A Numerically Scaled Spring-Friction System and Validation by Shaking Table Test

The seismic isolation efficiency of different friction-based devices needs verification by shaking table test, but faces problems in scaling before the test due to their frictional nonlinearity. To solve the scaling problems, a simplified civil structure, isolated by a self-centering spring-friction device, was numerically scaled in different ways considering the effect of friction action. The seismic responses of the scaled models were scaled back to those of the prototype and compared with the seismic responses of the prototype. The scaling problems and solutions were validated by a shaking table test on simply supported bridges using friction pendulum bearings (FPBs). The results show that both the unscaled gravity on a shaking table and the unscaled non-uniform friction distribution cause an inaccurate friction force in the structural motion equations of scaled models, and thus causing the scaling errors. One new and valid solution, i.e. changing the friction coefficient and scaling the non-uniform friction distribution to keep an accurate friction force for the scaled models, is put forward to avoid the scaling errors thoroughly. Another new solution shows that an increasing peak ground acceleration (PGA) can increase the other forces, while weakening the ratio of inaccurate friction force in the structural motion equations of the scaled models, which therefore reducing the scaling errors of acceleration and relative displacement responses, but not the scaling errors of residual displacement responses. In addition, the time-varying friction, the interface separation and collision of bearings, and other complex factors are found to cause scaling errors and need further investigation.