Abaqus vs Plaxis - 1D site response analysis

**michael27** · 2010-08-10, 22:18

Hi everybody,

I'm a long time reader, first time poster on these forums...

Basically I am in the process of comparing the results obtained from Plaxis and Abaqus for peak ground accelerations derived from three points (bottom, middle and top) within a soil column subject to ground shaking at the base.

Within the Plaxis model, I have specified Standard Absorbent boundaries to the left and right hand sides of the model in order to absorb propogating shear waves. And have applied standard fixities along the base of the column.

Within Abaqus, I have created a set of infinite elements within the Job *.inp file, to border the left and right hand sides of the model and to abosorb propogating shear waves. I have also specified a disp/rotation boundary condition at the base of the column.

Both models use the same linear elastic material properties. (no rayleigh damping applied).

Both models use the same Newmark implicit time integration method. (I have set newmark alpha and newmark beta values so that they match between the two programs)(ie within Abaqus: alpha = -0.1, and within Plaxis: alpha = 0.3025, Beta = 0.6)

Both models contain roughly the same number of elements and nodes.

And both models employ the same theory behind the above mentioned abosorbent boundaries (lysmer and khulemeyer) (I have set relaxation coeficients C1 and C2 to 1.0 within Plaxis to obtain this similarity).

And despite all os this, I am getting much smaller accelerations from Plaxis than I am from Abaqus!

My question is, is the plaxis dynamic analysis doing anything behind the scenes that I do not know about? or am I missing something crucial with definition of the Abaqus model, and Dynamic Step?

I have attatched my *.inp file if anyone is interested. (note: ive cut out the amplitude definition and deleted some elements to meet file size requirements..)

apologoes for the long post, I just got too carried away!

- Michael

**Jorgen** · 2010-08-23, 20:31

I am afraid that I am not familiar with Plaxis.

-Jorgen

**michael27** · 2010-08-23, 21:28

Well, Plaxis has a default and relatively trustworthy method for performing an earthquake site respoinse analysis. So my approach is to hopefully match the results obtained from an Abaqus analysis, with the Plaxis Analysis. Once ive doen this, ill know the Abaqus model is correctly set up, I can then extend this analysis to 2D and 3D.

At the moment the results from Abaqus are slightly larger than that from Plaxis.

Im putting this down to the dynamic alpha value I use within abaqus, which is responsible for introducing numerical damping to the implicit analysis.

I have manually changed the value for alpha within the input file from; its default value of alpha=-0.05, to alpha=-0.1 and to alpha=-0.3, and have witnessed no reduction in accelerations within the model.

Im unsure if changeing the value of alpha in this way, is actually applying it to the implicit analysis...

ive attatched my dynamic step if anyone is interested...