Energy storage is surging across America. Total installed capacity passed 1,000 megawatt-hours (MWh) during a record-setting , and the U.S. market is forecast to nearly double by adding more than 1,000 MWh new capacity in - adding as much capacity in one year as it did in the previous four.
You can find more information on our web, so please take a look.
However, this exponential growth has mainly been limited to vertically integrated utilities operating outside of the country’s organized power markets, which serve two-thirds of all U.S. electricity consumers. So how can energy storage plug into these markets?
In a word, revenue.
Energy storage can collect revenue in America’s organized power markets three ways: platforms, products, and pay-days. However, different projects will tap these potential revenue streams in different ways, and investors should seek nimble developers who can navigate a complex and evolving regulatory and market landscape.
In part two of this series, we’ll explore how storage will disrupt power markets as more and more capacity comes online, but first let’s cover the three ways it can tap the U.S. organized market opportunity.
Platforms: The Best Laid Plans…
Independent system operators (ISOs) go through a planning process where they identify opportunities for new transmission to improve reliability or market efficiency. Similarly, it’s normal to think about energy storage as a reliability asset, and it can become integrated as a lower-cost, non-transmission alternative to boost reliability.
Here's an example: A relatively isolated area on the grid must plan for losing a transmission line or local generator during peak demand. Rather than adding new transmission or local generation, building a storage project can carry a local grid through an emergency. If the economics add up, the project will then be built, and paid on a cost-of-service basis financed through transmissions charges.
If storage in this example plays the same role as transmission for so-called “reliability transmission expansion”, it should also enjoy an analog to “economic transmission” – transmission built to move surplus energy to constrained areas to create benefits for market buyers and sellers. But to date, only one such project exists within the U.S. independent system operators (ISOs), located near Baltimore on the PJM grid.
One reason ISOs have hesitated to fund such projects is that while “reliability” storage is tied to a definite risk of an emergency on the grid which determines how it will be used, “economic” storage requires instructions from the ISO about when to buy and sell power. ISOs worry this could challenge their market independence since the way they dispatch storage will invariably affect prices, and could make them look like self-dealing market participants.
However, ISOs already regulate power flow over transmission lines, which certainly affects power prices. When a new transmission project is proposed to relieve congestion in an area of the grid with high demand (and thus high prices), local generators are first in line to complain about lost revenue.
What preserves ISO independence in this case is transparent cost-benefit-analysis and security constrained economic dispatch with financial transmission rights – a standard methodology for fairly moving power across transmission lines and distributing revenue from arbitraging local price differences.
If or when markets start doing more multi-period dispatch, they can dispatch storage in the same way, according the transparent optimization, and assign financial storage rights to whomever pays the costs of economic storage.
Products: Fee for Services
While ISOs are uncomfortable paying for storage services through transmission access charges that passively incorporate storage into the grid, they have been receptive to storage competing to provide fixed services like fast frequency response, capacity, or regulation that projects can compete to provide on a “technology-neutral” basis. But keep in mind these services were defined by markets before batteries and other clean technologies like renewables changed the game.
Theoretically, fitting energy storage into these technology-neutral products should be simple. But storage resources are energy limited (they can’t just convert fuel to electricity ad infinitum), they must be charged, they take more energy to charge then they provide back, and they may be entirely driven by power electronics (no spinning inertia).
These differences mean existing market product definitions are often ill-suited to include storage, and while most incumbent participants often provide ancillary services for just a fraction of their revenues, storage projects dedicated to a single service (such as regulation) could have their entire business model upended by simple rule changes.
Storage resources also have attributes that are not always valued in markets, like how fast they can change their output, their ability to reduce air pollution, or the quick and modular pace at which they can be deployed. These attributes provide grid benefits but need revised power market rules to be properly valued. The standard equivalence for utilities between batteries and natural gas peakers seems to require a 1:4 power ratio, i.e. a 1 megawatt (MW)/4 MWh battery, so you might expect that product definition.
