Coordinated control of coastal multi-source multi-load system with desalination load: a review

1 Introduction

One of the most important aspects of energy internet (EI) is to attain large-scale utilization and sharing of distributed renewable energy sources (RESs) [1].Energy storage systems (ESSs) and controllable loads (CLs) are essential to stabilize the intermittent nature of RESs.Therefore, the coordinated control and flexible interaction of the “sourceload-storage” system after the distributed equipment connection is one of the most important research directions of the EI [2].Studies in this field focus on the methods to use the bi-directional interaction of EI information to introduce users’ concentrated demand responses into the virtual power plant dispatch operations, thereby reducing system operating costs, forced load losses, and additional losses of RES.In recent years, the related investigations on the intelligent power consumption have focused onmultiple independent techniques; these include advanced measurement systems [3-5], demand responses, and integrated technology practices such as intelligent power communities and interactive business halls.In terms of EI coordinated control, in foreign countries with large proportion of wind power grids, such as Denmark and Germany, under the guidance of real-time electricity tariff, many thermal power units have considered the configuration of heat storage devices to increase the operational flexibility and peaking capabilities in order to profit from the electricity markets.Especially in Denmark, thermal power plants participate in system peak shaving through heat storage to achieve flexible operation, which has become an important means to achieve a 100% renewable energy system in future [6].

An energy storage device functions as both a load and an energy supply system and is an indispensable part of the EI [7-9].Reference [9] demonstrated a variety of energy storage methods to improve the efficiency of renewable energy use.Reference [10] emphasized that hydrogen storage is an important energy storage method for the future, and energy storage equipment owned by data centers and electric vehicles could be used as energy storage units.Reference [11] proposed a hybrid electric energy storage device architecture that relies on information technology, and is also one of the future development directions of energy storage devices.

Multi-energy complementary consumption and efficient energy use have become an important part of achieving sustainable development of coastal economy.Owing to the increasing scarcity of fresh water resources, seawater desalination has emerged as one of the methodologies to meet the fresh water demand, especially around the coastal areas.The technique ensures freshwater supply and sustainable use of water resources [13-14].Coastal desalination, coastal aquaculture, and other related industries are developing rapidly.Consequently, their load demand is becoming increasingly intense, especially in areas lacking fresh water resources.Since the 1950s, seawater desalination technology has been widely applied in the Middle East; it developed and spread rapidly in several countries and regions outside the Middle East, becoming the world’s solution to water shortage [12].Coastal areas that are rich in RESs, have a great potential for development in terms of effective distribution of renewable energy, such as clean and pollution-free solar, wind, and tidal energy [15].Considering various loads (such as those required for desalination) to achieve time-domain matching and complementary utilization of coastal distributed renewable energy for large-scale coastal promotion of desalination and other industries, solving water resource problems, and eliminating distributed renewable energy, increasing and balancing the grid loads are of great significance.Moreover, this can lead to enormous economic and social benefits, promoting harmonious development of people and the environment [16].

In recent years, the combination of wind power generation [17], solar power generation [17-19] and seawater desalination system [20] has received increasing attention.Reference [21] proposed a new optimization algorithm combining harmony search and chaotic search.According to the load requirements of seawater desalination system, the installed capacity of each power generation system is optimized to obtain the optimal cost-effectiveness.This optimization method is based on a large amount of meteorological data; thus, it is difficult to adapt to the changing natural conditions.Reference [22] proposed the concept of “source-grid-load” interaction and emphasized that the balance capability of energy generation and consumption can be improved through their interaction.

This paper mainly reviews the coordinated control of coastal multi-source multi-load system and the flexible interaction between seawater desalination load and regional distribution networks.Section 2 introduces the coordinated control of coastal multi-source multi-load system with seawater desalination load.Next, Section 3 describes the flexible interaction between seawater desalination and regional distribution networks.Section 4 includes the joint regulation of coastal multi-source multi-load storage system with seawater desalination load.Finally, Section 5 presents the conclusions of this paper.

2 Coordinated control of coastal multisource multi-load system with seawater desalination load

2.1 Coastal load forecasting method with seawater desalination load

The objective of seawater desalination is to produce fresh water.At present, reverse osmosis (RO) seawater desalination technology is globally predominant with its advantages of low engineering cost, low energy consumption, and simple operation.Therefore, herein, the load prediction of RO seawater desalination load is focused on [23].Methodology of the RO seawater desalination is shown in Fig.1.

