Review of a Comprehensive Analysis of Planning, Functionality, Control, and Protection for Direct Current Microgrids

Abstract

Microgrids have emerged as a feasible solution for consumers, comprising Distributed Energy Resources (DERs) and local loads within a smaller geographical area. They are capable of operating either autonomously or in coordination with the main power grid. As compared to Alternating Current (AC) microgrid, Direct Current (DC) microgrid helps with grid modernisation, which enhances the integration of Distributed and Renewable energy sources, which promotes energy efficiency and reduces losses. The integration of energy storage systems (ESS) into microgrids has garnered significant attention due to the capability of ESS to store energy during periods of low demand and then provide it during periods of high demand. This research includes planning, operation, control, and protection of the DC microgrid. At the beginning of the chapter, a quick explanation of DC microgrids and their advantages over AC microgrids is provided, along with a thorough evaluation of the various concerns, control techniques, challenges, solutions, applications, and overall management prospects associated with this integration. Additionally, this study provides an analysis of future trends and real-time applications, which significantly contributes to the development of a cost-effective and durable energy storage system architecture with an extended lifespan for renewable microgrids. Therefore, providing a summary of the anticipated findings of this scholarly paper contributes to the advancement of a techno-economic and efficient integration of ESS with a prolonged lifespan for the use of green microgrids.

1. Introduction

1.1. Background and Motivation

The global energy supply is largely dependent on the utilisation of non-renewable sources of energy and fossil fuels to produce power. However, the tremendous amount of pollution caused by these sources has adverse effects on the planet. It is therefore necessary to reduce the strain on fossil fuels by replacing them with renewable energy sources (RESs). According to the International Energy Agency, the demand for renewable energy resources escalated by 3% in the year 2020, with the primary constituent of this increase being the growth in electricity generated using RES.

Countries such as Sweden and Scotland are leading this change by having almost all of their electricity needs produced by RES. Today many other nations have joined this goal of using only RES to fulfil all their energy requirements. The total renewable energy capacity of various countries is shown in Figure 1. China boasts the highest electricity generation capacity, with a total of 1020 GW, followed by the United States with 325 GW, Brazil with 160 GW, and India ranking fourth with 147 GW. These data underscore the substantial gap between the leading and second-ranking nations, highlighting the imperative for significant growth in total renewable capacity. According to the International Renewable Energy Agency’s Renewables Capacity 2022 Report, with an addition of 257 GW of energy-producing capacity, the renewable power generation (RPG) capacity of the world increased to 3064 GW at the end of 2021. The addition of different types of renewable energy on a global level is shown in Figure 2.

One way to make constructive use of RESs is by incorporating them within microgrid systems to produce energy. A microgrid is a self-supported grid system capable of producing and distributing energy locally over an area. Among the various types of RES available, the most popular are solar and wind energy due to their capacity of maximum power point tracking (MPPT) in addition to low costs of generation. Despite this, the use of such RES in microgrids remains minor due to their intermittent natures. It is challenging to integrate the various sources of energy into the microgrid system to be able to produce electricity effectively. The incorporation of solar and wind power generators with the DC bus of the DC microgrid is one option being investigated in relation to this issue.

Microgrids typically consist of five different components, namely, the utility grids, the energy storage system (ESS), controllable and uncontrollable generators and the load. The utility grids are interconnected with the help of transmission lines and are networks of synchronised power suppliers.The ESS provides a contingency measure to ensure a continuous and uninterrupted power supply in scenarios in which the electricity generated falls short of the energy demand. It consists of batteries that store electricity for further use.

The selection of DC Microgrids is motivated by several compelling factors in the context of contemporary energy and environmental challenges. With the gradual depletion of the ozone layer and the pressing need to mitigate carbon footprints, it has become imperative to incorporate RESs into power generation. Microgrids emerge as a practical and well-suited solution for achieving this goal. Additionally, economic constraints, rapid technological advancements, and heightened awareness of environmental impacts have led to an increased demand for distributed generation systems. These distributed generation systems, particularly Microgrids, have garnered significant attention due to their capacity to alleviate the strain on conventional power systems. Furthermore, addressing the intermittent and irregular behavior of RESs is critical to enhancing their penetration in the energy landscape. DC Microgrids offer an effective means to tackle this challenge. Lastly, the deployment of Microgrids is particularly advantageous in providing reliable electricity to remote locations, including defense installations and islands, where traditional grid infrastructure may be impractical. Consequently, the adoption of DC Microgrids aligns with the overarching goals of sustainability, resilience, and accessibility in the contemporary energy paradigm.

At the expense of efficiency, innovation has been compelled to tailor to AC power. One major example of this is the invention of the semiconductor and the extensive use of it in consumer electronics. All electronics that use semiconductor-based transistor technology require DC power to operate, such as Variable Frequency Drives for pumps, Heating, Ventilation and Air Conditioning systems, fans, elevators, mills and traction systems [2]. One notable example is the steel industry, in which DC electric arc furnaces are being adopted due to their reduced energy consumption and minimised light flicker in comparison to their AC counterparts [3]. These advancements in the implementation of DC power demonstrate its growing significance as an efficient and sustainable source of power in various industrial settings. As a consequence, DC power requirements apply to the majority of consumer appliances that are used in daily life. Along with consumer electronics being DC-based, technological advancements have developed distributed generation (DG) systems that are incompatible with the conventional AC grid. Photovoltaic (PV) systems are utilised all over the globe to transform renewable solar energy into electrical energy, but they naturally produce DC. In addition, DC electricity is produced by other DG technologies like fuel cells. The customer is unable to directly take advantage of these RESs without enduring efficiency losses caused by conversion because of the existing power distribution infrastructure. The integration of dispatchable DGs, such as diesel generators with flexible power generation, or energy storage units is crucial for the steadiness and dependability of microgrids with intermittent energy-based DGs [4].

DC microgrids produce DC energy for distribution and utilisation. In the current semiconductor-dependent world, the additional advantages are instantly apparent as DC microgrids only rely on DC power. The omission of the power converters, which are presently necessary for DC device interoperability, is the primary advantage of a DC microgrid [5]. DC microgrids also offer a number of design advantages compared to their AC counterparts. DC microgrids are neither impacted by zero crossings, nor do they require phase synchronisation. This type of microgrid also eliminates inductive losses and transmits only real power increasing efficiency. The implementation of a DC microgrid can lead to a quick increase in efficiency, potentially up to 20%, which is subject to the efficiency of the power sources and rectifiers being used [6].

The standardisation of DC microgrids is a critical aspect of ensuring their interoperability, safety, and reliability in various applications. The current status of DC Microgrid standardisation is that it is an ongoing process. The development and adoption of standards can take time due to the complexity of the subject matter and the need for consensus among stakeholders. These standards cover various aspects of DC Microgrids, including voltage levels, communication protocols, safety measures, and grid integration. Some of the most-used standards have been discussed in

Table 1. DC Microgrid Standards.

Standard	Description
IEEE 946	Supplementary DC control systems for power plants, emphasising DC equipment voltage ratings and presenting a standard DC system architecture.
IEEE 1547	Interconnection requirements essential for electric power systems featuring distributed energy resources, outlining their relevance to both islanding and grid-connected operational modes.
REbus	Open standard pertaining to clean power distribution based on DC.
EMerge Alliance	DC microgrid standards, design and control systems in building applications.
IEC SG4	LVDC distribution system with a maximum voltage of 1500 V, examining architectural options that include both pure DC and hybrid AC-DC configurations.
NFPA-70	The current electrical building codes in the US that apply to DC microgrids in Article 625, 690, 692, 694, 700, 705, 706, 710, and 750.

.

