Broad-Spectrum Technical and Economic Assessment of a Solar PV Park: A Case Study in Portugal

Abstract

While technical optimization focuses on maximizing the annual energy yield of utility-scale PV parks, the ultimate goal for power plant owners is to maximize investment profit. This paper aims to bridge the gap between technical and economic approaches by using simulation data from a real-case utility-scale PV park. It analyzes how changes in configuration parameters such as the DC–AC ratio and string length and PV technologies like solar tracking systems and bifacial modules impact the economic metrics of the project, i.e., net present value (NPV) and internal rate of return (IRR). PVSyst software was utilized as a simulation tool, while in-house developed software implementing appropriate technical and economic models served as a comparison platform and was used to validate the outputs generated through PVSyst. Results indicate that the commonly used horizontal single-axis tracking configuration may economically underperform compared with fixed-tilt setups. The optimal DC–AC ratio fell within the range of 1.30 to 1.35. Extending the string length from 25 to 28 modules improved economic indexes. Additionally, fixed-tilt bifacial modules can enhance project economics if a 10% cost premium compared with standard monofacial PV modules is considered.

1. Introduction

Energy yield, along with related parameters such as capacity factor and specific production, stands out as one of the key variables for evaluating the performance of solar photovoltaic (PV) assets. In the initial phases of developing a PV power plant, thorough simulations are conducted to assess energy yield. These simulations, performed by technical advisors and project developers, delve with meticulous detail into the impact of design decisions on the asset’s production.

Engineering teams responsible for these tasks often lack insight into the economic implications of their technical decisions, placing more emphasis on production-driven variables. When assessing the technical impact of emerging technologies like bifacial modules or dual-axis trackers, project designers frequently lack accurate information on how these advancements translate directly into project economic indicators such as net present value (NPV) and internal rate of return (IRR). While there may be some economic rationale present in industry practices, leading to the adoption of certain technologies over others, an economic optimization process is often overlooked among established and bankable technologies. As a result, basic technical decisions are often made with the primary goal of maximizing annual energy yield rather than optimizing economic indicators.

In the early 2010s, improving the efficiency of solar modules relied heavily on enhancing the efficiency of PV cells, with the technology being relatively straightforward and well documented. However, today, the landscape has evolved significantly, offering a plethora of options and configurations for PV park construction. Design decisions made during the engineering and feasibility phases can now substantially impact the project’s economic viability. Furthermore, the costs associated with solar energy production have seen a significant decline, accompanied by a reduction in balance of systems (BoS) costs, encompassing expenses related to wiring, mounting racks, and solar inverters, among other essential components. This reduction in costs across the value chain has heightened competition within the industry. These dynamic shifts in cost structures highlight the complexity of economic optimization exercises when building and operating solar assets.

The current literature on solar PV parks primarily focuses on optimizing energy yield, often ignoring the associated economic implications. Existing studies discussing both technical and associated economic consequences address only the change of one configuration parameter or technology. This paper fills this gap by conducting a comprehensive analysis considering both technical and economic aspects of several configuration parameters (DC–AC ratio and number of modules per string) and technologies (fixed-tilt, single-axis, or dual-axis trackers and monofacial versus bifacial modules).

The key contributions of the paper that, to the best of the authors’ knowledge, cannot be found in the literature are:

• Evaluation of the technical and economic implications of different unconventional technical choices (such as solar tracking technologies, DC–AC ratio, string size, bifacial modules) during the design phase of a large-scale PV park.
• A contrast of the technical and economic efficiency of these alternative design approaches within state-of-the-art projects, utilizing a real-world case study of a 48 MWp solar PV power plant situated in southern Portugal.

The current literature lacks recent studies that enable a comparison using the latest cost data for various unconventional solutions, as presented in this paper.

The simulations were conducted using PVSyst software (Satigny, Switzerland, https://www.pvsyst.com) [1]. PVSyst offers comprehensive simulation capabilities for PV system design, encompassing solar radiation modeling, shading analysis, PV module performance characterization, and electrical system modeling. This enables accurate performance predictions under various conditions. Through considering factors such as solar radiation, temperature, shading, system losses, and component characteristics, PVSyst accurately predicts the energy yield of PV systems. This aids in optimizing system design for maximum energy output. Users can refine the PV system design through evaluating configurations such as panel orientation, tilt angle, and system size, allowing performance comparison and the selection of the most cost-effective solution. Detailed financial analysis tools, including metrics like net present value (NPV) and internal rate of return (IRR), assist in assessing the economic viability of PV projects and making informed investment decisions.

It is worth mentioning that software code was developed by the authors, implementing the models outlined in Section 3, with the purpose of checking PVSyst output results. An error of less than 5% was found, validating the models presented in this paper. This error was explained using a simplified model for computing the energy yield of the PV park (one diode and three parameters) in the developed software, while PVSyst used a more detailed model (one diode and five parameters).

This paper is divided into six sections. In Section 1, the motivation and objectives and contributions of the work are described. Section 2 provides a review of the state of the art through the analysis of studies that have already been carried out in related subjects. The applied methodology, as well as the parameters used for the simulations carried out both in PVSyst and in the in-house developed software, are offered in Section 3. Section 4 presents the simulation offtakes of this work; the results are analyzed, and the economic model parameters are studied and compared. Section 5 provides an overview of the technical decisions currently being taken by promoters and subcontractors in projects in Portugal, framing them against the results obtained in this study. Finally, some conclusions are drawn in Section 6.

