Reactive Power Management in Renewable Rich Power Grids: A Review of Grid-Codes, Renewable Generators, Support Devices, Control Strategies and Optimization Algorithms

Power electronic converter (PEC)-interfaced renewable energy generators (REGs) are increasingly being integrated to the power grid. With the high renewable power penetration levels, one of the key power system parameters, namely reactive power, is affected, provoking steady-state voltage and dynamic/transient stability issues. Therefore, it is imperative to maintain and manage adequate reactive power reserve to ensure a stable and reliable power grid. This paper presents a comprehensive literature review on the reactive power management in renewable rich power grids. Reactive power requirements stipulated in different grid codes for REGs are summarized to assess their adequacy for future network requirements. The PEC-interfaced REGs are discussed with a special emphasis on their reactive power compensation capability and control schemes. Along with REGs, conventional reactive power support devices (e.g., capacitor banks) and PEC-interfaced reactive power support devices (e.g., static synchronous compensators) play an indispensable role in the reactive power management of renewable rich power grids, and thus their reactive power control capabilities and limitations are thoroughly reviewed in this paper. Then, various reactive power control strategies are reviewed with a special emphasis on their advantages/disadvantages. Reactive power coordination between support devices and their optimal capacity are vital for an efficient and stable management of the power grid. Accordingly, the prominent reactive power coordination and optimization algorithms are critically examined and discussed in this paper. Finally, the key issues pertinent to the reactive power management in renewable rich power grids are enlisted with some important technical recommendations for the power industry, policymakers, and academic researchers.

Nomenclature

AbbreviationExpansion

Acronyms:
ADP:	Approximate dynamic programming
AEMO:	Australian energy market operator
AG:	Asynchronous generator
ANN:	Artificial neural network
CASDM:	Current-mode asynchronous sigma-delta modulation
CMC:	Current mode control
CSC:	Current source converter
CVP:	Control vector parameterization
DFIG:	Doubly-fed induction generator
DPC:	Direct power control
DSO:	Distribution system operator
DSPWM:	Digital sinusoidal pulse-width modulation
DVR:	Dynamic voltage restorer
EA:	Evolutionary algorithm
EMTDC:	Electro-magnetic transient design and control
ESS:	Energy storage system
FACTS:	Flexible AC transmission system
FCWG:	Full-converter wind generator
FRT:	Fault ride-through
FSWG:	Fixed-speed wind generator
GA:	Genetic algorithm
GDA:	Gradient descent algorithm
GSC:	Grid-side converter
HVRT:	High voltage ride-through
LVRT:	Low voltage ride-through
MIMO:	Multiple-input and multiple-output
MRAS:	Model reference adaptive system
MPC:	Model predictive control
MAPSO:	Multi-agent based particle swarm optimization
MINLP:	Mixed-integer non-linear programming
MSC:	Machine side converter
NER:	National electricity rules
NLP:	Non-linear programming
NSGA:	Non-dominated sorting GA
OLTC:	On-load tap changer
OPF:	Optimal power-flow
OSLC:	Online supplementary learning control
PAC:	Pitch-angle control
PCC:	Point of common coupling
PEC:	Power electronic converter
PIR:	Proportional integral resonant
POC:	Point of connection
POI:	Point of interconnection
PMSG:	Permanent magnet synchronous generator
PMU:	Phasor measurement unit
PSO:	Particle swarm optimization
PSS:	Power system stabilizer
PSCAD:	Power system computer aided design
PWM:	Pulse-width modulation
REG:	Renewable energy generator
RPO:	Reactive power optimization
RSC:	Rotor-side converter
SCIG	Squirrel-cage induction generator
SMES:	Superconducting magnetic energy storage
SMC:	Sliding mode control
STATCOM:	Static synchronous compensator
SVC:	Static VAR compensator
SVM:	Space-vector modulation
TCR:	Thyristor-controlled reactor
TCSC:	Thyristor controlled series compensator
TSO:	Transmission system operator
ULTC:	Under-load tap changing
UPFC:	Unified power-flow controller
VSWG:	Variable-speed wind generator
VSC:	Voltage source converter
WRIG:	Wound rotor induction generator
Variables:
${S}$ :	Apparent power
${P}$ :	Real power
${Q}$ :	Reactive power
${V}$ :	Voltage
${I}$ :	Current
${E}$ :	Electromotive force
${s}$ :	Slip
${X}$ :	Reactance
$\delta$ :	Angle between stator voltage and internal emf
$\phi$ :	Angle between voltage and current
Subscripts:
PV:	photovoltaic
${R}$ :	rotor
${S}$ :	stator
${T}$ :	total
${g}$ :	grid
${c}$ :	converter
${o}$ :	output
pcc:	point of common coupling

SECTION I. Introduction

Due to the global drive towards renewable and sustainable energy systems, power electronic converter (PEC) interfaced renewable energy generators (REGs), such as wind generators and solar-PV systems have widely been adopted in power networks around the world. The Kyoto Protocol was one of the major catalyst for the global drive towards renewable energy generation [1]. In addition, due to the technical advances in the PECs and the electronic materials, manufacturing cost of REGs have significantly reduced during the past decade while encouraging wide-scale adoption of PEC interfaced REGs in power networks [2]. In 2016, REGs were accounted for the two thirds of the new generation added to power networks, and approximately 165 GW of renewable power generation capacity was added to power networks around the world. Among renewable energy sources, the solar-PV capacity was grew by 50%, while exceeding the total installed capacity beyond 74 GW, which is higher than the net annual growth in coal power generation [3]. Figure 1 illustrates the electricity capacity additions by fuel type for 2016.

FIGURE 1.

Electricity capacity additions by fuel type for 2016 [3].

Majority of the developed countries have set renewable energy targets in power generation, and also making policy directives to achieve these targets. For example, Australian government has set a renewable energy generation target of 33,000 GWh by 2020, which constitutes approximately 23.5% of total power generation [4]. During the last 10 years there was an average annual renewable energy growth of 5.3% in Australia [5]. Among the major renewable energy sources, solar-PV and wind generation had tremendous average annual growth rates of 59.3% and 23.5% respectively [5]. The Australian electricity generation share, by fuel-type is shown in Table 1.

TABLE 1 Australian Electricity Generation by Fuel Type

At the early stage of renewable energy integration, REGs could be either connected or disconnected from the power grid without significant impact on grid stability, due to their low penetration level. However, with the increased renewable power generation, it is no longer possible to connect or disconnect REGs at operators’ discretion, since it would adversely affect the power system stability and reliability. Therefore, requirements for grid integration of REGs are now being strictly stipulated in grid codes [6]–​[11]. Reactive power compensation and voltage stability have become major concerns for utility grid operators with significant renewable power penetration. Consequently, reactive power requirements are now becoming mandatory for REGs (e.g. wind farms [12]). Major blackouts were also caused due to voltage instability as a consequence of insufficient reactive power reserve in power networks [13].

With the large-scale integration of renewable energy sources to the power grid, reactive power reserve would decrease as they displace the conventional synchronous generators, and hence power grids are becoming more vulnerable for instability. Moreover, because of the intermittent and variable nature of some renewable energy sources (e.g. variable solar irradiation and wind speed), power system likely to become unstable during system contingencies. In transient fault conditions, without proper reactive power support mechanisms, the low inertial wind turbines, and the inertia-less solar-PV systems are unable to provide sufficient voltage support to the grid [14]. Furthermore, due to long distance between the load centers and large-scale REGs (e.g. MW-scale wind farms) transmission corridors likely to become unstable during system contingencies due to lack of reactive power support to stabilize the voltage [15]. Therefore, it is imperative to review reactive power management strategies reported in the literature for power grids with high renewable power penetration.

This paper presents a critical review of reactive power management in renewable rich power grids, with special emphasis on grid-codes, renewable generator capabilities, reactive power support devices, control strategies, and coordination & optimization algorithms. This paper is structured as follows: Section II explains the link between the power grid steady-state/ transient performance and reactive power, Section III delineates the reactive power grid-code compliance requirements set by the grid operators for REGs, Section IV analyses the reactive power capability and control schemes for various wind generator types, reactive power capability of solar-PV systems and other REGs are discussed in Section V, reactive power support devices and their capabilities are discussed in Section VI, the control strategies developed for reactive power management with REGs are discussed in Section VII, reactive power coordination and optimization strategies are summarized in Section VIII, in Section IX a case study is presented on reactive power management in a distribution feeder with solar-PV systems, in Section X, the key findings of the review are enlisted with some important technical recommendations for the power industry and policymakers, and finally, conclusions of the review are summarized in Section XI.

SECTION II. Reactive Power and Power Grid Performance

Reactive power plays an important role in power grid, particularly power grid voltage management and stability. This section presents the active and reactive power relationships with network voltage, and also delineates influence of reactive power on network stability.