1.2. Literature Review

A three-stage hierarchical model is introduced in [7] to enhance DC microgrids’ resilience, focusing on network outage management based on microgrid-reported data. Proactive measures are taken to bolster network readiness, optimise generation scheduling, mobile unit allocation, and distribution of feeder reconfiguration to minimize costs. The model shows that information sharing among microgrids also boosts supply service levels to critical loads by 48.16%, enhancing system resilience by 3.47%. In [8], a novel prediction-based mechanism is introduced in a hierarchical decentralised framework for dynamic demand response schemes tailored to smart prosumers’ behavior. The framework leverages deep learning-based forecasting and a risk-averse strategy based on the information gap decision theory to control scheduling risks. The results demonstrate reduced clearing prices for electricity and heat markets during peak hours, enhanced scheduling robustness, and the feasibility of a transport system-based structure for hydrogen exchange.

Particle Swarm Optimisation (PSO), a bio-inspired metaheuristic algorithm, is discussed in [9] to address the dynamic economic load dispatch problem in grid-connected microgrids. The findings underscore the substantial cost savings associated with microgrids’ power exchange with the grid and the integration of demand response, reducing operation costs by 21.77%. PSO outperforms other optimisation algorithms, such as grey wolf optimisation and backtracking search algorithm, in terms of operation cost optimisation. In [10], a mixed binary-continuous PSO approach to address the unit commitment (UC) problem within demand response-integrated microgrids, while accounting for uncertainties. The research explores six distinct scenarios and validates the potential of load curtailment as an incentive-based demand response strategy to demonstrate the superior performance of the proposed PSO method compared to benchmark optimisation algorithms across different scenarios. Reference [11] addresses UC in grid-connected microgrids with BESS, considering BESS degradation in the optimisation. It formulates UC as a constrained optimisation problem, utilising a Quadratic-PSO variant to solve it efficiently. Results show that incorporating BESS degradation increases BESS lifecycle from 350 to 500 cycles and minimum depth of charge from 5.5% to 34%. However, it leads to operation costs rising by 2.21%.

A bi-level bidding system is proposed in [12] for managing energy exchange among interconnected microgrids, considering traditional and smart consumers. In this model, microgrids create offers/bids in the upper level, while a community manager in the lower level sets market-clearing prices to maximize social welfare. The simulation results indicate that risk-taker scheduling, as opposed to risk-averse scheduling, reduces market-clearing prices and total system costs by 3.29%. It is also found that modifying the consumption patterns of regular consumers through a Demand Response Program also leads to cost reduction and improved comfort for smart consumers. The authors of [13] explore the integration of flexible resources, including fast-responsive demands and fast-start generators, to enhance microgrid flexibility and reduce operational expenses. The problem is formulated as a two-stage stochastic optimisation, optimising day-ahead and real-time decision variables simultaneously to minimise expected microgrid operation costs. The results indicate that these resources significantly decrease microgrid operation costs, enabling microgrids to adapt to various scenarios efficiently, and highlight their impact on electricity market interactions.

UC in combined cooling, heat, and power microgrids involves optimising the scheduling and operation of various energy generation and storage units, including combined heat and power units, to meet the electricity, heating, and cooling demands efficiently while considering factors such as costs, energy sources, and environmental objectives. A risk-aware information gap decision theory-based UC system with various constraints is implemented in [14]. It considers uncertainties in demands and wind power, modeling microgrid operation as a mixed-integer linear problem. In risk-averse UC, robust scheduling guarantees that the microgrid’s operation cost remains within a 6.7% deviation from nominal electric demand, while risk-seeking UC aims for minimal deviations to achieve a 2% cost reduction. The results of a similar model in [15] highlights the impact of battery energy storage, thermal energy storage, and battery charging stations on microgrid operation costs. Battery energy storage reduces costs by 7.42%, thermal energy storage by 1.82%, while battery charging stations increase operation costs by 20.19%.

In [16], limitations of the classical impedance-based approach are discussed as part of the analysis of grid-tied inverters under weak grids. The study introduces a line impedance cooperative stability region identification method that offers more detailed stability information. This method utilizes output impedance and input admittance matrices, creating a novel stability forbidden region that is less conservative than existing criteria. Through mapping techniques, it establishes the stability operation region and transforms it into the eigenvalue identification problem. Guardian map theory is applied to solve the detailed line impedance cooperative stability region efficiently. Results confirm the method’s effectiveness and reduced conservatism. Further, the challenge of achieving real-time current sharing based on the maximum capacities of wind and solar energy sources in hybrid systems is analysed in [17]. It introduces an accurate current sharing and voltage regulation approach based on a distributed adaptive dynamic programming approach. The method builds an equivalent wind/solar model to ensure complementary energy utilisation. It formulates the current sharing and voltage regulation as an optimal control problem, with each source agent aiming for optimal control variables. The results demonstrate the effectiveness of this adaptive dynamic programming approach.

In this paper, we aim to fill the gap in research by proposing a comprehensive review of DC microgrids. The main contributions of the proposed paper are summarised next:

• It provides an extensive overview of the entire lifecycle of DC microgrids, covering planning, operations, control strategies, and protection mechanisms in a single consolidated source, offering a holistic perspective of DC microgrids, which is often lacking in fragmented literature.
• It presents the latest research findings and emerging trends in DC microgrid technology, providing up-to-date insights for readers ranging from experts in the field to newcomers.
• It demonstrates practical applicability through the examination of real-world case studies and the exploration of implementation challenges, thereby providing valuable insights for researchers, practitioners, and policymakers alike.

Figure 3 displays an objective flowchart of the whole paper. Firstly, the detailed planning of DC Microgrid is elaborated in Section 2. Section 3 discusses about the different operations for load management and economic feasibility. In Section 4 various control strategies which ensure effective operation, stability, and reliability of DC microgrids are discussed. Section 5 explores importance of protection and various protection schemes. And finally, the conclusion is depicted in Section 6.

2. Planning of DC Microgrid

DC Microgrid planning involves designing and implementing a microgrid system that uses direct current for power distribution. Figure 4 demonstrates the standard DC Microgrid’s architecture. Recent works have looked into the planning of a DC distribution systems. Several factors should be considered while constructing a DC microgrid, especially if the utilised instrument was developed for AC uses originally. To ensure a reliable design, simplified models that describe load behavior during DC operation must be obtained as a key requirement. A proficient scheduling of the available load and generation components is imperative to enhance the robustness of the DC microgrid in instances in which it is disconnected from the main grid due to voltage disturbances, faults, or other disturbances.

Table 2. Real-world microgrid projects.

Serial No.	Microgrid Project Name	Scope of the Project
1	Energy Neutral Homes in Stroomversnelling	In this project, A collaborative initiative involving six housing corporations has been established to undertake a national effort aimed at transforming about 110,000 residences into zero-energy homes. These residences, referred to as “Nul-op-de-Meter” dwellings, are commonly known as “Zero-on-the energy-meter” homes.
2	Washington DC University DC Microgrid	In this project, the microgrid comprises a collection of solar panels with a total capacity of 2.5 MW, which are installed on various rooftops and parking garages inside the campus. Additionally, the microgrid includes a lithium-ion battery with a capacity of 1.2 MW/2.5 MWh, as well as a combined cooling, heat, and power system with a capacity of 4.5 MW. Gallaudet campus, in conjunction with scale microgrid solutions and urban inventiveness, has formed a collaborative effort to develop a solar plus storage microgrid. The primary objective of this initiative is to provide sustainable power to both the campus and the neighbouring residential areas, utilising a community solar plan.
3	Florida’s Neighbourhood Microgrid Project	In this project, the Medley at Hillsborough has a total of 37 residences, which have been selected for the purpose of assessing the incorporation of diverse renewable energy sources and mitigating peak energy demand. Furthermore, investigations have been conducted to assess the microgrid’s ability to withstand disruptions in the primary AC network.
4	DC Flex House	The project aims to propose a strategic approach for the replacement of traditional AC-based residential dwellings with DC technology. It also yields a methodology for the conversion of electrical installations in residential homes from AC to DC, incorporating the integration of novel components and products. A joint initiative between the industry and an institute, namely ABB, has been undertaken to develop components and establish mutual inter-connectivity. Additionally, the Hague University of Applied Sciences is involved in the project to contribute to the bottom-up vision creation for DC at the district level.
5	UK Microgrid Project on Residential Electrification	A project involving 162 houses is underway to establish a private microgrid by 2025 and the main goal is to achieve homes that are fully electric and equipped for zero carbon emissions, without the need for a gas connection.

discusses the various real-word case studies of microgrid implementations.

