Expectations Versus Reality: Economic Performance of a Building-Integrated Photovoltaic System in the Andean Ecuadorian Context

Abstract

This article presents an empirical evaluation of the technical and economic performance of a building-integrated photovoltaic (PV) system implemented at the Faculty of Architecture and Urbanism of the University of Cuenca, Ecuador. This study explores both stages of deployment, beginning with a 7.7 kWp pilot system and later scaling to a full 75.6 kWp configuration. This hourly monitoring of power exchanges with utility was conducted over several months using high-resolution instrumentation and cloud-based analytics platforms. A detailed comparison between projected energy output, recorded production, and real energy consumption was carried out, revealing how seasonal variability, cloud cover, and academic schedules influence system behavior. The findings also include a comparison between billed and actual electricity prices, as well as an analysis of the system’s payback period under different cost scenarios, including state-subsidized and real-cost frameworks. The results confirm that energy exports are frequent during weekends and that daily generation often exceeds on-site demand on non-working days. Although the university benefits from low electricity tariffs, the system demonstrates financial feasibility when broader public cost structures are considered. This study highlights operational outcomes under real-use conditions and provides insights for scaling distributed generation in institutional settings, with particular relevance for Andean urban contexts with similar solar profiles and tariff structures.

1. Introduction

According to the International Energy Agency [1], buildings account for approximately 30% of global energy use and are responsible for 26% of total carbon emissions. In Ecuador, the combined energy consumption of residential, commercial, and institutional buildings excluding industrial users represents 18% of the national demand. When industries are included, this proportion increases to 36% [2]. Due to the equatorial position of the country, seasonal variation is minimal, resulting in consistent energy requirements throughout the year.

In many parts of the equatorial Andes, cities and towns are located in temperate valleys that maintain stable climatic conditions year-round. In these regions, thermal comfort is often achieved passively or with very limited use of active systems, which contributes to relatively low energy consumption in buildings [3]. Despite this context, Ecuador’s energy system has experienced a growing strain in recent years. Reduced rainfall, likely associated with climate variability, has affected hydroelectric production, exposing the country’s dependence on a single generation source. According to the EMBER 2024 report [4], electricity production in Ecuador decreased by 0.6% between 2023 and 2024 even as demand continued to increase. This mismatch has forced the government to spend over USD 615 million on electricity imports from Colombia during that period [5], putting pressure on public finances and prompting discussion of alternative energy strategies.

Photovoltaic (PV) systems offer complementary characteristics to hydropower. During dry periods, when hydroelectric output tends to decrease, solar irradiance is generally high. In this context, water reserves in hydroelectric reservoirs can be conserved during peak solar production, enhancing operational flexibility. This synergy was explored in Brazil by Caldeira et al. [6], who analyzed hybrid systems that incorporate photovoltaic energy for reservoir recovery and energy balance. Spain provides another reference point: By 2024, its PV capacity reached 17,792 MWp, amounting to 15% of the total installed generation. This integration has enabled 100% renewable energy coverage on select days of the year.

However, large-scale photovoltaic installations often require extensive land areas, sometimes affecting ecosystems or competing with other land use cases [7]. In a country like Ecuador, where biodiversity conservation and population density are important concerns, rooftop PV systems represent a more suitable alternative. With high and stable solar radiation throughout the year, building-integrated generation offers high efficiency and reduces transmission losses. In addition, this decentralized model contributes to local grid resilience by generating power close to the point of use and facilitating peer sharing through low-voltage distribution networks. From a labor perspective, rooftop PV has also shown higher employment potential than utility-scale systems. In 2023, decentralized PV generation created 3.1 million jobs globally compared to 1.4 million in centralized installations [7]. Despite these benefits, Ecuador continues to rely heavily on electricity imports and fossil-fueled emergency generation [8,9], missing opportunities for local investment in clean energy infrastructure.

In Ecuador, photovoltaic self-generation was first regulated in 2018 through ARCONEL Regulation 042/2018 [10]. The most recent update, ARCONEL Regulation 005/2024 [9], retains the same technical conditions: The annual energy output of a grid-connected PV system must not exceed the consumer’s historic or projected annual demand. Although interest in these systems has grown since 2020, particularly in the city of Cuenca, deployment remains limited. A primary constraint is the relatively low cost of electricity—approximately USD 0.09 per kWh for commercial users and USD 0.11 per kWh for most residential consumers.

In contrast, the Ministry of Energy’s 2021 report estimates the average cost of generation at USD 0.1559 per kWh [8]. This result is based on a comprehensive methodology that incorporates generation, transmission, and distribution costs. The disaggregated values reported are as follows: 7.925 USD/kWh for energy generation, 1.257 USD/kWh for transmission, and 6.405 USD/kWh for distribution and commercialization. Although this valuation is official, it is often omitted in more recent publications, which typically limit their scope to operational and maintenance expenses. As a result, key financial components such as infrastructure depreciation, asset replacement, and future grid development are excluded. This narrowing of the cost structure conceals the real economic effort required to sustain the electricity system and weakens the basis for evaluating the fiscal benefits of distributed generation. A later report issued in 2023 [11] indicates a substantially lower figure of USD 0.09 per kWh, which does not reflect the full range of generation-related expenses. Specifically, this estimate only accounts for current infrastructure operation and maintenance, without considering total investment costs, long-term asset depreciation, or system expansion needs. This undervaluation contributes to a distorted perception of actual generation costs and discourages investment in new generation capacity. At the same time, private investors encounter high upfront costs for PV systems, which diminishes the economic viability of self-consumption models.

In 2021, a PV installation was proposed for the Faculty of Architecture and Urbanism at the University of Cuenca. The system was sized according to electricity consumption data from 2019, the last year with typical occupancy levels before the COVID-19 pandemic.

Table 1. Monthly power bills and electricity consumption (2019) for the Faculty of Architecture and Urbanism at the University of Cuenca.

Monthly Power Bills (USD)
Utility Account Code	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec	Total
2857694	1039	776	768	935	1032	1079	1147	630	787	952	961	934	11,041
Average monthly bill (USD): 920.08
Monthly Power Consumption (kWh)
Utility Account Code	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec	Total
2857694	12,763	9302	9875	12,172	12,415	13,043	13,777	7603	9554	11,489	11,654	11,308	134,954.6
Average cost per kWh (USD): 0.0818

summarizes the monthly electricity expenditure and consumption for that year.

The billed amount includes the electricity cost and additional fees, such as demand charges based on peak power, public lighting, and municipal services like waste collection and fire protection. For example, in November 2019, the total charge was USD 961.20, of which USD 724.50 corresponded exclusively to measured energy consumption. Excluding these additional charges, the effective price of electricity was approximately USD 0.062 per kWh—roughly USD 0.10 below the estimated cost of generation.

Based on these figures, a PV project was developed to offset as much of the building’s demands as possible. Meeting the 2019 annual consumption in full would have required a 100 kVA PV system. However, based on the rooftop geometry and the availability of 330 Wp panels at the time, 330 modules would be needed. Due to space constraints, the system was limited to 86.6 kWp [12]. This sizing limitation and its spatial implications have been discussed in more detail elsewhere.

The project was implemented in Cuenca, a city located in an inter-Andean valley. In this setting, commercial, institutional, and service-sector buildings contribute to just over 3% of total electricity use [13]. This relatively low share reflects the absence of energy-intensive heating and cooling needs commonly observed in regions with pronounced seasonal variation [1].

