A Metaheuristic Framework for Cost-Effective Renewable Energy Planning: Integrating Green Bonds and Fiscal Incentives

Abstract

The integration of non-conventional renewable energy sources (NCRES) plays a critical role in achieving sustainable and decentralized power systems. However, accurately assessing the economic feasibility of NCRES projects requires methodologies that account for policy-driven incentives and financing mechanisms. To support the shift towards NCRES, evaluating their financial viability while considering public policies and funding options is important. This study presents an improved version of the Levelized Cost of Electricity (LCOE) that includes government incentives such as tax credits, accelerated depreciation, and green bonds. We apply a flexible investment model that helps to find the most cost-effective financing strategies for different renewable technologies. To do this, we use three optimization techniques to identify solutions that lower electricity generation costs: Teaching Learning, Harmony Search, and the Shuffled Frog Leaping Algorithm. The model is tested in a case study in Colombia covering battery storage, large- and small-scale solar power, and wind energy. Results show that combining smart financing with policy support can significantly lower electricity costs, especially for technologies with high upfront investments. We also explore how changes in interest rates affect the results. This framework can help policymakers and investors design more affordable and financially sound renewable energy projects.

1. Introduction

This section begins by establishing the motivation for the research. It then develops the theoretical background necessary for clear reader comprehension and provides a review of the relevant literature. Finally, the section briefly introduces the main contributions of this project before outlining the organization of the remainder of the paper.

1.1. Motivation

The integration of NCRES is essential to the global energy transition toward a more sustainable, decentralized, and resilient power system. As global energy demand continues to grow, reducing dependency on fossil fuels is critical to mitigating climate change, improving energy security, and advancing environmental justice. Technologies such as solar, wind, biomass, and geothermal offer viable alternatives by providing clean, locally available, and low-emission energy solutions. Their widespread adoption not only lowers greenhouse gas emissions but also stimulates technological innovation and fosters inclusive economic development [1].

Small-scale NCRES-based generation projects have been increasingly deployed worldwide to address energy access gaps and regional energy security [2]. These initiatives have proven effective in reducing carbon footprints, diversifying the energy mix, and enhancing resilience in vulnerable communities. By decentralizing electricity production and utilizing local resources, these systems facilitate rural electrification and improve quality of life. Moreover, involving local stakeholders in their development fosters participatory governance aligned with the principles of a just energy transition [3,4].

The feasibility of implementing NCRES projects requires a comprehensive evaluation of socio-environmental benefits, financial risks [5,6,7], and cost structures. These are influenced not only by technical and labor factors, but also by fiscal mechanisms that promote equity, competitiveness, and sustainability [8,9,10].

1.2. Theoretical Background and Literature Review

To encourage the adoption of NCRES, various policies have been promoted globally. These policies include financial subsidies and incentives such as tax deductions [11,12,13], legal and regulatory frameworks [14], quotas, targets, and market mechanisms that address NCRES intermittency and establish transparent pricing [15], and government support and planning [16,17]. Despite these supportive policies, several challenges persist. These include the need for targeted policies that specifically promote promising technologies such as wind and biomass [14], the lack of clear and consistent pricing mechanisms, and the difficulties investors face in navigating the complexity of existing incentives [13].

Conversely, NCRES technologies face several inherent characteristics that hinder their competitiveness against conventional power generation. First, their upfront costs are often higher, contributing significantly to the overall lifetime expenses of a project [18,19]. Second, the intermittent and non-dispatchable nature of many NCRES can lead to fluctuations in generation costs due to mismatches between energy demand and supply [20]. Furthermore, the integration of NCRES involves additional costs associated with transmission grid expansions [15] and the necessity of ancillary services and reserves to compensate for their intermittency [21]. Moreover, emerging renewable energy technologies face significant uncertainties regarding both technical and economic performance due to their early stage of development [11,19]. This is compounded by a lack of adequate support from manufacturers and vendors as well as insufficient market competition [19].

NCRE technologies can be considered an alternative to meet the energy demands of disadvantaged communities in remote regions, which is directly related to the concept of energy justice. Energy justice offers a crucial perspective for evaluating the equitable distribution of energy resources and services, aiming to guarantee affordable, reliable, and clean energy access for all individuals irrespective of their geography, income, or social status [22]. This concept and framework fundamentally pursue fairness, equity, and inclusivity within energy systems, functioning both as an analytical tool to pinpoint injustices and a normative guide for shaping policy and practice [23,24]. This requires actively integrating distributional, procedural, and recognition justice principles into policy design and implementation at the local, national, and international levels [25]. The cost of energy generation is central to the energy justice discussion, profoundly impacting affordability and accessibility. Elevated generation costs can worsen energy poverty, disproportionately burdening low-income and marginalized communities. This directly relates to the sustainable development goals (SDGs), especially SDG7, as energy justice offers the ethical and analytical tools to ensure that energy development and the shift towards sustainable systems not only achieve energy and climate goals but are also inherently fair, equitable, inclusive, and beneficial for everyone. Furthermore, it addresses the root causes of energy poverty and marginalization [26,27,28]. Additionally, energy justice also intersects with other SDGs, including poverty reduction (SDG 1), gender equality (SDG 5), and climate action (SDG 13).

The economic viability of NCRES is widely recognized as the primary obstacle to achieving a world powered by them. For renewable sources to attract investment and become a viable substitute for conventional power plants, they must be economically competitive and carry risks that are well-understood and estimable [11,13]. Consequently, robust economic assessment tools are essential to evaluating the comprehensive impact of power plants on the power system [17,20,21].

Focusing on this crucial last point, technical-economic evaluation strategies such as the levelized cost of electricity (LCOE) can significantly aid in decision-making. The LCOE is widely used to assess the economic viability of power generation projects. It expresses the constant price of electricity (USD/kWh) required to achieve a net present value of zero throughout the lifetime of a project. However, LCOE calculations often exclude externalities and regulatory incentives such as income tax deductions, grace periods, and accelerated depreciation, all of which can significantly affect project costs [10,29]. Financial tools such as discounted cash flow (DCF) and real options (RO) have been proposed to complement LCOE by incorporating uncertainty and qualitative elements [12,30,31].

Essentially, the LCOE represents the stable price per unit of electricity needed to offset all costs incurred throughout a project’s economic lifespan. This price ensures a net present value (NPV) of zero and achieves a minimum acceptable rate of return [12,13]. Consequently, the LCOE enables direct comparison of different electricity generation technologies, even those with varying investment scales or operational durations. Although an abstraction of real-world complexities, the LCOE metric serves as a valuable benchmarking and ranking tool. It aims to eliminate bias when comparing different technologies by considering their total lifetime costs relative to the energy that they generate. Key components typically included in the LCOE calculation [11,19] are initial investment or overnight construction costs, fixed and variable operation and maintenance costs, fuel costs (relevant for fossil fuels, nuclear, and biomass), the economic lifespan of the plant, the capacity factor or capacity availability factor, and the discount rate. It is important to note that the LCOE is sensitive to macroeconomic indicators, price fluctuations, and risk assessments [17].

