Nanogrid sizing using nested integer linear programming and time-of-use based load management

Abstract

Electrical utility services are evolving from centralized conventional systems to distributed grids (DGs) attributing to clean energy production, customer participation and low energy cost. Integration of renewable energy (RE) systems into existing grids results in complex grid structure which requires optimization methods in planning and operational schemes. In RE system planning, capacity sizing and component placements are typically implemented using classical methods, application software and intelligent-based methods. The software-based methods are static, hence, cannot be tuned to a customized application. Whereas, intelligent-based methods produce results that are acceptable, however, not optimal. Linear programming (LP) based algorithms as classical methods are preferred due to its simplicity, speed and accuracy which yields global optimal results without branching at local solutions. The mixed integer linear programming (MILP) is used in microgrid’s components sizing. However, MILP has limitations of large formulations, high computational burdens and hardly consider multi-objective analysis. To overcome the MILP problems, nested integer linear programming (NILP) is proposed in this study to implement a multi-configurational sizing in residential nanogrid to achieve low energy cost. A residential located in sub-Saharan semiarid climates of northern Nigeria is chosen as a case study. The proposed NILP is implemented in a multi-stage hybridization of relaxation LP and MILP in a nested loop for nanogrid configurations using photovoltaic (PV), wind turbine (WT) and battery energy storage system (BESS). Effectiveness of the NILP is verified by comparison with the classical MILP and particle swarm optimization (PSO). Operation schemes in RE systems include power dispatch and demand side management (DSM). The DSM is preferred as it allows more options for customer participation and can simply follow supplies. DSM is implemented using the conventional time-of-use (CTOU) methods. However, the CTOU is time-bound, utility-centred, incur additional energy costs and affects customer comforts. To balance the conflicting objectives of energy cost and customer comfort, the time-of-use fitness (TOUF) which is an improved version of CTOU has been proposed. The method is introduced to achieve load management for the nanogrid’s optimal energy utilization and to reduce consumption cost. The proposed TOUF considered local RE supplies, BESS, grid interaction and customer demands based on a fitness function (Ffunction). The Ffunction is a demand response initiative used alternately for energy based on real-time energy cost to define a fitness costs (Fcost) as the energy consumption cost. Both the sizing and load management schemes are implemented using MATLAB programming. The NILP achieved reductions in nanogrid’s capacity, the levelized cost of energy (LCOE), and net present costs (NPC) as compared to the MILP. The PV/WT hybrid nanogrid configuration achieves NPC and LCOE reductions by 11% and 33% compared to MILP and PSO, respectively. The TOUF achieved up to 43.40% and 53.09% Fcost reductions under the BESS support. The autonomous nanogrid operations were analysed using the Markov Chains as a stochastic tool. The probabilistic information indicates thAat the proposed nanogrid is able to achieve up to 61.54% autonomy in a 25-year lifetime analysis.