2. Previous Research

Higher education institutions are frequently among the earliest adopters of energy technologies applied to the built environment. Their buildings often serve as testbeds for self-supply renewable systems, functioning both as components of institutional sustainability policies and as platforms for experimentation and applied research. In 2021, Kourgiozou et al. [14] conducted a study across 27 universities in the United Kingdom to assess energy efficiency measures. Among these, 18 campuses included local energy generation systems, even in a country with limited solar availability. Similarly, López-Ochoa et al. [15] examined retrofitting proposals and implemented projects across three university buildings in Spain. The analysis reported a reduction of 66% in non-renewable energy use and a 71% reduction in greenhouse gas emissions.

Abulibdeh [16] explored institutional strategies aimed at reducing carbon emissions across university campuses, drawing from a range of technological approaches such as renewable energy integration, energy efficiency, and waste minimization. Through artificial intelligence tools, the study identified over 1050 universities that had either implemented or planned decarbonization strategies. It also noted that, in 2023 alone, 205 scientific publications addressed Net-Zero or related initiatives in academic settings, with the U.S. Department of Energy contributing 26 of those works.

Globally, detailed assessments of photovoltaic integration on university campuses have been carried out. At Hitit University in Turkey, Şevik [17] examined various hybrid scenarios combining grid-connected PV systems, hydrogen production, and trigeneration, estimating a payback period of 6.94 years, with unit electricity costs between USD 0.061 and 0.065 per kWh. In Southeast Asia, Maity et al. [18] assessed various configurations on the Pahang Al-Sultan Abdullah University campus. Among forest-based, floating, and building-mounted PV systems, floating installations offered the highest generation yield due to favorable thermal and spatial conditions. In South China, Cai et al. [19] used simulation tools to evaluate rooftop PV and façade-mounted elements—such as solar pergolas and filters—tailored to the geometric envelope of campus buildings.

Other investigations in tropical regions include the study by Thoy and Go [20] in Malaysia. Their simulations showed rooftop installations outperforming façades in energy production. However, due to elevated cooling loads, rooftop systems alone covered just 49.3% of the campus demand. In Colombia, Castrillón-Mendoza et al. [21] analyzed a 301.25 kW PV system deployed at the Universidad Autónoma de Cali. The comparison between actual production and PVsyst simulations yielded a deviation of only 2.8%, indicating a strong alignment between design and operation. Additional works [22,23,24] have broadened the scope of technical and economic analysis in university PV integration, confirming that campus-scale projects continue to offer a valuable area for applied study.

The system examined in this article was developed in two stages. The first installation included 7.7 kWp, followed by a second phase adding 68 kWp. Two prior publications have addressed this case. The first [25] described the architectural and technical parameters, confirming that a total of 228 PV modules, each rated at 385 Wp, could be deployed across three flat rooftops to reach a capacity of 86.6 kWp. Based on meteorological and demand data from 2019, the projected annual generation was 119,563 kWh—approximately 88.6% of the building’s consumption in that year. This projected generation considered an ideal scenario without shading effects. However, minor shadowing occurs before 8:00 a.m. and after 5:30 p.m., primarily affecting the PV arrays installed at the east-facing block. This interference leads to an estimated annual loss of just 798 kWh, which corresponds to 0.67% of the total expected energy output. These losses were incorporated into the simulation and remain consistent with the equatorial location of the system, where solar altitude stays relatively stable and no adjacent structures produce substantial shading during central daylight hours.

Following the initial installation, a previous study examined the alignment between daytime energy use and photovoltaic production [12]. This analysis, based solely on simulations, found a close match between electricity demand and solar irradiance profiles, particularly on weekdays. Based on this observation, the feasibility of operating the system under a self-consumption scheme without a bidirectional meter was preliminarily assessed. The simulations projected that just 2.2% of the energy produced would exceed building demand, with most surpluses occurring during weekends and holidays. While this finding suggested limited value in exporting surplus energy, it relied exclusively on theoretical modeling.

The present study moves beyond this initial outlook by incorporating empirical monitoring data from a fully deployed 75.6 kWp system installed in two stages. It delivers a detailed evaluation of real-world photovoltaic performance, electricity usage, energy export profiles, and economic metrics. By contrasting monitored outcomes with the original simulations, the analysis uncovers discrepancies—such as higher-than-expected surpluses on certain days—and quantifies their financial implications. This approach adds depth to the prior work by validating and extending the simulated results under actual operating conditions. The findings contribute to a more grounded understanding of self-consumption feasibility and return on investment within the context of public institutional infrastructure in the Andean region.

In addition, this study contributes empirical evidence from a high-altitude equatorial location, a geographic and climatic condition that remains underrepresented in long-term PV performance research. The monitoring results presented here can support future institutional-scale deployments by offering realistic performance expectations under comparable solar and altitudinal conditions found in various Latin American cities, such as Bogotá, Quito, Arequipa, and Mexico City.

This research also contributes to the body of knowledge by providing a detailed analysis of actual PV performance, including losses and deviations that emerge in real-world operation, such as grid interruptions, partial dusting, and minor shadowing. It further incorporates a comparative cost evaluation between simulation-based projections and the actual financial outcomes after system deployment. These elements are particularly relevant in equatorial high-altitude settings, where few long-term monitoring datasets are available. In addition to well-documented capitals, many intermediate cities across the region—ranging from Mexico City to Ayacucho—share similar elevations between 2000 and 3000 m above sea level. The findings may therefore inform future planning and decision-making in a broad spectrum of urban environments.

3. Methodology

The implementation of a photovoltaic system in an institutional setting requires navigating a range of challenges, including budget constraints, grid interconnection conditions, and structural limitations inherent to the building. These factors can alter the project’s initial parameters. In the case examined, the system was installed in two distinct phases. The first became operational in June 2022, while the second phase—completing the originally planned capacity—was activated in October 2023. This sequential commissioning means that complete performance data for a full operational year is only available from 2024 onward. The data were retrieved from the Growatt cloud platform, which provides hourly resolutions on system behavior, enabling detailed analysis. These measurements are contrasted with previously conducted simulation results.

To improve clarity, the overall process followed in the study is now illustrated in Figure 1, which presents the main methodological steps, including installation phases, monitoring, data correction, simulation comparison, and financial assessment.

Several interruptions in data continuity were identified in April and during October–November of 2024. These disruptions were caused by nationwide power outages associated with a period of energy supply instability. To compensate for the missing values and ensure consistency in the annual analysis, the gaps were reconstructed using a temperature-dependent photovoltaic performance model based on real-time irradiance and ambient temperature data from a nearby meteorological station with uninterrupted power supply. This method ensures that the reconstructed values reflect realistic production under normal operating conditions, allowing for a more robust comparison with the baseline energy simulation.

Due to modifications in the final system setup—including changes in installed capacity and equipment specifications—the energy production values differ from the original project estimates. Consequently, the evaluation incorporates both energetic and economic metrics to understand actual system performance and cost-effectiveness relative to expectations.

For performance modeling, the system advisor model (SAM) was selected instead of RETScreen. While both tools are validated and widely used, SAM was preferred due to its capability to process hourly load profiles, detailed irradiance data, and location-specific system configurations. This level of granularity was essential to simulate conditions that align closely with the empirical monitoring strategy described below.

