α-Operator splitting time integration technique for pseudodynamic testing error propagation analysis

Abstract

The Pseudodynamic (PSD) test method combines the numerical integration of the motion equations of a structure condensed at selected degrees-of-freedom (d.o.f.) with the on-line measurement of the reaction forces needed to impose this computed motion. The results obtained with this testing method can be strongly corrupted by the spurious propagation of experimental errors. The α-operator splitting (α-OS) scheme, a convenient linearisation of the so-called α-method, is a non-iterative implicit time integration scheme. As such, it is a good candidate for PSD testing since it provides the unconditional stability sometimes needed when the number of d.o.f. is large, while preserving the implementation simplicity of explicit schemes. The properties of the α-OS scheme are presented with special emphasis on its error propagation characteristics. These properties are compared with the ones of corresponding iterative implicit schemes. It is shown in particular that, in contrast with what is observed using iterative schemes, the selective damping possibility associated with the α factor become a crucial ingredient for the success of the α-OS scheme.