The desalination device is mainly composed of a feed water pump, high pressure pump, RO unit, energy recovery device and reservoir (Fig.2); the high-pressure water pump is the main energy consuming unit [24].

Fig.1 Flow chart of the RO seawater desalination process

Fig.2 Schematic representation of a seawater desalination device

Several studies have focused on the intrinsic operation of seawater desalination devices.Reference [25] summarized the municipal applications of RO desalinated water in the Gulf countries, the United States, and China.Reference [26] analyzed the health impacts of the water produced via seawater desalination by RO to fully understand the quality of the as produced water.Reference [27] proposed two power coordinated control strategies: in grid-connected and isolated operation modes; the adjustable characteristics of desalination loads were applied to stabilize the battery DC voltage.

Taking seawater desalination load forecasting into account, following issues should be addressed: key factors affecting the coastal load and related influencing characteristics, important levels of each factor, and classification of these factors.Based on typical daily calculation of comprehensive forecasting weights, a load forecasting model should be established, and predictive load data obtained.Fig.3 shows the research process of the coastal load forecasting method.

Fig.3 Research process of the coastal load forecasting method

A method based on morphological clustering and LightGBM [28] for coastal load forecasting is proposed for typical coastal load characteristics.

(1) Morphological clustering via Industrial User Morphological Clustering (IUMC).

(2) Data preprocessing: For coastal load data, Grey Relational Analysis (GRA) method is used to select similar day.The coastal load characteristic quantity data is divided into historical data and data to be predicted.Degree of association between the feature quantity sequence of the historical day and the feature quantity sequence of the day to be predicted is calculated.

where X0 is the feature quantity sequence of the day to be predicted, Xm is historical day feature quantity, t is the number of coastal load characteristics, and m = 1,2,… n is the day for which the historical data is obtained.

First, the difference sequence between the historical day feature quantity sequence and the to-be-predicted daily feature quantity sequence is obtained.

where Δ is difference sequence.

Then, the maximum Δmax and the minimum Δmin of difference sequence is obtained.

where n is the total number of days, and t is the number of coastal load characteristics.

Finally, the gray correlation degree γ of the feature quantity sequence of the day to be predicted and the feature quantity sequence of the historical day is calculated.

where k is the k-th costal load characteristic, and ρ is resolution factor with a value of 0.5.

(3) Calculation of the cosine similarity coefficient of coastal load data;

(4) Calculation of the optimal validity index of morphological clustering.

RI is intraclass correlation index, and the intraclass correlation formula is

where g is the total number of classes, xi is the coastal load characteristic quantity of the f-class, cf is the morphological clustering center of the f-class, and nG is the coastal load characteristic quantity of the f-class.

RO is inter-class correlation index, and the correlation formula between classes is

where cij is the correlation coefficient between the morphological clustering center i and j.

The cluster validity indicator formula is

(5) Training and prediction of the various types of load data after morphological clustering using their corresponding LightGBM models.

2.2 Coordinated control of coastal multi-source multi-load system and multi-time scale energy scheduling method with seawater desalination load

The coastal multi-source multi-load system structure with desalination load mainly includes RESs (wind power generation, photovoltaic power generation, and others), energy storage systems, diesel generator sets, seawater desalination loads, and conventional loads [29].

Reference [30] proposed a multi-time scale coordinated automatic power dispatching mode by analyzing the characteristics of the predicting accuracy of wind power that increases level by level with different time scales and the inherent features of active power dispatch.Reference [31] presented a multi-timescale coordination method for ACE -advanced control, which can consider the different response speeds of various generators.Reference [32] discussed a real-time active power dispatching system, that considers two functions: spot balancing trading and real-time security management.Reference [33] presented a microgrid-seawater desalination system with traditional diesel power generation and energy storage systems, wherein the V/f control method was used to distribute the output power of the microgrid to provide stable voltage and frequency support for the desalination system.In their study, the operating cost of the system reduced to a certain extent; however, the controllability of the load of the desalination system was not considered.

Multi-time scale issues have been included in several studies to date.However, few studies have considered the multi-time scale energy scheduling in conjunction with seawater desalination.By collecting the energy information of RESs, state of charge and discharge of the energy storage device, and coastal load data as inputs, the scheduling time is established at the level of days, hours, and minutes.An energy scheduling control strategy with control scale and load matching, control domain optimization grouping, unit autonomous control timing, and logic optimization functions is established.The research process of coordinated control of coastal multi-source multi-load system and the multi-time scale energy scheduling method with seawater desalination load are shown in Fig.4.

