[Yade-users] [Question #194761]: is particle distribution true or close to real sand simulated by yade
New question #194761 on Yade: https://answers.launchpad.net/yade/+question/194761 Hi all, recently, I have read some papers using Yade or SDEC simulating triaxial test over sand, and made good calibration over the dem model with experimental models. However, two questions puzzled. 1), Triaxial box is usually a bit large than the box used in a true soil test box; 2), we insert particles into the box, usually 8000~1, I have estimated the mean particle size used in yade, usually 20-40mm in average, a bit lager than the usually sand particle size 0.5mm -1mm. A box composed of particles a bit larger than the sand grains used in real experiment could represent the real sand sample? My understanding over these questions, especially the secong is below if we use real particle distribution and size in the simulation, simulation should be a long long time. So the usual particle distribution is larger, maybe 10-100 times, than that in real sand. and also, numerical sample even don't care about the particle distribution, just uniform distribution. I think, in microscopic, sand grain has some statistical characters like distribution, average radius, morphorgy and roughness. at least the first 2 should be close or even equal to each other over yade and experiment. now the grain parameters in papers using yade could only have one parameters close to real sand, relative density, why and how to deal with this problem? -- You received this question notification because you are a member of yade-users, which is an answer contact for Yade. ___ Mailing list: https://launchpad.net/~yade-users Post to : yade-users@lists.launchpad.net Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
Re: [Yade-users] [Question #194761]: is particle distribution true or close to real sand simulated by yade
Question #194761 on Yade changed: https://answers.launchpad.net/yade/+question/194761 Status: Open = Answered jduriez proposed the following answer: Briefly : for practical reasons computer simulations can not reproduce the reality as it is, as you understood it. However, if you agree that good calibration over the dem model with experimental models can be reached, you should not see any problem in the method... (in the sense that the results it gives are ok, even if it is simple) So welcome in, and enjoy, the numerical world ! :-) There might be something in common with the real one... -- You received this question notification because you are a member of yade-users, which is an answer contact for Yade. ___ Mailing list: https://launchpad.net/~yade-users Post to : yade-users@lists.launchpad.net Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
Re: [Yade-users] [Question #194761]: is particle distribution true or close to real sand simulated by yade
Question #194761 on Yade changed: https://answers.launchpad.net/yade/+question/194761 Christian Jakob posted a new comment: 2), we insert particles into the box, usually 8000~1, I have estimated the mean particle size used in yade, usually 20-40mm in average, a bit lager than the usually sand particle size 0.5mm -1mm. A box composed of particles a bit larger than the sand grains used in real experiment could It depends on what you want to investigate. If you to make microscopic investigations (like I do) you should use real parameters (real grain distribution, shear modulus of quartz, poisson of quartz). In this case you can not validate your model, because calculation times are too high for macroscopic dimensions. For example you would need millions of particles to simulate a triaxial test. If you want to make macroscopic investigations you should use non-physical parameters (shifted grain distribution with higher particle radii, shear modulus of granular sand, poisson of granular sand). In this case you can validate your model, but parameters are different. Here you only need thousands of particles to simulate a triaxial test. -- You received this question notification because you are a member of yade-users, which is an answer contact for Yade. ___ Mailing list: https://launchpad.net/~yade-users Post to : yade-users@lists.launchpad.net Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
Re: [Yade-users] [Question #194761]: is particle distribution true or close to real sand simulated by yade
Question #194761 on Yade changed: https://answers.launchpad.net/yade/+question/194761 Luc Scholtès proposed the following answer: Hi, First, you should know that size does not matter in YADE, or, at least, should not. Indeed, interaction laws (at least, the basic ones like Law2_ScGeom_FrictPhys_CundallStrack or CohesiveFrictionalContactLaw for example) have been implemented so that they are size independent. You can try to run several triaxial tests with same parameters and loading conditions on samples with different sizes length to verify (be careful nonetheless to have enough particles in your samples, between 5000 and 1 is OK from my own experience). So, YES, people usually use bigger particles than real ones in order to speed up simulations, and NO, there should not be any influence on the results. Concerning now particle size distribution, I am not a specialist but it will certainly have an impact on you results, like for real granular materials. Things exist in YADE to manage particle size distribution. As you pointed out, an extended size distribution tends to penalize simulation time. Concerning now your last point about grain shape and roughness. Of course they influence the behaviour. You cannot, for example, reproduce real sand's internal friction angle with spherical particles as sand grains are not spherical in nature. That's one of the reason why some people use clumps or rolling moment law for example (polyhedral particles are not yet implemented in YADE). Please have a look here for more information if not already done: https ://yade-dem.org/doc/index-toctree.html Cheers Luc -- You received this question notification because you are a member of yade-users, which is an answer contact for Yade. ___ Mailing list: https://launchpad.net/~yade-users Post to : yade-users@lists.launchpad.net Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
