IEEE Electrification - December 2021 - 21

adequate to certify new systems for safety? Will smallscale
single cell testing, which may not simulate realworld
conditions, be sufficient or will full-scale testing,
which is prohibitively expensive, be necessary.
Engineering Safety
Safety engineering has a long history, and current industrial
practices in safety engineering were built, in part, out
of industrial or occupational safety requirements. Engineering
safety is to reduce risk of failures or mitigate
potential for accidents or incident with substantial effect
on life and property. In the energy storage domain, risk
management is a useful framework for considering engineering/testing
standards, along with siting/permitting
policy. Safety standards establish a list of falsifiable qualities
for an engineered system that, when checked, reduce
the risk of accidents. An example of this kind of standard
is the definition of size, separation, and maximum allowable
quantity limits for different battery types, as is done
in the 2020 version of NFPA 855. Enforcing these limits,
with specific exceptions, constrains the feasible worstcase-scenario
by preventing a fire in one block of cells
from spreading.
Battery Thermal Runaway
All forms of stored energy can be a hazard if that energy is
released without control. Batteries are inherently remarkable
systems, as they remain inert until an external circuit
is connected, and only then release energy as electric current.
Battery failure occurs when an event, either internal
or external to the battery, causes this stored energy to be
released in an uncontrolled manner. If that release is fast
enough, heating can create a feedback loop of increasing
heat generation and subsequent combustion called thermal
runaway. During thermal runaway, the release of
stored energy is occurring so rapidly that intervention is
effectively impossible. Fire is typically described as a triangle
among fuel, oxidizer, and heat. In a charged Li-ion battery,
the two electrodes effectively provide the fuel and
oxidizer within the cell, meaning the heat needed to initiate
a thermal runaway is the only item missing. While fire
resulting from combustible materials might be extinguished
through a variety of means, the only way to stop
the thermal runaway event is to remove the heat that is
sustaining it.
Broadly, single cell failures can be grouped into field
and abusive failure categories. Abusive failures are relatively
easy to identify as the failure is driven by the battery
being exposed to a condition it is not designed to withstand.
This can include mechanical damage or exposure
to temperature, voltage, or current outside the battery's
normal operating window. Field failures are more difficult
to predict and detect. They are roughly defined as an
unexpected failure while idle or during normal operation
of the battery, for example, an internal short circuit caused
by a contaminant in the cell.
The susceptibility of single cells or battery packs to
abusive failures as well as the potential consequences are
typically evaluated using standard industry tests (see Figures
1 and 2). Peak temperatures during failure, the conditions
where failure occurred, and video recordings of the
failure are all evaluated to determine the general severity
and susceptibility to abuse conditions. Taken as a whole,
this provides a view of the true safe operating envelope of
a particular battery as well as the worst-case scenario
during failure. This information can then be used to aid in
the development of a safe and reliable battery system.
The evaluation of field failures is more challenging as
statistical failure rates can vary by environmental and
operational conditions. Because these failures are typically
a rare and random process, it is difficult to evaluate the
risk of spontaneous failure of a grid connected energy
storage system. This is most easily evaluated in small,
commercial-off-the-shelf batteries from major manufacturers
as the large numbers of cells produced enables estimates
of a true failure rate, which is from where the
previously mentioned less than one in one million failure
rate comes. The large cell formats used in grid-scale storage
systems are typically much more specialized and produced
in relatively small quantities. This means
evaluating the vulnerabilities of grid storage cells to field
failure is difficult, and we are left with relying on battery
(a)
(b)
(c)
Figure 1. The failure of a 1-kWh battery pack from overcharging. (a) Test setup. (b) Battery fire resulting from overcharge. (c) Post-test condition.
IEEE Electrification Magazine / DECEMBER 2021
21

IEEE Electrification - December 2021

Table of Contents for the Digital Edition of IEEE Electrification - December 2021