A. Reactive Power vs Grid Voltage

To find the relationship of active and reactive power with the grid voltage, lets assume a Thevenin’s equivalent circuit of a node-k power system (see Figure 2). The apparent power can be calculated from the relationship, $S= V_{k}I_{k}^{*}$ , where $I_{k} = \frac {V_{k} \angle \delta _{k} - E_{Th} \angle 0}{Z_{Th} \angle \theta }$ and real and reactive power can be derived as:

$$\begin{equation*} S_{k} = V_{k}I_{k}^{*} = \left ({\frac {V_{k}V_{k}}{Z_{Th}} }\right)\angle \theta - \left ({\frac {V_{k}E_{Th}}{Z_{Th}}}\right) \angle (\theta + \delta _{k}) \tag{1}\end{equation*}$$ Real power: $$\begin{equation*} P_{k} = \left ({\frac {V_{k}V_{k}}{Z_{Th}} }\right) \cos {(\theta)} - \left ({\frac {V_{k}E_{Th}}{Z_{Th}}}\right) \cos {(\theta + \delta _{k})} \tag{2a}\end{equation*}$$ Reactive power: $$\begin{equation*} Q_{k} = \left ({\frac {V_{k}V_{k}}{Z_{Th}} }\right) \sin {(\theta)} - \left ({\frac {V_{k}E_{Th}}{Z_{Th}}}\right) \sin {(\theta + \delta _{k})} \tag{2b}\end{equation*}$$

FIGURE 2.

Thevenin equivalent circuit of a node-k power system.

Now, for small excursions from the nominal voltage, $\partial V_{k}$ , change in real and reactive power can be found as:

$$\begin{align*} \frac {\partial P_{k}}{\partial V_{k}}=&\left ({\frac {2V_{k}}{Z_{Th}}}\right) \cos {(\theta)} - \left ({\frac {E_{Th}}{Z_{Th}} }\right) \cos {(\theta + \delta _{k})}\tag{3a}\\ \frac {\partial Q_{k}}{\partial V_{k}}=&\left ({\frac {2V_{k}}{Z_{Th}}}\right) \sin {(\theta)} - \left ({\frac {E_{Th}}{Z_{Th}} }\right) \sin {(\theta + \delta _{k})}\tag{3b}\end{align*}$$

For small change in phase angle $\delta _{k}$ , active and reactive power relationships would become:

$$\begin{align*} \frac {\partial P_{k}}{\partial \delta _{k}}=&\left ({\frac {V_{k}E_{Th}}{Z_{Th}} }\right) \sin {(\theta + \delta _{k})}\tag{4a}\\ \frac {\partial Q_{k}}{\partial \delta _{k}}=&- \left ({\frac {V_{k}E_{Th}}{Z_{Th}} }\right) \cos {(\theta + \delta _{k})}\tag{4b}\end{align*}$$

In transmission systems, the reactance $X_{Th}$ is much greater than $R_{Th}$ projecting $\theta$ close to 90°. That is, $\cos {(\theta)} \approx 0$ , $\sin {(\theta)} \approx 1$ , $\cos {(\theta + \delta _{k})} \approx - \sin {(\delta _{k})}$ , and $\sin {(\theta + \delta _{k})} \approx \cos {(\delta _{k})}$ and for small changes of nominal voltage, $V_{k} \approx E_{Th}$ , $\cos {(\delta _{k})} \approx 1$ , and $\sin {(\delta _{k})} \approx \delta _{k} \approx 0$ . Based on these approximations:

$$\begin{align*} \frac {\partial P_{k}}{\partial V_{k}}\approx&0; \quad and \quad \frac {\partial P_{k}}{\partial \delta _{k}} \approx \left ({\frac {V_{k}E_{TH}}{Z_{Th}} }\right) \tag{5a}\\ \frac {\partial Q_{k}}{\partial V_{k}}=&\left ({\frac {2V_{k}}{Z_{Th}}}\right) -\left ({\frac {E_{Th}}{Z_{Th}} }\right)\approx \left ({\frac {E_{Th}}{Z_{Th}} }\right);\quad and \quad \frac {\partial Q_{k}}{\partial \delta _{k}} \approx 0 \\ {}\tag{5b}\end{align*}$$

Equation (5a) indicates a strong relationship between real power, $P_{k}$ and phase angle $\delta _{k}$ , and equation (5b) shows a strong coupling between reactive power, $Q_{k}$ and voltage, $V_{k}$ . The X/R ratio of the transmission system is always high due to high reactance (X), and hence, reactive power injection is necessary to boost the voltage at the end of the line. However, in distribution systems, the X/R ratio is usually low, approximately closer to 1 for overhead lines, and therefore, reactive power injection does not necessarily boost the voltage, and hence active power injection becomes more feasible for voltage management.

B. MV-LV Distribution Feeder Voltage Management

Due to large-scale integration of REGs in power distribution networks, steady-state voltage management has become a major planning and operation issue in modern power networks. REGs (e.g. wind generators and solar-PV systems) rated less than 50 MW are connected to the MV network, while the REGs rated less than 10 kW (mostly solar-PV systems) are connected to the LV distribution feeders. As the distribution feeder voltage might increase beyond the maximum stipulated limit in certain time periods of the day (e.g. during 12 – 1 pm for distribution feeders with high solar-PV penetration), conventional voltage regulation approaches are infeasible for regulating distribution feeder voltage with high renewable penetration. In conventional distribution feeders (i.e. without REGs), the voltage decreases from the LV side of the distribution transformer towards the end of the feeder [16]. Consider the distribution feeder shown in Figure 3, where a load ($P+jQ$ ) is attached at the receiving end, and the sending-end voltage can be approximated as:

$$\begin{align*} V_{s}=&V_{r} + I (R+jX); \quad where ~I = \frac {P-jQ}{V_{r}^{*}} \\=&\left [{V_{r} + \frac {RP+XQ}{V_{r}}}\right]+j\left [{\frac {XP-PQ}{V_{r}}}\right] \tag{6}\end{align*}$$

FIGURE 3.

LV distribution feeder model.

For distribution networks, the phase-angle deviation is very small due to low reactance, and therefore the imaginary part of the equation (6) can be neglected and sending-end voltage can be approximated as:

$$\begin{align*} V_{s}=&V_{r} + \frac {RP+ XQ}{V_{r}} \\ V_{s} - V_{r}=&\frac {RP+ XQ}{V_{r}} \\ \Delta V=&\frac {RP+ XQ}{V_{r}} \tag{7}\end{align*}$$

Therefore, it is evident from equation (7) that voltage drop, $\Delta V$ , of the distribution feeder depends on the power factor of the connected load and the impedance of the distribution feeder. Contrarily, by injecting reactive power in the opposite direction to the active power, voltage drop could be mitigated [17]. The active power losses in the line can be determined as:

$$\begin{equation*} Active Power Losses = I^{2}R = \left [{\frac {P^{2}+Q^{2}}{V_{r}^{2}}}\right]R \tag{8}\end{equation*}$$

According to equation (8) both active and reactive power of the load contribute to active power losses. By improving the lagging power factor of the load, the voltage drop will decrease, contrarily by improving the leading power factor of the load, the voltage drop will increase. However, irrespective of the load operating as either lagging or leading power factor, system losses decrease with the improved load power factor, and hence, reactive power-flow in the distribution feeder must be minimized to reduce the line losses. Reactive power management and voltage regulation for MV-LV distribution feeders with REGs have been extensively researched in [17]–​[20]. However, with increasing REG integration into distribution feeders, the MV-LV feeder voltage management is still a vibrant field of research.

C. Reactive Power Influence on Voltage and Transient Stability

Power grid stability is defined as the ability of the power grid to regain equilibrium after occurrence of disturbances or faults in the power system [21]. Power grid stability issues can be classified into three types: 1) Rotor-angle stability, 2) Voltage stability, and 3) Frequency stability. Rotor-angle stability can be further subdivided into transient stability, and small-signal stability. Transient stability is defined as the ability of the power system to remain in synchronism after severe transient disturbances or faults, and electro-mechanical oscillations should be damped within a reasonable time-frame [21]. Therefore, transient stability mainly deals with the rotor-angle stability of synchronous generators in the network. Voltage stability is defined as the ability of power grid to restore the nominal voltage levels in all network nodes after any disturbance or transient condition [22]. During fault conditions or disturbances, both active and reactive power interact very closely, and their relationship becomes very complex [23].

When REGs are integrated to the power grid, significant portion of the synchronous generation is displaced without adequately compensating for the reactive power provided by the synchronous generators. Consequently, voltage control capability of the power grid reduces significantly. Moreover, during transient disturbances the inertia-less solar-PV systems, and very low inertial wind generators can not provide reactive power support to the same extent as synchronous generators, which destabilizes the grid leading to serious voltage control stability issues [14]. If adequate reactive power is not provided during the post-fault period, then the grid enters into an unstable state, and subsequently grid voltage will collapse leading to a blackout [21]. Generally, if the injected reactive power couldn’t able to increase the voltage magnitude, then the system is considered to be voltage unstable. Aforementioned, voltage instability may lead to voltage collapse, which is a sequence of unstable voltage conditions leading to low-voltage profile in a large portion of the power network [12]. Ultimately, voltage instability would lead to transient instability, since it would create electro-mechanical power imbalance at the synchronous generator. Therefore, adequate dynamic reactive power reserve must be maintained in order to improve both voltage and transient stability of the power network [24].

