15.4.1 Oak Ridge National Laboratory

Oak Ridge National Laboratory (ORNL), Sentech, PNNL, and others conducted a study to look at the effects of incentives, regulations, and policies on the feasibility of the second use of EV batteries for stationary applications (Cleary et al., 2010). The ORNL study estimated the battery costs in 2020, along with the estimated availability of used batteries from HEVs, PHEVs, and EVs. Area regulation, transmission and distribution (T&D) upgrade deferral, and power quality were considered to be the most financially rewarding revenue streams for these used batteries. A financial analysis was done by taking into account the cost to refurbish the battery, transport it to the second-use venue, and operation and maintenance costs.

It was determined that energy-intensive applications such as arbitrage and capacity were not economically viable as stand-alone applications unless peak-to-valley price range was high and avoided T&D deferred costs were also high. The analysis was also done for a mix of applications after ensuring the duty cycles were not at odds with each other. For commercial customers, demand charges could be avoided by peak shaving, whereas residential customers could avoid time-of-use tariffs. Indirectly, benefits also accrue in terms of T&D deferral. A detailed analysis of refurbishment costs was carried out.

15.4.2 Pacific Northwest National Laboratory

Viswanathan and Kintner-Meyer performed an optimization study for the use of PHEV batteries for regulation services (up and down), which were assumed to be the highest value grid ancillary service (Viswanathan and Kintner-Meyer, 2011). A simplified model to estimate the battery state of health (SOH) was developed. For a battery life of 18 years, various combinations of battery use time in the vehicle were assumed. Assuming electricity consumption of 200 Wh/mile and an annual distance of 12,000 miles with 300 vehicle operating days, the optimization also varied the battery size to provide various depths of discharge. A total of up to three batteries was considered, and an optimum use profile was determined for various regulation rates in the $9–27/MW-h range. The depth of discharge in the vehicle was fixed by selecting the target SOH at the end of the first use. For the second use, the remaining years determined the number of cycles needed, which set the DOD for regulation. From the DOD, the energy per half cycle, and hence the power was calculated. Table 15.2 provides the highlight of optimum battery use for various values of regulation service rate, with the batteries taken out of the vehicle at low (65%), medium (75%), and high (85%) SOH. The net revenue is simply the difference between the net present revenue and the cost of the battery. The cost also included electricity charges during vehicle operation and regulation service use. Clearly, for low regulation rate, the batteries need to be used to the fullest, whereas for high rates, depending on the end SOH, the first and the second batteries are taken out of the vehicle in the 3-5-year time frame.

The United States Advanced Battery Consortium cost target for Li-ion batteries in 2020 is $150/kWh, whereas the Boston Consulting Group estimates a cost of $440/kWh. It was thereby assumed that the used batteries will sell at 50% of these values, or in the range of $75–220/kWh. Even if the resale value of Li-ion batteries is not high, the utilities could come out ahead by having at their disposal very low-cost energy storage options. General Motors (GM) and ABB have tested five used battery packs from Chevy Volts with air cooling after demonstrating the viability in the laboratory (Garthwaite, 2012). Duke Energy plans to use this battery to smooth and firm wind, along with other value streams. An additional option that researchers are investigating is using these batteries as community energy storage that is charged off peak or at night when there is excess wind generation to provide peak shaving by making the energy available for EV charging without upgrades to T&D.

In a National Assessment study, it was found that arbitrage as a stand-alone application was not cost-effective for storage (Kintner-Meyer et al., 2013). This is in line with the ORNL study described earlier.

