MAZARS model

GPSalachs

I found interesting the book "Computational Modelling of Concrete Structures" Gunther Meschke, Bernhard Pichler.

You can also look in the code-aster.org documentation for the examples that apply the Mazars Model!

Edit: Also look into this post from 2014 : code-aster.org/forum2/viewtopic.php?id=20005

GPSalachs

I have a question regarding the threshold of damage parameter of the Mazars Model.

When the model i am making is in milimetes, does the threshold of damage also need to be calculated in milimeters?

nirmaljoshi

Today i was running the model with similar case. The model ran only when i made mesh in meters.

GPSalachs

Did you use pilotage to make your case or DDL_IMPO ?

I am asking because i am trying to make it run by running a force with a ramp function.

Do you apply also self weight with Mazars Model of damage?

GPSalachs

If i have a continuous 3D medium and i want to use Mazars Model should i assign with the command AFFE_CARA_ELEM an element to the

medium and while at that should i define an initial analysis with MECA_STATIQUE and a secondary analysis with STAT_NON_LINE to get the

initial linear part and the secondary non linear part of the material right?

nirmaljoshi

It seems DEFI_MATER_GC (U4.42.07) is easier and convenient to assign Mazer model. Did you try?

GPSalachs

No haven't tried the DEFI_MATER_GC (U4.42.07) command because i want to remain and understand first better the DEFI_MATER command with mazars model .

nirmaljoshi

Actually DEFI_MATER_GC is converted to DEFI_MATER during analysis and parameters (used by DEFI_MATER) is printed in the output file e.g.

C30_EC2 = DEFI_MATER_GC(RHO=2400.0,


                          MAZARS=_F(CODIFICATION='BAEL91',
                                    FCJ=3.9000000E7,
                                    UNITE_CONTRAINTE='MPa',),

INFO=1,)

== Paramètres of the model MAZARS [MPa] ==

Partie elasticity:

E = 3.73031085E+06, NU = 2.00000000E-01, RHO = 2.40000000E+03,

Nonlinear part:

BT = 1.59414953E+00, AC = 2.79632486E-03, SIGM_LIM = 2.34000000E+07, AT = 9.00000000E-01,

BC = 1.68155462E+01, K = 7.00000000E-01, EPSI_LIM = 3.50000000E-03,

EPSD0 = 6.27293728E-01,

For information:

FCJ = 3.90000000E+07, FTJ = 2.34000060E+06, EPSI_C = 2.10253884E-01,

==================================================

This method may be useful to understand the material-model although this is like reverse engineering.. ;)

GPSalachs

I made some tests indeed it feels easier to do it with the DEFI_MATER_GC command!

On another part tho, the curves one produces in 3D should be taken from the forces and plastic deformations? So like FORCE_REAC and EPSP ? Or am i considering something wrong here? I ask that cause if i take the DX,DY or DZ components they give linear results so i don't know if i am doing wrong somewhere or am i confronting something wrong..

Thank you!

nirmaljoshi

You can use Linear plus Non-linear analysis with STAT_NON_LIN alone. No need to use MECA_STATIQUE. Also, you need to define COMPORTMENT in STAT_NON_LIN to make the material behave as non-linear.

GPSalachs

Hello, thank you for the response.

My problem is that i define the comportment as MAZARS in STAT_NON_LINE and run the analysis but the DEPL i get is almost always linear and i don't really understand the why.

I am attaching the .comm and .med files for further insights and discussion!

Attached file:
comp.tar.xz, 6720kb

nirmaljoshi

1D mazar model in excel with stress strain relationship.

Attached file:
mazarModel.xlsx, 42911kb

nirmaljoshi

I am fining difficulty to converge the calculation. Is there any special way to make the calculation convergent rapidely?

GPSalachs

Yes i am too finding it difficult to make converge this kind of calculation. I do not know if the problem lies with the elements of the mesh (HEXA8) or something else entirely.It could be also some mistake i am making in the STAT_NON_LINE command regarding the comportment or the algorithm for the resolution. I have tried concatenating two STAT_NON_LINE commands or piloting it but the results do not give me the post peak curve.

And i too am trying to understand how to make it converge..

Could it be an orientation issue here? Or maybe the '3D' type of modeling?

AsterO'dactyle

Hello,

Softening in your stress-strain curve => Newton cannot converge after the peak

Have a look on "PILOTAGE " methods (continuation algorithms).

nirmaljoshi

AsterO'dactyle wrote:

Have a look on "PILOTAGE " methods (continuation algorithms).

How to use PILOTAGE in the case of Mazars? For example in the model provided by GPSalachs.

I get error saying while useing the PILOTAGE with deformation.

! !

! One cannot control in the absence of forces of the type FIXE_PILO.

GPSalachs

The type of forces FIXE_PILO means that you have to select in the loads the FIXE_PILO property without a multiplying function.

On that end can it be done so i can run the analysis with PILOTAGE from the beginning or should i give for the piloting initial conditions with a beginning STAT_NON_LINE command?

In the event of successive forces do i have to pilot each force separately?

Just a reference thread i found : forum2/viewtopic.php?id=12930

nirmaljoshi

Did you solve the convergence issue? I am still struck with convergence. I found that convergence is more difficult in the tension.

If you found the solution can you please share.

GPSalachs

Hello,

i am trying the long arc method of piloting. I tried also the PRED_ELAS method because in the ''12-Continuation methods for non-linear analysis'' it is suggested for ''damage'' laws but it doesn't like Mazars model, unless i have to select certain properties to make it work.

For the long arc method now, i saw a Thomas DE SOZA (forum2/viewtopic.php?id=12930) post from a while back that indicates some the use of some properties:

PILOTAGE=_F( TYPE = 'LONG_ARC', << arc-length method

SELECTION='ANGL_INCR_DEPL', << how to determine solution when it is not unique, this one selects the one that minimizes the angle on the arc-length

GROUP_NO = 'TOUT_ELT', << which dofs to select (usually the whole structure)

NOM_CMP = ('DX', 'DY', 'DZ',), << which components (usually all)

ETA_PILO_MAX=5., << max multiplicative of the load

COEF_MULT = 120.), << the tricky part, it controls how fast you move on the arc-length, it is related to the time step, you should increase its value if you do not get past the snap-back (so as to decrease the increment on the arc-length).

Now i don't completely understand the correct use of these parameters and i am going through the test , fail and retest road.

I have seen some examples of piloting the model of mazar but those use the Deformation piloting method.

As far as the deformation method is concerned, do i have to get an initial state before continuing with the piloting by deformation?

nirmaljoshi

Regarding deformation method, you need to apply displacement by assigning the loading point as a support.

load = AFFE_CHAR_MECA(DDL_IMPO=_F(DY=-1e-06,


                                  GROUP_MA=('group_loadingEdge', )),

MODELE=model)

After running the analysis, you need to read the reaction (or sum of reaction) from the support nodes.

table0 = POST_RELEVE_T(ACTION=(_F(GROUP_MA=('group_obsFace', ),


                                  GROUP_NO=('group_loadingNodes', ),
                                  INTITULE='eqvalentreaction',
                                  NOM_CHAM='REAC_NODA',
                                  OPERATION=('EXTRACTION', ),
                                  RESULTANTE=('DY', ),
                                  RESULTAT=rxn),

)

I am not sure how the piloting method works though.