Design, Fabrication and Control of A Novel Shape Memory Alloy (SMA)-Wire-Based Flexible Composite Actuator

Abstract

Soft actuators are frequently seen in many design areas. They can adapt to the desired motion and do not harm the material they touch. These features give many ideas to designers. In this thesis, such a soft composite actuator, a spiral, has been designed. It is a novel design with its geometry, functionality and actuation stroke; providing automatic movement in its gripper-type implementation.

In the process leading to the design, first of all, the phenomena about material behavior of the shape memory alloy (SMA), which is the main component and provides the actuation, are discussed. In this study, the SMA wire is used as an actuator. SMA materials exhibit thermomechanical properties. In other words, the mechanical actuation is provided by the heating of the wire. The phase transitions of the SMA material follow different curves during heating (from the martensite phase to the austenite phase) and cooling (from the austenite phase to the martensite phase) due to hysteresis behaviour. Therefore, the model is separately constructed for both states. Transition start and finish temperatures are needed to model the phase changes. Another feature of SMA materials is that the phase transition temperatures also change according to the load hanging from the wire. Therefore, in the study, the characterization of the wire is carried out.

A thermal model is the first for the constitutive equations modeling the thermomechanical property of SMAs. In this study, it is shown that the specific heat coefficient, latent heat coefficient and convection coefficient of the heat conduction model can be estimated by the recursive least squares (RLS) method. It is particularly important that convection coefficient is predictable and considers relatively complex mathematics in its determination, since the convection coefficient varies with the environment. Additionally, the specific heat and the latent heat coefficient are not always available from the material supplier. In the study, the functionality of the method is shown for two different wires. It is also understood that by adding the latent heat term to the heat transfer equation, there can be a 28\% improvement of the model for one wire and a 35\% improvement in the model for the other.

The control of SMA wire under axial loading is studied. Reference tracking is provided with the PI controller. The performance of the controller is separately tested for the regions where phase transition exists or does not exist and with regard to increasing reference frequency. In addition, it is shown that the controller exhibits a robust performance under varying load.

In this study, the material used while creating the soft composite structure is thermoplastic polyurethane (TPU). The main structure, the spiral's geometry is obtained with a 3D printer. If compared to harder equivalent composite production methods such as curing and molding, this method makes the production much easier.

The spiral exhibits an elastic effect because of its circumvoluted structure. With the adding of the flexibility of the TPU, the composite can be achieved to return to its initial position after the actuation. In other words, a two-way actuator is obtained by means of the novel geometry.

The spiral has curvilinear motion. Moreover, it can be predicted that the composite structure will produce a highly non-linear response when TPU flexibility and SMA hysteresis are combined. However, the control of the free end of the spiral is provided by PID controller. The feedback here is displacement data taken from the one direction in the plane.

A load-bearing gripper is presented as an implementation in which two spirals are used. In similar designs, the actuator system that grasps the load is moved upwards by the user. After the grip, a 6.5gr load can be automatically lifted up by 6.5mm with the double spiral application.

Accordingly, the heat transfer equation for the spiral takes the natural convection as the only energy loss. The equation gives results that match the measured temperatures, thanks to the parameters estimated by RLS. Therefore, firstly, it is inferred from the comparison of the model results with the measurements that the model is suitable for the spiral. Secondly, it is shown that the RLS method is effective when estimated parameters are taken into account.