Edit

calibration of subloading surface model

The subloading surface framework provides advantageous features that conventional plasticity models do not offer when it comes to modelling cyclic behaviour.

In this example, we demonstrate the performance and how to calibrate the Subloading1D model.

We mainly use the data presented in this monograph.

To make life easier, we first define some utility functions.

from dataclasses import dataclass
from os import system
from h5py import File
import matplotlib.pyplot as plt


@dataclass
class Curve:
    name: str
    x: list[float]
    y: list[float]


def execute(name: str, content: str):
    with open("example.sp", "w") as f:
        f.write(content)

    system("suanpan -np -f example.sp")

    with File("RESULT.h5") as f:
        return Curve(name, f["dataset"][:, 0], f["dataset"][:, 1])


def plot(title: str, curves: list[Curve]):
    for curve in curves:
        plt.plot(curve.x, curve.y, label=curve.name)
    plt.legend()
    plt.tight_layout()
    plt.grid()
    plt.title(title)
    plt.show()

Smooth Transition

We consider a very basic, elastic-perfectly plastic response. We choose a moderate value 400 MPa400~\text{MPa} for initial yield stress.

The evolution rate uu of yield ratio controls the curvature. A higher value uu results in a faster approaching, making the response closer to conventional elastic-perfectly plastic. However, the conventional bi-linear response is not realistic, and the stiffness shows a discontinuity/jump at the yielding point.

The introduction of yield ratio zz ensures a controllable smooth transition from the conventional 'elastic' region to the 'plastic' one.

png

Isotropic Hardening

On top of the initial yield stress, a saturation can be added. The response exponentially approaches the saturation stress, this can be accompanied with a linear hardening as well.

Here, we choose a linear hardening modulus 2000 MPa2000~\text{MPa}, which is 0.10.1 of elastic modulus. However, this hardening modulus is measured with plastic strain, the total response would present a different slope. We further choose a saturation of 200 MPa200~\text{MPa}, so the total stress shall be 600 MPa600~\text{MPa}, in absence of linear hardening.

We shall see that misosm^s_\text{iso} controls saturation speed.

png

Kinematic Hardening

Similarly, kinematic hardening also saturated to aia^i. In additional to that, since it is defined that the saturation itself is not a constant, itself can be saturated again. We choose a ai=100 MPaa^i=100~\text{MPa} with potentially another 150 MPa150~\text{MPa} saturation, resulting in 500 MPa500~\text{MPa} to 650 MPa650~\text{MPa} in total, depending on the configuration.

png

Ratcheting

The original subloading surface model shows excessive plastic strain accumulation under cyclic loading. The extended version addresses such an issue with the assist of the elastic core. It is controlled by $z_e$ and $c_e$, the former controls the size, while the latter controls the rate.

To illustrate, we choose a bilinear hardening model as the basis.

png

Test Data

We show a few example calibrations based on test data shown in the monograph.

png
png

Caveats

It is not recommanded to define a large linear isotropic hardening.

Since the loading phase always incurs accumulation of plasticity, stable one sided cycles will accumulate plasticity, which further results in larger yield surface if the linear isotropic hardening is present.

The following shows an example with a small amount of linear isotropic hardening.

png

However, when a significant linear isotropic hardening is defined, the accumulated plasticity will yield significantly larger yield surface.

png
Previousthin-walled section analysis for frame structuresNextDynamics

Last updated