Exergy-Based Modeling of Wind and Solar Energy Systems under the Climatic Conditions of Istanbul
The demand for renewable energy sources is steadily increasing. However, these sources present significant challenges, the most critical of which is intermittency. This issue is particularly prominent in wind and solar energy systems, which are the most widely utilized among renewable technologies. One of the most effective (i.e., optimal) solutions to mitigate intermittency is the integration of these two sources into hybrid systems. By selecting suitable locations, the temporal and spatial variability of renewable energy generation can be minimized. In this study, a hybrid energy system capable of meeting the full energy demand of an average household has been installed in the Meteorological Park of Istanbul Technical University. Meteorological parameters obtained from this system were modeled with respect to spatial and temporal variations. In the modeling process, the point cumulative semivariogram method was employed for spatial representation of wind speed. Furthermore, wind energy and exergy analyses are to be conducted at eight designated measurement points. Similar modeling approaches have also been applied to solar radiation. As is well known, maps play a crucial role in spatial modeling. In recent years, the Weather Research and Forecasting (WRF) model has emerged as one of the most widely used tools in numerical weather prediction. In this thesis, the entire study area, encompassing Istanbul and its surroundings, was divided into grid cells of 3 km × 3 km. Based on these grids, spatial and temporal predictions were carried out. The second stage of the study involves calculating energy losses occurring in both wind and photovoltaic (PV) systems. To identify the locations of these losses within the system and quantify them thermodynamically, exergy analysis methods—which have gained increasing importance in recent years—were employed. In wind energy systems, the location and meteorological conditions play a significant role in loss mechanisms, while in photovoltaic systems, thermal losses are dominant. Therefore, meteorological and exergy-relevant parameters affecting such systems must be analyzed both spatially and temporally. Following the identification of losses, exergy efficiency models were developed for each system component and for the system as a whole. Based on these models, the energy losses were quantified, and recommendations for minimizing them were proposed. In summary, this thesis provides a comprehensive examination of wind–solar hybrid systems and assesses their applicability under the specific climatic conditions of Istanbul. In other words, a complete model and computational framework covering all critical aspects of hybrid system performance has been developed.
MUSTAFA KEMAL KAYMAK
Mitigation of Evaporation Losses Through the Use of Floating Photovoltaic Systems
In Turkey, total free-surface evaporation and transpiration losses are estimated at 274 billion cubic meters, a figure that is nearly 2.5 times greater than the country’s annual usable water potential, which stands at 110 billion cubic meters. Although reducing water losses due to plant transpiration is not yet technologically feasible, evaporation alone accounts for approximately 90% of total surface water losses. Therefore, reducing evaporation is of critical importance for sustainable water resource management. Evaporation is primarily driven by fluctuations in air temperature. According to IPCC climate scenarios, if current trends continue, the projected temperature increase for the mid-latitude regions, including Turkey, is 4–6 °C by the year 2070. In this study, a field experiment was conducted to evaluate the potential of Floating Photovoltaic Panels (FVPs)—a technology that has gained increasing global attention over the past decade—as a means to reduce evaporation losses. Two Class A evaporation pans, each filled with 20 cm of water, were used. One of the pans (T-PV) was partially covered (50% surface area) with a Floating PV Panel Platform (FPVP), while the other (T-NPV) was left uncovered to represent the control condition. The system operated under natural atmospheric conditions for 30 days, during which evaporation losses were monitored and compared with hydrometeorological parameters. Additionally, the actual electrical output of the 70 W multicrystalline PV module was measured via a custom-designed circuit and recorded continuously. These results were compared against the manufacturer’s lab-tested efficiency values. In terms of evaporation, 38.7 mm of water evaporated from the T-PV setup, while the T-NPV setup exhibited a 94.9 mm loss. The difference of 56.2 mm corresponds to a 59.2% reduction in evaporation due to the floating PV cover. The average electrical power output from the floating PV module was measured at 20.67 W. To further assess the effect of mounting surface on PV performance, a comparable PV module was installed on concrete blocks in two configurations: first on bare soil, and then over the water-filled pan. Measurements taken over 5 days for each setup revealed that PV modules on soil produced 10.8% more power than those on water. This result suggests that thermal interaction between the PV module and the water surface may negatively impact power generation efficiency. Finally, three water samples from each evaporation pan were analyzed microbiologically. The results were notable: in the T-PV setup, E. coli colony counts were measured as 9200, 14000, and 12800 CFU/100 mL, while in the T-NPV, the values were 400, 120, and 2300 CFU/100 mL, respectively. These findings highlight that the material composition of the floating platform is crucial, as it should not exert negative ecological pressure on the aquatic environment in practical applications.
