We recollect a conversation back in 2006 with the then CEO of a large diversified utility. In response to a question, he offered "I don't want to sound spiritual because I will likely not see it in my lifetime, but solar power and storage would be the potential gamechanger whenever it comes. Duration is the hurdle it will have to overcome but when that comes is tricky." We distinctly recollect the conversation because of what he said next, "I ain't often right, but I've never been wrong."

We never ignore anyone who can quote the Grateful Dead. Also, he was right. Solar plus storage technology has advanced much since then. Technologically, it still has a duration problem, but now it is disrupting conventional power.

We Expect Storage Deployments To Increase Substantially

As of December 2020, the U.S. had about 25 gigawatts (GW) of power in energy storage, a number that is still de minimus relative to the renewables already on the grid. Moreover, over 90% of this is pumped hydroelectric storage (23 GW), a tally that is not likely to grow much at all, due to high permitting and construction costs. In contrast, we estimate front-of-the-meter battery storage installations at about 1.3 GW in 2020. Overall, the U.S. has installed just over 1 GW of battery storage in the past five years, with a third of it in the Pennsylvania-Jersey-Maryland (PJM) Interconnection and a quarter in the California Independent System Operator (CAISO) region.

While growth has been robust over the past five years because the deployment of global installed battery capacity has increased at a 25% compound annual growth rate, battery storage solutions in the power sector have flattered to deceive. Each year, mass scale deployment appeared to be on the horizon only to be pushed three years into the future. However, we now believe that battery storage solutions in the U.S. are at an inflection point and that large scale deployment is imminent (see chart 1).

Chart 1

S&P Global Platts Analytics, an affiliate of S&P Global Ratings, expects the U.S. storage market to increase nearly nine times in 2020-2023, fueled largely by advanced battery energy storage, with cumulative deployment approaching 10 GW by 2024 (see chart 2). This is arguably an optimistic projection given installations have typically lagged expectations and will likely be influenced by the pandemic in the near term. We note that the Energy Information Administration (EIA) and SNL Energy project lower aggregate numbers through 2023 of the order of about 6.5 GW while Wood Mackenzie Power and the Energy Storage Association recently published a joint report putting their estimates at 7.5 GW (26.5 GWh) by 2024. Regardless, we think the secular trend is undeniable; the upward growth curve will likely get more convex even if the pace turns out somewhat slower than currently expected. In contrast, other forms of storage have seen their global installed capacity hold relatively flat over this same time period, demonstrating the attractiveness of modular and distributed solutions that advanced battery energy storage offers.

Chart 2

In a January 2018 commentary, we had focused on the economics of stand-alone battery operations in a power generation application (peaking unit) (see "Going With The Flow: How Battery Storage Economics Are Changing Power Consumption," published Jan. 11, 2018). We had also compared the less conspicuous, but more economically attractive, applications like commercial and industrial (C&I) peak demand shaving against the high-profile, yet less economic, applications like residential rooftop solar.

In this follow-up, we update our calculations for stand-alone battery peaker applications based on current cost-curve economies. Our focus now shifts to the economics of utility scale solar plus storage solutions, which we expect to be a gamechanger for the power industry.

What's Driving The Deployment Trend?

Battery installations in the past five years ranged between 150 megawatt (MW) and 225 MW a year, with the largest capacities located in California, the Electric Reliability Council of Texas (ERCOT), the PJM states, New York, and Hawaii. However, even before the pandemic struck, the near-term forecasts for battery installations had declined as uncertainty increased from the trade dispute with China and as some projects faced delays. We suspect there will be some uncertainty around the impact of the pandemic on 2021 deployments because states and the U.S. Congress are still discussing possible economic measures that could benefit storage and renewables more broadly. However, we expect annual deployments to accelerate starting in 2022 because several large projects already contracted will be commissioned.

A combination of reasons will likely spur battery deployments over the next three years, in our view. These include:

• State targets and support schemes, including utility procurement through integrated resource planning processes,
• Growth in storage capacity in the interconnection queue, and
• The economics of storage and the potential for revenue stacking.

