Design Optimization of Additively Manufactured Anisotropic Piezoelectric Lattice Structures by Gaussian Process Modeling

Piezoelectric materials have gained significant attention for numerous energy applications due to their ability to convert mechanical stress into an electrical response. Polyvinyl fluoride (PVDF) is a piezoelectric polymer known for its high flexibility and excellent piezoelectric properties, making it suitable for various fields including robotics, healthcare, and aerospace. However, conventional manufacturing methods have limitations in fabricating complex geometrical designs, leading to lower piezoelectric coefficients. Additive Manufacturing (AM) has emerged as an alternative for producing complex shapes with good mechanical properties. By leveraging AM, it becomes feasible to optimize designs and structures tailored to specific applications. Cellular structures represent a clear example of complex manufacturing designs achievable only through AM. Additionally, cellular structures offer a promising solution for optimizing the strength-to-weight ratio and increase directional piezoelectricity. This paper presents an optimization approach for gradient unit-cell of PVDF structures fabricated using fused deposition modeling (FDM). We propose a multiphysics finite element (FE) simulation to predict the output voltage response. Furthermore, we developed a Gaussian Process (GP)-based surrogate model using the simulation results as the training dataset with adaptive sampling techniques. The proposed surrogate model effectively predicts the output voltage of piezoelectric materials, enabling an optimum search over the design space, where we are aiming to minimize the volume while maintaining a high voltage output. The optimal results from the GP model were validated with experimental work, showing an accuracy above 90%.