Residential Energy Storage Economics
Inside this Article
If analysts’ projections hold true, behind-the-meter energy storage applications in general—and residential applications in particular—are poised for dramatic growth over the next 3 to 4 years.
Does this mean that stored energy costs are approaching the retail price of electricity?
In “US Energy Storage Monitor: 2015 Year in Review” (see Resources), GTM Research forecasts that the US annual energy storage market will reach 1.7 GW by 2020, a 26-fold increase from 2014. The projected growth by market sector is of particular interest to solar installers. According to data that GTM Research and the Energy Storage Association gathered for the report, customer-sited behind-the-meter energy storage deployments accounted for roughly 11% of the total market in 2014 and 16% in 2015, with utility deployments comprising the vast majority of installed energy storage capacity. However, GTM Research expects that by 2019 the energy storage deployed behind the meter in nonresidential and residential applications will eclipse the utility segment.
In this article, we consider some of the use cases and emerging markets for residential grid-interactive energy storage systems. We then present a simplified levelized cost of stored energy (LCOS) metric, useful for analyzing the economics in residential energy storage applications. In addition to exploring the sensitivity of LCOS in relation to various parameters, we consider the practical limitations of this metric. Throughout the article, we consider some of the ways in which the emerging energy storage market parallels the early days of the solar industry, and what this may tell us about its path forward.
Market and Applications
The projected growth of the residential energy storage market presents a tremendous opportunity for solar installation companies. After all, interconnecting a battery-in-a-box type of appliance is not all that different from making the grid connection for an interactive PV system. Further, adding solar to an energy storage system can improve the overall value proposition, both by leveraging tax credits and by adding another battery charging source. Finally, solar companies can offer energy storage as a retrofit solution to existing customers or as an optional upgrade to new customers, which are compelling ways to drive down customer acquisition costs and increase sales revenue.
The challenge, of course, is that the energy storage market has made less progress along the classic technology adoption and hype cycle curves, shown in Figure 1, than solar has. The California residential grid-tie solar market, for example, progressed from innovative pilot projects in the 1990s to a classic early adopter market at the beginning of the 2000s, when customer concerns about the environment or self-sufficiency drove sales more than economic self-interest. Now that residential solar has reached grid parity in California, the technology appeals more broadly to economic pragmatists, which is characteristic of an early majority market. While solar market maturity varies from state to state, GTM Research’s latest in its series of US solar market reports (see Resources) estimates that solar has reached grid parity in 19 states plus the District of Columbia, meaning that the levelized cost of energy for PV is equal to or lower than the retail price of electricity in these markets—after accounting for incentives and tax credits.
By comparison, these are very early days for the residential energy storage market in the US. The authors of the Rocky Mountain Institute (RMI) report “The Economics of Grid Defection” (see Resources) estimate that residential solar-plus-storage has the potential to achieve grid parity in Hawaii today under certain scenarios that assume improvements in demand-side load management and battery cost reductions. Under a more realistic base case scenario, the authors project the following: “Residential [solar-plus-battery] systems will reach grid parity as early as the early 2020s in Hawaii, late 2030s in Los Angeles, and late 2040s in [Westchester, New York].”
While Tesla succeeded in creating a lot of buzz about energy storage with its Powerwall announcements, it is reasonable to wonder whether current technology and business cases can meet customer expectations. If not, energy storage likely still has a hill to climb before it overcomes its own hype, as represented by the slope leading to the Peak of Inflated Expectation in Figure 1. By comparison, the leading solar markets in the US are more developed, advancing perhaps along the Slope of Enlightenment.
Part of the challenge facing residential energy storage in the US is that its use cases or applications are limited compared to those of maturing energy storage markets elsewhere in the world or even of domestic commercial deployments. The authors of a more recent RMI report, “The Economics of Battery Energy Storage” (see Resources), identify four customer services for behind-the-meter energy storage: backup power, increased PV self-consumption, time-of-use bill management and demand reduction. To put these applications in context, let us briefly review the goals and ideal economic environment for each scenario.
