Dynamic Analysis Using max Step and Substep

Dear Vu Ngo,

Sand boiling is caused by the volume reduction of the sand when it is shaked. Liquefaction occurs typically in loose sands: when the loose sands is being shaken it densifies, hence the void ratio decreases and therefore the pore water gets "trapped" causing the excess pore pressure to go up. Since the sand densifies it reduces volume, so the water needs to flow out and it will flow out towards the soil surface making it look like the sand "boils". This is just the physical process that occurs; I'm not sure what you exactly mean with "it is corrected"...? What did you assume is being corrected?
Liquefaction usually occurs close to the soil surface because there contact stresses between the sand particles are relatively small so it's easier for the grains to lose contact and densify. At larger depth, but also for instance below foundations, contact stresses are larger and so the sand particles are less likely to lose contact.

Excess pore pressure concentrations close to layer separations (especially if the layers have large stiffness difference) or boundaries are not uncommon and are caused by the movement limitation of the soil there due to the presence of a boundary or stiff layer. Generally the excess pore pressure concentration is very local and does not lead to liquefaction. Whether the effect is purely numerical or in fact also occurs in reality is unknown as there is simply no measurement data on this.

With kind regards,

Dennis Waterman

Thank you Dennis for the detailed explanation!

Understanding of the wave propagation in a model is not easy, specially in between the two very different material stiffness or between the soil and structure. That may be the reason why excess pore pressure (UBC3D-PLM) concentrated very high at the corner of the two materials. Beside, is it corrected with high negative pressure (positive value in Plaxis) below ground surface? I understand that this is the reason why the sand is usually boiled or flowed up during an earthquake. Am I right? Don’t mind if I asked the questions from the side of engineering work.

Thank you,

Dear Vu Ngo,

The proposed times steps and substeps by PLAXIS are based on a couple of criteria:

- The time step should not be larger than the time step used in the input signal (using signal from file)
- The time step should divide the smallest sine wave in at least 10 steps (using harmonic signal)
- The time step should be such the wave does not pass more than half a soil element per time step

If one of the fist 2 are critical there is not so much you can do because you really don't want to take time steps so large that some of the data of your signal is not being used.
If the 3rd is critical then it may be worth investigating. Small elements and high stiffness (hence high wave velocity) make the minimal recommended time step very small and thus the amount of substeps very high. However, if the critical element (either in size or in stiffness) is in a part of the project area that is not so important, you could decide to use less substeps and accept the inaccuracy in those parts. Or you could generate a bit larger elements there.
For instance, if you're doing a seismic analysis and at the bottom of your model you include some of the bed rock in your model ánd generate small elements there then the very stiff and small bedrock elements really blow up the amount of substeps. You could solve this by either generate somewhat larger elements in the bedrock, or maybe not include the bedrock at all and just use the top of the bedrock layer as bottom of your model thus allowing larger time steps.

With kind regards,

Dennis Waterman

Dear Dennis,

Can you suggest us a procedure to set the parameters for a reasonable accuracy and calculation time. I usually apply recommended max and sub steps by using retrieving, sometime it recommended quite different parameters such as very high max step and low sub step or both is in the medium range. I don’t know why? It may take half to a working day for a 2D calculation, so I usually consider low values and coarse mesh for preliminary calculations and sufficient values for calculations to be finalized and create an animation to present also.

By the way, how to create an animation with better quality? 

Thank you,

Dear Tolgahan,

There is no difference, a step and a substep are exactly the same in terms of calculation. The only difference is that for a step PLAXIS can save the results and for a substep PLAXIS cannot save the results.
So if you use max step = 3500, substep = 1 then you can potentially save 3500 steps to file and also make a curve of 3500 steps.
If you use max step = 1, substep = 3500 then you will only have 1 saved result (the last step) and your curves will also only have 1 point.
But the final result of both options is the same.

With kind regards,

Dennis Waterman