It is worth noting that DC Microgrids are still not as common as AC Microgrids and are mainly used for specific applications such as building, data centres, and islands. It is noted in [18] that there is a threshold ratio of DC loads above which the DC microgrid is economically preferable to the AC microgrid. If the proportion of DC loads is below a certain limit, adopting an AC microgrid would be more cost-effective. Conversely, if the ratio exceeds this limit, it would be more desirable to implement a DC microgrid. The ratio of DC loads was found to be the most important component in determining the sort of microgrid, and any changes to other parameters would not have an impact on the kind of microgrid as long as the ratio remained constant. An approach for the best planning and construction of PV-based DC microgrids with better distribution efficiency, typically 10% or greater, has been developed in [19]. The framework aims to provide economical solutions by selecting optimal sizes for solar panels, energy storage, and conductors. The analysis considers the impact of regional temperature and irradiance patterns on the sizing of the system.

Moreover, it is crucial to take into account the integration of the microgrid with the wider power grid, as well as the regulatory and standardisation requirements that pertain to the microgrid.

2.1. Load Forecasting

Load forecasting is an important aspect of operating a DC microgrid. Load forecasting pertains to the procedure of anticipating the forthcoming energy consumption in a microgrid, based on historical data and other factors such as weather, economic trends, and the presence of certain types of energy-intensive equipment.

Load forecasting can be performed using a variety of techniques, including time-series analysis, statistical analysis, and artificial intelligence algorithms. The precision of load forecasting is critical to the performance of a DC microgrid, as it affects the efficiency, reliability, and cost-effectiveness of the microgrid’s operations. In [20], fuzzy prediction interval models have been proposed as a solution for forecasting in microgrids. The approach performed better than linear regression models, especially for longer look-ahead forecasts, as shown by real data from a microgrid in Chile. The microgrid’s reserve requirements will be derived using the worst-case scenario inferred from the fuzzy intervals. Reference [21] discusses the application of a deep learning-based energy forecasting system for microgrids. The system utilises LSTM-autoencoders and a sliding window algorithm to facilitate fast learning of input data. The study concentrated on 10 min forecasts, which demonstrated adequate accuracy and reliability. Another technique for adaptive probabilistic load forecasting in electric power systems is demonstrated in [22], which adapts to variations in consumption behaviour and assesses load uncertainties. It employs learning methodologies that modify model configuration with a recursive approach and probabilistic forecasting methods that yield precise and versatile predictions. The proposed method was tested on multiple datasets and showed improved performance in relation to forecasting inaccuracies and probabilistic forecasts compared to existing techniques. In [23], a new bilevel prediction strategy for the Short-Term Load Forecasting (STLF) of microgrids is presented. In contrast to conventional forecasting techniques, a novel chance-based search approach is introduced to optimise the adjustable parameters of both a feature selection technique and a forecasting engine. The proposed strategy focuses on improving STLF, while the estimation of realised demand (difference of load demand and renewable energy generation) will be addressed in future work. A new method for STLF in microgrids is presented in [24]; it combines k-means clustering, long short-term memory networks (LSTM), quantile regression (QR), and kernel density estimation (KDE). The K-QRLSTM model was evaluated using real-world load data and showed improved accuracy, speed, and precision in comparison to other approaches like Back Propagation, LSTM, and QR. According to the findings, the proposed approach can generate accurate load forecasts utilising deterministic, interval, and probability density data.

Advanced load forecasting algorithms, such as machine learning algorithms, can provide more accurate predictions of future demand for electricity in a DC microgrid by taking into account a wide range of variables and historical data. A novel hybrid approach is proposed in [25] for STLF in microgrids, using a combination of the best-basis Stationary Wavelet Packet Transform (SWPT), Harris Hawks Optimisation (HHO), and Feed-forward Neural Network (FNN). The presented strategy is assessed using retrospective load demand, climatic, and data with correlation between working and non-working days across different seasons. The significance of STLF for the dependable and cost-effective functioning of a microgrid is emphasised in [26]. To overcome the difficulties brought on by the volatile and irregular load time series of microgrids, the authors suggest a new forecasting methodology. In order to enhance the capability of capturing the intricate non-linear patterns in the time series data, the method employs a Self-Recurrent Wavelet Neural Network (SRWNN) with a feedback loop. The proposed method was assessed using actual load data from a microgrid and contrasted with load data from two power systems. The findings demonstrated that the SRWNN model produces forecasts for volatile time series predictions that are more precise. The application of these algorithms can significantly enhance the operational performance of a DC microgrid, thereby guaranteeing its optimal functionality in terms of efficacy, dependability, and cost-effectiveness.

2.2. Topology Selection

The topology selection for a DC microgrid is an important design decision that affects its performance, reliability, and cost. The topology choice is determined based on the microgrid’s specific requirements, such as the size of the energy sources and loads, the level of redundancy sought, and the cost limitations. In [27], it was observed that small-scale ESSs can be employed in Networked Microgrids (NMGs) with greater financial advantages, as well as boosting dependability and resilience through energy exchange among microgrids in NMGs. An optimisation model with bi-level programming is utilised to maximize the annual net gains while considering the constraints of reliability and resilience. The results show that while NMGs have an annual net profit that is 20.66% higher than that of non-NMGs, their initial investment cost is 15.06% affordable. In contrast to non-networked microgrids, NMGs’ optimised ESS capacities are lower, but they still offer greater dependability and resilience.

2.3. Sizing of Energy Storage System

ESS play a pivotal role in DC microgrids by allowing the storage of excess energy produced by renewable sources, such as solar and wind, for later use. The proper sizing of ESS in a DC microgrid is considered a critical aspect to guarantee stable and efficient microgrid operation. The size of the ESS is determined by considering various factors such as the load profile, the renewable energy source availability, the grid connection and regulations, and the cost. The main objectives of ESS sizing are to balance the load and generation, to provide backup power during outages, and to optimise the utilisation of RESs. ESS can help reduce energy costs by allowing the microgrid to take advantage of lower energy prices during periods of low demand, and by reducing the need for peak power generation. They also provide backup power during outages, ensuring that the microgrid continues to operate even in the absence of a reliable grid connection.

In DC microgrids, the ESS size is typically calculated based on the capacity (kWh) and power (kW) requirements. The capacity requirement is determined by the amount of energy that needs to be stored to satisfy the energy requirements imposed by the load, while the power requirement is based on the maximum rate of energy transfer needed to support the load. The ESS sizing also takes into account the depth of discharge, which defines the minimum energy storage requirement in ESS. Reference [28] proposes an improved approach for determining the battery energy storage system’s (BESS) preferred technology, desired depth of discharge, replacement year, and the ideal size. The BESS service life and capacity degradation are taken into account when formulating the problem using the mixed-integer linear programming approach. The proposed approach was determined to be a plausible solution to the BESS sizing issue in a microgrid, reducing the total cost of the microgrid by 4.62% and 6.5% over 10-year and 15-year planning horizons, respectively. Another strategy is proposed in [29] for determining the optimal capacity of a BESS in a smart microgrid with high PV penetration. It considers system cost variability and uncertainties by incorporating a mean-variance Markowitz theory-based risk analysis resulting in the minimisation of system cost and optimisation of BESS sizing. In [30], a comprehensive solution for optimal sizing of BESS in microgrid applications is proposed. The study requires the implementation of a BESS degradation model, which relies on the association between the depth of discharge and the number of charging as well as discharging cycles. This model also includes islanded scenarios to determine microgrid reliability. The efficiency of the suggested model was confirmed by carrying out computational simulations on a test microgrid, using a resilient optimisation approach to tackle the uncertainty of load requirements and renewable power production.