15.4.3 University of California at Berkeley Transportation Sustainability Research Center

At the Second Life Workshop conducted on March 7, 2011, Haresh Kamath of EPRI gave a presentation on the potential for the second use of PHEV batteries (Kamath, 2011). Assuming a net worth for the battery and adding power conversion system (PCS) and balance of plant (BOP) costs, a scenario was possible wherein the cost of the used PHEV battery is higher than a new battery. The PCS and BOP costs were assigned the same for the used and new batteries. Even in this case, it makes sense to use these batteries because the end user simply procures the battery at a fraction of its residual value, thereby benefitting the utility customers. The key is that the net present value (NPV) of net revenues from various grid uses should be greater than the battery residual value at the end of its useful life in the PHEV. If it is not, the end user can simply sell the battery at its residual value. However, in our opinion, the residual value is simply tied to the revenue opportunities from grid services (or any other services) the battery can provide. The advantage of having a high residual value is that the monthly leasing costs are reduced as shown in Figure 15.4 (Williams, 2010). The residual value is a complex function of revenue opportunities and the anticipated new battery costs at the end of the useful life of the PHEV battery. In a recent communication, the author informed us that Figure 15.4 has been “superseded by subsequent work” (Williams, 2012; Williams and Lipman, 2011), which is expected to “show significant but more modest reductions in a different context.” In the subsequent work, the effect of various parameters such as unit power conditioning system cost, battery capital cost, power-to-energy ratio, balance of system cost, and amount of energy left at the end of useful life in the vehicle on battery lease payments was considered.

The University of California at Berkeley has a program at its TSRC to look at the viability of the second-use batteries for PHEVs (Lipman and Williams, 2011; Williams, 2010, 2011; Williams and Lipman, 2011). This is in collaboration with the National Renewable Energy Laboratory (NREL) on the NREL Secondary Use Project. The other project partners were the Center for Sustainable Energy (CSE) California, UC Davis, PIER, California Energy Commission, San Diego Gas & Electric, Sempra Energy, and AeroVironment.

Task 1 of this project was to identify the second-life applications. Task 2 was to acquire the following batteries from used PHEV battery packs:

 Lithium iron phosphate batteries from A123 Systems,

 Lithium manganese oxide batteries from EnerDel,

 Lithium titanate batteries from Altairnano.

Task 3 was to identify the benefit opportunities after accounting for battery degradation and repurposing costs.

Table 15.3 shows the repurposing cost and additional costs for three different battery systems (Williams, 2011). Note that these numbers may have changed since this study. For example, the battery energy for each type has increased significantly.

As seen from the repurposing and miscellaneous cost for the Leaf 32 kW/16 kWh battery cost, for an E/P ratio of 0.5, the cost is $2000/kWh or $1000/kW. The 10-year benefit for various grid applications is presented with permission (Williams, 2011, 2012; Williams and Lipman, 2011). Per this study, the results were not very sensitive to the state of health of the batteries at the end of their useful life in the vehicles.

Details of the battery repurposing were provided (Lipman and Williams, 2011). The capital equipment cost for testing 23 modules was estimated at $1.2 million. The various steps for repurposing were as follows:

 Collect used EV modules from dealerships and transport to refurbishing facility;

 Visual inspect modules for damage;

 Determine manufacturer ratings and age from label or barcode;

 Measure voltage and resistance to identify failed modules;

 Load module onto test equipment;

 Determine energy module capacity at C/3 rate at 100% DOD, four cycles;

 Determine power capacity at 50% SOC by discharging at two-thirds OCV for 30 s;

 Disconnect module from equipment;

 Sort modules by capacity, power, and age;

 Assemble into battery packs;

 Ship to ESS assembler/integrator.

When the researchers used an appropriate mix of applications based on battery state of charge and benefits, they found several combinations of use cases whose benefits can potentially exceed the cost. It was proposed, for example, that it would take 44,000 Chevy Volt batteries to provide the regulation requirement of 732 MW/year (2006–2008 average CAISO (California Independent System Operator) up/down regulation requirement). There is sufficient regulation capacity to meet these requirements. Table 15.4 provides the potential for benefits combining various applications for a Volt battery (Williams and Lipman, 2011). Hence, for the Volt, the net benefit or residual value is $3200, which translates to a NPV at the time of purchase of $1510 using a 10% discount rate. This reduces the monthly lease payment by 20% from $122 to $95 per month. A similar reduction is obtained for the PHV and Leaf batteries.