Backup power. The role of energy storage in backup power applications, of course, is to support loads in the event of a grid failure. Generally speaking, the installer relocates a subset of the home’s branch circuits to a new subpanel. In the event of a grid failure, loads connected to this backup- loads subpanel can operate for as long as the battery state of charge remains above its allowable depth of discharge. The duration of backup power varies based on the load profile and the battery capacity. A benefit of solar in this scenario is that it provides a battery charging source when the grid is down, which can extend backup power availability from a period of hours to days.
Demand for backup power is an important driver in residential energy storage applications, especially in the short term. However, this is also a classic early adopter market, one driven more by customer desire than by economics. It is simply not possible to develop a business case that justifies residential energy storage for backup power on a dollar per kilowatt-hour basis. Instead, customers willing to pay a premium price to have the latest technology or to ensure that they can keep lights on during an outage are the ones driving these sales.
PV self-consumption. The role of energy storage in a self-consumption or zero-export scenario is to store excess PV production and discharge this stored energy later. Self-consumption and zero-export applications always include solar. As compared to standard interactive or backup power systems, self-consumption applications require additional energy monitoring. In effect, the solar-plus-storage system needs to see the home energy consumption in real time to optimize energy inflows and outflows for maximum customer benefit. Moreover, zero-export systems need to curtail PV production whenever generation exceeds on-site loads and the battery bank’s capacity to store energy.
Markets where utilities value PV outflows at a rate that is lower than the retail price of electricity, or where they simply will not allow PV outflows, drive the demand for PV self-consumption or zero-export systems. In markets with net energy metering (NEM), customers can effectively store excess solar generation on the utility grid. Since these customers receive the full retail price for outflows, the only benefit of adding energy storage is as a hedge against the loss of NEM. In markets without NEM, however, PV customers can use a battery to shift delivery of excess PV generation in time for use later, as illustrated in Figure 2. To be economically viable, self-consumption applications need to offset the inherent efficiency losses associated with energy storage, as well as the substantial up-front costs associated with the additional hardware.
Time of use. As with PV self-consumption systems, the role of energy storage for time-of-use bill management is to store energy for use later. The primary difference between these scenarios is the logic they employ to define high-value versus low-value energy. Since NEM rules do not apply in a self-consumption scenario, customers get a retail credit only for PV generation that directly offsets household loads; they get a fraction of this value for PV outflows, which the utility might credit at the wholesale or avoided cost of electricity. In a time-of-use regime, the standard NEM logic applies; however, the retail rate structure itself values energy differently over the course of a day or year, as illustrated in Table 1. Utilities intentionally design these price differences to correspond with the relative demand for energy, as these signals can help level out electricity demand and even defer infrastructure upgrades.
In a time-of-use scenario, energy has the highest value during periods of peak demand, such as weekday afternoons or evenings in summer, and relatively lower worth at other times. The prime economic driver in these scenarios is the magnitude of the difference between on-peak and off-peak energy rates. In theory, if the spread between high-value and low-value energy is large enough, energy storage customers can achieve a return on investment by storing energy during off-peak periods and discharging it when on-peak prices are in effect. In practice, battery costs are relatively high and retail energy prices are relatively low in most of the US, which works against this business case.
Demand reduction. Demand-reduction applications use stored energy to reduce instantaneous power demand. Demand reduction is one of the most attractive business cases for behind-the-meter energy storage in commercial applications. Whereas utilities typically bill residential customers strictly based on monthly energy (kWh) consumption, they bill commercial and industrial users based on both energy consumption and peak power (kW) demand, typically as measured over a 15-minute interval. For commercial and industrial customers, peak demand charges are not only the fastest-growing part of their electric bill, but also may account for up to 50% of the total. In these applications, service providers can use advanced monitoring and control capabilities to discharge stored energy from the batteries coincident with peak loads.
Though demand reduction is an excellent use case for energy storage, very few utilities factor demand into residential rate structures. A notable exception is Arizona’s Salt River Project, which recently implemented a pilot program for a residential demand–based price plan. Unless residential demand charges apply, demand reduction is not a viable market for in-home energy storage.
