Allocation of dump load in islanded microgrid using the mixed-integer distributed ant colony optimization with robust backward\forward sweep load flow

Abstract

Reliable planning and operation of droop-controlled islanded microgrids (DCIMGs) is fundamental to expand microgrids (MGs) scalability and maximize renewable energy potential. Employing dump loads (DLs) is a promising solution to absorb excess generation during off-peak hours while keeping voltage and frequency within acceptable limits to meet international standards. Considering wind power and demand forecast uncertainties in DCIMG during off-peak hours, the allocation of DL problem was modelled as two problems, viz., deterministic and stochastic. The former problem was tackled using four highly probable deterministic generation and demand mismatch scenarios, while the latter problem was formulated within scenario based stochastic framework for uncertainty modelling. The mixed-integer distributed ant colony optimization (MIDACO) was introduced as a novel application in microgrids to find the optimal location and size of DL as well as the optimal droop setting for distributed generation (DG). Furthermore, to enhance the convergence of the proposed optimization technique, three robust and derivative free load flow methods were developed as novel extensions of the original backward\forward sweep (BFS) for grid-connected MGs. The three load flow methods are called special BFS, improved special BFS, and general BFS. The first two methods rely on one global voltage variable distributed among all DGs, while the latter has more general approach by adopting local voltage at each generating bus. The deterministic multi-objective optimization problem was formulated to minimize voltage and frequency deviation as well as power losses. Inversely, the stochastic multi-objective problem with uncertainty was formulated to minimize total microgrid cost, maximum voltage error, frequency deviation, and total energy loss. The proposed method was applied to the IEEE 33-, 69-, and 118-test systems as modelled in MATLAB environment and further validated against competitive swarm and evolutionary metaheuristics. Various convergence tests were considered to demonstrate the efficacy of the proposed load flow methods with MIDACO’s non-dominated solution. Likewise, different optimization parameters were utilized to investigate their impact on the solution. Moreover, the advantage of multi-objective optimization against single objective was provided for the deterministic optimization problem, while the effect of load model and droop response were also investigated. The obtained results in chapter 5 and 6 further demonstrate the fundamental role of DL in voltage and frequency regulation while minimizing costs and energy losses associated with DCIMG operation. Accordingly, an improved voltage and frequency profiles for the system after DL inclusion were attained in Figure 6.9 and Figure 6.10, respectively. To demonstrate the competitiveness of DL-based energy management system (EMS) against storage-based EMS, a brief cost benefit analysis considering hot water demand was also provided.