MEHMET SEREN KORKMAZ
Seasonal Weather Prediction Based on Fuzzy Logic: A Case Study for Istanbul
What will the weather be like tomorrow? How will it affect us? What should we expect? Humanity has been seeking answers to these questions for centuries. Weather conditions, regardless of class or species, are among the most influential phenomena affecting all forms of life. As the state of the atmosphere—both in the short and long term—directly shapes human activity and planning, it has consistently remained a subject of great interest and necessity. With the advancement of technology, efforts have been made to improve the accuracy of weather forecasts using observational data, and numerous models and software have been developed in this direction. However, the inherently dynamic and chaotic structure of the atmosphere, combined with the high number of interrelated meteorological variables, makes it challenging to define and predict atmospheric behavior accurately. This complexity also makes it difficult to establish reliable long-term climatological patterns, which are typically defined as the statistical averages of meteorological elements and atmospheric events over periods of at least 30 years. Seasonal weather forecasting has become an increasingly important and timely area of study, especially with growing urbanization and population density. Although extreme events may occur within any given season, seasonal forecasts differ from short-term weather predictions in that they focus on broad statistical summaries without abrupt changes. Achieving high-accuracy seasonal forecasts can significantly support early warning systems and the preemptive planning of critical sectors such as water resource management, renewable energy generation, agricultural productivity, transportation, tourism, maritime operations, and the mitigation of weather-related disasters, thereby reducing the risk of economic and human losses. Due to the chaotic nature of the atmosphere, expressing weather behavior through deterministic formulas and mathematical equations is inherently difficult. In such complex systems, Fuzzy Logic (FL) has emerged as a valuable modeling tool. Fuzzy Logic is already commonly used across many scientific fields and is particularly well-suited to meteorological applications. This is due in part to the qualitative nature of terms such as cold, cool, mild, warm, hot, very hot, which are inherently fuzzy in nature and cannot be precisely quantified. In this study, the Adaptive Neuro-Fuzzy Inference System (ANFIS) was employed as a modeling framework. Within the ANFIS environment, Sugeno-type fuzzy inference models were constructed using two-input, one-output and three-input, one-output architectures. The input data consisted of long-term temperature and relative humidity records, while precipitation was modeled as the output. Data from the Kandilli Observatory in Istanbul, covering the period 1960 to 2016 (a total of 57 years), were used in the analysis. These datasets were aggregated into monthly totals, and a separate ANFIS model was trained and tested for each month using the corresponding data. The modeling results indicated high predictive performance, particularly in monthly-scale forecasts. The model was applied to both training and test datasets, and performance was evaluated using relative error metrics, which confirmed the model's effectiveness in capturing seasonal precipitation patterns.