Taking into account the battery capital cost and the benefits of using a combination of applications, it was estimated that the net benefit for the PHV, Volt, and Leaf batteries was $850, $3230, and $6450, respectively (Williams, 2012).

Note that PNNL recently completed a demonstration of a 120 kW, 500 kWh Li-ion battery storage system integrated with wind and solar PV and located at a substation to provide multiple services (Beaston et al., 2014). Automatic dispatch was done to maximize revenues from various use cases that included peak shaving, time-shifting, schedule management, avoiding wind curtailment, PV smoothing, and PV firming.

15.4.4 National Renewable Energy Laboratory

A techno-economic analysis was performed by the NREL team funded by the US Department of Energy’s (DOE’s) Vehicle Technologies Program (Neubauer et al., 2012). This work built on the detailed analysis done by the Sandia National Laboratory on the feasibility of using EV batteries in stationary applications (Cready et al., 2003). The resale price of the used battery was proportional to a “health factor,” the discount applied to reflect its usage, and the estimated cost of a new battery. The health factor was calculated by dividing the present value of the use battery’s energy throughput during its second life to that of a new battery for the same application. To estimate the repurposing cost, module yield was defined as cell fault rate raised to the number of cells in a module, with fault rates of 0.01%, 0.1%, and 1% assumed. The repurposing cost and purchase price were calculated assuming a selling price of $132/kWh for the various yields as shown in Figures 15.5 and 15.6 (Neubauer et al., 2012).

Clearly, the results are highly sensitive to the cell failure rate. At very low failure rate, there is no limit on module size, whereas at 1% failure rate, the module size is limited to 2 kWh to minimize repurchasing cost.

Because the resale price is proportional to the new battery price, assuming a steep fall in battery prices, a resale price of $38/kWh requires the used battery owner to pay the buyer $10–$40/kWh for it to be repurposed, for cell fault rate of 1%. Clearly, having a good SOH monitor during EV operation would help to identify modules with healthy cells. Taken to the next level, if entire battery packs are healthy, this would drop repurpose costs further. Based on this analysis, Table 15.5 shows the cost basis for the end user (Neubauer et al., 2012).

Assuming a 15-year second-use project lifetime, and purchase of second-use batteries during this period as needed, the payback period increased from 7 years to more than 15 years as the E/P ratio of the battery increased from 0.5 to 4.5. The services offered include power quality and reliability, and demand charges and time of use for the Southern California Edison rate structure.

15.4.5 Automobile manufacturers and collaborators

GM and ABB have tested five used battery packs from Chevy Volts with air cooling for community energy storage after demonstrating viability in the laboratory (Garthwaite, 2012; Hampton, 2013). The batteries were artificially degraded to 85–90% of their original energy capacity for this demonstration.

Based on the test results, GM signed a memorandum of understanding with ABB to scale up to power 50 homes, using the five-battery module as a repeating unit. Duke Energy plans to use this battery to smooth and firm wind, along with other value streams. An additional option being investigated is using these batteries for community energy storage. These second-use batteries can be charged at low peak times and then be used to charge electric vehicles, thereby reducing the need for expensive T&D equipment upgrade. GM is also considering using these batteries as a fast charging source in urban areas.

In January 2012, Nissan North America, ABB, 4R Energy, and Sumitomo Corporation announced their plans to build a grid-scale storage system prototype using Nissan Leaf batteries. Plans are to build a 50-kWh prototype to power 15 homes. Indy Power Systems used various lead–acid battery types to build a 50 kW, 15 kWh storage system for peak shaving for Melink Corporation, a HVAC equipment vendor (Hampton, 2013).

A collaboration, led by Electricore, with San Diego Gas & Electric (SDG&E), CSE, BMW, and Ricardo, is underway to standardize EV battery packs, which will greatly simplify their redeployment for stationary storage at the end of EV life (CSE, 2014).

