Energy storage is playing an increasingly important role throughout the energy infrastructure, from powering hybrid and electric vehicles to offsetting the inherent intermittency of renewable energy generation. Unlike batteries for electronic devices, which can be charged using a pre-determined protocol simply by plugging them into the wall, many of these applications are characterized by highly variable charge and demand profiles. The Energy Storage Group headed by Craig Arnold is working to characterize how such variability in charging powers affects battery behavior in order to improve overall system efficiency and lifespan.

 


Impacts of stress in lithium-ion cells

This year, the Energy Storage group has identified several sources of mechanical stress within lithiumion cells and investigated the impacts of this stress on battery life and performance, identifying several potential areas for improvement.

Stress evolution in Li+ cells. Lithium-ion cells typically operate under some level of compression applied by a rigid constraint, for example a cylinder battery cell. The level of initial stress is fixed by the manufacturer, but this stress is not constant over the life of the cell owing to electrode expansion during charging/discharging as well as other effects. Knowledge of the stress evolution over the cell’s entire useful life is important for understanding and predicting battery cell degradation, which is heavily influenced by stress state.

Figure 10: Stress evolution of lithium-ion cells with different initial stack pressures shows as a function of cycle number long term trend of increasing mechanical stress.

Mechanical stress evolution was found to be a function of three competing mechanisms: viscoelastic stress relaxation of the polymeric battery components, volumetric changes of the battery active material due to lithiation, and growth of the SEI (a solid film produced by side reactions). The relative importance of these mechanisms were found to change depending on the initially applied stack pressure, with viscoelastic stress relaxation becoming a more dominant mechanism with increasing levels of stress. At long time scales, SEI growth became a dominant mechanism, increasing stress in all cells. (Figure 10) Future plans include development of more mechanically robust cells through improvements in materials selection and design.

Effects of stress on capacity fade. It is well known that mechanical stress (on the order of hundreds of MPa) that builds up within the electrode particles as a result of particle expansion during lithiation is linked to capacity degradation through particle failure. However, little attention has been given to the relatively modest pressures (tenths of MPa) that are applied to the entire cell during manufacturing and that build up during normal operation. It is commonly believed that these modest pressures have no negative effects on cell operation. However, due to the nature of the soft materials employed in some of the battery components such as the separator, these applied stress levels can result in major deformations over time which could potentially impact battery degradation over the lifetime of a cell.

Arnold and colleagues investigated the effect of applied stack pressure on the electrochemical performance and capacity retention characteristics of lithium-ion batteries. It was found that higher levels of stack stress resulted in higher rates of capacity fade. However, it was also shown that very small amounts of stress (on the order of hundredths of MPa) are beneficial to capacity retention through the prevention of electrode layer delamination. Upon disassembly of the cells it was discovered that growth of a surface film on the electrodes had occurred in the stressed cells, with highly stressed cells showing more film growth. This coupling between stress and chemistry had not been anticipated and will be a subject of future investigation.

Effects of stress on ion transport. High power battery operation requires very fast transport of lithium-ions through a liquid phase electrolyte between anode and cathode of a battery cell. In a real cell, this liquid phase is contained in a porous polymeric separator which is placed between the two electrodes to keep them from coming into contact and creating a short circuit. During battery manufacturing and operation, applied stresses build up which result in compression of the separator, which is relatively compliant compared to the battery electrodes. This separator deformation results in pore closure which ultimately restricts ion transport between the battery electrodes. Knowledge of how the impedance associated with this transport restriction varies with deformation is critical for predicting performance in high power cells.

Arnold’s group measured the impedance as a function of deformation of commercial separators by compressing a pouch cell containing separator wetted in electrolyte but no active battery material while simultaneously measuring impedance. Wet-manufactured, dry-manufactured, monolayer, trilayer, polypropylene, and polyethylene separators were tested. A relationship between deformation and impedance using the Bruggeman tortuosity-porosity relationship was derived and verified by curve fitting experimental data (Figure 11). Using the derived relationship the empirical Bruggeman parameters could be determined by curve fitting the experimental data, yielding fundamental information about transport in the separators. Future work will focus on development of separators that can sustain deformation without restricting transport.

Figure 11: Curve fits of tortuosity vs. porosity for different lithiumion battery separators used to determine both Bruggeman parameters.

Modeling the rate dependence of charge storage

The variability of wind, solar and other similar power sources necessarily means that batteries in these systems are charged over a range of different powers. Discharge efficiency is known to have a dependence on the discharge power; in Krieger and Arnold (2012) a similar effect on charge efficiency is modeled and experimentally confirmed for charging power. Both models have been expanded to account for an additional limitation to battery capacity as power increases: significant undercharging and underdischarging due to voltage limitations. As power increases, the charging voltage is offset higher and discharging voltage offset lower, leading to premature voltage cutoffs; this effect is more pronounced on charging due to the non-symmetric shape of the voltage curves. (see Figure 12) Energy-power relationships in battery charging and discharging are therefore found to be highly dependent on both the efficiency of charging and voltage limitations at any given power, which must be taken into account when designing battery systems operating over a variable range of powers.

Figure 12: Model for undercharge and underdischarge using experimentally determined value for Q(P) as compared to models without undercharge and
underdischarge. The experimental results are dominated by undercharge and underdischarge.

Battery degradation in off-grid renewable applications

The stresses of highly variable and incomplete charging in off-grid renewable energy systems result in rapid degradation of the lead-acid batteries typically used for these electrification projects, incurring large replacement costs over the lifetime of the system. To identify more promising energy storage technologies, Arnold and colleagues compared aging rates and mechanisms among constant-charge and wind-charged lead-acid, lithium cobalt oxide (LCO), LCO-lithium nickel manganese cobalt oxide composite, and lithium iron phosphate batteries.

Figure 13: Capacity fade in constant-charged, windcharged, and low frequency wind-charged lithium iron phosphate cells.

Accelerated aging studies conducted over the course of a year find that while constant-charged leadacid batteries last longer than wind-charged cells, lithium cobalt oxide cells last longer under variable and incomplete charging conditions than constant charging, and lithium iron phosphate cells show only 1-3% degradation under all charging protocols. While these last cells are more expensive per installed kWh than the lead-acid cells, their consistently good power and voltage performance and ability to withstand deep discharge and incomplete charging allow the systems to be sized smaller. Combined with their long lifespan in variable power conditions, these results suggest significant potential for lithium iron phosphate batteries to reduce system lifetime costs for off-grid renewables.

 


A frequency-based model for energy storage

Variable power energy storage requirements may be best met by a suite of energy storage technologies instead of a single device. Applications like electric vehicles require both slow delivery of energy and rapid absorption of power during regenerative braking. Variable powers, as seen in the previously described projects, affect the efficiency and lifespan of energy storage devices to different degrees. Batteries may be better at providing bulk storage, whereas ultracapacitors are good at handling high-power bursts. The metrics to understand these complex systems are limited, however. Energy storage devices may be characterized by energy or power density, or discharge time.

Arnold and colleagues are working instead to re-frame energy storage in the frequency domain. In systems where power supply demand has both rapidly changing and slowly changing components, this power profile can be translated into the frequency domain to quantify the energy contained in high, medium or low frequency oscillations. Energy storage devices are also classified by their ideal frequency range – e.g. high frequency for ultracapacitors, low for compressed air energy storage. Frequency analysis is performed using wavelet transforms, which can accommodate the nonstationarity characteristic of many of these systems. This classification of energy storage systems in the frequency domain allows for improved understanding of complex and hybrid systems.