Equation (1) formulates the BESS investment cost, where $P{C}_{BESS}$ represents the power cost of a BESS per MW, while $P_{BESS}^{max}$ signifies the BESS’s maximum power capacity. And $E{C}_{BESS}$ and $E_{BESS}^{max}$ are the BESS’s energy cost per MWh and energy capacity, respectively [31]. In [32], the sizing problem is formulated as MILP. The objective function is expressed in Equation (2), where $IC$ and $OC$ indicate the BESS investment cost and the microgrid operation cost, respectively.

$$I{C}_{BESS}=P{C}_{BESS}{P}_{BESS}^{max}+E{C}_{BESS}{E}_{BESS}^{max}$$

(1)

$$Min\left(\sum _{eϵE}I{C}_e+\sum _{tϵT}O{C}_t\right)$$

(2)

In conclusion, the sizing of ESS in DC microgrids is a complex process that requires a comprehensive analysis of the load profile, energy source availability, grid connection and regulations, and cost. An optimal ESS sizing can help to maximise the use of RESs while simultaneously ensuring the DC microgrid’s stable and effective functioning.

3. Functionality of DC Microgrid

DC microgrid involves the coordination of various components, including RES, ESS, and loads, to guarantee the efficient and reliable delivery of electricity. The functionality of a DC microgrid requires a robust and sophisticated control system, which can monitor the performance of the components, manage the flow of energy, and respond to changes in demand, supply, and weather conditions.

3.1. Unit Commitment

Unit commitment (UC) is a crucial process in DC microgrids that involves scheduling the operation of power generation units to meet the demand for electricity while minimising costs. In DC microgrids, the unit commitment process is similar to that in AC microgrids, but some key differences need to be considered.

One of the key distinguishing factors of DC microgrids is that they often rely heavily on green energy sources. These sources highly rely on weather conditions, which can be unpredictable, so the UC process must take into account their inherent variability and uncertainty. In [33], a stochastic mixed integer program for day-ahead UC is put forth, which facilitates the exchange of energy and associated services with the primary grid, while also accounting for the variability of PV generation. To handle computational complexity, Benders’ decomposition is used. With the use of Monte Carlo simulations, the commitment schedule is verified using an analysis of the solar situation. A more affordable UC approach for isolated microgrids is produced via Mixed Integer Quadratic Programming in [34]. It has linear constraints and a quadratic objective function. It is based on a day-ahead with Model Predictive Control (MPC) technique and incorporates changes in dispatchable unit outputs to better optimise dispatch settings. As a consequence, operational expenses are reduced without having a substantial impact on calculation times.

Adaptive UC is an extension of the traditional UC problem that incorporates forecast errors and real-time measurements to improve the scheduling of power generation units in a DC microgrid. A key advantage is that it can boost the accuracy of the UC schedule and reduce the need for manual interventions in the microgrid operation. By adapting the UC schedule to changes in renewable energy generation and demand, it can help to minimize the cost of microgrid operation while ensuring the stability and reliability of the system. One such approach is presented in [35], a contextual learning algorithm for unit commitment that is adaptively partitioned in microgrids with RESs. Without needing any a priori knowledge, the algorithm discovers the ideal UC schedule and lowers the cumulative cost. The algorithm adapts to uncertainties by considering strongly correlated context information such as the historical pattern of load demand and weather conditions. The paper proves the optimality of the proposed method and shows through simulations that it outperforms traditional UC algorithms even with small errors in the information. A data-adaptive robust optimisation strategy is described in [36] for the optimum UC in hybrid AC/DC power systems. The UC problem is a mixed-integer second-order cone programming problem which incorporates the spatio-temporal interdependencies of multiple wind farms. The proposed methodology successfully handles the multi-scenario problem by employing a column-and-constraint-generating algorithm. Findings indicate that the proposed method can manage the uncertainties brought about by wind power while lowering operational costs, minimising generator startup/shutdown times, and fully utilising the regulation capability of DC lines.

Overall, the UC process in DC microgrids involves optimising the operation of generators, loads, and energy storage devices to minimize costs while fulfilling the electricity demand and ensuring the stability and reliability of the system in the face of inherent variability and uncertainty in RESs.

3.2. Economic Dispatch

Economic dispatch (ED) is the process of allocating power generation among the available units in a microgrid with the objective of fulfilling the electricity demand at minimum generation cost. The energy management in a DC microgrid can be relatively more complex than that of an AC microgrid, primarily because of the limited number of generators and loads involved. Additionally, DC microgrids often rely heavily on RESs, which are variable and uncertain.

To perform ED in a DC microgrid, the operator must consider the real-time demand for electricity and any constraints on the system, such as voltage and current limits. The microgrid economic dispatch problem was investigated in the [37]. The study modified existing constraints for ED and added a new one to ensure stable islanded operation. Numerical simulations were performed on a test system with 15 DG units, considering factors like source type and part-load performance. The modified dispatch solution showed economically feasible operation in grid-connected mode and stable operation in islanded mode, even with a cost increase of up to 0.7%.

One approach to economic dispatch in a DC microgrid is to use optimisation algorithms to determine the optimal allocation of power generation among the available sources while minimising the cost of generation. This can involve solving a linear or nonlinear programming problem subject to a set of constraints. The optimisation algorithm takes into account the available power generation from each source, the cost of generation, and any constraints on the system. An heuristic optimisation technique is presented in [38] to reduce the costs of operation of a microgrid by dispatching resources economically. The optimisation problem incorporates the power flow model, accounting for transmission losses during generation dispatch, and demand response obligations from the utility. While taking part in demand response, it can increase the 380 V DC microgrid network’s system efficiency.

In [39], a finite-time consensus protocol-based power optimiser is designed. An increased convergence rate allows the controller to operate better. This study utilises a finite-time tracking scheme to address deviations in the average voltage of a microgrid and attain a balance between power generation and demand. The voltage regulator is employed to sustain the system’s equilibrium point, which represents the optimal operating point.

Overall, economic dispatch is an important process in a DC microgrid that ensures the efficient and cost-effective allocation of power generation among the available sources while meeting the real-time electricity demand.

3.3. Reserve Management

Reserve management in DC microgrids is an essential task to ensure the system’s stability and guarantee reliable operation. The primary objective of reserve management is to have enough reserve capacity to handle any potential disruptions in the system caused by variations in demand or supply. The reserve capacity is often used as a buffer to cope with unexpected disturbances or system failures.

Using energy storage devices like batteries or supercapacitors is one strategy for reserve management in DC microgrids. ESSs can be utilised to provide reserve capacity by storing excess energy generated during periods of low demand and discharging it during times of high demand. The reserve power from ESSs can be leveraged to regulate the voltage and frequency of the microgrid. Reference [40] outlines a distributed control approach for a hybrid ESS in a DC microgrid that can reduce bus voltage volatility and increase the lifespan of energy storages. To improve control precision, a multilevel energy management system (EMS) is presented. Although the control precision may suffer in such a case, the suggested distributed control approach can operate independently of the communication link, guaranteeing system operation even in the event of a communication breakdown. The method in [41] considers RPG output as the major uncertainty and formulates the inherent risks of load shedding and RPG restriction using the probabilistic distribution of predicted RPG values. The presented model uses a MILP problem, which is successfully resolved using a high-performance solver and views RPG output as the primary source of uncertainty. The study also examines how generated solutions are affected by RPG prediction accuracy and battery loss costs for electric vehicles. A hybrid microgrid EMS that uses a novel control approach to assess the power availability on the DC side before its transfer by determining its strength is proposed in [42]. Only when the electricity is sufficient is the conversion process activated; in the event of insufficient power, an auxiliary battery is utilised to store excess energy, resulting in a significant reduction in conversion losses and an 8–10% improvement in battery energy efficiency. Another EMS that operates in real-time and incorporates an adaptive energy calculator (AEC) that utilizes a moving average measurement method to effectively control bus voltage, eliminate high current pulsation, and reduce power and frequency fluctuation is established in [43]. The findings indicate that the suggested EMS utilising AEC method is superior in controlling the bus voltage in diverse load situations, along with mitigating the harmful consequences of pulsating load.

