In this paper, a methodology for energy management system (EMS) based on the multi-objective retreat horizon optimization (MO-RHO) is presented to find the optimal scheduling of hybrid renewable energy system (HRES). Keywords: Multi-objective descending horizon optimization, Energy management system, Optimal scheduling, Hybrid renewable energy system. This justifies the application of energy management system (EMS) to meet the demand profiles, while optimizing the technical and economic performance related to operational scheduling of HRES.
23] proposed the energy management method for a grid-connected residential microgrid, which consists of solar cell, wind turbine and battery energy. 35] proposed an energy management system to optimize the planning of a solar/wind hybrid renewable energy system that uses batteries and fuel cells as a storage system. Wang et al [38] presented the receding horizon optimization method for energy management of HRES in the chlor-alkali chemical unit.
Demand-side energy management is also added to their proposed EMS to capture the effect of demand profile changes on optimal scheduling in [39]. A drawing horizon optimization framework based on the two-stage methodology for building energy management (BMS) has been developed by Gruber et al. In this paper, a new energy management system (EMS) based on descending horizon optimization is presented.
The first step for the development of the HRES energy management system is the analysis of the different performances of the subsystems based on the input data profiles.
Energy storage system
Where Ppv is the output power, G(t) is the total solar radiation per hour, ηPV is the PV efficiency and APV. Ah throughput and cycle counting are two lifetime consumption methods in PDM that are needed to calculate battery operating costs. In this study, the Ah-throughput counting method is used to determine the battery lifetime consumption [45].
Said method assumes that a certain amount of energy can be cycled in the battery storage system before the battery needs to be replaced. In this method, λL is the estimated lifetime throughput of the battery storage system and is almost determined by the DoD-cycles-to-failure (CTF) curve (provided by the manufacturer) and the maximum battery capacity (EBat,max). To obtain the operating cost of a battery bank, the battery usage cost per cycle ($/kWh/cycle) is required.
The battery wear cost (CBat,w) is calculated by dividing the battery life cost, which is defined by the sum of the replacement costs of the battery bank in the HRES lifetime (CBat,life) (Eq. 13)) by the net energy consumption of the battery bank (λL) as shown in Eq. CBat,life is the net present values (NPV) [46] of the battery replacement costs that occurred at each replacement time of the battery for the duration of HRES operation (20 years) [44].
Backup system
MULTI OBJECTIVE RECEDING HORIZON OPTIMIZATION
Operating costs consist of fuel costs and battery wear costs, which are the two main functions of operating objectives in moving horizon optimization. The moving horizon optimizer is applied at each time step during the HRES runtime. The amount of energy in the battery (EBat) is a state variable that is calculated at each time step and used as an initial variable for the next time step.
Dynamic constraints provide the temporal relationship between the state variables, while technical constraints define the feasible region of the optimization problem. The first values of the power trajectory vectors are implemented as the optimal solution for the time step (t). In the next time step, the input profiles and trajectories are updated, and the new initial conditions and profiles are implemented in the next moving horizon until the final time step number is reached.
Finally, the main result of descending horizon optimization is to determine optimal operation scheduling using available renewable resources based on desirable objective functions. Fuel cost is the conventional operating objective that affects the economic performance of HRES. In this study, the weighted global criterion method is used, which is one of the conventional methods for the multi-objective optimization problem.
For this purpose, the distances of each Pareto solution to the positive and negative ideal solution (di+ and di-) are calculated [30]. The receding horizon method optimizes the control variables at each time step with respect to deterministic multi-objective optimization. The optimal trajectories are obtained as a result in the associated time step and the first values of these trajectories are reported as the final optimal solution in the current time step.
U=[PDG, Pch, Pdis] and x=[EBat] are the control variable and state variable (feedback variable) vectors imposed in the receding horizon optimization model. N represents the length of the receding prediction horizon, i is the counter of time steps in the specific horizon, which is updated at each horizon, and y* is the actual size of the HRES components, which is determined using the installed experimental setup. The mixed integer constraints are also implemented in the optimization model to consider the physical constraint of the charge/discharge state of the battery bank.
