Mitigating Extreme Wind and Solar Power Fluctuations using Portfolio Optimization

Project proposal Bright Minds Assistantships February 2022

Mitigating Extreme Wind and Solar Power Fluctuations using Portfolio Optimization

Department: Copernicus Institute of Sustainable Development Research group: Energy & Resource

Supervisor: Jing Hu

Email address: j.hu@uu.nl

Project description

Intermittent renewables such as wind and solar energy play an essential role in the energy transition.

However, the large-scale integration of wind and solar power into the power system also poses challenges to power system operation. Depending on weather conditions, the output profiles of wind and solar power fluctuate over time. The sudden fluctuations between two consecutive time steps create increased demand for flexibility resources (in terms of backup capacity, operating reserves, and storage) to maintain grid balance between supply and demand. Empirical data has demonstrated that the fluctuations of wind and solar follow a typical heavy-tailed non-Gaussian distribution, meaning that they are more prone to generate extreme fluctuations. As rare but high-magnitude events, extreme fluctuations can cause large disruptions in power supply and threaten system adequacy. At worst, blackout or brownout may happen, resulting in substantial socioeconomic costs. This is a particular risk when the availability of flexibility resources is low and when the power system heavily relies on wind and solar power.

It is well-known the collective outputs of wind and solar power can be smoothed out through optimizing the pooling of wind and solar generation assets. This is because complementarity in generation profiles exists between assets across space and technology.

In the proposed BMA project, the bright mind will help evaluate the effectiveness of optimal pooling in limiting extreme fluctuations events in wind and solar power. The focus is placed on downward

fluctuations (in terms of shortage) rather than upward fluctuations (in terms of surplus). Unlike upward fluctuations that can easily be solved through curtailment, downward fluctuations result in sudden power losses which must be compensated by backup capacity or storage. To this end, two novel portfolio optimization methods specifically targeting different downside risk measures (semi-variance and conditional Value-at-Risk) will be employed to develop the optimal pooling of wind and solar.

Extreme value analysis will also be used to predict out-of-sample extreme fluctuation events in the developed wind and solar portfolios and demonstrate the benefits obtained from portfolio optimization.

Taiwan island is selected as the case study area in this project. Taiwan has already set ambitious 2030 renewable targets to phase out nuclear power, boost energy security and reduce CO

emissions.

However, as a large islanding power system relying heavily on energy imports, Taiwan has limited availability in flexibility resources in absence of interconnection with mainland China. This makes Taiwan’s power system more susceptible to extreme fluctuations of wind and solar power.

The main results of this research will be fed into a quality academic publication with the student being

the main co-author. Relevant data inputs for this research have already been collected. The preliminary

methodology has also been developed in an early conference paper of the supervisor. The two novel

Project proposal Bright Minds Assistantships February 2022

portfolio optimization methods developed in this paper contribute to easing challenges associated with the system integration of wind and solar. They also enrich the toolbox of investors, system operators, policymakers, and energy modellers to support a system-friendly energy transition.

During the research, regular meetings between the student and supervisor will be arranged at least once per week for the discussion of progress, ideas and intermediate results. The meeting is primarily

organized at the university, but it must follow the COVID19 restrictions. Online meetings can be used as well via MS TEAMs.

Job requirements

The student should be able to use R or Python for data analysis and optimization with large data.

Experience in ArcGIS or similar tools is a plus. Basic knowledge about statistics, energy conversion and

portfolio theory are very much appreciated.