However, shoehorning batteries this way is not necessarily economically efficient – some peak needs may last longer, some may be more sporadic, and a battery’s highest value application may involve a different power ratio.
Collecting storage revenue by providing grid-need products will always be dependent on the fine print. As a new competitive entrant to most market, storage – especially battery storage – is not always in the best position to make sure rules value them at their best.
Pay-days: Profiteer or Just an Independent Businessman?
One way for storage resources to avoid being shoehorned into the wrong glass slipper is to compete directly in energy markets. What could be simpler than arbitrage: buy low, sell high?
Unfortunately, today’s markets just don’t provide enough revenue this way. Consider daily wholesale electricity price differentials in two ISOs with the most market spikes, California's CAISO and Texas' ERCOT, where crudely estimated annual revenues from buying low and selling high each day (with no roundtrip losses) come out to $10-20 per kilowatt-hours(kWh) year, not quite enough to be in the money yet but close to some of the prices we see coming out of vertical utilities like NV Energy’s recent announcement to add 100 MW of battery storage.
One thing is clear: The closer to a real-time market storage operates in, and the higher the power ratio, the more revenue is available from arbitrage. For example, a battery storage unit with a 4:1 power ratio and 20% round-trip losses operating in the Houston load-zone real-time market could be making as much as $57/kWh-year. This system would likely cost $300-400/kWh, making it an attractive investment, especially with high prices expected across ERCOT in coming summers.
This contrasts with other ISOs, and highlights the efficiency with which energy-only power markets can point to where investments have the most value.
Even if energy arbitrage revenues become sufficient to support storage investments, today’s markets still maintain some barriers. Not all ISOs offer the right kind of market “participation model” to offer efficiently in the markets. The Federal Energy Regulatory Commission’s (FERC) recent Order 841 directly addresses this, and the storage industry is eagerly awaiting new tariff structures and participation models in response.
Still, markets must contend with the fact that storage resources are energy limited, which begs the question: how should they play in markets? Most storage today bids on an opportunity cost basis, and will buy or sell from the market based on its state of charge: If the battery is low, its bids may not be structured to buy right away in case prices go lower, and if the battery is high and could provide power, its bids might make it more likely to wait for higher prices to discharge.
Opportunity-cost based bids may efficiently dispatch batteries for maximum system benefit, but such an approach inherently accepts a battery resource’s right to withhold its capacity. As more and more storage appears in markets as the marginal price-setting resource this may become an issue from a market-monitoring perspective.
The rapid pace of machine learning improvements mean storage bid patterns could be determined by software black-boxes that are impossible for market monitors and regulators to understand, or create strange market artifacts like stock market “flash crashes” we’ve seen with increased algorithmic participation.
One possible route to resolving these issues would be for ISOs to increase their use of probabilistic multi-period optimization in market dispatch algorithms. Then the ISO can be in charge of dispatching the battery in the most optimal way over time (hence multi-period), lowering the need for opaque and potentially problematic bid patterns.
Today’s Energy Storage Opportunity, Tomorrow’s Energy System Disruptor
Energy storage has jumped from tomorrow’s clean technology to today’s investment opportunity, but the industry’s true potential has yet to be tapped. As investors consider energy storage, they should seek nimble projects capable of navigating the complex and ever-evolving regulatory and market landscape.
And as more and more energy storage comes online, ISOs will need to evolve through new rules and market structures to accommodate the technology’s potential. In part two of this series, we’ll explore two ways energy storage will be a disruptor.
Can I use a simple energy calculation when selecting a supercapacitor for a backup system?
The simple energy calculation will fall short unless you take into account the details that impact available energy storage over the supercapacitor lifetime.
In a power backup or holdup system, the energy storage medium can make up a significant percentage of the total bill of materials (BOM) cost, and often occupies the most volume. The key to optimizing a solution is careful selection of components so that holdup times are met, but the system is not overdesigned. That is, one must calculate the energy storage required to meet holdup/backup time requirements over the lifetime of the application, without excessive margin. This article presents a strategy for choosing a supercapacitor and a backup controller for a given holdup time and power, considering the vagaries of supercapacitors over their lifetimes.