Re: [Yade-users] [Question #194761]: is particle distribution true or close to real sand simulated by yade
Question #194761 on Yade changed: https://answers.launchpad.net/yade/+question/194761 Chiara Modenese proposed the following answer: Hi Wangxiaoliang, The closer you get to the physics of the real material with your numerical model, the easier the calibration should become, up to the point where no calibration at all is needed (currently not possible due to the limit on the computational time, as Jerome said). The fact that you need to calibrate your model against experimental results does not mean that the model is wrong. With finite elements, you still need to calibrate the parameters of your mathematical constitutive model, for instance. Suggestion: Before deciding on the assumptions you want to make, you need to clarify what is the goal of your model and how you are going to do it. One example aong others: it is a bad idea to scale up the size of your particles if you are going to include attractive forces which scale with the particle size (the scaling would be incorrect then and the results difficult to interpret). Otherwise it should be safe doing it. Chiara On 26/04/2012 09:35, jduriez wrote: Question #194761 on Yade changed: https://answers.launchpad.net/yade/+question/194761 Status: Open = Answered jduriez proposed the following answer: Briefly : for practical reasons computer simulations can not reproduce the reality as it is, as you understood it. However, if you agree that good calibration over the dem model with experimental models can be reached, you should not see any problem in the method... (in the sense that the results it gives are ok, even if it is simple) So welcome in, and enjoy, the numerical world ! :-) There might be something in common with the real one... -- Chiara Modenese BSc MSc(Eng) DPhil(PhD) Candidate in Engineering Science Department of Engineering Science University of Oxford Parks Road, Oxford, OX1 3PJ, UK You received this question notification because you are a member of yade-users, which is an answer contact for Yade. ___ Mailing list: https://launchpad.net/~yade-users Post to : yade-users@lists.launchpad.net Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
Re: [Yade-users] [Question #194761]: is particle distribution true or close to real sand simulated by yade
Question #194761 on Yade changed: https://answers.launchpad.net/yade/+question/194761 Status: Answered = Open wangxiaoliang is still having a problem: thanks, all of your answers are helpful to me. I know that now if is impossible to derive macroscopic behavior from real grain level dynamics. So, to model one sand, Grenoble sand, for example, we have the physical parameters given, any reasonable technique exists on how to choose parameters for my particle or dem model? say like particle size, distribution or clump type. If not, how can we say my model represent Grenoble sand, maybe the same DEM sample with another set of parameter (kn ks miu) could model Ottowa sand. -- You received this question notification because you are a member of yade-users, which is an answer contact for Yade. ___ Mailing list: https://launchpad.net/~yade-users Post to : yade-users@lists.launchpad.net Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
Re: [Yade-users] [Question #194761]: is particle distribution true or close to real sand simulated by yade
Question #194761 on Yade changed: https://answers.launchpad.net/yade/+question/194761 Status: Open = Answered Hadda proposed the following answer: Hi, May be it would take you a lot of time to adjust your DEM parameters to reproduce an identical fingerprint of given natural sand (which is not obvious too). Also, nobody will expect such a correspondence using spherical or whatever shape bodies of cristal with a simple contact law. Nevertheless, by fixing some parameters in your DEM model you can put forward or emphasize a typical characteristic of your sand you may will for. I agree with the suggestion of Chiara, that is to fix the aim behind your DEM simulations and to have an idea about the influence of each DEM parameter on the macroscopic behaviour of an assembly of paticles (as Luc said, an assembly of 1 particles is large enough to start your tests with, even less if you want to go faster!) -- You received this question notification because you are a member of yade-users, which is an answer contact for Yade. ___ Mailing list: https://launchpad.net/~yade-users Post to : yade-users@lists.launchpad.net Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
Re: [Yade-users] [Question #194761]: is particle distribution true or close to real sand simulated by yade
Why do you think size matters? ___ Mailing list: https://launchpad.net/~yade-users Post to : yade-users@lists.launchpad.net Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
Re: [Yade-users] [Question #194761]: is particle distribution true or close to real sand simulated by yade
Question #194761 on Yade changed: https://answers.launchpad.net/yade/+question/194761 Status: Answered = Solved wangxiaoliang confirmed that the question is solved: thanks a lot, all of u. -- You received this question notification because you are a member of yade-users, which is an answer contact for Yade. ___ Mailing list: https://launchpad.net/~yade-users Post to : yade-users@lists.launchpad.net Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
