396 / 2022-03-15 19:59:21

Determining accelerated aging power cable spatial temperature profiles using Artificial Neural Networks

accelerated aging,power cable,finite element model,Artificial neural network,Temperature profile

Purpose

IEEE Standard 1407 outlines a methodology of accelerated aging experiments for Medium-Voltage (MV) extruded power cables using a water filled tank with no forced convection. Whilst aging, it is important to understanding the temperature distribution profile across the cable insulation to quantify the aging rate of the cables.

Modeling methods

In this study, a cable loop accelerated aging simulation model in a water filled tank is undertaken using Finite Element Analysis (FEA) where simulation of the temperature profile across the cable insulation is determined. The temperature at specific meshed points of the model can be extracted from the simulation results. To provide a spatial temperature profile prediction tool from the meshed results for any chosen point within the cables, a 10 hidden layer Artificial Neural Network (ANN) trained by a Bayesian regularization (trainbr) function is constructed.

Results

Figure 1 shows an example of the simulation temperature results of an accelerated aging power cable loop in a water filled tank. In the simulation model, there are four layers of cable structure including conductor, insulation, metal shield and outer sheath. Nature water convection is considered in the simulation. For a uniform cable core current density, the temperature is not uniformly distributed in both the water tank and within the cable length and cable cross section.

Based on the Arrhenius function, temperature has an essential impact on the power cable aging rate. To enable understanding of the precise impact of temperature at different locations then accurate estimation of the temperature profile at any point in the power cable can be predicted for future overall cable aging impact calculations. Figure 2 represents the predicted temperature profile in cylindrical-coordinate system.

Conclusions

It is demonstrated that the application of a neural network can effectively determine the temperature profile at all positions of a power cable within the simulation of an artificially accelerated water tank environment. This helps understand how different insulation positions will undergo different temperature exposures across the cable length.

Appendix