The second stage of the methodology centers on establishing a correlation between hourly PV output and building energy demand. To achieve this, a monitoring campaign was conducted over a full month, capturing both periods of academic activity and scheduled recess, each lasting two weeks. A Fluke 430-II Series analyzer, part of the instrumentation available at the Microgrid Laboratory of the University of Cuenca [26], was employed for this purpose. This portable, high-precision device is specifically designed for the analysis of three-phase electrical systems. It records root mean square (RMS) values for voltages and currents, electrical frequencies, power factors, and harmonic distortions with high temporal resolutions. The device supports simultaneous multichannel acquisition and synchronized waveform sampling, enabling detailed transient analysis. It integrates advanced calibration and signal filtering to reduce noise and measurement uncertainty. Its hardware complies with international standards such as IEC and ANSI, ensuring reliable performance in varied field conditions [27].

This monitoring phase supports the identification of energy flows in both directions—into the grid during surplus production and from the grid during times of shortfall. It also allows for the assessment of the load transferred to the transformer supplying the faculty, as well as the characterization of peak injection values and daily energy balances. In parallel, official electricity bills are analyzed to determine the quantities of energy injected into and purchased from the grid. This enables triangulation between physical measurements, utility meter readings, and economic transactions.

The final stage of the methodology compares the monthly energy production measured on-site with actual consumption throughout the year. This comparison allows for the calculation of the solar fraction—the proportion of demand covered by PV generation—and total energy usage and billing details. These metrics enable a comparison between the original projections and real-world outcomes, allowing for an evaluation of deviations and sources of discrepancy. In addition to validating the simulation approach used in the initial feasibility analysis, this stage offers a reference for estimating long-term system behaviors and for assessing investment recovery timelines.

4. Results

4.1. Background Project

The initial project’s objective was to supply the full annual electricity demand of the building complex through photovoltaic generation. However, during the design and implementation stages, several constraints emerged that limited this goal. One of the primary restrictions was the available rooftop surface suitable for PV module installation. Although flat terraces were initially identified for module placement, their usable area proved to be insufficient for reaching the system’s projected capacity. In addition, the amount of power permitted for injection into the public distribution grid was limited by the rating of the transformer feeding the faculty facilities, further restricting the system’s functional scope.

The distribution utility authorized a maximum output of 67 kW based on the specific capacity of the transformer connected to the Faculty of Architecture and Urbanism. This constraint directly limited the final installed capacity to 75.6 kWp, even though simulations had identified 86.6 kWp as necessary to fully cover the building’s annual consumption. Such technical restrictions are evaluated independently for each transformer, meaning that any campus-wide expansion would require separate injection assessments per faculty building depending on feeder topology, transformer loading, and voltage regulation margins.

An additional setback occurred in March 2023, shortly before the installation of the second system phase. The Balao earthquake affected multiple provinces in southern Ecuador, and one of the faculty’s building blocks presented structural damage after the seismic event [28]. As a precaution, the rooftop of this structure was declared to be unsuitable for equipment installation until a complete structural assessment and reinforcement plan could be executed. This development reduced the initially available area for PV deployment, compelling further design modifications.

According to the original plan, 228 PV panels rated at 380 Wp each were to be installed, yielding a total capacity of 86.64 kWp. The performance of this configuration was assessed using SAM, developed by the National Renewable Energy Laboratory (NREL, USA). Climate data specific to the city of Cuenca were incorporated into the simulation [29]. Based on this modeling, the estimated energy yield was 1357 kWh per installed kWp annually, resulting in a projected annual output of 117,624 kWh. Under these assumptions, the system would save approximately USD 9615 in electricity expenses per year. The projected payback period was 12.5 years, which under typical Ecuadorian financial conditions—characterized by relatively high commercial interest rates—renders the investment only moderately attractive.

However, if the analysis considers the actual average cost of electricity generation in Ecuador—estimated at USD 0.1758 per kWh [11], which includes USD 0.158 for generation and USD 0.02 for associated charges—the economic indicators shift accordingly. Under this framework, the projected payback period is reduced to approximately 5.5 years, with an estimated return on investment of 18%.

Public education in Ecuador, including universities, is largely financed by the state. In practice, utility costs associated with public academic infrastructure are covered through direct or indirect public funding mechanisms. According to national statistics on educational infrastructure [30], there are approximately 9900 public educational institutions across the country, including 31 public universities and 76 higher technical institutes. Although no consolidated data exist on electricity consumption at the national level, it is estimated that the Faculty of Architecture represents around 5% of the built area of the University of Cuenca. Based on this proportion, monthly and annual energy expenditures for similar institutions represent a substantial public budgetary allocation. Since the cost of electricity is paid through state subsidies, using the full system cost rather than the subsidized tariff provides a more accurate representation of fiscal savings.

From this perspective, the avoided annual expenditure from public funds due to the PV system is approximately USD 20,774. A comparative summary of the two evaluation scenarios is presented in

Table 2. Initial-stage comparison of economic outcomes based on billed price versus real electricity cost for an 86.6 kW photovoltaic system.

Metric	Billed Price Scenario	Real Cost Scenario
Annual energy (year 1)	117,624 kWh	117,624 kWh
Capacity factor (year 1)	15.5%	15.5%
Energy yield (year 1)	1357 kWh/kW	1357 kWh/kW
Performance ratio (year 1)	0.81	0.81
Levelized COE (nominal)	6.98 USD/kWh	6.98 USD/kWh
Levelized COE (real)	5.64 USD/kWh	5.64 USD/kWh
Electricity bill without system (year 1)	USD 11,154	USD 24,029
Electricity bill with system (year 1)	USD 1539	USD 3255
Net savings with system (year 1)	USD 9615	USD 20,774
Net present value	USD 17,551	USD 165,227
Simple payback period	12.5 years	5.5 years
Discounted payback period	17.8 years	6.3 years
Net capital cost	USD 115,150	USD 115,150
Equity	USD 115,150	USD 115,150
Debt	USD 0	USD 0

. Additional details on planning and implementation are available in a previous study [25].

Due to the technical and logistical limitations described above, the photovoltaic system underwent multiple adjustments throughout its planning and execution. These modifications occurred progressively during the design, pre-feasibility evaluation, formal approval, and construction stages. The final configuration diverged from the original design in both scale and physical distribution.

Although hybrid PV–thermal (PVT) systems, cogeneration technologies, and demand response incentives offer alternative pathways for improving energy management, their relevance in this project is limited by the operational characteristics of the faculty buildings. There is no demand for space heating, air conditioning, or hot water, which precludes thermal integration as a meaningful enhancement under current conditions. Previous analyses conducted by the authors on solar thermal deployment in Ecuador [31] demonstrate that low fossil fuel prices—particularly the subsidized cost of liquefied petroleum gas at USD 1.60 per 15 kg—discourage the adoption of thermal technologies. Should these subsidies be phased out, PVT could become viable in facilities with stable thermal loads. However, in this institutional context, improving electrical self-consumption remains the primary objective.

Figure 2 illustrates a comparative view between the initial 86.6 kWp system and the installed configuration. The original design envisioned the deployment of 228 PV modules arranged across three rooftop areas, targeting compliance with Net-Zero Energy criteria. However, after structural re-evaluations, limited surface availability, and funding delays, only 75.6 kWp could be installed. The project was implemented in two separate phases spaced approximately one year apart, reflecting the time required to secure institutional funding and allocate internal resources for procurement and installation.

4.2. Development and Outcome of First Stage

Institutional budget constraints necessitated the division of the project into two phases. The first stage involved the installation of a 7.7 kWp PV system connected to a 7 kVA inverter, located on one of the accessible rooftops of the Faculty of Architecture and Urbanism (Figure 3). This installation was completed and brought into operation in July 2022, remaining as a standalone system until May 2023 [12].

