Addressing extreme weather events for the renewable power-water-heating sectors in Neom, Saudi Arabia

Saudi Arabia's Vision 2030 aims for at least 50% renewable electricity generation, aligning with global trends towards clean energy. This transition, however, presents challenges regarding power system reliability and resilience due to the intermittency of renewable sources. While high-resolution weather data is available, incorporating it into optimization models poses computational difficulties. An integrated multi-sector (power, water, heat) approach is crucial due to the tight coupling between these sectors, offering improved technology integration and economic efficiencies compared to independent optimization. The nexus between water and power, particularly in water-scarce regions with energy-intensive desalination, necessitates co-optimization to manage demand and reduce peak power demand. Existing data reduction methods for optimization often smooth out extreme weather patterns, leading to unreliable system designs. This study addresses this gap by proposing a novel framework.

Many studies utilize clustering techniques to capture renewable resource variability for energy system optimization. However, these methods often smooth out outliers, like extreme weather events, which can significantly impact energy systems. Previous research has explored incorporating extreme days into capacity expansion planning, but primarily focuses on single-node residential systems or the power sector alone, lacking a comprehensive multi-sector approach and often employing deterministic models without exploring the impact of specific renewable technologies. This paper builds on these studies by proposing a more comprehensive framework that addresses these limitations.

The research employs a two-stage stochastic programming model to optimize a fully renewable multi-sector (power, water, heat) energy system for Neom, considering extreme weather days and uncertain demand. The model incorporates various technologies for each sector: power generation (CCGT, PV, CSP with TES, onshore wind, CHP, batteries), water (RO, MED, water storage), and heat (resistive heaters, ASHP, geothermal, CHP, CSP). The model considers both exogenous and endogenous demands. A k-means clustering algorithm divides the Neom region into nine nodes based on renewable resource data. The core methodology iteratively identifies and incorporates extreme days into the optimization. First, a base optimization is performed using representative days obtained via k-means clustering. Then, a simulation using the full weather dataset assesses system performance, identifying extreme days based on unmet demand. These extreme days are added to the representative days, and the optimization is repeated until the system reliably meets demand within a specified tolerance. The model minimizes total system cost, including investment and operating costs, subject to various constraints related to technology capacities, operations, and resource availability. The model considers uncertainties in demand through different load scenarios.

Applying the framework to Neom reveals significant insights. A fully renewable system optimized without considering extreme days failed to meet demand during simulations, requiring substantial electricity purchases from the national grid, particularly during winter months with low solar irradiance and wind speeds. Incorporating extreme days increased total generating capacity by 28%, with wind and PV capacity increases of 68% and 28%, respectively. Water storage capacity nearly tripled. While this increased resilience, it also increased total system costs by 13%. Allowing up to 10% fossil-based generation significantly improved system reliability and reduced costs, dramatically decreasing electricity purchases from the grid. CSP proved cost-effective but less reliable than PV/wind, especially during extreme events due to the limitations of TES compared to battery storage. Geothermal heating offered significant cost savings and reduced power demand, highlighting its potential for sustainable heat provision. The results show a clear trade-off between cost and reliability, with the consideration of extreme days leading to increased cost but enhanced reliability.

The findings highlight the critical need to consider extreme weather events when designing renewable energy systems. Ignoring these events leads to inaccurate capacity estimations and unreliable systems, potentially necessitating costly external energy supplies. The study demonstrates the value of integrated multi-sector modeling, revealing synergistic interactions between power, water, and heat sectors. The results underscore the importance of incorporating uncertainty and extreme events in planning processes, leading to more robust and reliable renewable energy systems. The trade-off between cost and reliability emphasizes the need for a balanced approach, potentially incorporating small amounts of fossil fuels as a safety net to minimize reliance on costly external energy during extreme conditions.

This study demonstrates the effectiveness of an integrated clustering-optimization framework for designing reliable renewable energy systems in regions with abundant renewable resources but high variability. The key finding is the crucial need to incorporate extreme weather events into the planning process to avoid costly reliance on external energy sources. The analysis of different renewable energy technologies highlights the trade-offs between cost and reliability and suggests potential strategies for optimization. Future research could explore advanced clustering techniques, incorporate more detailed modeling of transmission constraints, and consider the impact of climate change on the frequency and intensity of extreme events.

The study focuses on Neom, Saudi Arabia, and may not be directly generalizable to other regions with different climatic conditions or energy demands. The model simplifies certain aspects, such as transmission losses and the uncertainty in technology prices. Further research could incorporate these factors to enhance the model's accuracy and applicability.