This thesis reports on a novel approach to cable thermal field and ampacity computations using a newly proposed concept of perturbed finite-element analysis, which involves the use of derived sensitivity coefficients associated with various cable parameters of interest. It uses the sensitivity coefficients to achieve optimal cable performance. The proposed model provides a quick methodology, based on the finite element model, to assess the cable thermal performance subject to variations in the cable thermal circuit parameters. Furthermore, an optimization model for an underground power cable thermal circuit, based on generated gradients was developed, where subsequent utilization of the derived sensitivities as gradients of objective functions in a general framework of power cable performance optimization is presented. This comprehensive model uses the more accurate perturbed finite element method, which enables calculation of the objective function value and its gradients, without sacrificing the model accuracy. The algorithm developed was applied to various benchmark cable systems with their actual configurations, for different practical cable performance optimization objectives of interest to power utilities operators. The thermal field of an underground power cable sample directly buried in the soil was observed in the laboratory using a developed full size experimental setup. The investigation involves all parts of the thermal circuit parameters including cables composition, surrounding soil and boundaries phenomena. This experimental set was used to validate the developed simulation model by comparing the simulation results with the real laboratory measurements. Such experimental verification confirmed the accuracy of the newly introduced finite element sensitivity methodology. This thesis reports on a novel approach to cable thermal field and ampacity computations using a newly proposed concept of perturbed finite-element analysis, which involves the use of derived sensitivity coefficients associated with various cable parameters of interest. It uses the sensitivity coefficients to achieve optimal cable performance. The proposed model provides a quick methodology, based on the finite element model, to assess the cable thermal performance subject to variations in the cable thermal circuit parameters. Furthermore, an optimization model for an underground power cable thermal circuit, based on generated gradients was developed, where subsequent utilization of the derived sensitivities as gradients of objective functions in a general framework of power cable performance optimization is presented. This comprehensive model uses the more accurate perturbed finite element method, which enables calculation of the objective function value and its gradients, without sacrificing the model accuracy. The algorithm developed was applied to various benchmark cable systems with their actual configurations, for different practical cable performance optimization objectives of interest to power utilities operators. The thermal field of an underground power cable sample directly buried in the soil was observed in the laboratory using a developed full size experimental setup. The investigation involves all parts of the thermal circuit parameters including cables composition, surrounding soil and boundaries phenomena. This experimental set was used to validate the developed simulation model by comparing the simulation results with the real laboratory measurements. Such experimental verification confirmed the accuracy of the newly introduced finite element sensitivity methodology. Thesis Doctor of Engineering (DEng)