2. State of the Art Review

As mentioned before, there is not much literature that addresses the economic impact of each technical decision aimed at optimizing the energy yield of a PV park. This difficulty in finding relevant articles is also related to the scale of the parks being the object of study, since it is possible to observe several works published in relation to projects for self-consumption (up to 3 kW) [2], self-sufficient communities or commercial rooftop PV modules (up to 100 kW) [3], and small power plants (of a few hundred kW, but less than 1 MW) [4]. Applications in practical cases of utility-scale plants are scarcer, given that their proliferation is a more recent event, and their study is still quite limited to the industry itself. Another factor in the inadequacy of the studies already published is the aggressive change in the costs associated with the technological elements of PV projects. An approximate 90% drop in the cost of PV modules since 2010 makes the premises of most technical–economic analyses obsolete.

Silva et al. [5] conducted an optimization study of a large-scale power plant, similar to the one proposed here, although taking a different approach regarding some of the technical parameters—DC–AC ratio was not optimized, for example, and a standardized approach was followed instead. Also, the maximization of economic viability and specifically the values for the variation of CAPEX and OPEX with the different technologies were directly supplied by the partner company without an analysis and benchmarking of the industry values having been carried out.

In another study, Ullah et al. [6] evaluated and compared several possible sites for solar projects from a technical and economic point of view, a component that the current study did not consider as the contracted land and grid connection permit were binding elements within the decision. Some studies have tried to include an even broader approach but in different directions. Ryan et al. [7] proposed a multidisciplinary approach considering, in addition to the technical and economic component, the maximization of social welfare and benefits for the final consumer, presenting contributions made by the projects in this aspect as important non-price criteria to be taken into account, together with a levelized cost of energy (LCOE)-based analysis.

Regarding the DC–AC ratio, several studies have aimed at perceiving the effect of oversizing on the economic viability of projects. Mondol et al. [8] explored the optimization of PV–inverter sizing ratios in multiple locations in Europe, concluding that the optimum ratio varied from 1.1 to 1.3. The takeaways were obtained from changing the inverter’s input-rated capacity and keeping the PV-rated capacity constant, an approach substantially different from the one carried out in this study. In this context, the inverter-rated capacity was kept the same while more modules were added or removed from the PV array. A similar approach was proposed by Mahmoud et al. [9], using an iterative method to obtain an optimum DC–AC ratio of 1.42 for a 30 kWp PV array, although the optimization process used was based entirely on the maximization of conversion efficiency, based on irradiance and temperature, and did not consider any economic input.

Regarding the maximum number of modules in series, thoroughly developed studies are scarce, and the subject has often remained an under-explored optimization in specialized magazine articles. Sporadic reports, such as the one from the group led by Ladd [10] have explored maximum voltage calculations based on manufacturer-provided temperature coefficients, stating that they are unnecessarily conservative, and offering room for optimization. A statement supported by Brooks [11] explains that the record low temperature is usually too conservative for design calculations because temperature is only one of two major factors that impact an array’s open-circuit voltage—the other being irradiance. These are theoretical concerns that raise relevant points but do not translate into a practical exercise of optimization. That type of practical approach was developed in the work conducted by Karin and Jain [12]. There, a new methodology was proposed to develop longer strings and therefore lower total system costs. Using historical weather data, strings with a size around 10% longer were projected, still maintaining system voltage within the electrical limits.

Regarding the optimization of systems using bifacial modules and tracking systems, Rodriguez-Gallegos, et al. [13] computed single-axis tracking bifacial LCOE in multiple locations across the globe, comparing it with the dual-axis tracking bifacial method. Although the energy produced was higher on the second type of system, the LCOE was lower than for the single axis given the associated initial investment and operation and maintenance (O&M) cost. Talavera et al. [14] also studied the cost competitiveness of five PV projects with the same objective as the present study: to understand the optimum balance between the additional cost of trackers and yield increases. Although that study had the merit of being conducted with the novelty of including reference to electricity tariffs, that information was not relevant to the present study as a fixed power purchase agreement (PPA) price was assumed for electricity in this case. Their conclusions stated that all five projects registered a lower or equal LCOE with the fixed-tilt solution, but the study indicated that for similar LCOE, it was preferable to install a single-axis tracking PV system due to its higher energy yield.

The higher costs of tracking systems were also addressed in a study from the American National Renewable Energy Laboratory (NREL), by Lisell and Mosey [15], which emphasized the difference in O&M expenditure between fixed and tracking systems. That work concluded that the necessary moving parts and a higher rate of demanded maintenance were found to result in a 100% increase in the operational expenditure of tracking systems compared with fixed ones. Regarding extra unusual variables to be analyzed while assessing the viability of fixed versus tracking systems, Adinoyi and Said [16] pointed out that tracking systems could be beneficial in reducing dust accumulation by 50%, due to the motion effect that helps counter-balance the higher temperature of modules that use trackers.

A burgeoning technology attracting researchers’ attention is Floating PV (FPV), involving the installation of PV parks on water bodies like lakes, hydropower reservoirs, and ponds. Islam et al. [17] examined the techno-economic feasibility of a 10 MWp FPV in Malaysia using PVSyst, reporting a capacity factor of 20.5%, an investment of EUR 0.85/Wp, and an LCOE of EUR 49/MWh. Similarly, a study in Bangladesh [18] investigated a 50 MWp FPV, yielding a capacity factor of 15.5%, with an investment of EUR 1.12/Wp and an LCOE of EUR 48/MWh. Additionally, research in Iran [19] demonstrated that a 10 MWp FPV installed in a hydropower plant reservoir could significantly reduce water evaporation, providing an added benefit.

