Users Forum
ADONIS Category => ADONIS Users => Topic started by: simone_stano on January 17, 2017, 12:37:41 am
-
I don't know if other think the same but it would be very appreciated. It could be focused basically in the process of building a simple model with special attention to setting of boundary and initial conditions and on plot tab features.
I met some trouble to do it but could be my limited experience in geotechnical fem software.
Thank you
Simone
-
have you checked this?
http://geowizard.org/forum/index.php?topic=25.0
-
Ok that is useful. One more with MC costitutive model with stratified soil?
Simone
-
you can redo the Mohr-Coulomb and Hoek-Brown by yourself by simply replacing the material creation line in the script.
-
Well... basically the problem is to apply correctly a K0 procedure...I'm thinking about it.
Simone
-
Ok now I think I understood the right way...
-
I'm going to upload more scripts later on .. keep in touch
-
Thank you for the second script.
In the meanwhile I tried to build up a very simple model which consist of two submerged layers (a NC clay overlying a NC medium dense sand) with Mohr-Coulomb constitutive model. The goal was to verify how good is Adonis in the calculation of stresses within the soil (vertical effective stress, pore pressure and horizontal effective stress as well).
To clarify I put the script of the model in attachment. Anyway the model is very simple with water table at ground level and horizontal layers. I posted also a plot which reports the correct stresses in the ground in comparison with Adonis's prediction. As anyone can see (note in this particular case sigma'_v0=sigma1’ and sigma'_h0=sigma3’) the prediction for pore pressure and vertical effective stress are good with probably a border effect which affects a bit the data on the top and bottom of the model. For horizontal effective stress the prediction seems not so good as the previous one.
Doing this I noticed a couple of things:
The first thing is MC properties tab has just a single value of density. This can conduct, expecially in sandy soil, to an inaccurate evaluation of stresses of course depending on the presence and quote of water table. To overcome this, probably we just need to introduce one layer more with gamma_sat density (or gamma density), but I think this could represent a serious limitation in the next development of Adonis especially facing with more complex model when for example there will be the need of simulate a dewatering and Adonis will not able to switch from a gamma_sat density to a gamma density. This imply that Adonis must be able to be sensitive of the quote of the groundwater table and switch the density accordingly. I think this can be added to the list “things to do”.
The second thing I want outline is about the process of adding initial conditions to a material. Generally is invocated the so called k0 procedure if we are not facing with an unbalanced geometry of the model for which a gravity loading procedure is the right choice. If I’m not wrong a gravity loading procedure is used in the second script posted in the forum.
Regarding the K0 procedure, it seems Adonis can incorporate initial stresses (horizontal and vertical) for any specific material layer by introducing an initial value and a gradient (in x and y ), nevertheless it seems this method is not properly able to introduce initial stresses in a stacked model because equations in this case are not just in the form y=mx but also in the form y=mx+c which are not implemented in Adonis at the moment (please refer to initial_stresses_pdf attached in which equations of horizontal and vertical stresses are explicitated through M (gradient) and C (intercept)). At the moment I don’t see any workaround if not a fix about this point but please consider my limited experience with Adonis and consider also my possible misunderstanding of the "initial condition" building process.
-
Sorry this is the pdf with equations.
-
ok I'll take a look and get back to you
-
first of all thanks for the detail comparison. honestly I really am looking for this kind of verification.
second, I went through your calculation and I think there is inconsistency in your calculation. so I recalculated the effective stress based on your input data and here is the procedure (please take a look at attached file):
1- calculated total stress and pore pressure from density given in the script file
2 - back calculated Poisson's ratio (nu) from shear and bulk modulus you provided which turned out to be 0.2 for all.
3 - from "nu" I calculated k0 = nu/(1-nu) which is lateral stress coefficient if you are using numerical analysis (not providing by yourself)
4 - from there I calculated effective vertical and horizontal stress which is pretty close to what ADONIS provides.
please let me know if need additional information or comments.
-
simone,
thanks to your example I was going through water table code and found out a bug which need to be fixed in next release.
unfortunately I did not have time to verify all aspect of program so your examples make me go back take a look at program carefully. I hope you keep it going.
-Roozbeh
-
Ok now is clear! The gap comes from a different K0. I got it from Jaki equation, but I missed that is related to nu which I assumed equal to 0.2 to get Bulk and Shear modulus from Young Modulus (I usually start from E). This procedure generated the inconsistency you noted. Now the correct procedure in Adonis is clear. Maybe some routine from other software can produce error here and this is an important discover for me.
From another point of view the problem still remain in the sense that Jaki's formula is ok for NC soils and the procedure you show bring to an unique value of K0 which is not in line with the stratigraphy of the model clay and sand. We expected 2 distinct value of k0...
-
maybe I missing something...I need more time to think about it
-
mohr-coulomb constitutive model has no idea about k0 value (Jaki or others) usually it's consistent with "nu/(1-nu)" providing model is still elastic (not failing) but if you really want to get correct lateral pressure you need to initialize the lateral stress and specify the correct stress to get correct K0 at the end. or use some sophisticated constitutive models that can calculate correct K0 value ..
-
i believe you could repeat this test with more coherent values this way:
- at first calculate k0 with jaky equation from fi;
- then calculate nu from relation nu = k0/(1+k0) that is the same used by adonis, but inverted;
- define coherent shear and bulk modulus with the above values;
- assign all the calculated values to the soil material used in the simulation.
-
Hi reversi,
I thought at that possibility but can we consider it theoretically correct?
-
Ok following the hints of reversi this is the new model (script + comparison).
It's evident the prediction of horizontal effective stress is now fairly accurate with a significant general improvement in the range expected.
So this could be the right way...
-
Hi all,
this is a slightly modified model from previous one. Geometry can be inspected in the script attached. The goal was to show the necessity of add an extra layer in case of water table hanging through a layer which will assume a saturated density on the submerged part. This need comes because at the moment we have the possibility to insert an unique value of density. As I told above is auspicable Adonis in the future can switch to two distinct density according to the position of water table.
Anyway at this moment, in case of model without water level changes, this limit can be easily overcame as shown in the "comparison" pdf attached as well.
-
Simone
would you please move the latest topic to "Useful scripts" board? thank you.
-
To be honest I'm not sure to be able to do that... :-\
-
what is the problem?
-
this is a slightly modified model from previous one.
could you publish the values of nu, E and the relations between moduli you used?
i've tried with usual elastic relations but my values don't match with yours.
-
I used the table given in the following page
https://en.wikipedia.org/wiki/Shear_modulus
go to the bottom of the page and find the correlation between E,nu and Shear (G) and Bulk (K) modulus.
there is possibility that you find error in my sheet since I generated it very quick. please double check it and let me know.
-
i've quoted simone.
-
what is the problem?
I don't know how to do it
-
this is a slightly modified model from previous one.
could you publish the values of nu, E and the relations between moduli you used?
i've tried with usual elastic relations but my values don't match with yours.
-
maybe it was too late when i checked those values.
i used the relations published in this document: http://ascelibrary.org/doi/abs/10.1061/9780784412763.026
assumed nu = (1-sin fi)/(2+sin fi) = 0,35
because i didn't know E, i used this relation: 2(1+nu)G = 3(1-2nu)K so for G=1119 and nu=0,35 i've found K=3357.
using nu=0,34 they match fine.