Hart et al. (2014) modeled the use of the second-life batteries in a conventional and hybrid CERTS microgrid. It was found that the increase in impedance due to aging did not impact the performance of parallel connected batteries both in islanded and gird-connected mode.

Heymans et al. (2014) identified a significant opportunity for smaller storage systems for residential use for peak shaving. Decentralization using storage also serves to increase grid flexibility. It was determined that government incentives are needed for residential customers to take on installation and procurement costs.

Lih et al. (2012) used a simple approach to estimate the residual value at the end of the first-use life in the vehicle. The product of residual capacity and calendar life fractions was multiplied with a parameter that was set to 0.9 in the study. The cost to upgrade the battery was described by another equation, which includes an assumed improvement in performance for the upgraded battery and an upgrading cost, which was set at 20% of the residual value. It was not clear how the performance of the battery could be improved by upgrading, as the chemistry cannot be altered. To estimate the profit, the lease income in each stage was calculated. For the first stage, the difference between the new product price and the residual value was added to 15% of the residual value. For the second stage, the lease income was set to be the difference between the upgraded value and the scrap value. The total initial cost of the battery was subtracted from the total lease income to determine the profit over the life of the battery pack. For this case, the battery was in the vehicle for 3 years, followed by 12 years’ use in stationary application. A profit of 34% was obtained over the 15-year life of the battery. It is not clear why the cost for upgrading the battery pack was not considered in the total cost.

15.6.1 Design limitations and disadvantages of battery repurposing

Battery systems do not fail at a linear rate from 100% to 0% state of health. When the batteries are taken off the vehicles, they have 80% of performance left in terms of power deliver and/or energy capacity. Typically, the energy capacity is around 80% of initial capacity at the rated continuous power when the batteries are considered to be at 0% SOH in terms of ability to perform in the vehicle. Because the battery management system recognizes this condition to be a fully discharged state, a separate BMS needs to be developed that resets the SOH appropriately. For example, the SOH for stationary can be defined so that when the battery reaches 60% of original energy capacity, its SOH is set to 0. Assuming a linear decline in SOH from 100% of initial energy capacity to 60% of energy capacity, the battery can be assumed to start at 50% SOH when it is taken off the vehicle. Life estimation is preferably done from calendar life and cycle life as a function of DOD. As an approximation, the SOH at the end of vehicle life can be reset to 100%, with the remaining energy capacity set as the initial capacity for stationary use. The number of cycles to end of life can be estimated by extrapolation of the cycle life curves obtained earlier. In parallel, vendors can determine cycle life to the end-of-life energy capacity for the second use. The remaining duration will be set as the calendar life for the second use. For situations when data are absent, an algorithm needs to be developed to estimate life remaining until the energy capacity drops to 60% of initial value. Identification of the onset of a steep fall in performance is needed to provide a headspace for the battery system to avoid unanticipated failure. A summary of this approach is described in the literature (Viswanathan and Kintner-Meyer, 2011).

It is also desired that when batteries are taken off the vehicle they are given a fresh BMS to estimate remaining life. In addition to SOC and SOH, safety issues related to aging need to be considered. Aged Li-ion batteries are prone to a greater rate of thermal runaway because of a variety of factors such as solid electrolyte interface layer breakdown, higher impedance, and delayed melting of separator relative to onset of exothermic reactions (Underwriters Laboratories). It is critical to understand the safety aspects of aged batteries to enable them to be used safely and effectively.

Deployment of used batteries becomes an issue when using them in applications that have a duty cycle significantly different from that of the vehicle. An example is an HEV battery used for time-shifting, or an EV battery used for frequency regulation. Although it is straightforward to avoid this situation for a single application, a combination of energy storage applications with multiple duty cycles can be more demanding in terms of battery system performance requirements. For example, if the storage is used to provide regulation, load following, and time-shifting, the superimposition of these duty cycles could result in a duty cycle that can be challenging for the used HEV and EV batteries. Use of PHEV batteries may provide the required balance needed for the multiple-use application. Depending on the application, an EV battery may be the preferred choice as it can provide the power of an HEV battery with about six times the energy.

