3.2.6 Verification of creep integration

Product: ABAQUS/Standard  

This two-part example is intended to verify the algorithms used to integrate creep constitutive behavior by comparison with closed-form solutions. In the first part the constitutive creep behavior for simple creep and relaxation tests are verified. In the second part solutions for the Mises, the Drucker-Prager, and the extended Drucker-Prager/Cap creep and plasticity models are verified for tests in which the load is ramped from zero over a given period of time.

Problem description

Material

For all test cases the elastic material properties use a Young's modulus of 138 GPa (20 × 106 psi) and a Poisson's ratio of 0.3. In the creep and relaxation tests (first part of the example) the material definition uses creep behavior with the Mises stress potential and the equivalent uniaxial creep strain rate defined by where q is the Mises equivalent stress, t is time in the time hardening case, or—in the strain hardening case—

A, n, and m are constants, which are defined here as  1.6 × 10–16 MPa–5sec–0.8 (2.5 × 10–27 psi–5sec–0.8),  5, and  –0.2.

For the strain hardening case t can be eliminated from the creep strain rate definition, giving

The time hardening creep law defined above with  1.6 × 10–16 MPa–5sec–0.8 (2.5 × 10–27 psi–5sec–0.8) is specified for the coupled Mises and the coupled Drucker-Prager models. The plasticity hardening curve is given by

where is the equivalent plastic strain,  69 MPa (1 × 104 psi), and  0.2. The Drucker-Prager model is reduced to a Mises model by specifying the material angle of friction,  0.0, and the dilation angle,  0.0. No intermediate principal stress effect is used (i.e.,  1.0), as is required by this type of model.

A Singh-Mitchell type creep law is used for the case employing the modified Drucker-Prager/Cap model, activating the cohesion mechanism only. This creep strain rate is defined by

where is the equivalent uniaxial compression creep stress, and the constants  2.5 × 10–5 sec–1 (2.5 × 10–5  sec–1),  1.45 × 10–2 MPa–1 ( 0.0001 psi–1),  1.0, and  0.0.

Solution control

ABAQUS begins the analysis with explicit integration and continues to use that method unless its stability limit appears to be too severe a restriction on the size of the time increment or if plasticity occurs. If one of these conditions occurs, ABAQUS switches to the backward difference method; thus, the integration is unconditionally stable, and the only limitation on time increment size is solution accuracy. This approach is usually the most economic method for applications involving this type of material behavior.

The accuracy of the time integration of the creep behavior is determined by the size of the time increments chosen by the automatic time incrementation scheme, which is controlled by the value assigned to the CETOL parameter on the *VISCO option (Quasi-static analysis, Section 6.2.5 of the ABAQUS Analysis User's Manual). CETOL limits the difference between the creep strain increments computed from the creep strain rates calculated from conditions at the beginning and at the end of the increment. In a case such as this, where the creep strain rate depends strongly on the stress, the usual guideline for setting CETOL is to decide on a value that represents a small error in the stress and then divide that value by the elastic modulus to determine CETOL. For the creep and relaxation tests we have used 0.69 MPa (100 lb/in2) as a small stress; hence, CETOL is chosen as 5 × 10–6. For the coupled Mises and coupled Drucker-Prager tests a CETOL of 1 × 10–4 showed sufficient accuracy, and for the modified Drucker-Prager/Cap test a CETOL of 5 × 10–6 was selected.

Exact solutions

For the one-dimensional cases observed here, the uniaxial stress is equal to the effective stress, q, and also equal to the equivalent creep test stress, In the creep test the creep law can be integrated directly to give

This solution is the same for both time and strain hardening. It is plotted in Figure 3.2.6–2.

In the relaxation test the strain is constant, so

For the time hardening assumption this gives

which integrates to give

where is the stress at the start of the event. This solution is shown in Figure 3.2.6–3.

For the strain hardening assumption the governing equation becomes

Since the strain is constant, and at any time. Thus, the governing equation defines the stress by

so q is defined by

This equation is integrated numerically, using a fourth-order Runge-Kutta scheme, with a time increment of one second, which should provide a solution of high accuracy. The solution is plotted in Figure 3.2.6–3. For the relaxation test the solutions provided by the time hardening and strain hardening models are slightly different.

In the second part of this example, the closed-form solution for the creep strain of both the Mises and the Drucker-Prager models can be obtained by integrating the strain rate. The following is the exact solution for the total strain prior to yielding:

where is the initial yield stress occurring at time . After the onset of yield, the exact solution for the total strain is as follows:

For the case employing the modified Drucker-Prager/Cap model, a very high value of yield stress is specified to prevent yielding. Thus, a closed-form solution of the total strain can easily be obtained:

Results and discussion

Input files

Figures

Figure 3.2.6–1 Typical elements for creep and relaxation tests.

Figure 3.2.6–2 Creep test history.

Figure 3.2.6–3 Relaxation test history.

Figure 3.2.6–4 Mises and Drucker-Prager creep and plasticity models.

Figure 3.2.6–5 Modified Drucker Prager/Cap model (stress applied as a ramp function).

Figure 3.2.6–6 Modified Drucker Prager/Cap model (stress applied as a step function).