STORED ENERGY COSTS
The US Energy Information Administration (EIA) tracks average electricity prices by state and market sector over time. In October 2015, EIA published its recently aggregated annual energy cost data, which indicate that the average retail price of electricity in the residential sector was 12.52 cents per kWh in 2014. Drilling down on the state-level data, Washington, which gets 65% of its electricity from hydro, had the lowest residential electricity prices (8.67 cents per kWh), whereas Hawaii, which gets the lion’s share of its electricity from fuel oil, had the nation’s highest prices (37.04 cents per kWh).
So how does the cost to store energy in residential applications compare to retail electricity prices? In reality, this depends largely on the energy storage system’s use and cycling practices. However, it is possible to use the simplified formula in Equation 1 to derive a back-of-the-napkin value for the LCOS:
LCOS = Cost ÷ (Usable Capacity x Cycles x Efficiency) (1)
Cost. In the SolarPro article “Levelized Cost of Energy” (April/May 2012), Tarn Yates and Bradley Hibberd note that LCOE calculations “should include all costs that the project incurs—including construction and operation—and may incorporate any salvage or residual value at the end of the project’s lifetime.” They go on to explain how to account for project financing costs, discounted cash flow and depreciation in these calculations. As the formulas in the 2012 article evidence, detailed LCOE calculations can become quite complicated.
For the purposes of this article, we intentionally strip away many of these layers of complexity. We do not factor the time value of money into our LCOS calculations; we do not consider the residual value of the equipment at the end of its life cycle; we do not even consider O&M costs. Instead, we use the initial installed cost of the energy storage system and the associated power electronics as a proxy for the total life cycle costs.
To justify this simplified approach, we look specifically at the LCOS for a representative set of lithium-ion energy storage solutions and applications. Since vendors advertise these advanced battery appliances as maintenance-free solutions, we assume that the user does not incur any additional costs to maintain and operate the energy storage asset. Note that if you wanted to use this formula to analyze the LCOS for an energy storage system with flooded lead-acid batteries, which have lower up-front costs than lithium-ion batteries, you would have to factor in O&M costs over the life of the system. These costs vary based on labor rates and travel times.
Table 2 illustrates the sensitivity of LCOS to costs. On one hand, Green Mountain Power, Vermont’s largest electric utility, is selling turnkey Tesla Powerwall solutions—complete with a SolarEdge StorEdge inverter, autotransformer and energy meter—for an installed price of $6,500, made possible in part by a bulk purchase order for 500 units and the fact that the utility can use its own technicians to install the systems. On the other hand, TreeHouse, an Austin, Texas–based sustainable home improvement store, is the nation’s first retailer to offer Tesla’s Powerwall, which it provides through a network of licensed installation partners. According to one of these installation partners, the initial turnkey cost for a StorEdge plus Powerwall system in Austin will likely be “closer to $9,000.” This $2,500 difference in up-front costs results in a $0.16/kWh difference in the LCOS between the utility provider and the retail provider.
Usable capacity. An energy storage system’s usable capacity is primarily a function of nameplate kWh capacity, the allowable depth of discharge and battery capacity degradation over time. For example, Adara Power (formerly JuiceBox Energy) sells a residential energy storage system with a nominal 8.6 kWh lithium-ion battery. According to the company’s CEO, Neil Maguire, the maximum allowable depth of discharge for these nickel manganese cobalt oxide batteries, which Samsung manufactures, is 75%. Based on this allowable depth of discharge, the usable capacity of Adara’s energy storage system is 6.45 kWh (8.6 kWh x 0.75). However, this is the nominal battery capacity on day 1 only. This battery capacity will invariably decline over the life of the system, based in part on aging and in part on usage.
There are challenges associated with deriving a realistic value for an energy storage system’s usable capacity. Notably, the energy storage industry lacks nameplate and datasheet reporting standards and independent third-party verification requirements. This scenario is not unlike the early days of the US solar industry, when some manufacturers batched modules according to very tight nameplate power tolerances, such as +5% to −0%, while others had a very wide window, such as +5 to −10%. It is also reminiscent of the days before the CEC established independent verification test requirements for PV modules (PTC ratings) and inverters (CEC-weighted efficiency).
