**1. Introduction**

Wind diesel isolated microgrids (WDIMs) are microgrids that combine wind turbine generators (WTGs) with diesel generators (DGs) to supply electricity to remote consumers. WDIMs are in many cases the retrofitting of an existing isolated diesel microgrid with WTGs when there is an available wind resource at the diesel microgrid location. By means of the WTG-supplied power, the DG-demanded power is reduced so that the fuel consumption and the CO2 emissions are also reduced. Figure 1 shows a WDIM scheme where the main components of a WDIM, namely the wind turbine generator (WTG), diesel generator (DG) and consumer load, can be seen. Only one DG and one WTG are shown for simplicity, but WDIMs can include several DGs and/or WTGs. The DG consists of a diesel engine (DE) driving the rotor of a synchronous machine (SM). The DE converts fuel energy into shaft mechanical energy by the combustion of the fuel in the DE cylinders. The DE speed is controlled by the DE speed governor which actuates on the fuel rate incoming into the DE cylinders to control the DE-produced mechanical power. The DE mechanical power is converted into electrical power by the SM, which also provides the sinusoidal grid voltage waveform of frequency f and amplitude V. The SM output voltage amplitude V is controlled by an automatic voltage regulator by regulating the SM reactive power injected in the microgrid. The WTG consists of a wind turbine (WT) driving the rotor of an electrical generator. The WT converts the wind power into shaft mechanical power. The WT mechanical power is converted into electrical power by the WTG electrical generator. The WTG electrical generator can be connected directly to the grid or through a power converter. The dump load (DL) consists of a resistor bank connected to the grid through power switches or an electronic power converter. The DL behaves as a controlled active

power consumer. The energy storage system (ESS) consists of short-term energy storage technology suitable for grid applications [1] (based on flywheels, batteries, ultracapacitors, etc.) connected to the grid through an electronic power converter. The ESS behaves as a controlled active power producer/consumer. The reactive power block supplies reactive power to the isolated grid. It can be a synchronous condenser, a static VAR compensator or integrated into the ESS electronic power converter.

**Figure 1.** Wind diesel isolated microgrid scheme.

## **2. WDIM Operation Modes**

All WDIMs have two modes of operation: diesel only (DO) and wind–diesel (WD) [2,3]. High-penetration WDIMs are capable of working with the DGs not running working mode known as Wind Only (WO)). The three modes of operation and the conditions to transition among them are shown in Figure 2.

In DO mode, the WDIM behaves as an isolated diesel power plant, and all the consumers' active and reactive power demands are supplied by the diesel generators (WTGs are disconnected in DO mode, IT = off in Figure 1). The regulation of the frequency is performed by the DE speed governors. The regulation of the voltage is performed by the SM voltage regulators.

The WTGs can supply active power with wind speeds above the WTG cut-in speed and below the cut-off speed. As Figure 2 shows, when the wind speed is inside the previous limits during a predefined TWD time interval to confirm enough wind resource, the WDIM control orders the connection of the WTGs, performing a DO-to-WD mode transition. Conversely, the WDIM control orders the disconnection of the WTGs when the wind speed is outside of the cut-in/out limits (WD-to-DO transition).

In WD mode, both WTGs and DGs supply active power. The regulation of the WDIM voltage and frequency is performed as in DO mode. The DE governor can only control frequency if the DE produces positive power, and a DL is required to guarantee this constraint. The WTG-produced active power *PT* can exceed the power consumed by the load *PL* (*PT* > *PL*), and if there is no means in the WTG to reduce its power (e.g., by varying the blade pitch), the WDIM control must order the DL to dump the necessary active power *PD* to keep the DG power positive *PG* and thus to avoid DG reverse power. Therefore, this DL use increases the WDIM reliability. The ESS can also load the system in WD mode to keep the DG power positive when *PT* > *PL* [4], with the advantage that the WTG excess power stored in the ESS can later be mostly recovered. ESSs can provide fast reserve power [5] that can be used to reduce the spinning reserve needs in both DO and WD modes. Additionally, for DO/WD modes, ESSs can improve the efficiency of DGs by increasing DG loading (low-loaded DGs have low performance).