Another approach to reserve management in DC microgrids is through the use of demand response programs. Demand response programs encourage customers to use less energy during peak hours, which lowers system demand and frees up reserve capacity. When the system is at risk of instability and there is a limited amount of reserve capacity, this strategy can be especially helpful. In [44], a two-stage stochastic programming method was put forth to overcome the RES’s uncertain behaviour. The objective was to identify the optimal energy and reserve solutions to lower the microgrid’s operating costs. The network restrictions of the microgrid were modelled using a DC Optimal Power Flow, and the technique was evaluated using actual generation data on a 37-bus distribution grid. The outcomes demonstrate strong computing efficiency and reasonable accuracy to the system’s inherent non-linear behaviour. The study demonstrates the proposed strategy’s potential for resolving energy and reserve market issues in microgrids with high RES penetration.

In summary, effective management of reserves in DC microgrids is crucial to ensure system stability and reliability. The utilisation of strategies such as ESSs and demand response programs can aid microgrid operators for regulation of power supply and demand, particularly during times of high demand or unexpected disturbances.

4. Control of DC Microgrid

Control in a DC microgrid refers to the process of monitoring and adjusting the flow of electrical power within the microgrid to ensure that it operates efficiently and reliably. The goal of control in a DC microgrid is to maintain stability, ensure security, and enhance the standard of the electrical supply. The selection of a control technique for a DC microgrid will be contingent on the particular specifications of the microgrid and the goals of the control system. In general, a combination of different control methods is often employed to attain peak performance and stability in a DC microgrid.

4.1. Control Architectures

4.1.1. Centralised Control

Centralised control in a DC microgrid refers to a system-level control architecture in which a single central controller manages and coordinates the operation of all components in the microgrid as illustrated in Figure 5. The central controller is accountable for observing the system state and making decisions about how to distribute energy and power within the microgrid. This approach involves a hierarchical structure in which the central controller receives feedback from the system components and makes decisions based on the received information. The instructions issued by the central controller are subsequently transmitted to the individual system components for implementation. The central controller is responsible for ensuring that the microgrid operates efficiently, reliably and within safety constraints. This approach is typically used in larger microgrids in which a centralised control system provides better coordination and optimisation of the microgrid components.

In a DC microgrid, centralised control refers to a control approach in which a single centralised controller is responsible for controlling and coordinating the various components of the microgrid, for instance, distributed energy resources (DERs), ESS, and loads. In [46] the control and stability of DC islanded microgrid systems with communication delays are addressed. The Lyapunov–Krasovskii theorem is used to ensure stability, while the H1 disturbance attenuation requirement is used to ensure robustness. For various delay scenarios, the performance of the controllers is assessed with variable loads and erratic input voltage. The results show that the predictor-based robust controller performs best in maintaining voltage regulation and stability under different conditions. The controllers’ gains are obtained through linear-matrix-inequality (LMI) constraints and the performance is verified through simulations and analysis

These techniques can be utilised to control the power flow, voltage, and frequency in the DC microgrid, ensuring that it functions efficiently, reliably, and satisfies the requirements of the end-users.

4.1.2. Decentralised Control

Decentralised control refers to a system architecture in which the control and decision-making functions are distributed among multiple agents, rather than centralised in a single entity as shown in Figure 6. In the context of a DC microgrid, decentralised control involves distributing control and decision-making among various components such as PV panels, ESS, and loads. Each component is capable of operating independently and communicating with other components to coordinate and optimise the overall system performance. In order to attain effective control of the DC bus voltage, predefined load sharing, and minimisation of circulating current, a decentralised control technique for DC microgrids is introduced in [47]. The proposed method utilises the back-stepping technique to convert the original output-constrained system into an unconstrained one, while also limiting transient tracking errors. Meanwhile, a cooperative decentralised control strategy is proposed in [48] for power control in a DC microgrid with various energy sources, such as PV and conventional DGs. The approach consists of a hierarchical structure of primary, secondary, and tertiary controllers to effectively manage power flow. The cooperative control strategy mitigates voltage oscillations caused by constant power loads and reduces the overall generation cost by controlling the voltage output of each power converter.

In a decentralised DC microgrid, the individual components can monitor and control their power production and consumption, as well as the flow of power within the system. This offers greater efficiency and flexibility by allowing for real-time response to changing conditions. Additionally, it enhances system reliability by enabling independent operation of other components in case of a component failure. A decentralised control method utilising MPC is proposed in [49] for maintenance and regulation of PCC voltage of Distributed Generation Units (DGUs) while accounting for uncertainties, disturbances, and unmodeled dynamics. The method employs state-space-based MPC and its efficacy in maintaining the PCC voltage within a reasonable range, despite load disturbances and changes in the reference voltage, was demonstrated through simulation. The advantage of using MPC in this study is its ability to consider disturbances and uncertainties in regulating the PCC voltage of DGUs. Another strategy is proposed in [50], a decentralised and communication-less coordination control method for a PV-BESS in a microgrid. The proposed technique utilises decoupled control loops and existing BESSs to facilitate dynamic compensation even in generation-dominating mode. In the control system, both the PV converter and BESS converter are taken into account. The PV converter controller allows for smooth switching between MPPT control and droop control. In [51], a decentralised optimal primary voltage control system for autonomous DC microgrids is put forward. It has unique features such as robust stability and performance under uncertainty, optimal performance resulting from a unique convex optimisation problem, no limitations on Lyapunov matrices for decentralised approach, and no need for pre-filter design. It demonstrates robust performance in various scenarios including PnP operation, topology changes, load variations, and the occurrence of constant power loads (CPLs).

Overall, decentralised control in a DC microgrid can enhance the efficiency, adaptability, and dependability of the system, making it a more attractive solution for various applications.

4.1.3. Distributed Control

Distributed control, which only communicates with nearby units using available digital lines of communication, combines the benefits of both centralised and decentralised controllers, as can be seen in Figure 7. This approach refers to a control system in which the control logic is divided among multiple devices in the network and the decisions are made collectively. This results in a more robust and flexible system, as individual devices can act locally based on the information they have, reducing the impact of failures. This type of control system is used to govern the flow of power within the microgrid, ensuring a reliable and efficient distribution of energy. A distributed optimal control technique is formulated in [52] to reduce operation loss in DC microgrids. The communication network for optimal control involves limited information exchange and is the same as that of the electrical network. The traditional hierarchical scheme is improved for real-time optimisation, energy efficiency, and to avoid deviation from optimal operating points. The approach derives an equivalent optimality condition using the Karuch–Kuhn–Tucker optimisation challenge, which is then utilised to develop a distributed control algorithm. The efficacy of the control methodology was demonstrated through switch-level simulations in DC microgrids. It has been found that the proposed controller can handle a communication delay of no more than 1500 µs. In [53], a control paradigm for DC Microgrids is a distributed secondary/primary control framework that uses cooperative control. The voltage and current regulation modules on every converter utilise a noise-resistant voltage observer to approximate the global mean voltage and proportional load distribution. The limited communication network facilitates data transfer between converters. The created framework offers accurate global voltage control and proportional distribution of loads while being resistant to noise and link malfunctions. In [54], a novel distributed discrete time control approach for multi-bus DC microgrids that is able to handle CPLs, constant impedance loads and constant current loads. The proposed approach aims to accomplish regulation of the weighted average bus voltage and proportional sharing of load current. It is completely distributed and only requires data from its immediate neighbours. The proposed controller is robust against communication network imperfections and plug-in and plug-out operations of DGs and can tolerate a communication delay of up to 2 ms or less. Simulation investigations confirm the functionality of the planned controller and demonstrate its ability to concurrently regulate voltage and current.