Without access to standardized data verified by a third party, the battery warranty may provide the best approximation of usable capacity over the life of an energy storage system. For example, Tesla nominally rates its Powerwall at 6.4 kWh when used for daily cycling. If we assume that a Powerwall cycles to capacity 3,650 times over its 10-year warranty period, the best-case LCOS falls in the $0.31–$0.43/kWh range, depending on system cost. However, if we review the Powerwall warranty, which accounts for the stepped degradation over time shown in Figure 3, we see that Tesla guarantees “18 MWh of aggregate discharge” from the battery cells. If we calculate LCOS based on the warranted battery capacity, LCOS increases to $0.40–$0.56.
Cycles. Battery life depends strongly on usage patterns, as anyone with a cell phone or laptop computer can attest. So a manufacturer might rate a particular battery for 4,000 cycles at a 70% depth of discharge or 3,000 cycles at an 85% depth of discharge. Ideally, system integrators and customers should have access to these data, either in tabular form or in graphs that plot battery cycle life in relation to depth of discharge. In practice, these data are difficult if not impossible to find.
One of the reasons battery cycle ratings are so important is that customers get the most value out of a battery that is the right size for their application. For example, sonnen guarantees all of its sonnenBatterie eco series of energy storage systems for “10,000 cycles or 10 years.” Since daily cycling accounts for only 3,650 cycles over a 10-year period, you would need to cycle these batteries an average of 2.7 times per day to approach 10,000 cycles over the warranty period. To achieve this level of usage, you would likely need to deploy the energy storage system in an application where it provides more than one service, a practice known as application stacking. An example might be an application where you are using an energy storage system for both self-consumption and time-of-use bill management or demand reduction.
Table 3 illustrates the sensitivity of LCOS to the total number of cycles. According to Greg Smith, sonnen’s senior technical trainer, the retail price for a sonnenBatterie eco 6 is roughly $12,000, and installation costs (including relocating circuits to a protected-loads subpanel) are likely to fall in the $1,500–$3,000 range. Using conservative cost assumptions—and ignoring, for the moment, battery capacity losses over 10 years—we can see that the best-case LCOS varies dramatically depending on whether the user is full-cycling the battery once per day ($0.79/kWh at 3,650 cycles) or for maximum usage ($0.29/kWh at 10,000 cycles). While this is an extreme example of how the number of cycles relates to LCOS, this basic relationship holds for all energy storage applications. Anything less than 100% resource utilization drives up the LCOS.
It is worth noting that the best-case LCOS values in Table 3 do not take into account battery capacity losses over the life of the installation. In this case, sonnen does not publish a warranted maximum aggregate discharge value or provide any information about capacity retention over time. A simple way to account for the inevitable effects of battery degradation is to apply a capacity adjustment factor that accounts for battery aging and usage. In this case, sonnen’s warranty guarantees 70% of the original rated capacity after 10 years. If we assume that we reach 70% of capacity in the maximum cycling scenario and that capacity degrades linearly over time, then the average battery capacity over the 10-year period is 83.6%. For the daily-cycling application, we assume 0.5% of battery degradation per year due to aging and 1% due to usage, which works out to a 92.6% adjustment factor. If these assumptions seem overly optimistic or conservative, simply increase or decrease the adjustment factor accordingly.
Efficiency. Charging and discharging a battery incurs an internal cost. If you charge a battery from the grid during off-peak hours and discharge it on peak, you lose some amount of energy along the way. Conversion losses in both the battery and the inverter, as well as voltage drop losses in the conductors and electrical connections, occur during periods of active charging and discharging; the battery even has self-discharge losses when it is doing no work at all. The efficiency value in Equation 1 accounts for the fact that we do not get all of the energy out of a battery that we put into it.
These round-trip efficiency losses are significant in solar-plus-storage systems, especially in comparison to losses in an interactive PV system. Today’s non-isolated string inverters have weighted efficiencies in the 96%–98% range. By comparison, Tesla states that the “beginning of life” round-trip efficiency for its Powerwall is 92.5%. While round-trip efficiency data can be difficult to find—and third-party verified weighted data reflecting real-world scenarios do not exist—these losses depend somewhat on power processing and battery configuration.
