Viscoelastic Viscoplastic Damage Model Development, Parameter Estimation, and Comparison to PBX9501 Experimental Data
Abstract not provided.
Engineering topics
Publications and source records attributed to Buechler, Miles A..
Abstract not provided.
This report provides the necessary details to implement a viscoelastic-viscoplastic-damage model into an implicit finite element code. It should be noted that this model has already been implemented into an Abaqus subroutine. The intent is not to fully describe the theory, though enough detail is presented to guide the reader towards the final equations. An additional report providing the full theoretical development may be written on request.
This report briefly summarizes an interesting set of experiments performed on PBX9501 in FY13. It also describes the use of the experimental data in a constitutive model development program. The experiments were requested as a follow-on to a set of cyclic compression experiments performed by Darla Thompson in 2008. Thompson originally performed the experiments to demonstrate a capability and to stimulate discussions on what experimental data W13 needed. Those experiments proved crucial in demonstrating that the material exhibited permanent deformation after removal of load, viscoelastic recovery, and some form of damage. Further experiments were requested to flesh out a viscoelastic-viscoplastic-viscodamage model.
This report documents three improvements made to the implementation of a historically significant material model used in W13 for simulations involving components made of PBX9501. ViscoSCRAM combines linear viscoelasticity with isotropic damage evolution based on fracture mechanics concepts. The original implementation was focused on short duration transient events, thus an explicit update scheme was used. For longer duration simulations which employ relatively larger time step sizes the explicit update scheme is inadequate. This work presents a new semi-implicit update scheme suitable for simulations using relatively large time steps. The algorithm solves a nonlinear system of equations to ensure that the stress, damaged state, and internal (Maxwell) stresses are in agreement with implicit update equations at the end of each increment. The crack growth is advanced in time using a subincremental explicit scheme, thus the entire implementation is semi-implicit. Additionally, modifications to the crack growth kinetics are proposed and implemented obviating the need for an empirical tensile damage growth rate (TDGR) multiplier used in previous implementations. Finally, an automatic time stepping algorithm is implemented to adjust the time step size based on error estimates for the generalized Maxwell model update scheme. Model parameters are fit to existing data and example problems are simulated using both new and legacy implementations to assess comparative advantages and disadvantages.
A constitutive model is developed which incorporates linear viscoelasticity and combined isotropic and kinematic rate dependent hardening plasticity based on the Drucker-Prager yield criterion and flow potential. Abaqus currently has intrinsic material models for linear viscoelasticity, isotropic hardening Drucker-Prager plasticity, and J2 plasticity with combined hardening, but there is no facility to combine these behaviors in one constitutive model. An Abaqus UMAT is developed using the equations developed in this document and then a code to code comparisons are performed using the Abaqus intrinsic models and the new material model with parameters selected such that only the behaviors Abaqus can produce are present. The comparisons show that the new model can produce the individual behaviors very well. A 1D development of a damage model that could be easily incorporated in this framework is also discussed.
This memo discusses observations that have been made in regards to a series of monotonic and cyclic uniaxial experiments performed on PBX9501 by Darla Thompson under Enhanced Surveilance Campaign support. These observations discussed in Section Cyclic compression observations strongly suggest the presence of viscoelastic, plastic, and damage phenomena in the mechanical response of the material. In Secton Uniaxial data analysis and observations methods are discussed for separating out the viscoelastic effects. A crude application of those methods suggests the possibility of a critical stress below which plasticity and damage may be negligible. The threshold should be explored because if it exists it will be an important feature of any constitutive model. Additionally, if the threshold exists then modifications of experimental methods may be feasible which could potentially simplify future experiments or provide higher quality data from those experiments. A set of experiments to explore the threshold stress are proposed in Section Exploratory tests program for identifying threshold stress.