In the time dimension, the active scheduling strategy of the coastal multi-source multi-load coordination system with seawater desalination is decomposed into minute, hour, and day [34].As the time scale increases, the impact of uncertainties in the demand for RESs and wind/solar power output forecasting also increases, with concomitant decrease in the accuracy of prediction.Therefore, the daily plan cannot meet the actual energy balance needs.The energy flow of each unit in the remaining period is cyclically corrected to maximize the energy utilization of the system.As the load participates in the demand response in different ways, and the demand response resources are widely distributed, dispersed, and small, it is necessary to participate in energy scheduling by using a schedulingstrategy with control domain allocation, as direct energy control is not feasible.

Fig.4 Research process of coordinated control of coastal multi-source multi-load system and multi-time scale energy scheduling method

In the process of studying the control strategy, the following objective function can be considered.The goal is to minimize the overall cost of system operation, including micro-power operation and environmental governance costs.

where Crun is the overall cost for the system operation, Cm is the operating cost for the micro-power supply, and Ce is the environmental treatment cost for the diesel generator operating cycle.

The micro-power operation cost refers to the cost incurred by the micro-power unit during the power generation process.where i is the type of distributed power supply, Cm,i is running unit cost for distributed power of island microgrid, Pi(t) is output power for distributed power supply at t-hour, and T is system operation cycle.

The environmental governance cost model not only enhances the advantages of renewable energy output, but also reduces the pollution gas emissions of diesel generators.

where n is the type of pollutant gas, αn is unit cost for pollution gas treatment of diesel generators, nφ is pollution gas emission coefficient, and Fdg(t) is the t-hour diesel consumption of diesel generators.

3 Flexible interaction between seawater desalination and regional distribution network

3.1 Connotation of “source-grid-load” system interaction

“Source-grid-load” interaction refers to multiple interactions between power supply, distribution network, and load to achieve more efficient and safe operation of power systems.“Source-grid-load” interaction is essentially an operational mode that maximizes the use of energy resources [35].The traditional power system operation control mode is adjusted by the power supply tracking load change, and no obvious interactive relationship has been formed.In future, power supply, power grid, and load will have flexible features, resulting in a comprehensive “source-grid-load” interaction with multiple interaction modes such as source-source complementation, sourcegrid coordination, grid-load interaction and source-load interaction [35], as shown in Fig.5.

Fig.5 Connotation of “source-grid-load” system interactions

“Source-source complementary” involves complementing the output timing and frequency characteristics of different power sources to promote the absorption of renewable energy.“Source-grid coordination” means that the grid operation control is based on comprehensive application of load forecasting, unit combination, daily planning, online scheduling and realtime control at different time scales to achieve coordinated control of power supply and power grid.“Grid-load interaction” refers to changing the distribution of power flow through planned and active adjustment of load when the grid appears or is about to occur, ensuring safe andreliable operation of the grid.“Source-load interaction” means that both power supply and load can participate in power supply and demand balance control as schedulable resources.The flexible change of load becomes one of the important means to balance power supply fluctuation.

“Source-grid-load” interaction has been validated in several example systems as well as real pilot projects [36].Technical issues, such as demand response and power balancing, have been discussed.Reference [37] developed a multi-objective robust optimization model of power systems considering demand response.Reference [38] discussed the optimization of load control and smart demand response mechanism for system services.Reference [39] evaluated voltage quality for urban power network considering vehicle to grid (V2G), which developed dynamic probability models of basic load and electric vehicle’s power.

Based on the interaction characteristics and interaction behavior of the “source-grid-load”, an interactive model should be established first.Considering the interactive ability of the “source-grid-load”, complementary features, and the time difference identification model, the “sourcegrid-load” operation scheme in flexible interactive environment has been proposed [40].Based on the flexible interaction ability, the probability of safe operation and the level of economic efficiency, the effectiveness evaluation method of the flexible interaction model is developed, as shown in Fig.6.

Fig.6 Flexible interaction between seawater desalination load and regional distribution network

3.2 Multi-source multi-load system and regional distribution network interaction behavior modeling method

Modeling the interactive behavior of a multi-source multi-load system and regional distribution network requires consideration of the stochastic factors and models at the multi-temporal and multi-spatial scale for hybrid interactive systems; such modeling can assist subsequent analysis and control of the multi-source multi-load system.In coastal “source-network-load” interactions with seawater desalination loads, the responses of the interacting subjects to the interaction vary with temporal scale.There are fast processes and slow processes.The spatial scales of also differ among the interactive subjects.For example, the widearea complementarity of renewable energy is much higher than that between energy storage devices and renewable energy sources, and that between intermittent power supply and pumped storage is higher than that between power and load, on a particular time scale.Therefore, there is a need to propose different ways of obtaining the patterns, content, and effects of interactive behavior.Considering the complexity of modeling, it is necessary to study modeling methods suitable for different types of interaction behaviors, such as hybrid modeling methods [41], random probability modeling methods [42] and adaptive hierarchical modeling methods.

