(373y) On Including Weather Extremes in the Design and Operation of Renewable Power Systems

The optimal design and operation of renewable energy systems are paramount to minimizing the high investment costs of renewable conversion and storage technologies, plan power transmission extensions, and minimize the risk of unserved energy while contributing to mitigating carbon dioxide emissions. This study addresses the impact of weather variability and extremes on the design and operation of these energy systems.

Developing comprehensive renewable energy models entails using adequate time resolutions to capture the fine-scale variability of energy demand and renewable sources, such as wind and solar power, and sufficient long-term horizons to capture their seasonal and inter-annual variability. Otherwise, the generation and storage capacities obtained may fail to ensure a reliable and secure supply of energy over specific weather conditions. However, a comprehensive description of weather resources renders complex and large-scale models, which has motivated the adoption of methods to compress weather time series. Two well-known approaches are clustering methods and using a typical meteorological year (TMY).

For example, clustering methods and their assessment are discussed in [1], where alternative methods were applied to define representative periods for two operation optimization problems: a battery case and a gas turbine one. Clustering methods were also adopted to determine representative days for the duration of load curves and renewables power output as inputs to a power system capacity expansion problem [2]. Recently, a methodology was proposed in [3] to design renewable utility plants generating electricity and steam, including a stage for clustering weather resources and electricity demand. The proposed method involves a sampling strategy to generate multiple batches of representative days to perform the design optimization and a post-optimization analysis based on an out-of-sample analysis. Despite using many representative days in the optimization stage, the out-of-sample analysis still indicated that energy from an external grid was required in specific weather conditions. Those results indicate that the representative periods used did not capture the total weather variability of the input data.

In this study, we propose an adapted methodology to incorporate weather extremes in the optimization models and present three case studies related to the design of renewable energy systems. The first case illustrates the impact of reducing the weather variability and extremes in the input data on the design and operation of hybrid renewable systems based on wind, solar, battery storage, and a connection to an external grid; the full details are available in [4]. Two important conclusions are drawn from the results: 1) underestimating weather variability leads to a significant underestimation of battery storage capacity, up to 65%; and 2) the connection to an external grid mitigates the impact of reduced weather variability datasets. However, that work lacks an approach to incorporate weather extremes into the input data. The following two case studies consider a methodology based on the iterative process proposed in [5], also used in [6].

In summary, the methodology involves the following steps: 1) determine the weather input data using clustering or a TMY; 2) perform the design optimization of the renewable system; 3) evaluate the obtained design over historical weather data and determine the periods when the system cannot meet the energy demand; 4) if the system can meet the energy demand then stop; otherwise 4) extend the initial weather input data with the periods that the system fails to meet the demand in step 3) and repeat the optimization as in step 2). This methodology is based on the following ideas: 1) the weather input data obtained through clustering or TMY does not fully capture the historical weather variability or extremes; and 2) the specific extreme periods depend on the renewable conversion technologies and storage capacity selected. The second and third case studies apply variants of the above methodology to incorporate weather extremes in the weather input data.

The second case study focuses on designing a renewable power system for a community, involving the sizing of wind turbines, solar photovoltaic panels, a concentrated solar power plant, and battery storage. The corresponding model spans one year with an hourly resolution, and a TMY represents the local weather conditions. The methodology described to include weather extremes is adapted in an iterative process, such that specific days of the TMY are replaced by extreme days. The method applied identifies the days the system failed to meet energy demand and the corresponding previous day that led to the failure, which is the extreme day. The results show that only three extreme days (from a 25 historical weather dataset) are included in the initial TMY to obtain a system that meets the electricity demand over the historical data. Note that in this approach, the size of the weather input data is the same at each iteration. Detailed results include the conversion and storage costs and technology capacities obtained without and with extremes and the corresponding extreme periods.

The third case study involves planning a power-water-heat system to supply an entire region using an integrated extremes-aware, clustering-optimization framework. It considers a greenfield approach with annual population and demand growth and the design of the following systems to meet water, power, and heat demands: cogeneration technologies (concentrated solar power and combined heat and power), direct use of geothermal for heat, wind turbines, photovoltaic panels, multiple electricity-heat conversion technologies, and multi-effect distillation and reverse osmosis plants to produce fresh water. In contrast to the second case, the renewable resources and demand are described by representative days, as described in [7]. At each iteration, the set of representative days is extended with a specific number of representative days corresponding to extreme days detected in the post-optimization. The results strongly indicate the positive outcome of the first iteration of the proposed method, whereby a 91% decrease in required external power is obtained.

In the three case studies, we also identified a pattern in periods where external power was required, mainly days during winter. Additional investment costs are required in the three cases to cope with weather variability and extremes, mainly allocated to additional storage capacity if a wind-photovoltaic system is considered and additional generating capacity if CSP is adopted. The methodology adopted is generic, though extreme weather events depend on local weather patterns and on selected renewable energy conversion technologies.

References

[1] Teichgraeber H., Brandt A.R., 2019. Clustering methods to find representative periods for the optimization of energy systems: An initial framework and comparison. Applied Energy 239, 1283 – 1293.

[2] Lara C.L., Mallapragada D.S., Papageorgiou D.J., Venkatesh A., Grossmann I.E., 2018. Deterministic electric power infrastructure planning: Mixed-integer programming model and nested decomposition algorithm. European Journal of Operational Research 271 (3), 1037–1054.

[3] Pérez-Uresti S.I., Lima R.M., Martín M., Jiménez-Gutíerrez A., 2023. On the design of renewable-based utility plants using time series clustering. Computers & Chemical Engineering 170, 108124.

[4] Souayfane F., Lima R.M., Dahrouj H., Dasari H.P., Hoteit I., Knio O., 2023. On the behavior of renewable energy systems in buildings of three Saudi cities: Winter variabilities and extremes are critical. Journal of Building Engineering 70, 106408.

[5] Teichgraeber H., Lindenmeyer C.P., Baumgaertner N., Kotzur L., Stolten D., Robinius M., Bardow A.,Brandt A.R., 2020. Extreme events in time series aggregation: A case study for optimal residential energy supply systems. Applied Energy 275.

[6] Li C., Conejo A.J., Siirola J.D., Grossmann I.E., 2022. On representative day selection for capacity expansion planning of power under extreme conditions. International Journal of Electrical Power & Energy Systems 137.

[7] Riera J.A., Lima R.M., Hoteit I., Knio O., 2022. Simulated co-optimization of renewable energy and desalination systems in Neom, Saudi Arabia. Nature Communications 13 (1), 3514.