**Figure 2.** Wind diesel isolated microgrid (WDIM) operation modes and transitions between modes.

As Figure 2 shows, if the WTG-produced power exceeds the load consumption (*PT* > *PL*) (increased by a security factor k) during a predefined TWO time interval, to confirm the WTG active power excess, and the WDIM is a high-penetration one, the WDIM control orders the DGs to stop, performing a WD-to-WO mode transition.

In WO mode, the active power is supplied only by the WTGs. In this mode, as the DGs are not running, no fuel is consumed and the WDIM voltage and frequency control must rely on auxiliary components. In a no-storage WDIM, the condition for WO mode is that the WTG-produced power *PT* must exceed the load-consumed power *PL* plus the system losses constantly. The active power generated by the WTGs ranges from partly controlled to totally uncontrolled depending on the WTG type. As the consumer load is uncontrolled, the DL is ordered to consume the WTG power excess to match generation with consumption to regulate WDIM frequency. When the WO mode condition is not met (*PT* < *PL*), there will be an active power deficit that will be detected by a system frequency falling, and the WDIM must transition from WO to WD, as Figure 2 shows. The WDIM control will order the DG to start and connect it to the isolated grid. Once the DG is connected, the WDIM will be in WD mode, and the DG will supply power to balance the WDIM active power to restore the frequency to its rated value [6]. If an ESS is included in the WDIM, the WTG power excess can be stored in the ESS or dumped in the DL, or both actions can be taken in a coordinated way. In addition, when the current load is above the WTG power, the ESS can supply power *PS* up to its rated power *PS-NOM* (temporally until the ESS is discharged). In a WDIM with an ESS, the condition for WO mode is that the WTG-produced power *PT* plus *PS-NOM* (ESS rated power) must exceed the load consumption *PL* (neglecting the losses). As Figure 2 shows, if the previous condition *PT* + *PS-NOM* > *PL* is not satisfied, the WDIM control will start and connect a DG [7], changing from WO to WD mode.

If an SM is running and connected to the grid in WO mode, the SM will generate the microgrid voltage, and the voltage regulation will be performed by the SM voltage regulator as in DO and WD modes. The WDIM active power active unbalances will be translated into frequency variations [8]. If the voltage waveform is generated by a voltage source inverter (VSI) which connects either a WTG or an ESS to the grid, and the VSI works with constant frequency, the WDIM active power unbalances will be translated into voltage variations [9].

#### **3. WDIM Dynamic Simulation Review**

WDIM simulations can be of two types: logistic and dynamic. The logistic simulations allow long-term WDIM simulations in terms of energy balance and can be used to test a particular WDIM architecture and for the sizing of the different components of the WDIM. Examples of logistic simulation software are Hybrid2 [10] and Homer [11]. The input data to these programs are the WDIM architecture, the estimated load consumption and the WDIM location wind speed data. These logistic simulation tools produce as output important results such as the fraction of energy that the WTGs supply, the fuel savings achieved, the total DG run time, the number of DG starts and the cost of energy. There are also articles that deal with the generator sizing of a specific WDIM, such as [12], with the aim in this case of optimizing the dynamic power-sharing between the DG and WTG.

WDIM dynamic simulations allow short-term simulations in order to obtain detailed electrical variables waveforms, such as system voltage and frequency and active and reactive power in the different elements, so that the WDIM power quality and stability can be tested. For WDIM dynamic simulations, the most used software are PSCAD [13], which is focused on power system modeling and simulation, and Matlab–Simulink along with the Simscape Electrical [14] library (Simscape Electrical includes power system modeling). The RPM-SIM Simulator [15] has also been used in WDIM dynamic simulation.

In the dynamic simulation of big power systems, the system frequency is assumed to be constant, and this assumption allows the use of low-order dynamic models for the electrical machines in big power system modeling. WDIMs are low-inertia isolated power systems, and therefore, the WDIM frequency can suffer strong variations [7]; thus, according to [16], low-order electrical machine models should not be used in WDIM modeling. Therefore, to obtain precise voltage waveforms in the dynamic simulations, the WDIM modeling must use high-order-type electrical models for the electrical machines included in the WDIM (these high-order models are available on both Simscape Electrical and PSCAD).

This section classifies articles according to several factors, which are dealt with in the following subsections.