Despite its widespread adoption, the LCOE is not a flawless metric. It represents a static assessment, relying on assumptions and “best estimates” for its input values. Consequently, this single value fails to capture the inherent uncertainty associated with a project’s financial outcomes. Moreover, a significant drawback of the LCOE is its failure to account for variations in electricity generation timing or the dispatchability of different technologies. This makes direct comparisons between intermittent renewable sources and fully dispatchable options particularly challenging [20]. Additionally, the LCOE overlooks grid interaction complexities and lacks the granularity to reflect dynamic market prices and the devaluation of generation in grids with a high penetration of variable renewable energy [32].

Green bonds are debt instruments intended to finance projects that deliver environmental and climate-related benefits, with a strong focus on initiatives in the energy sector [33,34]. They allow investors to support green initiatives while earning a return, and provide issuers with dedicated capital for sustainable energy projects [35,36]. Green bonds can lower costs by matching long-term clean energy needs with capital, thereby reducing search friction for green investors and enabling pricing advantages where buyers accept lower returns for environmental responsibility. They play a key role in the direction of funds to grow clean energy and reduce environmental risks [37]. The effects of green bonds on the viability of implementing renewable energy solutions have been extensively studied, covering their sources of financing, financial market interactions, and growth drivers [38,39]. Other studies have suggested that green bonds can help to bridge financing gaps in immature capital markets by improving the diversity of financial portfolios [40]. In addition, green bonds can help to mitigate private investments, which face high capital costs due to policy, market, and technology uncertainties, which are more pronounced in developing countries and emerging markets due to weak institutions and immature clean energy markets [41]. In the long run, this financial aid can also alleviate the burden on innovators and producers facing initial less competitive and costly implementations. This is particularly the case in NCRES, for example, bioenergy [42,43] and geothermal production [44]. Finally, in addition to renewable energy, green bonds can fund other impactful sustainability and development sectors, such as blue economy and water management projects [45].

In addition to their environmental benefits, green bonds can act as diversifiers, protect investments, and offer a safe haven against global geopolitical and economic uncertainties. However, the intricacies of their interconnectedness and behavior under varying market conditions and time frames need further research [46,47].

1.3. Contributions and Content

This study proposes a novel methodology to improve LCOE-based analysis by integrating fiscal incentives and green bonds into the investment evaluation of NCRES projects. In this context, green bonds act as both financing mechanisms and instruments for recognizing environmental and social co-benefits, particularly in underserved areas.

The main contribution of this work lies in the integration of tax credits, accelerated depreciation, and green bonds into an LCOE optimization framework adapted to the Colombian regulatory context. The proposed approach is solved using three metaheuristics—TL, HS, and SFLA—which identify cost-optimal investment structures for each technology. This combination provides flexible decision-making tools for stakeholders involved in clean energy planning.

The rest of this paper is organized as follows: Section 2 presents the mathematical formulation and optimization methods; Section 3 describes the case study and key results, including a sensitivity analysis; finally, Section 4 provides concluding remarks and outlines future research directions.

2. Methodology

This section outlines the methodological framework designed to assess the LCOE in power generation projects utilizing NCRES. The proposed approach integrates economic and fiscal incentives such as investment tax credits (ITC), accelerated depreciation schemes, and green bonds as key components in the financing structure. A generalized mathematical model of the LCOE is developed which accounts for capital investment, fixed and variable operation and maintenance (O&M) costs, fuel expenses, reliability charges, and both positive and negative externalities. The model also incorporates multiple financing sources—including equity, debt, and green bonds—under a set of decision variables and constraints that reflect regulatory, technical, and financial conditions appropriate to each country’s context. This framework enables the use of metaheuristic techniques to identify optimal investment strategies that minimize LCOE across diverse technologies and scenarios.

2.1. Problem Formulation

The proposed mathematical formulation aims to minimize LCOE by optimizing the investment structure of NCRES-based generation projects. The model captures key cost components such as initial investment, annual O&M expenses, fuel costs, externalities, and income from reliability charges while integrating fiscal policies such as ITC and accelerated depreciation. Green bonds are considered both as a financing mechanism and as contributors to positive externalities. Constraints are imposed to reflect realistic bounds on decision variables, such as capital allocation proportions, debt terms, depreciation periods, and incentive utilization limits. The complete set of equations describing the LCOE components is presented in Equations (1)–(5) following the formulation established in previous works [8,9], and serves as the basis for the metaheuristic optimization process developed in this study:

$$LCOE=LCO{E}_I+LCO{E}_V+LCO{E}_F+LCO{E}_{FL}\pm LCO{E}_E$$

(1)

where:

• $LCO{E}_I$: Investment component (USD/kWh)
• $LCO{E}_V$: Variable O&M cost component (USD/kWh)
• $LCO{E}_F$: Fixed O&M cost component (USD/kWh)
• $LCO{E}_{FL}$: Fuel cost component (USD/kWh)
• $LCO{E}_E$: Externality component (USD/kWh)

$$LCOE={\displaystyle \frac{I_0+{\sum }_{t=1}^n\frac{C_t}{{(1+i)}^t}}{{\sum }_{t=1}^n\frac{E_t}{{(1+i)}^t}}}$$

(2)

where:

• $I_0$: Initial investment cost (USD)
• $C_t$: Annual operating costs (USD)
• $E_t$: Energy produced in one year (kWh)
• i: Discount rate (annual effective)
• n: Operational lifetime (years)
• t: Useful life of the project (years).

2.2. Objective Function and Constraints

Equation (3) presents the enhanced formulation of the LCOE derived from Equation (2) by incorporating realistic financial structures that include multiple sources of capital, namely, equity, debt, and green bonds, as well as fiscal incentives such as ITC and accelerated depreciation. This reformulated objective function (FO) is minimized using metaheuristic optimization techniques.

$${\displaystyle \mathrm{FO}=\mathrm{LCOE}=\frac{1}{\mu {\sum }_{t=1}^n\frac{E_t}{{(1+i)}^t}}·\left[{\alpha }_{1}I+\sum _{t=k}^{l+k}\frac{{\alpha }_{2}I{(1+g)}^k+\mu i_{D}L}{L{(1+E)}^t{(1+i)}^t}+\sum _{t=1}^n\frac{\mu C_t}{{(1+i)}^t}+\sum _{t=1}^d\frac{D_{j}I(\mu -1)}{{(1+E)}^t{(1+i)}^t}+\frac{I{\alpha }_3+\mu Y_{GB}}{{(1+E)}^r{(1+i)}^r}\pm \sum _{t=1}^n\frac{\mu E_x}{{(1+i)}^t}\right]}$$

(3)

$$\mu =1-\left(1-\sum _{t=1}^p{I}_t\right)\beta$$

(4)

$$Y_{GB}=I{\alpha }_3{(1+i_{GB})}^r-I{\alpha }_3$$

(5)

In the above equations, the denominator represents the discounted sum of annual energy production $E_t$ over the project’s operational lifetime n adjusted by the fiscal benefit factor $\mu$ derived in Equation (4). The numerator includes all the cost and benefit components related to investment, financing, operation, fiscal deductions, and externalities. In this sense, the LCOE can be understood as the total cost of generation (CG) associated with one specific option among the possible alternatives.

