There are three methods to test for transition temperature.  They each have their own strengths and weaknesses, so let us help you decide which best suits your project.  In practice, most projects use all three tests at different stages of the project.

This test is governed by ASTM-F2082.  For standard tests, a sample is bent around a mandrel with a radius nominally 20x the thickness of the part.  In practice, the bend radius can be changed to better replicate your project.  The sample is then scanned over a range of temperature that exceeds the range of interest.  When motion begins, the austenite start temperature is recorded and when motion ends, the austenite finish temperature is recorded.  This test can be conducted in our lab with great accuracy, or you can replicate the test with a pan of water and a hot plate to keep your costs low.

This test is governed by ASTM-F2004.  For this test, a section of your part is removed and crimped into a DSC pan.  The sample should be weighed accurate to 0.01mg.  For standard tests, the sample should be heated at a rate of 10°C per minute.  In practice, it is important to replicate the heating rate of your application as nitinol performs differently at different rates of heating.  Data should be collected from at least 20°C below the martensite finish temperature to at least 20°C above the austenite finish temperature to ensure that the instrument does not contaminate the data.  In practice, the data from the first heating/cooling cycle should be discarded as it is simply removing the thermal history of the sample.  The data from the second cycle is valid but a third cycle should be run to validate the data to identify errant data.  This test is extremely accurate (our DSC is accurate to 0.05°C and 0.04µW) and can reveal all of the critical temperatures as well as the latent heat of transformation.  This test has a moderate cost per sample, but it has two major weaknesses: first that it is a destructive test and second that it is a stress-free test so it may not accurately replicate your system response.

This test is not governed by any ASTM standard because each test must be individually designed.  For this test, a fixture is designed which replicates the forces on your application.  The sample is then loaded under constant force and then heated across the temperature range of interest.  Force data is collected through the use of a load cell attached to the fixture used to deform the sample.  When heating, the active austenite start temperature is recorded when the displacement begins to increase and the active austenite finish temperature is recorded when displacement ends.  When cooling, the active martensite start temperature is recorded when displacement begins and the active martensite finish temperature is recorded when the displacement ends.  Generally speaking, this is the most costly of the tests for transition temperature and it can only be used to identify the start and finish temperatures for martensite and austenite.  However, this sheds significant light on the behavior of your system that may be very difficult to obtain otherwise.  This test can be conducted in either air or water (antifreeze) based on your application.