State mandates and targets

Similar to renewable portfolio standards that ushered in massive renewable deployments over the past decade, key drivers for the expected surge in battery deployments are the surging mandates and targets. Increasingly, more states--now seven--have implemented an aggregate 11 GW of battery deployments (see table 2). These incentives are particularly important in regions with limited current economic potential for storage, either due to cheaper incumbent technologies or a limited need for new peaking capacity.

Table 1

Storage in the interconnection queue

Anecdotal evidence of a potential tsunami comes from the rapid increase of battery projects in the interconnection queue (see chart 2). This capacity is largely speculative as developers line up projects and reserve slots in the queues in anticipation of utility requests for proposals. Projects with an executed interconnection agreement have a higher likelihood of eventually commissioning. For comparison, the battery capacity in regional transmission organization (RTO)/ISO interconnection queues had increased to 85 GW by June 2020 from 14 GW in June 2018. The three RTO/ISOs with the highest increase are CAISO, PJM, and ERCOT.

Chart 3

In CAISO, the speculative nature of the interconnection queue is underscored by the decline in battery capacity in 2020. Many counties have seen stand-alone or solar plus battery contracts signed with community choice aggregators (CCAs) and one factor that results in a decline in capacity is when a developer, with a project in the queue, fails to secure contracts with those CCAs. Several developers plan to install batteries at existing wind or solar PV assets to benefit from the existing facilities and interconnection capacity while deriving additional revenue for their assets through participation in ancillary service markets.

In PJM, the median capacity MWs of stand-alone projects ranges between 25 MW and 50 MW. Further, only about 55% of stand-alone projects have requested interconnection as a capacity resource, which would allow these units to participate in PJM's capacity market. In contrast, almost all co-located solar PV plus storage solutions have applied as capacity resources. However, the current 10-hour duration limitation to qualify in the capacity market limits the potential revenue for stand-alone batteries. Shorter duration batteries would have to de-rate their capacity to qualify for the capacity market, receiving the capacity market clearing price only for a fraction of their capacity.

Finally, in ERCOT, most battery projects are located in counties that expect large renewable solar PV projects. Moreover, most projects co-located with solar PV installations have a battery capacity ranging from 50%-150% of the PV nameplate capacity. This indicates that developers are seeking to contract with utilities for the provision of peaking capacity.

The potential for revenue stacking

Perhaps the most significant development since we last published on storage in 2018 has been the continually declining cost curve. We believe lower capital costs for storage over the next few years, together with declining PV solar installed capital costs, as well as lower operating and maintenance (O&M) costs, will result in solar plus storage significantly disrupting conventional generation.

But before we get into the cost dynamics, we'll revisit some basics:

Storage has many uses.  In contrast to generating units, energy storage supports a wide range of applications, many of which do not directly correspond to delivering energy to the grid. Storage can be deployed both on the electric grid and at the consumer's home. Its economics are shaped by the customer type, location, customer load shape, rate structure, and nature of the application. For instance, storage can have many different applications, like energy arbitrage, rate arbitrage, generation capacity, or transmission/distribution capacity deferral, etc. It is also uniquely flexible in its ability to change its dispatch to serve different needs over the course of a year or even an hour.

Ancillary services.  Battery application in the ancillary services market was enabled by the Federal Energy Regulatory Commission (FERC) through Orders 890 and 755. These FERC orders effectively allowed non-generator resources to compete with generating units by selling ancillary services as a stand-alone product. These applications typically require storage of 15 minutes to an hour.

Ancillary services are increasingly required as electricity grids experience continuous imbalances between power generation and consumption as millions of devices are turned on and off in an uncorrelated way. These imbalances cause electricity frequencies to deviate, which can hurt sensitive equipment and, if left unchecked and allowed to become too large, even affect the stability of the electric grid. To avoid disruptions and blackouts, power has to flow within the grid at quasi-constant frequency (or voltage). Historically, thermal plants would provide that by quickly varying the kinetic energy of their generation units. Now, storage systems are particularly well suited for frequency regulation applications because of their rapid response time and ability to charge and discharge efficiently.