D. Grid Stability Improvement by Reactive Power

Several measures can be taken to improve static and dynamic reactive power reserves in the power grid. Usually it is achieved by deploying reactive power support devices, such as on-load tap changing (OLTC) transformers, excitation control, switchable and non-switchable shunt capacitors/ reactors, synchronous condensers, and flexible AC transmission system (FACTS) devices (e.g. static synchronous compensators (STATCOMs)) [14], [21]. Various techniques have been employed by researchers using these elements to stabilize the power grid, and provide adequate reactive power support to network [25]–​[29].

Some wind generators based on asynchronous machines (e.g. squirrel-cage induction machines (SCIMs) in fixed-speed wind generators (FSWGs)) can not contribute to the voltage regulation as they absorb reactive power during steady-state operation [30]. However, variable-speed wind generators (VSWGs) with PEC interface, such as the doubly-fed induction generator (DFIG) can provide reactive power [31]. Unfortunately, rotor converter rating of the DFIG is limited to only steady-state requirements to keep this technology within a reasonable cost margin. Therefore, the reactive power capability of the DFIG is not adequate as the primary safeguard during transient conditions. Similar limitations could be experienced with the full-converter wind generators (FCWGs). Hence, FACTS devices are used in wind farms to improve voltage stability using their dynamic reactive power capability.

Excitation controllers also play an important role in reactive power compensation in power systems [32]. Yet this type of controllers lack accuracy as they are designed considering static load models [33], [34]. Although VSWG technologies, such as DFIGs are more widely used due to their superior control capabilities, they have very limited dynamic reactive power reserve in comparison to the synchronous generators [24]. Nevertheless, PEC interfaced STATCOM devices could be used to improve the dynamic reactive power capability of wind farms to comply with grid-codes [28].

SECTION III. Grid Code Reactive Power Compliance Requirements for REGS

With the increasing renewable power penetration levels in power networks, the grid operators (e.g. transmission system operators (TSOs) and distribution system operators (DSOs)) have started to stipulate strict grid-codes for REGs on fault ride-through (FRT), reactive power management and voltage control. Aforementioned, reactive power strongly influence on network steady-state voltage, and voltage recovery during system contingencies, hence grid-codes specify both steady-state and dynamic reactive power capabilities for REGs. The grid-code specifications for FRT and voltage control are also closely related with the static and dynamic reactive power requirements for REGs. Therefore, reactive power grid-code requirements set for the wind generators and PEC interfaced generators (e.g. solar PV) are discussed in following subsections.

A. Reactive Power Requirements for Wind Generators

Almost all the grid codes reviewed in this paper specify steady-state reactive power requirements for wind generators. However, these requirements vary w.r.t. point of common coupling (PCC), voltage level at the connection point, specification of the actual capability of the system, and whether the reactive power requirement is expressed in terms of the power factor, or fraction of the rated power output etc.

In Danish grid code, for generators rated greater than 25 MW must have a reactive power capability of +/−0.3 p.u. for active power range between 0.2 to 0.8 p.u., and that has been progressively decreased from +/−0.3 p.u. to +/−0.2 p.u. when the active power level increases from 0.8 p.u. to 1 p.u. [35]. This indicates a reduced reactive power requirement at high active power levels, which is a reasonable reactive power specification in terms of the cost and technical limits of wind generators. A similar reactive power specification is stipulated in other grid codes for wind generators. The reactive power requirement specified in different grid codes for wind generators is summarized in Table 2. These reactive power requirements are usually expressed as P-Q diagrams (i.e. available active power versus available reactive power). Figure 4 illustrates reactive power requirements stipulated in grid codes for some European countries in a P-Q diagram [36].

TABLE 2 Reactive Power Requirements Specified in Different Grid Codes for Wind Generators

FIGURE 4.

Grid code reactive power requirements specified for wind generators in some European countries.

The Australian energy market operator (AEMO) is the responsible authority to operate Australia’s electricity market and power network. The National Electricity Rules (NER) require wind generators to have reactive power control capability of +/−0.93 power factor at full output at the point of connection (POC), throughout the full operating range of active power, and +/−10% of nominal voltage. However, the minimum access standard specifies no or zero reactive power capability for either reactive power supply or absorption. In case of South Australia, wind farms should have a +/−0.93 power factor capability at their full output, and 50% dynamic reactive power capability (as a fraction of rated power) should also be available at wind farms [37]. More information on grid code requirements for wind generation are given in [36] and [38]–​[43].

B. Grid Code Specifications for PEC Interfaced Energy Systems

The grid operators are yet to implement strict grid code specifications for different types of PEC interfaced energy systems (wind generation is excluded here), such as small-scale solar-PV systems, fuel cells, battery energy storage systems etc. According to AS/NZS 4777.2:2015 standard for four possible voltage ranges, namely V₁, V₂, V₃, and V₄ having Australian default voltage values of 207, 220, 244, and 255 V respectively, should have 30% leading power factor capability for V₁, 30% lagging power factor capability for V₄, and no regulation (i.e. 0%) is required for V₂ and V₃ [10].

In German grid-code, the generating plant should able to provide reactive power at the POC with 0.95 lagging power factor to 0.95 leading power factor. The reactive power generation can either be fixed or adjustable over different values of active power [46]. For low voltage (LV) generation unit, such as solar-PV, the operation range can be divided into three levels [47]:

• $S_{PV} < 3.68~kV\!\!A$ : the system should operate in between $\cos \phi = 0.95$ (under-excited/ lagging power factor) to $\cos \phi = 0.95$ (over-excited/leading power factor)

• $3.68~kV\!\!A < S_{PV} < 13.8~kV\!\!A$ : the system should accept any set point from DSO in between $\cos \phi = 0.95$ (under-excited/ lagging power factor) to $\cos \phi = 0.95$ (over-excited/leading power factor)

• $S_{PV} > 13.8~kV\!\!A$ : the system should accept any set point from DSO in between $\cos \phi = 0.90$ (under-excited/lagging power factor) to $\cos \phi = 0.90$ (overexcited/leading power factor)

Contrary to the German grid-code, the French grid-code distinctly mentions that the low voltage solar-PV systems should not absorb any reactive power at its entire operating range [48].

C. Dynamic Reactive Power Requirement for FRT

Aforementioned, small and medium-scale REGs (rated less than 50 MW) are connected to distribution networks (i.e. LV or MV), which is not typically designed to transfer power into the transmission grid [14], [49]. Therefore, voltage will increase during periods of high active power production from REGs. This eventually increases the need for dynamic reactive power support and fault ride-through (FRT) capability, due to weak dynamic voltage regulation capability of distribution networks. In some grid codes, the FRT requirements are specified as low voltage ride-through (LVRT) and high voltage ride-through (HVRT) for smooth operation of the power grid during symmetrical or asymmetrical fault conditions [35], [37], [45].

When a grid fault occurs, voltage decreases significantly around the fault node, and subsequently voltage depression propagates across a wide-area of the network until the fault is cleared [12]. During the fault, asynchronous wind generators demand more reactive power (e.g. squirrel-cage induction generator (SCIG) based FSWGs and crowbar activated DFIGs) while worsening the voltage levels across the network [50]. If the wind penetration level is high, and it is not supported by adequate dynamic reactive power reserve, then wind generators will start to disconnect from the grid due to decrease of their terminal voltage below the LVRT voltage specification, while leading to a catastrophic voltage stability issue in the power network [51]. A similar kind of issue could happen for solar-PV systems under fault conditions [46], [47], [52]–​[54]. Therefore, dynamic reactive power specifications are given in grid codes to improve LVRT capability of REGs. For example, German grid code requires REGs to provide 100% reactive current (w.r.t. nominal current), when there is a ≥50% voltage drop at their terminal [9].

On the other hand, voltage swell could occur when a large amount of load disconnect from the grid within a very short time-span or during significantly intermittent active power production from REGs (e.g. solar-PV systems or wind generators) [55]. Inefficient switching of capacitor banks or reactive power sources can also lead to voltage swell. To solve this issue, REGs are usually switched off during voltage swells. However, with increased penetration of renewable power generation in power networks, by switching off large amount of wind generators or solar-PV systems would lead to frequency stability issues. Therefore, nowadays in most grid codes, the HVRT requirements are specified for REGs [6]–​[10], [35], [37], [39], [45]. To met these HVRT requirements, REGs should essentially have reactive power absorb capability.

SECTION IV. Reactive Power Capability of Wind Generators

Wind generators are typically categorized into four (4) types: 1) Type-1: Fixed-speed wind generator (FSWG) (based on SCIG), 2) Type-2: Limited variable-speed wind generator (based on wound rotor induction generator (WRIG)), 3) Type-3: Doubly-fed induction generator (based on WRIG), and 4)Type-4: Full-converter wind generator (FCWG). The FCWGs can be further subdivided depending on the generator type (e.g. permanent magnet synchronous generator (PMSG) and electrically excited synchronous generator). Figure 5 shows typical wind generator configurations. It must be noted that both the SCIG and the WRIG machines are also known as the asynchronous generator (AG).

FIGURE 5.

Typical wind generator configurations: i) Fixed-speed wind generator (FSWG); ii) Limited variable-speed wind generator; iii) Doubly-fed induction generator (DFIG); iv) PEC interfaced fully-fed AG based FCWG; v) Electrically excited synchronous generator based FCWG; and vi) Permanent Magnet synchronous generator (PMSG) based FCWG.

