Semi-active dampers and actuators hold significant promise for their ability to add supplemental damping and reduce structural response, particularly under earthquake loading. However, to date, very little large-scale design, development or testing has been done with these devices, limiting the knowledge of what practical obstacles may stand between theory and successful implementation. In this research, a one fifth scale semi-active, resettable device is designed and tested to determine the efficacy of this controllable form of supplemental damping. Resettable devices are essentially non-linear spring elements that are able to reset their rest length actively, releasing stored energy before it is returned to the structure, thus creating a semi-active form of supplemental damping.

A novel device design that utilises each chamber independently allows more flexible control laws than previous resettable devices. It also enables better performance for large-scale devices and structural control testing, as it is better able to account for significant times to release stored energy than previous designs. More importantly, this approach allows the hysteretic behaviour of the structure to be actively modified by design and re-shaped to increase damping without increasing base shear forces, which is a potentially important advantage for retrofit applications.

The designed device characteristics, with air as the working fluid, are determined and a non-linear analytical model developed. The design stiffness is 250 kN m-¹, with the prototype having a stiffness of 185–236 kN m-¹. The peak force achieved by the prototype is in excess of 20 kN at a piston displacement of 33 mm. The model is experimentally validated and used to experimentally determine the effect of the actuator in a virtual structure through an iterative, hybrid form of dynamic testing, avoiding the need for full structure shake table testing at this stage of development. Hence, different semi-active control laws can be examined prior to physical testing using the experimentally validated model and the device.

Finally, manipulation of the force–displacement hysteresis curve via innovative control laws is demonstrated both experimentally and in simulation for three different control laws, focusing on different quadrants of the force–deflection hysteresis loop. The results for this form of stiffness-based supplemental damping are clearly evident in significant reductions of up to 60% in displacement and acceleration response spectra, particularly for periods of 0.5–2.0 seconds, which is the region of concern for earthquake resistant design. In addition, finite times to release energy relative to structural or ground motion dynamics are seen to limit performance and must therefore be accounted for in design. Overall, this research demonstrates that large-scale resettable devices can be implemented practically using very simple designs to deliver measurable supplemental damping and resistive forces, and the issues that must still be overcome are clearly delineated.