Degree Name

Doctor of Philosophy


School of Electrical, Computer and Telecommunications Engineering


The integration of distributed generation (DG) units in distribution systems is gaining momentum due to their network support capabilities and modular designs. It is one of the effective and viable planning options for improving the supply quality and reliability of the network with ever increasing loads. It is expected that renewable DG systems may play a key role in future power distribution systems for sustainable and emission free energy supply. However, the incorporation of active DG units into conventional passive distribution systems poses difficulties to distribution system planners due to alternations in existing infrastructure and operational strategies. Moreover, with the integration of renewable DG units such as wind and solar photovoltaic based generation systems, system planners have to deal with the uncertainty in power availability under grid-connected and islanded modes of operation. This thesis aims to develop methodologies for optimal distribution system planning with distributed and renewable generation systems. Moreover, the distribution system reliability assessment has been also conducted with the integration of renewable DG units. In order to tackle stochastic variables in the optimisation formulation, the uncertainties associated with load demand and renewable power generation are modelled using probabilistic models. A clustering algorithm is proposed for conducting distribution system planning and reliability studies efficiently with uncertainty considerations.

In relation to distribution system planning considering renewable DG units, the theoretical analysis is carried out to investigate the impacts of active and reactive power injections (by DG units) on system voltage profile and energy losses. The analytical methods are also developed for finding the optimal operating power factor and optimal allocation of DG units considering maximum voltage support and maximum energy loss reduction, respectively. The main highlight of this research work is that the results obtained by the proposed analytical methods are in close agreement with the global optimum found by exhaustive search methods. Moreover, a new planning framework considering technical and economic aspects is proposed with incorporation of reactive capabilities of DG technologies, and uncertainties in renewable power generation and load demand. An integrated solution algorithm with intelligent based TRIBE particle swarm optimisation (TRIBE PSO) and statistical based ordinal optimisation (OO) is developed to determine optimal and near optimal solutions that provides multiple options to the system operator for comparing and deciding the feasible solution for practical implementation. The results emphasise the reduction in total cost with the consideration of DG reactive capabilities at the planning stage and effectiveness of proposed solution algorithm in terms of computational time and accuracy. Distribution system planning can be modelled as a multi-objective (MO) optimisation problem with different techno-economic considerations. A multi-objective TRIBE PSO is used to explore a set of trade-off options known as non-dominated solutions. A selection criteria based on Pareto dominance is applied to identify the most attractive solution for system planners.

For customer side reliability assessment in case of distribution networks with renewable DG units, a probabilistic based analytical approach, accounting time-varying load demand and stochastic power generation over the restoration period, is proposed. The results indicate that the proposed analytical method is capable of evaluating distribution system reliability indices comparable to those obtained by Monte Carlo simulation. Furthermore, reliability assessment of distribution systems with renewable DG units is also carried out with special emphasis on system uncertainties and optimal restoration strategies. The uncertainties associated with the total power output from renewable resources, time varying load demand, stochastic prediction errors, and random fault events are all considered in the restoration optimisation formulation for reliability evaluation. A dynamic encoding scheme is proposed in conjunction with TRIBE PSO method for ensuring optimality with less computational burden. The results indicate that the inclusion of time-varying load demand and generation in restoration optimisation improves distribution system reliability significantly.