Modular PV System for Applications in Prosumer Installations with Uncontrolled, Unbalanced and Non-Linear Loads

Abstract

This article proposes a modular system for prosumer installations composed of photovoltaic (PV) panels and energy storage (ES) integrated with the low voltage (LV) network through a common 4-wire AC/DC inverter. The novel idea is a control strategy for the inverter in which additional functionalities are incorporated. Apart from transmitting an active power generated by the PV source, the same inverter is used to manage energy generated by the PV and to compensate for the current unbalance, harmonics (including subharmonics and interharmonics) and reactive power of the prosumer loads. As a result of the algorithm operation, the currents flowing to the prosumer installation are sinusoidal, symmetrical and purely active, which results in voltage balancing and improving voltage waveforms at the point of common coupling (PCC). In this way, with the widespread use of this solution among prosumers, the impact of the prosumer installation on the distribution network is minimized, and power quality (PQ) disturbances such as unacceptable voltage rises, voltage unbalance and harmonics are avoided. The presented approach may be a solution to the problems network operators face nowadays due to the uncontrolled connection of PV sources. The proposed modular system is also beneficial for the prosumer as the instances of unacceptable overvoltage and, consequently, shutdowns of prosumer installations are reduced. The features of the proposed method are shown in relation to other means applied for PQ improvement in the networks with distributed generation. A principle of the control and the involving algorithm for the inverter is presented. The efficiency of the control strategy was tested in a simulation developed in the PSCAD/EMTDC program. The results of simulations are presented, and the proposed solution is concluded.

1. Introduction

The technological development of renewable energy sources (RES), as well as financial support systems and the quest to increase energy efficiency, have enabled the emergence of prosumer power generation [1,2]. It can be observed that photovoltaic (PV) panels are the most frequently chosen energy sources to be used in single-family houses due to their increasing efficiency, simplicity of installation and constantly decreasing prices [3]. In recent years, the number of PV systems connected to low-voltage (LV) distribution networks has increased considerably.

According to the Polish grid connection codes [4], the installation of PV panels with capacity up to the customer’s contracted power does not require the permission of the distributed network operator (DSO), and so, the network hosting capacity is not verified. In these conditions, the high penetration of PV sources may negatively affect the network operating conditions and result in deterioration of the quality of the energy delivered to end-users [5,6,7,8,9]. One of the most serious threats is an excessive voltage rise in the network nodes caused by the change of electricity flow during periods of high PV generation when excess power in the prosumer installation is introduced to the network. Voltage fluctuations, resulting from the variability of primary energy, and voltage unbalance which is an effect of the use of single-phase PV inverters, also appear [10,11,12]. The presence of unbalanced and non-linear loads utilized in prosumer installations contributes to the deterioration of power quality (PQ) caused by the operation of renewables.

It is obvious that the worsening of the PQ affects all end-users. In particular, an unacceptable voltage rise may lead to switching off the PV inverters [11,12,13], which reduces the use of renewable energy by prosumers. Poor PQ also results in additional power and energy losses in the network and worsening of the power factor. The above technical and economic problems are well-known for distributed system operators (DSOs), and their mitigation is a challenge that they must face nowadays.

Typical LV networks are characterized by relatively low voltage regulation capabilities. They come down to the voltage step changes by means of off-loading tap changers with which the distribution transformers are equipped. The regulation is performed manually and requires a power supply. Obviously, this method cannot be the online solution for the occurring PQ problems. Reinforcement of the network comprising the replacement of distribution transformers with ones having on-load tap changers (OLTC) could help (an example is presented in [14]), but it is difficult due to the high cost of OLTCs and the outage needed for the transformer upgrades [14,15].

The methods applied by the DSOs to mitigate voltage quality problems are based on changing the power flow in the network. This can be obtained through reactive power using PV inverters that operate according to Q-U strategies [16,17,18]. If a voltage reduction is needed, the PV inverter increases the reactive power drawn from the network (cosφ inductive), resulting in a higher voltage drop on the network impedance and, thus, decreased nodal voltages. For networks with many inverters participating in the control, a central system is needed to fairly coordinate the operation of the devices [19]. It should be noted that reactive power generation from the inverter can be limited by its total current carrying capacity. Furthermore, the regulation of voltage by reactive power contributes to an increase in the energy losses in the network and may be ineffective due to the low X/R ratio of the LV networks. In this case, the limitation of active PV power generation is imposed. Obviously, the curtailment of active power reduces benefits for prosumers.

