Calibrating paramters under different confining pressure

Hi,
Just adding more info according to today's simulation :

1. When I apply the set of parameters which agrees well with experimental results at100 KPa confining pressure to 400 KPa condition, the overall stiffness and peak strength in stress -strain curve is relatively lower than the experiment results, the volumetric response shows more compaction than experiment results with 400 KPa confining pressure.

2. When I apply the set of parameters which agrees well with 400 KPa experiment results to 100 KPa condition, the overall stiffness is relatively higher than the experiment results, the peak strength is OK, the volumetric response shows less compaction and a more quick transition from compaction to dilation.

Thanks,
Leonard

Hi,

I understand you're happy with your first, calibration, comparisons, and unhappy with your second, validation, comparisons. It "just" means the blind-prediction capabilities of your (DEM) model are somewhat deceiving for you, even though the model could be tuned to correctly reflect a real observation in the first place. That is a quite common situation for honest scientists...

You probably do not have much choice from there.
Either you lower your expectations, or you change your model (particle's shape or contact laws or everything that could contribute to a discrepancy between the real and DEM worlds).

Thanks jduriez,

I understand that there must be some distance between simulation and experiment results. It is difficult to make one set of parameters for all conditions.

I will still appreciate that you guys share more opinions or ideas about how you may deal with this problem.

e.g., making a compromise by selecting a set of intermediate parameters so that both conditions (100 KPa and 400 KPa) are roughly met but the fit is less accurate?

How is it usually handled?

Cheers,
Leonard

Hi Leonard,

like you said, it is difficult to create one set of parameters that will provide proper results for different conditions.

In my work, I used TriaxialCompressionEngine to perform some triaxial tests. As far as I'm concerned, when I wanted to properly prepare samples for two different confining pressures e.g. 100 kPa and 400 kPa, I had to use different friction angles between the particles (higher one for higher pressure), so the initial porosity was equal for both pressures (higher pressure will better squeeze
the specimen). I used the same approach to reach different initial porosities.

As far as I know, you can try to change your parameters a little, e.g.:
- increasing micro friction angle will increase only the peak value of the stress-strain curve and the residual value will stay on the same level,
- increasing young modulus will increase the global stiffness of the response.
- changing moments law etaRoll parameter can change the placement of the peak value on the stress-strain curve.

Cheers,
Aleksander

Hello,

First be sure that your model or the calibration is suitable for
describing the soil you consider.
For instance:
- if the soil is cohesive a calibration on a single triaxial test may
not be enough or a model without inter-particle adhesion will be useless.
- if soil particles are crushable and your model does not take that into
account, again it is useless.

Then, calibration of the mechanical parameters of the model may not be
enough. The initial state of the soil, and its reproduction with the
discrete model is also very important. If the initial state is not
simulated correctly then the model, even with calibrated mechanical
parameters may give a wrong responses.

Finally, be sure that the experimental results have a good repeatability
and are not to much affect by experimental "imperfections" (boundary
conditions, strain localization, etc ...): experimental results are not
necessarily representative of the constitutive behaviour of the soil,
they may be far from that for different reasons.

If all the above points (and others?) are OK then you should be able to
reproduce more or less the soil responses for different confining
pressure with the same mechanical parameters, see for instance:
Sibille et al. (2019) Quantitative prediction of discrete element models
oncomplex loading paths. IJNAMG
https://hal.archives-ouvertes.fr/hal-02059431/document

Luc

--
Luc Sibille
Université Grenoble Alpes / IUT1 de Grenoble
Laboratoire 3SR: Sols, Solides, Structures, Risques

Tel lab.: +33 (0)4 76 82 63 48
Tel IUT: +33 (0)4 76 82 53 36

Hi Aleksander,

Many thanks for your comments.

>> Higher pressure will better squeeze the specimen.
This makes sense for using relatively higher friction angle for higher confining pressure.

Parameters study is ongoing:)

Cheers,
Leonard

Hi Luc,

Many thanks for your sharing which lets me know more about DEM.

I got a lot of things to learn.

Cheers,
Leonard

For those who may need:

Besides the one Luc recommanded, paper [1] also gives lots of discussions above this question (e.g., Fig 10).

Cheers,
Leonard

[1]https://link.springer.com/article/10.1007/s10035-016-0687-0

See also further directions in [Suhr2017]
http://dx.doi.org/10.1007/s40571-016-0119-2

Bruno

On Thu, 2 Apr 2020 at 13:18, Leonard <email address hidden>
wrote:

> Question #689592 on Yade changed:
> https://answers.launchpad.net/yade/+question/689592
>
> Leonard posted a new comment:
> For those who may need:
>
> Besides the one Luc recommanded, paper [1] also gives lots of
> discussions above this question (e.g., Fig 10).
>
> Cheers,
> Leonard
>
> [1]https://link.springer.com/article/10.1007/s10035-016-0687-0
>
> --
> You received this question notification because your team yade-users is
> an answer contact for Yade.
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-users
> Post to : <email address hidden>
> Unsubscribe : https://launchpad.net/~yade-users
> More help : https://help.launchpad.net/ListHelp
>
>
>

--
--
_______________
Bruno Chareyre
Associate Professor
ENSE³ - Grenoble INP
Lab. 3SR
BP 53
38041 Grenoble cedex 9
Tél : +33 4 56 52 86 21
________________

Email too brief?
Here's why: email charter
<https://marcuselliott.co.uk/wp-content/uploads/2017/04/emailCharter.jpg>