What is energy storage?

Solar panels and wind turbines at dusk
Energy storage is “…the capture of energy produced at one time for use at a later time to reduce imbalances between energy demand and energy production.” [1] This can apply across the full spectrum of timebase and energy density, which means energy can be stored for nanoseconds or years and in applications ranging from picowatts to gigawatts of power.

The figures below compare utility-scale batteries to supplement solar photovoltaic (PV) energy capturing to the kind of board-level components that act as smaller, more temporary energy storage for filtering, bootstrap, (near) lossless energy commutation, resonance, and other circuit-level applications.
Solar panels (left) and circuit solution (right)
Fig. 1: Comparison of Macro (i.e., utility scale) to Micro (i.e., board/circuit scale) Energy Storage Solutions
Most discussions around energy storage tend to focus on battery storage solutions, but we should also remember that there are numerous forms of energy storage available even if a deep dive on all is a limitation in this format. From an electronics perspective, energy can be stored in the form of electrochemical potential in rechargeable batteries, as voltage in capacitor and supercapacitor devices, as well as current in magnetic devices (i.e., inductors and transformers).

Dynamic energy can be stored in both kinetic (i.e., flywheels) and potential forms (i.e., water reservoir on top of a mountain). Energy is stored in one form for either convenience, efficiency, or cost purposes, even if it is eventually converted into electrical energy. An example of this can be thermal energy stored in heated salts for extraction later, and boiling of water to spin a turbine, also known as a form of phase-change material energy storage. It is not uncommon to use compressed air as a form of energy storage, even for utility-scale applications [2].

While the definition may not be as clear, there can be more direct forms of energy storage and utilization, such as repurposing the “waste” heat from a large data center to be pumped directly into nearby households exposed to severe cold weather in winters.

Given that energy storage covers a vast expanse of components and applications, it is important to breakdown just what is implied by certain solutions in terms of their anticipated performance and design nuances, to ensure they are optimally utilized and safely executed. In this discussion, we shall compare/contrast a variety of solutions, but only after we first define some key terminologies and understand the important factors to consider when looking into various energy storage components and solutions.

It should be noted there are some aspects of energy storage that apply universally, and there are some that are very component- and/or application-specific. For instance, there is no Moore’s Law to energy storage as it is mostly at the whim of chemistry and physics. The point is energy storage density only doubles roughly every decade, where the semiconductor world is more accustomed to densities doubling every 18 to 24 months.

From a design perspective, this involves consideration with the expectation of extremely disaggregated development schedules and roadmaps for energy storage, where elements on the load side (e.g., system power budgets/densities) advance on a timetable more closely aligning to those associated with a Moore’s Law-like trajectory.

From the universal perspective, load demand only tends to increase over time. Hence, this tends to put an increasing demand on the energy density of the energy storage solution(s), though the overall impact very much depends on the application. Even if the trend is universal in nature, the methodologies for dealing with the trend can vary greatly.

For instance, critical backup power for a data center or building will be architected and distributed very differently from safety capacitors used to filter and shunt energy in a common-mode electromagnetic interference (EMI) filter. Narrowing the window of consideration to only energy storage in power supplies still provides a wide consideration of energy storage solutions, design parameters, and therefore reliability impacts as well. In power supplies, energy storage devices may also act as critical safety devices, and therefore be subject to specific standards for design de-rating, test/qualification requirements, and assumptions for thermal impacts on component lifetime calculations.

Batteries vs. Capacitors

In terms of electrical charge storage, solutions tend to be under the categories of either a capacitor or a battery, although there are devices that combine both features. Practically any two, parallel plates with an insulating but polarizable material, or dielectric in between serve as the fundamental design of a capacitor. A capacitor stores charge physically as an electric field between the plates. However, if the cell instead stores the charge electrochemically rather than physically, by using an ion-conductive electrolyte between the terminals, then we tend to refer to this as a battery. More specifically, a battery is composed of the half-cell potentials (e.g., voltages) between dissimilar materials that combine to form a cell unit with a terminal voltage.

Batteries are classified as “wet” or “dry” cells by the characteristics of their electrolyte. A wet cell is one having a liquid electrolyte and a dry cell contains a solid-state electrolyte. Dry cells tend to last longer and fail less destructively since the electrolyte does not tend to degrade over time, due to aging/thermal effects and/or form catastrophic (e.g., thermal runaway) shorting conditions, even if the isolation barrier between cathode and anode electrodes is compromised.

