Predictive Modeling on the Piezoelectric Properties of 3D Printed Functional Nanocomposites Using the Data Analytics Approach
This paper presents research done on prediction modeling using a data analytics approach to determine various factors affecting the piezoelectric properties of the 3D printed pressure sensors. Previously, the material extrusion 3D printing technique was used to fabricate pressure sensors composed of multiwall carbon nanotubes (CNT), barium titanate (BT), and polyvinylidene fluoride (PVDF) using simple fabrication and low-cost methods. This sensor produced a voltage output of 725 mV (0.13 pC/N) which is enough for pressure sensing applications. However, a holistic study to determine impacts of all factors was not carried out in the previous model. In this study, the design of experiments (DOE) approach was conducted on two set of variables of the fabrication process: material percentages and printing parameters. The weight percentage (wt.%) of BT and CNT in the PVDF matrix and the temperature of the printing bed and the extrusion nozzle were examined. Acquired data were analyzed through regression analysis and neural network (NN) to find the most significant factors as well as relevant interactions between the factors. It was found that (wt.%) of BT is the most significant factor to increase the piezoelectric properties among all parameters. Additionally, CNT nanoparticles introduced to the nanocomposite system as the stress reinforcing agent has substantially enhanced the piezoelectric properties by interacting with BT. It was observed that an increase in heating bed temperature has a negative impact on the piezoelectric properties. The extrusion nozzle temperatures did not have any individual impact. However, an increase in the extrusion temperature has a negative interaction with the BT for output response. This study also provides a prediction model to estimate sensor capabilities based on the combination of materials and printing parameters. Among the regression and the NN approaches, NN provides a model with higher prediction accuracy. ^
Industrial engineering|Mechanical engineering
Islam, Didarul, "Predictive Modeling on the Piezoelectric Properties of 3D Printed Functional Nanocomposites Using the Data Analytics Approach" (2018). ETD Collection for University of Texas, El Paso. AAI10930666.