Result and discussion
As can be seen, the optimal trajectory for six future hours is presented in time step 26, and its trend is generally followed by the trajectory corresponding to time step 27. The first values in each trajectory corresponding to each time step are implemented as the optimal power flow of DG , which represents the final solution for the optimization of the receding horizon during operation. It is worth emphasizing that by using receding horizons it is possible to capture load changes in future time steps (hour: 31) and generation profiles to estimate optimal power flows. 5-B) and (5-C) show the optimal battery charging/discharging power flows with two sample trajectories (time steps: 36 and 37 for charging and time steps: 41 and 42 for discharging).
The optimal initial values at each trajectory are reported as the optimal solution and the optimal trajectories of control and state variables at each time step are implemented as the initial condition for future receding optimization. As shown in this figure, increasing the length of the forecast horizon increases the total share of renewable energy (direct renewable + indirect renewable) in the load supply. When the longer horizon is considered, more insight into the future can be captured, and therefore also into the future.
As can be seen, diesel generator power flow is reduced from 41.1 kWh to 24.3 kWh throughout the week by increasing the length of the horizon from 6 to 12 hours. As previously indicated, increasing the length of forecast horizon leads to better management of battery scheduling and increases the share of renewable energy. Fig (8), shows the effect of length of forecast horizon on conflicting behavior of diesel fuel cost and the battery wear cost.
Recent research has proven that it is a trade-off between the choice of length of prediction horizon and CPU runtime [39]. As the length of the horizon increases, the corresponding optimal operating cost decreases, while at the same time the time for running the optimization problem increases. For the analysis of the effect of weather data profiles on the optimal schedule of HRES, the sensitivity analysis is performed with respect to weather seasonal variations.
For this purpose, the input weather data profiles are collected from a sample week in summer (Fig. 10)) and the results are compared with previous results related to sample week in winter. 11-A) and (11-B) show the optimal current flows regarding three control variables which are DG power generation and battery charge/discharge current flows for a trial week of winter and summer respectively. However, increasing the length of the predicted horizon in the summer week (Fig. 11-C)) increases the indirect share of renewable energy (battery discharge) and thus improves the economic one. Table (6) summarizes the results of seasonal sensitivity analysis and increasing the length of the predicted horizon on optimization results.
Conclusion
Optimization of an off-grid hybrid PV–Wind–Diesel system with different battery technologies using genetic algorithm. Long-term optimization based on PSO of a grid-connected renewable energy/battery/hydrogen hybrid system. Building optimization and control algorithms implemented in existing BEMS using a web-based energy management and control system.
An energy management strategy based on short-term generation planning for a renewable microgrid using a hydrogen storage system. Optimal energy management system for stand-alone wind turbine/photovoltaic/hydrogen/battery hybrid system with fuzzy logic based supervisory control. Improving the long-term performance of energy sources in off-grid hybrid systems based on renewable energy sources, hydrogen and batteries.
Real-Time Experimental Implementation of an Optimal Energy Management System in a Standalone Microgrid Using Multilayer Ant Colony Optimization. A new strategy for optimizing the operational plan of energy systems in the conditions of uncertainty of renewable energy sources and changes in demand. Optimal management of electrical and thermal energy of residential energy hub, integration of demand response and energy storage system.
A technology selection and operation (TSO) optimization model for distributed energy systems: Mathematical formulation and case study. Multicriteria decision making based on the TOPSIS method and similarity measures between intuitionistic fuzzy values. Real-time integration of optimal generation scheduling with MPC for the energy management of a renewable hydrogen-based microgrid.
A drawdown economic horizon optimization approach for energy management in the chloralkali process with hybrid renewable energy generation. Modeling and optimization of a power generation island based on renewable technologies and hydrogen storage systems. Multi-objective optimization of solid oxide fuel cell arrays considering parameter effects: Fuel utilization and hydrogen cost.
The optimal share of the energy supply side to meet the daily demand profile at two length horizons. The optimal accumulated share of the energy supply side to meet the weekly demand profile at two length horizons.