NİDA DOĞAN ÇİFTÇİ
The steadily increasing global energy demand is an undeniable reality. Initially, fossil fuels were preferred and widely utilized to meet the demand for electricity. However, the oil crisis of the early 1970s, followed by the intensifying effects of global warming in the 1990s due to excessive fossil fuel consumption, triggered a significant shift toward renewable and clean energy sources. This transition has led to the development of hybrid systems that combine wind and solar energy technologies—two of the most accessible and widely implemented renewable energy sources. In this study, wind and solar energy potentials were analyzed based on measurement data collected from various regions in and around Istanbul. The results included monthly and hourly generation profiles for wind, solar, and hybrid systems. Additionally, annual revenue values were calculated based on the wholesale electricity prices, and corresponding monthly-hourly income profiles were established. A total of five locations—Ömerli, Ormanlı, Bağırganlı, Melen Road, and Kıyıköy—were examined in detail for wind, solar, hybrid energy production, and revenue analyses. The same brand and model of wind turbines, inverters, and photovoltaic panels were used for consistency across the analysis. For each selected region, the capacity factors of wind–solar hybrid systems were determined by evaluating the region-specific wind and solar potentials. Wind potential analyses were conducted using AWS Openwind, which provided hourly wind generation estimates. Similarly, solar potential was analyzed using PVsyst, yielding hourly PV output data. These generation results were converted into hourly data sets covering one full year, enabling a comprehensive hybrid performance assessment. In Ömerli (on the Anatolian side of Istanbul), wind and solar capacity factors calculated through Openwind and PVsyst were 26% and 23%, respectively. The hybrid capacity factor, based on hourly data, was determined as 24%. Scatter plots of hourly average values for each month and their correlations with other months were analyzed to determine the degree of similarity in production patterns. The highest correlation was observed between July and August. In Ormanlı (on the European side of Istanbul), the wind and solar capacity factors were calculated as 34% and 23%, respectively, with a hybrid capacity factor of 28%. For this location, the months with the strongest correlation in hybrid output were found to be June and July. The monthly contributions of wind and solar energy sources to the total hybrid system output were also analyzed. Results showed that during months when wind contribution was high, solar contribution tended to be low, and vice versa. In some cases, both sources contributed nearly equally. These findings demonstrate that wind and solar resources complement each other, enhancing the reliability of hybrid systems. To further model and evaluate the system performance, the Adaptive Neuro-Fuzzy Inference System (ANFIS) was employed. Using fuzzy logic, relationships between the input datasets and the average values were assessed and modeled. ANFIS was used to estimate hourly energy production profiles for wind, solar, and hybrid systems throughout the year. The relationship between real data and ANFIS-estimated values was analyzed, and the model's predictive capability was verified. The study concluded that ANFIS can effectively capture the relationship between wind, solar, and hybrid energy production, and is capable of accurately estimating next-month hybrid output based on current production data. This confirms the robustness of ANFIS-based modeling in hybrid renewable energy systems under Istanbul's climatic conditions.
ESRA GÜN
In recent years, the increasing global population has led to a significant rise in energy consumption, which in turn has intensified the observable impacts of global climate change. Rising global average temperatures have accelerated evaporation from freshwater sources, creating a potential risk of failing to meet the demand for potable water. As freshwater resources are finite, the need to protect them has become increasingly critical worldwide. One of the major contributors to freshwater loss is evaporation from large open water surfaces, which is influenced by various meteorological parameters. Controlling or minimizing evaporation from such reservoirs, particularly those used for drinking water, plays a vital role in preserving freshwater supplies. In response to this challenge, several mitigation strategies have been explored globally. Among these, Floating Photovoltaic Systems (FPVS) have emerged as a promising solution. These systems are designed to partially cover the water surface with photovoltaic (PV) panels mounted on floating platforms, thereby reducing direct solar radiation exposure and minimizing evaporation, while simultaneously generating clean electrical energy. In this study, the objective is to quantify the wind and wave loads acting on a test FPVS installed on the surface of Lake Büyükçekmece, a major drinking water reservoir in Istanbul. The study aims to assess the potential of FPVS in both reducing evaporation and generating electricity, while also providing valuable insights to support the design and optimization of future systems. The wind and wave interaction analysis of the FPVS system was carried out using Computational Fluid Dynamics (CFD) simulations in ANSYS Fluent®. Various wind speeds were considered to compute the associated wind and wave loads. The platform was modeled in two dimensions, and a finite element mesh comprising 196,559 triangular elements was generated to conduct the analysis. Four boundary conditions were defined to represent the ambient environment: air inlet, air outlet, atmospheric boundary layer, and the platform surface. Given that the FPVS is placed on the water surface, it is directly affected by wind and hydrodynamic forces. To simulate these interactions, Multiphase Flow and Open Channel Flow models were utilized within the CFD environment to represent wave generation and surface interactions. The k–ε Realizable turbulence model was employed to characterize turbulent flow behavior. According to the results, high wind speeds induced wave deformation and accumulation, causing flow streamlines to compress and resulting in regions of high-velocity air flow. Wind-driven waves with increased kinetic energy exerted stronger impact forces on the platform surface. The turbulent kinetic energy was observed to be highest near the front-facing panels of the platform and gradually decreased toward the rear. Eddy formations, caused by turbulence, were more concentrated in the lake water between the platform’s front face and the inlet boundary. Under a wind speed of 30 m/s, the maximum instantaneous total force exerted on the platform due to wind and wave loads was calculated as 38,677 N, while the maximum negative force was –2,345 N. For the PV panel surfaces mounted on the platform, the maximum instantaneous total pressure was found to be 7,654 N, with a negative reaction force of –1,537 N. These findings emphasize that the front surfaces of the FPVS are exposed to significantly greater forces due to direct wind and wave impact, highlighting the critical importance of improving structural design in this region for enhanced system durability and performance.
