There are several sources of short-run risks and long-run deep uncertainties for the T100RE power grid. The latter are defined as uncertainties that could profoundly affect power grid performance but for which meaningful probabilities cannot be assigned because of a lack of mechanistic understanding and historical data. Usually, but not always, the term refers to uncertainties that will unfold over time scales of decades.
Short-run risks of variable renewable energy resources (VREs) are well recognized. However, weather patterns not only affect the output of VREs, but being distributed and networked, those uncertainties can change power system dynamics, thus motivating new control means for VRE-rich power grids (see Thrust 2). In the long-term, planners must cope with uncertain availability and mixes of VREs climate-relevant timescales, as well as risks associated with economics, technology evolution, and policy. Short- and long-run uncertainty factors are tightly related. For example, climate change has been shown to impact availability of hydro- and wind power resources and, thus, their contributions to power production; this affects the reliability and economic benefits of further investments in those sources. Further, climate change can increase the variability of weather and the frequency and severity of extreme events, and will therefore drive the need for enhanced T100RE grid resiliency.
In addition to climate-related factors, deep uncertainty manifests itself in the evolution of technology, policy, and economics, whose social natures make them especially difficult to project. Some technological factors that could greatly influence decarbonization pathways are techno-economic characteristics of VREs and storage resources which are not yet deployed at scale; the ability to effectively aggregate small-scale DERs with high accuracy, power grid operation and market visibility, and controllability; and whether modular nuclear technology will turn out to be a competitive means of mitigating emissions. In turn, future choices of policy instruments can dramatically affect the attractiveness and mixes of alternative resource investments. Furthermore, policy uncertainty arises around carbon emissions, capture, and decarbonization pledges, which prescribe end goals but do not inform specific implementation routes.
This research direction will develop a multi-scenario modeling framework based upon best practices utilized in time-series forecasting models. These models would build upon the Vector Autoregression and Vector Error-Correction Models (VAR and VECM) models used for environmental market analysis to model joint distributions of key model inputs such as fuel prices and demand growth, while including structural uncertainties concerning climate, technology development, and policy evolution. Statistically rigorous multi-variate forecasts can then be combined with planning models to produce statistically rigorous estimates of distributions of key outcomes such as costs. They are also well suited for environments where the objectives are based upon probabilistic criteria, such as minimizing the probability of a transmission contingency or supply deficiency, or minimizing expected resource and transmission cost in multi-scenario stochastic programs, which will be co-developed with Thrust 2 group. As a demonstration of the value of such methods, sets of scenarios developed by the above methods will be used for two types of analyses central to planning the energy transition. First, they will be integrated into long-run transmission planning models that the investigators have used in the UK, US, and Australia, and the benefits of doing so will be compared with traditional methods. Second, building on earlier sensitivity analyses of transition pathways, we will derive rigorous uncertainty characterizations for the cost, technology trajectories, and emissions from energy transition models, which are widely used for policy analyses but usually in a deterministic or sensitivity analysis mode. The characterizations will be fully probabilistic, providing likelihood information that is more useful for risk management, research planning, and valuation of hedges and options, instead of simple sensitivities.