Does Electricity Storage Innovation Reduce Greenhouse Gas Emissions?

RESEARCHERS: Joshua Linn, Jhih-Shyang Shih

SUMMARY

While public policy and innovation in large-scale electricity storage are anticipated to reduce costs and improve performance, the effect of storage costs on greenhouse emissions depends on the supply responsiveness of fossil and renewable generators.

KEY FINDINGS

• Under typical US conditions, lower electricity storage costs are more likely to reduce emissions when wind investment responds to electricity prices, and less likely to reduce emissions when solar investment responds to electricity prices.
• Introducing a pricing on carbon dioxide emissions may increase the likelihood that lower storage costs will reduce emissions.
• Policies incentivizing storage investment and R&D subsidies that reduce storage costs have ambiguous effects on emissions in the medium run.
• Storage investment is positive when storage costs approximately half of what it did in 2010. This suggests that further storage innovation will be needed for storage to be economically viable for arbitrage purposes.

ABSTRACT

In the electricity sector, innovation in large-scale storage is anticipated to reduce costs and improve performance. The effect on greenhouse gas emissions of lower storage costs depends on the interactions between storage and the entire grid. The literature has disagreed on the role of storage in reducing emissions. Using a stylized model, we show that the effect of storage costs on emissions depends on the supply responsiveness of both fossil and renewable generators. Under typical conditions in the United States, lower storage costs are more likely to reduce emissions when wind investment responds to equilibrium electricity prices and when solar investment does not. Simulations of a computational model of grid investment and operation confirm these predictions. Moreover, because of its effect on coal and natural gas–fired generation, introducing a carbon dioxide emissions price may increase the likelihood that lower storage costs will reduce emissions.

Introduction

In the absence of policy intervention, private decisionmakers do not consider the external costs of greenhouse gas emissions, such as using electricity generated by fossil fuel combustion. Standard economic theory suggests that setting an emissions price equal to social damages, via either an emissions tax or cap-and-trade, is the welfare-maximizing approach to addressing this market failure. However, policymakers seeking to reduce greenhouse gas emissions have demonstrated a preference for subsidizing low-emitting technologies rather than fully pricing emissions. A vast array of explicit and implicit subsidies for low-emitting technologies exists, such as renewables tax credits and requirements that renewables provide a specified fraction of electricity generation. Although some policymakers have adopted an emissions price, the price rarely if ever fully internalizes the costs of greenhouse gas emissions. For example, the current US regulation of the electricity sector will approximate an emissions price of $10 per ton of carbon dioxide (EIA 2014), which is likely to be substantially lower than the external cost of emissions (Greenstone et al. 2013; Nordhaus 2014).

The literature has demonstrated that these subsidies are not economically efficient. Subsidizing research and development (R&D) and adoption of low or zero-emissions technologies reduces the private costs of adopting these technologies, in both the short and the long run. However, several recent studies (e.g., Holland et al. 2009; Fell and Linn 2013) show that these policies can have ambiguous effects on emissions and social welfare. For example, subsidizing wind- and solar-powered electricity generators can reduce electricity prices, increasing consumption and generation from fossil fuel–fired generators. This effect can offset the emissions reductions from such policies, reducing efficiency compared with an emissions price.

Under the rationale of subsidizing low-emitting technologies, subsidies for R&D and adoption of large-scale electricity storage are also becoming more widespread. For example, since 2009 the US Department of Energy has provided roughly $200 million in funding for such storage research. Storage subsidies are commonly supported by the view—which many recent studies (e.g., de Sisternes et al. 2016) have largely confirmed—that electricity storage reduces the costs of achieving very high levels of renewables generation and limiting greenhouse gas emissions, as well as providing other benefits to the electricity system. Because electricity production from wind- and solar-powered generators is more difficult to control than production from conventional technologies, integrating large amounts of wind and solar increases the challenge of balancing electricity demand and supply. Storage can address this challenge by charging the storage device when electricity supply is abundant relative to demand, and discharging when supply is scarce. According to this view, subsidizing storage R&D and adoption reduces the costs of integrating renewables and reduces emissions (or, alternatively, reduces the cost of meeting an emissions objective).

However, Carson and Novan (2013) and Graff Zivin et al. (2014) provide an alternative view of storage. They show that adding an incremental amount of storage can increase greenhouse gas emissions by causing a shift in generation from lower- to higher-emissions sources, such as from natural gas– to coal-fired generation. Thus, a central question for storage policies is whether anticipated reductions in the cost of storage will reduce emissions—that is, whether the widespread view of storage as facilitating emissions reductions is valid. The literature has provided conflicting views on this question.

We reconcile these opposing views of storage by taking an alternative approach, in which we consider potentially large amounts of storage and renewables capacity added to the existing grid, and analyze the effects of storage costs on emissions. Previous studies differ in the time horizon, either considering an incremental amount of storage added to the existing grid (e.g., Carson and Novan 2013) or redesigning the entire power system in the long run (e.g., de Sisternes et al. 2016). A short-run analysis is confined to the interaction between storage and existing generators, and cannot assess whether storage reduces the cost of integrating renewables. In contrast, we consider the medium run, a timeframe of 10 to 20 years, and include the interaction between storage and investment in new electricity generators. Moreover, the medium run, rather than the long run, is the relevant timeframe for studying current policies that affect storage costs and near-term investment in generation and storage capacity.