The first and most simple configuration is the FSWG, which directly connects the SCIG to the grid, and a gear box is used in the drive-train to maintain the constant rotational speed. This type of wind generators produce real power when the shaft rotational speed is greater than the electrical frequency of the grid (i.e. when producing a negative slip), however these generators consume reactive power. For a given wind speed, the operating speed of the turbine varies linearly with the torque. The mechanical inertia of the drive-train limits the rate-of-change-of electrical power output under varying wind conditions. This configuration is depicted in Figure 5 (i). There is no active or reactive power control scheme, except the pitch angle control (PAC) scheme maintains the maximum power point (MPP) and curtails the wind power extraction at high wind speeds. To avoid high transient starting current, a soft-start device (e.g. back-to-back thyristors) is used in FSWGs.

Figure 5(ii) shows a limited variable-speed wind generator (Type-2), which is almost similar to the FSWGs. However, variable resistors are connected to the rotor circuit of this type of wind generators to provide limited variability in rotational speed. The variable resistors can control the rotor current depending on the wind gust conditions, and can also improve the dynamic response during grid disturbances.

The Type-3 wind generators are commonly known as the doubly-fed induction generators (DFIGs), and the configuration of the DFIG is illustrated in Figure 5(iii). In this type of wind generators, the stator circuit is connected to the grid directly, and the rotor is connected via a back-to-back PEC interface, by making it a doubly-fed machine. Because of the superior active and reactive power controllability of the DFIG, this wind generator type is heavily being used in the wind power industry, and hence substantial research has been conducted on DFIGs during last 15 years to improve their performance [56]–​[62].

The Type-4 wind generators (also known as the FCWG) use a fully-rated PEC interface to connect with the grid, and three different configurations are shown in Figure 5(iv)-(vi). The Figure 5(iv) shows a FCWG based on the AG, and the WRIG is mostly used as the AG. The FCWG configurations based on synchronous generators can either be excited electrically via slip rings as shown in Figure 5(v), or they can be self-excited permanent magnet synchronous generators (PMSGs) as shown in Figure 5(vi).

A. Reactive Power Capability of Type-1: FSWG

During 1990s, FSWG was the dominant wind generation technology, which is comprised of a SCIG directly coupled to the grid [63]. A FSWG wind farm usually has multiple induction generators, hence each has its own reactive power compensation device (e.g. capacitor bank), and its reactive power demand depends on the wind speed and the local voltage. This type of wind generators usually consume a large amount of reactive power during the initial magnetization of the machine. During low voltage conditions, FSWGs consume more reactive power to keep them magnetized while making the situation more worse [12]. The capacitor bank placed at each FSWG busbar supplies fixed no-load reactive power compensation for the FSWG ($Q_{cap}$ ) to comply with the grid code. Figure 6 depicts how the reactive power requirement of a FSWG depends on active power generation [64]. There is no possible control mechanism for FSWGs to control reactive power absorption or supply, without having an additional reactive power compensation device. However, capacitor banks, static VAr compensators (SVCs), or STATCOMs can be installed at FSWG for reactive power support. Therefore, the reactive power capability of the FSWG depends on the capability of the externally added device.

FIGURE 6.

FSWG reactive power characteristics.

B. Reactive Power Capability of Type-3: DFIG

Aforementioned, in DFIGs the stator is directly connected to the grid, while the rotor is connected to the grid via a back-to-back PEC interface. Both the grid-side converter (GSC) and the rotor-side converter (RSC) have independent controllers, which can control active and reactive power independently. Therefore, this generator operates either in sub-synchronous or super-synchronous modes, and hence could operate in a wider speed range (e.g. 0.7 – 1.2 p.u.). In sub-synchronous mode, the stator winding supplies power to the grid while the rotor circuit absorbs power from the grid depending on the operating slip of the generator. In case of super-synchronous mode, both the stator and the rotor supply power to the grid. The typical configuration of a DFIG is illustrated in Figure 7.

FIGURE 7.

A typical configuration of a DFIG.

The reactive power output of the DFIG ($Q_{W}$ ) is the sum of reactive power from the stator ($Q_{s}$ ), and reactive power from the GSC ($Q_{g}$ ). Both $Q_{g}$ and $Q_{s}$ are controlled at their reference values ($Q_{W,g}$ and $Q_{W,s}$ ) by the GSC and the RSC controllers respectively. Usually, the GSC controller controls the active power transfer either from the grid or to the grid to maintain the DC-link voltage constant. The RSC controls the active and reactive power output from the stator side of the DFIG. The equivalent circuit of the DFIG is illustrated in Figure 8.

FIGURE 8.

Equivalent circuit diagram of the DFIG [12].

From the equivalent circuit of the DFIG, the stator active and reactive power can be derived as:

$$\begin{align*} P_{S}=&3\frac {1}{X_{S}}EV_{S}\sin \delta \tag{9a}\\ Q_{S}=&3\frac {1}{X_{S}}EV_{S} \cos \delta - 3 \frac {V_{S}^{2}}{X_{S}} \tag{9b}\end{align*}$$

Similarly, the rotor active and reactive power can also be obtained as:

$$\begin{align*} P_{R}=&-3s\frac {1}{X_{S}} EV_{S} \sin \delta \tag{10a}\\ Q_{R}=&- s\left ({3 X_{R}I_{R}^{2} + 3\frac {1}{X_{S}}EV_{S} \cos \delta - 3 \frac {E^{2}}{X_{S}}}\right)\tag{10b}\end{align*}$$

The relationship between the stator and the rotor active power can be derived as:

$$\begin{equation*} P_{R} = -sP_{S}\tag{11}\end{equation*}$$

The above equations indicate that, when the machine is operating in super-synchronous mode $(P_{S} > 0)$ , the rotor power becomes positive when the slip (s) is negative. This means that the rotor power is fed into the grid when the generator is working at super-synchronous mode. However, the rotor power becomes negative when the slip is positive, and power is fed into the rotor from the grid at the sub-synchronous mode.

Now, total active power can be calculated by summing the total stator power and rotor power as follows:

$$\begin{align*} P_{T}=&P_{S} + P_{R}\tag{12a}\\[-2pt] P_{T}=&(1-s) P_{S}\tag{12b}\end{align*}$$

The total reactive power can not be calculated by summing up the stator and rotor reactive power. This is because there is no reactive power flow from the rotor (via RSC) to the DC-link. However, the GSC can control the reactive power, and in typical commercial DFIGs, the reactive power reference for the GSC, $Q_{W,g}$ is kept at zero. Therefore, the total reactive power output from the DFIG is given by equation (9b). However, the GSC can be used to provide reactive power to improve the reactive power capability of the DFIG by some other methods proposed by researchers [50]. Some of these emerging control techniques employed in DFIGs for reactive power support are summarized in Table 3. The proposed methodologies can make the GSC act like a dynamically controlled reactive power source. Therefore, the total reactive power supplied to the grid by a DFIG can be obtained as:

$$\begin{equation*} Q_{T} = Q_{S} + Q_{GSC}\tag{13}\end{equation*}$$

TABLE 3 Reactive Power Control Objectives for the DFIG

The reactive power capability curve of a typical 1.5 MW DFIG-RSC and GSC are obtained from [65], and are illustrated in Figure 9. It can be seen from the figure that, the DFIG-RSC can operate between +/−0.95 power factor without additional reactive power support from the GSC. However, +0.90 power factor operation is constrained to 0.90 pu active power output. Therefore, additional reactive power must be provided by the GSC. Besides, the reactive power capability reduces with an increase in DFIG active power output. The reactive power capability chart of the GSC indicates that, ±0.48 pu average reactive power capability for a 50% converter rating across its operating range. Hence, a 50% converter rating means a combined reactive power capability of 1.28 pu during zero active power production. During full active power production this value reduces upto 0.83 pu, thus, substantial reactive power capability is available at the DFIG to support the grid during contingencies.

FIGURE 9.

Reactive power capability curves of the DFIG (a) DFIG-RSC capability, and (b) GSC capability.

C. Reactive Power Capability of the Type-4: PMSG

The PMSG wind generators have a fully rated back-to-back converter system. This converter system consists of a rectifier (AC/DC converter or machine-side converter (MSC)), followed by a DC link capacitor, and an inverter (DC/AC converter or GSC). The configuration of a typical PMSG is shown in Figure 10. This wind generator completely decouples the generator from the grid, which helps conceal the generator and the turbine during fault conditions. When the grid voltage decreases in a fault condition, the power balance at the GSC and the MSC is disturbed, and the excess power is accumulated at the DC-link, and subsequently increases the DC link capacitor voltage. Usually, energy storage systems (ESSs) are incorporated in the DC link and activated during fault conditions to absorb or supply the power imbalance. Under normal operating conditions, the PMSG can either absorb or supply reactive power depending on control techniques applied to the power converters. The reactive power capability of the PMSG is not constrained by either the machine properties or characteristics.

FIGURE 10.

A typical PMSG system.

