[Solved] - Load control Cyclic Path

[gaetano_cmr]

Dear STKO Team,

I'm modelling masonry axial experimental test and I need to do a cyclic analysis but using "load control".
Implementing the same analysis in displacement control is quite easy I've tried it defining a Path.TimeSeries and everything went well.
Now I would like to do the same but in load control analysis, so using a vector with imposed values of forces.
Is it possible to do it in STKO?
Thanks in advance.

[STKO Team]

I don't understand what you want to achieve, however, you have 3 scenarios:

• You want to control the displacement cycles (you expect strain softening):
  • Apply a load with a linear-time-series
    Use a DisplacementControl integrator
    Choose the Cyclic option
    Use a path-timeSeries that defines the displacement cycles
  • Apply an sp non-homogeneous constraint (imposed displacement), with a path-timeSeries that defines the displacement cycles
  • Use a LoadControl integrator
• You want to control the load cycles (you do not expect strain softening):
  • Apply a load with a path-timeSeries that defines the load cycles
  • Use a LoadControl integrator

[gaetano_cmr]

yes, I would like to achieve the second option.
I already achieved a displacement control with a cyclic displacement path.
I tried to setup the same analysis but with a Cyclic load path and using load control, but how can I assign the load cycle to the control point?
I mean using in the analisy the option "displacement control" I can choose the control point and the cycle displacement path, this option however is not available in load control analysis option.

[STKO Team]

If you refer to this option

GC_LC1.png (17.05 KiB) Viewed 2692 times

You can control the intensity of the load just by:

• Assign a reference load, let's say for example a pressure of 10 MPa
• Imagine you want a cycle like [0 MPa, 5 MPa, -8 MPa, 10 MPa]...
• Create a Path-timeSeries that defines the load multiplier like [0, 0.5, -0.8, 1.0], define a dT such that the last point is a time 1, and define the maximum duration of your load-control analysis = 1
• Define a loadPattern for the reference load that uses this path series.

If you are not sure, please share your file here and I will have a look

[gaetano_cmr]

I was able with a copy of this model to do a cyclic analysis with imposed displacement but I'm still not able to do the same thing with an imposed cyclic force/pressure and a load controlled analysis.
I'm attaching the model file and thank you for your help.

061022_masonry_with_load_cycle.zip

(202.04 KiB) Downloaded 160 times

[STKO Team]

Here's an example of imposed load with load control:

T1_load_control_imposed_load.zip

(187.61 KiB) Downloaded 175 times

Keep in mind that in this way you can only be in the strain-hardening regime.

There were few issues in your model:

• In a 2D model it is important that your local Z vectors point to the positive global Z axes. Otherwise the jacobian of your quad elements will be negative and you will obtain garbage results.
  In your case, you can fix it either by rotating the whole model about the Y-axis, or using the Reverse command

  

  masonry_1_error.png (63.16 KiB) Viewed 2663 times

• You were fixing rotational DOFs in 2D continuum elements that do not have rotational DOFs
• You were trying to impose a pressure load using the sp(imposed nodal DOF) command on the P (pressure) DOF. However, in U-P elements used for soil models, the P DOF refers to the water pore pressure, which is treated as a nodal DOF using a Mixed-FEM formulation. It is NOT the pressure as a force.
  I simply changed it to an Edge Load (because you are in 2D) whose value is the Pressure (F/L^2] time the Thickness (F/L)
• You need a higher number of increments. When you use the IMPLEX integration scheme to improve convergence you need small time steps.
• The timeSeries for cyclic loading had 4 points with a dT = 1. So the total duration of the timeSeries was 4 seconds. But your analysis only has a total duration of 1 second.
  So I simply made the dT of the timeSeries = 0.25. Then I also added an extra point equal to the last one, because if for some round-off error the analysis time is slightly large and 1.0 (say 1.0 + small_epsilon) the load multiplier from the timeSeries will be dropped to 0

Let us know if this is what you wanted to achieve

[gaetano_cmr]

Every was really clear, I solved my problem.
Thank you very much STKO Team

[STKO Team]

You're welcome