15.6.2 Mitigation of issues related to aging

Several issues related to aging can adversely impact use of transportation batteries in the stationary energy storage space. Safety-related issues can be addressed by ensuring the battery is not abused significantly during operation. This could be achieved by having tight tolerance on operating temperature range, limiting the power in and out of the battery when battery impedance is high, and ensuring thermal runaway conditions are not approached. The latter can be ensured by characterizing batteries at various aged conditions. The characterization methods comprise accelerated rate calorimetry to determine onset of thermal runaway (Ottaway and Mores, 2001), determination of temperature increase at various multiples of rated power (Williford et al., 2009), and estimation of total heat generation from first principles (Viswanathan et al., 2010; Williford et al., 2009). Oversizing the storage system to provide a degree of flexibility would be another method to circumvent capacity loss related to aging. However, this is probably not a practical solution owing to higher capital costs. Another alternative is to have the storage output power decrease with aging during its second life, thus accounting for the decreasing performance with time. This scenario can be attractive when capital spending on T&D is spread out over a few years, thereby requiring less storage performance with each passing year.

The energy content for HEV batteries is ~ 2 kWh, PHEV batteries have ~ 8–16 kWh, and EV batteries have more than 16 kWh, with a pulse power requirement for all the batteries greater than 25 kW. The continuous rated power for PHEV and EV is about 15 kW for 1–3 h. Hence, it is possible to mix and match the various types of transportation batteries with the end application. For example, for a use case involving frequency regulation and time-shifting, it is conceivable that HEV batteries could be used for the former and PHEV/EV for the latter. The HEV batteries can also be used to supplement the power needs of an energy-intensive application such as time-shifting, by selectively providing/absorbing the additional power needed during conditions of excessive peak demand or supply.

Depending on the depth of discharge of these batteries, their cycle life would vary. For a 10- to 15-year life, the number of cycles needed is in the 3500–4500 range (Amjad et al., 2010). The cycle life depends on the depth of discharge for any chemistry. As shown in Figure 15.8, Li-ion batteries can undergo 4000 cycles at 50% DOD, whereas Ni–MH has a 70% DOD for the same number of cycles. This is simply an example, as cycle life at various DOD depends significantly on the specific Li-ion chemistry.

Li-ion batteries are expected to degrade when maintained at high SOC, high internal temperature, and when operated at a high depth of discharge. The following figure provides an example of effect of DOD and temperature on cycle life of an Altairnano battery (Lipman and Williams, 2011). Each Li-ion chemistry has its own cycle calendar life, energy, power, safety, and cost characteristics. There is no one size fits all. Hence, depending on the mix of applications being considered, different battery chemistries would be preferred candidates based on the desired property.

The BMS for the transportation batteries does Coulombic counting to estimate the SOC, with periodic checks on accuracy of these estimates. State of health determination is done by periodic capacity checks after a full charge, along with measurement of voltage efficiency at a fixed current charge/discharge at a fixed SOC (say 50%). An alternate means allow determination of internal resistance by measuring the voltage drop (or rise) at constant current discharge (or charge) for a few seconds. For HEV batteries, the voltage efficiency at 50% SOC and/or internal resistance at 50% SOC provides an indication of battery SOH. For PHEV and EV, along with this approach, the periodic capacity check provides an added criterion for battery SOH. As BMS systems get more sophisticated, the battery SOH can be more accurately estimated. This includes the effect of ambient temperature and calendar life on battery SOH, thereby taking into account the degradation of the battery during its life in the vehicle purely due to degradative effects of temperature and time.

The other issues are logistical in terms of how to track the various batteries seamlessly. According to Pablo Valenci of GM, the same system used to track and refurbish/resell auto parts such as water pumps and starters can be deployed for used batteries (Garthwaite, 2012).