Reference [43] discussed the principle of interaction between a large power system and a micro-grid, optimal planning, operation strategy, relay protection, and pattern of electrical market.Reference [44] developed a control strategy and a transient stability analysis model for three typical flexible AC transmission system devices.Reference [45] developed a demand response algorithm, called the weighting coefficient queuing algorithm, by modifying the colored power algorithm and state-queuing model to improve RES utilization.

4 Combined control of coastal multi-source multi-load storage system with seawater desalination load

4.1 Energy transmission and information interaction in a “source-load-storage” system

In terms of energy transmission in a “source-loadstorage” system, the EI can significantly improve the adaptive capacity of the power grid and realize greater flexibility and intelligence of the multi-energy network access ports.This, and flexible access ports, energy routers, multi-directional energy automatic configuration technology, and energy carrying information technology, reduces the possibility of conflicts and the possibility of blocking multi-energy cross-flow in the network.In the event of a system failure, the EI can accelerate the rapid reconstruction of the network and re-adjust the distributionand direction of energy flow.

Reference [46] developed a bi-level-programming- based model combining planning and operational issues at different time scales to determine the optimal installation sites and capacities of ESSs.Reference [47] proposed a planning model that includes thermal power, wind power, and energy storage plants and employs a strategy that allows the fluctuation of wind power within a certain range to reduce investment and operation costs in energy storage plants.Reference [48] proposed a universal mathematical description of the power system multiple time scale flexibility “supply/demand” balance and an evaluation index to assist in determining the optimal flexibility for complex “Generation-Grid-Load-Storage” scenarios.However, few studies have considered a “source-load-storage” system with seawater desalination loads.

4.2 Combined control of coastal multi-source multi-load storage system

Considering the multiple constraints on a coastal multi-source multi-load storage system with seawater desalination load, such as the power balance constraint, energy storage capacity constraint, distributed power output constraint, interactive standby capacity constraint, and grid output constraint, a joint control optimization model was established.The energy and economic environment costs and their utilization rates were optimized to achieve “source-load-storage” joint regulation.

Reference [49] developed a multi-objective optimal placement model for energy storage systems in an active distribution network with the three objectives of peak shaving capacity, power self-regulation capacity, and voltage quality.Reference [50] designed a correction method for the power injection probability model for nodes considering changes in the model that included wind power and ESS.Reference [51] proposed a multi-stage joint planning model with the objective of minimizing the total investment cost, operation cost, and load loss cost, considering the integration of electric vehicle charging loads and distributed ESSs.

During the peak and valley load periods, it is necessary to consider constraints such as power balance, energy storage capacity, distributed energy output, interactive standby capacity, and grid output for energy transmission from and information interaction in the “source-loadstorage” system in order to maximize energy utilization and minimize economic and environmental costs.The research process corresponding to the combined control of a coastal multi-source multi-load storage system with seawater desalination load is shown in Fig.7.

Fig.7 Research process associated with combined control of a coastal multi-source multi-load storage system with seawater desalination load

5 Conclusion

Taking into account the EI technology and the potential of distributed RESs, such as coastal solar energy and wind energy, the combined coordinated control seawater desalination loads, various conventional loads, and distributed RESs could be obtained.Distributed RESs and distributed CLs are located over large areas in coastal spaces and require flexible interaction mechanisms for interaction with grid connection points and grid energy storage systems to ensure real-time energy transfer.To ensure the quality and efficiency of use of desalinated seawater, this study considered the seawater desalination incremental load as the base load and made full use of the characteristics of ESS technology.Through ESS capacity adjustment and CL regulation, coastal CL and ESS could be adjusted.The grid load could be balanced and increased, and the safety and economy of the coastal power grid could be improved.The coordinated and comprehensive application of power grids, renewable energy, and coastal loads, the comprehensive development of coastal areas, and the requirements for the consumption of RESs can therefore be realized.

Acknowledgements

This work was supported by the State Grid Science and Technology Project, “Study on Multi-source and Multiload Coordination and Optimization Technology Considering Desalination of Sea Water” (No.SGTJDK00DWJS1800011).