According to the operational data recorded by the Growatt monitoring platform [32], a total of 8952.9 kWh was produced over the ten months for which complete generation records were available. To estimate the system’s performance over a full year, the production for the two missing months was extrapolated using the average generation from April and May. These months were selected due to their similar irradiance profiles, which in the local context typically represent periods of lower solar availability. The resulting annualized production estimate is 10,432.3 kWh.

Dividing this figure by the installed capacity yields a specific production of 1354.8 kWh per installed kW for the first year. This value aligns very closely with the simulation projection of 1357 kWh/kWp/year, indicating high predictive accuracy in the initial performance modeling.

Table 3. Electricity generation achieved over ten months and extrapolated to annual production, estimating values for June and May.

Month	Jun	Jul	Aug	Sep	Oct	Nov	Dec	Jan	Feb	Mar	Apr	May	Total
Energy (kWh)	727.6	729.1	870.0	973.0	933.4	978.8	1036.1	942.4	839.6	900.2	750.3	751.8	10,432.3

summarizes the monthly energy production values and highlights the interpolated estimates for June and May.

The pro forma installation cost, as calculated in December 2021, amounted to USD 9946.00. Based on this estimate and considering a gradual annual reduction in module efficiency (1%) and a modest annual increase in electricity prices (1%), the projected net payback period was 12.6 years. This conservative inflation assumption contrasts with the 3% annual increase estimated by the International Energy Agency but reflects the effects of longstanding electricity subsidies in Ecuador.

Due to the limited scale of the installation, the cost per installed watt was relatively high. This, combined with a marginally lower-than-expected yield, initially suggested limited economic competitiveness. However, when recalculating returns based on the actual cost of electricity to the state, the investment’s financial outlook changes substantially. Under this adjusted scenario, the return on investment is estimated at 18%, with a shortened payback period of 5.5 years. From this perspective, the installation stands out as a sustainable initiative and an efficient allocation of public resources in the energy sector.

4.3. Development and Outcome of the Second Stage

Eleven months after the implementation of the initial 7.7 kWp system, the installation of the remaining capacity was completed, thereby finalizing the project. During the permitting process, an application was submitted to obtain approval for the maximum generation capacity permitted based on the rating of the existing service transformer (75 kVA). The electricity distribution utility authorized the connection of an additional inverter rated at 60 kVA, which would operate in conjunction with the previously installed equipment. As a result, the total inverter capacity reached 67 kVA.

The new system configuration was based on the installation of 172 photovoltaic modules, each rated at 395 Wp. These modules were integrated into the existing infrastructure, complementing the 20 modules already in place. This upgrade brought the total system capacity to 75.64 kWp, distributed across five inverters with different orientations and nominal powers. The final configuration is illustrated in Figure 4, which shows the physical distribution of the modules and the separation between the original and expanded systems.

During the 2024 operational year, the system recorded a total production of 97,435 kWh, based on real-time monitoring by the Growatt inverter platform (

Table 4. Monthly energy production (kWh) per inverter in the 75.6 kWp system during 2024.

Inverter Serial Number	Orientation	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec	Total (kWh)
EKDUCGXXXX (56.9 kWp, 50 kVA)	N/S/E/W	7009.9	6317.3	6453.4	5694.7	5884.0	6010.2	5677.8	6793.3	5857.2	4496.1	6712.6	6720.7	73,627.3
FPH2B1XXXX (2.3 kWp, 2 kVA)	W	290.3	256.7	256.5	227.6	236.6	236.8	220.1	257.9	217.9	184.1	259.3	263.8	2907.6
THG2BDXXXX (5.4 kWp, 5 kVA)	N	625.7	583.8	610.5	550.0	581.1	606.7	564.1	681.3	560.6	404.4	592.6	594.7	6955.4
YDDQD4XXXX (5.5 kWp, 5 kVA)	W	658.8	592.7	599.1	534.4	558.7	575.7	538.7	615.6	521.5	417.5	620.3	627.5	6860.6
ZMG3BMXXXX (5.5 kWp, 5 kVA hybrid)	E	673.4	609.6	619.5	544.3	559.6	570.6	550.3	641.0	572.8	430.8	669.2	643.2	7084.2
Monthly Total (kWh)	–	9258	8360	8539	7551	7820	8000	7551	8989	7730	5933	8854	8850	97,435.1

).

To better reflect the expected performance under uninterrupted conditions, adjusted monthly values were derived. For the months affected by nationwide outages, including April, September, October, November, and December, the generation data were reconstructed using physical estimations based on meteorological inputs. Rather than relying on adjacent-month averages, the correction used irradiance and temperature data from a nearby weather station with UPS backup to simulate the expected PV output through the temperature-dependent model.

For validation, a parallel estimation was performed using meteorological data collected by a nearby weather station with UPS backup, ensuring uninterrupted measurements. The estimation applied a temperature-dependent performance model adapted from Skoplaki and Palyvos (ref. [33]), enabling the calculation of the expected photovoltaic output under real environmental conditions. Specifically, the DC power output was estimated using the following expression:

$$P_{DC}\left(t\right)\left[\mathrm{kW}\right]=G\left(t\right)·{\displaystyle \frac{62.807}{1000}}·\left[1-0.004·(T_{\mathrm{amb}}\left(t\right)-25)-0.004·\left({\displaystyle \frac{25·G\left(t\right)}{800}}\right)\right]$$

(1)

In this expression, $P_{DC}\left(t\right)$ represents the instantaneous direct current power output in kilowatts (kW): $G\left(t\right)$ is the global horizontal irradiance at time t, expressed in watts per square meter (W/m²); $T_{\mathrm{a}}\left(t\right)$ denotes the ambient temperature at time t, given in degrees Celsius (°C). The numerical coefficients correspond to the scaled nominal capacity of the actual system in kilowatts and the global efficiency, while the terms within the brackets account for temperature and irradiance-related performance losses.

This method yielded a total of 104,452.0 kWh. The close agreement with the adjusted value (less than 0.5% deviation) confirms the validity of the correction methodology. The integration of both data-driven correction and physical modeling offers a more reliable estimation framework than interpolation alone, particularly under conditions of unstable grid availability.

Figure 5 illustrates the comparison between the recorded and estimated monthly generation, demonstrating the deviation caused by power interruptions. Meanwhile, Figure 6 provides the full monthly breakdown of measured and adjusted production, including the reconstructed values for April and September to December.

Overall, the difference between the real monitored output and the estimated generation based on environmental data was approximately 7.2%, which aligns with the cumulative impact expected from the five grid interruptions.

According to the total solar irradiation received on the 441 m² PV surface, the efficiency projected by the SAM simulation was 16.9%, based on an expected output of 117,624 kWh from 695,218 kWh of annual irradiation. In comparison, actual measurements indicated a generation of 101,583 kWh, based on a total incident irradiance of 585,446 kWh, resulting in an operational efficiency of 17.6%. This increase may be due to the use of updated PV modules with slightly improved characteristics compared to those used in the initial simulation inputs.

The simulation included azimuthal angle deviations as part of the loss assessment. According to the SAM loss report, the estimated impact from orientation misalignment under local solar path conditions is 3.02%. This loss was incorporated into the performance yield forecast.

The inverter configuration used in the SAM simulation reflects the actual layout with Growatt units and includes default efficiency parameters. The expected energy loss in conversion was 1.94%, with an additional 0.388% attributed to operational consumption and standby power draw, particularly overnight. These values were incorporated into the model as part of the total system derating and aligned closely with manufacturer specifications.