So, in the near term, it has all been about short-duration opportunities (frequency regulation and spinning reserve markets), which continue to be key drivers behind fully or quasi-merchant projects. Of these, the most economic opportunity remains the frequency response market. In PJM, this is Regulation D, i.e. the ancillary services market. While almost all batteries in PJM have been commissioned to provide frequency regulation, it is interesting to see that even in CAISO most batteries currently commit most of their capacity to the provision of frequency regulation. In New York ISO and ISO-New England as well, batteries have been commissioned mostly to perform frequency regulation. While this use for batteries is limited, we see compensation as attractive and see the bulk of near-term storage development activity oriented toward earning revenue in this $100 per kilowatt-year (/kW-year) market from short duration 15-minute to 1-hour battery products. Near term, this service will likely support early deployments in more markets--such as in the Southwest Power Pool and the Midwest Independent System Operator (MISO). However, ancillary services, in general, are shallow markets, prone to rapid saturation. Yet, ancillary services can provide some initial growth in volumes that can help drive down battery prices.

Peaking capacity.  Revenue stacking remains key to achieving a positive return on investments with long-duration batteries. New deployments will have to tap into new revenue streams (price arbitrage, spinning reserve, capacity, and transmission and distribution deferrals) as we have seen happen in the U.K and Germany (see chart 4). Just as one would expect of nascent technologies, we've seen storage solutions enter the market as 15-minute products that have now grown to economical peak-shifting four-hour applications. The total value from storage is a blend of several benefit streams from different sources and might not always be driven purely by economics.

Chart 4

We believe the provision of peaking capacity is the next big use because utilities are starting to recognize the value of storage in their integrated resource planning, particularly in the vertically integrated markets. Actually, we expect most of the large projects recently contracted by utilities to provide peaking capacity (with or without solar PV). Generally, the first utilities contracting for storage were located in regions with high solar PV insolation but now it is increasingly across all RTO regions. However, most projects will have a four-hour duration and not come online before 2022.

Low reserve margins in ERCOT will likely drive the first wave of merchant projects. ERCOT has the highest revenue potential due to a scarcity pricing mechanism that increases power pricing at times of low reserve margin. For instance, ERCOT saw more than 100 hours in 2019 when energy prices exceeded $1,000 per megawatt-hour (/MWh). As reserve and energy are co-optimized, spinning reserve prices reached very high levels, making it an attractive service for batteries to provide. ERCOT procures between 2,500 and 3,000 MW of spinning reserve and about 250 MW of upward and downward regulation, on average.

Stand-alone battery economics

Battery costs have declined enough to replace gas-fired peaking generation in certain markets such as Hawaii and California. With continually declining capital costs, the economics of a four-hour grid-level battery storage unit is within the neighborhood of a gas peaker plant. But with no moving parts, the battery bank needs less maintenance (yet, battery asset life is shorter). It also requires no fuel, making its long-term cost of operation more stable and predictable. A battery bank can respond to power demand almost instantly--less than a millisecond rather than several minutes. Where a gas turbine is strictly an energy generator, a battery bank can also store surplus energy. Finally, a battery bank is scalable: more units can be added as needed without dramatic cost increases. We discuss how these factors factor into our ratings assessment in the commentary Credit FAQ: What Factors Power A Rating On A Utility-Scale Battery Storage Project?, published Jan. 12, 2018

Since we have the highest visibility of Lithium-ion economics, we'll take that as an example. First, when we think in terms of capital costs for batteries, units are in kilowatt-hours (kWh) of operation--i.e., it's in $/kWh because we expect batteries to be duration products for peak shifting (or peak shaving) solutions. Stated differently, since a battery is used to store energy, the capital cost of a battery is expressed in dollars per unit of energy stored, or $/kWh. To be clear, this is different from energy prices, where the cost of electric energy is expressed in cents/kWh.

