The following three subsections introduce concepts which are central to the research focus of this article.
2.1. Background Concept 1: Electromagnetic Compatibility
One area of emerging research examines the electromagnetic compatibility (EMC) aspects of PV installations—this is the overall domain within which this article focusses. The concept of EMC is achieved when a device is able to function “satisfactorily in its electromagnetic environment without introducing intolerable disturbances” [
1]. This typically involves three facets: susceptibility, emissions, and coupling.
Susceptibility studies, such as in [
2] or [
3], for example, may involve the simulation of the propagation and resulting effects of small- and large-signal surges within PV installations. These surges could be caused by electrical fault conditions (either from within the installations themselves, or originating from the grid to which the PV installation is connected), or by external environmental stimuli (such as a direct lightning strike to the installation).
Emissions refers to the generation of electromagnetic energy by a device or installation, and is generally less of a concern for utility-scale PV installations as the chosen equipment typically complies to an accepted EMC standard (such as [
4]), and does not produce interference which would hamper the correct operation of other nearby equipment. An exception, however, is where nearby equipment may be particularly sensitive to electromagnetic interference (EMI). One example of where EMI mitigation strategies had to be employed in order to attain EMC is the Australian Square Kilometre Array Pathfinder (ASKAP) hybrid solar-diesel power station, which was designed in such a way so as not to interfere with the ASKAP and Murchison Widefield Array (MWA) radio telescopes [
5].
Coupling studies can be more nuanced, and, for example, may investigate the susceptibility of a PV installation as a result of the coupling between a radiating electromagnetic source (such as a nearby lightning strike) and the PV installation. This particular topic is investigated using a high-current pulse generator in [
6,
7]. It is investigated in [
8] by mathematically deriving a circuit representation of a lightning-channel-to-PV-module coupling model, and then including this within a circuit simulation. In [
9], the partial element equivalent circuit method (PEEC) is used to produce a circuit representation of the coupling between an actual PV plant and an overhead high-voltage transmission line—the measured and simulated results are then compared.
The aforementioned topic has also been investigated using computational electromagnetic (CEM) simulation packages. For example, CEM results for the induced currents within a single PV module are compared with those from a high-voltage laboratory in [
10]. These CEM results are then extended to a four-module array in [
11]. More cases, including examination of the induced voltages, as well as mitigation strategies, are investigated in [
12]. This type of simulation produces results which are heavily dependent on the accuracy of the impedances of the models employed—especially if the simulations cover a wide bandwidth. The inclusion of nonlinear devices, of which a PV cell is an example, further increase the complexity of the simulations. For a truly comprehensive simulation of this nature, a full-wave 3D CEM simulation needs to be combined with a nonlinear circuit model—this has not been done yet. CST is an example of a CEM software package which offers a circuit solver—similar to Simulation Program with Integrated Circuit Emphasis (SPICE)—which it can (unidirectionally or bidirectionally) couple to its electromagnetic solvers [
13]. It is for this compatibility reason that a SPICE implementation was selected for this study, and an appropriate circuit model is the first step in creating a model which accurately links external electromagnetic stimuli to induced currents and voltages.
In order to begin the process of constructing a suitable circuit model for this particular purpose, the following subsection covers the state-of-the-art of PV circuit modelling.
2.2. Background Concept 2: Circuit PV Modelling State-of-the-Art
When employing circuit simulations for EMC-centred purposes, adequate component model accuracy is pivotal. The most simple circuit which is representative of PV cell operation is put forward by [
14] and comprises a real diode,
D, in parallel with a current source,
. This current source represents the photocurrent, and the magnitude of the current corresponds to the level of irradiance that the PV cell is exposed to. PV cells are only able to produce around 0.6–0.7 V [
14]. Therefore, in practice, multiple cells are generally connected in series. For weatherising purposes, these series-connected cells are often covered by a glass pane, which is surrounded by an aluminium frame—thus forming a PV module. Multiple series-connected modules form a string. When considering the series-connected nature of modules and strings, a quandary arises: how does the aforementioned model respond when different cells are exposed to different irradiance intensities—such as with full or partial-shading? This issue is demonstrated in
Figure 1, where
when the cells experience differing levels of solar irradiance. As each diode would be reverse biased in this circumstance, zero current would flow through each of the real diodes (i.e.,
). Thus, the total module (and string) current would be zero, as would the current delivered to the load,
. Consequently, the power produced by the string, as well as the power delivered to the load, would also be zero. In practice, this is not the case; therefore, a more representative model is needed.