The 5 kWh battery unit, linked to a hybrid inverter and managing 5.46 kWp of the total capacity, was integrated for educational purposes within the architecture faculty. Its main function is to support student training in residential-scale energy storage systems. Due to its small scale, the battery does not alter building-wide performance profiles and was not included in simulations of long-term yield or storage behavior. With an investment cost of USD 2550, it accounts for only 6.6% of the total installed capacity and does not influence the system’s financial indicators or daily self-consumption performance.

Through this adjustment procedure, the system’s monitoring records indicate that the energy produced during April 2024 was 7551 kWh. However, under a scenario without power interruptions, the estimated output for that month increases to 8190 kWh based on adjusted projections. A similar discrepancy was found in October, when registered production was 5933 kWh. By correcting this value using a model based on local meteorological conditions, a revised estimate of 9538 kWh is obtained.

The corrected monthly values result in an overall increase of 4244 kWh in annual energy yield, raising the total from 97,435 kWh to 104,452 kWh. This corresponds to a 4.4% increase in reported output, derived exclusively from meteorological reconstruction in the five months affected by grid outages.

Considering the total annual irradiation available on the PV surface of 441 m² where the 192 modules were installed, the projected efficiency based on SAM simulations was 16.9%, calculated from an expected output of 117,624 kWh under 695,218 kWh of incident solar energy. The measured system performance yielded a higher conversion rate of 17.6%, which may be explained by the updated characteristics of the modules installed, differing slightly from those modeled in the simulation. When the real system output is compared against the measured irradiance of approximately 585,446 kWh for the same area, the resulting operational efficiency increases to 17.6%. This higher value may be explained by the improved characteristics of the newer modules actually installed, which slightly outperform those initially specified in the simulation model.

4.3.1. Instrumentation and Maintenance Assumptions

To evaluate the dynamic interaction between onsite photovoltaic generation and electricity drawn from the utility grid, a detailed power monitoring campaign was carried out. A Fluke power quality analyzer was installed at the main distribution panel of the Faculty of Architecture and Urbanism, enabling the acquisition of high-resolution measurements of active power (in kW) over a continuous one-month period.

Based on the 0.5% annual degradation rate assumed in the SAM simulation, the projected energy output in year 25 decreases to 92,162 kWh, which represents 88.25% of the baseline generation. This long-term performance projection supports the financial assumptions used in the payback analysis. The model includes an annual maintenance cost of USD 15/kW for system upkeep, with an additional USD 2.25/kW allocated for potential inverter replacement and USD 1.75/kW reserved for the battery component. Although small, the battery’s presence is considered in the cost structure. However, its functional configuration and limited scale do not influence the hourly power profiles or load compensation metrics presented in this section. Consequently, storage efficiency and degradation were not included in the dynamic performance assessment. A 2% annual increase in maintenance cost was applied to reflect inflation over the analysis horizon.

4.3.2. Power Flows and Daily Profiles

Figure 7 illustrates the resulting power exchange profile. Positive values represent electricity consumption from the grid, while negative values correspond to surplus photovoltaic generation injected into the local transformer. The plot inherently reflects net consumption, as the power exported to the grid is recorded after subtracting the immediate onsite use.

The analysis reveals that electricity drawn from the grid occurs predominantly during nighttime and early morning hours, where a relatively stable base load between 15 and 17 kW is observed. With the onset of daily activities around 7:00 a.m., the demand rises sharply, commonly reaching 25 kW and peaking at 35 kW on weekdays. A single instance during the measurement window recorded a peak of 40 kW. Occasional late-afternoon peaks of approximately 25 kW were also identified, though they were limited to four instances across the month.

Regarding grid injections, excess generation occurred on all monitored days, confirming that the photovoltaic system regularly surpasses building demand during daylight hours. The most pronounced injection peaks were observed on weekends, denoted with “S” for Saturdays and “D” for Sundays in Figure 7, where three out of four Sundays exhibited surpluses near 55 kW. During weekdays, injections typically ranged between 30 and 40 kW, occasionally surpassing this threshold. Only two weekdays recorded lower injection values, remaining below 10 kW.

4.3.3. Weekly Trends and Grid Dependency Reduction

The temporal distribution of these events is governed by two primary factors: the variability of solar irradiance—principally influenced by cloud cover—and fluctuations in building demand. A consistent midday drop in consumption is visible on working days between 13:00 and 15:00, aligning with the scheduled academic recess when teaching and administrative activity is paused.

During the same monitoring period, several representative days exhibit contrasting energy dynamics shaped by both meteorological and operational conditions. One clear example is Sunday, March 17, characterized by minimal cloud cover and very low building occupancy. As illustrated in Figure 8, this date represents an optimal scenario for photovoltaic performance with minimal onsite demand. According to the Growatt Server platform, the system reached a peak output of 63,759.2 Wp. However, the Fluke monitoring equipment registered a maximum export to the utility grid of 55,080 Wp, which indicates a concurrent base consumption of 8679 Wp. When energy flows are integrated over the 24 h period, the daily net surplus amounts to 107 kWh. The highest momentary surplus, 193 kWh, occurred around 17:00, and then progressively diminished due to nighttime consumption.

In contrast, Friday, 22 March, exemplified the opposite condition. A weekday with overcast skies and correspondingly low solar irradiance, combined with full campus activity, resulted in a marked imbalance between consumption and generation. Peak photovoltaic output was registered at approximately 10:25 a.m., with a modest export of only 7170 Wp. Grid injection was limited and brief throughout the day. The overall daily energy balance reveals a deficit of 302.4 kWh, underscoring the sensitivity of energy performance to both demand and weather variability. As shown in Figure 9, the contrast with the surplus observed on 17 March reaches nearly 500 kWh.

A broader weekly review of the data confirms consistent patterns. From Monday to Thursday, electricity demand is the highest, with daily consumption ranging between 497 and 564 kWh. Friday shows a slight decrease, while weekends—particularly Sundays—register the lowest values, fluctuating between 263 and 321 kWh. Notably, export surpluses were recorded on only three occasions across the month: two Sundays and one Saturday. The highest daily surplus reached 130 kWh.

This variability is further examined in Figure 10, which contrasts weekday generation and consumption under low-irradiation conditions, emphasizing the role of real-time monitoring in identifying both production deficits and self-sufficiency opportunities.

Finally,

Table 5. Comparison of electricity consumption from the public distribution grid in the periods 2019 (without PV system) and 2024 (with PV system).

Month	2019 (kWh)	2024 (kWh)
January	12,763.26	2399.04
February	9302.40	1380.06
March	9874.62	3855.60
April	12,171.66	4336.02
May	12,415.38	5351.94
June	13,042.74	4537.98
July	13,777.14	4920.48
August	7603.08	0.00
September	9553.86	2844.79
October	11,489.28	2844.78
November	11,653.50	0.00
December	11,307.72	0.00
Annual Total	134,954.64	34,494.69

presents a comparison of monthly electricity consumption from the utility grid between the pre-PV year (2019) and the post-implementation year (2024). The results clearly indicate a marked decrease in purchased energy, dropping from 134,955 kWh in 2019 to 34,495 kWh in 2024. This 74.5% reduction highlights the measurable economic and operational impact of the system under real operating conditions.