The real and reactive power of the PMSG can be derived as:

$$\begin{align*} P=&\frac {V_{g}V_{c}}{X} \cos (2 \pi - \phi) \tag{14a}\\ Q=&\frac {V_{g}V_{c} \sin (2\pi - \phi) - V_{g}^{2}}{X} \tag{14b}\end{align*}$$

From the equation (14b) it is evident that, the reactive power capability of the PMSG depends on the grid voltage, $V_{g}$ , grid reactance, $X$ , and converter voltage, $V_{c}$ , and power factor, $\cos \phi$ . With the decrease of grid voltage, the reactive power capability will also decrease. In case of grid reactance, the lower the reactance, the higher the reactive power support from the PMSG. Also, reactive power capability depends on the available GSC capacity after transferring active power to the grid. However, PMSG based FCWGs can still provide reactive power support at rated active power, since GSC is usually rated at 110% of the active power rating of the FCWG.

Various reactive power control/management techniques have been developed by the researchers for the PMSG and they are summarized in Table 4.

TABLE 4 Reactive Power Control Techniques of PMSG Systems

SECTION V. Reactive Power Capability of Solar-PV Generators

Having no rotating magnetic field or coil arrangements, the solar-PV systems supply power through an inverter. Solar-PV panel itself does not have any reactive power support as it produces electricity using photovoltaic effect. However, the inverter used for DC/AC conversion can provide significant amount of reactive power support during normal operating conditions or even in fault conditions. The solar-PV inverter also provides other ancillary services, such as, MPPT control, LVRT etc. Although, reactive power support is not yet mandatory for solar-PV systems in most grid codes, as the penetration level increases more controllability over active and reactive power will become a necessity. A typical single-phase grid connected solar-PV system is illustrated in Figure 11.

FIGURE 11.

A typical single-phase grid connected solar-PV system.

There are several reactive power compensation techniques implemented by researchers for solar-PV systems. Traditionally, this is done by employing a control scheme in the inverter control circuit. These techniques along with some others are discussed in the following subsections.

A. Various Controllers Used in Solar-PV Inverters

Implementing a control scheme by means of a controller at the solar-PV inverter for active and reactive power control is one of the simplest way for reactive power compensation. The control schemes are usually implemented either using digital signal processors (DSPs), field programmable gate arrays (FPGAs), or microcontrollers. FPGAs are renowned for their low power consumption and ability to achieve high level of parallelism [89]. Hassaine et al. [90] proposed a reactive power compensation methodology for grid tied solar-PV system using FPGAs. They have implemented the control strategy based on digital sinusoidal pulse-width modulation (DSPWM) and the phase-shift between inverter and grid voltages. With this control technique the injected reactive power can be dynamically modified and controlled. A similar kind of implementation using FPGA can be found in [91]. DSPs can also be employed to design reactive power controller and maximum power point tracking (MPPT) unit for solar-PV systems. Libo et al. [92] proposed a modified incremental conductance MPPT controller and a reactive current controller using a DSP. In [93] and [94], simplified reactive power control schemes are proposed using a microcontroller, where current-mode asynchronous sigma-delta modulation (CASDM) is employed to improve the dynamic response. Besides these, a large number of researchers have employed different techniques and controllers in the inverter control circuit to implement reactive power support schemes [95]–​[97], and few of them are summarized in Table 5.

TABLE 5 Control Schemes in the Inverter Control Circuit of the Solar-PV System for Reactive Power Support

B. Using Cascaded Multilevel Converters

Because of the modular structure, scalability, and enhanced energy harvesting capability, cascaded multilevel converters are gaining popularity in solar-PV system applications. Liu et al. [52], [107], [108] proposed cascaded multilevel converters to enhance reactive power support for solar-PV systems. They have developed a reactive power compensation algorithm suitable for different types of cascaded solar-PV systems. In this proposed strategy, they have first converted the output voltages from each solar-PV module in d-q reference frame. Then, they obtained the active power of each module from MPPT control. Consequently, the output voltage from each converter module is also analyzed, and active and reactive power is distributed in each converter module accordingly. A block diagram of a typical cascaded multilevel converter based solar-PV system is shown in Figure 12.

FIGURE 12.

A typical cascaded multilevel converter based solar-PV system.

C. Using ESS

Coordinated use of ESS and solar-PV inverters are also being proposed as a solution for the voltage and reactive power control of the distribution network. Droop-based ESS is used and analyzed along with solar-PV inverters to mitigate the voltage rise issue and reactive power support by Kabir et al. [109]. They have found that if the R/X ratio of the line is 4.5 to 5, then reactive power compensation alone is not sufficient for voltage regulation. Therefore, urban areas where the R/X ratio is close to unity, the solar-PV inverters are sufficient to provide reactive power support. However, in rural areas where the R/X ratio is much higher, the ESS should be added along with the solar-PV inverters for better reactive power support and voltage control. The authors have also investigated both constant and variable droop based reactive power control schemes for the ESS, and found that constant droop based scheme requires a large battery, whereas, variable-droop based scheme requires a smaller battery.

D. Using High Frequency Link Converter

Because of the low cost, high power density, and capability to provide isolation between the solar-PV panels and the grid, high frequency link converters are being used in grid-tied solar-PV systems. This type of converters also have the bidirectional power flow capability from the grid to the DC source, which enable it to provide reactive power support and voltage regulation. Robles et al. [110] implemented reactive power compensation scheme in a grid tied solar-PV system using a high frequency link converter. Using the push-pull topology they have investigated the performance of the proposed system and validated the results through simulation studies.

E. Using Transformerless Solar-PV Inverters

Among transformerless solar-PV inverters H5, H6 and HERIC inverters are most commonly used. Usually they do not have reactive power compensation capability because of the absence of freewheeling path in the negative power region. However, modulation techniques are proposed by the researchers to provide bidirectional freewheeling current path [111]–​[113]. In [111] space-vector based PWM modulation strategy is proposed, which is operated in two stages; 1) Inverter modulation, and 2) Reactive power modulation. They also designed a proportion-integration-resonance (PIR) current controller to subdue zero-crossing current distortion. Freddy et al. [112] proposed a PWM technique to implement reactive power support capability in H5 and HERIC inverters. This is similar to the sinusoidal PWM with the exception that it requires additional duty-cycle generators for each switch. The proposed PWM scheme is illustrated in Figure 13.

FIGURE 13.

The PWM technique used to implement reactive power support capability in H5 and HERIC inverters.

F. Reactive Power Capability of Other REGS

Other than solar PV, and wind generators, the renewable generators, such as hydro, wave energy, tidal power, bio fuel, or fuel-cells can also be connected to the grid by means of PECs or synchronous generators. Separate reactive power compensation strategy is not required for renewable generators, which uses synchronous machines to produce power. Besides, the renewable generators which are connected to the grid through a PEC, can use reactive power compensation techniques used in solar-PV converters.

SECTION VI. Reactive Power Support Devices

Besides the internal reactive power control schemes implemented in REGs (i.e. machine or converter level), there are several reactive power control devices which can be connected at the PCC or some other place for reactive power support and voltage stability of the power grid. Usually, FACTS devices, such as STATCOM, SVC, DVR etc. as well as conventional devices, such as OLTC transformers, and capacitor banks are used for reactive power compensation. However, FACTS devices provide better controllability and flexibility compared to conventional reactive power compensation devices. Reactive power support devices used in the power grid are illustrated in Figure 14. Both the conventional and contemporary reactive power control devices are discussed with elaboration on their current research progress in following subsections.

FIGURE 14.

Reactive power support devices used in the power grid.

FIGURE 15.

Schematic diagram of a STATCOM.

A. OLTC Transformers

OLTC transformers are used to regulate system voltage by changing the turns ratio of the transformer under loaded condition [114], [115]. However, the mechanically switched OLTCs are not fast enough to provide reactive power support for dynamic loads connected to power system. Combination of OLTCs with other reactive power control devices are usually used to provide efficient voltage control. For instance, in [116], a similar combined approach using the OLTC along with the SVC is proposed for reactive power compensation. A coordinated control of the OLTC transformer and local wind turbine controllers is implemented in [117]. In [118], the OLTC is modeled as a finite machine and coordinated controller is designed for both the DFIG controller and the OLTC transformer. Parallel operation of the OLTC transformer and solar-PV inverters is described in [119].

B. Capacitor Banks

Parallel switched capacitor banks are usually installed at the PCC of the REGs, to enhance the reactive power support (mainly in Type-1 and Type-2 wind turbines). They behave as reactive power sources during transient conditions. However, there are some major advantages and disadvantages of using capacitor banks as a source of reactive power. Among the advantages, the power quality enhancement by power factor improvement, and thus reduction in thermal losses and increase in system capacity are the main. However, there are some disadvantages, such as capacitor switching creates strong transients propagating throughout the network. It also creates high-frequency harmonics. The inductive lines and capacitor banks form RLC circuits which may create resonance issues [120]. Because of these issues additional harmonic filters are required, which leads to additional cost and system complexity. A lot of research studies have been conducted on the optimal positioning and switching mechanisms of the capacitor banks for reactive power compensation [121]–​[124]. In addition, the capacitive reactive power is proportional to the square of the terminal voltage, hence capacitor banks are not a very good dynamic reactive power source.