To change the power flow in the network, one can use energy storage systems (ESS) [20,21]. Both active and reactive power may be included in the regulation process. However, if the storage system belongs to a prosumer, it is primarily applied for local energy management aimed at the maximization of renewable energy self-consumption and reduction of energy costs. Using it for voltage regulation purposes interferes with this task and may be done as an ancillary service provided by the prosumer to the DSO, which requires a special service agreement between the parties [22]. In such a case, the aspect of additional battery wearing should be considered [20].

Numerous publications address voltage regulation through active and reactive power with the use of PV or energy storage inverters. Most of them focus on the control strategy, especially if both devices are used [21,23,24,25,26], control algorithms [20,27,28] and the effects of control in dependence on the network impedance [29]. It is shown that these methods and strategies presented are effective in limiting voltage rises and unbalances. However, regardless of what devices are used to change the power flow in the network and how the controllable power is determined, the main disadvantages of these methods remain the same, i.e., additional costs related to appropriate dimensioning and selection of inverters, requirements for measurement and communication systems, knowledge on the network impedances, the need of coordination if many inverters participate in the control and increased network losses. All these features make the voltage control by means of power flow changing complex and costly.

Taking this into account, the authors propose another approach to solve voltage problems in LV distribution networks with RES. This approach is based on the fact that voltage disturbances result from current disturbances that arise in end-user installations and are entered into the network. Thus, the idea is to eliminate current disturbances at the place of its origin using the capabilities of the devices installed there rather than to affect the voltage at the network nodes.

In previous work, authors have dealt with the extended control of photovoltaic inverters [30], including the functionality of compensation of current unbalance and reactive power in the prosumer installation, making it balanced and purely active. Research proved the purposefulness of such a solution and showed that eliminating current unbalance leads to significant mitigation of voltage unbalance at the point of common coupling PCC, which results in leveling voltage phase values and maintaining them within the permissible range. It was demonstrated that with all prosumer installations having the current balancing capability, the network would operate with the required power quality and with minimal power.

This paper presents the results of further work on this topic. It proposes a modular system composed of a PV and energy storage (ES) integrated with the prosumer installation through a common AC/DC inverter. A novelty of this work is the method of inverter control and the control algorithm adapted to the considered system. According to the previous assumption, independently of the load character and changes in load power consumption and PV power generation it makes the currents flowing to the prosumer installation sinusoidal and symmetrical, which in turn contributes to the voltage being sinusoidal and balanced. New elements in this research comparing the authors’ previous work are the system structure and extension of the control with harmonic compensation, including subharmonics and interharmonics.

The proposed control makes the prosumer installation balanced, linear and purely active. If the presented modular system was applied as a common standard, it could provide effective mitigation of voltage disturbances and required power quality in the network in a relatively simple way and moreover would ensure network operation with minimal losses.

The paper is arranged as follows. In Section 2, the system under study is presented, then in Section 3 the proposed control strategy for AC/DC inverter is described. Section 4 applies to the analysis of simulation results. The paper ends with conclusions and a summary in Section 5.

2. System under Study

The considered infrastructure is a prosumer installation connected to the LV distribution network, as shown in Figure 1. In a general case, the installation consists of unbalanced and non-linear loads. PV panels and an ES are integrated into the prosumer installation by a three-phase, 4-wire inverter. The PV panels (equipped with a DC/DC converter with MPPT—Maximum Power Point Tracking module) and the ES are connected to the common DC busbar.

Battery energy storage is primarily used to increase the self-consumption of energy generated in photovoltaic panels and to minimize energy costs incurred by the prosumer.

A simulator was built using the PSCAD/EMTDC environment to illustrate the operation of the AC/DC inverter with the developed control algorithm [31]. The components of the prosumer installation were modeled using elements from the PSCAD library. The diagram of the model is presented in Figure 2.

