It depends upon the methods used to measure the properties.
For ultimate tensile strength measured on a standard specimen using a universal testing machine, I would expect the mean of multiple samples from each lab to be within 5% of each other. One percent would be nice, but not likely. For example, if the tensile strength is 500 MPa, I don't see two labs reporting values within 5 MPa.
For chemical composition of a steel alloy measured by an emission spectrometer, I would expect the composition's to be within 1% of each other (e.g. steel with 0.500 mass percent Mo should be 0.500% at lab 1 and 0.505% at lab 2).
Regards,
Cory
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.
In my dealings with reviewing numerous material Code Case submittals for ASME Boiler and Pressure Vessel Code applications, the spread in mechanical property data can vary by as much as 10%. ASME B&PV Code does provide some limited guidance regarding the spread in mechanical property data for production heats of material. This spread is typically stated as a maximum of 10%.
Keep in mind that the variation within a heat is likely to be more than 5%, so it is not correct to assume that the samples are the same, and only the test methods vary. This is why statistical methods should be employed when testing important features, so that you can have an expectation of the normal variation. I agree with CoryPad and Metengr regarding lab testing differences, although elements that segregate (e.g., C & Mn in steel) will not be as uniform as other alloying elements.
Lab to lab variations are expected and should be accepted within about 5%.
After analysing a sample,please resubmit the same sample to them for confirmation and the values will not be the same. Especially in C,S and P. The lab technician refuse to accept these variations and insists that his readings are precise.!!
In mechanical testing there will be a variation in YS and elongation or in impact testing always controversial to say the least.
Hence the need for statistical analysis to arrive at the right inferences.
