Lifetime cost | Storage Lab

Projecting the future lifetime cost of electricity storage technologies

There is consensus to use levelized cost of energy (LCOE) as a lifetime cost metric to compare energy generation technologies, such as solar, wind, and coal plants. However, there is no universally applied metric for calculating the cost of energy storage technologies. As a result, manufacturers have a hard time explaining cost advantages over their competitors, investors struggle to make educated decisions for financing, and end-users are unsure about which technology to choose.

Energy storage technologies can be used in a range of applications (e.g. frequency response, energy arbitrage, power reliability). These different applications have different operational requirements (e.g. duration of energy supply, number of activations per year) and each storage technology is differently suited to these applications based on their individual cost and performance parameters.

The best approach is to compare storage technologies for clearly defined application requirements using storage-specific lifetime cost. These lifetime cost account for all technical and economic parameters affecting the cost of delivering stored electricity. There are two forms of lifetime cost which matter:

Levelized cost of storage (LCOS) quantifies the discounted cost per unit of discharged electricity (e.g. USD/MWh) for a specific storage technology and application. It divides the total cost of an electricity storage technology across its lifetime by its cumulative delivered electricity. By doing that, the metric describes the minimum revenue required for each unit of discharged energy for the storage project to achieve a net present value of zero. The metric is used for applications that value the provision of electric energy (e.g. MWh)

Capacity cost quantifies the discounted cost per unit of power capacity provided for a certain timeframe. If represented per year this gives the annuitized capacity cost (ACC).

Figure 1 compares the LCOS of the four most competitive storage technologies in the application peak capacity (discharge duration: 4 hours, annual cycles: 300). It also shows the probabilities for each technology to be most cost-efficient calculated at 1-year intervals from 2015 to 2050. The benchmark in this application would be a gas peaking plant with a 10-15% utilisation rate, which is reported at levelized cost of energy (LCOE) of 120-200 USD/MWh.

Pumped hydro had the lowest LCOS in 2015 at just below 200 USD/MWh median (range: 150-225 USD/MWh), followed by compressed air at 250 USD/MWh median (range: 200-300 USD/MWh). However, the strong anticipated investment cost reductions for battery technologies mean that by 2030 vanadium redox flow and lithium ion are likely to be the most cost-efficient solutions for this application.

Figure 1 – Lifetime cost projections for providing peak capacity at ~15% capacity factor. Top-left: Application requirements. Bottom: Explicit LCOS projections for the four most competitive technologies, including uncertainty ranges based on Monte Carlo simulation of LCOS calculation. Top-right: The probability of these technologies having the lowest LCOS. Top-right chart includes the median LCOS of the technology with highest probability to be most cost-efficient (black line). Note that LCOS projections are based on future investment cost reductions only and disregard potential performance improvements.

The median LCOS of the most cost-efficient technology reduces from just below 200 USD/MWh (the current upper LCOE bound of gas peaker plants) in 2015 to 175 and 150 USD/MWh in 2030 and 2040 respectively. This is in line with findings of other studies and means that from 2030 energy storage solutions may be the most cost-effective solution to provide peak capacity services, in particular when accounting for the uncertainty in future natural gas prices.8 When charging for less than 50 USD/MWh (e.g. solar PV in sunny locations) and providing additional grid services, battery solutions can already be the most cost-efficient solution much earlier.

The same analysis can be conducted for 13 selected archetypical applications. The resulting overview of all technologies’ probabilities for lowest OS, and the median LCOS of the most cost-efficient technology is shown in Figure 2.

The overview reveals that the incumbent technologies which dominated electricity storage applications in the past will lose their competitiveness, e.g. pumped hydro for peak capacity, compressed air for seasonal storage or lead acid for power reliability.

Instead, by 2030 lithium-ion batteries will be the most cost competitive option in 7 out of the 13 applications. Note that these are all the applications with <4 hours discharge and <300 annual cycles. For specific applications with requirements outside of these ranges, other storage technologies will come to dominate:

High throughput, short-discharge: Flow batteries for congestion management (discharge: 1 hours, annual cycles: 1,000)
High throughput, long-discharge: Pumped hydro for renewables integration (discharge: 8 hours, annual cycles: 300)
Very long discharge: Hydrogen systems for seasonal storage (discharge: 700 hours, annual cycles: 10)
Very high annual cycle applications: Flywheels for high cycle application (discharge: 0.5 hours, annual cycles: 5,000)

Lifetime cost for frequency regulation, frequency response and high cycle is displayed as annuitized capacity cost. These services are usually reimbursed for the power they provide instead of energy. Also, for network services that require <10 seconds response time and for the small-scale consumption services, pumped hydro and compressed air are excluded from the analysis.

Figure 2 – Lowest lifetime cost probabilities for 9 electricity storage technologies in 13 applications from 2015 to 2040. Probabilities reflect the frequency with which each technology has lowest cost accounting for the uncertainty ranges identified with the Monte Carlo simulation. Left axis displays probability. Right axis displays median lifetime cost of the technology with highest probability for lowest cost. Cost are usually displayed as levelized cost of storage (LCOS). Note that there are different scales between panels.