#### *3.1. WTGs Included in WDIMs*

The three types of generator systems most used in WTGs [17] are shown in Figure 3. The low-cost one is the WTG that equips a squirrel-cage induction generator (SCIG), which has the stator directly connected to the grid (Figure 3a). This combination of a WT driving a SCIG is called a constant-speed WTG, because for generator operation the SCIG speed range is very narrow, typically between 1.01 and 1.02 the synchronous speed [18]. This WT–SCIG set does not allow the variation of its speed to optimize the wind energy capture [17], but it has remarkable features for the remote locations of WDIMs, such as robust construction and simple maintenance [19]. Since the SCIG consumes reactive power, a capacitor bank is added for reactive power compensation. Additionally, the SCIG-produced torque is proportional to the SCIG slip in the WT–SCIG working speed range [20], and due to the SCIG being directly connected to the isolated grid, the WTG inertia increases the system inertia. These two WT–SCIG features provide significant frequency support, moderating system frequency deviations [21]. This WT–SCIG type is used in the WDIM simulations of [22–29].

**Figure 3.** SCIG, DFIG and PMSG Wind Turbine Generators types.

Variable-speed WTGs allow the optimization of the wind energy capture and are mainly of two types [17]: the double-feed induction generator (DFIG) (Figure 3b) type and the permanent magne<sup>t</sup> synchronous generator (PMSG) type (Figure 3c).

The DFIG type has its stator directly connected to the grid, and its rotor is connected through a slip ring to an AC–DC–AC converter to the grid. This rotor converter controls the rotor frequency and therefore the rotor speed that can be varied in a range of ±30% around the synchronous speed [18]. The rotor converter rated power is only 25–30% of the DFIG rated power, which is an advantage when compared with other variable-speed WTG types. The slip rings need periodic maintenance, which is a drawback for the remote sites of WDIMs. This WT–DFIG type is used in the WDIM simulations of [30–32]. The SCIG and DFIG WTG types need a gearbox to adjust the low WT speed (10–15 rpm) to the high IG rotor speed (for example, 750 rpm for a four-pole-pair IG and 50 Hz grid frequency). The gearbox has friction losses that decrease the WTG efficiency, and it needs periodic maintenance.

The third WTG type considered consists of a WT that directly drives (no gearbox) a PMSG connected to the grid through an AC–DC–AC double converter. The gearbox disadvantages are avoided, but as the SG rotates at the low WT speed, the SG must produce high torque to deliver the rated power; therefore, a larger size of the generator is needed [17]. The double converter makes the WT–SG rotor speed independent from the grid frequency and allows a rotor speed range from 0 to rated one. The double converter rated power has to be equal to the PMSG rated power as all the WTG-produced power has to pass through the converter, so the converter losses are greater than in the WT–DFIG type. In [33–35], the simulated WDIMs include a WT–PMSG type. In the WDIM of [36], the WT drives an electrically excited SG instead of a PMSG. Another interesting feature of the WT–PMSG type is that the DC–AC grid power converter can generate the isolated grid AC voltage and regulate the AC voltage amplitude and frequency, making the WO mode architecture simpler.

Variable-speed operation in WTs has additional advantages such as the reduction in torque peaks in the gearbox and shafts in the DFIG–WTG type and the possibility of using the kinetic energy of the blades to absorb wind power fluctuations, which is an interesting feature in WDIMs. Variable-speed WTGs can also support system frequency if this feature is included in their controllers [21]. Additionally, the AC–DC–AC converter of the variablespeed WTGs allows a battery to be connected to the DC side of the converter [37] so that the battery ESS (BESS) is embedded into the WTG. This embedded BESS can be used to improve the frequency support previously mentioned or to smooth the WTG power output.

#### *3.2. Use of Energy Storage in WDIMs*

The aim of the ESSs in a WDIM can be logistic, in order to reduce the start–stop cycles of DGs [38] and/or to improve the WDIM stability and power quality [8]. A significant reduction of the cycling of DGs can be achieved with an ESS with a rated power around the average WDIM load and with a storage time of several minutes, having an acceptable reduction with just 1 min of storage time [38]. In [39], the logistic modeling SW Hybrid2 [10] is used to determine an optimum storage time of 10–15 min for the average WDIM load to reduce the DG cycling. The rated power of WDIMs goes from tens of kW to MW, so this is also the power range of ESSs for WDIMs. In this section, short-term ESSs based on batteries (BESSs), supercapacitors (CESSs) and flywheels (FESSs) are briefly considered as these types comply with the ESS rated power and storage time requirements previously mentioned. Additionally, BESSs, CESSs and FESSs are compared in terms of cycle life, as any ESS in WDIMs must withstand a high number of charging–discharging cycles with variable depth of discharge as the ESS must continuously smooth the wind power, and wind power is of random nature. The structures of the three ESS types considered are presented in Figure 4.