PV parks are also being utilized as renewable energy sources for green hydrogen production. Park et al. [20] developed an optimization method to size electrolyzer capacity relative to PV capacity in the United States, China, Australia, and Korea, aiming to maximize the benefit–cost ratio. Additionally, the economic potential of green hydrogen production from PV electricity that would be otherwise curtailed has been explored [21].

A 1600 MWp PV park with solar tracking in Egypt was technically and economically assessed [22], revealing a capacity factor of 27%, an investment cost of EUR 1.38/Wp, and an LCOE of EUR 75/MWh. In India, Boddapati et al. [23] conducted a techno-economic analysis of a 50 MWp solar PV park, finding a capacity factor of 24% and an investment of EUR 0.6/Wp. Furthermore, the integration of a utility-scale PV park with a battery system on the island of Mauritius was examined [24], with evaluation of technical impacts such as battery size, inverter loading ratio, and curtailment on the LCOE, demonstrating comparable costs to fossil-fired peaking generators, the conventional technology used to supply peaks.

The literature review did not reveal evidence of integrated approaches incorporating assessment of the economic implications of changing multiple configuration parameters and PV technologies. Through proposing four different variations to the base case of a real-world project, this work aims at merging scattered techno–economical assessments and industry knowledge.

The design of PV parks is typically conducted using standardized solutions that can be swiftly deployed. These solutions are widely accepted with minimal discussion, accompanied by a traditional economic analysis. While these solutions are technically proven, widely used, and often economically attractive, there is no guarantee that better design alternatives do not exist. This paper aims to demonstrate that state-of-the-art solutions might not always lead to the best economic outcomes. Therefore, evaluating the lifetime economic implications of other available alternatives is worthwhile, as it provides stakeholders with a holistic view of the project, explores different approaches, and enables more informed decision making.

An example may help illustrate this point. The state-of-the-art configuration typically deploys 25 modules in series per string. This number is based on the maximum open-circuit voltage, calculated using a conservative minimum temperature that is rarely reached, and does not consider irradiance. If long-term recorded values of temperature and irradiance are taken into account, 28 modules could be connected in series. An economic assessment of this option is necessary to determine whether this unconventional configuration is more economically attractive than the state-of-the-art solution. This type of analysis is conducted in this paper.

3. Methodology

This work’s main objective is to propose an approach to assess the impact of technical options on the economic performance of a utility-scale solar PV park, using PVSyst software [1] as a simulation tool. PVSyst is a widely adopted software in the PV industry, used by both companies and academics. As mentioned before, in-house developed software was built incorporating the models presented in this Section and its results were compared against PVSyst. The input parameters were the same for both pieces of software.

In the present work, the imported specifications reflected the project data of the real-case development asset made available by the project owner. This information made it possible to structure the base case, which consisted of the project configuration as it was originally projected and as is expected to be built. The base-case specifications consisted of a 47,992 kWp project, using fixed-tilt structures at an angle of 20° with 25 monofacial modules in each string. The DC–AC ratio was 1.24, and the considered albedo was 0.2.

To study the impact of technical variations in terms of energy production and economic impacts, this work evaluates the energy yield over the 30-year project lifetime. To accomplish this, the key component “PV Degradation Rate” of PVSyst was used, allowing the simulation of equipment ageing, considering a progressive loss of efficiency. The range of values predicted in the literature for the annual degradation factor is wide. Given the absence of a paradigm establishing an accurate value to be considered, this paper keeps in line with the industry-standard value of 0.5% per year. A value of 1% was also considered for light-induced degradation losses—the loss of performance in the first hours of operation, derived from the initial exposure to the sun that affects the functioning of the crystalline modules.

3.1. Economic Model

The developed financial model includes, as the main parameters to be evaluated, the net present value (NPV) and the internal rate of return (IRR). These are the most common indicators considered in the evaluation of such investments and are widely referred to in the literature.

NPV is given by Equation (1), and consists of the difference between positive and negative cashflows over the project lifetime, discounted to the current moment:

$$NPV=\sum _{j=1}^n\frac{R_{Nj}}{{\left(1+a\right)}^j}-\sum _{j=0}^{n-1}\frac{I_j}{{\left(1+a\right)}^j}$$

(1)

where $a$ is the discount rate, $n$ is the project’s lifetime, $I_j$ is the investment in year $j$, and $R_{Nj}$ is the net revenue for year $j$, i.e., the difference between the gross revenue derived from energy selling and the annual O&M expenses.

The net revenue is computed through ${R_{Nj}=p}_{sj}{E}_{aj}-c_{O\&Mj}$, where $p_{sj}$ is the selling price of each unit of PV produced electricity in EUR/MWh and in year $j$, $E_{aj}$ is the annual electricity produced by the PV park in MWh in year $j$, and $c_{O\&Mj}$ is the annual operation and maintenance cost in EUR in year $j$. The investment in year $j,I_j,$ is the corresponding capital expenditure (CAPEX) in EUR. Details of the values used are given in Section 3.2 and Section 3.3.