Figure 3 moves away from the concept of clearly defined applications with discrete discharge and cycle requirements to allow a more overarching view to be taken on technology competitiveness and lifetime cost variability. It shows the technology with lowest lifetime cost and the explicit lifetime cost for all possible combinations of discharge duration and frequency requirements for 2020. The positions of selected archetypical applications are now indicated by circled letters in the spectrum.

Pumped hydro and compressed air are most cost-efficient for applications with more than 2 hours discharge duration due to relatively low energy-specific investment cost. Above ~300 hours discharge, hydrogen with even lower energy-specific cost takes the lead. Lithium ion is most cost-efficient in applications with below 2 hours discharge and below 300 cycles per year. The longer cycle life of vanadium redox flow makes it more cost-efficient between 300 and 1,000 annual cycles. Above that, flywheels take the lead due to even higher cycle life.

Figure 1 – Left: Competitive landscape showing energy storage technologies with highest probability to have lowest LCOS relative to discharge duration and annual cycle requirement in 2020. Circled letters represent the requirements of the 13 archetypical applications: BS - black start, FS - frequency response, DR - demand charge reduction, FG - frequency regulation, CM - congestion management, HC - high cycle, RL - power reliability, SC - self-consumption, PC - peak capacity, EA - energy arbitrage, TD - transmission/distribution investment deferral, RE - renewables integration, ST - seasonal storage. Colour indicates the technology with the lowest LCOS. Shading indicates how much higher the LCOS of the second most cost-efficient technology is; meaning lighter areas are contested between at least two technologies, while darker areas indicate a strong cost advantage of the dominant technology. Both axes are on logarithmic scale: x-axis with base 10 and y-axis with base 2. Right: LCOS of the most cost-efficient technologies relative to discharge duration and annual cycle requirement in 2020.

LCOS falls with higher cycles (i.e. discharge frequency). This is intuitive since more energy is discharged for the same energy/power capacity installed (i.e. investment capital deployed). The strong impact is a result of the high share of investment cost in the LCOS, which gets diluted with higher energy throughput. In addition, LCOS falls with increasing discharge duration (i.e. energy-to-power ratio). This increases the energy discharged with each cycle, however, investment cost does not decrease proportionately as it is only the energy-specific cost component which is affected. This effect diminishes at higher discharge durations where energy-specific costs already make up the majority of the total investment cost.

Figure 2 makes it even clearer that LCOS reduces with increasing utilisation (i.e. discharge hours per year). This is driven by hours per discharge (energy capacity) and discharges per year (cycle frequency). However, at maximum utilisation, energy storage applications with lower duration requirements enable lower cost solutions. Given current technologies, it is cheaper to purchase a 1-hour system and discharge it 4,380 times per year than to pay for a 4,380 hour system and discharge it once per year. The lowest LCOS in 2020 is achieved by pumped hydro at maximum utilization between 4 and 10 hours discharge duration.

Figure 2 – Graphic representation of LCOS drivers. Increase in duration (energy capacity) reduces LCOS. Increase in frequency (annual cycles) also reduces LCOS. Combination of both leads to lowest LCOS due to optimisation of investment cost (i.e. high share of energy-specific cost) and high number of annual cycles to recoup the investment. Speech bubbles indicate hours of discharge per year, i.e. hours per discharge multiplied with discharges per year.

Figure 3 shows lowest cost technologies for power provision and the corresponding minimum annuitized capacity cost (ACC) in USD/kW-year. The dominance of lithium ion for short-duration applications is more pronounced in ACC, and even expands to applications below 4 hours discharge and 1,000 annual cycles. This confirms the high uptake seen already in ancillary service applications around the world. In contrast, vanadium redox-flow batteries are not competitive anymore. This is because depth-of-discharge is not relevant for the monetization of this service as it only reimburses power capacity provided and not energy discharged. It is therefore not accounted for in this metric. This is to the advantage of lithium ion, which has a lower depth-of-discharge than its ‘direct’ competitors (i.e. pumped hydro, compressed air, vanadium redox flow).

Figure 4 highlights that low ACC values are achieved in applications with short-duration discharges and few annual cycles. Because storage technologies in power applications get reimbursed for available power capacity, rather than energy discharged, any additional cycle reduces lifetime without leading to additional revenues. Black start and frequency response are cases in point. Also, any additional energy capacity increases investment cost without directly enabling additional revenues. Indirectly, more energy capacity allows providing power for longer, which is an advantage in some services (e.g. de-rating in capacity markets). This is not accounted for in ACC and should be considered as boundary condition if applicable. However, the majority of services that reimburse for power are in the ancillary services market and require discharge durations below 1 hour.

