Coupled vibration control (CVC) systems, which typically consist of two frames and connectors, such as stiffness and damping elements, can be used for the seismic protection of coupled or adjacent structures. This study aims to experimentally and numerically elucidate the response behavior and control effects of CVC structural systems comprising a mainframe and subframe incorporating passive negative stiffness (PNS) at the connection subjected to earthquakes. Shake table experiments were conducted using CVC system specimens incorporating a PNS device (PNSD) consisting of curved-leaf springs under dynamic excitations. A significant seismic response reduction for the mainframe of the CVC specimen under simulated waves was achieved by incorporating the PNSD at the connection. Moreover, a response simulation was performed using six-story CVC structure models connected by negative stiffness and viscous damping elements subjected to simulated and observed seismic wave inputs. The displacement and acceleration peak response of the mainframe using the numerical CVC models was lower than that of the uncontrolled model. Therefore, incorporating negative stiffness and viscous damping connectors into CVC systems can reduce the seismic response. This study contributes to the advancement of aseismic technologies for coupled or adjacent buildings utilizing the PNS.
Coupled vibration control (CVC) structural systems typically consist of two frames and connectors, such as stiffness and damping elements. CVC systems with a proper connecting mechanism that depends on the structural properties of the two frames can be a solution for protecting coupled or adjacent structures against earthquakes , , , , , , , , , , , , , , , , . To date, various CVC structures incorporating various connecting devices, including steel hysteresis dampers , , fluid dampers , , magnetorheological (MR) dampers , , , , tuned mass damper inerters , , viscous inertial mass dampers , and friction dampers , have been proposed and investigated to mitigate the response when subjected to seismic input motions. Passive , , , , , semi-active , , , , and active ,  control methods for CVC systems have been studied theoretically, numerically, and experimentally, and the results have proved a potential effectiveness for the seismic protection of buildings. However, employing only zero or positive stiffness at the connectors of CVC systems to achieve an optimal tuning and obtain a better control performance may be difficult depending on the properties of the two frames .
In contrast to positive stiffness, negative stiffness generally generates a decreased (negative) restoration force as the displacement increases. The use of negative stiffness can be considered a potential response control strategy for buildings against dynamic disturbances, such as earthquakes. During the last few decades, many studies have been conducted to explore negative stiffness devices (NSDs) and structural systems containing NSDs , , , , . Iemura et al.  used the self-weight of the structure and developed a negative-stiffness damper based on a slide bearing with a convex curve. Nagarajaiah et al.  and Sarlis et al.  studied a passive NSD based on a vertical preloaded coil spring placed between two chevron braces and a horizontal-gap spring assembly. Wang et al.  investigated a structural system with a negative-stiffness amplifying damper consisting of a passive NSD , , positive-stiffness spring, and dashpot. Shirai et al.  proposed another passive NSD comprising curved-leaf springs. This passive NSD exhibited an initial negative stiffness created by the pre-compressed strain energy stored in the curved-leaf springs, followed by an increased second negative stiffness with the occurrence of snap-through buckling. However, these previous studies on NSDs explored the seismic response of individual, that is, nonadjacent or noncoupled, structural systems with NSDs.
This study experimentally and numerically investigated the response behavior and control effects of CVC structural systems with a PNS spring as a connector subjected to dynamic excitations. Shake table experiments were conducted using CVC system specimens incorporating PNSDs consisting of curved-leaf springs. In addition, time-history earthquake response simulations were performed using six-story CVC building models connected by negative stiffness and viscous damping elements. The following conclusions were drawn:
(1) In the shake table experiments, PNSD-1 and PNSD-2 exhibited an initial negative stiffness owing to the pre-compression effect, and the three PNSDs exhibited an increased second negative stiffness owing to the onset of snap-through buckling in the subtracted hysteresis loops of the PNSDs.
(2) An effective response control performance for the mainframe under seismic excitations was attained by incorporating a PNSD as the connector of the CVC specimen.
(3) The CVC specimen with PNSD-2 exhibited the most effective control performance among the three PNSDs for mitigating the response displacement and acceleration under the simulated wave input.
(4) In the numerical simulations, both the CVC-NV and CVC-V models exhibited a reduced mainframe peak response displacement and acceleration compared to that of the uncontrolled model (UC-MF).