The IRR consists of the discount rate to be considered to obtain a null NPV. A null NPV represents the discount rate that allows recouping of all the costs and receiving an associated premium, depending on the discount rate used, without any additional monetary benefit. This index allows comparison of different projects to improve the allocation of the available capital (from an opportunity–cost comparison perspective). The IRR is computed as follows:

$$0=\sum _{j=1}^n\frac{R_{Nj}}{{\left(1+IRR\right)}^j}-\sum _{j=0}^{n-1}\frac{I_j}{{\left(1+IRR\right)}^j}$$

(2)

A discount rate of 6%, a project lifetime of 30 years, and a pool price of EUR 38/MWh were considered, as these values were used in the economic assessment carried out for the real PV park being followed in this comparative study. The energy pool value is the price at which energy is sold under a PPA contract.

3.2. Project Costs

CAPEX values were based on the market and a review of the literature validated as applicable by the project owner (

Table 1. Breakdown of CAPEX costs for fixed-tilt systems.

Category	CAPEX [EUR/Wp]	CAPEX [%]
PV modules	0.18	30.1%
Support for PV modules	0.095	15.9%
Grid connection	0.08	13.4%
Settings and others	0.055	9.2%
Electrical components	0.05	8.4%
Inverters	0.045	7.5%
Civil works	0.045	7.5%
Mechanical assembly	0.03	5.0%
Insurance	0.013	2.2%
Studies and analysis	0.0055	0.9%
Transport, accessories	Included	n/a

n/a = not available.

), while OPEX values were calculated based on a review of the literature (

Table 2. Literature review of O&M costs for fixed-tilt systems.

Source	OPEX [EUR/kWp/Year]
[14]	2.25
[15]	8.50
[25]	20.00
[26]	14.20
[27]	21.25
Average	17.04

). These values were then determined for each section, considering the proposed variations to the base case previously described. The value for the initial investment considered in the calculations was obtained through multiplying the number of peak watts (Wp) installed by the cost per Wp. This cost totaled EUR 598.5/kWp and included the components described in Table 1.

Table 2 presents a summary of the literature reviewed, from which was determined the average value to be considered for the O&M costs, also known as OPEX, of fixed-tilt systems—EUR 17.04/kWp/year.

The CAPEX and OPEX costs mentioned previously were then used as a starting point to reflect the impacts resulting from each technical variation considered in this study. New configurations such as horizontal single-axis tracker (HSAT) systems and bifacial modules, among other changes, naturally have great impact on these values, as discussed below.

3.3. Open Circuit Voltage

One of the assessments proposed ahead entails the use of a model to calculate the open circuit voltage of each PV module. Over the years, multiple ways of calculating the number of modules in series that guarantee proper functioning of systems have been developed. Researchers have come up with a simple yet safe framework that solar array engineers can apply to design the maximum number of modules in series—a component that has great impact on the solar park layout. Each PV string is often associated with a single table structure, which supports the corresponding string modules. The definition of the maximum number of modules in series is used by designers to establish the standard structure unit, which is then multiplied across the available area, impacting the possible arrangements. Utility-scale solar parks tend to have hundreds or thousands of these tables arranged around natural exclusion areas, sloped terrains, and other occurrences. Their arrangement is, therefore, of great importance in the engineering phase.

The maximum open circuit voltage ($V_{oc}^{Max}$) is determined based on the minimum expected module temperature ($T^{min})$ and the open circuit voltage temperature coefficient (${\mu }_{Voc})$, available in the module datasheet, which specifies how much the open circuit voltage is decreased if the module temperature increases by one degree Celsius. It is given by Equation (3) [28]:

$$V_{oc}^{Max}=V_{oc}^{STC}\left[1+{\mu }_{Voc}\left(T^{min}-T^{STC}\right)\right]$$

(3)

where $V_{oc}^{STC}$ is the open circuit voltage specified by the manufacturer under standard test conditions (STCs: irradiance = 1000 W/m² and module temperature = 25 °C), and $T^{STC}$ is the temperature under STCs (25 °C). The obtained value is then used to calculate the maximum number of modules to be associated in series, as per Equation (4):

$$MaxStringSize=Floor\left(\frac{V_{Inv}^{Max}}{V_{oc}^{Max}}\right)$$

(4)

where $V_{Inv}^{Max}$ is the inverter maximum DC input voltage. The floor operation rounds the result to the greatest integer less than or equal to it, giving the number of modules to be associated in series.

Equation (3) represents an arguably conservative approach as it considers an extreme value for the minimum module temperature, usually −10 °C, independently of the recorded historical temperatures. Furthermore, the influence of irradiance on the open circuit voltage is not reflected in this equation. It assumes the STC irradiance (1000 W/m²) for the incident irradiance, which normally does not occur when the module temperature is −10 °C—as the minimum temperatures are registered during the night when the irradiance is null.

The alternative proposed in this work entails using Equation (5) [28] to calculate the open circuit voltage, considering site temperature and irradiance calculated in the reference meteorological year dataset:

$$\begin{array}{c}{V}_{oc}^{Max}=MAX\left[V_{oc}^{STC}\left[1+{\mu }_{Voc}\left(T_i-T^{STC}\right)\right]+m{V}_{Ti}\ln \frac{G_i}{G^{STC}}\right]\hfill \\ i=1,2,\dots ,8760\hfill \end{array}$$

(5)

where $m$ is the diode’s ideality factor, $V_{Ti}$ is the thermal voltage for temperature $T_i$, $G^{STC}$ is the irradiance at STC conditions (1000 W/m²), and $G_i$ and $T_i$ are, respectively, the irradiance and temperature registered each hour at the specific site in the reference meteorological year dataset. The maximum open circuit voltage is the computed maximum value obtained from each of the hourly temperature and irradiance records registered in the reference meteorological year dataset. This approach allows one to consider the impact of both temperature and irradiance on the open circuit voltage, using on-site recorded measurements.