15.6.3 Barriers and proposed solutions

Specifically, four barrier categories were identified, along with some proposed solutions during a one-day workshop at UCLA for the second use of EV batteries. Participants at the workshop included automakers, utilities, storage developers, business leaders, and public officials (Elkind et al., 2014).

Barrier Category 1: Economic uncertainty regarding residual value leading to lower purchase or lease costs:

 Uncertainty in the use cases for energy storage applications;

 Competition from cheap energy storage;

 Expensive repurposing of battery pack;

 All of the above pose risk to obtaining financing.

The solutions proposed were as follows:

 Reduce uncertainty through demonstration projects;

 Improve the quality of such demonstration projects to achieve specific goals;

 Remove regulatory barriers to allow such pilot projects access to specific grid markets;

 Offer federal and state tax rebates and financing for second-life demonstration projects;

 Ease liability for faulty batteries.

Barrier Category 2: Difficult regulatory landscape

 Utility regulations very complex;

 Transport of used batteries, classified as hazardous waste, have additional regulations;

 Some incentives for storage make used assets ineligible;

 At the customer side, state and local permitting authorities may raise safety-related issues to prevent or slow down process.

The solutions proposed were as follows:

 Develop “clear and consistent regulations” for the 2018–2028 time frame at the federal and state levels;

 State policy makers fund effort to identify critical regulatory needs that affect this use;

 Federal and state programs should ensure used batteries are eligible for incentives;

 Grid operators should modify rules to allow second-life battery participation.

Barrier Category 3: Uncertain liability picture for second-use batteries

 The automakers that own the battery may discourage its use in grid applications because of potential for damages from unknown duty cycles.

The solutions proposed were as follows:

 Liability shields that protect OEMs for reused, refurbished automotive parts such as engines and transmissions;

 Develop technical performance standards for second-life batteries. The model used by the PNNL/Sandia-led DOE-OE–sponsored performance protocol for energy storage systems can be followed here;

 These standards can be used as a basis for liability protection;

 Insurance companies can provide liability coverage for the second-life market.

Barrier Category 4: Difficulty in estimating battery life

 Lack of data on battery performance in the vehicle;

 Unknown duty cycles for second use.

The proposed solutions were as follows:

 Make the battery performance and SOH data in the vehicle available;

 Make the data obtained by universities and national laboratories on the second-use application available

– Fund the collection and distribution of this data;

– Share utility infrastructure needs and database of energy data to allow creation of use cases for these batteries.

Nine barriers for widespread adoption of energy storage were identified by the California Public Utility Commission (Malashenko et al., 2012). Some were grid-related such as a well-defined interconnection procedure between the storage and the grid. Some related to the lack of operating experience and poor transparency of incentives for storage. Among the barriers identified, the ones that appear most critical are insufficient clarity of regulations, unclear use cases/applications duty cycle. To this we would add lack of a uniform standard to evaluate storage systems for grid applications. The potential to use storage for multiple applications creates additional complexity, as some of the applications may be at odds with the other in terms of compatibility. Energy storage procurement targets such as AB 2514 serve as a catalyst for higher storage demand.

15.6.4 Safety, warranty, and controls architecture

The US DOE Office of Electricity Delivery and Energy Reliability has created a Strategic Plan for Energy Storage Safety. As a part of this effort, PNNL has published two documents that provide an overview of issues affecting energy storage safety (Conover, 2014a) and an inventory of safety-related codes and standards related to energy storage systems (Conover, 2014b).

The SOH of the used battery determines to a large extent its residual value and ability to provide the required service during its second use. The uncertainty in terms of duty cycles for stationary applications is further expected to complicate the terms and conditions for warranties related to battery use and performance.

The controls’ architecture for an energy storage system is expected to be multilayered. A BMS resides on the DC side and an energy management system resides on the AC side, while a storage control system resides either on-site or in a remote location to interface with the grid. All of these control layers need to be compatible with each other, with a clear understanding of which layer has ultimate control and override capability.