FEM and ANN based simulations to study the effect of electrical field distribution on water tree affected polymeric cables

• Main Authors: Sulaiman S., Abidin I.Z.
• Other Authors: 36562570400
• Format: Conference paper
• Published: 2023
• Subjects: Cable insulation; Finite element methods and neural networks; Trees (insulation); Backpropagation; Bioelectric phenomena; Cables; Database systems; Error analysis; Finite element method; Materials science; Neural networks; Numerical analysis; Permittivity; Photoacoustic effect; Sulfate minerals; Trees (mathematics); Water analysis; Wire rope; Artificial neural network (ANNs); Back propagation (BP); Combined stresses; Electrical field distributions; Electrical fields; Finite element method (FEM); Finite element method (FEM) software; International (CO); Life spans; Optimisation; Permittivity values; Polymeric cables; Water treeing; Water-tree degradation; Waveforms; Electric network analysis

Earlier researches have confirmed that polymeric cables suffer from serious water tree degradation under combined stresses of electrical, thermal and chemical, thus shortening the lifespan of the cables. This paper aims at proposing a methodology using Finite Element Method (FEM) and Artificial Neural Network (ANN) based simulations to estimate the effect and create a database of the electrical field distribution in water tree affected polymeric cables. The water tree lengths in the affected polymeric cables are varied to assess their effect on the level of the electrical field distribution. The permittivity values within the water tree structures are varied accordingly to the lengths measured in water tree growths. The electrical field distribution waveforms arising out of the variation in water tree growths are plotted using FEM software and the results are analysed. Analysis shows that the electrical field distribution levels increases exponentially with the rise in water tree lengths. The results are then translated into electrical field ratios and fed to an ANN simulation. The simulation attempts to create a database for water tree degraded cables with two (2) categories of training inputs i.e. permittivity and water tree length, and a target output of electrical field ratio. The error in the ANN simulation is then marginalised to an error value of less than 1% using the back-propagation method for optimisation of weights. � 2006 IEEE.