ALİ OSMAN MUT
Throughout history, humans have observed and sought to understand natural phenomena. While many of these events have been explained through scientific inquiry, others—particularly in earlier times—were attributed to supernatural causes. Phenomena such as lightning and thunder, which were once interpreted through mythology and deities in various cultures, have persisted as subjects of human curiosity and now receive detailed physical explanations. These phenomena, which once influenced societies even to the extent of shaping the names of days of the week, continue to impact modern life across a wide range of sectors including energy, agriculture, aviation, logistics, urban planning, tourism, mining, and insurance, as well as everyday activities. The increasing demand for accurate and operational lightning forecasting has led to the integration of numerical weather prediction models and statistical techniques. In this thesis, lightning and thunderstorm events over Istanbul were examined using such approaches. The analysis was based on lightning observation data obtained from the Turkish State Meteorological Service (TSMS). Initially, the monthly and seasonal distribution of lightning and thunderstorm observations for the years 2018, 2019, and 2020 was analyzed. These full-year datasets were processed to determine the number of events per month, frequency distributions, and seasonal observation percentages. To develop predictive capabilities, further numerical studies were conducted. Given that lightning is closely associated with convective atmospheric processes, accurately modeling convection is essential. The WRF (Weather Research and Forecasting) regional atmospheric model was employed for this purpose. Initial and boundary condition data for WRF simulations were obtained from the GFS global model to allow forecasting. Microphysical parameterizations previously used in lightning prediction studies were selected to better represent convective processes. Two nested domains were defined to improve model resolution and performance. A coarse outer domain covering an area larger than 1000×1000 km was used for initial spin-up, while a fine inner domain was activated three hours prior to peak lightning activity in each case study. The fine domain, used for detailed analysis, was constrained to lie within 26 km of the observation network’s spatial coverage. Simulation time steps were set based on spatial resolution: 15 seconds for the inner domain (3 km resolution) and 45 seconds for the outer domain (9 km resolution). Output was recorded at 15-minute intervals, ensuring consistent temporal alignment across case studies. Input data from GFS were assimilated at 3-hour intervals, allowing WRF to develop independent model dynamics with minimal external influence. Observation data from the Lightning Detection and Tracking System (LDTS) were mapped onto grid points within the fine-resolution domain. For each time step, lightning strike frequency was assigned to the corresponding grid point. After defining the model structure and parameters, the study proceeded with the analytical methods. Historically, lightning forecasting has often relied on convective indices and instability parameters. In this study, two additional methods based on the physical mechanisms of lightning formation were applied: the Lightning Forecast Algorithm (LFA) and the Lightning Potential Index (LPI). Given the regional focus on Istanbul, LFA was selected due to its adaptability and stronger local performance. While LPI is standardized, it lacks regional calibration and can show varying effectiveness across different areas. The LFA is based on two microphysical variables: graupel flux and cloud ice content. These variables are calculated independently for each time step. For calibration, the maximum graupel flux, maximum cloud ice content, and maximum observed lightning strikes are recorded. Scatter plots between observations and each variable are constructed, and linear regression equations are derived using the least squares method. A weighted combination of the two variables—0.95 for graupel flux and 0.05 for cloud ice—is used to form the final lightning prediction equation, consistent with previous literature. Recognizing the potential regional variability in optimal weightings, a second approach was tested using the non-negative least squares (NNLS) method. This technique allows for an objective estimation of the weighting coefficients by fitting a single multivariate regression model. The results showed that the NNLS approach yielded predictions very similar to the traditional LFA method, confirming its validity. Additionally, the analyses demonstrated the importance of higher spatial resolution in capturing lightning activity. The overall findings underline that lightning prediction is a complex process that requires integration of various disciplines: hazardous weather meteorology, micrometeorology, synoptic meteorology, remote sensing, numerical modeling, mesoscale meteorology, cloud physics, programming (e.g., shell scripting), and advanced statistics. Thus, successful lightning forecasting demands comprehensive data processing and interdisciplinary expertise.