• Explanation of Terms in the Objective Function:

• ${\alpha }_{1}I$: Portion of the total investment I financed with equity capital, where ${\alpha }_1$ is the equity share.
• ${\sum }_{t=k}^{l+k}{\displaystyle \frac{{\alpha }_{2}I{(1+g)}^k+\mu i_{D}L}{L{(1+E)}^t{(1+i)}^t}}$: Represents the cost of debt financing, where ${\alpha }_2$ is the debt share, g is the debt interest rate, k is the grace period, L is the loan term, $i_D$ is the annual debt payment, and E and i respectively represent the inflation and discount rates.
• ${\sum }_{t=1}^n{\displaystyle \frac{\mu C_t}{{(1+i)}^t}}$: Present value of operational costs $C_t$, including fixed and variable O&M, administrative expenses, fuel costs, reliability charges, and income from externalities. Adjusted by the fiscal benefit factor $\mu$.
• ${\sum }_{t=1}^d{\displaystyle \frac{D_{j}I(\mu -1)}{{(1+E)}^t{(1+i)}^t}}$: Reflects the benefits of accelerated depreciation, where $D_j$ is the depreciation rate, d is the number of depreciation years, and $(\mu -1)$ represents tax reductions due to depreciation.
• ${\displaystyle \frac{I{\alpha }_3+\mu Y_{GB}}{{(1+E)}^r{(1+i)}^r}}$: Cost and benefit associated with the issuance of green bond, where ${\alpha }_3$ is the green bond share, $Y_{GB}$ is the bond yield defined in Equation (5), and r is the maturity period.
• $\pm {\sum }_{t=1}^n{\displaystyle \frac{\mu E_x}{{(1+i)}^t}}$: Present value of externalities $E_x$, which may be positive (e.g., social or environmental benefits) or negative (e.g., emissions), adjusted by $\mu$.

• Subject to the following constraints:

$$\begin{array}{cc}\hfill \sum _{t=1}^p{I}_t& \le {\lambda }_i\hfill \end{array}$$

(6)

$$\begin{array}{cc}\hfill 0\le p\le & \theta \hfill \end{array}$$

(7)

$$\begin{array}{cc}\hfill 0\%\le {\alpha }_1& \le 100\%\hfill \end{array}$$

(8)

$$\begin{array}{cc}\hfill 0\%\le {\alpha }_2& \le 100\%\hfill \end{array}$$

(9)

$$\begin{array}{cc}\hfill 0\%\le {\alpha }_3& \le 100\%\hfill \end{array}$$

(10)

$$\begin{array}{cc}\hfill {\alpha }_1+{\alpha }_2+{\alpha }_3=& 100\%\hfill \end{array}$$

(11)

$$\begin{array}{cc}\hfill K,L,r\ge & 0\hfill \end{array}$$

(12)

$$\begin{array}{cc}\hfill r& \le \delta \hfill \end{array}$$

(13)

$$\begin{array}{cc}\hfill {\sigma }_i\le d& \le {\sigma }_f\hfill \end{array}$$

(14)

$$\begin{array}{cc}\hfill D_j& \le \varphi \hfill \end{array}$$

(15)

where:

• $I_t$: Annual Investment Tax Credit rate; p: ITC application period; ${\lambda }_i$: maximum rate of the ITC; $\theta$: validity period of the ITC.
• $\delta$: Maturity period of the green bond
• ${\sigma }_i$: Number of years before asset depreciation initiation
• ${\sigma }_f$: Number of years before asset depreciation completion
• $\varphi$: Maximum depreciation rate per year.

The decision variables in this model are ${\alpha }_1$, ${\alpha }_2$, ${\alpha }_3$, d, k, L, and r.

2.3. Roadmap and Implementation in Other Markets

Figure 1 depicts a policy implementation roadmap. It begins by identifying the unconventional renewable energy sources of interest. Next, the existing fiscal and economic policies in each country are reviewed, including those related to green bonds. Following this, the investment, management, and operational costs of generation technologies are calculated, along with key financial parameters such as the discount rate, inflation, and tax rate, among others.

This information is then input into the metaheuristics, which include the LCOE model, to determine the optimal combination of decision variables (${\alpha }_1$, ${\alpha }_2$, ${\alpha }_3$, d, k, L, r) that minimizes the LCOE. If the resulting combination of variables representing the capital structure of each unconventional renewable energy source does not meet the investor’s financial conditions, a new simulation can be run to generate an alternative combination of decision variables. Finally, a sensitivity analysis is conducted to allow the investor to assess the risk and make decisions based on both this analysis and the energy and market policies.

Although the LCOE model was developed using Colombia’s legislative and energy context as a reference, its structure is inherently general and highly adaptable. LCOE is a globally recognized methodology rooted in universal economic principles; it compares the total lifecycle costs of a power generation technology with the amount of energy it produces. Therefore, the validity of the model is not country-specific, depending instead on the relevance and accuracy of the input data.

The proposed model can be applied in any country as long as key parameters are adjusted to reflect local conditions. These include the discount rate, inflation, tax rate, capital and operating costs, administrative expenses, and availability of policy instruments such as green bonds or other financial incentives.

Furthermore, when the model incorporates optimization techniques such as metaheuristics, it can determine the most efficient investment configuration tailored to a country’s specific regulatory, economic, and energy landscape. This flexibility makes the model both versatile and replicable. In summary, provided that inputs are appropriately localized, the proposed approach is methodologically robust, scalable, and suitable for international application.

2.4. Metaheuristics Applied to LCOE Calculation

Metaheuristic techniques are effective tools for solving combinatorial, non-convex, and nonlinear optimization problems, delivering high-quality solutions within relatively short computational times. These techniques are particularly well-suited for identifying optimal or near-optimal solutions to objective functions—whether maximization or minimization—subject to complex constraints [48,49,50].

In this study, the aim is to minimize CG through a non-convex and nonlinear combinatorial optimization problem. The objective function and associated constraints are defined in Equations (3) and (5)–(12), respectively. The optimization involves a set of decision variables (${\alpha }_1$, ${\alpha }_2$, ${\alpha }_3$, d, k, l, r), and the goal is to find the optimal combination of these variables that results in the lowest possible CG.

To solve this problem, three metaheuristic algorithms are applied: teaching–learning (TL), harmony search (HS), and the shuffled frog leaping algorithm (SFLA). These algorithms were selected for their effectiveness in complex optimization tasks. TL is known for its fast convergence and balanced exploration–exploitation behavior; HS offers robustness and simplicity, which is particularly useful for handling nonlinear problems; and SFLA stands out for maintaining population diversity while delivering stable results across iterations. The selection rationale considered key performance metrics such as convergence speed, stability, and solution quality.

