skip to Main Content

The quality of machining samples for mechanical testing in laboratory conditions is a critical consideration when choosing a testing facility.



In order to comply with industry norms and standards when administering mechanical testing procedures in laboratory conditions, it is imperative that full compliance of machined samples for relevant codes and standards is implemented. This means that all dimensional and surface tolerances as prescribed by applicable codes/standards, must be done in accordance to the full range and scope of technical specifications.


Sharing our experience and lessons learnt – through the years in various industry sectors

In metallurgical testing it is our experience that little attention is given to the quality of the machining of mechanical test samples. In fact, the machining of the samples is as critical to the result as the testing itself. Machining of the samples to the appropriate standard is also not a quick or a cheap action. Indeed, if it is claimed that this is the case, then there should be very serious concern as to whether the samples are machined to the correct standards.

The ASTM standard for tensile testing outlines the requirements for the geometry of tensile test samples. For example, the reduced section needs to be parallel with a tolerance of +/- 0.1mm. The center point of the radius from the grip section to the reduced section needs to be parallel also within a tolerance of +/- 0.1mm. The tolerance for the gauge length is +/- 0.25mm. The schematic shows the associated tolerances in mm for each of the geometric elements for the tensile test specimen


Likewise, the schematic below shows the tolerances associated with the geometric elements of the impact specimens. The dimensional tolerances on the section of 10mm by 10mm as well as the length of 55mm is +/-0.01mm. The various other tolerances are shown on the schematic.

Many people are quite careless when it comes to bend test samples which are used to qualify welders and welding procedures. The standard specifies that the bend test sample should be parallel on both sides with a tolerance of +/- 0.5mm. In other words, a simple saw cut from a coupon is just not good enough. The purpose of the bend test is to detect any defects in the welded joint. If the surface finish is not 2 microns on the welded joint, defects can go undetected. This could mean that a poor welding procedure could pass and not fail, as the defects are not detected. If a welder is qualified by way of a bend test the welder could pass by way of defects not being detected. Without attention to this detail, there cannot be 100% confidence in a passed bend test.

What happens if this is not the case?

In the first instance, the test is NOT compliant with the standard. If the geometry is not as per the description in the relevant ASTM standard or ISO standard or BS standard, then the test result does not comply with the standard. It is that simple. The reason the standards give such requirements is that the measurement of stress and strain which is the purpose of the tensile test will not be valid. So, if the samples are not machined according to the applicable standard, the result will not be deemed valid. This means the result for the measured yield stress, ultimate tensile stress and elongation to failure will not be valid. We have had firsthand experience of measuring inaccurate results because the sample did not comply to the geometric requirements. It is for this reason that we do not allow our clients to machine their own samples, as we are responsible for the accuracy of the test result and cannot guarantee this unless we are in control of the machining. Further, we only use approved and qualified workshops that have demonstrated to us that they can comply with the necessary geometric and surface finish specifications.

In the case of the impact test results, non-compliance with the tolerance requirements and surface finish requirements will cause the measured impact energy will reflect a higher than acceptable degree of uncertainty. This means impact tests could fail when they could have actually passed.

In order to properly notch an impact test in a welded joint in the right location, the sample must be polished on the right side, without affecting the dimensional tolerances of the sample. This face should be etched properly to locate the fusion line in the weld to position the notch correctly. Without this attention to detail the test result is invalid and can lead to a failed WPQR when it could have passed.

ISO 17025 deals in section 7.7 with the validity of the results. At One Eighty we have dealt with the dimensional tolerances of the test specimens to comply with this ISO 17025 requirement, because if the sample does not comply with the standard that the test method is based on, then the test result is in fact invalid. A set of invalid test results for a WPQR for example means that the weld procedure has in fact not been qualified at all. To use such a WPQR in the case of a pressure vessel for example could have very severe consequences particularly if a failure occurs and results in loss of life.

In my opinion there is only one way to do things – and that is the correct way. It is impossible for me to understand why any departure from the right way should be acceptable.



Dr Janet Cotton

Founder and Director of One Eighty Materials Testing


PhD(Materials Engineering UCT)


Back To Top