The single diode model (SDM) circumvents this issue by including a parallel resistance,
, to model a P-N junction leakage current,
[
15]. Furthermore, the resistances of the wire leads (connected to the PV cells), as well as the contact resistances between the wire leads and the PV cells, are modelled by a series resistance,
, which conducts a current
[
14,
15]. The extraction of the SDM parameters has been the subject of much research. A well-cited example of this is [
16], where the authors propose and demonstrate the accuracy of a convenient method for extracting the SDM parameters using datasheet parameters at the following points: (1) open-circuit, (2) short-circuit, and (3) maximum power. Furthermore, as a PV installation is generally expected to function for multiple decades, PV models for long-term yield forecasting purposes should also consider the effects of PV cell ageing; [
17] investigates the impact of changes in the ideality factor (due to changes in
and
) on power production. For completeness, it is noted that the current path through a fully-shaded PV cell provided by
introduces a practical problem—heat. Current forced through a shaded PV cell is dissipated as heat. This is referred to as the hotspot phenomenon, which results in both a decrease in the total power production of a module or string, and also a risk of permanent damage to the shaded PV cell should a threshold (known as the critical power dissipation,
) be reached.
To circumvent this issue, a bypass diode may be incorporated. This diode, usually a Schottky diode, is placed parallel to a cell or module to provide an alternative current path in instances of shading [
18].
Figure 2 illustrates this concept—where three PV cells (each depicted by an instance of the SDM) are shown, each in parallel with their own bypass diode,
. The current through a bypass diode is represented by
.
Figure 1.
Current inconsistency encountered with the simplest photovoltaic (PV) cell model during instances of differing solar irradiance. Developed from [
18].
Figure 1.
Current inconsistency encountered with the simplest photovoltaic (PV) cell model during instances of differing solar irradiance. Developed from [
18].
Alternative PV models, such as the two diode model, are sometimes employed to either increase the simulation accuracy (as in [
19]) or to reduce the computational load (as in [
20]). The models discussed thus far, however, are only valid for direct current (DC) conditions (i.e., a single operating point).
For alternating current (AC) conditions, the model shown in
Figure 3 is often referenced [
8,
21,
22]. This model, referred to in this article as the small-signal AC sub-model (SSACSM), incorporates a parallel capacitance,
, and a series inductance,
, to model the PV cell response under dynamic conditions.
models the effect of junction, diffusion, and breakdown capacitance—which are all nonlinearly dependent on the operating point and temperature [
21]. These capacitances also dominate at different operating points.
Junction capacitance is dominant at small positive and negative voltages; diffusion capacitance is dominant above the maximum power point voltage; and breakdown capacitance is dominant when the cell experiences reverse breakdown [
21].
models the series inductance of the wire leads connected to the PV cells [
21]. Other parameters, such as
,
, and
from the SDM, are linearised and replaced by a single equivalent resistance
. As both
and
are dependent on the voltage across the P-N junction,
, as well as environmental conditions (such as temperature and the level of solar irradiance) this model is only valid for small-signal perturbations [
21]. Voltage
refers to the voltage at the output terminals for this small-signal case. Even assuming constant environmental conditions, a large-signal perturbation would result in a change in
, requiring subsequent parameter changes of
and
. One particular strength of this model is its usefulness in the extraction of circuit parameters which either do not, or negligibly, change with environmental conductions (e.g.,
and
). Therefore, they can be derived using a small-signal-based measurement setup.
Figure 4, adapted from and put forward by [
21], illustrates a dynamic PV cell model which implements a variable capacitance,
, that adjusts with bias voltage and environmental conditions, as does the photocurrent,
. Voltage
refers to the instantaneous voltage at the output terminals. This model also aims to improve the reverse bias accuracy of the model with the inclusion of a reverse-biased diode,
, the operation of which is offset by a DC voltage,
. This model, implemented in MATLAB [
23] by the authors, offers sufficient abstraction from the underlying semiconductor physics without overly compromising simulation accuracy for practical purposes. In [
21], DC parameters were extracted using a DC source, and unbiased AC parameters were extracted using a signal generator and an oscilloscope. DC-biased AC parameters were extracted using a combination of the aforementioned instruments, as well as an isolation transformer, in order to allow for the simultaneous application of both an AC and a DC signal. Three frequency points were used:
was extracted at 500 Hz,
was extracted at 100 kHz, and
was extracted at 1 MHz [
21]. Due to its MATLAB-based implementation in [
21], this model is difficult to combine with a CEM simulation—which, as initially explained in
Section 2.1, is the next goal of this line of research. Furthermore, the lack of measured phase information and the risk of sampling complexities (explained later, in
Section 3.2.1) supported the need for the research documented in this article.
This section introduced the commonly-used circuit models for PV cell, module, and array modelling under different conditions. However, when producing a circuit model for the behaviour of a geometrically-complex circuit, especially at high frequencies, one has to consider the possibility of the presence of undesirable circuit elements. These are discussed in the following section.