C. ESS

ESS improves the reliability, and dynamic stability of the power system by enhancing the power quality and transmission capacity of the grid. There are various types of energy storage systems, such as battery energy storage systems, super-capacitors or ultra-capacitors, flywheel energy storage systems, pumped hydro energy storage systems, compressed-air energy storage, and electrochemical energy storage, such as fuel cells etc. Battery energy storage is the most widely used ESS, and usually used for active and reactive power support for REGs in distribution networks [125]–​[127]. Currently, ultra-capacitor/ super-capacitor is also becoming very popular for active and reactive power support [128]. For example, in [129], ultra-capacitor is added into the DC-link of the converter of wind or solar-PV systems for better reactive power support capability.

D. STATCOM

Hingorani et al. [130] first proposed the concept of STATCOM in 1976. A STATCOM is a FACTS device usually consisted of a VSC, a controller, and a step-up transformer or coupling reactor as shown in [51, Fig. 15]. It is typically used at the PCC of a wind farm or solar-PV generator for reactive power compensation and voltage control. By turning on/off the VSC switches (e.g. IGBTs) of the STATCOM, the output voltage of the VSC is regulated, and hence the output current can be controlled. The current and power equations of the STATCOM are given in equation (15) and (16).

$$\begin{align*} \qquad \qquad \qquad I=&\frac {V_{o} - V_{pcc}}{X_{s}} \tag{15} \\ P=&\frac {V_{o}V_{pcc}}{X} \sin (\alpha - \theta)\tag{16a}\\ Q=&\frac {V_{o}(V_{o} - V_{pcc} \cos (\alpha - \theta))}{X}\tag{16b}\end{align*}$$

It is evident from the above equations that, either capacitive or inductive current can be achieved by regulating the VSC output voltage, $V_{o}$ . For the values of $V_{o}$ larger than $V_{pcc}$ , the STATCOM will operate in the capacitive mode, whereas for the values of $V_{o}$ smaller than $V_{pcc}$ it will operate in inductive mode. The active and reactive current characteristics of a STATCOM are illustrated in Figure 16. STATCOM has been extensively used along with REGs for reactive power support [131]–​[135]. The STATCOM is capable of providing strong dynamic reactive power support in comparison to capacitor banks and other conventional devices.

FIGURE 16.

Active and reactive current characteristic curves of a STATCOM.

E. SVC

SVC is a parallel connected static var absorber or generator which can be controlled to stabilize the grid voltage. SVC can be used to provide dynamic reactive power to the grid. SVC contains a voltage measurement circuit, and a voltage regulator, and their output is fed into a thyristor control circuit. A schematic diagram of a typical SVC, employed with a thyristor controlled reactor (TCR), a thyristor switched capacitor (TSC), a harmonic filter, a mechanically switched capacitor and a mechanically switched reactor, is shown in Figure 17. The active and reactive current characteristic curves are shown in Figure 18. SVCs are used with REGs in distribution networks for reactive power compensation and voltage stability improvement [136]–​[140].

FIGURE 17.

Schematic diagram of a typical SVC.

FIGURE 18.

Active and reactive current characteristic curves of a SVC.

F. DVR

The DVR is a FACTS device which contains a VSC having an energy storage system (ESS) connected to the DC-link. It is connected to the power network in series with a transformer and coupling filters as shown in Figure 19. DVR is capable of either generating or absorbing real and reactive power independently. It is used along with REGs for voltage control and LVRT improvement [141]–​[144].

FIGURE 19.

Schematic diagram of a DVR.

SECTION VII. Control Strategies Developed for Reactive Power Management in REGs

For reactive power management in REGs, various control strategies, such as sliding mode control (SMC), model predictive control (MPC), droop control, current mode control (CMC), synchrophasor based control, and soft computing based control strategies are used. Figure 20 illustrates these control techniques. In the following subsections application of these control strategies for reactive power control are discussed.

FIGURE 20.

Various control techniques used for reactive power management in REG integrated power grid.

A. Sliding Mode Control

Sliding mode control (SMC) was first introduced in 1962 based on B. Hamel’s idea of nonlinear compensators [145]. Now, it is the most widely used nonlinear control strategy for reactive power compensation in REGs. In SMC, usually three steps are defined to design the control scheme: 1) A sliding surface is identified; 2) The existence of such a surface is tested, and 3) Stability analysis is done inside that defined surface [146]. There are some variants of SMC applied in literature for reactive power support in REGs. For example, Yang et al. [147] proposed perturbation and observe (P&O) based SMC for maximum active power extraction and reactive power control in DFIG based wind generators. A fuzzy SMC is used by Wang et al. [148] for reactive power compensators, such as SVCs. In [149], a fuzzy SMC is also implemented for transient stability improvement and reactive power compensation of the system. Discrete SMC was adopted by Pande et al. for real and reactive power control, and used discrete representation for system dynamics [150]. Second or higher order SMC is deployed for reactive power compensation in a DFIG [151]–​[154]. Besides these, SMC is also adopted by researchers extensively for active and reactive power support for converter based REGs [74], [155]–​[159].

B. Model Predictive Control

As the name suggests, the model predictive control(MPC) uses a model explicitly to predict the output of the process in future time instants, and then the objective function is minimized by calculating a control sequence [160]. However, finding an appropriate model of the process is the most daunting task in this type of control scheme. Yaramasu et al. [161] proposed a MPC algorithm using the discrete time model of an inverter for a wind energy conversion system. A MPC controller is proposed for modular multilevel converters, and other converter types in [162]–​[165]. A MPC controller is implemented for a DFIG in [166], for a PMSG in [167] and [168], and for a solar-PV system in [169].

C. Droop Control

Among linear control strategies, droop control is the most commonly used control technique for reactive power compensation in REGs. As the output of REGs are variable and intermittent in nature, the controller has to respond accordingly to compensate this variation in active power. A variable droop gain control to enhance reactive power support for wind farms is described in [170]. Virtual flux based droop control is deployed for reactive power control in [171]. Besides these, droop control method is largely used in literature for active and reactive power control of REGs [172], [173]. A review of droop control techniques can be found in [174].

D. Current Mode Control

Current mode control uses sensed inductor-current ramp in the PWM modulator and has a two loop structure compared to its counterpart, voltage mode control, which has a single loop structure [175]. This control scheme is incorporated mostly in converters of REGs for active and reactive power control. For instance, peak current mode control is used for a solar-PV converter in [176] and [177]. Both voltage mode control and current mode control were deployed and compared for a dual active bridge converter in [178].

E. Synchrophasors Based Control

In synchrophasor based control, phasor measurement units (PMUs) perform digital signal processing to estimate phasor components from measured analog waveforms, which is then used in control algorithms for various control purposes [179]. Jiang et al. [180], [181] proposed an auxiliary coordinated control, and multiple-input and multiple-output (MIMO) model-predictive control (MPC) using synchrophasor measurement data for a distribution system with high penetration of renewable generation. Synchrophasor based secondary voltage control scheme is described in [182]. A PMU based wind farm monitoring application can be found in [183]. Current trends on synchrophasor based control applications are summarized in [184] and [185].

F. Soft Computing Methods

Soft computing methods are the emerging group of problem solving methods, which strive to imitate the intelligence found in nature [186]. Actually, these methods exploit tolerance for imprecision, uncertainty, and partial truth to achieve tractability, low cost, and robustness. Among the notable soft computing methods, fuzzy logic, neural networks, genetic algorithm, particle swarm optimization, and wavelet theory are more widely being used in control applications. The use of various soft computing methods for reactive power control are discussed in the following subsections.

1) Fuzzy Logic

Fuzzy logic controllers are being used extensively in recent control applications because of their robustness, ability to handle imprecise inputs, non-linearity and their ability to work without an accurate mathematical model [187], [188]. A fuzzy logic controller was developed for a fixed-speed wind energy conversion system by Krichen et al. [189] for active and reactive power control. Medjber et al. [190] proposed a fuzzy logic controller to control active and reactive power of a DFIG. A fuzzy logic supervisor is deployed to control a flywheel energy storage system of a DFIG based wind energy conversion system in [191]. Fuzzy logic is also used to tune the parameters of a unified power-flow controller (UPFC) for reactive power compensation of a stand-alone wind-diesel-tidal hybrid system [192]. Rezaei and Esmaeili [193] employed a decentralized voltage control method based on fuzzy logic, and optimized it by gradient descent algorithm (GDA) to control reactive power of distributed solar-PV and wind based power system. Besides these, fuzzy logic controllers are also used for reactive power control of REGs [194]–​[197].

2) Artificial Neural Network

The biologically inspired computational model, artificial neural network (ANN), consists of elements (called neurons) processing and identifying connections between the elements along with their coefficients. These element connections make neuronal structure, and training and recall algorithms attached to them [198]. Bansal et al. [199] tuned the parameters of an SVC controller using ANN for an autonomous wind-diesel hybrid power system. An ANN based thyristor controlled series compensator (TCSC) controller is developed for reactive power compensation in wind-diesel-PV hybrid system in [200]. ANN is also used to tune the PI gains of a STATCOM controller of an autonomous wind-diesel hybrid system in [201]. Saxena and Kumar [202] used ANN to control reactive power of a STATCOM in a decentralized hybrid power system. A similar kind of work with ANN based STATCOM controller is proposed by Mauboy et al. [203] for power system stability enhancement.