KERİM ATİLLA KORKMAZ
Greenhouse gases (GHGs) present in the atmosphere are of great importance for the Earth’s climate system. In the absence of these gases, the Earth would experience scorching daytime temperatures and extremely cold nights, rendering the planet less habitable. However, due to natural processes and the presence of GHGs in the atmosphere, Earth maintains a livable average temperature of approximately 15°C. Since the Industrial Revolution, atmospheric GHG concentrations have increased dramatically. While the pre-industrial level of carbon dioxide (CO₂) was around 285 ppm, this figure reached 417 ppm in 2022. The energy sector is the primary contributor to the increase in anthropogenic GHG emissions, followed by agriculture, forestry and land use (AFOLU), industry, transportation, and construction. As the impacts of climate change become more pronounced, scientific studies and policy efforts addressing this issue have intensified. The first major international treaty to address climate change was the United Nations Framework Convention on Climate Change (UNFCCC), which aimed to reduce anthropogenic influence on the climate system and lower GHG concentrations. Although the convention did not impose legally binding commitments, the subsequent Kyoto Protocol introduced quantified emission reduction obligations. The most recent Paris Agreement established, for the first time, a long-term temperature goal: to limit global warming to well below 2°C above pre-industrial levels, ideally keeping it to 1.5°C. Global warming has triggered significant changes on a planetary scale, including rising land and ocean surface temperatures, sea level rise, glacier retreat, increased frequency of extreme weather events, and loss of biodiversity. These developments underscore that the climate crisis caused by the increase in GHGs is a shared global concern, prompting international cooperation. Countries, institutions, and individuals around the world have begun to quantify their atmospheric impact, typically measured as carbon dioxide equivalent (CO₂e)—commonly referred to as the carbon footprint (CF). Tracking and calculating carbon emissions—whether at global, city, institutional, or individual scales—is the basis for developing emission reduction strategies and action plans. A carbon footprint is defined as the total amount of GHGs emitted into the atmosphere, expressed in CO₂ equivalents, as a result of human activities. CF is typically categorized into two components: Primary footprint: direct emissions from fossil fuel consumption, such as transportation and energy use. Secondary footprint: indirect emissions from the production, use, and disposal of goods throughout their lifecycle. Over time, increasing CF poses both regional and global threats to ecosystems and human well-being. Therefore, accurate quantification is critical for taking preventive actions. According to global data, total GHG emissions reached 34.807 MtCO₂e in 2020, while Turkey’s emissions increased from 220 MtCO₂e in 1990 to 524 MtCO₂e in 2020. This rise is largely due to population growth, economic development, transportation expansion, and increased consumption. Although not legally bound under international frameworks, Turkey has participated in Voluntary Carbon Markets and initiated national and local-scale emission calculations. These calculations, especially at the city level, form the backbone of Turkey’s climate change mitigation efforts. Cities are now recognized as major sources of GHG emissions, as urban areas surpass rural ones in population, energy demand, industrial activity, transportation, and waste generation—all contributing to a higher CF. For this reason, estimating and comparing city-scale GHG emissions is essential. Various international guidelines have been developed for such analyses. In this thesis, emissions were calculated based on the IPCC 2006 Guidelines, which propose three methodological tiers: Tier 1: a basic method relying on default emission factors (EFs) and fuel consumption data. Tier 2: uses country-specific EFs for improved accuracy. Tier 3: the most detailed method, requiring data on combustion technologies, operating conditions, emission control systems, and fuel properties. Due to limited data availability for Tier 3, this study applies both Tier 1 and Tier 2 methods. The focus is on Ankara, Istanbul, and Izmir, as they consistently rank among the top three Turkish cities in terms of population, GDP, vehicle use, heating fuel consumption, industrial output, and waste generation. These cities were selected due to their differing geographic, climatic, and socioeconomic profiles. Emissions for the years 2010–2020 were calculated and compared. The following sectors were analyzed per IPCC 2006 classifications: stationary combustion, mobile combustion, enteric fermentation, solid waste disposal, and wastewater treatment. Emissions of CO₂, methane (CH₄), and nitrous oxide (N₂O) were estimated. Based on IPCC’s Fifth Assessment Report, CH₄ and N₂O have global warming potentials of 28 and 265, respectively; thus, all emissions were converted to CO₂e for comparability. In the stationary combustion sector: Residential