Conference Agenda

The trend towards higher power density motors and electronics has resulted in manufacturers looking for more efficient ways to cool their products during operation. Due to the size of these components and the need to keep them light weight, polymer solutions for heat dissipation have been growing in popularity. Polymers are already present in these designs due to their ability to protect from environmental hazards, function as an adhesive, and provide electrical insulation. These same polymers can be modified to be thermally conductive through the addition of various fillers. However, the effects of thermally conductive resins when used in a thin film are still unclear. In this study, the thermal conductivity of an epoxy-based system with traditional silicas vs high performance alumina-based fillers at various loading levels were measured. These thermally conductive resins were then coated as a thin film onto copper magnet wire windings using a dip method. An electrical current was then applied to the windings and the dissipation of heat was observed using an infrared camera. The data obtained from observing the heat removed from the windings when under load was used to determine the thin film effects of thermally conductive polymers.

In view of the great impact of sulfur hexafluoride (SF6) on global environment, the development of eco-friendly dielectric replacement gases has attracted considerable experimental and theoretical attentions. It is extremely difficult to identify an alternative gas that outperforms SF6 in all aspects because combining various mutually exclusive properties including dielectric strength, boiling point, global warming potential, arc interruption capability, toxicity, and flammability remains challenging. A variety of structure‐activity relationship models were developed to estimate the specific properties of the potential alternatives. However, these techniques either showed inadequate accuracy or employed the computationally intensive quantum chemical calculations. More reliable and efficient models are desired for the large-scale systematic virtual screening on the enormous number of all conceivable molecules. In this work, a valence-bond based structure‐activity relationship method is proposed for the first time to assess the dielectric performance from aforementioned six aspects by means of artificial neural network. From the mechanistic point of view, the dielectric-related physicochemical properties are determined predominantly by the electron-molecule and molecule-molecule interactions, which depend strongly on the inherent characters of the relevant chemical bonds as augmented by the bond-bond interactions. The chemical bonds are extracted straightforwardly from the simplified molecular input line entry specification (SMILES) and designated to be the descriptors for property prediction. Good correlations between theory and experiment have been obtained through the extensive machine-learning training on the well-established databases. The correlation coefficients for all the properties exceed 0.9. Using the present theoretical model, a multidimensional assessment on the performance of the potential alternative gas could be carried out efficiently with good accuracy in a consistent manner, shedding new light on the characteristic sought in SF6 replacement gases.