This Tech Talk will discuss this new feature and provide a number of examples of workflows that can be modestly reconsidered with significant improvement.
Abaqus has historically had a paradigm of "history dependency" in analysis a general step follows prior general steps, and permanent effects "carry over". While this is a generally practical approach most often consistent with "reality", there are certain classes of problems for which such an approach is not ideal.
In the case of structures exposed to a variety of "worst-case envelope loads", it may be reasonable to treat these various loading scenarios as "independent" so as to not accumulate prior stresses and strains. Other scenarios such as parametric studies of different variables into system performance (e.g. changes of friction and/or rotational speeds in brake assemblies) may yield performance benefits if these different "branch studies" can be calculated independently, while still benefitting from relevant prior states.
In the past, the "RESTART" function has been a sufficient workaround to address such approaches, particularly when pre-loading effects impact all solutions. More recent features in Abaqus present a superior solution to this classic approach, while also allowing "one model and one set of results" to be created. This results not only in numerical efficiencies, but also in post-processing efficiencies by automatically unifying all results under one analysis model.