When we use Type B estimation, there are different distribution such as triangle, rectangular...etc. Could anyone pls advice how to choose which distribution to use?? Thx.

For Type B, most of the time you can ignore triangle. That is not the textbook answer but is reality.

Usually, you only need to concentrate on normal or rectangular. If you have uncertainties from an outside cal, divide them by two and plug the number in. For resolution error, the conservative route is to take one half of the least resolution of the instument.

Hope this helps.

Hershal

I just use the rectangular distribution (for all analysis) because the rectangular is the most conservative.

Here's a great link to show the different types of MU distributions and when to apply each. Very helpful.

http://www.hn-metrology.com/distrib.htm

Daryl Montie