Hi, I am working at Aalborg University. We have done extensive testing on dense sand, undrained conditions, and found it has response slightly beyond what currently is "state of the art". Namely - we concentrated on the after loading when 1000 cyclic loads sustained by the sample. We do the tests on a single diameter height test under slipping boundary conditions - the kind of triaxial tests that give homogeneous stress state inside the sample, leading to results exceptionally well suited for FEM or numerical simulation.
We reached a point where classic theory no longer provides the necessary assumptions on what leads to the response observed. Undrained cyclic loading was found to increase or decrease the friction angle, depending on the load average and cyclic load... we found a slight decrease in density can lead to a stable state where granules "interlock" and provide higher strength with changing dilation and contraction properties...
How far can DEM go in triaxial test simulation? It is extreeeemely complicated to measure any kind of structural integrity change inside a real sample, If a simulation would provide the full range of response - that would be a breakthrough. I have all the stress paths and responses out on interactive Matlab plots but currently available FEM soil models just do not cut it well enough when the paths are fitted...
It takes me 6-8 hours to setup and run a full 3h long cyclic load on a sample. Is there any chance that a numerical model based on DEM could provide a faster computation? 1000 sinusoidal load cycles are applied on each specimen, at intervals of approx 10[s]...
DEM is very interesting but it is not a part of the standard curriculum here. I'd love to hear about your experience in applying it. I have all the measurement data that needs to be fitted, but other tools I've encountered so far seem to be highly underdeveloped. Wonder how far DEM can take us at this point in time.
Tomas Sabaliauskas