Figure 3 – Left: Competitive landscape showing energy storage technologies with highest probability to have lowest ACC relative to discharge duration and annual cycle requirement. Colour indicates the technology with the lowest LCOS. Shading indicates how much higher the LCOS of the second most cost-efficient technology is; meaning lighter areas are contested between at least two technologies, while darker areas indicate a strong cost advantage of the dominant technology. Right: ACC of most cost-efficient technologies. In both panels, circled letters represent the requirements of the 13 archetypical applications.

Figure 4 – Graphic representation of ACC drivers. Longer discharge duration (energy capacity) or higher discharge frequency (annual cycles) add no value to power provision. Therefore, lowest ACC are achieved when minimizing both. Black start capability (BS) is an example for a service where this can be achieved.

Figure 5 and 6 project the technology competitiveness landscape up to 2040 for LCOS and ACC respectively, accounting for potential investment cost reductions of each technology. The left-hand panels include all storage technologies, while the right-hand panels exclude pumped hydro and compressed air. They are excluded, because building these technologies may not be an option for selected projects due to limited response times or limited geographic flexibility.

With continued investment cost reduction, lithium ion is projected to outcompete pumped hydro and compressed air below 8 hours discharge to become the most cost-efficient technology for most of the 13 displayed applications by 2030. At the same time, hydrogen storage becomes more cost-efficient than compressed air for long-discharge applications. Vanadium redox-flow batteries dominate in high-throughput applications with 300-3,000 annual cycles and up to 4 hours discharge.

Excluding pumped hydro and compressed air reveals that hydrogen storage would have already been most cost-efficient in 2020 for discharge durations beyond 12 hours. The remaining application space that would have been covered by pumped hydro is then dominated by lithium ion and vanadium redox flow.

Figure 5 – Competitive landscape showing storage technologies with lowest LCOS relative to discharge duration and annual cycle requirements for all modelled technologies (panels a, c, e) and excluding pumped hydro and compressed air (panels b, d, f). Circled letters represent the requirements of the 13 selected archetypical applications. Colours represent technologies with lowest LCOS. Shading indicates how much higher the LCOS of the second most cost-efficient technology is; meaning lighter areas are contested between at least two technologies, while darker areas indicate a strong cost advantage of the dominant technology. White spaces mean LCOS of at least two technologies differ by less than 5%. The modelled electricity price is 50 USD/MWh. Discount rate is 8%.

Figure 6 shows lowest cost technologies for power provision and the corresponding minimum annuitized capacity cost (ACC) in USD/kW-year. The dominance of lithium ion for short-duration applications is more pronounced in ACC, and even expands to applications below 4 hours discharge and 1,000 annual cycles. This confirms the high uptake seen already in ancillary service applications around the world. In contrast, vanadium redox-flow batteries are not competitive anymore. This is because depth-of-discharge is not relevant for the monetisation of this service as it only reimburses power capacity provided and not energy discharged. It is therefore not accounted for in this metric. This is to the advantage of lithium ion, which has a lower depth-of-discharge than its ‘direct’ competitors (i.e. pumped hydro, compressed air, vanadium redox flow).

By 2040, only three technologies cover the full application space, with hydrogen most cost-effective in applications above 16-64 hours, flywheels above 1000-3000 cycles, and lithium ion taking all the rest.

Figure 6 – Storage technologies with lowest ACC relative to discharge duration and annual cycle requirements for all modelled technologies (panels a, c, e) and excluding pumped hydro and compressed air (panels b, d, f). Circled letters represent the requirements of the 13 selected archetypical applications. Colours represent technologies with lowest LCOS. Shading indicates how much higher the LCOS of the second most cost-efficient technology is; meaning lighter areas are contested between at least two technologies, while darker areas indicate a strong cost advantage of the prevalent technology. White spaces mean LCOS of at least two technologies differ by less than 5%. The modelled electricity price is 50 USD/MWh. Discount rate is 8%.

The future projection of LCOS for the most cost-efficient technology at all discharge and frequency combinations is displayed in Figure 7. The lowest LCOS is achieved at maximum utilisation of the storage systems between discharge durations of 1-64 hours and discharge frequencies of 100 to 5,000 cycles per year. The LCOS range of 100 to 150 USD/MWh corresponds to the levelized cost of storage from new pumped hydro facilities.

The future projection of LCOS shows a proportional cost reduction across the entire discharge and frequency spectrum, despite the changing technologies that achieve these LCOS. As a result, LCOS of 100-150 USD/MWh will be achieved in five of the 13 modelled archetypical applications by 2040. The lowest ACC is achieved for short discharge duration and few annual cycle applications. For example, the application black start could be serviced for 65 USD/kW-year in 2020 and below 25 USD/kW-year by 2040.

Figure 7 – LCOS (panels a, c, e) and ACC (panels b, d, f) of most cost-efficient technologies relative to discharge duration and annual cycle requirements for all modelled technologies. Circled letters represent the requirements of the 13 archetypical applications. Colours represent LCOS range. The modelled electricity price is 50 USD/MWh. Discount rate is 8%.

Produce your own 'competitive landscape' using your technology assumptions at www.EnergyStorage.ninja