Reliable energy storage systems are critical to the success of intermittent renewable energy sources (i.e. wind and solar). The development of such energy storage technologies complements efforts in carbon reduction, capture, and storage during energy generation and is one of the “scouting” areas within CMI. Craig Arnold and colleagues are working to assess and optimize usage of existing technology as well as develop new technologies to meet emerging demands for energy storage. Their research is a combination of experimental characterization, materials development and numerical modeling to understand these systems and develop methods for optimizing their implementation.


Battery characterization

Under most circumstances, batteries are characterized by their ability to deliver energy. For instance, charging is done under standardized conditions and the challenge is optimizing the amount of energy delivered to a potentially changing load, such as in a laptop computer or electric vehicle. However, when considering energy storage for intermittent power generation, it is important to understand the effects of variable charging conditions in addition to the variable discharging conditions.

In this area, the Arnold Group has been working to detail the relevant metrics of energy storage, including the response time or power density (i.e. how fast a given technology can store energy and how fast it can release that energy) and the energy density (i.e. how much energy can be stored) as a function of system parameters such as amount of energy stored or frequency and amplitude of power fluctuations. Although the main focus is on batteries and supercapacitors, the researchers consider other existing technologies including pumped hydro, compressed air, and mechanical flywheels. Given a better understanding of the benefits and limitations of current devices, they have begun to examine methods of integrating multiple storage technologies that have the ability to work synergistically to provide more stable and higher efficiency storage.

In approaching the topic of energy storage for fluctuating power, the team has focused on combining battery storage with wind power generation. One of the main challenges is that fluctuations in output power from wind (or solar) have multiple time scales, whereas most batteries have an optimal charge/discharge rate. In addition, wind can exhibit large scale diurnal cycles which require many hours worth of storage to provide reliable off-grid power throughout a 24 hour period, as well as seasonal cycles and short duration, large intensity intermittencies. Since the output power scales with the wind velocity to the third power, relatively small gusts (or calms) can produce significant increases (or decreases) in output power over short times. Therefore, it is not only necessary to store excess energy in order to level out fluctuations, it is necessary to optimize the energy storage technology so that it can function efficiently over the many different time scales that it will experience over its lifetime. A non-optimized or unmatched storage device leads to excessive loss and a decrease in charge storage.

Arnold and colleagues simulate power fluctuations experimentally and explore their effect on the amount of energy stored, lifetime, and the efficiency of charge storage in different batteries. 30 Their studies include common battery chemistries such as lead acid, lithium ion and nickel metal hydride cells. The researchers then use this characterizaton to select appropriate batteries for different applications, more accurately predict battery lifespan, and improve controls systems to increase battery efficiency. They find that Li-ion batteries (including Lithium cobalt oxide and Lithium iron phosphate) are robust under variable current situations with fluctuations between seconds and minutes, regardless of their state of charge. Conversely, nickel metal hydride batteries (like those currently used in some hybrid vehicles) experience sharp declines in charge storage efficiency under similar conditions. A deeper analysis of the material response offers insight into improved design for non-constant charge applications. Improved design, the integration of alternative storage mechanisms like supercapacitors, and intelligent controls will significantly increase battery efficiency and lifespan in the field.


Analysis of hybrid energy storage systems

Armed with a better understanding of the underlying physics and chemistry, the researchers have been developing numerical modeling approaches to optimize the use of these battery systems in grid level applications. In these cases, one not only has to deal with the physical response of the batteries, but one also has to deal with the stochastic nature of the wind fluctuations. Namely, it is hard to predict when the wind speed will change.

To deal with these challenges, Arnold and colleagues have been working in collaboration with Prof. Warren Powell to create approximate dynamic programming algorithms to model energy allocation for an array of storage devices. In their approach, the notion of a single monolithic battery to meet all the system demands is rejected in favor of engineering a combination of devices, each with different characteristics such as response time, power density, or capacity. This paradigm shift allows for the engineering of a hybrid energy storage system whereby each component is individually optimized for maximum capacity, lifetime, and rate behavior, while the overall system is designed for optimal implementation under real-world conditions.