IEEE Power & Energy Magazine - September/October 2017 - 27

table 4. Representative installations of VRB.
Company

Location

Basic Specification

Application

Year of
Installation

Prudent Energy

Zhangbei, China

2-MW/4-MWh VRB

Balancing wind and solar power

2011

Rongke Power

Liaoning, China

5-MW/10-MWh VRB

Coupling with wind power (smoothing/
ramping, transient active support, etc.)

2012

UniEnergy
Technologies

Washington,
United States

1-MW/3.2-MWh VRB

Load shifting, frequency regulation,
voltage regulation

2015

Sumitomo Electric
Industries

Hokkaido, Japan

15-MW/60-MWh VRB

Coupling with wind and solar power
(e.g., frequency fluctuation suppression)

2016

EVs (e.g., Tesla's Model S); for home and utility energy storage, Tesla's Powerwall and Powerpack adopt Li-ion battery
technology as well. The Tesla Gigafactory plans annual Li-ion
battery production capacity of 35 GWh and anticipates driving down the cost of an Li-ion battery pack by more than 30%.
It is clear that the Li-ion battery is among the most promising
technologies in electrical energy storage, and its installed capacity is well poised to increase.

Battery Management
Notwithstanding significant progress in battery chemistries
and materials, effective and dependable battery management
systems (BMSs) are still needed for the condition monitoring,
charge/discharge regulation, thermal control, cell balancing,
health prognosis, and safety protection of large-scale battery
energy storage. The absence of such systems is most likely
the reason for the conservative use of batteries (e.g., 20-50%
excess energy capacity, evoking undesirable weight, volume,
and costs). Without appropriate BMSs, catastrophic hazards and
premature failure, such as thermal runaway, may occur owing
to poor electrical and thermal operating and maintenance practices. Battery management thus plays a critical role in the integration of battery energy storage into the electric grid in terms
of performance, safety, reliability, and economy. As shown in
Figure 6, the main functions of a BMS include the following:
✔✔ data acquisition: the measurement and collection of
data on current, voltage, temperature, etc.
✔✔ state estimation: high-accuracy gauging of SOC, state
of power (SOP), state of health (SOH), state of temperature, etc.
✔✔ charge/discharge control: charge current/voltage regulation, power electronics interface, etc.
✔✔ cell balancing: passive or active state-of-charge and
voltage equalization
✔✔ thermal management: control of the maximum temperature and temperature deviation among cells inside
a battery pack
✔✔ safety protection: hardware setup for avoiding overcharge/
overdischarge and overheating, as well as hardware/
software redundancy for proactive fault diagnosis/isolation
and alarming.
september/october 2017

The key enabling technologies for these functions in BMSs
are dedicated to battery modeling, SOC/SOP/SOH estimation,
cell balancing, charging control, and fault diagnostics. Because
batteries' internal states are generally inaccessible through present in situ sensing techniques, models are established with the
aim of mimicking their dynamics and constituting a duplicate of
a real battery. Then, based on such models, battery state/parameter behavior can be probed and manipulated by sophisticated
estimation and control approaches. Additionally, large-scale
battery storage invariably consists of thousands of cells in series
and parallel connections to satisfy power and energy requirements. Challenging but critical BMS functions maximize the
potential of a whole battery system by balancing the charge on
cells and avoiding the bucket effect caused by the weakest cell.
The BMS also realizes system supervision by conducting fault
diagnosis and prognosis of the state of the cells.

Battery Modeling
Mathematical modeling of battery dynamics is fundamental and is also an efficient tool for advanced battery management. The related work can be categorized into three types:
white-, black-, and gray-box models. The white-box models
are developed from first principles and accurately capture
internal battery dynamics such as ion diffusion, intercalation
kinetics, and electric potentials. Mathematically, these models are initially subject to partial-differential equations and
can be computationally intractable for online implementation.
Consequently, model-based algorithms in this field are often
fused with reduced-order modeling, resulting in minimal sacrifice of precision and physical interpretations. Alternatively,
data-driven or so-called model-free approaches can be used
to model battery characteristics, such as neural-network and
machine-learning approaches. The foregoing two methods
represent end points on a spectrum. A scheme in between is
a gray-box model (also referred to as an equivalent circuit
model) that approximates internal battery dynamics with some
voltage sources, resistors, and capacitors. These models are
relatively easy to implement but suffer from the lack of physically meaningful parameters and insights. As a result, precise
constraints that achieve safe and optimal operations are difficult to impose with gray-/black-model-based algorithms.
ieee power & energy magazine

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - September/October 2017