How do you store renewable energy so it’s there when you need it, even when the sun isn’t shining or the wind isn’t blowing?
Giant batteries designed for the electrical grid – called flow batteries, which store electricity in tanks of liquid electrolyte – could be the answer, but so far utilities have yet to find a cost-effective battery that can reliably power thousands of homes throughout a lifecycle of 10 to 20 years.
Now, a battery membrane technology developed by researchers at the U.S. Department of Energy’s Lawrence Berkeley National Laboratory (Berkeley Lab) may point to a solution.
As reported in the journal of Joule, the researchers developed a versatile yet affordable battery membrane – from a class of polymers known as AquaPIMs.
This class of polymers makes long-lasting and low-cost grid batteries possible based solely on readily available materials such as zinc, iron, and water.
The team also developed a simple model showing how different battery membranes impact the lifetime of the battery, which is expected to accelerate early stage R&D for flow-battery technologies, particularly in the search for a suitable membrane for different battery chemistries.
“Our AquaPIM membrane technology is well-positioned to accelerate the path to market for flow batteries that use scalable, low-cost, water-based chemistries,” said Brett Helms, a principal investigator in the Joint Center for Energy Storage Research (JCESR) and staff scientist at Berkeley Lab’s Molecular Foundry who led the study.
“By using our technology and accompanying empirical models for battery performance and lifetime, other researchers will be able to quickly evaluate the readiness of each component that goes into the battery, from the membrane to the charge-storing materials
This should save time and resources for researchers and product developers alike.”
Most grid battery chemistries have highly alkaline (or basic) electrodes – a positively charged cathode on one side, and a negatively charged anode on the other side.
But current state-of-the-art membranes are designed for acidic chemistries, such as the fluorinated membranes found in fuel cells, but not for alkaline flow batteries.
(In chemistry, pH is a measure of the hydrogen ion concentration of a solution. Pure water has a pH of 7 and is considered neutral.
Acidic solutions have a high concentration of hydrogen ions, and are described as having a low pH, or a pH below 7.
On the other hand, alkaline solutions have low concentrations of hydrogen ions and therefore have a high pH, or a pH above 7.
In alkaline batteries, the pH can be as high as 14 or 15.)
Fluorinated polymer membranes are also expensive. According to Helms, they can make up 15% to 20% of the battery’s cost, which can run in the range of $300/kWh.
One way to drive down the cost of flow batteries is to eliminate the fluorinated polymer membranes altogether and come up with a high-performing yet cheaper alternative such as AquaPIMs, said Miranda Baran, a graduate student researcher in Helms’ research group and the study’s lead author. Baran is also a Ph.D. student in the Department of Chemistry at UC Berkeley.

Getting back to basics
Helms and co-authors discovered the AquaPIM technology – which stands for “aqueous-compatible polymers of intrinsic microporosity” – while developing polymer membranes for aqueous alkaline (or basic) systems as part of a collaboration with co-author Yet-Ming Chiang, a principal investigator in JCESR and Kyocera Professor of Materials Science and Engineering at the Massachusetts Institute of Technology (MIT).
Through these early experiments, the researchers learned that membranes modified with an exotic chemical called an “amidoxime” allowed ions to quickly travel between the anode and cathode.
Later, while evaluating AquaPIM membrane performance and compatibility with different grid battery chemistries – for example, one experimental setup used zinc as the anode and an iron-based compound as the cathode – the researchers discovered that AquaPIM membranes lead to remarkably stable alkaline cells.
In addition, they found that the AquaPIM prototypes retained the integrity of the charge-storing materials in the cathode as well as in the anode. When the researchers characterized the membranes at Berkeley Lab’s Advanced Light Source (ALS), the researchers found that these characteristics were universal across AquaPIM variants.
Baran and her collaborators then tested how an AquaPIM membrane would perform with an aqueous alkaline electrolyte.
In this experiment, they discovered that under alkaline conditions, polymer-bound amidoximes are stable – a surprising result considering that organic materials are not typically stable at high pH.
Such stability prevented the AquaPIM membrane pores from collapsing, thus allowing them to stay conductive without any loss in performance over time, whereas the pores of a commercial fluoro-polymer membrane collapsed as expected, to the detriment of its ion transport properties, Helms explained.

This behavior was further corroborated with theoretical studies by Artem Baskin, a postdoctoral researcher working with David Prendergast, who is the acting director of Berkeley Lab’s Molecular Foundry and a principal investigator in JCESR along with Chiang and Helms.
Baskin simulated structures of AquaPIM membranes using computational resources at Berkeley Lab’s National Energy Research Scientific Computing Center (NERSC) and found that the structure of the polymers making up the membrane were significantly resistant to pore collapse under highly basic conditions in alkaline electrolytes.