Another key classification of batteries is whether they are considered primary or secondary storage cells. A primary battery is a single-use type that is non-rechargeable, while a secondary battery is a multi-use type that is rechargeable. In just about any energy storage application, secondary batteries are preferred over primary ones for purposes of reuse and mitigation of waste (especially the hazardous materials contained in most battery chemistries), though primary batteries (typically alkalines) are still very prevalent due to being considered more economical if they can meet a system’s lifetime power needs with little to no consideration for replacement. When thinking of lower-power applications such as in the Internet of Things (IoT) or Industrial IoT (IIoT), the push to mitigate primary batteries is especially important when we consider scales of many billions or even 1 T devices in the near future.

Do Not Underestimate Rechargeable Batteries (Pun Intended!)

It may be common to think of batteries as some simple 2-terminal dc-source device, but there is so much more to them than that. Many energy storage solutions are oversimplified and thought of in these terms, though the reality of device characteristics may dictate a very different approach. A secondary battery is the perfect example of this point since there are many parameters and figures of merit that determine everything, from capacity to impedance to cycle life to safety performance. Some of these items are listed/defined in the table below.

TERM/FIGURE OF MERIT DEFINITION IMPACT
STATE OF CHARGE (SOC) Battery charge level relative to capacity (based on open-circuit terminal voltage), 0–100%
  • Overall, most common figure of merit for characterizing a battery’s remaining capacity
C-RATE Battery charge or discharge rate, typically a min/max spec given by battery’s spec sheet and expressed as a ratio of the battery’s capacity (e.g., a 2.0 C max discharge rating for a 40 mAh-rated cell means the max discharge rate is 80 mA)
  • Determines the min/max charging/discharging rates the battery will support and still guarantee specs/reliability
  • Typically, values are given for short pulses (e.g., much higher currents) and steady-state currents
FAST CHARGE RATE Current limit (typically set by the battery management system [BMS]) for the constant current portion of the charge cycle
  • Determines how quickly most of the battery’s capacity is charged up before transitioning to constant voltage charging mode
  • Typically a compromise between charge time and overall cycle/capacity life
DEPTH OF DISCHARGE (DOD) Battery discharge level relative to capacity, 0–100% (opposite of SOC)
  • Same usage as SOC, but in terms of characterizing how much of a battery’s capacity has been utilized
CYCLES Number of charge/discharge cycles supported before battery is considered out of spec (minimum capacity)
  • Cycle life is one of most important battery characteristics, impacted by numerous variables intrinsic to the chemistry and impacted by application/environmental factors
EQUIVALENT SERIES RESISTANCE (ESR) Intrinsic, internal resistance (typically ac or frequency-dependent resistance) of cell as measured at terminals
  • Determines self-discharge (a.k.a. shelf life) of battery
  • Explains why batteries (such as Li-Ion) tend to heat up as their ESR increases exponentially with decreasing SOC
CONSTANT-VOLTAGE CHARGING The BMS controller applies a constant voltage to the battery, while the cell organically draws current based on charge transfer
  • Typically toward the end of a charging cycle or in the “top-off” mode
CONSTANT-CURRENT CHARGING BMS controller applies a constant current to the battery, while the cell charges to end-of-charge target potential
  • Typically toward the beginning of a charging cycle when battery is in a low SOC
CELL BALANCING Battery packs (even dual-cell supercaps) may require cell terminal voltages to be within certain range of adjacent cells, even if there is variability in part-to-part capacity
  • Balancing is done for optimal operation and reliability performance by bringing all cells in a pack to a relatively similar SOC
  • Balancing also helps to mitigate ESR mismatches, which decreases risks of unsafe usage
Table 1: Common Terms Related to Secondary Batteries & Typical Usage

As a reminder, this table is not fully comprehensive in terms of all the parameters and terminologies associated with batteries, but it covers the key items. Now that we have a better understanding of some common terms/metrics, we can to look at an example of a secondary battery discharge curve, as shown in the figure below.