Each algorithm was applied independently to solve the defined combinatorial, non-convex, and nonlinear optimization problem. The performance of each algorithm was evaluated in terms of its ability to find cost-effective configurations of decision variables. The results were analyzed and compared to assess the quality, convergence behavior, and practical implications of each solution, providing valuable insights for investment planning in non-conventional renewable energy systems.

It is worth emphasizing that the primary contribution of this study is not the creation of novel optimization algorithms but rather the strategic application of well-established metaheuristic methods to a specific and impactful problem, namely, the formulation of optimal investment strategies for renewable energy systems. By leveraging these tools within the LCOE framework, this study demonstrates how fiscal and economic incentives can be effectively integrated to achieve cost-efficient configurations for NCRES.

Furthermore, in more complex or highly regulated markets the optimization problem may involve a greater number of constraints, such as varying policy frameworks, market structures, or technology-specific regulations, which may affect the convergence time of the selected algorithms. As the complexity of the problem increases, it may be necessary to fine-tune the algorithms’ parameters in order to maintain optimal performance.

2.4.1. Teaching–Learning

TL is a metaheuristic technique inspired by the natural process of education in which knowledge transfer occurs between a teacher and learners. In TL, the optimization process is divided into two phases, called the teacher phase and learner phase. The teacher phase represents the teacher’s role in improving the learners’ knowledge or skill levels, akin to guiding the population of solutions toward better positions in the search space. This is achieved by introducing the teacher’s influence, which represents the best solution in the population, and moving other solutions closer to the teacher. The learner phase, on the other hand, is based on the interaction among the learners themselves, where they enhance their knowledge by sharing information and learning from peers. This phase promotes diversity and exploration in the solution space, helping to avoid premature convergence [51,52].

One of TL’s strengths lies in its simplicity and parameter-free nature, making it an attractive option for solving complex optimization problems across various domains. It eliminates the need for additional algorithm-specific parameters, thereby reducing computational overhead and simplifying implementation. TL has been successfully applied in engineering design, scheduling problems, and multi-objective optimization, showcasing its flexibility and effectiveness.

Despite its advantages, TL is not immune to challenges; like other metaheuristic techniques, it may struggle with high-dimensional or highly constrained problems. Researchers often hybridize TL with other algorithms or incorporate problem-specific knowledge to address such limitations. As the field of optimization evolves, TL continues to inspire innovative approaches, solidifying its role in the toolbox of problem solvers seeking efficient and adaptive solutions. Through its teaching–learning mechanism, TL metaphorically demonstrates the power of education and collaboration in achieving optimization excellence. Its essence lies in learning and improving, as we strive to do in everyday life.

2.4.2. Harmony Search

HS is a metaheuristic optimization algorithm inspired by the improvisation process of musicians seeking a harmonious state. Just as musicians aim to find aesthetically pleasing harmonies by considering different note combinations, HS seeks optimal or near-optimal solutions to complex optimization problems by exploring a search space in a structured yet flexible manner. The core idea is to simulate a “harmony memory” which stores a number of good candidate solutions. At each iteration, a new solution (or “harmony”) is generated by combining components of existing solutions in the memory, with a certain probability of random variation or pitch adjustment to promote diversity and avoid local optima [53,54].

The algorithm uses three main parameters: the harmony memory considering rate (HMCR) controls how often values from memory are reused; the pitch adjustment rate (PAR) sets the probability of making small changes to selected values; and the number of improvisations defines how many iterations the algorithm performs. Together, these parameters balance the use of known good solutions with the search for new ones.

HS is particularly well-suited for solving nonlinear, multi-dimensional, and multimodal problems where traditional optimization methods may struggle due to the absence of gradients or the presence of many local optima. Its simplicity, flexibility, and ability to be hybridized with other techniques make it an attractive choice in engineering, energy systems, and economic optimization.

In the context of renewable energy planning and cost minimization, harmony search offers a powerful tool for navigating complex design spaces influenced by multiple technical, economic, and regulatory constraints. Its capacity to adapt and converge towards cost-effective solutions while incorporating external incentives makes it valuable for long-term planning and decision-making in sustainable energy systems.

2.4.3. Shuffled Frog Leaping Algorithm

The SFLA is a population-based metaheuristic inspired by both natural evolution and the memetic behavior of frogs searching for food. It combines the global exploration capabilities of genetic algorithms with local search mechanisms driven by information sharing within subgroups [55]. In the SFLA, a population of candidate solutions (referred to as “frogs”) is partitioned into several groups called “memeplexes”. Within each memeplex, frogs interact and share information, allowing poor-performing individuals to update their positions by mimicking better-performing peers. This local search process encourages convergence towards improved solutions within each subgroup.

After a number of local search iterations, the memeplexes are shuffled and the population is restructured. This step enhances diversity by allowing frogs to exchange information across the entire population, helping to avoid premature convergence to local optima. The alternation between focused local search and global mixing is a key strength of the SFLA, allowing it to effectively balance exploration and exploitation.

This algorithm is particularly advantageous in solving complex, nonlinear, and constrained optimization problems where the search space is large and rugged. The SFLA’s structure promotes parallelism and robustness, making it suitable for engineering applications such as system design, energy optimization, and resource allocation.

In the context of renewable energy systems and LCOE estimation, the SFLA provides a versatile framework for identifying configurations that minimize long-term generation costs while accounting for technical constraints and economic incentives. Its ability to escape local optima and explore multiple regions of the solution space is especially useful when optimizing systems with discrete variables, policy-driven parameters, or non-convex cost structures. As a result, the SFLA stands out as a promising approach for tackling the multi-objective and multidisciplinary challenges inherent in sustainable energy planning.

3. Testing and Results

This section presents the results obtained from the application of the three metaheuristic techniques described in the previous section to the minimization of CG within the LCOE framework. Because these metaheuristic techniques explore the solution space differently, they do not necessarily converge to the same combination of decision variables. As a result, each technique provides a distinct investment portfolio capable of minimizing CG according to different capital structures or investor preferences.

3.1. Description of the Test Case

The general methodology presented in this work ws tested using a specific case applied to the Colombian context. In 2014, the Colombian Congress approved Law 1715, which promotes investment in NCRE projects through a series of fiscal incentives. These incentives, which are incorporated into the proposed model, include the following:

• A 50% reduction in the value of the investment over a 5-year period (Investment Tax Credit, ITC).
• A value-added tax (VAT) exemption on equipment and services related to NCRE projects.
• Exemption from import duties.
• Accelerated depreciation of assets (up to 20% annually).

In 2019, Law 1955 extended the validity period of the ITC to 15 years. Subsequently, Law 2099 of 2021 increased the maximum allowable depreciation rate to 33.33%.

Based on the above, for the Colombian case, the objective function is subject to the following specific constraints:

$$\sum _{t=1}^p{I}_t\le 50\%$$

(16)

$$0\le K+L\le n$$

(17)

$$r\le 10.$$

(18)

In addition, the debt interest rate (g) is assumed to be 17.89% effective annually [56]. The operational lifetime of the project (n) is set to 20 years. The inflation rate (E) is considered as 13.12% effective annually [57].