A screen test for better batteries
While evaluating AquaPIM membrane performance and compatibility with different grid battery chemistries, the researchers developed a model that tied the performance of the battery to the performance of various membranes.
This model could predict the lifetime and efficiency of a flow battery without having to build an entire device. They also showed that similar models could be applied to other battery chemistries and their membranes.
“Typically, you’d have to wait weeks if not months to figure out how long a battery will last after assembling the entire cell. By using a simple and quick membrane screen, you could cut that down to a few hours or days,” Helms said.
The researchers next plan to apply AquaPIM membranes across a broader scope of aqueous flow battery chemistries, from metals and inorganics to organics and polymers.
They also anticipate that these membranes are compatible with other aqueous alkaline zinc batteries, including batteries that use either oxygen, manganese oxide, or metal-organic frameworks as the cathode.
Researchers from Berkeley Lab, UC Berkeley, Massachusetts Institute of Technology, and Istituto Italiano di Tecnologia participated in the study.
Due to cost decreases1,2, renewable energy is experiencing greater use (https://www.eia.gov/outlooks/steo/pdf/steo_full.pdf). Many jurisdictions have policies in place to incentivize renewable use (http://www.dsireusa.org/). These policies are often intended to decrease the carbon-intensity of electricity production.
The role of energy storage in aiding the integration of renewable energy into electricity systems is highly sensitive to the renewable-penetration level3. California, for instance, is experiencing days during which demand is too low to accommodate all of the solar energy that is available midday4. This overgeneration-related renewable curtailment can be exacerbated by thermal generators having limited flexibility in how quickly they can adjust their production or how low their production levels can go5.
The development and deployment of grid-scale energy storage is advancing due to technology development and policy actions, such as California’s energy storage mandate6,7. Energy storage can provide a variety of services and its economic rationale is highly application-dependent8. Numerous studies optimize the size and operation of energy storage within a specific power system to achieve the best economic or environmental outcome.
However, there are no studies in the extant literature that investigate systematically the economic viability of using energy storage to alleviate renewable curtailment for the purposes of decarbonizing electricity production.
Moreover, the existing literature does not examine the impacts of emissions policy, such as a carbon tax, on the economics of energy storage for mitigating renewable curtailment. Detailed analysis is required to estimate the value of energy storage that is used for different applications, including renewable integration9.
This study addresses this gap by optimizing the investment in and operation of nine currently available energy storage technologies to minimize cost of the California and Texas power systems. We assume varying renewable penetrations and different CO2-tax policies.
Energy storage technologies have different characteristics and potential applications10–13. As such, no single technology excels on all characteristics. Integrating energy storage into the grid can have different environmental and economic impacts, which depend on performance requirements, location, and characteristics of the energy storage system14–16.
The cost of energy storage systems and regulatory challenges are major obstacles to their adoption13,17–19. Braff et al.20 examine the value of using energy storage to increase the price at which wind and solar energy can be sold in wholesale markets.
They find that many energy storage technologies are currently too costly for this application and determine the cost reductions that are needed to make this application economically viable. Other works21–25 examine the environmental impacts of energy storage, showing that it depends upon how it is operated and the technical characteristics of the power system into which it is integrated.
Thus, there is a need to optimize the operation of energy storage to achieve desired economic and environmental outcomes. Many studies optimize the operation and size of an energy storage system for a given grid application based on economic criteria26,27. Others propose optimization models for sizing and operating energy storage to minimize total electricity cost or to maximize investor profits28–30. Another set of studies model emissions and economic considerations in optimizing energy storage use31–33.
Our study extends the existing literature by evaluating the role of energy storage in allowing for deep decarbonization of electricity production through the use of weather-dependent renewable resources (i.e., wind and solar).
The model optimizes the power and energy capacities of the energy storage technology in question and power system operations, including renewable curtailment and the operation of generators and energy storage.
This is done to minimize total system costs, which consist of the capital cost of energy storage, generator-operations costs, and CO2-emissions costs. Technical constraints in the model include operating limits of generators and energy storage and load-balance requirements. We examine nine currently available energy storage technologies: pumped-hydroelectric storage (PHS), adiabatic (ACAES), and diabatic (DCAES) compressed air energy storage (CAES), and lead-acid (PbA), vanadium-redox (VRB), lithium-ion (Li-ion), sodium-sulfur (NaS), polysulfide bromide (PSB), and zinc-bromine (ZNBR) batteries.
Our model allows us to determine which energy storage technologies are most cost-effective in aiding renewable integration and the extent to which the cost of a currently uneconomic technology must come down to make it cost-effective. We use two case studies, which are based on the California and Texas power systems in 2010–2012, and consider up to 20 GW of wind and 40 GW of solar capacity added to the system. We also consider the impact of a CO2 tax of up to $200 per ton.