The prosumer non-linear and unbalanced load was modeled using three single-phase components. A source of subharmonics and interharmonics was located in phase L1 as a current source generating the sinusoidal current of constant amplitude (5 A) and frequency varying from 5 Hz to 550 Hz and to a triangle wave. A 4 kW resistive load was connected to the L2 phase, and a 3.3 kW Graetz bridge rectifier was connected to the L3 phase.

A three-phase, 4-wire AC/DC inverter was modeled as current controlled. The hysteresis PWM technique was used to generate the reference current waveform determined in a microcontroller according to the developed algorithm. The PV panel was connected to the DC busbar through a DC/DC converter. The MPPT module for the DC/DC converter applied the ‘Perturb and Observe’ tracking algorithm.

The battery storage was connected directly to the DC busbar. The direct connection was possible because it was assumed that the rated voltages of the DC busbar and ES were equal (850 V—the DC value was acquired experimentally). Because the main objective of the study was to develop the inverter control algorithm, the type of battery, its capacity and energy losses were not considered further.

3. Inverter Control Strategy

The AC/DC inverter, with an implemented control algorithm, integrates the PV and ES in the prosumer installation. As assumed, it should also mitigate the negative impact that the prosumer infrastructure may have on the operation of the LV distribution network. It is done through the compensation of current unbalance, harmonics (including sub- and inter-harmonics), as well as reactive power of the loads.

According to the above assumption, the currents of the prosumer’s installation ${\mathit{i}}_{pros}$ (see Figure 1 and Figure 2) should form a balanced, sinusoidal and purely active system expressed by the following equation.

$${\mathit{i}}_{pros}=I_{pros}\ast {\mathrm{e}}^{j{\omega }_{o}t}$$

(1)

where

• ${\mathit{i}}_{pros}={\left[i_{L1}{i}_{L2}{i}_{L3}\right]}^T$—currents vector;
• $i_{L1}{i}_{L2}{i}_{L3}$—phase total currents of the prosumer installation;
• $I_{pros}$—the currents module;
• ${\omega }_o$—pulsation of the supplying voltage.

As can be seen in Figure 1, the total installation currents are the sum of load and inverter currents.

$${\mathit{i}}_{pros}={\mathit{i}}_{load}+{\mathit{i}}_{inv}$$

(2)

where

• ${\mathit{i}}_{load}={\left[i_{load L1}{i}_{load L2}{i}_{load L3}\right]}^T$—load currents vector;
• ${\mathit{i}}_{inv}={\left[i_{inv L1}{i}_{inv L2}{i}_{inv L3}\right]}^T$—inverter currents vector.

Taking into account Equations (1) and (2), the inverter currents can be calculated.

$${\mathit{i}}_{inv}=I_{pros}\ast {\mathrm{e}}^{\mathit{j}{\mathit{\omega }}_{\mathit{o}}\mathit{t}}-{\mathit{i}}_{load}$$

(3)

The presented algorithm controls the AC/DC inverter in such a way that, regardless of the instantaneous power consumption and the PV generation, the total currents of the prosumer’s installation are balanced, sinusoidal and purely active. The amplitude of these currents results from the active power exchanged with the network, which determines the operation schedule of the ES according to the prosumer’s needs. Obviously, to minimize energy costs incurred by the prosumer, it should depend on the applied energy tariff. Typically, the ES is discharged during high energy prices and charged during low energy prices.

The algorithm that should be responsible for determining the power exchange set point $P_{SetPoint}$ and the resulting value of the total current amplitude $I_{SetPoint}$, here called a ‘prosumer program’, is not considered in this article. It is assumed that it would be a part of an external controller—such as a Building Management System (BMS). That controller would be equipped with technical measures to acquire valid energy tariffs (especially if dynamic tariffs are applicable) and be easily adoptable by the prosumer (through HMI—Human Machine Interface). Summarizing, we assume that in prosumer installation with ES connected to the AC/DC inverter, the current $I_{pros}$ is set to be equal to the externally obtained $I_{SetPoint}$ and it allows us to calculate the inverter currents ${\mathit{i}}_{inv}$ according to Equation (3).