Four generation technologies were selected in order to evaluate the proposed methodology’s ability to identify their optimal capital structures:

• Battery energy storage systems (BESS)
• Utility-scale solar (USW)
• Small-scale photovoltaics (SP)
• Wind power (WP).

Table 1. Summary of the main characteristics of the selected generation technologies.

Technology	Capacity	Annual Electricity	O&M Cost	Fuel Cost	Externality Income	Initial Investment	Lifetime
	(MW)	(GWh)	(¢ USD/kWh)	(¢ USD/kWh)	(¢ USD/kWh)	(MUSD)	(Years)
BESS	10	21.9	0.33	0	0.00	23.55	16
USW	56	441.5	6.10	0	0.77	371.9	20
SP	10	21.17	1.03	0	0.00	10.80	20
WP	10	26.3	1.70	0	0.00	21.10	20

summarizes the technical and financial characteristics of the generation technologies analyzed in this case study.

To apply the methodology, the CG of each technology is first estimated using the proposed LCOE approach. Then, each metaheuristic algorithm determines the optimal combination of decision variables (${\alpha }_1$, ${\alpha }_2$, ${\alpha }_3$, d, k, L, r) that minimizes CG.

Because metaheuristic techniques tend to converge to different solutions, the output includes three distinct combinations of decision variables for each technology, resulting in three possible CG values. If any of these solutions meet the investor’s expectations, the process ends. Otherwise, the investor may reinitiate the process to obtain new CG estimates using the LCOE model and metaheuristic algorithms. As these techniques are stochastic in nature, repeated simulations may produce new combinations of decision variables.

The investment costs ($I_o$) and operation and maintenance (AOM) costs, both fixed and variable, were updated using Equations (19) and (20) based on the producer price index (PPI) and consumer price index (CPI) published by the Banco de la República (2023a, 2023b). Additionally, the USW technology includes a positive externality related to revenue from solid waste disposal services [58]:

$$I_{o,b}=I_{o,a}\left({\displaystyle \frac{{\mathrm{PPI}}_b}{{\mathrm{PPI}}_a}}\right)$$

(19)

$$I_{o,b}=I_{o,a}\left({\displaystyle \frac{{\mathrm{CPI}}_b}{{\mathrm{CPI}}_a}}\right)$$

(20)

where:

• a: The year in which the costs were originally reported
• b: The update year (2021).

To evaluate the effect of fiscal incentives on the feasibility of the selected generation technologies, two base cases were considered:

• Calculation of CG using the LCOE method without applying any fiscal incentives
• Calculation of CG using the LCOE method considering fiscal incentives, with ${\alpha }_1=100\%$ and $d=3$ years.

3.2. Results

Table 2. Comparison of CG with and without fiscal incentives for each technology.

Technology	Without Incentives	With Incentives	Reduction	TL (CG)	Reduction (TL)	HS (CG)	Reduction (HS)	SFLA (CG)	Reduction (SFLA)
BESS	16.6	14.4	13.25%	7.1	57.23%	7.2	56.63%	7.1	57.23%
USW	16.8	15.5	7.74%	10.8	35.71%	10.3	38.69%	10.3	38.69%
SP	8.0	7.2	10.00%	4.0	50.00%	3.7	53.75%	4.0	50.00%
WP	12.7	11.4	10.24%	6.4	49.61%	6.1	51.97%	6.4	49.61%

presents a comparative analysis of the CG calculated for four renewable energy technologies with and without the application of fiscal incentives. The purpose of the table is to highlight the economic impact of fiscal policies on the feasibility of each technology and to assess the performance of the TL, HS, and SFLA metaheuristic optimization techniques. Costs are provided in ¢USD/kWh.

Table 2 is structured to compare the CG of four renewable energy technologies (BESS, USW, SP, and WP) under different fiscal and optimization scenarios. The first column lists the technologies analyzed. The second column presents the baseline CG, expressed in ¢USD/kWh, calculated using the LCOE method without applying any fiscal incentives. The third column shows the CG when fiscal incentives are included, specifically assuming ${\alpha }_1=100\%$ and a deduction period of $d=3$ years. The fourth column quantifies the percentage reduction in CG resulting from the implementation of these incentives. The next set of columns (5 and 6) shows the CG and its corresponding reduction achieved using the teaching–learning (TL) algorithm. Columns 7 and 8 report the same metrics obtained using the harmony search (HS) algorithm, while columns 9 and 10 display the CG and percentage reduction calculated through the shuffled frog leaping algorithm (SFLA). This layout allows for a comprehensive comparison of the impact of both fiscal policies and metaheuristic optimization methods on the economic viability of each technology.

For BESS, the application of fiscal incentives reduces the baseline generation cost (CG) from 16.6 to 14.4 ¢USD/kWh, representing a 13.25% decrease. When combined with metaheuristic optimization, the CG is further reduced to 7.1 ¢USD/kWh when using both the TL and SFLA algorithms, achieving total reductions exceeding 57%. In the case of USW, the impact of fiscal incentives is more modest, with a 7.74% reduction from the baseline; however, the application of HS and SFLA lowers the CG to 10.3 ¢USD/kWh, resulting in an additional reduction of nearly 39%. SP begins with a relatively low CG of 8.0 ¢USD/kWh, which is further reduced to 3.7 ¢USD/kWh through HS optimization, yielding a total cost reduction of more than 50%. WP follows a similar trend, with the TL and SFLA techniques driving the CG down to 6.4 ¢USD/kWh, corresponding to total reductions of up to 56.69%.

This table demonstrates the significant economic benefit of combining fiscal incentives with metaheuristic optimization, particularly in capital-intensive technologies such as BESS and USW.

Table 3. LCOE results and CS by technology and metaheuristic technique.

Technology	TL (LCOE)	TL (CS)	HS (LCOE)	HS (CS)	SFLA (LCOE)	SFLA (CS)
BESS	7.1	[1,1,8,9,10,10,10]	7.2	[1,4,5,9,10,10,10]	7.1	[1,1,8,9,10,10,10]
USW	10.8	[1,6,3,9,10,10,10]	10.3	[1,1,8,9,10,10,10]	10.3	[1,1,8,9,10,10,10]
SP	4.0	[1,1,8,9,10,10,10]	4.2	[1,4,5,9,10,10,10]	4.0	[1,1,8,9,10,10,10]
WP	6.4	[1,1,8,9,10,10,10]	6.7	[1,4,5,9,10,10,10]	6.4	[1,1,8,9,10,10,10]

presents the results of the LCOE and the corresponding capital structure (CS) obtained for the four renewable energy technologies using the three different metaheuristic optimization techniques.

In Table 3 the CS is represented using a vector of seven decision variables, denoted as $[{\alpha }_1,{\alpha }_2,{\alpha }_3,d,k,L,r]$, where ${\alpha }_1$, ${\alpha }_2$, ${\alpha }_3$ are the percentages of tax incentives or deductions, d is the duration of deductions, k is the loan repayment period, L is the loan term, and r is the interest rate or return expectation, as defined in Section 2.1. Each LCOE value represents the minimized generation cost achieved by each algorithm for a given technology.

