Amit:

All the previous answers provide valid and useful information in response to your question. However, if everything is to standard, i.e, the sizes of specimens are large enough, there are no measurement errors, etc., etc., there is still one property which will depend upon the specimen size.  For a metallic material, for example, the yield strength, UTS, % reduction in area, and work hardening coefficient all you should give you consistent values, essentially independent of specimen size, if the uniaxial tensile test is done correctly.  However, the % elongation (or strain to failure) will be a function of the gauge length.  The uniform strain up to maximum load should be the same, but the post maximum load strain, or non-uniform strain where necking takes place, is a local phenomenon which no longer scales with the gauge length. The result is that the total elongation will be different depending upon the length of the specimen.

ROR

Hi Amit, It would be more appropriate if you would let know the scale of the specimens. Theoritically, the engineering stress should be same as it is F/A and with the change of A the force also change for the same material. However, when you change the scale of the specimen largely, you introduce many other things, such as grain size, anisotropy, also machine error as mentioned before by Guilherme Soares etc. So, specifially if we know the scale of the testing specimen, material type, we should be more confident in understanding the effect of dimension change on the stress-strain plot.

Thanks

Agree with Soud. The scale of change in dimension is important. If  you change the width of a 1mn thick sheet metal specimen from 12.5mm to 14mm, the results may not change significantly. But if you change it to 20mm then you may see a significant difference because the stress state will change towards plane strain. That is why standard tests are devised. The best would be to perform standard test but if it is necessary to deviate then do not deviate too much.

HI Amit, actually all the previous answers already gave a good explanation. I'd actually like to know what do you mean by "primary dimension of the specimen" and which rate of change you mean to do on that dimension, because this additional information may change the proper answer to your question.

However, I'd line anyway to add some additional considerations:

1) When you perform simple tension test on, let's say, 5 sample with the same test condition (same theoretical geometry of the specimens, same material, strain rate, temperature, etc.) you anyway measure the dimension of the specimens in their gauge length before the test, in order to have a more precise estimation of the cross-sectional area, thus of the stress/strain curve. This measurements are required to take into account the small variation of the specimens cross-section area which may occur during the milling operation of the specimen (this is required whatever process you use to make the specimens). Even small changes in the specimen dimensions (due to the production process), compared with the standard dimension you thought to have, may change the resulting stress/strain curve. This is a partial answer which may make you think of how important the dimensions of the specimens are, even though they theoretically comply with the standard regulations.

2) I personally experienced a problem where I could not use the standard specimens dimensions because the metal sheet which i wanted to test was not wide enough to make 45° and 90° specimens (required to test for anisotropy) with the prescribed dimensions. What i did was to get the 0° samples both with standard dimensions and "smaller dimensions" I designed (which could fit the sheet also in 45° and 90° directions). I tested all of them, trying to understand whether there were variations in the stress/strain curve due to the variation of the specimens dimensions. The conclusion (which I also verified by FEM analysis) was that, if you do not violate the plane stress condition (this is the key-point), the resulting stress/strain curve won't be affected by this variation of the specimen dimensions.

As for the anisotropy, I could not test whether this change of specimen dimensions actually influenced it or not. However, since the anisotropy is actually a "comparison" between stress/strain curve under different direction compared with the one you normally call "0° direction" (which, in my case, was the rolling direction of the sheet), if your tests are all made with the same sample dimensions at, lets say, 0°, 45° and 90° directions, the changes in the stress/strain curve you may see, should be due to anisotropy, regardless the dimensions of your samples.

The material properties characterisation topic is pretty complicated and sometimes even tricky, so if you can provide more information I'll try to give you a more detailed answer. Regards. 

To add to the previous answers:

The simple methods which are typically considered for conversion of the data are far too simple to account for the scale effect and therefore, based on them you shouldn't expect a difference before the necking.

On the scale effect, you will note that the ratio of grain size (and other structural features) to the gauge diameter and length are not the same for specimens of different dimensions and therefore the results should be different in a range sensitive to the feature sizes.

Experimentally, it could make a difference due to the following parameters:

Stiffness of your test rig; specially when measuring the Young Modulus

Potential directional properties of the sample material

just to name a few.

Amit:

All the previous answers provide valid and useful information in response to your question. However, if everything is to standard, i.e, the sizes of specimens are large enough, there are no measurement errors, etc., etc., there is still one property which will depend upon the specimen size.  For a metallic material, for example, the yield strength, UTS, % reduction in area, and work hardening coefficient all you should give you consistent values, essentially independent of specimen size, if the uniaxial tensile test is done correctly.  However, the % elongation (or strain to failure) will be a function of the gauge length.  The uniform strain up to maximum load should be the same, but the post maximum load strain, or non-uniform strain where necking takes place, is a local phenomenon which no longer scales with the gauge length. The result is that the total elongation will be different depending upon the length of the specimen.

ROR