Utility scale battery economics are currently at about the $240/kWh price point, or about $950/kWh for a battery peaker plant that provides a four-hour peak shift. Costs for the balance of the system are about $300/kW for equipment like inverters/rectifiers, transformers and power control equipment, and various safety equipment. So a utility scale battery would currently cost about $1,250-$1,300/kW (240 times 4 plus $300-$350)--we think those costs are comparable with the cost of building a natural gas-fired peaker plant in California. We present a simplified calculation (see table 2) of the installed cost of a battery park that could replace a peaking unit.

Table 2

From market sources, we are given to understand that a $11-$12/kW-month fixed all-in tolling payment (i.e. capacity, energy, etc.) is roughly in line with contracts with stand-alone battery projects. We think batteries as capacity and energy resources are still some time away from being economically viable in a merchant market compared with both ancillary services and C&I peak-shaving applications. However, as a natural transition until they become competitive as peaking applications, such investments are increasingly being considered as part of a utility's rate base or supported by a contract in fulfillment of state initiatives.

Stand-alone solar PV economics

Due in part to the capital cost declines, the economics of pairing storage with solar to provide a near firm power product are now favorable in many regions. Utility resource plans are including co-located storage as a new resource option; we have seen increases in announced storage projects as a result.

However, before we talk about co-located solar plus storage solutions, we'll discuss how solar PV economics have recently trended. We calculated the levelized cost of energy (LCOE) of a competitive solar unit--one whose capital and operating costs are in the 25th percentile. Our base case assumptions (see table 3) for overnight capital costs and capacity factors are $1,050 per kW and 28%, respectively--with a cautious note that capacity factors do vary significantly by region (i.e. irradiance/insolation). Capacity factors are lower for projects in the PJM and MISO, where 25% is more appropriate, for instance.

Table 3

Our base case LCOE estimate for stand-alone solar PV, including an ITC, is about $23.00/MWh. A no-ITC case increases our estimates by about $5.50/MWh to $8.50/MWh, depending on how low (or high) the base case O&M and capital costs assumptions are (see table 4). These LCOEs appear low because they represent the LCOE of the lowest cost units, also because of the prevailing financing environment. We have presented a matrix that provides a snapshot of how LCOEs would move if operating costs, or utilization, were higher or lower than base case expectations. For example, in urban areas near major cities, property taxes will make operating costs higher than our assumptions.

Table 4

Our results are consistent with LevelTen Energy's P25 power purchase agreement (PPA) price index (see chart 5). This PPA index reports the prices that solar project developers have offered for standard, flat, as-generated (i.e. unit-contingent) power from their projects. The P25 refers to the most competitive 25th percentile PPA offer prices. The P25 of levelized PPA offer prices is $25.10 and $26.70 in CAISO and ERCOT, respectively, even as the aggregate prices were higher because of higher offers in the PJM Interconnection and MISO.

We note that the PPA index prices were lower still and have risen recently. Permitting and interconnection are becoming increasingly costly and difficult in many U.S. markets, which creates a bottleneck, constraining supply and putting upward pressure on prices. As the best locations get scooped up, identifying and securing permits and interconnection positions--in the absence of broad-based reform in these two areas--becomes more challenging for new projects. In addition, economically competitive projects that had secured permits, and safe-harbored equipment for the full ITC, found contracts with offtakers, leaving higher-priced projects available in the market.

Chart 5

Peak Shifting Economics

The problem with solar power is that it is interruptible. Moreover, all solar units produce at the same time. This means that as solar installed capacity goes up, its effective load carrying capability (ELCC) goes down. This is because the ELCC values depend on when electricity shortages are most likely to occur, and the type and quantity of resources already on the grid have a significant impact on the likely timing of electricity shortfalls. As the grid adds more solar plants that are all producing electricity at the same time, it reaches a point where all those solar plants are preventing daytime reliability issues so effectively that the remaining reliability challenges move into the evening hours when solar can't produce. At this point, adding more solar does very little to prevent electricity shortages, and its ELCC often falls to 0%. This is why co-location of storage with solar has gained importance.

