate the fully ferritic case.

Material was available in the form of single pass submerged arc bead-in-groove deposits in 25mm thick 31:6 austenitic stainless steel plate penetrating approximately 15mm into the plate and having a ferrite number of 10 as measured on a transverse section using a Magnegage. Cylindrical samples were machined parallel to the axis of this weld with diameters of 1, 2, 4 and 8mm. For each diameter two 10mm lengths were produced and inserted into bakelite mounts and mild steel blocks as shown in Fig.1. Both the bakelite mounted samples and the steel block assemblies were ground and polished before making Magnegage measurements.

Fig.1. Test blocks for measurement of ferrite in small samples using the Magnegage ( d, diameter of test cylinder = 1, 2, 4 or 8mm):
a) Simulating weld surrounded by nonmagnetic material
b) Simulating weld surrounded by magnetic material

The Elcometer was not used for measurements on these samples because it was not possible to aim the magnet with sufficient accuracy. With this instrument the contact between the sample and the magnet is obscured by a rubber cup of 15mm outside diameter.

The ferrite number of each cylinder was determined in the normal manner, aiming the magnet as accurately as possible for the centre of each weld metal cylinder. To facilitate this, a small plastic tube was used to minimise magnet movement, as the Magnegage magnet is normally able to swing freely to some extent. The instrument was calibrated (using National Bureau of Standards coating thickness standards) with this guide tube in position. These arrangements enabled measurements to be made on all the test cylinders except the i mm diameter cylinder mounted in mild steel, the attraction of the steel mount making it impossible to locate the magnet on the sample.

Specimen thickness

The effect of weld deposit thickness is particularly important in the case of stainless steel claddings on ferritic materials. This situation was simulated by stacking 15 x 60mm coupons cut from strip welding electrodes 0.5mm in thickness on a 50mm thick steel block and making Magnegage and Elcometer measurements. The Magnegage was again calibrated using NBS standards whilst the Elcometer was calibrated using the plastic films supplied with the instrument by the manufacturer. The strip electrodes were Ni-Cr alloy and 347 and 309Nb austenitic stainless steels with Magnegage ferrite numbers (as measured in a 15mm thick stack) of 0, 7 and 18 respectively.

Instruments of this type calibrated as above are intended for measuring weld metal ferrite but the microstructural distribution of ferrite is different in wrought material, and there is a tendency to underestimate the ferrite content.^[7] However, in the present context the absolute numbers are less important than the way measured values change with specimen size and thus wrought material can be used even though true ferrite numbers are not determined.

Results and discussion

The results of the ferrite measurements are plotted in Fig.2 and 3. The marked effect of specimen size is shown by the measurements on cylinders ( Fig.2). It is apparent that even at a cylinder diameter of 8mm, the measurements have been influenced by the surrounding material. For the case of specimen thickness ( Fig.3) both the Magnegage and Elcometer gave similar results indicating that a thickness of about 5mm or above was required for valid measurements although the limit depended on ferrite level, lower ferrite levels being more sensitive to thickness. The results from the thickness and diameter experiments are in fact self-consistent if it is assumed that the magnetic fields of the measuring magnets are approximately hemispherical at their tips because it is the volume of ferromagnetic material adjacent to the end of the magnets which determines the pull off force. Thus specimens 10mm in diameter and 5mm thick contain similar volumes of metal contributing to the measurements as indicated in Fig.4.

Fig.2. Effect of specimen size on ferrite number measured by the Magnegage for submerged arc weld metal with a nominal ferrite number of 10

Fig.3. Magnegage and Elcometer measurements on stacks of strip welding electrode coupons on mild steel base plate

Fig.4. Schematic section showing volume of specimen contributing to measurement as indicated by the experimental results

These results may seem rather restrictive on size, particularly with respect to specimen diameter, but in practice, considerations of geometry and accuracy in many cases reduce the diameter required for useful measurements. The experimental data were generated for the extreme cases of dissimilar surrounding materials. An example of a more typical measurement might be on the surface of a weld in an austenitic steel plate as shown in Fig.5a. Two geometrical factors may reduce the effect of sample size. Firstly the austenitic parent material may contain some ferrite. This is usually at a lower level than in the weld but its presence reduces the diluting effect which would be given by fully austenitic material. A similar effect acts for thickness in the case of clad layers of different ferrite contents ( Fig.5b).

Fig.5. Geometrical influences tending to reduce the effect of specimen size measured in this work:

a) When measuring a 6mm wide weld the result is equivalent to measuring a cylinder >6mm in diameter because extra material (cross hatched, in the welding direction) contributes to the measurement
b) When measuring a layer of cladding 3mm thick the result will be distorted to a much smaller extent if there is another layer of different ferrite content below it than if a single layer directly deposited on to ferritic steel is measured

Secondly, measurements on a weld bead 6mm wide are equivalent to those on a cylinder >6mm in diameter because the specimen has an effectively infinite dimension (for welds longer than 10mm) in the welding direction ( Fig.5a).

As regards accuracy, it must be observed that ferrite levels, whether determined from compolsition ( i.e. using the Schaeffler or Delong diagrams) or instrumentally, are not considered precise quantities and are also usually small numbers. Thus for a weld of 5FN, AWS A4.2-74 quotes an expected range of 5 ± 0.6FN for 95% of 'type A' gauges (such as the Magnegage) and such an accuracy (within 12% of the true figure) could be obtained from a sample 4mm in diameter surrounded by completely dissimilar material.

The greatest effect of sample size is experienced when making measurements on transverse sections of small welds but, even here, beads of 5mm diameter give measurements of sufficient accuracy for most practical purposes provided that the magnet is placed carefully in the centre of the specimen.

These results are specific to the individual instruments used for the tests. However, both instruments have previously been compared with others of the same type, and were found to give a similar response to samples of different ferrite levels. ^[5] Hence, the data should be of reasonably general application. Moreover, in the case of the Magnegate used in this work, the pull-off force of the magnet has been measured and found to comply with IIW magnet strength requirements, so the present data should be typical for this type of instrument.

Summary

Experiments with The Welding Institute's Magnegage and Elcometer Inspector Gauge instruments indicate that a hemisphere of material of approximately 5mm radius contributes to measurement of ferrite content. However, adequate measurements of low ferrite levels in austenitic welds or claddings are possible for most applications in samples of 5mm diameter with the Magnegage and/or 3mm thickness with the Magnegage and Elcometer.

Acknowledgements

References