SolarEdge and Fronius, for example, offer multiport inverters that work with high-voltage lithium-ion batteries such as Tesla’s Powerwall. In these systems, the PV array and battery both connect to the dc bus of a transformerless inverter, which is very efficient but provides modest surge capacity for motor loads. By contrast, Adara Power and sonnen have designed their energy storage systems around transformer-based inverter platforms (from Schneider Electric and Outback Power, respectively) that use a 48 V nominal battery bank. While these systems are somewhat less efficient, they offer excellent surge ratings for backup loads, which is critical to customers who need to run essential equipment—say, a well pump—during a power outage. To add solar to a sonnenBatterie, integrators must ac-couple an interactive inverter with the battery-based inverter via the ac bus in the backup-loads subpanel. The Adara Power system accommodates both ac- and dc-coupled configurations. The former is most cost-effective in retrofit applications where an existing interactive inverter can process PV power. The latter uses a 600 V Schneider charge controller to integrate the PV power source. SMA, meanwhile, is releasing a new Sunny Boy Storage inverter—designed especially with the retrofit market in mind—that will allow integrators to ac-couple solar with a high-voltage battery.
As the market matures, we will likely see increased demand for something parallel to an Energy Star rating method targeting home energy storage systems. This may be a long time coming in practice, however, based on the slow progress in the multiyear international efforts to develop comparative tests and ratings for PV modules. In the meanwhile, system integrators and developers may want to take vendors’ round-trip efficiency claims with a grain of salt. Self-reported efficiency values are suspect, if only because they likely reflect best-case scenarios. In the real world, inverter loading and battery cycling is highly variable, which could reduce round-trip system efficiency in the field.
Table 4 illustrates the sensitivity of LCOS to round-trip efficiency. In this example, we assume that Adara Power’s energy storage system, warranted for 10 years or 4,000 cycles, costs $11,500 fully installed, and that the PV system bears the cost of the charge controller or inverter processing power from the PV array. According to Adara, the round-trip efficiency of each battery charge and discharge cycle is 98%, which suggests that it has chosen to publish the efficiency value for the battery management system and battery chemistry. After all, the Schneider Electric XW+ 5548, which Adara uses in its systems, has a CEC efficiency rating of 93%. Moreover, the manufacturer-reported efficiency of Schneider Electric’s 600 V charge controller is 96% in a 48 V application. Based on these efficiency ratings, the best-case round-trip efficiency for a dc-coupled Adara Power solar-plus-storage system is roughly 87.5% (0.98 x 0.93 x 0.96).
To estimate the round-trip efficiency for the ac-coupled configuration, we assumed that the interactive inverter is 96% efficient. We then looked at the battery-based inverter’s charging efficiency, which Schneider Electric describes in detail in its user manual for the XW+ 5548. In comparison to the inverting efficiency curve, the charging efficiency curve has a slightly lower peak value and a more pronounced downward slope. Based on a comparison of these curves, we estimate that the weighted average charging efficiency for the XW+5548 is roughly 91%. So in the ac-coupled solar-plus-storage configuration, we estimate that the round-trip efficiency is closer to 81.2% (0.96 x 0.91 x 0.93), which means the unsubsidized LCOS is about $0.04/kWh more than that for the dc-coupled configuration. While the storage component of a solar-plus-storage system does not automatically qualify for the 30% solar Investment Tax Credit (ITC), the post-ITC LCOS value in Table 4 illustrates the impact of the federal tax credit on qualifying systems.
LCOS USES AND LIMITATIONS
As these case studies illustrate, you can use LCOS calculations to quickly compare energy storage solutions for a specific application or to evaluate the impact of changing certain design variables and assumptions. As useful as this may be, it is important to recognize that the LCOS metric has some limitations as a basis of comparison and a project assessment tool.
Cost vs. value. In late 2015, the financial advisory and asset management firm Lazard published a detailed cost comparison of various energy storage technologies in a wide variety of applications on both sides of the meter. “Lazard’s Levelized Cost of Storage Analysis—Version 1.0” (see Resources) not only models LCOS based on current technology prices, but also looks at how LCOS might change in the next 5 years based on projected capital cost decreases. While the authors compare the LCOS for various storage technologies, including lithium-ion batteries, to that of a gas peaker plant as a baseline, they also note that LCOS does not “purport to provide an ‘apples-to-apples’ comparison to conventional or renewable electric generation.”