To model the PV modules, the one diode and three parameters (1D + 3P) model was used [29]. The fundamental equation of this model states that the current changes with the voltage, as follows:

$$I=I_{SC}-I_0\left(e^{\frac{V}{m{V}_T}}-1\right)$$

(6)

The 3 parameters of the model are the short circuit current, $I_{SC}$, the inverse saturation current, $I_0$, and the diode’s ideality factor, $m$. To determine these parameters, the fundamental Equation (6) includes the short circuit ($V=0$), open circuit ($I=0$), and maximum power ($V=V_{MP};I=I_{MP}$) points, respectively. After some manipulation of the equations, the following relationships are obtained:

$$I_0=\frac{I_{SC}}{e^{\frac{V_{OC}}{m{V}_T}}-1}$$

(7)

$$m=\frac{V_{MP}-V_{OC}}{V_T\ln \left(1-\frac{I_{MP}}{I_{SC}}\right)}$$

(8)

It is important to highlight that the 3 parameters of the model can be computed solely based on datasheet open information. The short circuit current does not need to be determined, because it is obtained directly from the datasheet.

For a given irradiance and module temperature, the electrical DC output power, $P$, is as follows:

$$P=VI=V\left[I_{sc}-I_0\left(e^{\frac{V}{m{V}_T}}-1\right)\right]$$

(9)

The maximum power is obtained when $dP/dV=0$, which allows to obtain the maximum power voltage, $V_{MP},$ as follows:

$$V_{MP}=m{V}_T\ln \left(\frac{\frac{I_{sc}}{I_0}+1}{\frac{V_{MP}}{m{V}_T}+1}\right)$$

(10)

The maximum power current is obtained from Equation (6) as $I_{MP}=I\left(V_{MP}\right)$ and the maximum power is $P_{MP}=V_{MP}{I}_{MP}$. As Equation (10) is a non-linear equation, its solution requires iterative methods, e.g., the Gauss method.

3.4. Bifacial Modules

Bifacial photovoltaic technology consists of PV modules that convert light to electricity, both on the traditional front side and also on the reverse side of the modules. The main distinguishing feature of bifacial modules is that they take advantage of the radiation reflected on the ground and other adjacent modules, and also the diffuse radiation that originates from separation processes in the atmosphere and after being reflected. As a result, more energy is produced per area unit.

Regarding the use of bifacial systems as an optimization technique, and to correctly compare bifacial systems with their monofacial counterparts in terms of associated economic impacts, a model proposed by the International Energy Agency (IEA) [30] was used to assess economic variations. This approach provided a comparison between two simulations, one with bifacial modules and one with monofacial modules, with identical properties. The method suggests that the economic impacts of a bifacial system deployment should be adapted to express these changes in units of +Δ or −Δ, where Δ is roughly equivalent to the percentage of the bifacial gain of the system. This approach was used to calculate new CAPEX and OPEX values to be incorporated into the economic model, following two methods proposed by the IEA:

• Method A: Keeping the number of modules in the bifacial system as it was on the base case with monofacial modules, the BoS components were fitted to hold the increased current and yield. Associated economic impacts were reflected in cabling, inverter, and transformer costs, which vary with a percentage Δ, roughly equivalent to the percentage of system bifacial gain, defined as follows [31]:

  $$B{G}_{sys}=\frac{E_{bif}-E_{monof}}{E_{monof}}$$

  (11)

  where $B{G}_{sys}$ is the system bifacial gain, $E_{bif}$ is the energy yield simulated or measured through the bifacial solution, and $E_{monof}$ is the energy yield simulated or measured through the monofacial solution. The methodology suggested by IEA was therefore applied to inverter price, electrical supply, and installation and grid connection costs, all being components of the unitary PV costs in EUR/kWp.
• Method B: Reducing the number of bifacial modules to keep the same annual yield as produced by the monofacial system, associated economic impacts were translated via the reduced installed capacity multiplied by the original unitary PV cost and then reflected as total CAPEX.

3.5. Summary of the Methodology

As mentioned before, the base case was an existing solar PV park, fixed-tilt, DC–AC ratio (the ratio of the DC peak power to the inverter AC power) equal to 1.24, 25 modules per string, and equipped with monofacial modules. Each of these four parameters was changed and the impact on the NPV and IRR was assessed. The fixed tilt system was compared with horizontal single-axis tracking (HSAT) and dual-axis tracking (DAT), then, the DC–AC ratio was varied from 1.0 to 1.5, 25 modules per string were assessed against 28 modules, and finally, standard monofacial modules were compared with their new bifacial counterparts.

A summary of the methodology is offered in Figure 1.

4. Results and Discussion

4.1. Fixed vs. Tracking Systems

Constituting by far the most adopted group of trackers [32], single-axis trackers represent the current trend being implemented industry-wide. In the single-axis trackers category, HSAT systems have been the most implemented. Therefore, an HSAT system was simulated and compared with fixed-tilt and DAT structures.