4.3.4. Integration and Functional Scope of the Lithium-Ion Battery System

The hybrid component of the building-integrated photovoltaic (BIPV) system consists of a Growatt (Growatt New Energy Technology Co., Ltd., Shenzhen, China) SPH 5000TL BL-US inverter configured for split-phase 240 V outputs, integrated with a 5.46 kWp PV input and a 5.0 kWh Growatt AXE lithium-ion battery system. This configuration was selected to emulate a residential-scale hybrid setup within an institutional context [34]. Its implementation supports educational objectives, allowing students and instructors to analyze energy storage operation, hybrid control behavior, and architectural constraints of limited-capacity battery systems. The inverter supports dual MPPT inputs, accepts up to 550 V DC, and allows for 75 A charge/discharge from the battery bank. Its maximum output is 4999 VA, with a conversion efficiency of 97.6%, and it complies with North American interconnection standards including UL1741 and IEEE1547.

The AXE 5.0L battery system provides a nominal energy capacity of 5.0 kWh at 51.2 V and uses LiFePO₄ cells with integrated BMS. Its round-trip efficiency is approximately 92%, and it supports continuous operation at 60 A. Although modest in size compared to the building’s energy demands, the battery enables short-term load shifting and offers a functional platform to explore CAN-based communication, charge control, and state-of-charge tracking within real deployment conditions [35].

Figure 11 illustrates the wiring configuration and logical flow of the hybrid subsystem, including PV input strings, battery interconnection, grid interface, and load management boundaries. The inverter operates in a grid-connected mode, while the battery is configured to prioritize self-consumption based on a time-of-use logic.

Given its limited scale, the hybrid subsystem does not alter the building’s total demand curve significantly and was excluded from hourly simulations focused on aggregated grid interaction. Nonetheless, its integration provides didactic value and helps evaluate practical constraints and the expandability of low-voltage storage in institutional environments. The connected load consists of a low-consumption design studio classroom, which benefits from uninterrupted power supply during daylight and moderate evening hours. This configuration establishes a test bench for energy self-sufficiency, demand-side efficiency, and autonomous load operation, with the goal of gradually expanding hybrid functionalities across other educational spaces. Future system iterations may scale the battery array or explore alternatives with higher energy density and improved cycle efficiency.

4.3.5. Power Quality Assessment at the Grid Connection Point

The hourly monitoring campaign also enabled a power quality assessment of the system at the grid interconnection point, in accordance with the technical criteria defined by the Ecuadorian grid code (Regulación No. ARCERNNR 002/20 [36]). During a full representative week, three-phase indicators were calculated to verify compliance with regulatory voltage and harmonic distortion thresholds.

The following parameters were computed, as defined in the regulations: voltage deviation $\Delta V_k$, total harmonic distortion of voltage (THD), and total demand distortion of current (TDD).

The voltage deviation index was calculated using Equation (2):

$$\Delta V_k={\displaystyle \frac{V_k-V_N}{V_N}}\times 100$$

(2)

The THD and TDD values were computed as follows:

$$TH{D}_k={\left[{\displaystyle \frac{1}{V_{h,1}}}\sum _{h=2}^{50}{\left(v_{h,k}\right)}^2\right]}^{1/2}\times 100$$

(3)

$$TD{D}_k={\left[{\displaystyle \frac{1}{I_{h,1}}}\sum _{h=2}^{50}{\left(i_{h,k}\right)}^2\right]}^{1/2}\times 100$$

(4)

The variable $\Delta V_k$ in Equation (2) represents the relative deviation of the voltage at connection point k with respect to its nominal value $V_N$, where $V_k$ is the average voltage measured over a 10 min interval. In Equation (3), $TH{D}_k$ quantifies the total harmonic distortion of voltage at point k, where $v_{h,k}$ denotes the root mean square (RMS) value of the voltage component of harmonic order h, and $V_{h,1}$ corresponds to the RMS value of the fundamental component. Similarly, Equation (4) defines the total demand distortion of current, $TD{D}_k$, based on $i_{h,k}$; the RMS value of the current harmonic of order h; and $I_{h,1}$, the RMS value of the fundamental current component at the same point. The summations extend from harmonic order $h=2$ to $h=50$, in accordance with standard power quality guidelines.

The values obtained are summarized in

Table 6. Power quality indicators at the public grid connection point.

Parameter	Phase A	Phase B	Phase C
$\Delta V_{\max }$ [%]	4.20	3.80	3.19
THD [%]	0.94	1.00	1.03
TDD [%]	2.47	3.37	4.49

. For the low-voltage connection level applicable to the Faculty of Architecture, the thresholds established by the regulation are as follows: $\pm 8\%$ for $\Delta V_k$, 8.0% for THD, and 5.0% for TDD. All measured values remained well within these permissible limits.

The frequency was also monitored throughout the week. It remained stable between 59.99 Hz and 60.16 Hz, with a mean value of 60.085 Hz. As this parameter is governed by the national bulk power system, it is not influenced by the local photovoltaic system.

The power quality indicators at the connection point comply with all thresholds established by the Ecuadorian grid code. This confirms that the integration of the 75.6 kWp system has no adverse effect on network voltage or harmonic conditions.

4.3.6. Comparison with a Nearby PV Installation

To complement the performance assessment of the main 75.6 kWp PV system, a second installation operated by the Microgrid Laboratory of the University of Cuenca was considered for comparison. Located approximately 3.3 km from the study site, this facility includes a 35 kWp grid-connected array composed of three segments (SFV1, SFV2, and SFV3). Only SFV3, which accounts for 14% of the total installed capacity, features single- and dual-axis solar tracking, while the remaining modules consist of fixed monocrystalline and polycrystalline arrays facing north [26].

In 2024, the Microgrid Laboratory registered an actual total production of 37,039 kWh. Since this installation was also affected by nationwide power outages, it was necessary to apply a production adjustment procedure based on reliable meteorological records to enable a fair comparison. This adjustment allowed recalculation of net production at 44,002 kWh, yielding a capacity factor of 14.35% (Figure 12). For comparison, the FAUC system produced 97,435 kWh with a nearly capacity factor of 14.71%. Despite differences in scale, configuration, and module type, the two systems share comparable environmental exposure and were affected by the same nationwide grid outages. The monthly production histograms reveal similar seasonal dynamics and amplitude, reinforcing the consistency of the recorded data.

Although this is not a formal multi-site performance study, the inclusion of this local reference strengthens the interpretation of the monitored outcomes. The alignment in capacity factors underlines the coherence of the empirical results and supports the validity of the methodology applied to correct incomplete months. This local comparison adds contextual relevance to the findings, especially within high-altitude equatorial environments. Furthermore, the use of a physics-based estimation model grounded on continuously recorded meteorological variables, combined with the comparison against a nearby 35 kWp PV installation, provides additional support for the reliability of the cloud-based monitoring platform.

4.3.7. Financial Analysis

This section evaluates the economic performance of the system by contrasting installation costs and expected savings across two implementation stages: the initial 7.7 kWp installation and the full 75.6 kWp configuration. The financial figures are derived from the actual invoices issued for both installations, ensuring a grounded comparison based on real expenditures.

The first installation, rated at 7.7 kWp and operational for a full year, incurred a cost of USD 10,236.40, which equates to approximately USD 1.33 per installed watt. In contrast, the total installed cost of the complete system reached USD 99,741.36, yielding a marginally lower unit cost of USD 1.32 per watt. Although economies of scale typically suggest more substantial reductions at higher installed capacities, the cost decrease in this case was minimal due to three distinct factors affecting the second-stage deployment.

First, the 7.7 kWp system benefited from a more favorable electrical connection arrangement. It was installed on the upper floor and connected to an adjacent distribution panel, which reduced the need for extended runs of both DC and AC cabling. Conversely, the larger inverters of the 67.9 kWp system were installed at ground level due to space and power requirements, necessitating a connection to the building’s main distribution board, which involved more complex wiring infrastructure.