**Figure 4.** Structures of energy storage systems (ESSs) based on (**a**) batteries (BESSs), (**b**) supercapacitors (CESSs) and (**c**) flywheels (FESSs).

Battery energy storage systems (BESSs) (Figure 4a) consist of a battery bank, a threephase bidirectional power converter to convert the DC battery voltage into the AC grid voltage and additional filters to smooth the battery current and to limit the grid harmonic injection. The power converter works as an inverter/rectifier to discharge/charge the battery to/from the grid. The modeling of one of these converters working in current-controlled mode can be seen in [40]. In Figure 4a, an elevator transformer matches the battery voltage with the AC grid voltage. This solution is feasible as batteries change voltage little when they discharge, but other solutions instead use a DC–DC buck–boost converter to elevate the battery voltage up to the needed DC input voltage in the inverter [41]. Reference [42] shows different dynamic BESS models. The battery types that have been used in WDIMs are Pb–acid, Ni–Cd, Ni–MH and Li-ion [43]. Reference [44] presents an electrical model for batteries consisting of a variable voltage source in series with a constant resistor. The DC variable voltage source follows the volts–state of charge (SOC) battery discharge curve. Reference [45] improves a previous model by proposing different resistances for charging and discharging, these resistances being a function of SOC. Battery SOC calculation is not simple. Reference [46] proposes battery SOC estimation, integration of the battery current and tracking of battery voltage variations. In [47], the Volterra models are applied to calculate battery SOC. The relationship between the battery number of cycles and the depth of discharge is that of an exponential decrease [48]. The maximum number of cycles for a battery depends on the battery type. WDIM simulations including a BESS in their architecture can be found in [4,19,34,40,49].

Supercapacitor energy storage systems (CESSs) (Figure 4b) consist of a supercapacitor bank, a bidirectional DC–DC converter, a DC–AC three-phase bidirectional power converter and a grid connection filter. Unlike batteries, supercapacitors present significant voltage drop as they discharge, so a DC–DC converter is needed in this case [50] to adapt the supercapacitor DC voltage to the DC voltage needed at the input of the DC–AC grid converter. The simplest electrical model for supercapacitors consists of a series of a capacitor and a resistance along with a parallel resistance which takes account of the autodischarge process. Compared to batteries, supercapacitors have higher specific power (W/kg) and much higher cycle life and allow simpler measurement of the state of charge (by means of the supercapacitor voltage), but they have lower specific energy (Wh/kg) and much higher cost. Reference [51] includes a CESS in its WDIM simulations. By combining the high power density of supercapacitors with the high energy density of batteries, an optimum ESS is achieved [52]. In [35], the WDIM includes a mixed CESS–BESS.

FESSs (Figure 4c) consist of a rotating flywheel, an electrical machine that drives the flywheel and a double AC–DC–AC power converter to connect the electrical machine to the AC grid. Filters are needed to connect the double converter to the grid and to the electrical machine. A capacitor bank is connected to the DC side of the double converter. The energy is stored in the kinetic energy of the flywheel, and the electrical machine acting as a generator/motor converts the kinetic energy into electrical energy and vice versa. The bidirectional electrical machine side converter transforms the DC capacitor bank voltage into AC voltages for the electrical machine. The bidirectional grid side converter transforms the AC grid voltages into the DC voltage of the capacitor bank. The machine converter controls the electrical machine torque in order to maintain a constant DC voltage in the DC-link, so the control of the grid converter works in the same way as a battery converter. Compared to batteries, flywheels have higher cycle life and higher specific power (W/kg) and allow simpler measurement of the state of charge (only flywheel spinning speed is needed), but they have lower specific energy (Wh/kg) and much higher cost.

Reference [53] deals with the modeling of the FESS double power converter. WDIM simulations which include a FESS can be found in [24,25,53,54].

Another type of short-term ESS that has been used in WDIM simulations is superconducting magnetic energy storage (SMES). Reference [55] uses SMES in the presented WDIM. A review on SMES and its potential applications in power systems can be seen in [56].