Whereas most other studies analyze the effects of an exogenous increase in storage capacity (e.g., Walawalkar et al. 2007; Sioshansi et al. 2009; Nyamdash et al. 2010), we consider a context in which storage investment depends on storage costs and other market factors. This focus is motivated by several considerations. First, in practice, storage investment depends on decisions made by individual investors (in some cases with regulatory oversight) in response to market conditions, but most previous studies have treated storage as exogenous to the model. Second, the focus on storage costs is relevant to storage policies, which primarily reduce storage costs in the short run (via investment incentives) and long run (via R&D subsidies). Third, partly but not entirely because of storage policy, technological innovation over the coming years is likely to reduce storage costs (Kintner-Meyer et al. 2010). In the context of declining costs, the most relevant question for the future of storage is how the anticipated reduction in storage costs will affect emissions and other outcomes.

We analyze the medium-run effects of storage costs on carbon dioxide emissions from electricity generation, and characterize conditions in which a decrease in storage costs reduces emissions. We focus on storage used for arbitrage purposes, charging when wholesale electricity prices are low and discharging when prices are high (wholesale prices are the prices received by electricity generators supplying electricity to retailers or utilities). We begin by using a simple, stylized model of a wholesale power market that generalizes Carson and Novan (2013) to include investments in wind and solar power generation. We show that the effect of storage costs on emissions depends on relative supply responsiveness—mathematically, the derivative of generation with respect to electricity price—of fossil fuel–fired and renewables generation plants.

To provide intuition for this result, we note that storage charging and discharging raises equilibrium prices during what would otherwise be low-price periods and reduces prices during what would otherwise be high-price periods; in the extreme case of free storage, equilibrium prices are equal across periods. Storage therefore has two effects on operation and investment of generators. The first is that storage raises generation from existing fossil-fired generators in low price periods and reduces generation from existing fossil-fired generators in high-price periods. As we show, coal-fired generation is typically more price responsive than is natural gas–fired generation during low-price periods, whereas natural gas–fired generation is typically more price responsive than is coal-fired generation during high-price periods. Therefore, reducing storage costs raises storage capacity and causes a shift from natural gas– to coal-fired generation. Because coal-fired generation is more emissions intensive than natural gas–fired generation, a decrease in storage costs is likely to raise emissions; this is the effect that Carson and Novan (2013) identify.

The second effect is novel: it is the response of renewables investment to storage. For wind- and solar-powered generators, it is useful to focus on the responsiveness of investment with respect to the generation-weighted average electricity price. Renewables generation may be positively or negatively correlated with electricity price changes caused by storage, depending on the availability of the underlying resource and other factors. For example, in many regions wind generation peaks during the nighttime, when electricity demand and prices tend to be low. In that case, storage would increase nighttime electricity prices, and wind generation would be positively correlated with the electricity price changes. When renewables generation is positively correlated with electricity price changes, lower storage costs raise the generation-weighted average electricity price and therefore renewables investment, displacing fossil fuel–fired generation and emissions. In this case, the more price responsive is renewables investment, the more likely that lower storage costs reduce emissions.

In contrast, when renewables generation is negatively correlated with electricity price changes caused by storage, reducing storage costs reduces the generation-weighted average electricity price, causing renewables investment to decrease. The renewables generation is displaced by fossil fuel–fired generation, and the more price responsive is renewables investment, the more a reduction in storage costs reduces renewables investment and raises emissions. This case often, but not always, pertains to solar.

The stylized model suggests that the effect of storage costs on emissions depends on the price responsiveness of fossil fuel–fired and renewables generation. We test these predictions using a more detailed optimization model that endogenizes storage operation, dispatch of coal and natural gas–fired generators, and investment in storage, wind, and solar. The model accounts for the nondispatchability of renewables and includes calibrated supply curves for renewables and fossil fuel–fired generators. Applying the model to the Texas power system (i.e., the Electric Reliability Council of Texas, ERCOT), we first consider the case of zero renewables investment, which is comparable to Carson and Novan (2013). We find that lower storage costs increase storage investment and raise emissions precisely for the reasons identified in the stylized model: coal-fired generation is more price responsive during low-price periods, and natural gas–fired generation is more price responsive during high-price periods. As in much of the United States,in ERCOT wind generation is positively correlated with storage-induced electricity price changes, and solar generation is negatively correlated. If we allow for wind and solar investment, lower storage costs raise emissions, as in the case without investment. However, changes in the price responsiveness of wind and solar investment have the predicted effects: a reduction in storage costs is more likely to reduce emissions the more price responsive is wind investment, and a reduction in storage costs is less likely to reduce emissions the more price responsive is solar investment.

An extension of our stylized model suggests that a carbon price has an ambiguous effect on the likelihood that lower storage costs reduce emissions. On the one hand, the carbon price causes fossil fuel–fired generation to be less price responsive relative to wind generation, which raises the likelihood that lower storage costs reduce emissions. On the other hand, the carbon price causes fossil fuel–fired generation to be less price responsive relative to solar generation, which reduces the likelihood that a reduction in storage costs reduces emissions. In the baseline model calibration, adding a carbon price causes lower storage costs to decrease emissions for most levels of storage costs considered.

Our paper contributes to the literature in several ways. First, we characterize an internally consistent set of supply conditions under which lower storage costs reduce carbon emissions. In contrast, Carson and Novan (2013) hold fixed renewables investment, and long-run studies do not characterize these supply conditions. Second, because wind investment responsiveness is central to the relationship between storage costs and emissions, we present the first attempt to estimate this responsiveness directly from observed investment decisions. By comparison, other studies rely on engineering-based cost estimates. Third, our computational model endogenizes investment in storage, wind, and solar capacity; previous studies have treated one or more of these as determined outside the model. This allows us to consider the question most relevant to storage policies: how a reduction in storage costs affects emissions, given the makeup of the existing grid.

Read more ...

Source: http://www.rff.org/research/publications/does-electricity-storage-innovation-reduce-greenhouse-gas-emissions