While there is far too much information here to fully review in detail, one should note a handful of things: 1) The voltages and currents much be carefully controlled based on the various charge states, which is typically the main job of the BMS circuit/power management IC (PMIC); 2) Precondition current thresholds tend to be rated for 10–20% of the max (a.k.a. fast charge) current, which can start to enter the noise floor of the BMS in low-capacity batteries; 3) From a low, starting SOC (i.e., low terminal voltage), a cycle starts in constant-current mode and transitions to constant-voltage mode when nearly recharged (i.e., SOC > 80–90%), equivalent to a “top-off mode,” and it is also used for cell balancing (where appropriate).
Charging step diagram and current/voltage curves
Fig. 2: Example Secondary (Rechargeable) Li-ion Cell Discharge Curve [3]
As if all these terms and charge control states are not confusing enough, now consider all these characteristics are different for each and every kind of battery chemistry out there. This means one must completely understand these operational and charge/control nuances for their specific chemistry, in order to properly implement (in a cost-effective and safe manner). Lithium-ion/polymer (Li-Ion/Li-Po) is treated differently than nickel-metal-hydride (NiMH), lithium-iron-phosphate (LiFePO4 or LFP), or sealed lead acid (SLA) solutions.

Perhaps you have designed and built a battery-based system before and found huge discrepancies between the calculated capacity/life and the one demonstrated. This “sticker shock” scenario of measured data falling far short of the early calculations is very common, because there are so many different factors to consider that most necessary variables and de-rate factors are not initially taken into consideration for system energy storage calculation needs. Also, many batteries need to be “conditioned”—passed through several controlled charge/discharge cycles after initial manufacturing—before they can reach their full potential (pun intended!)

In general, a designer will try to determine a reasonable, steady-state average current draw (regardless of actual loading profile) and divide by the spec-sheet and max-rated capacity to estimate battery life. As alluded to above, this simple estimation methodology likely neglects the specific properties of the battery chemistry proposed. Furthermore, this approach likely also dismisses other manufacturing and environmental factors that can have huge impacts on battery performance, which will leave critical deration factors unaccounted for.

The extremes of temperature alone can necessitate wild swings in capacity calculations to ensure the meeting of system load demands under all operating and environmental conditions. All these design gaps tend to increase the error between calculated and demonstrated life.

Scales of Energy Storage

Emergency power supply diagram with response times
Fig. 3: Data Center Ascending Backup Power Levels and Response Times, courtesy of PowerRox [4]
When architecting an energy storage solution, it is important to not only consider the amount of energy required to supply but also the time scales at which the energy needs to be supplied. This energy storage response time tends to directly correlate between the amount of energy required and the response time for delivering said energy.

A data center is a great case study in this regard because it tends to have multiple layers of energy storage, each serving a different purpose, and therefore on a different response time scale. The graphic below reviews this spectrum going from a smaller energy demand with the lowest response time to indefinite energy/time backup. The cost ($/Wh) of these solutions tend to be inversely proportional to the amount of energy and response time. In other words, the more available modular solutions tend to be the most expensive ones too.

It should be noted energy storage requirements are sometimes determined purely through regulatory or governmental policy agencies, even if not necessarily the most pragmatic in the application. A good example of this is Californian State Senate Bill 431 [5], which was introduced in 2019 in a panicked response to major power outages due to wildfires, which disrupted telecommunications.

The idea was to require major network operators (MNO) to provide a minimum of 72 hours of backup power for all base stations. While well-intentioned, this flat requirement does not necessarily make sense (for example, does not matter if the base station is powered when the rest of the network/core remains unpowered), yet it is very costly in terms of resources to implement and maintain. The false peace of mind is a more political topic, beyond the scope of this discussion.

A lot of confusion has been attributed to the many different regulations for grid-tied inverters, and how each state dictates the utilization of energy storage, especially for home PV applications in which the owner may not necessarily mandate how stored energy is distributed and when. Though these are focused examples, the point is that energy storage is sometimes “thrown” at a problem as a knee-jerk response without considering the overall pragmatics or taking the time to meet application needs, while reducing expensive energy storage solutions as much as possible to meet those needs.

Different Applications within Energy Storage

Storing energy can have a wide variety of applications and value-added purposes. Critical or backup energy storage is probably the most common, especially at higher power levels. In power supplies, energy storage devices are not only used to store energy for use, when demanded by the load, but also for filtering, meeting transient requirements, biasing control ICs, and even facilitating near-lossless energy commutation in resonant topology applications.

From individual systems up to utility-scale levels, energy storage is increasing for usage in peak shaving applications. This means localized, responsive, rechargeable energy storage (typically Li-Ion batteries, though supercaps also perform this function when the extremes of response and cycle life are required) is used to meet only infrequent, peak demand needs. Therefore, the rest of the infrastructure (including backup power) can be minimized because it can all be designed to (what is typically) a much lower, steady-state demand, while using the peak power storage to guarantee uptime/availability/reliability requirements.