Second, the expanded system included a hybrid inverter rated at 5 kVA and a 5 kWh battery module. These additional components contributed notably to the total system cost and were not present in the first-stage configuration. While the battery was included for training applications, its limited capacity and cost contribution (USD 2550) did not materially affect the financial performance metrics of the complete system.

Third, the integration of the 67.9 kWp installation required connection to the medium-voltage utility network, which imposed the added requirement of installing a dedicated transformer to comply with local interconnection standards.

Additionally, the monthly generation variability was analyzed to assess system stability. The standard deviation and variance for each month are included in

Table 7. Monthly energy production for 2024 with standard deviation and variance.

Month	Adjusted Energy (kWh)	Std. Dev. (kWh)	Variance (kWh2)
January	9258	20.78	431.95
February	8360	18.91	357.41
March	8539	19.94	397.58
April	8190	19.84	393.53
May	7820	20.83	433.92
June	8000	20.18	407.27
July	7551	21.46	460.69
August	8989	22.89	523.98
September	8586	23.57	555.71
October	9538	21.46	460.63
November	10,494	21.40	457.81
December	9127	20.47	418.87
Total	104,452	–	–

as performance descriptors. These values show relatively low dispersion, with standard deviation values mostly below 23 kWh, indicating stable output patterns across the year. Slightly higher variability was observed in August and September, likely due to increased cloud cover. Overall, the statistical consistency supports the system’s suitability for long-term financial and operational planning.

In Ecuador, electricity tariffs vary across user categories, including residential, commercial, industrial, and service-sector customers. Each group is subject to differentiated rates based on demand type and consumption thresholds. For instance, residential users exceeding 1000 kWh/month face unsubsidized pricing, while commercial and institutional users often benefit from partially subsidized rates. The case analyzed in this study corresponds to an institutional service user—specifically the educational category—which is among the lowest tariffs available. Therefore, the economic indicators presented (e.g., payback period, NPV) reflect a conservative scenario. In applications involving higher-tariff users, such as residential or commercial facilities, financial performance would likely improve, reinforcing the replicability and scalability of similar systems under adjusted pricing conditions.

Table 8. Estimated annual electricity expenditures and system payback periods under three pricing scenarios.

Scenario	Annual Expense 2019 (USD)	Annual Expense 2024 (USD)	Payback Period (Years)
At USD 0.065/kWh	8772.05	2242.15	15.3
At USD 0.0818/kWh	11,039.29	2821.67	12.1
At USD 0.1559/kWh	21,039.43	5377.72	6.4

summarizes the annual electricity expenditures under three pricing scenarios and presents the resulting payback periods for the 75.6 kWp system. These scenarios were formulated to conduct a sensitivity analysis of the system’s financial viability under varying electricity prices. The first scenario reflects the subsidized electricity tariff applicable to the Universidad de Cuenca, set at USD 0.065 per kWh, as per national law. However, this base tariff does not include ancillary charges typically present in electricity billing, such as public lighting fees, demand charges, administrative costs, waste collection, and firefighter contributions. When these additional items are considered, the effective cost per kWh rises to USD 0.0818.

The third pricing scenario accounts for the real cost of electricity in Ecuador, estimated at USD 0.1559 per kWh. This figure, published by the Agencia de Regulación y Control de Energía y Recursos Naturales no Renovables (ARCERNNR) in its 2023 annual report [11], includes generation infrastructure and operational expenditures.

Under the subsidized rate, the annual electricity expense in 2019 was USD 8772.05, which decreases to USD 2242.15 in 2024 following PV integration. At the adjusted effective rate, annual expenditures decrease from USD 11,039.29 to USD 2821.67. When the real cost of electricity is applied, the reduction is even more pronounced—from USD 21,039.43 to USD 5377.72. Correspondingly, the system’s payback period varies from 15.3 years at the subsidized rate to just 6.4 years under the real-cost scenario. This scenario reinforces the argument that electricity subsidies, while intended to protect access, can obscure the real value of distributed generation and delay investment decisions in microgeneration systems. As shown in this analysis, even a 2 USD/kWh increase in the electricity price leads to a reduction in the payback period of more than three years, indicating the high sensitivity of financial returns to energy cost assumptions. Considering that the university is a public institution, the latter case offers a more realistic reflection of national fiscal savings, since subsidies represent state-funded expenses.

5. Discussion

The outcomes obtained from the economic and energy monitoring of the BIPV system installed in the Faculty of Architecture and Urbanism provide a multifaceted view of performance under real operational conditions. Unlike simulation-based studies, this project combined measured energy balances with institutional billing data, highlighting inconsistencies between invoiced and actual electricity values, as well as fluctuations in daily consumption patterns linked to academic schedules and solar resource variability.

One of the key findings is the divergence between electricity consumption recorded by the utility company and the actual energy delivered to or drawn from the grid. The monitoring equipment revealed that the building frequently exports energy during daylight hours, especially on weekends and during midday breaks. However, this exported energy does not translate directly into economic returns due to the absence of a net metering scheme that fully compensates for energy injected into the grid. Consequently, the actual economic benefit derived from self-consumption exceeds the compensation received for energy exports.

Additionally, the hourly analysis underscores the mismatch between solar production profiles and institutional demand curves. While generation peaks around noon, consumption is typically higher during the early morning and early evening, when PV output is minimal. This mismatch limits the building’s ability to increase self-consumption without complementary technologies such as storage or demand-side management. Although a battery was included in the second stage of the system, its capacity (5 kWh) remains insufficient for reshaping load curves on a daily basis.

The year-on-year comparison of electricity use before and after system deployment confirms a substantial drop in grid dependency—particularly in the months with consistent irradiance and when academic activities align with solar production. However, this reduction is not uniform across all months. During the rainy season, lower PV output combined with increased energy use due to lighting and heating requirements resulted in a higher proportion of energy drawn from the grid. These seasonal dynamics emphasize the importance of long-term monitoring to understand production–consumption interactions beyond annual averages.

The financial analysis, built upon actual invoicing and installation costs, provides further context. While the unit cost per installed watt remained relatively stable across both implementation stages, differences in component configuration, wiring complexity, and interconnection standards introduced cost variances. The inclusion of a hybrid inverter and the need to connect to the medium-voltage network increased capital expenditure in the second stage. Nevertheless, when examining electricity expenditure under various tariff schemes—including subsidized, adjusted, and real cost scenarios—the payback period is compressed considerably under real-cost assumptions. This outcome reveals how generalized energy subsidies can obscure the economic feasibility of self-generation, delaying the adoption of PV systems that would otherwise be viable under transparent cost structures. In this context, shifting from uniform subsidies to targeted support mechanisms could enhance the competitiveness of distributed solar generation, particularly in public institutions with favorable technical conditions but limited incentives.

As observed in Table 8, even a marginal increase in the electricity unit cost—less than USD 0.02 per kWh—results in a payback reduction of over three years. This underscores the sensitivity of PV system feasibility to the reference tariff used. Although the figure of USD 0.1559 per kWh includes distribution and commercialization components, their inclusion is justified when analyzing the public expenditure impact in state-funded facilities. If only the generation cost is considered—approximately USD 0.08 per kWh as reported in the national energy plan—the projected payback period extends to nearly 14 years.

Moreover, none of the evaluated tariffs include environmental externalities such as carbon emissions or land degradation, which ultimately result in public health costs or remediation expenses. This omission further underestimates the potential economic advantages of clean distributed energy systems.