Electrostatic double-layer capacitors (EDLC), or supercapacitors (supercaps), are effective energy storage devices that bridge the functionality gap between larger and heavier battery-based systems and bulk capacitors. Supercaps can tolerate significantly more rapid charge and discharge cycles than rechargeable batteries can. This makes supercaps better than batteries for short-term energy storage in relatively low energy backup power systems, short duration charging, buffer peak load currents, and energy recovery systems (see Table 1). There are existing battery-supercap hybrid systems, where the high current and short duration power capabilities of supercapacitors complement the long duration, compact energy storage capabilities of batteries.
Table 1. Comparison Between EDLC and Li-Ion Batteries *To preserve reasonable lifetime Feature Supercapacitors Li-Ion Battery Charge/Discharge Time <1 s to >10 s 30 min to 600 min Termination/Overcharge — Yes Charge/Discharge Efficiency 85% to 98% 70% to 85% Cycle Life 100,000+ 500+ Min to Max Cell Voltage (V) 0 to 2.3* 3 to 4.2 Specific Energy (Wh/kg) 1 to 5 100 to 240 Specific Power (W/kg) 10,000+ to Temperature (°C) –40°C to +45°C* 0°C to +45°C charge* Self-Discharge Rate High Low Intrinsic Safety High LowIt is important to note that higher temperatures and higher cell voltages in supercaps decrease a supercap’s lifetime. It is important to ensure that the cell voltages do not exceed temperature and voltage ratings, and that these parameters remain within desired operation levels in applications where supercapacitors are stacked or when the input voltage is not well regulated (see Figure 1).
It can be difficult to achieve a robust and efficient solution using discrete components. In contrast, integrated supercap charger/backup controller solutions are easy to use and typically provide most or all of these features:
Analog Devices has an extensive lineup of integrated solutions that incorporate all necessary circuitry to cover the fundamentals of your backup system in a single IC. Table 2 summarizes the features of some Analog Devices supercap chargers.
Table 2. Feature Summary of Integrated Supercap Charger Solutions *Can be configured for more than four capacitors LTC LTC LTC LTC LTC VIN (V) 1.8 to 5.25 2.9 to 5.5For applications with 3.3 V or 5 V supply rails, consider:
You will get efficient and thoughtful service from Powrloo.
For applications with 12 V or 24 V supply rails, or if you require backup power beyond 10 W, consider:
If your system requires a main buck regulator for 3.3 V or 5 V rails with a built-in boost converter for backup using a single supercapacitor or other energy source for temporary backup or ride-through, you should consider:
Analog Devices also has many other constant current/constant voltage (CC/CV) solutions that can be used to charge a single supercapacitor, electrolytic capacitor, Li-Ion battery, or NiMH battery. You can find more supercapacitor solutions on analog.com.
For more information on other solutions, please contact your local FAE or regional support.
When designing a supercapacitor energy storage solution, how big is big enough? To limit the scope of this analysis, let’s focus on the classic holdup/backup applications used in high end consumer electronics, portable industrial equipment, energy metering, and military applications.
A good analogy for this design task would be a hiker who wants to determine how much water to carry on a day-long hike. Less water at the start certainly makes going uphill easy, but he may run out of water too early, especially for a difficult hike. On the other hand, a hiker carrying a large bottle of water must endure the additional weight, but will likely stay hydrated throughout the duration of the trip. The hiker may also have to take weather into account: more water on a hot day, less when cool.
Choosing a supercapacitor is very similar; holdup duration and load are important, as is ambient temperature. Furthermore, one must take into account the lifetime degradation of the nominal capacitance and the inherent ESR of the supercapacitor. Generally, the definition of the end-of-life (EOL) parameters for supercapacitors are:
These two parameters are important to the following calculations.
To size your power components, it is important to understand your holdup/backup load specifications. In the case of a power failure, for example, the system might disable noncritical loads, so that energy can be shuttled to key circuits, such as those that save data from volatile to nonvolatile memory.
