•Experimental test of a novel bracing system equipped with SMA bars is presented.•Seismic performance of a building equipped with such bracing system is evaluated.•The Closed-Loop Dynamic testing method is proposed to perform seismic test.•Very good agreement between the simulation and the experimental results was achieved.•The performance of the system under strain rate effect showed very good response.•A building with such bracing showed self-centering behavior under seismic load. This study investigates the seismic performance of a newly developed self-centering bracing system using a novel experimental technique named as closed-loop dynamic (CLD) testing. The bracing, named piston-based self-centering (PBSC) apparatus, employs Ni-Ti superelastic shape memory alloy (SMA) bars inside a sleeve-piston assembly for its self-centering mechanism. During cyclic tension-compression loading, the SMA bars are only subjected to tension avoiding buckling and leading to flag-shaped symmetric force-deformation hysteresis. Initially, a braced frame building fitted with PBSC is seismically designed and the preliminary sizing of the brace is determined. For testing, considering the lab capability, the brace is fabricated at a reduced scale. The process of “Closed-loop dynamic testing” starts with the brace test (step 1) under strain-rate loading to characterize the numerical model parameters (step 2), which are then scaled-up as per similitude law and implemented in a finite element software, S-FRAME’s PBSC brace model (step 3). Then the braced frame building is analyzed under an earthquake (step 4) and the axial force-deformation response of the brace under consideration is captured (step 5). In order to further understand and validate the actual response of the brace under earthquake type loading, the axial deformation obtained from S-FRAME is scaled-down (step 6) and used as input parameters for testing the reduced scale brace (step 7). The obtained response (step 8) is further scaled-up and used to match the S-FRAME’s PBSC model for validation (step 9). Iterations from step 3 to step 9 will be required until the experimental and numerical results converge. Convergence criteria used for this validation include both the energy dissipation capacity and initial stiffness within 10% accuracy. Reasonable agreement between the numerical and experimental results is achieved in the closed-loop dynamic testing. The PBSC brace shows excellent self-centering capability under various earthquake loadings.