We see co-located solar plus storage projects of varying configurations. Broadly, the ratio between the solar PV capacity and the battery capacity can provide an indication of the use of the battery. While the ratio varies widely by developers, lately, we have seen this increase, suggesting that battery usage is slowly moving into longer duration applications like peak shifting and peaking capacity. Ratios above 50% indicate that batteries are used to provide some peaking capacity (by shifting midday generation to evening peaks). For example, National Grid USA commissioned a 6 MW/48 MWh battery in Nantucket, Mass. in October 2019, to be used to meet summer peak demand and defer the need for an additional underwater supply cable to the island. A 20 MW/100 MWh battery at the Lawai Solar and Energy Storage Project, commissioned by The AES Corp. (BBB-/Stable/--) in December 2018 in Hawaii, provides dispatchable solar PV. Similarly, Vistra Corp. has developed the world's largest battery energy storage project, the 400-MW/1,600-MWh Moss Landing energy storage facility, in California.

Note that utility-scale solar-plus-storage installations can take advantage of the ITC as long as the energy storage system is at least 75% charged by the on-site solar unit. Our assumptions (see table 3) for co-located batteries with solar are similar to those for a stand-alone unit but with additional ITC benefits from the storage component to offset initial capital expenditures. Also, the storage is typically priced either through a $/kW-month fixed tolling payment basis, as with stand-alone storage, or as a $/MWh adder to the total project PPA. However, as a result of the ITC benefit, the fixed toll is typically lower for co-located storage than for a stand-alone storage unit.

Based on our analysis, we arrived at a LCOE value that adds about $10-$11/MWh to the LCOE ranges of stand-alone solar, or a premium of about 50% to stand-alone solar levelized costs. However, an LCOE analysis in the context of storage is less relevant. What matters more in storage installation is the type of unit that the solar plus storage unit is displacing. In regions that are served by a preponderance of base load units and peaking units, without any meaningful intermediate units in the power supply stack, solar plus storage could provide a valuable firm power product. For instance, in a location like Hawaii, offtakers will be willing to pay a higher power price if the unit that is being mitigated is a high-cost oil-fired peaker. Thus, it is better to evaluate the internal rate of returns (IRRs) for co-located storage projects. We chose to model a solar plus storage unit that employs a $/MWh storage premium adder of 55%. (see storage premium adder rationale in table 3).

Table 5

We believe that continued cost reductions support further penetration, with current all-in utility-scale storage costs trending around $240-$250/MWh for larger procurement. Reliability remains a key driver, with value in storage being located close to load in supporting grid stability in light of transmission constraints, especially in markets like California. However, given relative complexity compared with solar assets, managing the operating risks of storage assets will be key to enabling reasonable IRRs.

Implications For The Power Sector

We think that with substantially lower capital costs for both solar and storage, solar developers will increasingly deploy co-located projects. Not only does adding storage provide the potential for offering a firm power product, it has meaningful up-front benefits from qualifying for ITCs. Storage effectively doubles the capital costs of a solar project, 26% of which can then immediately attract ITC benefits and also avail modified accelerated cost recovery system depreciation. We believe adding storage to existing sites offers significant growth prospects to companies like NextEra Energy (A-/Stable/--) because it already has the advantage of established interconnections.

We see AES Corp. as another company that would benefit given its widespread adoption of storage at renewables development sites, as well as at existing thermal sites. In particular, Fluence, AES' battery energy storage joint venture with Siemens AG, has disclosed that it expects to make $500 million in revenue in 2020 and increase it by 45% each year over the next five years. The company recently announced the acquisition of Advanced Microgrid Solution's (AMS) software and digital intelligence platform. AMS' technology uses artificial intelligence, advanced price forecasting, portfolio optimization, and market bidding to ensure energy storage and flexible generation assets respond optimally. We think this will support Fluence's ability to provide competitive fully integrated storage solutions given the need to provide optimized "revenue stacking" around storage assets. This is particularly important in markets like ERCOT to capture merchant value given real-time price volatility and lack of a capacity market. We note that at year-end 2020 Fluence received a $125 million private investment from Qatar Investment Authority for a 12% minority stake. After the transaction, AES and Siemens will remain major shareholders, with each maintaining a 44% stake. The transaction implies a $1 billion valuation, or roughly twice 2020 sales for the company.