This is true in part because cost tells only a piece of the story. An interactive PV system, for example, clearly has a lower levelized cost of energy than an energy storage system. However, the energy storage system can keep the customer’s lights on in the event of a power outage, which is very important to some customers. A cost-oriented metric such as LCOS does not capture that value. Similarly, some energy storage systems can support specific loads or applications that others cannot. For example, adding inverter capacity to a sonnenBatterie increases system costs and negatively impacts LCOS, but could improve the value proposition for the customer if the expanded system is able to power a deep-well pump in the event of an outage. Value-based considerations are very important when comparing energy storage systems, which residential applications often deploy as a means of improving service reliability. In many energy storage scenarios, the cheapest solution may not provide the best value.
Stacking revenue and benefits. Even if we exclude factors that are difficult to quantify in dollars and cents—such as reliability or environmental attributes—it is impossible to understand the value proposition for energy storage without considering the revenue side of the equation, which can quickly get complicated. In an ideal use case, an energy storage system provides multiple revenue-generating services via application stacking, as illustrated in Figure 4. Some of these revenue streams, including time-of-use and demand reduction savings, depend entirely on variable load profiles and utility tariffs. To model these revenues, you need access to specialized software, detailed interval meter data and a database of utility rate structures.
Perhaps the most important challenge facing residential energy storage is the need to unlock additional revenue streams, which is as much a policy problem for utility regulators as it is a technology problem for manufacturers and vendors. The authors of RMI’s report on battery economics note that when you use an energy storage system for a single application, such as self-consumption or backup power, that leaves something like 50%–99% of the battery capacity unused over the life of the system. The bad news, of course, is that resource underutilization leaves potential value on the table and increases the LCOS. The good news is that removing the regulatory barriers that prevent application and revenue stacking can tilt the economics in favor of behind-the-meter energy storage.
Green Mountain Power’s pilot program offering Tesla Powerwall batteries to its customers is a good example of how an innovative utility can leverage additional value from residential energy storage systems. According to public filings, the utility estimates that its net present value for a leased Powerwall is roughly $50 per system per month over a 10-year term. (A $37.50 per month customer fee offsets the additional monthly costs associated with the Powerwall deployments.) To create this revenue stream, the utility will discharge the Powerwall batteries during “times of high market prices to help lower its energy costs,” as well as during “times of peak load to reduce significant capacity and transmission expenses.” Green Mountain Power expects that in addition to providing backup power for end users, the Powerwall deployments will “smooth grid impacts caused by a high penetration of solar energy, potentially avoiding more expensive, traditional upgrades.” The company is also deploying 10 additional units as part of a pilot microgrid project, which will “contribute to improving the reliability of the Rutland 46 kW subtransmission network during system contingencies.”
While current business models typically leverage one or two use cases for energy storage, RMI identifies “thirteen fundamental electricity services” that can benefit “three major stakeholder groups” (system operators, utilities and end users). The report also notes that “the further downstream battery-based energy storage systems are located, the more services they can offer to the system at large.” Utilities can even aggregate a network of residential energy storage systems and operate this as a virtual power plant. At the end of the day, grid parity for energy storage is more about leveraging and monetizing these many value streams than it is about achieving a LCOS lower than the retail price of electricity.
Matthias B. Krause / Berkeley, CA / matthiasbkrause [AT] gmail.com
David Brearley / SolarPro / Ashland, OR /solarprofessional.com
GTM Research and the Energy Storage Association, “US Energy Storage Monitor: 2015 Year in Review,” March 2016, greentechmedia.com
GTM Research and the Solar Energy Industries Association (SEIA), “US Solar Market Insight: 2015 Year in Review,” March 2016, greentechmedia.com
Lazard, “Lazard’s Levelized Cost of Storage Analysis—Version 1.0,” November 2015, lazard.com
Rocky Mountain Institute, “The Economics of Battery Energy Storage,” October 2015, rmi.org
Rocky Mountain Institute, “The Economics of Grid Defection,” February 2014, rmi.org