• Tii
  • Member
  • Offline
  • Registered: 2013-04-15
  • Posts: 14

Topic: Simulating uniaxial compression test

Hi all,

I'm trying to simulate a uniaxial compression test at a concrete cube (3D). I'm using the EC2 C50/60 so the compression strength is fc=-50 MPa. The modelled cube (just one 3D solid element) is loaded with a displacement till it will yield and with the maxmin-Volume-monitor the minimum stress and strain can be displayed.
The problem is that the cube yields at ca. -52,4 MPa not at -50MPa. Boundary conditions are right and the load steps are really small.
So why is the yielding strength not -50 MPa?

Quote Post 1

  • dpryl
  • CC software developer
  • Offline
  • Registered: 2007-07-13
  • Posts: 1,107

Re: Simulating uniaxial compression test

Dear Tii, without seeing your model, I can't check if the BCs are right. However, I don't know a way to correctly simulate the classic cube test, where friction at the top and bottom of the sample is NOT eliminated, using a single finite element! Moreover, the stress in the classic experiment is not measured, but a calculated average (force/reference cross section area).

Please also read ATENA Troubleshoting, 2.1.21 The tensile (or compressive) stresses exceed the tensile (or compressive) strength or yield stress – what is wrong?

dpryl's Website

Quote Post 2

  • Tii
  • Member
  • Offline
  • Registered: 2013-04-15
  • Posts: 14

Re: Simulating uniaxial compression test

Hi,
thank you for your fast reply.

My problem is that if I take the CC3DNonLinCementitious2 material with a concrete strength of fc=-50 MPa and I just have one load direction (uniaxial) the finite element has to yield at the minimum principal stress of -50 MPa (its kind of a very simplified patch-test).

I am taking a 8 node solid element. 4 nodes are loaded with a uniaxial displacement till the element will yield. The other 4 nodes are fixed with boundaries in load direction. One of these nodes is fixed in the other two directions too (for numerical forces in these directions).

Quote Post 3

  • dpryl
  • CC software developer
  • Offline
  • Registered: 2007-07-13
  • Posts: 1,107

Re: Simulating uniaxial compression test

Hello,
1. if you like us to check your model, please send it along with your ATENA User ID (WR) and a description and/or sketch.
2. for a single element in uniaxial compression, the rigid body rotations have to be blocked. Your description of the conditions sounds like allowing rotation around the vertical axis. Normally, one would model 1/4 of the sample and apply symmetry conditions.

dpryl's Website

Quote Post 4

  • Tii
  • Member
  • Offline
  • Registered: 2013-04-15
  • Posts: 14

Re: Simulating uniaxial compression test

Hi,

one of the nodes is fixed in each direction so rigid body rotations are blocked.

Maybe I will explain you my intention:
I want to do a patch test on a single solid element to evaluate the bi- and triaxial compressive strength. Then I want to change the eccentricity factor e (from the Willam-Menetrey failure surface) to evaluate its influence on the bi- and triaxial strength.

Is this possible with just one element? I think of a displacement on three sides of the element (for biaxial failure) and on five sides for triaxial failure. In the bi-axial state the compressive strength has to be around 1.2 times of the uniaxial strength (e=0.52), this is what I want to check with the patch-test.

Quote Post 5

  • dpryl
  • CC software developer
  • Offline
  • Registered: 2007-07-13
  • Posts: 1,107

Re: Simulating uniaxial compression test

Yes, a single-element model is a good approach for this. The typical (and recommended) modelling is 1/4 symmetry. Please do not forget the effects of non-zero Poisson's number when preparing your boundary conditions. Please also note the confinement effect is very sensitive to even very small tensile or compressive stresses in the 3rd (or 2nd+3rd) dimension(s). As these stresses are easily smaller than 1% of the max. compressive stress, it can make sense to tighten the convergence tolerances in the solution settings.

dpryl's Website

Quote Post 6

  • Tii
  • Member
  • Offline
  • Registered: 2013-04-15
  • Posts: 14

Re: Simulating uniaxial compression test

Hi,

now I have a good single element model but the results for the biaxial stress state are nearly 8% above the analytical solution of the willam-menetrey failure law.

In a second model I modelled 1/4 symmetry but I think there is no chance to simulate a biaxial failure with it because of the confinement effect.

It would be great if I could send you my GiD model. Should I send it to the cervenka contact mail-adresse on your homepage or can I send it directly to you?

Quote Post 7

  • dpryl
  • CC software developer
  • Offline
  • Registered: 2007-07-13
  • Posts: 1,107

Re: Simulating uniaxial compression test

> now I have a good single element model but the results for the biaxial stress state are nearly 8%
> above the analytical solution of the willam-menetrey failure law.

As already mentioned, you may need to tighter the convergence criteria as the confinement effect can be very sensitive to the lateral stresses.

> In a second model I modelled 1/4 symmetry but I think there is no chance to simulate a biaxial
> failure with it because of the confinement effect.

Why??? The BCs other than the symmetry ones are defined and act the same as for the full model.

> It would be great if I could send you my GiD model. Should I send it to the cervenka contact
> mail-adresse on your homepage or can I send it directly to you?

Please send as instructed in ATENA Troubleshooting, 2.1.1 I have a problem not listed here.

dpryl's Website

Quote Post 8