Design of Experiments (JMP use research)
- Ishan Saini
A very important part of material testing is identifying different factors that affect the end result of that test. Design of Experiments (DOE) is a systematic method that uses statistical analysis to determine the relationship between factors affecting a process and the output of that process or in other words, is used to find cause and effect relationships with the use of mathematical predictions. Unlike excel where a relationship of only two factors can be graphed and understood at once, DOE tends to identify and graph multiple variables (controllable variables like thickness, material strength and uncontrollable variables like human error) to show the contributions of each factors within an experiment or a test.
A software that could be used to identify these relationships in our testing procedures is called JMP, which is used widely in the manufacturing quality and process departments of many well known companies. The software offers a wide variety of statistical analysis, but it should be noted that it costs $150/month, which is a considerable amount. The software also offers a free trial version for 30 days without some of the paid features, but for the purposes of Midnight Sun’s testing the free trial should be enough to show the usefulness of this software.
The article attached below should give an idea of a high level use of JMP to find results from a tensile test. To provide a brief overview of this article, an aluminum foam sandwich was introduced to a tensile test using a universal testing machine to notice effects of the skin to core ratio on the stress-strain behavior of the material using the desirability function in JMP version 11. Other factors like weight, part thickness and core thickness and effects of increasing/decreasing them were analyzed to find the contribution of these factors during the test. The article can be viewed below.
The graphs shown above are taken directly from JMP and the tables are examples of sample data that could be used to run a study in JMP.
Now, to give an idea on how these learnings could be applied to Midnight sun, let’s take an example of single lap shear test. For this test, experimental parameters like the adhesive (epoxy or methacrylate), surface pre-treatment (acetone solvent wipe or silane coupling agent), bond line spacer shim thickness and adhered thickness are a few examples of variables that can be changed to find a test with the most optimal results. By collecting the data for these variables and running a multivariable study in JMP, the software can recognize the contribution of these factors in an experiment and show the factors that need to be used to improve the results for single lap shear test.
From the specific study linked below, an example of the relationships found based on a study looks like this: “It was found that that effective strength increases with decreasing bondline thickness, increasing adherend thickness, with an overflow fillet, and with a robust surface pretreatment were confirmed by taking standard average values. The average values for effective -4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4 PC2 PC1 SG300, no fillet SG300, fillet 28 strength also imply that the epoxy adhesive is most sensitive to the surface pretreatment, while the methacrylate is influenced the greatest by adherend thickness.” These results may vary for different studies based on the data presented for each study, which further proves the importance of a design of experiment in materials testing.
An experiment that ran DoE studies for a single lap joint can be viewed below.
https://www.sciencedirect.com/science/article/abs/pii/S0143749618302574