Simulations conducted on PVSyst showed the impact of trackers on the energy yield of the considered PV project. Figure 2 summarizes the results, allowing comparison of the energy yield with the different tracking technologies proposed.

These results suggest that an HSAT system would generate 9.06% more energy than the fixed-tilt system. A dual-axis system would be the most beneficial approach from an energy yield perspective, with an increase of almost 40% compared with the base-case scenario.

The critical item to be examined in addition to energy yield is the impact of technology on both CAPEX and OPEX values. For the fixed-tilt scenario, CAPEX values were obtained through multiplying installed capacity in kWp by the total unitary PV cost in EUR/kWp (see Table 1). For the other two scenarios, the reviewed literature data [14,15,28,33,34,35] showed an average necessary additional initial investment of EUR 360.33/kWp associated with the deployment of single-axis tracking systems, and of EUR 1015.42/kWp for dual-axis tracking systems. For each system, the additional initial investment was multiplied by the total amount of installed capacity in kWp and the result was added to the original fixed-tilt CAPEX costs to ontain the new CAPEX value including single-axis and dual-axis installation premiums.

Regarding OPEX costs, a similar approach was used; the total O&M cost in terms of EUR per installed kWp per year was reviewed and compared with the base case. Given the recent progress and cost decrease in O&M associated with greater adoption of tracking systems [36], and to fully understand their economic potential, best-case scenarios of EUR 17.85/kWp/year for a single-axis system and EUR 29.75/kWp/year for dual-axis were considered as inputs in the economic model.

Higher values of O&M in single-axis systems are associated with a more frequent need for servicing moving parts, alignment, and calibration, among other periodical activities, an effect that has even greater impact on dual-axis trackers. A similar procedure as for CAPEX was followed for OPEX, multiplying installed capacity by total unitary O&M costs to find the new annual OPEX. The results of the economic model are summarized in

Table 3. IRR and NPV for fixed-tilt, horizontal single-axis tracking (HSAT) and dual-axis tracking (DAT) technologies. Simulation conditions: DC ratio = 1.24, monofacial modules, 25 modules per string. Legend: BC = base case.

	Fixed-Tilt (BC)	HSAT	DAT
IRR [%]	7.56%	4.03%	0.90%
NPV [EUR M]	4.64	−8.83	−34.32

.

Conclusions indicate that the increased amount of energy production is not enough to cover the higher installation and operation costs. The additional gain from an HSAT system appears to be ineffective. The situation is even more significant with dual-axis systems. In the case of single-axis systems, given the adoption rates previously mentioned, the conclusions obtained may be questionable since the industry trend is to trust single-axis tracking as the option to follow when terrain conditions allow.

4.2. DC–AC Ratio

DC–AC ratio is the quotient between the DC power installed in PV modules, measured under STCs, and the AC power installed in inverters, the power injected in the grid being limited by the inverters’ AC capacity. In the design of PV parks, the DC–AC ratio is taken as higher than 1, to take advantage of STC hardly occurring, not to say never. This supports the possibility of optimization, consisting of exploring the oversizing of the system (DC–AC ratio higher than 1) to maximize solar PV production.

To measure the advantages of oversizing, distinct DC–AC ratio scenarios were considered and compared with the 1.24 original DC–AC ratio of the base case. Following the technical assessment, these results were included in the economic model, with energy yield values being directly exported from PVSyst. A new optimized scenario considered adding or removing new modules to already installed inverters, which resulted in a correlated impact on CAPEX costs. OPEX values were calculated assuming that the base case cost (EUR 17.04/kWp/year) did not change.

Table 4. IRR and NPV for different DC–AC ratios. Simulation conditions: fixed-tilt, monofacial modules, 25 modules per string. Legend: BC = base case.

DC–AC Ratio	1.00	1.10	1.15	1.20	1.24 (BC)	1.30	1.35	1.40	1.50
IRR [%]	7.40%	7.48%	7.48%	7.54%	7.56%	7.56%	7.52%	7.46%	7.27%
NPV [EUR M]	3.38	3.96	4.22	4.46	4.64	4.83	4.92	4.89	4.58

summarizes these results.

As shown, the best configuration in terms of IRR was found to be a DC–AC ratio of 1.30, and in terms of NPV a DC/AC ratio of 1.35. These conclusions represent an increase in comparison to the industry standard of 1.20. This analysis showed that the cost decline in the PV module chain might have an impact on the optimal configuration of module-to-inverter distribution, questioning the one-size-fits-all approach defining the DC–AC ratio at around 1.20.

A report by IRENA [37] collected DC–AC ratio data from 2010 to 2020 comprising 202 GW of capacity from 6836 projects, showing that in the USA, the median DC–AC ratio grew 9% between 2010 and 2019 to reach 1.31 in 2019.

4.3. Number of Modules in Series

As mentioned in Section 3, the maximum number of PV modules connected in series depends both on the inverter’s maximum input DC voltage and on the maximum module open circuit voltage (Equation (4)). The state-of-the-art solution to compute the maximum open circuit voltage is to use Equation (3), leading to 25 modules per string, which was implemented in the real PV park followed in this study. An alternative that allowed more modules to be connected in series in a string was to use Equation (5), which, in this case, led to 28 modules per string. The simulation results obtained for the two methods under assessment, after overcoming the limitations of PVSyst regarding these simulations, are presented in

Table 5. IRR and NPV for different string lengths. Simulation conditions: fixed-tilt, DC ratio = 1.24, monofacial modules. Legend: BC = base case.