The first column in Table 3 presents the four generation technologies evaluated in this study: BESS, USW, SP, and WP. Columns two and three display the LCOE in ¢USD/kWh and the corresponding optimal capital structure obtained through the TL technique. Columns four and five report the same metrics calculated using the HS algorithm, while columns six and seven present the LCOE values and optimal capital structures identified by the SFLA.

Each row corresponds to a specific technology, allowing for a comparative assessment of the cost-effectiveness and optimal financial structuring identified by each optimization algorithm. For example, for BESS, the TL and SFLA algorithms both achieve an LCOE of 7.1 ¢USD/kWh with similar capital structures, whereas the HS method results in a slightly higher LCOE of 7.2 ¢USD/kWh with a different set of decision variables.

This table highlights how different metaheuristic approaches can lead to varying configurations and costs, providing flexibility to investors based on their preferences or financial strategies.

3.3. Sensitivity Analysis

To assess the robustness of the proposed investment strategies and their sensitivity to changes in financial conditions, two sensitivity analyses were conducted. The first analysis was conducted by varying two parameters, namely, the bank interest rate and the yield rate of green bonds. These two parameters were selected due to their significant influence on the cost of capital, and consequently on the LCOE. The analysis considered a range of ±30% around the nominal values used in the previous simulations, permitting evaluation of LCOE fluctuations under diverse market conditions. By maintaining the capital structure reported in the previous section, the impact of interest rate volatility on the economic feasibility of each technology can be systematically explored. This approach provides valuable insights into the financial resilience of renewable energy projects under uncertain macroeconomic scenarios. As a result, the values of the bank interest rate and the green bond yield rate are as follows:

• Bank interest rate (%): [12.52, 14.31, 16.1, 17.89, 19.68, 21.47, 23.3].
• Green bond yield rate (%): [9.26, 10.58, 11.9, 13.23, 14.55, 15.87, 17.19].

The second sensitivity analysis was performed by swapping the green bond rate and the mortgage credit rate (bank interest rate: 13.23% effective annually; green bond yield rate: 17.89% effective annually).

3.4. Discussion and Analysis of Results

For this sensitivity analysis, the reference values are based on the lowest LCOE results presented in Table 3. Using these values as a baseline, the bank interest rate and the green bond yield are varied while keeping the capital structure fixed.

The resulting rate values span a broad range, allowing for an assessment of the impact of market condition fluctuations on the LCOE and the economic feasibility of the technologies under study.

Figure 2 illustrates the sensitivity of the LCOE for BESS to changes in two key financial parameters: the bank loan interest rate and the green bond yield rate. These variables were independently varied within a range of ±30% from their nominal values (17.89% and 13.23%, respectively) in order to capture the potential impact of market volatility on project feasibility. The bars in the figure illustrate the percentage variation in the LCOE resulting from changes in the bank interest rate and the green bond yield rate. Additionally, the LCOE value in ¢USD/kWh for each scenario is displayed next to each bar. The financial rates considered in the analysis range from 12.52% to 23.3% for the interest rate and from 9.26% to 17.19% for the green bond yield. This comprehensive range allows for the assessment of BESS performance under optimistic, baseline, and pessimistic financial conditions.

The TL, HS, and SFLA optimization strategies were used to determine the capital structure that minimizes LCOE. The legend label “BESS–TL–SFLA” corresponds to the minimum LCOE of 7.1 ¢USD/kWh obtained using both TL and SFLA algorithms, with a shared capital structure of [1,1,8,9,10,10,10]. The label “BESS–HS” indicates the result from the harmony search method, which achieved a slightly higher LCOE of 7.2 ¢USD/kWh with a different structure of [1,4,5,9,10,10,10].

The lowest LCOE of 5.0 ¢USD/kWh is observed when both the interest rate and bond yield are decreased by 30%, reaching values of 12.52% and 9.26%, respectively. This represents a 30.56% reduction from the nominal scenario, and was achieved using the HS algorithm. Conversely, the highest LCOE of 10.8 ¢USD/kWh was obtained under a +30% increase in both financial variables (23.3% and 17.19%), also using HS. The difference between the maximum and minimum LCOE (equivalent to a 116% spread) underscores the significant sensitivity of BESS economics to financial market conditions.

Importantly, the minimum LCOE achieved in this analysis (5.0 ¢USD/kWh) is comparable to the global benchmark of 10 ¢USD/kWh for storage technologies, as reported by IRENA in 2021 [59]. This confirms the potential competitiveness of BESS projects in Colombia when supported by favorable financial structures and incentive mechanisms.

Figure 3 presents the sensitivity analysis of the LCOE for USW technology in response to variations in the bank loan interest rate and the yield of green bonds. As in the BESS analysis, these financial parameters were independently varied within a range of ±30% around their nominal values (17. 89% for the bank interest rate and 13. 23% for the yield of green bonds) in order to simulate a spectrum of real-world market conditions. Figure 3 displays bars representing the percentage change in LCOE due to variations in the bank interest rate and green bond yield. The corresponding LCOE value in ¢USD/kWh is also shown alongside each bar. The analyzed financial scenarios span interest rates from 12.52% to 23.3% and bond yields from 9.26% to 17.19%, allowing for a comprehensive assessment of how financing dynamics affect the viability of USW projects.

The legend label “USW–TL” identifies the result obtained with the TL algorithm, which yielded an LCOE of 10.8 ¢USD/kWh under a capital structure of [1,6,3,9,10,10,10]. The label “USW–HS–SFLA” corresponds to the identical minimum LCOE of 10.3 ¢USD/kWh found independently by both the HS and SFLA techniques, each using the capital structure [1,1,8,9,10,10,10]. The consistency between these two techniques supports the robustness of the proposed investment configuration.

In terms of extreme cases, the lowest LCOE of 8.8 ¢USD/kWh was obtained using the TL algorithm under a −30% reduction in both financial rates (12.52% and 9.26%), reflecting an 18.5% reduction from the base case. Conversely, the highest LCOE of 14.3 ¢USD/kWh was recorded using the HS algorithm when both rates were increased by 30% (23.3% and 17.19%), implying a 32.4% rise. The total LCOE spread of 61.5% between these two scenarios emphasizes the significant sensitivity of USW projects to financing conditions.

Notably, the minimum LCOE achieved in this analysis (8.8 ¢USD/kWh) is competitive when compared to the global benchmark of 10.0 ¢USD/kWh for utility-scale solar PV reported in [60]. This suggests that with optimal financing and incentive design, USW projects in Colombia can achieve globally competitive costs.

Figure 4 shows the sensitivity analysis of the LCOE for SP systems in response to variations in the bank loan interest rate and the yield on green bonds. These financial parameters were adjusted within a ±30% range around their nominal values (17.89% for the bank interest rate and 13.23% for the green bond yield) in order to simulate the effects of different market scenarios. As previously, Figure 4 shows bars indicating the percentage change in the LCOE as a result of fluctuations in the bank interest rate and green bond yield. Each bar is accompanied by the corresponding LCOE value in ¢USD/kWh. The evaluated financial values range from 12.52% to 23.3% for the loan interest rate and from 9.26% to 17.19% for the bond yield.