consumption of natural gas, coal, fuel oil, and electricity; Commercial and public sector electricity and gas usage; Industrial and energy sector electricity/gas use; Public lighting and agricultural irrigation emissions were included. Key results: Ankara: from 16,000 ktCO₂e in 2015 to 20,000 ktCO₂e in 2020 Istanbul: from 43,000 ktCO₂e in 2015 to 45,000 ktCO₂e in 2020 Izmir: from 22,300 ktCO₂e in 2015 to 22,800 ktCO₂e in 2020 Mobile combustion emissions were also evaluated: Ankara: from 5,300 ktCO₂e in 2010 to 8,200 ktCO₂e in 2020 Istanbul: from 11,500 ktCO₂e to 14,300 ktCO₂e Izmir: from 4,000 ktCO₂e to 5,500 ktCO₂e Tier 2 methods yielded lower emissions for road transport. Air transport emissions were calculated using takeoff/landing data, with Istanbul showing the highest figures. Enteric fermentation emissions were dominated by cattle, with Izmir recording the highest and Istanbul the lowest values. Solid waste disposal emissions were calculated based on waste quantities and composition within city boundaries, primarily generating CH₄, which in many cases is recovered for electricity generation. Istanbul, having the largest waste volume, had the highest emissions. Emissions from composting facilities in Istanbul were also assessed. Forest carbon storage for 2010 was estimated using forest management plans, revealing Izmir as the leading city in carbon sequestration. Finally, Monte Carlo simulation was used to perform uncertainty analysis for activity data and emission factors, ensuring calculated emission estimates fall within acceptable confidence intervals. Some municipalities that are party to the Covenant of Mayors have begun developing inventories and mitigation plans. Since inventory preparation requires collecting data from multiple institutions, streamlining data acquisition processes in Turkey would facilitate broader and more consistent emission reporting nationwide.
SENA ECEM YAKUT ŞEVİK
Since the beginning of human existence, individuals have acted on the instinct to meet their basic needs: shelter, clothing, and nourishment. From prehistoric times, the tendency of humans to live communally and fulfill their needs through cooperation led to the formation of societies and a shift toward production-oriented consciousness. From the discovery of fire onward, humanity has consistently required energy. Advancing civilizations concurrently expanded their domains of production. Following the Industrial Revolution and the technological developments that accelerated through the 20th century, new concepts and technologies emerged to comprehend energy and to meet growing energy demands. As global population levels and civilizational demands increased, traditional methods struggled to satisfy energy needs. The rise of energy-intensive technological activities and the discovery of new energy sources have fundamentally transformed global dynamics. Energy generation resources are now broadly classified into fossil-based and renewable types. The growing demand for energy has also necessitated interdisciplinary studies to enhance the efficiency of production systems. Meteorology has emerged not only as a critical science for forecasting energy demand but also as an integral factor in planning fossil-based energy generation. With the evolution of energy markets, the importance of demand forecasting has led to the establishment of national and international energy markets. Beginning in 2016, market liberalization in Turkey has driven energy sector stakeholders to improve the accuracy of generation planning and demand forecasting. Consequently, the significance of meteorology in the energy sector has become increasingly apparent. For example, wind direction and speed are crucial for wind turbines, solar radiation and ambient temperature for solar power plants, and precipitation and snowmelt for hydropower plants. This thesis focuses on the dynamics of Turkey’s energy market and the influence of meteorological conditions on various generation technologies. Scenario development based on meteorological classifications—wet, dry, and climatological normals—was conducted using anomaly maps from ECMWF, GFS, Meteociel, and EURO-4. The projected impacts of these meteorological conditions on both supply and demand were analyzed. Capacity factor profiles for wind, solar, and hydropower plants were assigned based on historical weighted production averages under varying meteorological scenarios. The case study model was constructed using the PLEXOS integrated energy modeling software. The primary objective of this study is to evaluate how source-based generation facilities in Turkey respond to seasonal and meteorological variations, and how these responses influence future national energy generation projections. For instance, during the hydrologically wet year of 2019, Turkey's run-of-river hydropower plants contributed substantially to electricity generation. Storage hydropower plants conserved water to meet elevated summer demand—mainly due to tourism—thereby minimizing the reliance on natural gas plants. The simulated core objective of this research is to identify how much each generation type might contribute under future