The distributed control system is capable of real-time monitoring of the microgrid’s state and can make informed decisions based on the data obtained from its diverse components, such as the PV panels, ESS, and loads. The utilisation of a distributed approach enhances the resilience and robustness of the microgrid, since it can persistently operate even when some components fail. The aim of the study described in [55] to upgrade and enhance the efficiency of DC microgrids by overcoming the limitations of traditional droop control through a distributed control scheme. This control method considers the impact of line resistances and loading conditions and is implemented locally in each source to maintain the decentralised structure of the microgrid. Furthermore, the efficiency of the considered control method is illustrated in an interconnected microgrid by using a centralised proportional integral controller (PIC) to regulate the tie-line power flow. Another approach is presented in [56], a distributed optimal controller for DC microgrids is proposed to achieve simultaneous bounded voltage regulation and minimised total generation cost. The distributed structure of the controller reduces the data exchange between neighbouring controllers. Through the use of Lyapunov synthesis, the resilience of the system is ensured, and the mathematical optimisation of the operating point is established. Using rigorous switch-level simulations, the effectiveness of the control method is evaluated, demonstrating that even with a 5 ms communication delay achieving the control objectives is possible, but voltage and current oscillations may still occur.

The control architecture of a DC microgrid is of paramount importance, as it defines how the various components within the microgrid communicate, coordinate, and make decisions to ensure efficient and reliable operation.

Table 3. Comparison of different control architectures.

Architecture	Advantages	Disadvantages
centralised	1. It depends on only one main controller called the Microgrid central controller and the data from several units are sent to the central control unit via the communication link and the control signals are sent back to each unit. 2. It is easy to implement. 3. The centralised controller has the capability to coordinate different energy sources to achieve the critical and non-critical loads.	1. The foundation of centralised control is communication, which also makes it vulnerable. 2. The key component in centralised architectures is prone to single point failure. 3. The response time is very low. 4. Centralised control structures incur the greatest computational burden, while communication failures within these structures can lead to system failures.
Decentralised	1. Each unit is controlled by its local controller, which only receives local information. 2. No Communication is required; hence, the response time is very high. 3. The failure of one controller does not have catastrophic impact since replacement capacity can be brought online immediately. 4. The system is more scalable, flexible, and reliable. 5. Communication burden is low hence computation cost is very low.	1. There is no centralised control unit. 2. The necessity for effective coordination among units within decentralised structures still remains a concern in need of more research. 3. Information security is also major issue in this controller
Distributed	1. The use of distributed control inside a microgrid facilitates the autonomous decision-making capabilities of a DER by utilising local measurements and restricted communication with other DERs. 2. Each unit is controlled by its local controller, which only receives local information. 3. The system is less flexible and more reliable compared to Decentralised. 4. The communication burden is lower than that of the centralised controller; hence, the computation cost is also low.	1. As the number of DERs rises, the distributed coordination process will require more iterations to converge. As a result, the response time will gradually increase. 2. There is no risk of single point failure due to absence of central control unit. 3. This architecture difficult to implement due to its complex design.

highlights the advantages and of the abovementioned architectures.

4.2. Control Schemes

4.2.1. Droop Control

Droop control is a distributed control approach widely used in DC microgrids to ensure coordinated operation of multiple power sources without relying on a centralised controller. Its primary purpose is to regulate the amplitude and power sharing among DC buses in microgrids. Voltage and current droops are two commonly employed techniques in droop control, in which the reference current is generated when the DC bus voltage droops, and the reference voltage is generated when the load voltage droops. To achieve effective control, the optimal power supply-based voltage droop control technique is utilized. However, voltage and current oscillations may still occur despite achieving the control objectives.

In DC microgrids employing droop control, a voltage or current controller is installed in each power source to regulate its output according to the voltage or current measurement at the PCC. The PCC voltage or current serves as an indicator of the overall electricity demand in the microgrid. If the demand increases, the PCC voltage drops, prompting the controllers in the power sources to boost their output. Conversely, if the demand decreases, the PCC voltage rises, leading the controllers to reduce their output. The authors of [57] introduce an observer-based control method for DC microgrids that combines voltage droop and current feed-forward control. The proposed method improves the dynamic response of voltage control by incorporating an observer. It also enhances load demand-sharing accuracy while maintaining system stability by addressing the trade-off between these two factors, enabling larger droop gains and improved load-sharing accuracy. Reference [58] introduces a novel current-limiting droop control method for DC microgrids to distribute power among multiple parallel DC-DC boost converters. Compared to conventional droop control, the proposed controller provides improved power sharing, load voltage regulation, and inherent current protection.

There are multiple techniques that can be used to implement droop control in a DC microgrid, including PIC controllers, voltage-sourced converters, and current-sourced converters. The selection of the droop control technique will depend on the individual criteria of the microgrid, such as its size, the number of energy sources and loads, and the desired level of stability and reliability. In [59], an adaptive droop scheme is implemented in a DC microgrid that includes a PV system to overcome non-linearity and achieve accurate power sharing. It is simple and adaptive and improves dynamic performance. The Sliding Mode Control technique is employed in the proposed scheme to achieve a quick dynamic response by controlling the output voltage and input current of every converter. In [60], a novel technique is proposed to assess the dynamic stability of DC voltage control in DC microgrids, which employs a reduced-order modelling approach. This method involves providing a modular modelling framework that is expandable for droop-based DC voltage control unit development. The study also investigates the effect of control specifications on the performance of the DC link voltage using an analytical solution of dynamic performance indices.

A DC microgrid can experience inaccuracies in load sharing caused by line resistances, which can be overcome by implementing an adaptive droop control algorithm, as found in [61]. The proposed algorithm not only addresses load sharing issues but also includes a distributed secondary controller that helps suppress circulating currents. This results in precise distribution of the load between converters with varying power ratings, which consequently enhances the stability and overall operation of the DC microgrid. In contrast, the approach proposed by [62] involves utilising a nonlinear droop curve coefficient in the heavy load range and a linear droop function that incorporates a negative droop resistance in the light load range. To further improve voltage regulation and enable smooth switching between light and heavy loads, the authors also employ a boost-up technique and curve-fitting methodology. The proposed technique has been verified through experimentation on two DC converters connected in parallel and can be generalized for DC microgrids that comprise multiple converters.

Droop control technique is commonly employed in DC microgrids owing to its simplicity, robustness, and ability to sustain a steady power flow between energy sources and loads. Additionally, droop control can be implemented using low-cost components and does not require a centralised control system, making it well-suited for use in small-scale and decentralised microgrids.

4.2.2. Hierarchical Control

Hierarchical control is a method of controlling the operation of a DC microgrid by dividing the control tasks into several hierarchical levels. The idea behind hierarchical control involves delegating control responsibilities to different control units, each with a distinct function in the microgrid’s operation. The hierarchical control architecture consists of three levels: primary, secondary, and tertiary controls. The primary control level focuses on regulating voltage and frequency, ensuring power balance, and implementing fault protection mechanisms. The secondary control level is responsible for tasks such as load forecasting, economic dispatch, and management of energy storage systems. The tertiary control level monitors and optimizes the microgrid’s performance over a more extended time horizon, undertaking duties such as performance monitoring, data analysis, and long-term planning.