The label “SP–TL–SFLA” refers to the minimum LCOE of 4.0 ¢USD/kWh, which was obtained using both the TL and SFLA techniques. These results were achieved under an identical capital structure of [1,1,8,9,10,10,10]. Meanwhile, the “SP–HS” label indicates the result obtained with the harmony search (HS) algorithm, which produced a slightly higher LCOE of 4.2 ¢USD/kWh with a different capital configuration of [1,4,5,9,10,10,10].

Under the most favorable financial scenario, resulting a 30% reduction in both the interest rate and the bond yield (12.52% and 9.26%, respectively), the lowest LCOE achieved was 3.2 ¢USD/kWh, representing a 23.8% reduction from the nominal case. This value was consistently reached using all three optimization techniques (TL, HS, and SFLA). Conversely, the HS algorithm achieved the highest recorded LCOE of 5.9 ¢USD/kWh, which was observed when both rates increased by 30% (23.3% and 17.19%). The difference of 54.2% between the minimum and maximum LCOE highlights the sensitivity of distributed solar systems to financing conditions, although to a lesser extent than more capital-intensive technologies.

Finally, the lowest LCOE achieved in this study was 3.2 ¢USD/kWh, demonstrating strong competitiveness relative to the global benchmark of 4.8 ¢USD/kWh for small-scale PV systems reported by IRENA in 2021 [59]. This highlights the cost-efficiency of SP installations under optimized financial structures.

The legend label “WP–TL–SFLA” refers to the minimum LCOE of 6.4 ¢USD/kWh achieved by both the TL and SFLA algorithms using the same capital structure of [1,1,8,9,10,10,10]. Meanwhile, the label “WP–HS” indicates the result obtained by the harmony search (HS) algorithm, which produced a slightly higher LCOE of 6.7 ¢USD/kWh with a different capital structure of [1,4,5,9,10,10,10].

Figure 5 presents the sensitivity analysis of the LCOE for WP technology in response to variations in the bank loan interest rate and the yield of green bonds. These two key financial parameters were varied within a ±30% range around their nominal values (17.89% for the loan interest rate and 13.23% for the green bond yield) in order to evaluate the impact of financial volatility on project feasibility and electricity generation costs. The figure depicts bars that represent the percentage change in LCOE caused by adjustments in the bank interest rate and green bond yield, with the LCOE value in ¢USD/kWh shown beside each bar.

The legend label “WP–TL–SFLA” refers to the minimum LCOE of 6.4 ¢USD/kWh obtained using both the TL and SFLA techniques, with a shared optimal capital structure of [1,1,8,9,10,10,10]. In contrast, the label “WP–HS” denotes the result obtained with the harmony search (HS) algorithm, which yielded a slightly higher LCOE of 6.7 ¢USD/kWh with a different configuration of [1,4,5,9,10,10,10].

In the most favorable financial scenario, which features a 30% reduction in both interest and bond yield rates (12.52% and 9.26%, respectively), the lowest LCOE of 5.1 ¢USD/kWh was achieved using the HS algorithm. This corresponds to a 23.9% decrease compared to the nominal case. On the other hand, when both rates increased by 30% (to 23.3% and 17.19%, respectively), the LCOE rose to 9.4 ¢USD/kWh, reflecting a 40.3% increase. The total range of variation of 54.3% demonstrates the significant impact of financing conditions on WP projects.

It is worth noting that the minimum LCOE of 5.1 ¢USD/kWh is relatively close to the global benchmark of 3.3 ¢USD/kWh reported by IRENA in 2021 for onshore wind projects [59]. This discrepancy is primarily attributed to the low capacity factor used in the Colombian case study (below 30%), which affects the competitiveness of wind generation in the local context.

These findings emphasize that financial parameters such as interest rates and bond yields substantially affect project viability even for moderately capital-intensive technologies such as WP. Moreover, the selection of metaheuristic algorithms can influence outcomes, as seen in the divergence between the HS and TL/SFLA solutions. Thus, these results underscore the importance of minimizing debt costs and optimizing the structure and maturity of green bonds in order to reduce long-term electricity costs.

In summary, investments in BESS, USW, SP, and WP technologies must consider low financing rates in order to achieve LCOEs that are comparable to global benchmarks for conventional generation technologies. The debt interest rate represents the cost that a renewable energy project must bear in order to access financing. As interest rates rise, borrowing becomes more expensive, which increases the overall capital cost. This increase directly affects the LCOE, as higher financing costs translate into greater expenditures on operations, maintenance, and capital investment, ultimately raising the cost per unit of electricity generated.

Meanwhile, green bonds are financial instruments specifically designed to fund sustainable and renewable energy projects. Their yields reflect the return expected by investors. If green bond yields rise, project developers face increased financing costs, as they must offer more attractive returns in order to attract investment. This increases the overall cost of capital, consequently raising the LCOE.

While economic incentives such as favorable debt interest rates and green bond yields have the greatest influence on LCOE, fiscal incentives also play an important role. In the Colombian context, these include a 50% deduction on taxable income for capital investment, VAT and import duty exemptions, and accelerated depreciation. Together, these fiscal measures can result in an estimated investment cost reduction of approximately 41.5%, broken down as follows: 17.5% from the ITC, 19% from VAT exemption, and 5% from import duty exemption.

Accelerated depreciation is implicitly accounted for through the ITC, and contributes to the reduction in income tax payments over the project’s operational lifetime. One of the key contributions of this work is the ability to determine the optimal number of depreciation years that maximizes the benefit of fiscal incentives while minimizing tax liabilities. Additionally, this study proposes a methodology for optimizing two key economic incentive parameters—the proportion of financing obtained through green bonds and their maturity period—using the LCOE formulation proposed in this paper.

To further explore the influence of financing conditions on the LCOE, an alternative scenario was considered in which the roles of the green bond yield and the bank loan interest rate were reversed. In this configuration, the bank interest rate was set at 13.23% effective annually, while the green bond yield was increased to 17.89% effective annually. This adjustment simulates a market condition where access to traditional bank loans is more favorable than issuing green bonds, reflecting potential shifts in investor preferences or macroeconomic conditions. The resulting LCOE values under this new financial scenario are analyzed below, providing additional insights into how the relative cost of debt instruments affects investment viability and capital structure optimization for non-conventional renewable energy projects.

Table 4. LCOE results and capital structure by technology and metaheuristic technique.