3) Genetic Algorithm

Genetic Algorithms (GA) are evolved from biological concepts, and are being used in various control applications [204]. Vrionis et al. [205] tuned the GSC and the RSC controller of a DFIG for reactive power compensation and LVRT operation using GA. In [206] GA is employed to optimize reactive power in wind generators. For solar-PV systems, a multi objective GA is used for volt-var control in [207]. More on GA’s application for reactive power compensation in wind, solar PV, and wind-solar hybrid systems can be found in [208]–​[210].

4) Particle-Swarm Optimization

Kennedy [211] first proposed the particle swarm optimization (PSO) algorithm in 1995. It is a population based stochastic search, and this optimization technique can avoid local optimum like other evolutionary algorithms (EAs), such as GA [212]. Sayadi et al. [213] performed the optimal scheduling of an OLTC transformer, and shunt capacitors of a solar-PV system for reactive power control using PSO method. Similar kind of research studies using adaptive PSO have been conducted for reactive power management in offshore wind farms [214], [215]. In [216] wind and solar DGs are placed optimally based on the reactive power loadability using the PSO algorithm. Further research studies on reactive power compensation for REGs in distribution network using PSO can be found in [217]–​[219].

G. Conventional Methods

A comparison of all control techniques (based on the complexity and the response time) discussed above are enlisted in Table 6.

TABLE 6 Comparison of Control Techniques Used in Reactive Power Control of REGs

SECTION VIII. Reactive Power Coordination & Optimization Strategies

Reactive power compensation for REGs can be implemented at three different levels: a) At the machine level, i.e. inside the REGs, such as in GSC of the DFIG, b) At the PCC level, i.e. connecting FACTS devices or energy storage systems (ESSs) at the PCC and controlling them using various control methods, and c) At the overall distribution network level, i.e. connecting the reactive power compensation devices away form the PCC or at the load connecting point and controlling & optimizing them efficiently. Reactive power can also be controlled centrally in the distribution network, or can be managed at local generation and load. The objective of reactive power optimization in an AC power system is to determine the best values for control variables(e.g. generator voltages, transformer tap positions, and reactive power compensator’s output) within given constraints(e.g. active and reactive power flow limits, and voltage deviation range). This problem can be divided into two distinct parts; 1) the optimal placement of reactive power compensators, 2) the optimal operation of the existing reactive power compensators, as shown in Figure 21. The reactive power optimization of distribution networks with REGs is usually performed with the well-known optimal power flow (OPF) method. It combines an objective function with the power flow equations to form an optimization problem [221]. Usually, the system losses decrease with the increase in reactive power capability up to a certain point, and after that minimum point further increase in reactive power will increase the system losses as shown in Figure 22 [220]. Therefore, an optimization problem is solved to find that optimal point at which the system losses become minimum. More information on the OPF problem formulation for voltage control and reactive power optimization with REGs can be found in [222]–​[225].

FIGURE 21.

The reactive power optimization types.

FIGURE 22.

System loss variation for various DFIG reactive power capability [220].

Reactive power coordination and optimization approaches (both conventional and advanced methods) reported in the literature are discussed in following subsections.

1) Using Linear Programming

Linear programming methods are reliable techniques to obtain solution for optimization problems characterized by linear constraints and linear objects. They are usually robust techniques applicable to electric power systems, but sometimes they provide with incorrect evaluation of the system losses and get trapped in a local optimal solution [12]. Guggilam et al. [226] constructed a quadratic constrained quadratic program by leveraging on linear approximation of the power flow problem to develop an OPF problem with solar-PV systems in distribution networks. However, a large amount of literature can be found on reactive power management using linear programming method [227]–​[229].

2) Using Nonlinear Programming

As constraints of the reactive power planning are nonlinear, the nonlinear programming would be the most practical method for solving the optimization problem. Sequential quadratic programming, extended Lagrangian method, generalized gradient method, and interior-point method are mostly used non-linear programming methods in electric power systems [12]. Meegahapola et al. [220], [230] solved OPF and voltage constrained OPF problems for a DFIG based wind power system using the Newton Lagrangian method. According to their study the wind farms should dispatch optimal reactive power to improve active power losses. Chen et al. [231] used nonlinear programming to find an optimal size of the centralized capacitor banks, and to control them for reactive power management in distribution networks.

3) Using Mixed-Integer Nonlinear Programming

Mixed-integer nonlinear programming methods are used to solve optimization problems containing nonlinear functions in the objective function. They combine the difficulty of optimizing discrete variable sets with nonlinear functions, which means that they include both nonlinear programming and mixed-integer linear programming as subproblems [232]. Kulmala et al. [233] used mixed-integer nonlinear programming to optimize distribution network voltage control. They assumed all the optimization variables to be continuous, and solved the problem using MATLAB optimization toolbox. Genetic algorithm was used in [234] to solve the mixed-integer nonlinear programming problem to optimize the reactive power requirements. Branch flow model based relaxed OPF is used to formulate a mixed-integer second order conic programming problem for active and reactive power optimization in [235] and [236]. Tiwari et al. [237] first formulated the reactive power optimization problem as a mixed integer dynamic optimization, which is then converted into mixed integer nonlinear problem by means of simultaneous discretization. In [238] reactive power elements, such as capacitor banks, voltage regulators, and under-load tap changing (ULTC) transformers are considered as control variables, and reactive power optimization is achieved through mixed-integer nonlinear programming. Nick et al. [239] presented a control technique for optimal sizing and placement of the ESS for reactive power control in distribution networks using Benders decomposition method. They have also considered the stochastic nature of the renewable energy sources and the load demand.

4) Using Nonlinear Dynamic Optimization

In nonlinear dynamic optimization, linear optimization is first achieved, and then linear optimization values are used as initial guesses for nonlinear optimization. In [241], a dynamic optimization approach called control vector parameterization (CVP) is used to find the optimal location and amount of reactive power support required for a distribution network. The CVP approach was also upgraded by trajectory sensitivity analysis, singular value decomposition, and linear optimization programming. Jwo et al. [244] proposed a constrained dynamic optimization model using quadratic objective function for reactive power and voltage control problem in a distribution network. Their work has explicitly taken into account the time-varying projection operation of constrained dynamic optimization.

A. Advanced Methods

1) Using Simulated Annealing Methods

Simulated Annealing method is inspired by the physical process of heating a material and then cooling it slowly to decrease defects, thus minimizing the system energy. It is a local search algorithm having the ability to escape from local minimum by making probabilistic moves [245]. In this method, a new point is randomly generated after each iteration having a proportional probability distribution scale of temperature. This algorithm accepts all new points that either lower the objective or raise the objective. By doing this, the algorithm can escape local minima in early iterations, and can explore globally for better solutions. Liu et al. [246] used simulated annealing to solve reactive power planning problem using the IEEE-30 test bus system. They have also improved the simulated annealing method using a modified definition of neighborhood selection. More applications of simulated annealing algorithm in reactive power management of power grids can be found in [247]–​[254].

2) Using TABU Search Method

The Tabu search method was first developed by Rueda-Medina et al. [255] and Golshan and Arefifar [256] to solve combinatorial optimization problems. This method works similarly to human memory, which is flexible and able to eliminate the local minima, and can search beyond a local minimum [257]. Tabu search algorithm was used in [258] for reactive power support in distribution networks with capacitor banks and distributed generators. They minimized the power losses subjected to various network constraints. In [259], distributed generation, reactive sources and network-configuration are optimized using Tabu search method for power and energy-loss reduction. Tabu search method was also implemented broadly in reactive power optimization & planning in [260]–​[265].

3) Using Evolutionary Algorithms

Evolutionary algorithms are inspired by the paradigm of biological evolution, for example, mutation, reproduction, recombination, and selection [266]. Evolutionary algorithms are developed as a combination of several techniques, such as, genetic algorithms, evolutionary programming, evolutionary strategies, and genetic programming [267]. A comprehensive study on optimal reactive power planning using evolutionary algorithms is presented in [268]. They implemented evolutionary programming, evolutionary strategy, and genetic algorithm to solve the optimal reactive power planning problem. Malachi and Singer used genetic algorithm for active and reactive power optimization using linear approximation of load flow equations, and heuristic selection of participating controls [269]. Among the evolutionary algorithms, genetic algorithms are the most widely used algorithm for reactive power optimization in power grids [242], [243], [250], [270]–​[280]. However, other evolutionary algorithms are also used for reactive power optimization in power grids, such as, evolutionary programming in [281]–​[291], evolutionary strategy in [268], [292], [293], and genetic programming in [294].

4) Using Swarm Algorithms

‘Collective intelligence’ is the basis of swarm algorithms which emerges through the cooperation of large number of homogeneous agents in the environment. For example, flocks of bird, colonies of ants, and schools of fishes. Among the swarm algorithms, particle swarm optimization (PSO) is the baseline algorithm for many variant swarm algorithms, such as ant colony algorithm, bees algorithm, and bacterial foraging optimization algorithm etc [245]. Zhao et al. [295] deployed a PSO algorithm for reactive power and voltage control to handle a mixed integer nonlinear optimization problem consisting of continuous and discrete control variables, such as OLTC tap positions, number of reactive power compensation equipment, automatic voltage regulator operating values of generators etc. A multi-agent based PSO is presented in [240] as a solution to the reactive power dispatch problem. Besides these, particle swarm optimization algorithms was used extensively in the literature for reactive power optimization [296]–​[300].