Implications for Energy Subsidy Reform in Ecuador

The monitoring results and financial simulations developed in this study invite a broader reflection on the structure of energy subsidies in Ecuador and their implications for institutional-scale photovoltaic systems. Based on the empirical data obtained and the broader policy context, we propose a set of considerations to guide future reforms aimed at reconciling fiscal responsibility with distributed energy deployment:

• Redesigning Subsidies with Institutional Self-Generation in Mind: The current uniform subsidy scheme discourages investment in PV systems, even in public buildings where solar availability, consumption patterns, and infrastructure compatibility are favorable. Replacing flat-rate subsidies with performance-based incentives—such as verifiable energy offset credits—could align public sector investments with national energy goals while improving transparency in electricity spending.
• Creating Budget-Neutral Mechanisms: Institutional incentives do not necessarily require increased fiscal expenditure. As demonstrated in studies such as Schaffitzel et al. (2020) [37], part of the savings achieved through targeted subsidy reduction can be redirected toward self-generation programs in sectors with high social returns, including education and healthcare. These programs could operate through competitive funds, rebates, or long-term procurement contracts with performance verification [37].
• Incorporating Avoided Infrastructure Costs into Fiscal Planning: Distributed generation reduces the need for expansion in transmission and distribution infrastructure, especially in urban areas with stable demand and aging grid components. Including these avoided costs in the financial assessment of public investments—as suggested by Camino-Mogro and Arias (2024)—can shift the cost–benefit balance in favor of photovoltaic projects, even when using conservative tariff assumptions [38].
• Transitioning to differentiated tariffs based on consumption and user type: Evidence indicates that generalized subsidies benefit higher-income users disproportionately. Implementing tiered tariffs, where public institutions with generation capacity are assigned real-cost or near-cost rates, could encourage the more efficient use of electricity budgets and accelerate the adoption of PV systems. In this context, compensation schemes such as net billing or time-based feed-in payments—evaluated in the residential sector by Benalcázar et al. (2025)—could be adapted for institutional users, balancing cost recovery with fair access and fiscal neutrality [39].
• Ensuring administrative simplicity and policy continuity: For these reforms to succeed, mechanisms must be easy to apply, stable across political cycles, and linked to transparent monitoring. In the case of the BIPV system studied here, even small-scale installations benefit from real-time monitoring, verified generation, and traceable billing—all of which can support a reliable framework for institutional incentives.

These reflections arise from direct experience with a real-world pilot system, reinforced by recent evidence on subsidy reform, fiscal impact, and social acceptability. The authors believe that focusing support not on energy consumption per se but on demonstrated efforts to reduce public energy dependence through generation and efficiency can lead to a more equitable and resilient electricity sector.

In summary, the monitoring and economic analysis performed in this study illustrate the interplay between system design, consumption behavior, tariff structure, and meteorological factors. While the installation meets its annual production targets and reduces public grid dependency, operational challenges—such as load profile mismatches and regulatory constraints—continue to influence its performance. Addressing these challenges requires technical adaptations and regulatory and fiscal reforms that promote investment in microgeneration and optimize public resource allocation. These observations advocate for the integration of energy management strategies that extend beyond PV generation itself and highlight the value of empirical data when evaluating long-term project viability. In this regard, recent studies emphasize the importance of reforming compensation mechanisms and subsidy structures to encourage institutional participation in decentralized generation. Policy approaches such as differentiated tariffs, net billing for public prosumers, and investment-linked incentives offer a path toward more transparent and fiscally coherent support schemes that reflect actual system performance.

6. Conclusions

This study assessed the real-world performance of a building-integrated photovoltaic system installed at the Faculty of Architecture and Urbanism at the University of Cuenca, encompassing both technical and financial dimensions. Unlike purely theoretical evaluations, the analysis was based on metered data collected over an extended monitoring period, which allowed for a close examination of hourly energy balances, institutional consumption patterns, and deviations between billed and actual values.

One key outcome was the identification of discrepancies between recorded generation, grid injection, and utility billing. Despite frequent electricity exports during midday hours—especially on weekends and during academic recess periods—these exports are not economically compensated under the current regulatory framework. As a result, most financial returns are obtained through direct self-consumption, a condition that underscores the importance of aligning load profiles with generation patterns.

The addition of a second-stage system increased the overall capacity to 75.6 kWp, enhancing the building’s annual generation. However, daily mismatches between solar availability and peak institutional demand, particularly during early mornings and evenings, limit the share of self-consumption. The integration of a 5 kWh battery during the expansion stage contributed marginally to offset this issue, but its storage capacity is insufficient for addressing demand shifts on a broader scale. Nonetheless, the battery subsystem supports the uninterrupted operation of a low-consumption design studio classroom, effectively establishing a self-powered space within the academic facility. This configuration serves as a test bench for evaluating energy autonomy, demand-side efficiency, and decentralized load control strategies, providing a replicable model that could be scaled to other campus environments in future stages.

Financially, the installed system demonstrates economic viability over time, even under a subsidized tariff structure. When evaluated using actual electricity production costs reported at the national level, the system recovers its investment in less than seven years. This finding gains relevance in the context of public institutions, where reduced energy purchases translate into lowered public expenditure and not just operational savings. Under real retail price conditions, the investment recovery period would shorten to approximately 6.4 years, demonstrating that feed-in tariff schemes are not essential to support microgeneration. If complemented by technology subsidies or financial incentives, adoption would likely increase, especially among users for whom the initial capital cost remains the primary barrier.

Among the main constraints identified are (i) the limited capacity of the battery system to reshape consumption curves, (ii) the absence of a comprehensive net metering policy that accounts for energy exports, and (iii) the seasonal variability of solar irradiance, which introduces fluctuations in monthly savings and grid dependency. Moreover, the system’s performance relies heavily on behavioral and scheduling factors, including occupancy patterns and academic calendar events. Although no inverter tripping or operational downtimes were recorded during the monitoring period, partial shading remains a potential limitation under altered rooftop conditions or long-term infrastructure changes.

Future studies could focus on exploring demand-side management techniques, such as adaptive load shifting, to increase the proportion of self-consumed energy. Likewise, evaluating the impact of a larger storage system under similar operational conditions could provide further insight into optimizing energy use within educational institutions. In addition, expanding the monitoring campaign across multiple buildings would allow for a broader evaluation of photovoltaic integration at the campus scale. There is also a need for deeper research on political and regulatory aspects, considering scenarios that account for evolving electricity pricing models, investment in grid infrastructure, and the integration of distributed energy under different fiscal frameworks.

Possible directions include assessing how technology cost reductions—driven by local market expansion and industrial competition—could further improve the financial outlook of small-scale PV systems. Under a more competitive and accessible market structure, the adoption of these technologies would likely become more attractive, especially in public institutions with constrained budgets but high technical potential.

Additionally, further research is recommended to evaluate the thermal performance of photovoltaic modules under equatorial high-altitude conditions. In these environments, lower atmospheric pressure and increased ultraviolet exposure may affect module behavior differently than in lowland tropical areas. On-site temperature measurements would be necessary to quantify potential performance deviations and confirm simulation-based estimations.

The experience documented in this study offers a practical reference for academic institutions aiming to implement distributed energy systems in comparable climatic and urban settings. While energy self-sufficiency remains constrained by technical and regulatory boundaries, monitoring data reveal a measurable reduction in electricity demand from the public grid, confirming the system’s ability to meet a meaningful portion of the building’s annual electricity needs. These findings support broader discussions on subsidy reform and institutional incentives, suggesting that well-structured compensation mechanisms and targeted fiscal support could accelerate the adoption of photovoltaic systems in the public sector without requiring generalized financial transfers.