In a hierarchical control system, each level of control has its objectives and operates independently from the other levels. However, the different levels are linked by a communication network that allows information to be shared and decisions to be made at the appropriate level of control. The authors of [63] propose a hierarchical approach for controlling DC microgrids, comprising two main control strategies: crowbar control and load shedding control. The objective of this approach is to simplify the control methodology for DC microgrids. In [64], the authors propose a hierarchical control strategy for the coordinated control of grid-connected DC microgrids, with a focus on power management. The control strategy involves two levels of control. To facilitate dynamic power sharing and state-of-charge (SOC) balance among different storage elements (SE) interfacing converters, SOC-based droop management is recommended. At the secondary level, droop-shifting methods based on SOC and VDC are employed. The experimental results demonstrate the effectiveness of this approach in compensating for DC bus voltage deviation, achieving SOC control of battery SE units, and enabling power management and coordinated control of grid-connected DC microgrids. A new three-level control hierarchy for DC microgrids is discussed in [65] to address the challenging converter-level management and the simple system-level management. The sophisticated converter topologies are represented as a black box at the converter control level, the first level. Utilising DC voltage signals, the second stage, known as voltage coordination, manages power distribution. The energy management level, which is the third level, improves electricity quality and flow through communication. This architecture balances robust and smart control functions, achieving a reliable power supply and optimal power management.

An SOC-implemented battery management system is presented in reference [66]. Equation (3) represents the formulated mathematical model for SOC calculations, where $Siz{e}_B$ denotes the battery capacity in kilowatt-hours (kWh), $P_b\left(t\right)$ signifies the power supplied to the battery, and $V_{dc}$ represents the dc-link voltage of the battery module. To ensure that the SOC never reaches 100%, a function is employed to decrease the power supplied to the battery when the SOC exceeds 90%, as expressed in Equation (4). Conversely, when the SOC is below 90%, the charging current remains equal to its predetermined value.

$$SOC\left(t\right)=SOC(t-1)+\frac{1}{3.6}\frac{1}{Siz{e}_B}\frac{P_B\left(t\right)}{V_{dc}\left(t\right)}\Delta t$$

(3)

$$I_{charging}=I_{predetermined}\frac{100-SOC(\%)}{100}$$

(4)

In [67], a hierarchical control approach is formulated for ESS in a DC microgrid to boost system reliability. The strategy is hierarchical and combines both centralised and distributed control methods. The control strategy proposed in the referenced paper involves a centralised approach, which generates power references through iteration, aiming to maximise the ESS ramp rates and power capacities. In addition, secondary voltage regulation is employed to minimise the DC bus voltage deviation, while autonomous SOC recovery is utilised to reduce the slack terminal SOC variation. The DC bus voltage is considered a key indicator for power balance, and droop relationships are integrated into a unique distributed control algorithm to manage cases of communication loss. The system’s net power breakdown and ESS power dispatch are enabled through localized low-pass filters, while SOC recovery is achieved by adjusting the slack terminal threshold voltage. The proposed hierarchical control strategy demonstrated higher system reliability, faster response time, and lower system control accuracy compared to centralised control. Reference [68] proposes a hierarchical control architecture for islanded DC microgrids that consists of three distinct layers: primary, secondary, and tertiary. The primary layer is composed of decentralised voltage controllers that are connected to DGUs. The secondary layer translates power references into appropriate voltage signals for the primary layer by solving an optimisation problem. Finally, the tertiary layer integrates a MPC-based energy management system to provide an overall control framework for the microgrid. The effectiveness of the proposed control framework is evaluated through simulations conducted on a modified 16-node DC microgrid, demonstrating its proficiency in ensuring reliable power supply to the connected loads.

Hierarchical control is well-suited for use in DC microgrids because it provides a flexible and scalable approach to controlling the microgrid’s operation. Additionally, hierarchical control allows for the implementation of advanced control algorithms, such as machine learning algorithms, which can improve the performance of the microgrid.

4.2.3. Fuzzy Control

Fuzzy control, which employs fuzzy logic principles, is utilised in DC microgrids to enhance the stability and performance of the system. Fuzzy control is a type of non-linear control method that can handle complex and uncertain systems, making it well-suited for use in microgrids where there are multiple energy sources and loads, and where the system’s behaviour is subject to change.

In a fuzzy control system, the control inputs are represented as fuzzy sets, and the control decisions are made based on a set of fuzzy rules that are derived from expert knowledge or data analysis. The fuzzy control system is designed to mimic the reasoning of a human expert, making it well-suited for use in microgrids where the system’s behaviour is difficult to model or predict. In [69] a novel control scheme for DC microgrids with RESs and an ESS based on fuzzy logic is introduced. Unlike conventional methods that require a sophisticated mathematical model of the system, the proposed scheme’s straightforward design makes it possible to implement without such a model. The fuzzy logic controller is capable of minimising power-sharing errors between the RESs while regulating the bus voltage under various operational conditions and network topologies.

In the field of DC microgrids, fuzzy control is a popular method for regulating power flow between various energy sources and loads. A fuzzy control system can be implemented to govern the power output of photovoltaic (PV) systems, as well as to oversee the charging and discharging of energy storage systems (ESS) based on the microgrid’s voltage and load levels. In [70], an event-based distributed fuzzy control approach is suggested for DC microgrids equipped with DC/DC converters. It uses a Lyapunov function and convexification approaches to derive a controller design in terms of LMIs. The result of the proposed method uses fewer communication resources and stabilises the DC microgrid while achieving voltage tracking synchronisation. In [71], a non-fragile fuzzy controller is proposed for DC microgrids with a single CPL to enhance the system’s transient stability by enabling power circulation between the DC link and an ESS. The proposed controller outperforms existing methods, as evidenced by the improved settling time and reduced oscillation and overshoot against uncertainties.

Fuzzy control is a flexible and scalable control method that can be adapted to the specific requirements of the microgrid. Additionally, fuzzy control can be integrated with other control techniques, such as droop control and hierarchical control, to enhance the overall performance of the DC microgrid. A fuzzy-MPC that can be used to stabilise a DC microgrid networked with a power buffer is discussed in [72]. The use of two network delay compensators ensures robustness against network delays and reduces the computational burden. A DC microgrid feeding a CPL showed improved robustness against network delays and practical feasibility, as demonstrated by hardware-in-the-loop simulations.

Overall, fuzzy control is a powerful control method that has the ability to upgrade the performance, stability, and reliability of DC microgrids.

5. Protection in DC Microgrids

Any power system must prioritise protection, and DC microgrids are no different. The advantages of DC microgrids include greater efficiency, better power quality, and improved reliability. These advantages, meanwhile, may only be realised if the system is protected against faults, overloads, and other disruptions. Fault isolation, location, and detection are a few of the numerous protection strategies employed in DC microgrids. A few difficulties are also encountered when creating protection plans for DC microgrids, and solutions to these difficulties are discussed in

Table 4. Protection schemes in DC Microgrid.