Note that the inverter can also operate without ES, implementing the assumed functions. In such a case, to ensure that all energy is transferred from the PV modules, it is necessary to control the energy of the capacitor connected to the DC busbar by stabilizing the DC busbar voltage $U_{DC}$. The voltage $U_{DC}$ should be equal $U_{DCnom}$, at which the inverter is able to generate all the required harmonics. The value of $U_{DCnom}$ was acquired experimentally and is equal to 850 V. Thus, in the case of the PV system without ES, the amplitude of total prosumer installation currents $I_{pros}$ is defined by Equation (4).

$$I_{pros}=k{\int }_0^t({U_{DCnom}-U}_{DC})dt$$

(4)

where

• k—a scale coefficient;
• $U_{DCnom}$—the nominal voltage required on the DC busbar;
• $U_{DC}$—the voltage measured on the inverter DC busbar.

It should be emphasized that in the system with the ES connected, the DC busbar voltage is stabilized by batteries (the battery voltage, related to the state of charge, determines the DC busbar voltage).

The flow chart of the presented algorithm is shown in Figure 3. Its layout was divided into blocks. At first, the variables ‘DSO’ and ‘B’ are set to a default value equal to 1. The first variable is used to consider possible power limits imposed by the network operator, and the second is the coefficient used to protect the inverter against overloading.

In block A, the amplitude of the total current of the prosumer installation $I_{pros}$ is memorized in the variable I_{pros_mem}. The availability of the ES is verified, and the value of the prosumer current $I_{SetPoint}$ is acquired (from the external controller). Then, in the system with ES, the $I_{pros}$ is set to $I_{SetPoint}$ and if the ES is not connected, the $I_{pros}$ is calculated based on the measured DC voltage (Equation (4)).

In block B, the vector of load currents ${\mathit{i}}_{load}$ and the vector synchronized with the network phase voltage ${\mathrm{e}}^{\mathit{j}{\mathit{\omega }}_{\mathit{o}}\mathit{t}}$ are determined.

In block C, the reference current vector ${\mathit{i}}_{ref}$ for the AC/DC inverter ${\mathit{i}}_{inv}$ is calculated. Because the inverter should not be overloaded, the control algorithm compares the calculated currents, ${\mathit{i}}_{inv}$, with its ratings, i_lim. If the calculated currents exceed the limits, the algorithm reduces the B coefficient (the default value is ‘1’). The algorithm operates in the loop, so it provides both a reduction of the B coefficient and the withdrawal of the limitations.

In block D, the active power generated by the prosumer installation can be decreased on a request from the network operator in the event of excessive power in the system. If such a request is provided, block A is omitted, and the excessive power from PV sources is stored in ES, or the efficiency of the PV module is decreased. The logical condition ‘DSO < 0.95 and I_{pros_mem} < 0’ determines the operation of the algorithm. If the network operator requests to decrease the generation (DSO < 0.95) while the prosumer sends power to the LV network (I_{pros_mem} < 0) block A is omitted (the loop consists of blocks B, C and D). Otherwise, blocks A, B and C are applied. This way, data calculated by blocks D and A cannot reach block B at the same time.

4. Simulation

The simulation was carried out using the prosumer installation model presented in Figure 2. The simulation lasted 35 s, during which the AC/DC inverter was tested in configuration with PV and with PV and ES. During the simulation, the loads were switched on and off in cycles of different time durations. The load cycle duration in phase L1 was 0.66 s. In phase L2, the load was switched on and off every 0.833 s, and in the L3 load cycle, the time duration was 1.25 s. The simulation results for selected time windows are presented in the figures below.

Figure 4 corresponds to the system configuration with PV panels. The colors of the load current waveforms are as follows: the L1 phase—pink (neon), the L2 phase—green and the L3 phase—brown. During the first 50 milliseconds (from 16.20 s to 16.25 s), there was no solar radiation and the inverter acted as a compensator of load phase current unbalance and distortion (Figure 4c). The radiation (‘Radiation’—red) appeared in 16.25 s (Figure 4b), and the AC/DC inverter started to transfer power. At first, it resulted in the decrease of the power P_pros and then the change of its sign (the power began to flow into the LV network), Figure 4c.

