Geotech Investigation-how to approach soil models

[neforun]

requesting advise on how to prepare a soil model from a site investigation, this is my first geotech investigation. the site consisted of a saturated multi layered soil (not to get into to much detail it was 5layers of soil on bedrock). SPT's were taken for all but one till layer. I have some moisture content, Pl,Ll,Pi,Cu, bulk density and results from a triaxial for a sand layer above the till.
Im looking for some direction on where to start basically how to build a mohr-coulomb and hardsoil model with what i have. should try to back calc other parameters, find similar soil model or correlate as much as i can so i can input the models into plaxis 8.
thank you
n

 

[BigH]

Are you attacking this problem like a university assignment or a practical practicing engineering point of view? What kind of structure is going on the site? Do you know what kind of foundations might be the most appropriate? For your "first" geotechnical assignment, I doubt your organization would be giving you one that would require fancy "modeling" and Plaxis 8 analysis. One thing in practice - KISS principal when you can (assuming normal type project) - how did you get a triaxial on sand? Reconstitute it or undisturbed?

 

[dgillette]

BigH said it a little kinder than I was going to. You have asked a question that is both broad and vague, without providing sufficient context.

Building a Plaxis model is a bit odd to assign as someone's first project. Where is your leadership coming from? From a geotechnical engineer who knows how to run Plaxis, and will be able to tell you if the input and results make sense? Why can you not get guidance from within your company or agency? With complicated models, there are so many things that can get messed up that one cannot simply read the manual, run it, and accept the results. If you cannot get guidance and checking within your own organization, you should consider simpler methods, or find someone who can teach you. Programs like Plaxis and FLAC can be useful, but if you put garbage in, you will get garbage out. You can spend a lot of labor hours on them without ever producing a credible result. You may therefore wish to step back and look at whether that is the best method.

There have been many, many publications that give correlations among soil properties. You can probably download the US Navy's NAVFAC DM-7.1 and -7.2 for no cost, as well as Corps of Engineers EM-series manuals. DM-7 contains a number of correlations for predicting commonly used material properties like Mohr-Coulomb parameters. For any constitutive model more complicated than elasto-plastic Mohr-Coulomb, there are many parameters that can only come from lab testing, and few engineers have a "feel" for them as they do for phi', for example. BigH has given you good advice about the KISS principle - start with the simple methods and add Plaxis if you need it.

 

[neforun]

thanks for the info first of all. the project is for university so im working on my own except for the odd meet up with supervisor, so im not able to simply turn to the next desk and ask advise. the site consisted of several test piles. im modelling the piles firstly with plaxis and then hopefully looking into the pile/soil interaction once i get to grips with the latter.
i have a large amount of info from the GI but its patchy spt's where taken for all but 1 layer and lab test where carried out on only one layer so to me the info is patch.im trying to build a soil model, mohr-coulomb and hardsoil, but finding it difficult to determine the parameters that i require from the GI.
im fully aware that i have to be careful with what info i input into a software package thats why i want to get this correct before i start inputing. also what is KISS principle?
thanks again the advice

 

[BigH]

KISS = Keep It Simple, Stupid (well, that's what they always told me!)

Is this research for thesis? or . . . . Doubt that most projects would use such software for pile capacity - or even to determine pile settlements - they'd stick with the traditional. You have the strata - you have SPT values (for sands you can determine phi approximately from the N value) - for the clays- with CU data (undrained shear strength) you would have corrections as per Tomlinson and others. The till - you didn't say if clayey till or not - but I would use presume the till, if clayey, would be very stiff to hard - with Su values in the order of 200 kPa or so; if "sandy" you would have phi values in the order of 40deg. For modeling, you need to be able to assign (estimate by correlation; experience in the area, etc) to get the strength parameters - and even more correlations for the elastic properties. You seem to have a "run of the mill style" GI so it is all, to a large degree, a crap shoot.

 

[neforun]

thanks for your help.