Material curves based on Material Certifivates

[blckwtr]

Can you from a standard material certificate make a rough approximization of a stress-strain curve? The parameters used are: Yield stress, tensile stress, reduction of area at break and elongation at break...

Regards

Tommy

 

[arunmrao]

It will be too ambitious in my opinion to guess stress strain curve with just 3 data points,load at yield,breaking load and final elongation of the sample.


I have not failed. I've just found 10,000 ways that won't work." — Thomas Edison
_____________________________________

 

[blckwtr]

Most of our analysis are based on rough approximisations in the first steps in our analysis, and occasionally we run into problems that needs to be solved in the plastic area. For this problem, there are very small local highly stressed areas, and naturally these are very high since only the linear material model is loaded into the program and stresses are calculated from strain. We can choose to use another failure criteria, but we need a plastic material model (linear), and was wondering how to make the best assumption.

Note that we are not material scientists, but of course we must check such problems more thoroughly in the final design phase. When working with both materials and limited space for our applications we must tune our design throughout the whole project.

 

[CoryPad]

You can do this, the results may be off by 25 % or so.

Coordinate system is strain (abscissa) vs. stress (ordinate).

1) Straight line from 0,0 to YS/E,YS

where

YS is yield strength
E is Young's modulus obtained from a handbook or equivalent

2) Parabola from YS/E,YS to EL,YS

where

EL is elongation at break

The stress at break is just for convenience and symmetry. The maximum of the parabola can be placed at MEAN(YS/E,EL),TS

where

TS is the tensile strength

 

[blckwtr]

OK, but is this engineering stress? I thought you had to use true stress in a FEM program?! Well, if you can define a curve for engineering stress, maybe it can be done for true stress also?!

 

[arunmrao]

Corypad, I had an opportunity to review a PhD thesis,where a curve was drawn using just 2 data points!!

Precisely,in the non linear region the flow curve need not be as what we conventionally think. It could depend on many other factors too.

I have not failed. I've just found 10,000 ways that won't work." — Thomas Edison
_____________________________________

 

[CoryPad]

You can convert engineering stress-strain to true stress-strain with basic equations. This also can introduce another error. The following links provide the equations.

http://en.wikipedia.org/wiki/Stress_(mechanics)

http://en.wikiversity.org/wiki/Nonlinear_finite_elements/Stress_and_strain_in_one_and_two_dimension

 

[cloa]

Wouldn't the actual alloy type allow a better approximation? Doesn't highly ductile austenitic stainless steel such as as annealed 304 have a much wavering stress-strain curve?

 

[MSUKeith]

Some other tensile test parameters that can help are the R-Value and particularly the n-value which can help describe the curve through the plastic region. Sometimes these are available on mill certs.

Another factor to consider is material thickness. Predicting highly localized stress concentrations and resulting strains in thick cross sections using the simple variables from the tensile curve can introduce much error, particularly when FEA analysis is a shell, not a solid. Mesh size is also a factor.