Reduce your testing errors TODAY!
Materials Testing: Understanding & Reducing Sources of Error in Your Results
The need to ensure the repeatability and reliability of mechanical test results between individual test systems, whether those systems are located in the same lab or different labs or a comparison is being made between multiple suppliers, has long been a concern throughout the materials testing industry. This is not surprising since manufacturing companies have a strong focus on quality, and product quality assurance depends on the ability of testing systems to provide accurate results. Further, as many industries expand R&D and manufacturing operations into different parts of the world, data comparisons have become increasingly more important and complex. As a result, the materials testing industry wants to know, “why aren’t our test results the same?”
Consider the following example. Medical tubing manufactured in two different locations and the mechanical testing reveal differences in peak load and strain at failure. Assuming the specimens tested in both locations are thought to be the same and that differences in the results are not because of part-to-part variation or production processes, how could this be? Careful investigation of each system configuration, as well as the software methods and data, usually allows one to identify configuration and / or operator differences which can explain discrepancies. Good news! Most of the time, with the right set of eyes and attention to detail, these discrepancies can be resolved (without the need to purchase new equipment)!
The fishbone diagram drawn in header shows the six major categories that define a complete test system: method, measurement, operator, material, machine and environment. Within each category are the details that define the system, and when comparing these details between systems, one can identify where and how to make the appropriate changes.
Method: The method defines both the software method created to control the test system as well as the procedural or operational method used by the operator. When software methods are created by different people and for different systems, they are often not exactly the same. It is important to examine definitions for test speed, data collection rate and calculations used to generate results. Although operational procedures are often documented, lack of detail in those procedures can lead to self-interpretation and therefore, differences in procedure between operators.
Measurement: Measurement most often refers to calibration of system transducers such as load cells and extensometers but also includes the verification of speed and displacement of the crosshead as well as calibration of any specimen measurement devices. International testing standards recommend yearly calibration and verification of these systems. Failure to do so can be the cause for uncertainty in measurements.
Operator: Differences between operators can be the most difficult to isolate because they can usually only be identified by one who observes the testing of all operators and is trained to identify procedural errors. Each operator often employs a unique technique in handling specimens, preparing specimens, inserting specimens into the grips and conducting tests that can have a significant affect on results.
Material: In order to get repeatable and reliable results between systems, one must ensure that the materials tested are the same. It is important to know where the materials have been, when they arrived in the test lab, how they were prepared and to what accuracy. Differences in material handling and specimen preparation can have a significant effect on test results.
Environment: Some materials are affected by environment conditions, such as temperature and humidity, and therefore, must be controlled. But physical location in which the system is installed may also be important. For example, some test systems are located on the manufacturing floor where the air quality and surrounding conditions would be significantly different than that of a clean room.
Machine: New test systems are configured with the ideal set of software and accessories for a particular testing application. Although some of these accessories can be used for other applications, some can not, and it is important to learn the “signs” of a poor test configuration. The signs can include slipping due to insufficient grip pressure or worn jaw faces and specimens breaking at the grip jaws. In any test system, it is critical that alignment of the specimen within the grips is achieved and that slack within the load string is eliminated. Finally, all test systems require regular maintenance and care to ensure repeatable results over the life of the equipment.
Are you seeing sources of error in your data that you can't resolve? Start with an audit of your lab and use the diagram above as a checklist. Observe the operators. Use a standard specimen (like a spring) to remove error that may be caused from part-to-part variation. Still not working? You may want to consider running a GR&R study. Or contact me! We have investigated challenges of these sorts many times with great success and satisfaction from our users.
Note: this article was the joint effort between myself and Meredith Platt (now VP/GM of Buehler).
This is a nice, concise article on the potential sources of errors or differences in testing results. You may want to include specimen preparation rather than assume specimens are "identical". It covers the ground well for tensile types of tests. It could be expanded to include other types like torsion and compression or shear. It could be expanded for other classes of testing such as Rockwell hardness and micro-hardness. This is a very useful article.