String Length	25 Modules (BC)	28 Modules
IRR [%]	6.71%	7.56%
NPV [EUR M]	1.96	4.64

.

The impact of this optimization on the IRR was an increase of 0.85%, representing an important variation in a project of this size and reiterating the winning perspective of the optimization approach. The industry-standard calculation is not only unnecessarily conservative in terms of project safety but also hides the possibility of greater economic benefit.

4.4. Bifacial Modules

The two methods proposed by the IEA were applied to assess the economics of bifacial PV modules.

The annual IRENA Renewable Power Generation Cost report from 2019 stated that bifacial module costs were 56% higher than those of monofacial modules that year. The same report from 2020 mentioned that bifacial crystalline modules sold for 21% higher than high-efficiency monofacial modules in December 2019. It also added that this cost premium fell to 6% during December 2020. The report from 2021 pointed to a cost premium of 5% in December 2021. Despite these reports pointing to lower cost premiums of bifacial modules relative to their monofacial counterparts, in this paper, a price premium of 10% for bifacial modules, a fixed-tilt system, and an albedo of 0.2 were considered, to be on the conservative side. An albedo of 0.2 is typically considered for Portugal. It is recalled that the albedo is the percentage of radiation that reaches the ground and is again reflected into the atmosphere; it is heavily dependent on typology and ground cover.

Table 6. IRR and NPV for monofacial and bifacial modules. Simulation conditions: albedo = 0.2, fixed-tilt, bifacial overcost = 10%, DC ratio = 1.24, 25 modules per string.

	Monofacial (BC)	Bifacial Method A	Bifacial Method B
IRR [%]	7.56%	7.50%	7.66%
NPV [EUR M]	4.64	4.68	4.88

depicts the results obtained with bifacial modules when compared to monofacials.

The results for the fixed-tilt system reiterated the importance of considering a range of values for the bifacial module premium. The system with bifacial modules achieved slightly better economic indexes compared with the base case when the additional price paid for bifacial modules was 10% higher. This happened for both Method A and Method B, which indicated similar conclusions, and contributed to the robustness of the model within the two different approaches. This conclusion makes clear the fact that an effective adoption of a bifacial system is highly dependent on the type of procurement deal that the company closes with the manufacturers. With bifacial module costs in recent years continuing to drop to scenarios closer to a null bifacial premium, this technology might therefore continue to see greater rates of adoption, with a cost breakdown that can justify the additional investment.

5. Link to the Portuguese Case

In Portugal, 18 utility-scale projects were subject to the environmental impact assessment (EIA) procedure by the Portuguese Environment Agency during the first half of 2021. The procedure was mandatory for projects with installed power greater than 50 MWp (or 100 hectares of occupied land in the revised version of the legislation). Among the projects considered were more than 15 different project developers, a diversity that met the strong demand and that allowed a better understanding of the identity of the main players operating in Portugal. The majority were large multinational groups that obtained licenses through auction procedures launched by the government in 2019 and 2020. The analyzed projects represent a total of 4.1 GW of installed peak power.

Horizontal single-axis trackers (HSATs) are a common choice in Portuguese PV projects, usually combined with bifacial modules. One-third of the projects use fixed-tilt technology, while all the others opt for an HSAT strategy. The use of trackers is heavily dependent on the slopes of the terrain where the project is located. A flat terrain is more suitable for installing trackers. In terrains with uneven topography, earthmoving may be a solution, but this type of terrain change would greatly increase the project costs and may prevent the use of such technology. Results obtained in this work indicated that HSAT adoption is not cost-effective, but it is important to understand that the economic modelling considered a specific discount rate and energy price, which may vary among the projects addressed in the table. In all the studied cases of projects that used bifacial modules, these were installed jointly with HSATs, except one case.

Regarding the ratio between installed peak power and grid injection power, the values recorded oscillated between 1.11 and 1.41. The average value of 1.24 was close to the optimal range calculated with the economic model in this work and it was exactly the base-case value. The reasons for such a wide range of values can vary. At the lower limit, with a ratio of just 1.11, the cause may be related to the lack of usable area that prevents the placement of a greater number of PV modules for a given licensed power of injection into the grid. As an example, if the terrain topography does not technically allow the installation of structures in a certain area, the maximum value for the installed peak power would therefore be limited. At the upper limit, the involved variables would need to be further analyzed to understand the reasons behind such an oversizing, but low irradiance locations could be one of the factors that impact this type of configuration. A higher DC–AC ratio can help mitigate this site problem.

Of the 10 projects for which environmental impact studies have included published information allowing calculation of the open circuit voltage, it was found that only two proposed a string length with more modules than the conservative (minimum temperature of −10 °C) scenario. A thorough analysis considering meteorological studies for each site location would have to be conducted to acquire values for temperature and irradiance and consequently calculate new thresholds for the number of modules in series. The reason for this lack of optimization may be related to outdated industry practices, but the number of modules in series might also have been calculated exclusively using automatic calculation software such as PVSyst, which assumed the conservative scenario as well.

