A123 Grid Battery System Single Rack Evaluation

Southern California Edison (SCE) is performing accelerated life cycle testing on a subsystem of A123’s Grid Battery System (GBS). SCE’s laboratory testing is being performed on one rack containing 6 modules that provided 100 kW and 23 kWh. The full GBS is a 2MW 500kWh system composed of 18 racks with 8 modules in each. A test profile composed of 1C and 2C 100% depth of discharge (DOD) cycles was produced to apply 9 cycles per day. Assuming the full system will be used once per day, four months of testing corresponds to the cycles that would be applied over three years of operation. To help track battery performance throughout the testing, Reference Performance Tests (RPTs) are performed every 300 cycles (which corresponds to just over one month). The rack is contained in an environmentally controlled chamber at 20°C during all cycling to simulate the thermal conditions in the GBS container. Over 3000 test cycles have been completed on the GBS system. Approximately 10% decay in capacity has been observed for all power levels. At 80% DOD the available power has decreased only 3%. After approximately the 1200th cycle, a four month pause in testing occurred. Upon restarting the cycling, a significant decrease in capacity was observed. However, the capacity did not decay as quickly. After about the 2000th cycle, the capacity trend resumed the initial rate of decay. This pause in testing has produced interesting conclusions relating to the relationship of calendar life and accelerated cycling which could have significant impacts on electric vehicle or utility application of lithium-ion batteries. This report focuses entirely on the laboratory testing of the GBS single rack. Future analysis will combine these results with those of other SCE tests to produce conclusions regarding the application of the GBS on SCE’s grid.


Introduction
A123 Systems' Grid Battery System (GBS) units apply technology used in hybrid electric vehicles to meet the power sector's growing need for energy storage [1]. The GBS hybridizes a power plant by adding a multi-megawatt energy storage system to the plant. The GBS uses Smart Grid technologies with energy storage to provide grid stabilizing ancillary services such as frequency regulation, fast ramping, T&D deferral and spinning reserve to increase power plant efficiency.
GBSs increase the capacity, responsiveness, and efficiency of individual power plants and whole power systems. Testing by Southern California Edison's (SCE's) Advanced Technology Organization has been performed on a partial GBS. The system being tested is a single rack with 6 modules installed. The complete system is rated at 2 MW for 15 minutes and includes 18 racks with 8 modules per rack. In parallel, a full system is being tested at SCE's Large Energy Storage Test Apparatus (LESTA). The laboratory test is focusing on the fundamental technology that composes the GBS (the battery and battery management system), while the full system test will focus on grid integration and control mechanisms. The single rack being tested in the laboratory for this report does not contain A123 Systems' Smart Grid Domain Controller (SGDC) which is typically installed in a complete GBS such as the GBS testing at the LESTA station. The GBS is composed of A123 System's lithium-ion Nanophosphate® technology. This chemistry is designed for high power capability and was initially used primarily in hybrid-electric buses and handheld power tools. Based on this proven performance, A123 believes they could be effective for stationary storage applications as well. SCE is evaluating the performance of the GBS single rack to identify the benefits that this technology can provide. While the full GBS can be used in all of the applications mentioned above, SCE is primarily interested in the system's use on our distribution system. The power capabilities of the system make it useful to provide relief to overloaded circuits and potentially help defer distribution upgrades. SCE will also evaluate the full GBS for renewable integration. This system could help smooth photovoltaic or wind turbine output.

System Characterization Testing
This project aims to thoroughly understand and measure the performance of the single rack of the A123 GBS. To do so, SCE performed an initial preliminary characterization of the rack. This characterization was performed by running several profiles on the system and recording detailed data during the entire test. SCE used this data to: Verify the capacity of the rack when discharged at various rates Calculate the roundtrip efficiency of the system Evaluate effectiveness of the battery management system (BMS) Determine the thermal performance

System Life Cycle Testing
After the initial characterization of the GBS single rack, SCE used the performance data (electrical and thermal) to assemble a test cycle that could be run continuously on the pack. This profile has been used for over two years to help determine the cycle life of the system. Periodic reference performance tests (RPTs) are performed throughout the cycle testing. These RPTs are used to measure the performance of the system after each interval and identify any decrease in performance as the system ages.

Battery Cycler
Tests are conducted using the Aerovironment AV900 battery cycler.
Voltage and current measurements are obtained from the cycler. Temperature measurements for all battery modules are obtained from the rack's CAN Bus. The battery cycler was calibrated locally prior to and periodically throughout testing to ensure the cycler is within factory standards.

Environmental Chamber
The ambient temperature during all testing is maintained through the use of an Environmental Chamber. Throughout the course of testing, the GBS rack has been placed in two different chambers, and each was set to maintain at an average of 20° C. The full GBS includes an HVAC system that will maintain the full system at a similar temperature. In the first chamber the temperature ranged from 16 to 24°C, while the second controlled the temperatures to within 1° (19 to 21° C). During the installation of the new chamber, the rack was removed and placed in the lab from June 18 through October 27, 2010. The rack was fully charged prior to this resting period. Over the summer, the temperature in the lab varied greatly. Fig. 1 Table 2 is averaged from two discharges at each rate to 100 % DOD. From the data we can determine that there was essentially no difference in capacity as the system is discharged at higher rates. The lower capacities seen at the lower rates most likely can be attributed to the system requiring more time to fully discharge. Since the passive balancing system is always active in the rack, the balancing circuit is constantly removing energy from the cells. Thus the lower rate discharges had a slightly lower capacity due to the energy lost during cell balancing. The thermal evolution and resulting lower impedance of the cells may also be a factor in this behavior.

System Efficiency
The capacity retention test was performed as specified in 3.2.2. From Table 3 it is evident that there is a significant capacity loss in a 24 hour period. This loss can be attributed to the load imposed on the battery modules by the BMS (for balancing). Although the energy loss may seem significant for one daily cycle, it is not significant if the battery is cycled continuously during the same period of time. It should be noted that since the rack is not part of a full GBS, this is an abnormal operating condition. A123 has provided a firmware update to reduce decay by cell balancing. However, this update has not been implemented in this test. Overall system efficiency was calculated at different discharge rates. Ancillary loads (such as BMS and system fans) were not taken into account for these calculations as they were powered externally. Elapsed time was defined as the time the discharge is started to the time the charge is completed. A one hour rest was included between the charge and discharge. The difference in system efficiency at different rates was small, with only a four percent difference between the efficiencies of the 1C to 4C discharge rate (Table 4). The lower efficiency rate observed at the 0.5C discharge rate may have been due to the longer period of time during which the cell balancing circuits were active.

Thermal Performance of the System
Temperature was measured throughout each cycle to determine the temperature rise for each rate in Table 5. For both charges and discharges, the temperature rise was calculated using the maximum temperature at or after the end of each charge or discharge. In some cases the maximum temperature did not occur until several minutes after the charge or discharge was completed; this maximum value was used. However, cooling was seen during the constant voltage portion of some of the charges (as the current tapered). In this case, the temperature at the end of the charge was used, and the charge rises may take into account some of this cooling (and not reflect the absolute maximum temperature rise observed during the charge). In addition, the negative temperature rise for the 1C charge can be attributed to cooling of the GBS after an elevated initial temperature. Fig. 3 below shows the temperature profile for the 4C discharge. .2 Continu roughout the ay in capaci le has been s correspond ry 1,000 cy acity versus . 5. s figure also the testing. ting after the bient temper e rack was lef the temperatu orded tempe wn in Fig. 6