Our analysis of the cost reductions that are necessary to make energy storage economically viable expands upon the work of Braff et al.20, who examine the combined use of energy storage with wind and solar generation assuming small marginal penetrations of these technologies. Conversely, we examine their economics at significant renewable penetrations that are necessary for deep decarbonization of electricity production.
Our findings show that renewable curtailment and CO2 reductions depend greatly on the capital cost of energy storage. Moreover, increasing the renewable penetration or CO2 tax makes energy storage more cost-effective.
This is because higher renewable penetrations increase the opportunities to use stored renewable energy to displace costly generation from non-renewable resources. Among the energy storage technologies that we consider, PHS and DCAES are deployed in more of the scenarios that we examine.
This is due to the lower capital costs of these technologies. Other technologies see deployment under some scenarios. We also find that relatively modest reductions in the capital costs of other energy storage technologies can make them cost-effective for this proposed application.Go to:
Results
Energy storage deployment
Supplementary Table 1 summarizes the energy capacity of the energy storage technologies that are installed with different wind- and solar-penetration levels and CO2 emissions-tax regimes in 2012 in the base case with a 7.0-GW minimum-dispatchability requirement in the California Independent System Operator (CAISO) system.
Supplementary Table 2 summarizes the same for the Electric Reliability Council of Texas (ERCOT) system under the base-case 8.2-GW minimum-dispatchability requirement. The tables show that higher renewable penetrations or emissions taxes tend to improve the economics of energy storage deployment. Due to their relatively low capital costs, PHS and DCAES are deployed in more scenarios and with greater capacity than most of the other technologies. Conversely, a technology that is currently more-expensive but has a higher round-trip efficiency, such as Li-ion batteries, is not deployed in any of the scenarios that are summarized in these two tables.
Table 1 shows the results of a sensitivity analysis, in which lower cost assumptions for Li-ion batteries lead to significant Li-ion deployment and resultant curtailment and emissions reductions. Supplementary Data 1 summarizes the amounts of energy storage that are installed in the other years and with the other minimum-dispatchability requirements that we analyze.
Table 1
Changes in renewable curtailment and CO2 emissions resulting from lower Li-ion and PSB costs and higher NaS costs as a percentage of renewable curtailment and CO2-emissions levels with the baseline costs
Technology | Renewable Curtailment (%) | CO2 Emissions (%) | ||
---|---|---|---|---|
CAISO | ERCOT | CAISO | ERCOT | |
Li-ion | −27.1 | −56.9 | −22.3 | −1.6 |
NaS | 60.2 | 89.6 | 43.9 | 1.4 |
PSB | −15.1 | −34.4 | −8.9 | −0.5 |
Results shown are for 2012 assuming 20-GW wind- and 40-GW solar-penetration level, a $200 per ton CO2-emissions tax, and base-case minimum-dispatchability requirements of the CAISO and ERCOT systems
Supplementary Tables 1 and 2 show that irrespective of the carbon-tax level, energy storage is not cost-effective in California for the application that we model without added renewables. This is because California’s fossil-fueled generators are all natural gas-fired. Thus, there is limited value in using energy storage for energy arbitrage, because of small differences between on- and off-peak marginal generation costs.
In California, the value of energy storage stems primarily from its ability to reduce renewable curtailment, thereby displacing fossil-fueled generation. This benefit is greater with a higher carbon tax, because fossil-fueled generation is more costly in the presence of a tax. Recent estimates from the California Energy Commission show that as of October 2017, California has 5.6 GW of wind and 16.2 GW of solar installed (https://www.energy.ca.gov/almanac/renewables_data/wind/). Thus, California is approaching renewable-penetration levels at which a number of energy storage technologies are cost-effective for mitigating renewable curtailment.
Even in the absence of renewables, deploying some energy storage technologies in Texas is cost-effective under higher emissions-tax rates.
This is because the ERCOT system has a more mixed generation fleet, with both coal- and natural gas-fired units that have very different generation costs. Moreover, the differences in the carbon contents of coal and natural gas gives larger differences in marginal generation costs between coal- and natural gas-fired units with higher CO2-tax rates.
Renewable curtailment
Figure 1 shows total annual renewable curtailment with and without access to energy storage with different amounts of renewable capacity and CO2-emissions taxes in 2012 in California under the base case 7.0-GW minimum-dispatchability requirement. Figure 2 shows the same for Texas under its 8.2-GW base case minimum-dispatchability requirement. The curtailment results for other minimum-dispatchability requirements and years are provided in Supplementary Data 1. The figures show that California has much higher renewable-curtailment rates relative to Texas.
This is because California has much higher outputs from inflexible resources (e.g., nuclear, geothermal, biomass, and hydroelectric units) and energy imports compared to Texas. This greater inflexibility makes it more challenging for the CAISO system to absorb wind and solar generation.
