No correlation between hardness and ultimate tensile strength ! 2

[stanislasdz]

I have a metallurgical problem : when I take different parts from a steel bloc i don't found correlation between hardness (HB) and ultimate tensile strength (UTS)

How can I explain this ?

 

[stanweld]

Do not expect an absolute relationship. If you plot a least mean squared statistical analysis of the data, you can clearly see the relationship of increasing UTS with increasing hardness.

 

[EdStainless]

That looks like a good correlation to me.
Run the statistics, make sure to account for precision of each test, and you have good results.

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube

 

[arunmrao]

All the hardness data provided is in a narrow band of 35 -400BHN. It is easy to err ,when testing the samples. Also,since you are close to 400BHN,can you consider using carbide indentor,instead of steel one.

The data as represented fits neatly in a band,do not expect to get a straight line relationship.

 

[stanweld]

The formulae for calculating UTS from BHN are based on statistical analysis of empirical evidence produced over half a century ago.

 

[JoeyV87]

Aye, but it's a good rough guide, in all fairness.

Explaining it is simple, stanislasdz. Hardness testing is a compressive test, tensile testing is a tensile test (surprisingly!). Compressive tests are pretty much (although not totally) unaffected by inherent flaws in the material. Tensile testing is affected.

Take a titanium aluminide as an extreme example. Compressive strength is huge, but the brittleness of the sample due to cracks formed on solidification causes the UTS to be rubbish.

I should point out that this is the TiAl one company I worked for made, the stuff can be very good, apparently.

 

[Guest102023]

There are a number of factors involved that have nothing to do with the presumed hardness-UTS correlation:

1) You mentioned 'different parts of the block' - metallic materials are rarely perfectly homogeneous in both composition and properties.
2) Tensle tests inherently have a certain amount of scatter.
3) Hardness tests have measurement error, especially if indentation size is visually measured. This increases at smaller indentation size; i.e. higher hardness (where this data is).

Hardness-UTS correlations (and hardness conversions) should always be taken with a large grain of NaCl. The many points collected to generate these curves have some scatter, and the curves are 'best-fit'.

Also, you are looking at data points within fairly small parts of the respective HB and UTS ranges. The scatter will be visually exagerrated.

 

[stanislasdz]

Great Answer !
Thanks a lot Brimstoner !

 

[davefitz]

I guess we forget that most correlations found in applied engineering texts are based on widely scattered test data.

If this particular data set was plotted on log-log paper and extrapolated back to the origin, it would look like a smooth fit.

In the case of hardness vs UTS, the UTS test data is usually from a sample that is long enought to fit in a test rig, while the hardness test zone is only a surface phenomena . Considering how a material is cooled down following N+T , the outer surface zone cools much faster than the inner zones, one would expect a different microstructure at the surface vs internal zones. Review a typical TTT plot and correlate that to a temp vs time vs depth thru the cooled part ( nased on a conductive heat transfer model) and see for yourself.