While storage creates opportunities for many companies, it also poses a threat to companies that have significant conventional generation. As renewable costs continue to decline, they weigh heavily on both peak and off-peak power prices. We think this changing market paradigm affects gas-fired assets in different ways depending on their location or their ability to respond quickly to market signals. We think that in several regions of the country gas-fired assets will continue to coexist with renewables. For instance, Energy & Environmental Economics (E3) believes that the least-cost electricity portfolio to meet the 2050 economy-wide greenhouse gas goals for California includes 17 GW to 35 GW of natural gas generation capacity for reliability. The market consultant concluded that firm capacity is needed even while adding very large quantities of solar and electric energy storage. We think power companies with gas-fired assets could witness cash flow backwardation as they cede market share slowly to renewables but have adequate time to pivot strategies.

However, companies that have coal-fired generation exposure could face substantial asset retirements. Given continued low power prices, increased renewable penetration, and higher costs to operate coal units, we see nearly 85 GW, or about one-third of existing capacity, at risk (see link to our November 2019 report below). It is no surprise then that companies such as Talen Energy Corp. have announced substantial capital plans. In a November 2019 strategic shift, Talen announced that its Montour generation facility located in Pennsylvania and its Brandon Shores and H.A. Wagner coal generation facilities located in Maryland will cease coal-fired operations by the end of 2025 and repower with alternative fuel. The company simultaneously announced that it would undertake about 1 GW of renewable energy (both solar and wind) and battery storage projects currently under development across its existing asset footprint.

Similarly, in September 2020, Vistra Corp. announced a comprehensive plan to accelerate its transition to clean power generation sources and advance efforts to significantly reduce its carbon footprint. The company's plans include the development of six new solar projects and one battery energy storage project that total nearly 1 GW and represent a capital investment of approximately $850 million in ERCOT. Vistra simultaneously announced its next phase of coal plant closures in Illinois and Ohio. The company plans to retire seven plants with a combined capacity of more than 6.8 GW between 2022 and 2027. In aggregate, Vistra and its subsidiaries have retired or announced the retirement of more than 19 GW at 23 coal and natural gas plants since 2010.

"Once in a while you get shown the light, in the strangest of places if you look at it right"

So continue the lyrics of that fan favorite Grateful Dead song that we alluded to in our introduction. Like many market participants, we have focused on the cost curve of storage for some time and have been watching the rapid increase of battery projects in the interconnection queue with some interest. However, we now believe that the market has been spending too much time monitoring the declining cost curve of batteries and, in doing, so missing the forest for the trees.

We think this is not just about storage cost curves. While that is an enabler, the real story is in the improving economics of solar power generation. In 2020, overnight capital costs of solar PV installations declined below that of onshore wind generation. Moreover, operating expenses for solar PV projects have also declined substantially over the past three years. As a consequence of these developments, the PPA index prices for solar power have declined below those of wind power (for the first time in the first quarter of 2020). Given batteries would provide the much-needed firming that solar power needs, we think large-scale co-located solar plus storage deployments are now imminent, which would be a gamechanger for the power industry.

Chart 6

Related Research

• S&P Global Sharply Lowers Its Recovery Values For Coal-Fired Power Projects, Nov. 7, 2019
• Credit FAQ: What Factors Power A Rating On A Utility-Scale Battery Storage Project?, Jan. 12, 2018
• Going With The Flow: How Battery Storage Economics Are Changing Power Consumption, Jan. 11, 2018
• Uncharged: The Possible Market Impact Of Electricity Storage, Jan. 10, 2018
• Future Shock: Will Better Batteries Dim Electric Utilities' Prospects?, Nov. 3, 2017
• Storage: The Final Piece In The Global Energy Transition Puzzle, June 10, 2016