2.3. Background Concept 3: Undesirable Circuit Elements
In this article, an undesirable circuit element refers to a circuit element which results in unintended operation of the device under test (DUT). Three types of undesirable circuit elements are introduced: parasitic, radiating, and mutually-coupled.
A parasitic element is an unintended and somewhat-unavoidable circuit element which possesses undesirable characteristics. For example, a circuit designer may wish to make use of a typical through-hole technology (THT) capacitor in their circuit. A practical capacitor, however, does not only possess a capacitance (as is the case for an ideal/theoretical capacitor). Individually, wire leads will have both resistance and inductance. A parasitic capacitance will also form between the wire leads. This concept is illustrated in
Figure 5 below, where the ideal capacitance is represented by
, and the parasitic inductances, resistances, and capacitance are represented by
,
, and
, respectively.
The circuit designer may expect the impedance of the real capacitor to decrease as frequency increases, however, if the parasitic inductances are great enough that the ideal capacitance no longer dominates the overall impedance, then the circuit designer will discover that the capacitor behaves more like an inductor at high frequencies. The influence of the parasitic elements, can, however, be diminished by paying attention to the component choice and layout. In the example of the THT capacitor, the wire leads can be shortened prior to installation in order to decrease the parasitic inductances, resistances, and capacitances. Parasitic elements may or may not be within the control of a circuit designer, however, their presence should always be considered if they noticeably influence the operation of the designed circuit.
The second type of undesirable circuit element discussed here is a radiating element. This type of element directly radiates electromagnetic energy. In this case, the DUT may become resonant and actively radiate energy in the form of electromagnetic waves—thereby acting as an (unintentional, in this case) antenna. This may happen as physical size of conductive pathways within the DUT become similar to the wavelengths involved in the testing.
Figure 6 illustrates this concept. In this figure, a DUT is being analysed by a VNA. The VNA produces a signal and interprets the reflected signal of the DUT in order to obtain the desired parameter(s). The DUT, however, contains a radiating element, which broadcasts electromagnetic energy to space. In the context of a transmitting antenna, optimal radiation is desirable. However, an unintentional antenna in an electromagnetically-sensitive environment, such as the area near the aforementioned ASKAP radio telescopes, would be consequential.
Finally, the concept of mutual coupling is introduced. Mutual coupling describes the electromagnetic interaction between circuit elements. This is the core concept in a transformer, where a primary coil produces an electromagnetic field which couples into, and produces a response, in the secondary coil—exploiting the concept of mutual inductance (generally referred to by parameter
M).
Figure 7 illustrates this concept, where
and
represent the self inductance of the primary and secondary coils, respectively, and
M represents the mutual inductance between the coils.
The concept of layout-based parasitic elements, radiating elements, and mutually-coupled elements becomes applicable later, in the results section (
Section 5). The following subsection introduces the particular PV modules selected for this study.
2.4. Studied PV Modules and Motivation
A large and a small PV module, shown in
Figure 8, were selected for this study in order to assist readers in choosing the circuit modelling parameters most appropriate for their particular application. The most-used datasheet parameters of both PV modules are included in
Table 1 below.
The large PV module, a BYD 310P6C-36 (BYD Co. Ltd., Shenzhen, China), is composed of 72 series-connected polycrystalline PV cells [
24]. This topology is seen in many utility-scale PV power plants around the world. Within this module, a bypass diode is generally placed in parallel with every 24 cells. However, for this study, the bypass diodes were detached from the module in order to remove their effect on the measured impedance between the output terminals. For completeness, nonlinear dynamic modelling of bypass diodes is covered in detail in [
18].
The small PV module, an ACDC SLP005-12 (ACDC Dynamics, Edenvale, South Africa), is composed of 36 series-connected polycrystalline PV cells [
25]. This smaller class of PV module is often used in off-grid outdoor lighting and Internet of Things (IoT) based applications, where little power is required yet no method of convenient grid integration exists. This module does not include any bypass diodes.
This section provided the necessary background information for this study. Firstly, the concept of EMC was introduced—including an explanation of the need for accurate circuit models in CEM simulations. Secondly, existing models for PV cells, modules, and arrays under DC, small-signal AC, and transient conditions were discussed—including their shortcomings. Thirdly, the concept of undesirable circuit elements was introduced. Finally, an overview of the PV modules chosen for this study was presented. Altogether, this section highlighted the need for a circuit PV model which: (1) is SPICE-compatible, (2) able to handle both small-signal and large-signal stimuli. These two points became the primary objectives for this study. The following section discusses a proposed model for these two PV modules, suitable for use in SPICE.