A short increase of exported power P_pros after 16.65 s was caused by turning off the load in the L2 phase. The subsequent reduction in the power was the result of the power limitation (by 50%) imposed by the DSO in 16.7 s, Figure 4b.

The instantaneous values of the AC/DC inverter phases and neutral currents are presented in Figure 4d.

The operation of prosumer installation with PV panels and ES connected with the AC/DC inverter is depicted in Figure 5. In a system with ES, the level of power exchange between the prosumer installation and the network (P_setpoint) is determined by an external controller in a prosumer program. The AC/DC inverter maintains this value regardless of the load power and PV generation. As shown in Figure 5c, from 20.9 s to 21.00 s, the P_setpoint was equal to 5 kW and power was drawn from the network. In 21.00 s, the P_setpoint was changed to −10 kW, and the power began to flow to the grid. In addition to the fact that the exchange power was kept at the set value, the phase currents formed a balanced and sinusoidal system, Figure 5c. Only the effect of the rectifier commutation in phase L3 (high-frequency currents) was not fully compensated.

To better illustrate the effectiveness of the inverter operation, Figure 6 shows the waveforms for a shorter time range (four periods). During the considered time, the load current amplitudes and the solar generation were constant.

Small distortions of the L3 waveform, Figure 6c (also visible in Figure 4 and Figure 5) result from the strongly non-linear nature of the load in this phase—almost square shape (Figure 6a).

The presented results show that the AC/DC inverter of the PV system was able to compensate the currents from the prosumer installation without significant delay after each change in the loads and PV power. The step changes of loads, PV generation and DSO control signal were chosen on purpose, as they pose the greatest challenge for the operation of the control system. The expected effects were obtained for both tested configurations, i.e., the system with PV only and with PV and ES, without the need to modify the implemented control algorithm.

The AC/DC inverter effectively compensated all components of the total prosumer currents that were the cause of phase unbalance and harmonics in steady and transient states.

5. Conclusions

In LV distribution networks characterized by high penetration of volatile renewable energy sources and in the presence of non-linear and unbalanced loads, preserving the required power quality is an increasing challenge for the network operators. Because traditional methods of power quality improvement require costly investments and are time-consuming, it is justified to look for alternative solutions.

The main contributions of the presented work are as follows:

• The paper introduces the modular system intended for PV and ES integration and local management of power in prosumer installations. It consists of a three-phase, 4-wire AC/DC inverter with a common DC busbar for both PV panels and ES. The AC/DC inverter can operate in connection with both PV panels and ES or only with photovoltaics with the same algorithm.
• The presented algorithm for the inverter control enables balancing currents in the prosumer installation, eliminating current harmonics, subharmonics and interharmonics and compensating reactive power. This control makes the prosumer’s installation a controllable unit and mitigates its negative impact on the supplying network in terms of voltage quality. In other words, instead of ‘treating’ voltage disturbances, the proposed system eliminates its causes.
• The tests performed using the PSCAD simulator of the prosumer installation proved that the system operates as intended. Its responses to even harsh (step) power changes are satisfactory when executing commands from DSO.

From the point of view of the network operator, the prosumer installation equipped with the proposed modular system becomes ideal because its negative impact on the network in terms of power quality is negligible. Moreover, the network energy losses are reduced, and the network hosting capacity is increased.

The presented approach is also beneficial for prosumers. As the voltage parameters at the PCC are within acceptable limits, there are no automatic shutdowns of PV inverters and noticeable losses due to energy nonproduction. A possible disadvantage is related to the additional power losses resulting from the increased number of functions performed by the AC/DC inverter. The losses may strongly vary. For example, losses depend on the efficiency of selected electronic components, which is expected to increase due to recent developments in electronic systems (including adopting gallium nitride—GaN and silicon carbide—SiC in compound semiconductors). This aspect is not discussed in this paper and will be considered in future work.

In the author’s opinion, the proposed modular system, if commonly applied in prosumer’s installations, could contribute to solving many problems that DSOs currently experience in LV networks with high levels of renewable generation. Operators may obtain the effect of improved power quality, reduced energy losses and additional hosting capacity without the need to implement advanced management systems, which generally require efficient communication infrastructures. If so, it would be advisable for DSOs to offer some financial support to prosumers to encourage them to invest in the proposed solution.