meteorological conditions and to establish the relationship between such conditions and generation performance. The study utilized historical production and demand data, installed capacities and technical specifications classified by generation source, and macroeconomic indicators projected by the International Monetary Fund (IMF) through 2031—including GDP, CPI, PPI, minimum wage index, population, and foreign exchange rates (USD and EUR). This research integrates atmospheric sciences and energy meteorology to assess the long-term influence of weather on source-based generation in Turkey. Data analysis and anomaly interpretation were conducted using meteorological modeling within PLEXOS. Renewable generation profiles were derived based on expected meteorological patterns, and transparency data from EPİAŞ, historical records from TEİAŞ, and data from the Load Dispatch Information System were incorporated. Scenario modeling encompassed meteorological projections (normal, wet, dry) in conjunction with demand scenarios (low, base, high). A mixed-integer programming approach was used for model construction. The resulting forecasts were validated against observed historical data. To improve model accuracy, normal distribution-based error corrections were applied, significantly increasing the success rate of the projections. Ultimately, this thesis demonstrates the significant role of meteorological conditions on Turkey’s energy generation by source type. Model results from PLEXOS were compared to TEİAŞ’s 10-year supply and demand forecasts, revealing consistent outcomes that respected supply security considerations. The importance of manual anomaly map interpretation and the ability of meteorologically trained engineers to engage with stochastic simulation tools were also affirmed. This work highlights the growing need for meteorology-informed strategies in energy planning and policy-making.
BAHADIR KARABEKİROĞLU
Since the earliest periods of human history, the need for energy has been a constant driving force behind societal and technological development. With the advent of the Industrial Revolution and the rapid acceleration of technological progress in the 20th century, energy demand has increased significantly, prompting the search for more efficient and diverse energy sources. Today, energy production is broadly categorized into fossil fuels and renewable sources. The efficiency and sustainability of these sources have become major areas of scientific research. Meteorology plays a crucial role not only in forecasting energy demand but also in optimizing energy supply—especially for renewable energy systems such as wind, solar, and hydroelectric power. In Turkey, the liberalization of the energy market since 2016 has emphasized the need for accurate, meteorology-informed forecasting in both energy generation and consumption. Meteorological parameters—such as wind speed and direction, solar radiation, precipitation, and snowmelt—directly affect the operational efficiency of renewable energy facilities. This thesis explores the influence of meteorological scenarios (dry, baseline, and wet conditions) on Turkey’s energy generation capacity across various resource types from 2022 to 2031. The analysis integrates meteorological datasets (e.g., ECMWF, GFS, Meteociel, EURO-4 anomaly maps) and simulates different climate scenarios using the PLEXOS energy modeling software. Based on historical meteorological and energy production data, production profiles were assigned to each energy source under the projected meteorological conditions. The study uses a case-study methodology and applies weighted average production trends to evaluate how seasonal climate variations affect electricity generation. Key input data includes historical generation and demand statistics, technical specifications of power plants, and economic indicators provided by the International Monetary Fund (IMF), such as gross domestic product, inflation indices, minimum wage trends, population forecasts, and currency exchange rates. Scenario-based modeling was conducted in PLEXOS through mixed-integer programming, with verification through comparison to real-world data and the 10-Year Supply and Demand Forecasts published by TEİAŞ (Turkish Electricity Transmission Corporation). Forecasting accuracy was improved through statistical calibration using normal distribution functions. The study finds that the PLEXOS model provides robust projections for energy supply security under different meteorological conditions. Results indicate that in hydrologically favorable years (e.g., 2019), hydropower plants can dominate energy supply, reducing reliance on fossil fuels such as natural gas. This highlights the potential of meteorological forecasting to inform strategic energy planning and policy development. The thesis emphasizes the interdisciplinary nature of energy meteorology, underscoring the critical role of meteorologists in managing data-intensive forecasting systems and in interpreting anomaly maps. The findings demonstrate the value of incorporating meteorological expertise into energy sector planning and decision-making.
RUKİYE AYBÜKE AYDEMİR
English 
Turkish