Reference No.	Protection Type	Methodology	Remarks
[73]	Superimposed Current	During a fault, superimposed currents from a line segment’s two ends are exhibited on the i-plane to determine a protective measure	Fast fault detection; reliable selectivity
[74]	Travelling Wave	The characteristics of the waveform and the direction of the first travelling wave are measured in the local vicinity following the occurrence of a fault.	Quicker than existing techniques; works without communication; operates as primary and backup protection; resilient against high-resistance faults
[75]	Overcurrent	Blocking and inter-tripping schemes are implemented using the communication infrastructure of smart grids to select the appropriate relays for protection.	Reduced cost compared to differential protection; offers a secondary layer of protection for relays; overcomes coordination issues by utilising communication infrastructure.
[76]	Centralised	Utilises the current of each segment of a line to evaluate the similarity of the current alteration at both ends of the segment. Based on this analysis, the protection decision is made.	Sturdy and resistant to communication delays and link failures; determines accurate defaulted segments during communication channel fail; fault location time less than 3ms
[77]	Differential Current	A revised cumulative sum average technique is utilized for fault-type classification, whereas the non-iterative Moore–Penrose pseudo-inverse method is used to determine fault distance.	Swift and accurate determination of faults and their distance; suitable for a variety of faults
[78]	Fuse based	Framework calculates current and capacitance needed for proper fuse operation using I2t value. Capacitance limits set for selectivity: minimum for VBCs, maximum for LCs. Experimental validation was carried out.	Basic; low cost; standardized; lower losses; reliable protection method; capacitance upgrade may increase system cost in some cases.
[79]	Centralised	The proposed method employs a communication-assisted approach for detecting faulty zones and self-healing to facilitate network restoration. Differential-based protection can be implemented using the smart grid’s communication infrastructure.	Minimum DCCBs for selective protection; self-healing for network restoration; real-time simulations for performance analysis
[80]	Localised	A protection relay for current limiting purposes is developed, which relies on the characteristics of current and voltage. The estimation of fault resistance is formulated using power-sharing methodology to achieve greater precision in fault detection.	Sturdy protective plan; resistant to interferences; effective and dependable fault detection and resistance estimation with precision and tolerable margin of error; no communication latency implicated.
[81]	Least-Square based	The proposed method uses the LS technique for fault direction identification and communication between IEDs for internal fault identification. Results show method accuracy for various fault conditions.	Precision for various fault resistances and close-in-fault conditions; evaluated on a reduced-scale hardware configuration; enhanced performance for high-resistance faults.
[82]	Oscillation Frequency and Transient Power-based Unit Protection	The first cycle’s frequency and transient power are used to detect faults and identify faulted sections in a DC microgrid. A relay coordination scheme is proposed, and communication failure can be averted using local data.	Precise for various fault categories and two-way power transmission scenarios; local data application avoids relay malfunction from communication breakdown
[83]	Localized Non-Unit Protection	The scheme uses first and second-order current derivatives to detect low and high-impedance faults and discriminate between different faults. Analytically calculated thresholds are used in the algorithm.	Straightforward protective scheme; absence of communication latency; no communication delay
[84]	Zone-IED based	In a ring-bus-based microgrid system, the suggested protection technique employs zone IEDs to detect fault current and isolate the bus segment. To increase system dependability and MTBF, a fault-location algorithm utilising a probe power unit is given.	Fault identification; fault isolation and location; enhanced system dependability; fault diagnosis prior to reclosure

.

5.1. Fault Detection

Detecting and identifying abnormal conditions in the system is called fault detection. Monitoring the current and voltage data at various locations in the infrastructure is usually the way fault detection is accomplished in DC microgrids. The most frequent faults in DC microgrids are short-circuiting, ground, and overcurrent failures.

When a higher current runs via a system’s low-resistance pathway, a short-circuit fault occurs. Ground faults occur when a fault current flows to the ground through a path other than the intended current path. Overcurrent faults occur when the current in a circuit goes beyond its rated value.

There are different techniques that can be used to detect faults in DC microgrids, such as waveform analysis, time-domain analysis, and frequency-domain analysis. Monitoring the current and voltage waveforms for changes indicating a fault is called waveform analysis. Time-domain analysis involves examining the time-domain characteristics of the signals to detect fault conditions. The frequency-domain analysis involves examining the frequency content of the signals to detect faults.

The effective management of fault occurrences in DC microgrid systems is crucial for maintaining normal operation and providing stability support to the main grid, while also ensuring rapid recovery performance. As previously stated, the fault management of direct current microgrids include the processes of fault detection, fault identification, fault placement, fault isolation, and fault reconfiguration. The fault detection and management are the two major functions that can be used to handle the faults of a DC system [85]. A more rapid and intelligent fault detection method with a suitable grounding design is necessary for DC systems protection. The following are some of the recent fault detection and classification techniques viz safety/protection methods, model-based methods, methods based on artificial intelligence and machine learning, and methods based on signal processing [86,87]. The various fault detection techniques are clearly discussed in

Table 5. Example of Fault Detection in DC Microgrid.

Fault Detection Technique Type	Example Case	Key Characteristics and Contributions
A signal-processing based [88,89]	Discrete Wavelet Transform (DWT) method is implemented to detect any surges in DC Microgrid	To extract the feature vector from the DC signal and transfer it to the neural network during the surge, DWT is used. The feature is then encouraged to be related to a fault current surge caused by a short circuit that occurred across the DC link capacitor. It establishes the signal’s irregularity as it approaches fault detection in DC systems.
	Wavelet transform and S-tranform based method	Used to detect faults in grid-connected systems. During islanded mode of operation, it is used to examine voltage stability at the site of common coupling with a nonlinear load-connected
AI and ML based [90,91,92]	ANN based detection and location of a fault in DC microgrid	The faulty bus segments may be quickly identified and pinpointed using ANN technology. DC microgrids benefit from faster detection and isolation times without having to completely shut off power.
	Fuzzy Logic	Fuzzy Logic algorithm is used to find out if there is an imbalance in power or a change in voltage. DC microgrids work better because of this.
	Hybrid ANN-SVM	For the purpose of the protection of microgrids, it is utilised for the detection and localisation of faults. It is capable of teaching itself new things and learning on its own.

.

5.2. Fault Isolation

Fault Isolation is the method used to break off the DC microgrid’s faulty segment from the rest of the system. This detaching is necessary to contain the failure and prevent its spread and damage to other system components. In DC microgrids, circuit breakers or switches are commonly used for isolation.

The primary obstacle in isolation for DC microgrids is the absence of natural zero-crossing in the DC voltage. This absence makes it difficult to disconnect the faulted section without causing harm to the system. To overcome this challenge, some DC microgrids employ high-speed DC circuit breakers that can operate quickly and reliably.

5.3. Fault Location

Fault location is a crucial process in DC microgrids, as it helps in identifying the location of the faulted section, which is necessary for restoring power to the rest of the system. The most common technique used for fault location in DC microgrids is time-domain reflectometry (TDR). It works by transmitting a signal through the transmission line and measuring the reflection of the signal at the faulted section. By calculating the time delay between the transmission and the reflection, the distance to the faulted section can be determined.

When it comes to DC microgrid protection, designers face various challenges. One of the significant challenges is the lack of standardisation in DC microgrid technology. Since DC microgrids come in different topologies, voltage levels, and applications, designing protection schemes that work well for all systems can be difficult.

6. Conclusions

In conclusion, this comprehensive analysis of planning, functionality, control, and protection for DC microgrids underscores the pivotal role that sustainable energy solutions play in the contemporary energy landscape. This review paper highlights the following conclusions:

• Smart Grids are the future of traditional grid networks, and a better approach to actualise them involves advancements such as proper planning, design, and control in microgrids. This article only focuses on DC microgrids, which are gaining lot of advantages in terms of system efficiency, reliability and cost compared to the traditional use of AC microgrids or hybrid AC microgrids.
• This review addressed several critical DC microgrid concerns, including architecture, communication technologies, control structure, and EMS classifications.
• Nonetheless, a significant amount of information is provided on fault classification and analysis. The strategies widely used to manage fault in DC microgrids include signal processing-based fault detection, wavelet-based fault detection, background noise-based fault detection, and several other methods.
• However, still the proper fault detection in DC microgrids becomes the primary focus of studies after increasing dependence on the use of renewable sources, in accordance with the characteristics of DC microgrids during fault scenarios.
• Further research is required to study the impact of different control systems on the overall performance of DC microgrids and to develop advanced control algorithms to enhance the efficiency and reliability of the system.