A modularized implementation of this is something called a battery backup unit (BBU) inverter for systems in which only the individual system backup is supplemented by a BBU that sits directly on the shared output voltage bus, on the main power supply. The BBU can handle short-term holdup with literally no delay time, which enables upstream energy storage (such as a flywheel or building-scale storage), and more time to initialize and take over the load. Disaggregating this short-term holdup from the system power supplies allow the supplies to be smaller, cheaper, and denser with the mitigation of bulky, (typically) electrolytic capacitors that accomplish the same backup purpose when internal to the power supply.

In many residential PV applications, energy storage serves more of an economic role than a technical one. Before the days of ubiquitous, inexpensive PV panels, produced in high volumes for residential applications, home-scale or building-scale PV was practically only for off-grid applications, in which the utility grid was either not available or not economically viable to extend. This means PV installations were very expensive, and the user was typically dependent on this configuration (now known as an islanded microgrid) for their power needs. An inverter would attempt to source as much power as possible from the PV array during the day, then store excess energy in the batteries to be available when the array was not producing (i.e., at night or overcast day) and/or the load demand exceeded supply. Most residential PV installations today are for folks that are already tied to the grid and are just looking to supplement their utility bill. In other words, people who use PV to run their grid-tied power meter backwards, perhaps even generating excess credit that pays out at the end of the year. Economical optimization of this arrangement could mean storing the energy at times when supply is plentiful, then waiting to sell it back to the utility company when the dynamic cost of energy is higher.

Energy Storage for Sustainability

Energy storage can be utilized for meeting many sustainability objectives. As already discussed, mitigating the need for primary batteries and reducing overall infrastructures are worthy objectives for energy storage solutions. In fact, most major initiatives for improving sustainability commonly have a major dependency on energy storage in one form or another. With the fight against climate change comes a big push for renewable energy sources and energy storage is absolutely critical, whether it be used to stabilize/balance the grid against intermittent energy sources (i.e., wind, sunlight) or store excess energy for transfer/export to other regions and applications that otherwise use less-sustainable sources.

The major industries you hear about leading the charge on sustainability are focused on electrified transportation and/or manufacturing automation. It is fairly obvious how electrified vehicles and airplanes are dependent on energy storage, but perhaps less obvious is how energy storage increases its value-add in the “factory of the future” (a.k.a. Smart Factory or Industry 4.0). Factory automation typically implies an increase in the deployment of robotics, which tend to be dense and dynamic loads, and the exacerbation of the need for the peak shaving techniques described above. The increased density of the energy footprint will also be driven by increased load demand from higher levels of computation and networking, all of which can be sustainably optimized with intelligent placement and exploitation of energy storage solutions. At a certain point, it will be impractical to try and size all the utility infrastructure to deliver peak demand at all times, and therefore will make localized energy storage a critical enabler for energy-dense load footprints.

Though there are storage solutions comprising organic, non-toxic materials, the majority of most commonly used solutions today still contain hazardous materials. This is in addition to limited resources such as rare-earth metals and finite resources. It is common for energy storage solutions to be produced by manufacturing processes that are energy intensive and use dangerous chemicals as both inputs to and outputs of these processes. To assess the true sustainability of energy storage applied to a particular use case, these end-to-end or “embodied energy” considerations must also be taken into account.

References

[1] Wikipedia contributors, "Energy storage," Wikipedia, The Free Encyclopedia, https://en.wikipedia.org/w/index.php?title=Energy_storage&oldid=1086908512 (accessed May 11, 2022).
[2] H. Lund, G. Salgi, "The role of compressed air energy storage (CAES) in future sustainable energy systems". Energy Conversion and Management. (May 1, 2009). 50 (5): 1172–1179. doi:10.1016/j.enconman.2009.01.032.
[3] Qorvo ACT2801 Spec Sheet. [Online]. Available: https://www.qorvo.com/products/d/da006751.
[4] B. Zahnstecher, “Module PR-2: Data Center Structure Overview,” PowerRox Training Module. Last updated 1/25/19.
[5] M. McGuire, S. Glazer, "SB-431 Telecommunications service: backup electrical supply rules." CALIFORNIA LEGISLATURE—2019–2020 REGULAR SESSION. [Online]. Available: https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=201920200SB431.
용도