Battery switch array system with application for JPL's rechargeable micro-scale batteries

2009 30th IEEE Real-Time Systems Symposium

2009

Abstract-Electric vehicles operate inefficiently with a naive battery management system that charges or discharges battery cells in a pack based solely on application load demands. The battery pack's operation-time and lifetime can be extended significantly by effectively scheduling (the cyber part) battery charge, discharge, and rest activities, based on the battery characteristics (the physical part). We propose a set of policies for scheduling battery-cell activities, called the weighted-k roundrobin (kRR) scheduling framework. This framework dynamically adapts battery-cell activities to load demands and the condition of individual cells, thereby extending the battery pack's operationtime and making them robust to anomalous voltage-imbalances. The framework comprises two key components. First, an adaptive filter estimates the upcoming load demand. Then, based on the estimated load demand, the kRR scheduler determines the number of parallel-connected cells to be discharged simultaneously. The scheduler also effectively partitions the cells in the pack, allowing the cells to be simultaneously charged and discharged in coordination with the battery reconfiguration system we developed earlier [17]. Besides the kRR scheduling framework, we characterize the discharge and recovery efficiency of a Lithiumion battery cell. The kRR scheduling framework is shown to outperform three alternative scheduling mechanisms with respect to the operation-time by 7-56%, and improve the tolerance of voltage-imbalance by up to 50%.

Section: A Reconfigurable Battery Systemmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Scheduling of Battery Charge, Discharge, and Rest

Kim

2009 30th IEEE Real-Time Systems Symposium

2009

“…To tolerate the failure of battery cells connected in series, researchers proposed addition of programmable switches around battery cells to bypass faulty cells [1], [6], [18]. While these methods work well for small-scale batteries via static configuration, they are not designed to reconfigure large-scale batteries in real time.…”

Section: Introductionmentioning

confidence: 99%

“…Pack-sizing determines not only the number of cells in a pack but also the total number of cells. Existing solutions [1], [18] do not account for the effect of pack size on the cost. They also fail to compute the number of necessary backup battery cells, thereby either compromising the ability to provide the required power reliably, or requiring more backups than needed in order to maintain the required level of power in the presence of cell failures.…”

Section: Introductionmentioning

confidence: 99%

Pack Sizing and Reconfiguration for Management of Large-Scale Batteries

Jin

2012 IEEE/ACM Third International Conference on Cyber-Physical Systems

2012

Abstract-Battery systems for electric vehicles (EVs) or uninterruptible micro-grids-prototypical cyber-physical systems (CPSs)-are usually built with several hundreds/thousands of battery cells. How to deal with the inevitable failure of cells quickly and cost-effectively for vehicle warranty or uninterruptible service, for instance, is key to the development of largescale battery systems. Use of extra (redundant/backup) cells to cope with cell failures must be minimized so as to make the target systems cheaper and lighter, while meeting the reliability requirement that is directly related to, for example, the vehicle warranty. Existing reconfigurable battery systems do not scale well because they incur a long delay in properly setting a large number of switches to bypass faulty cells or adapting to dynamically changing power demands in large battery systems for such applications as EVs.In this paper, we propose a scalable solution, not only to reduce the required number of backup cells and the total cost of a battery system, but also to facilitate recovery from cell failures and adapt to changing power demands while increasing battery utilization. Specifically, we optimize the pack-size by striking a balance between various types of cost in order to reduce the overall cost. We also configure battery packs and optimize their connection topology, reducing delays in failure recovery and power reallocation. Our in-depth evaluation has shown that the time to recover from cell failures remains constant irrespective of the number of cells involved, which is important to scalability. The proposed pack-sizing also reduces the cost and the size of battery systems. Moreover, fast power reallocation is achieved by utilizing prior knowledge of power usage patterns.

“…In such a modular battery-management system, individual electronic control units (ECUs) collect information-such as cell voltage and current, temperature, etc.-on their serially-connected battery packs via an equalizer connected to each battery cell, and then process and report the collected information to the central ECU responsible for making the local ECUs work properly. To handle the case of any battery cell becoming open-circuited, the battery pack can be made configurable [8] by adding additional controllable switches around each battery cell, detouring any faulty battery cell [1,6,22]. These solutions are, however, dedicated to micro-scale batteries based on static configuration and hence limited to physical processes, requiring more interactive computations online to cope with a large-scale battery-management system.…”

Section: Introductionmentioning

confidence: 99%

On Dynamic Reconfiguration of a Large-Scale Battery System

Kim

2009 15th IEEE Real-Time and Embedded Technology and Applications Symposium

2009

Abstract-Electric vehicles powered with large-scale battery packs are gaining popularity as gasoline price soars. Large-scale battery packs usually consist of an estimated 12,000 battery cells connected in series and parallel, which are susceptible to batterycell failures. Unfortunately, current battery-management systems are unable to handle the inevitable battery-cell failures very well. To address this problem, we propose a dynamic reconfiguration framework that monitors, reconfigures, and controls large-scale battery packs online. The framework is built upon a syntactic bypassing mechanism that provides a set of rules for changing the battery-pack configuration, and a semantic bypassing mechanism by which the battery-cell connectivity is reconfigured to recover from a battery-cell failure. In particular, the semantic bypassing mechanism is dictated by constant-voltage-keeping and dynamicvoltage-allowing policies. The former policy is effective in preventing unavoidable voltage drops during the battery discharge, while the latter policy is effective in supplying different amounts of power to meet a wide-range of application requirements. Our experimental evaluation has shown the proposed framework to enable the battery packs to be 9 times as fault-tolerant as a legacy scheme.