Schmertmann vs. Plaxis 1

[bdbd]

I am sure that some of you made this comparison.

When I calculate with Schmertmann, due to theory, calculations ends 2B below the foundation. When I model the same foundation in Plaxis, it is clear that there is settlement below 2B.

Do you really think there will be no settlement below 2B? What is your experience about this comparison? Do you know any resource, paper?

Regards,

b.

 

[BigH]

The 2B is used (for square footings) as that point where the pressure bulb is at 10% of the applied pressure. That has been normal geotechnical practice since "way back" - as a point where below settlement induced is very small compared to the assumptions made in modeling so is neglected. Plaxis, although I don't use it, obviously doesn't follow this "rule" - and continues below the 2B. For strip footings it is generally taken as 4B.

 

[bdbd]

Thank you for the answer.

I know that Plaxis does not follow this rule. Actually Plaxis and other finite element softwares do not follow any rules that are based on empirical or numerical approach. They are following the finite element concept. And I don't know your opinion, but I am a fan of finite element. So, I am trying to understand the reason of difference.

 

[Ron]

bdberkdemir....finite element analysis can be a very powerful tool; however, you must understand that when applied to such a variable material as soil, it has some limitations. The finite element approach can be difficult to apply to soils because of the need to discretely identify boundary conditions that usually don't exist in soils. Much of geotechnical engineering is observational, historical, empirical and tested over many years. The premises put forth by Terzaghi, Cassagrande, Peck, and others have been proved over and over by subsequent geotechnical leaders such as Sowers, Bowles, Lambe, Seed and others, as well as accepted geotechnical engineering practice for many years.

Don't let the "accuracy" of the finite element approach mislead you to think that it's "accuracy" is also correct.

I go back to an old saying that certainly applies to geotechnical engineering and much of engineering in general.....

We are guilty of measuring with a micrometer, marking with a crayon, and cutting with an axe.

 

[bdbd]

Thank you for your answer too Ron.

So, to come to the reality, thus would you guys limit the depth of the model to 2B -for square footing- so we do not see any more settlement below that? Of course, the settlement over 2B would be relatively large compared to Schmertmann but it may be a useful assumption.

 

[Ron]

One other way to approach this from a settlement analysis is to treat it as a step function.....take the resulting increase in stress at the 2B depth and apply that over the larger area (from 2B) and compute the settlement below that level.

 

[BigH]

@bdberkdemir: Congratulations on your using Plaxis. I've never had the opportunity of using it - I sort of came up in the generation after the "founders" but before the computers have taken over and my projects seem to have no money for such programmes. I do not know your age - your background and that . . . but I love history and the history of geotechnical engineering is highly enjoyable. I have wonderful texts from Tschebotarioff (1951), Krynine and Judd (1957) among many others - I've read many of the papers by Bjerrum and Skempton, Peck and others . . . seeing how geotechnical engineering has developed and, as Poulos stated in his Turkish SOTA paper - the old methods still work.

Plaxis and other FE programmes can be a wonderful tool - but it is a tool and is not "exact". I've attended conferences in Singapore where papers have put forth very interesting Plaxis analyses - 3D and all that - all based on E being derived from an SPT N value (uncorrected or not - no one specifies). Therein lies the rub, if I may. Determining the "real" values of E or other parameters for the FEM are very difficult. In the analyses, changing E by 10% can change your result by 10%. If you can't get the "real" value of E, then your result is a wonderful indication of the behaviour - but not of the reality. I've pointed out in other threads that settlement computations cannot be taken as "exact" - try four or five different computational methods for settlements in sand and see what range of answers you obtain. This is why geotechnical computations are fraught with vagueness/fuzziness - our COV values are in the order of 25 to 50% whereas steel and concrete at in the range of 5%. And getting back to Singapore, I asked if anyone, based on the "real" measurements of excavation support movement tried to back-calculate the E values and then to their N values so they can get better fits . . . the response was "No. Don't have the time."

Enjoy having you on the site - regards and