Technology	Teaching L	Structure (Teaching L)	Harmony S	Structure (Harmony S)	SFLA	Structure (SFLA)
BESS	5.4	[1,8,1,9,10,10,10]	5.5	[1,8,1,8,10,10,9]	5.7	[1,8,1,6,10,10,7]
USW	9.6	[1,8,1,7,9,10,9]	9.8	[1,7,2,9,10,10,9]	9.8	[1,7,2,7,10,9,10]
SP	3.9	[1,6,3,9,10,10,10]	4.1	[1,5,4,9,10,10,10]	3.7	[1,8,1,7,9,9,6]
WP	5.5	[1,8,1,9,10,10,10]	5.9	[1,7,2,9,10,10,10]	5.5	[1,8,1,9,10,10,10]

presents the LCOE values and associated optimal capital structures obtained for each of the BESS, USW, SP, and WP renewable energy technologies using the three different metaheuristic optimization techniques of TL, HS, and SFLA. Each column reports the minimum LCOE achieved by each method along with the corresponding vector representing the capital structure, which includes the percentage of fiscal and financial incentives such as the share of green bonds, tax deductions, and loan components.

For BESS technology, the TL method yielded the lowest LCOE of 5.4 ¢USD/kWh with a capital structure of [1,8,1,9,10,10,10], closely followed by HS and SFLA with slightly higher LCOEs of 5.5 and 5.7 ¢USD/kWh, respectively. In the case of USW, all three techniques achieved similar performance, with LCOE values ranging from 9.6 to 9.8 ¢USD/kWh and differences primarily reflected in the allocation of capital components.

SP exhibited the lowest LCOEs across all technologies, with the SFLA producing the best result of 3.7 ¢USD/kWh. This was slightly lower than the 3.9 and 4.1 ¢USD/kWh obtained by TL and HS, respectively. WP yielded equivalent LCOE values of 5.5 ¢USD/kWh using both TL and SFLA, with HS showing a slightly higher value of 5.9 ¢USD/kWh.

The consistency in capital structures between TL and SFLA, especially for WP and BESS, suggests convergence toward similar optimal configurations, reinforcing the robustness of the results. Overall, this table highlights the comparative performance of metaheuristic techniques in minimizing the LCOE under different capital structuring strategies and confirms that SP is the most cost-effective technology under the evaluated conditions.

Table 5. Comparison of CG with and without fiscal incentives by technology and metaheuristic technique.

Technology	Without Incentives (¢USD/kWh)	With Incentives (¢USD/kWh)	Reduction	TL (CG)	Reduction (TL)	HS (CG)	Reduction (HS)	SFLA (CG)	Reduction (SFLA)
BESS	16.6	14.4	13.25%	5.4	67.47%	5.5	66.87%	5.7	65.66%
USW	16.8	15.5	7.74%	9.6	42.86%	9.8	41.67%	9.8	41.67%
SP	8.0	7.2	10.00%	3.9	51.25%	4.1	48.75%	3.7	53.75%
WP	12.7	11.4	10.24%	5.5	56.69%	5.9	53.54%	5.5	56.69%

summarizes the impact of fiscal incentives on CG for each technology and metaheuristic technique. The table compares CG values with and without fiscal incentives, specifically the investment tax deduction, VAT and tariff exemptions, and accelerated depreciation, then quantifies the percentage reduction achieved in each case.

In the first three columns, the table presents the CG for each technology without and with incentives along with the overall percentage reduction. Implementation of fiscal incentives results in significant cost savings across all technologies: 13.25% for BESS, 7.74% for USW, 10.00% for SP, and 10.24% for WP. These reductions highlight the important role that fiscal policy plays in improving the competitiveness of renewable energy projects.

The remaining columns report the CG obtained using the three metaheuristic techniques of TL, HS, and SFLA and the corresponding cost reductions relative to the CG without incentives. For BESS, the most substantial reduction was achieved using the TL algorithm (67.47%), while SFLA showed the lowest reduction (65.66%). A similar trend is observed in WP, where both TL and SFLA achieved the same CG (5.5 ¢USD/kWh) with identical reductions of 56.69%, while HS performed slightly worse with 53.54%.

In the case of SP, the SFLA yielded the best result, with a CG of 3.7 ¢USD/kWh, as well as the highest reduction (53.75%), outperforming both TL (51.25%) and HS (48.75%). For USW, although the overall incentive impact was more moderate, the TL algorithm again led to the highest reduction (42.86%).

These results underscore the effectiveness of combining fiscal incentives with metaheuristic optimization to reduce electricity generation costs. Moreover, they demonstrate that the selection of the optimization technique can significantly influence the extent to which such incentives are capitalized within the project’s financial structure.

4. Conclusions

This study highlights the critical role of fiscal and economic incentives in the viability of non-conventional renewable energy sources (NCRES), particularly through their capacity to reduce the levelized cost of electricity (LCOE). By incorporating instruments such as investment tax credits, VAT and tariff exemptions, accelerated depreciation, and green bonds into capital structures, this research demonstrates how LCOE assessments can be optimized under current regulatory frameworks. Using different metaheuristic techniques (TL, HS, SFLA), this study identifies capital structures that significantly reduce generation costs. Fiscal incentives alone can lower the LCOE by up to 13.25%, while their combination with optimization methods leads to reductions exceeding 50% for capital-intensive technologies such as battery energy storage systems (BESS, 5 ¢USD/kWh), solar photovoltaic (SP, 3.2 ¢USD/kWh), and wind power (WP, 5.1 ¢USD/kWh). SP emerged as the most competitive option with an LCOE of 3.2 ¢USD/kWh, well below the global benchmark of 4.8 ¢USD/kWh set by IRENA. A key differentiator of this study is its ability to minimize the LCOE by defining an optimal capital structure tailored to the specific financial and technical conditions of each project. In addition, the proposed model can identify the most efficient investment configuration suited to a country’s unique regulatory, economic, and energy landscape. This adaptability enhances both the model’s precision and its practical applicability, making it a versatile and replicable tool across different markets. These types of incentives are especially relevant in the global context, where the energy transition is a strategic priority for climate change mitigation. Lowering the tax burden throughout the lifecycle of renewable energy projects enhances operational liquidity, which in turn facilitates reinvestment in expansion, modernization, and technological upgrades. These reinvestments not only boost expected returns for investors but also raise standards of quality, safety, and continuity in electricity supply. This study also underscores the influence of financing conditions; fluctuations in interest rates and green bond yields can cause LCOE variations of over 100%, emphasizing the need for stable and supportive fiscal environments. Our sensitivity analyses confirm that adjusting decision variables (${\alpha }_1$, ${\alpha }_2$, ${\alpha }_3$, d, k, L, r) within the optimization models can empower investors to explore a broad range of capital structures, helping to enhance profitability. Furthermore, integrating fiscal incentives can support sustainable urban solutions such as energy recovery from municipal solid waste (USW), which achieved an LCOE of 8.8 ¢USD/kWh, demonstrating the potential of circular economy approaches in clean energy generation. These incentives accelerate the adoption of renewables while also strengthening the resilience of energy systems, contributing to global decarbonization targets, and addressing key urban challenges such as efficient waste management. Finally, the proposed framework offers a robust and adaptable strategy for optimizing renewable energy costs and accelerating the transition to clean energy. It highlights the urgent need for well-designed public and private policies to support NCRES investment, thereby advancing energy diversification, carbon neutrality, and long-term sustainable development. Future research might extend this model to dynamic regulatory contexts or incorporate uncertainty analyses into the financial parameters.