Regarding the use of bifacial modules, although only a few projects have already been built and are in operation, the adhesion to this recent technology is surprisingly high, with 50% of the projects incorporating this technology in layouts developed for licensing purposes. As seen in this work, it is now possible to obtain better economic performance using this type of technology and some sources have already mentioned increases in costs of 10% or less, so project developers should aim for the objective of closing procurement deals with manufacturers that allow them to meet lower price premiums. Contracts signed for several projects simultaneously might be an important strategy to capitalize on scale economies and influence the cost of technologies such as bifacial modules.

Table 7. Project characteristics of Portuguese utility-scale PV projects in 2021. Legend: FT—fixed-tilt.

Project	Pp (MWp)	Mod. Pp (Wp)	Tracking	DC/AC Ratio	String Size	Modules
A	282	510	HSAT	1.26	26	Bifacial
B	126.5	530	FT (20°)	1.15	27	Bifacial
C	126.4	450	FT (20°)	1.26	n.a.	Monofacial
D	36.53	400	FT (18°)	1.22	n.a.	Monofacial
E	200	440	HSAT	1.41	n.a.	Bifacial
F	265	500	FT (20°)+HSAT	1.20	26	Monofacial
G	63.5	550	HSAT	1.27	36	Bifacial
H	128	450	HSAT	1.28	27	Bifacial
I	98	460	HSAT	1.17	26	Monofacial
J	284	445	HSAT	1.32	27	Bifacial
K	1008.50	440/530	FT (15°)	1.11	27	Monofacial
L	240.7	435	FT (n.a.)	1.17	n.a.	n.a.
M	144	525	HSAT	1.20	n.a.	Bifacial
N	100	405	HSAT	1.20	n.a.	Bifacial
O	150	405	HSAT	1.20	n.a.	Bifacial
P	63.5	410	HSAT	1.30	n.a.	n.a.
Q	189	330	HSAT	1.40	21	Monofacial
R	558	440	HSAT	1.16	6	Monofacial

n/a = not available.

summarizes the main characteristics of Portuguese utility-scale PV projects in 2021. The table shows the total peak power, the module peak power, the type of tracking system, the DC–AC ratio, the string size, and the type of modules.

6. Conclusions

The conducted work sustains the thesis that after a technical analysis, the associated economic impact must also be addressed to select the most viable option. Alternatives that increase energy yield are not always translated into greater economic benefits and a failure to incorporate this component might endanger project viability. This is especially relevant to the four subjects covered in this work: the use of tracking mechanisms, DC–AC ratio definition, string length sizing, and the use of bifacial modules.

The strongly adopted configuration of including horizontal single-axis trackers was shown to underperform in terms of economic behavior compared with the original fixed-tilt base case. HSAT represented a 9.06% increase in energy yield, but the trade-off meant a reduction in the IRR and NPV from 7.56% to 4.03% and from EURO 4.64 M to EUR −8.83 M, respectively. Dual-axis technology usage was shown to be unfeasible, with an IRR of 0.9% and an NPV of EUR −34.32 M, justifying the fact that this technology is not being deployed for utility-scale projects in Portugal.

In terms of DC–AC ratio, simulations showed that energy yield was maximized with increasing DC array oversizing, but the optimum point in terms of IRR and NPV was found at ratio values of 1.30 and 1.35, respectively, with subsequent drops in these indicators for further increased values of the DC–AC ratio. The studied system configurations represented a surge in NPV when compared with the base case, improving the NPV from EUR 4.64 M to EUR 4.92 M, while the calculated IRR was similar.

String length extension was also proven to be a more effective way of using the available resources to harvest additional energy without significant additional economic effort. Overriding outdated conservative project methods translated into an 11.80% energy yield increase, with the corresponding reflection in the economic model, increasing IRR and NPV from 6.71% to 7.56%, and from EUR 1.96 M to EUR 4.64 M, respectively.

Regarding bifacial modules, simulations were conducted for the fixed-tilt configuration. A premium of 10% in cost compared with their monofacial counterparts resulted in an average IRR of 7.58% and an average NPV of EUR 4.78 M. All the considered results reflected the available room for improvement in relation to the overall quality of deployed projects.

The 30-year-long nature of projects and the distinct interests among project developers and long-term asset managers might sometimes be the root cause of a project’s under-optimization. The complexity involved in the licensing and development stage often involves tight deadlines to submit project documentation, layouts, and technology specifications to several entities, meaning companies end up opting for standardized approaches and limited evaluation of pathways to economic enhancement. Although this might be a safer approach to guarantee licensing of deliverables, it means that less time and fewer resources are invested in research, simulation, and optimization, which may jeopardize further gains along the project’s lifetime.

It is concluded that the integration of technical and economic components in solar PV park development adds significant value through providing a holistic understanding of projects’ feasibility, performance, and profitability. This integrated approach allows informed decision making throughout the project lifecycle, from improved initial design and configuration to long-term operation and maintenance. Ultimately, through considering both technical and economic aspects together, developers can maximize the overall value of solar PV park investments while ensuring sustainable and profitable operation over time.

One aspect that is gaining more relevance is the adoption of large-format modules, which are beginning to approach peak powers of around 700 Wp. This technical solution may be relevant in projects where the terrain limitations are significant and require a greater allocation of power per area, something that later ends up affecting the entire arrangement and configuration of strings and inverters. To compare the performance of these modules with the state-of-the-art ones, both from a technical and economic perspective, will be necessary in future work. Another aspect that is worth more investigation is the economic performance of integrating wind turbines and PV capacity in a single location, the so-called hybridization of wind and solar parks, to explore energy transmission infrastructure synergies.