Power failures come in many forms, but generally backup/holdup power must enable the system to gracefully shutdown in the face of a persistent failure or continue to operate through a transitory power failure.
In either of these cases, the component sizing must be worked out based on the sum of the loads that requires support during backup/holdup and the time those loads must be supported.
The amount of energy that is required to holdup or backup the system:
The stored energy in a capacitor:
Common sense design dictates that the energy stored in the capacitor must be greater than what is required for holdup or backup:
This approximates the size of the capacitor, but is not sufficient to determine the size for a truly robust system. Key details must be determined, such as the various sources of energy loss, which ultimately translate to greater required capacitance. Energy losses fall into two categories: those due to dc-to-dc converter efficiency, and those from the capacitor itself.
The efficiency of the dc-to-dc converter must be known for the condition where the supercapacitor is powering the load during holdup or backup. Efficiency depends on the duty cycle (line and load) conditions and can be obtained from the controller data sheet. The devices noted in Table 2 above have a peak efficiency of 85% to 95%, which can vary over the load current and duty cycle during the holdup or backup.
Supercapacitor energy loss amounts to the energy we cannot extract from the supercapacitor. This loss is determined by the minimum input operating voltage of the dc-to-dc converter. This is dependent on the topology of the dc-to-dc converter and is called the dropout voltage. This is an important parameter to consider when comparing integrated solutions.
Taking the earlier calculation for the energy of a capacitor and subtracting the energy unavailable below VDropout results in:
What about VCapacitor? It seems obvious that setting VCapacitor to near its max rating would increase the stored energy, but this strategy has serious drawbacks. Often, supercapacitors have an absolute maximum voltage rating of 2.7 V, but the typical value is 2.5 V or less. This is due to the lifetime consideration of the application and its specified ambient temperature of operation (see Figure 2). By using a higher VCapacitor in a higher ambient temperature, the lifetime of the supercapacitor is degraded. For robust applications requiring a long operating lifetime or operation at relatively high ambient temperatures, a lower VCapacitor is best. Individual supercapacitor suppliers usually supply characteristic curves for estimated lifetime based on clamping voltage and temperature.
The third effect that must be taken into consideration is not so obvious: the maximum power transfer theorem. To obtain maximum external power from a supercapacitor source with an equivalent series resistance (see Figure 3), the resistance of the load must equal the resistance of the source. This article uses the words out, backup, or load interchangeably as all three mean the same thing in this case.
If we take the diagram in Figure 3 as a Thevenin equivalent circuit, we can easily calculate the amount of power dissipated across the load via:
To find the maximum power transfer, we can take the derivative of the previous equation and then solve for the condition when it is zero. This is the case when RSTK = RLOAD.
Allowing RSTK = RLOAD, we can obtain:
This can also be approached intuitively. That is, if the resistance of the load is greater than the source resistance, the load power is reduced, since the total circuit resistance goes up. Likewise, if load resistance is lower than source resistance, then most of the power is dissipated in the source due to a lower total resistance; similarly, the amount dissipated in the load is reduced. Therefore, deliverable power is maximized when source and load impedance are matched for a given capacitance voltage and a given stack resistance (ESR of the supercapacitors).
There are implications with regard to the usable energy in a design. As the ESRs of the stacked supercapacitors are fixed, then the only value that varies during backup operation is the stack voltage and, of course, the stack current.
To satisfy the backup load requirements, as the stack voltage decreases, the required current to support the load increases. Unfortunately, increasing currents beyond the defined optimum level reduces the available backup power, as it increases the losses in the ESR of the supercapacitors. If this effect occurs before the dc-to-dc converter reaches its minimum input voltage, it translates into additional loss of usable energy.