A summary of recent researches on reactive power optimization and the use of REGs are tabulated in Table 7. Reactive power optimization algorithms discussed above are enlisted along with their references in Table 8.

TABLE 7 Summary of Recent Researches on RPO and Use of REGs

TABLE 8 Reactive Power Optimization Algorithms

SECTION IX. A Case Study - Reactive Power Management in a Distribution Feeder with Solar-PV Systems

Performance of various reactive power control strategies used for distribution feeder voltage control is demonstrated in this case study. The investigated strategies include tap-changing transformers, capacitor banks, voltage controlled single-phase solar-PV systems and energy storage systems. The simulation was carried out in DIgSILENT Power Factory for a three-phase LV distribution feeder, and each phase consisted of eight domestic household loads and each household is separated by 43.05 m as shown in Figure 23. It is assumed that every household has a 3 kW (3.3 kVA) solar-PV system. The distribution line has an impedance of $0.315+j0.259 \Omega /km$ .

FIGURE 23.

Distribution feeder model for the case study.

The investigated operation scenarios are listed in Table 9. For each operation scenario, a load-flow calculation is performed using DIgSILENT Power Factory software tool.

TABLE 9 Solar PV Operation Scenarios for Each Household

According to Australian standard AS60038 distribution grid voltage should be maintained between +10% to −6% of the nominal voltage, which is 230 V. The node voltages for each operation scenario are illustrated in Figure 24.

FIGURE 24.

Distribution feeder model for the case study.

As illustrated in Figure 24, for the first scenario (scenario 1) voltage limits are not violated. This represents the conventional scenario without any solar-PV connection to the LV network. However, in scenario 2, where the load demand is increased from 2.4 kW to 12 kW, representing peak evening load, the lower voltage limit has been violated, and the voltage has decreased below the stipulated minimum limit in the standard. In this situation, distribution transformer’s tap position can be increased to improve the voltage. This scenario with optimized tap position of the OLTC transformer is shown in scenario 3, and from Figure 24, it is evident that voltage profile is well within the standard voltage limit once the tap position is optimized for the distribution transformer. It has to be noted that, the optimized tap positions for all scenarios are given in Table 9, and the additional voltage available per tap is 2.5.

A much better voltage profile can be achieved, once a switched capacitor bank is installed in the feeder. Capacitor banks along with OLTC transformers can provide an even better voltage regulation, and also can reduce the burden on tap changers while increasing their lifetime. The node voltages after adding a capacitor bank at each phase are depicted in scenario 4. A 60 kVAr capacitor bank is used in this case study, which has six 10 kVAr steps. With the reduced tap position (i.e. Tap position - 2), the optimum value for capacitor bank is 50 kVAr. It can be seen from Figure 24 that the node voltages in scenario 4 are within standard voltage limits.

Now, in scenario 5, assume that all single-phase solar PV systems are connected, and generating 3 kW. It can be seen from Figure 24 that node voltages have increased beyond the stipulated maximum limit due to the bidirectional power-flow in the feeder. However, in practical situations during the mid-day, the load demand will be lower, hence the majority of the power generated from the solar-PV units will be dispatched to the network. Therefore, in scenario 6, the load demand is set at 2.4 kW, and in order to control the voltage rise issue, the capacitor banks are also disconnected from the feeder. However, if the OLTC tap position remained in the previous optimized value (i.e. Tap position - 4), the voltage profile will not be within the voltage range stipulated in the standard, as shown in Figure 24. The inverter of the solar-PV generator can act as a reactive power support device; however, the inverter should have a high apparent power rating than the active power rating to provide reactive power support across its entire operating range. In scenario 7, a typical load scenario with reactive power support from the solar-PV inverter is illustrated. It is apparent that, the reactive power support provided by the solar-PV systems has reduced the voltage rise by a small amount, however could not able to lower the voltage significantly due to inverter apparent power limit. This manifest the fact that PEC interfaced REGs can provide reactive power support through their converter, but this support is not sufficient enough to achieve a perfect voltage profile for the feeder. However, in this situation an ESS can provide active or reactive power support, and keep the voltage inside the standard voltage limit. This is demonstrated in scenario 8. A three-phase 50 kVA ESS is attached with the system, and it is clearly visible from Figure 24 that the distribution feeder voltage profile is maintained within stipulated limits.

SECTION X. Key Findings and Recommendations for Future Research

• Grid operators should adhere to grid-code standards to maintain stable voltage levels (i.e. upper and lower voltage limits) in their networks. These limits have been breached during high renewable power penetration levels, particularly in distribution networks. Therefore, rigorous technical assessments should be conducted prior integrating REGs in distribution networks to develop strategies to maintain network voltage within stipulated limits.

• Large-scale integration of REGs may lead to voltage and transient stability issues in the power grid, which can be either compensated by the REGs, or by using additional reactive power support devices. Therefore, rigorous technical studies must be conducted by grid operators to accurately quantify the additional reactive power requirements for the network.

• Although, there are variations in reactive power requirements, almost every grid code now requires reactive power capability from REGs. The current reactive power requirements can be achieved by REGs, however in future these requirements must made more stringent to par with the conventional synchronous generator capabilities.

• Among wind generators, PMSGs can provide better reactive power support than DFIGs (assuming only RSC reactive power capability), but a large converter is used in PMSGs, which make it more expensive than the DFIG. DFIG converter ratings are smaller, and thus cost effective, and can provide additional reactive power support by the GSC. Thus, GSC has to be controlled effectively to improve reactive power capability of the DFIG.

• Reactive power support can be enabled in solar-PV inverters by implementing various control schemes. However, their reactive power capability is limited in comparison to synchronous generators. Thus, an ESS can be used at the DC-link of the solar-PV inverter for extra active and reactive power support. Such enhanced capabilities should be encouraged for solar-PV systems through financial intensive schemes.

• FACTs devices (e.g. STATCOM, and DVR etc.) can provide superior reactive power support compared to conventional devices, such as OLTC transformers, and capacitor banks, but utility grid operators should make substantially high investment to acquire such devices in power networks. Thus, it is essential to conduct in-depth techno-economic studies prior deploying FACTs devices in the network.

• Various control schemes have been developed for reactive power management in REGs, such as sliding mode control, model predictive control, current mode control, and soft computing methods. All these control schemes have their own complexity issues and response speed limits. The selection of an appropriate control scheme depends on the controller device (i.e. DSP, FPGA), complexity of the system, and ease of implementation.

• There are various reactive power coordination schemes reported in the literature, such as centralized or distributed coordination. Implementation of centralized schemes are expensive, and hence distributed schemes should be used for better coordination of various reactive power support devices in the power network.

• Reactive power planning issues can be solved using either linear programming, or nonlinear programming. More advanced heuristic methods, such as simulated annealing, tabu search, and evolutionary algorithms can also be used with the increased computational complexity and time.

• Reactive power optimization is necessary for efficient management of the power grid, since more reactive power devices are likely to be installed in the power grid with the increased penetration of renewable power generation. Such optimization schemes would determine the optimal sizes (i.e. device ratings) for reactive power support devices, and optimal reactive power dispatch level for power dispatch intervals.

SECTION XI. Conclusion

A comprehensive review of recent literature reported on reactive power management in power grids with high penetration of REGs was presented in this paper. According to the review, many grid codes specify steady-state reactive power requirements for REGs, however only few grid-codes specify dynamic reactive power requirements. Nonetheless, with the increasing renewable power penetration levels, it is becoming a necessity for all grid operators to specify dynamic reactive power requirements for REGs in their grid codes to maintain a stable and a reliable power grid. The Type-3 (DFIG) and Type-4 (FCWG) wind generators can provide both steady-state and dynamic reactive power to the grid, however this capability is substantially limited at high active power levels. Therefore, additional reactive power compensation devices should be installed at wind farms to provide reactive power capability comparable to synchronous generators. The large-scale solar-PV generation should also provide with similar reactive power support. In addition, reactive power can be controlled in PEC interfaced REGs to achieve various control objectives, such as LVRT, stability and power quality improvement. Selection of a specific reactive power control objective depends on the requirements at the installed location of the REG, and these requirements should be carefully determined by the grid operator.

Various reactive power support devices are also used in power grids, and PEC interfaced devices (e.g. STATCOMs) offer much better dynamic reactive power compensation capability in comparison to conventional devices, such as capacitor banks etc. However, selection of suitable reactive power support device also depends on the economic considerations, since PEC interfaced reactive power support devices are substantially expensive than the conventional devices. Two major factors should be considered when selecting a reactive power control technique; 1) response speed, and 2) control complexity. Although complex algorithms provide better accuracy, they substantially reduce the response speed. Hence, a suitable reactive power control technique should be selected based on requirements of the primary control objective. As power grids require many reactive power support devices under high renewable power penetration levels, coordinated operation of these devices is vital for efficient power network operation, hence power grid operators should implement coordinated control schemes for reactive power devices for optimal operation of the power grid.