The figures show that with relatively low emissions taxes (i.e., $50 per ton or less), PHS and CAES are the only economically viable technologies for averting renewable curtailment. However, with higher emissions taxes, all of the energy storage technologies (except for Li-ion batteries) become cost-effective for this application. This is consistent with Supplementary Tables 1 and 2, which show that most of the energy storage technologies are deployed in some of the renewable-penetration scenarios if the CO2-emissions tax is sufficiently high.
Consistent with real-world experience4, renewable curtailment is greatest in the spring. This is due to the spring having relatively low electricity demand and many days with good midday solar availability. California has experienced recently an increasing number of spring days on which these factors require solar curtailment.
CO2 emissions
Figure 3 summarizes the benefits of energy storage in decarbonizing in-state electricity production in California in 2012, under the base case 7.0-GW minimum-dispatchability requirement. Figure 4 shows the same in Texas under the base case 8.2-GW minimum-dispatchability requirement. Results for other minimum-dispatchability requirements and years are provided in Supplementary Data 1. Without any added renewables or energy storage, California can achieve negligible 0.2% CO2-emissions reductions with a sufficiently high carbon tax through dispatch switching. In Texas, dispatch switching can decrease emissions by 24% without added renewables.
California’s fossil-fueled generators have negligible emissions-rate differences. With a carbon tax, generating loads can be switched to units that have higher operating costs and lower emissions rates. Texas, conversely, has a mix of coal- and natural gas-fired generating units. A sufficiently high carbon tax switches the merit order between these units.
Without any access to energy storage, California’s 2012 CO2 emissions could have been reduced by 72%, through deployment of renewables with a 7.0-GW minimum-dispatchability requirement and a $200 per ton CO2 tax. However, energy storage decarbonizes electricity production to a greater extent by reducing renewable curtailment. Li-ion batteries would have provided essentially no emissions improvements in 2012, due to their high capital costs. Conversely, DCAES yields the greatest emissions reductions in California in 2012.
Texas shows similar trends. Without energy storage, renewable deployment, in conjunction with a $200 per ton CO2-emissions tax, can reduce CO2 emissions by 54% in 2012 with the base case 8.2-GW minimum-dispatchability level. As in California, DCAES yields the greatest emissions reductions in Texas.
Figure 5 summarizes energy storage’s impacts on renewable curtailment and CO2 emissions in California in the 3 years that we analyze. The results that are shown in the figure assume 20-GW and 40-GW wind- and solar-penetration levels, respectively, a $200 per ton CO2-emissions tax, and the base-case 7.0-GW minimum-dispatchability requirement. The results are similar for other minimum-dispatchability requirements. Renewable curtailments are shown as percentages of potential renewable production while emissions reductions are reported as percentages relative to a no-renewables case.
The figure shows significant interannual variability in renewable-curtailment rates, which stem from differences in electric loads. 2012 has significantly higher loads compared to 2010, meaning that California can accept more renewable generation in 2012. Each of the energy storage cases that is shown in the figure corresponds to the technology that achieves the greatest curtailment or emissions reduction. PHS achieves the greatest curtailment reductions in all of the years that are analyzed and the greatest emissions reductions in 2010.
However, DCAES achieves greater emissions reductions in the other 2 years. These results suggest that if curtailment reduction is the goal of deploying energy storage, PHS is a relatively stable technology choice in California. Conversely, if emissions reduction is the policy priority, there is less technology robustness.
DCAES is, conversely, a more robust technology in Texas, achieving the greatest curtailment and emissions reductions in all of the years and with all of the minimum-dispatchability requirements that we examine. However, energy storage delivers smaller incremental benefits in reducing Texas’s CO2 emissions.
Figure 5 shows that without energy storage, adding 60 GW of renewables yields emissions reductions that range between 71 and 92% across the years that are analyzed. Energy storage increases these emissions reductions to between 90 and 97%. ERCOT achieves 52–56% emissions reductions from adding 60 GW of renewables without energy storage. DCAES increases these emissions reductions to 56–59%.
This relatively small impact of energy storage in Texas is because there is relatively little renewable curtailment compared to California. As such, energy storage has a more limited role in increasing the use of renewable energy in Texas relative to California. Instead, the emissions-reduction benefits of DCAES in Texas largely stem from helping to shift some generating loads from coal- to natural gas-fired generators.
More information: Miranda J. Baran et al, Design Rules for Membranes from Polymers of Intrinsic Microporosity for Crossover-free Aqueous Electrochemical Devices, Joule (2019). DOI: 10.1016/j.joule.2019.08.025
Journal information: Joule
Provided by Lawrence Berkeley National Laboratory