Figure 5 shows the available power as a function of VSTK, assuming an optimal resistance matching to the load, and the graph of 25 W of backup power. This graph can also be viewed as a unitless time base: as the supercapacitors satisfy the 25 W of required backup power, the stack voltage decreases as it discharges into the load. At 3 V, there is an inflection point at which the load current is beyond the optimum level, decreasing the available backup power for the load. This is the maximum deliverable power point of the system, and at this point, losses in the ESR of the supercapacitors increase. In this example, 3 V is significantly higher than the dropout voltage of the dc-to-dc converter, so unusable energy is due entirely to the supercapacitor, leaving the regulator underutilized. Ideally, the supercap reaches the dropout voltage, so the system’s ability to provide power is maximized.
Taking the earlier equation for PBACKUP, we can solve for VSTK(MIN). Likewise, we can also take into consideration the efficiency of the boost converter and add it to this equation:
With this lower limit VSTK(MIN), we can establish a capacitor utilization ratio αB, which is derived from the maximum and minimum cell voltage:
Not only is the supercapacitor capacitance vital for determining the backup time, but the ESR of the capacitor is as well. The supercapacitor’s ESR determines how much of the stack voltage can be used for the backup load, also known as utilization ratio.
As the backup process is a dynamic process in terms of input voltage, output current, and duty cycle, the complete formula for required stack capacitance is not as simple as the earlier versions. It can be shown that the final formula is:
where η = Efficiency of the dc-to-dc converter.
The concepts and calculations to this point can be translated into a supercap backup system design methodology:
For a system that must reach a certain lifetime, the previously described methodology must be modified with EOL values, generally 70% of CNOM and 200% of ESRNOM. This complicates the math, but existing spreadsheet tools are available on product webpages for most ADI supercapacitor managers.
Let’s use a simplified methodology with example using the LTC:
Based on the initial guess of 25 F capacitance, we obtain the required four seconds of backup time (with an additional 25% margin) using nominal values. However, if we consider the EOL values of ESR and capacitance, our backup time drops to almost half. To obtain four seconds with the EOL values of the capacitors, we must modify at least one of our input parameters. Since most of them are fixed, the capacitance is the most convenient parameter to increase.
The necessary increase toward 45 F seems large since the nominal values provide a comfortable nine seconds of backup. However, with the addition of CAPEOL and ESREOL, and the resulting minimum stack voltage of 6.2 V, there is a sharp degradation to half of the backup time at EOL. Nevertheless, this meets our four second requirement for holdup time with an additional 5% margin.
The LTC and LTC offer additional telemetry features via an integrated ADC. These parts can measure the system voltages, currents, capacitance, and ESR of the supercapacitor stack. Capacitance and ESR measurements are performed with minimal impact to the system while it is online. Device configuration and measurements are communicated via I2C/SMBus. This enables the system processor to monitor important parameters over the life of the application, ensuring that available backup power meets the system requirements.
The LTC’s and the LTC’s capability to measure the capacitance and ESR of the supercapacitor stack in real time enables the user to reduce the clamp voltage when the capacitors are new and easily meet the backup requirements. The processor receiving the telemetry data can be programmed to implement the previously shown calculations. This would enable the system to calculate, on-the-fly, the minimum necessary clamp voltage to satisfy the backup time, considering real-time capacitance and ESR. This algorithm would further enhance the lifetime of the supercapacitor backup system, because, as shown in Figure 2, at elevated temperatures, the lifetime of the supercapacitors can be significantly increased by even a small decrease in the clamp voltage.
Lastly, the LTC features a hot swap controller function for protection purpose. The hot swap controller uses back-to-back N-channel MOSFETs to provide foldback current limiting, which reduces inrush current and short circuit protection in highly available applications.
Calculating the capacitance values required to meet backup specifications can be approached as a simple power needed, power stored problem by using the basics of energy transfer at nominal values. Unfortunately, this simple approach falls short when you consider the impact of maximum power transfer, a capacitor’s EOL capacitance, and ESR. These factors greatly impact the available energy in a system over its lifetime. Using ADI’s integrated supercapacitor solutions and a number of available backup time calculation tools, analog engineers should have the confidence to design and build reliable supercapacitor backup/holdup solutions that meet design requirements over an application’s lifetime with minimum impact on cost.
For more information, please visit energy storage sports equipment.