Skip to content
RT
Skip to main content
Provenance:Dataset:v2025.1Run:1 Jan 2025
PhD Dissertation · 2025~5 min read

RL Optimisation Results

Part of Hybrid AI for Turkey's 2050 Energy Transition — Seyit Önderol, supervised by Prof. Dr. Faik Bilgili

In this section the performance of the RL system dimension along four dimensions is presented, viz. (1) achievement of 2050 target and success of the multi-objective optimization (2) technology portfolio evolution and capacity expansion trajectory (2025-2050), (3) regional distribution and spatial optimization, (4) comparison of algorithm comprehensiveness (MAPPO vs. 12 benchmarks). The outcomes indicate that Turkey's goals for net-zero have been successfully achieved with 97.34% renewable share (target: 90%), 14.30 MtCO₂ emissions (90.6% reduction from the baseline of 152 MtCO₂, target: 87% reduction), total investment of $226.79B, capacity of 220 GW, 99.2% reliability. Thus, the attainment of these results confirms that the long-term energy transition planning of Turkey's power generation system can be performed with the use of the hybrid AI framework.

Target Achievement and Multi-Objective Performance

Primary objectives: The Net-Zero Climate Strategy of Turkey for 2053 has set three binding targets for 2050. (1) At least 90% share of renewable electricity. (2) At least 87% reduction of CO₂ emission across the sector relative to 2025 baseline (172 MtCO₂). (3) Reliable electricity supply (at least 99% adequacy). Secondary objectives: To ensure regional equity and maximize technology diversity, minimize total system cost.

MAPPO-optimized results (final year 2050):

-------------------------------------------------------------------------------------------------
Renewable Share97.34%≥90%+7.82 pp✅ Exceeded
CO₂ Emissions14.30 Mt≤22.4 Mt (87% reduction)-8.10 Mt✅ Exceeded
Total Capacity220 GW≥218 GW+2 GW✅ Met
Reliability (Demand Met)99.2%≥99%+0.2 pp✅ Met
Total Investment (2024–2050)$226.79BN/A (minimize)--
Renewable Capacity215.37 GWN/A--
Fossil Capacity0.00 GWN/A--
Nuclear Capacity4.8 GW4.8 GW (Akkuyu)0 GW✅ Met

: MAPPO 2050 Results: Target Achievement and Multi-Objective Performance {#tbl:mappo-2050-target-achievement}

[Figure]

*2050 target achievement visualization: Multi-dimensional performance across capacity (220 GW), renewable share (97.34%), emission reduction (90.6%), and investment ($226.79B). All targets exceeded or met.*

Figure Figure shows a multi-dimensional perspective on target achievement across key metrics. All targets were met or exceeded: capacity (220 GW, target ≥218 GW); renewable share (97.34%, target 90%); emission reduction (90.6%, target 87%), demonstrating comprehensive success.

[Figure]

*2050 technology portfolio distribution: Balanced strategy with solar (63.48 GW, 35.1%), wind (58.52 GW, 32.4%), hydro (57.09 GW, 31.6%), geothermal (7.98 GW, 4.4%), biomass (9.96 GW, 5.5%), nuclear (4.8 GW, 2.7%), and no fossil backup (0.00 GW, 0.0%).*

The 2050 technology portfolio as the result of a balanced strategy indicates 63.48 GW (35.1%) solar, 58.52 GW (32.4%) wind, 57.09 GW (31.6%) hydro, 7.98 GW (4.4%) geothermal, 9.96 GW (5.5%) biomass, and 4.8 GW (2.7%) nuclear. There is a total fossil 0.00 GW (0.0%) for the 2050 energy portfolio.

[Figure]

*2050 capacity summary: Comprehensive capacity summary showing total capacity (220 GW), renewable capacity (215.37 GW), and technology breakdown. Summary includes capacity factors, generation potential, and regional distribution.*

Figure Figure is a capacity summary for 2050 depicting the total capacity as 220 GW, renewable capacity as 215.37 GW, a comprehensive technology breakdown showing capacity factor, generation potential, regional distribution and more.

[Figure]

*2050 final portfolio comparison algorithms: Side-by-side comparison of 2050 portfolios across 13 algorithms. MAPPO shows balanced portfolio, other algorithms show different technology preferences. Comparison reveals algorithm-specific optimization strategies.*

An example of side-by-side analysis comparing the 2050 portfolio across the 13 algorithms. It appears that each algorithm has a particular strategy to optimize its holdings across technology and fuels.

[Figure]

*2050 final portfolio density bubble: Bubble chart showing 2050 portfolio with bubble size representing capacity (GW) and color representing technology category. Large bubbles for solar, wind, and hydro indicate dominant technologies. Bubble visualization enables quick portfolio assessment.*

The 2050 portfolio is represented in the bubble chart above, with bubble size denoting capacity (GW) and bubble color denoting technology category, easy to interpret.

[Figure]

*2050 final portfolio heatmap algorithms: Heatmap showing technology capacity (GW) across algorithms in 2050. Darker colors indicate higher capacity. MAPPO shows balanced distribution, other algorithms show concentration in specific technologies. Heatmap reveals algorithm preferences.*

Figure Figure is a heatmap showing technology capacity (GW) for each algorithm in 2050. The results indicate algorithm-specific optimal solutions revealed by the algorithms used in this study.

[Figure]

📎 Cite this — RL Optimisation Results

Önderol, S. (2025). RL Optimisation Results. In Electricity cost optimization in Turkey's transition to renewable energy (Chapter). Erciyes University.
RL Optimisation Results — Thesis Viewer | Renewable Transition