Abstract
Evolving technologies such as EVs and smart grids
require their batteries to be operated safely and cost-effective
and a battery management system is always used to achieve
that. This study focused on building a passive balancer for
four cells connected in series of a battery pack. Two voltage-based
algorithms for passive balancing were tested and results were
compared.
Batteries premature failure can be caused by not making
use of their full capacity. Variations in the physical volume,
internal impedance and different self-discharge rates of the
cells are a cause [2]. Then, as not all the cells are identical,
they will charge and discharge at different rates and the
charging/discharging process will stop as soon as the first
one reached or finished its full capacity. To prolong the cell's
life, their state of charge can be balanced during a charge or
discharge process. The main methods available are passive and
active balancing. Passive balancing dissipates the energy
from the strongest cells discharging them through a resistor,
until the weakest ones reach their full capacity. The same
principle is applied by active balancing, only that the energy
is not wasted but transferred from the strongest to the weakest
cell. In [3] the SoC profile for a car battery is compared before
and after cell balancing over a 12h driving cycle. After one
driving cycle, the cells' voltages vary and after a relaxation
period, they will start to drift further away. These variations
were translated in quite big SoC variations in individual cells.
For the same current profile, cell balancing was introduced.
After driving for 0.5h, the highest and the lowest charged
cells were balanced, continuing with multiple cells balancing.
Comparing the graphs, it can be seen how the active balancing
circuit reduced the cells' variations. Balancing algorithms are
based on SoC calculation. This can be achieved by voltage
translation and/or coulomb counting" [1]. The main limitation
of the voltage-based algorithm is that it assumes that the cells
of different voltages are of different SoC, when the terminal
voltage of a battery is not the same as the OCV (open circuit
voltage). A voltage drop across its internal resistance can rise
or drop the terminal voltage of the cell while charging or
discharging. In their application notes for cell balancing ASICs (application specific integrated circuit), Texas Instruments also
state that terminal-voltage-based balancing may not be accurate,
due to the IR contribution of the internal resistance of
the cells while charging/discharging. The relaxed voltage after
charging (the true OCV) shows imbalances between the cells.
Figure 1 is used to illustrate this [4].Some studies have discussed the problem associated with deriving the SoC from the terminal voltage. For example, David Andrea in his book [5], shows how the terminal voltage
curve vs. the SoC is different for various discharge and charge
rates as the current variation leads to a variation in the IR drop.
However, knowing the current and the resistance for each cell,
the OCV can be calculated. [6] also tried to approximate the
SoC directly from the terminal voltage for a Lead Acid battery
and have concluded that for different currents, the parameters
for the approximation curve were different. Then, it proposes
two equations to model the SoC as a function of both the
terminal voltage and the internal resistance of the battery, as
both vary with the SoC and the error for both were compared.
Other methods for deducing SoC from the OCV are [7-9]
or [10-11] and they require a long time for the voltage to
relax to its open circuit value. However, all the studies mentioned above do not assess the performance of the SoC estimations for a balancing system. The novelty of this study is that it aims to confirm the importance of the internal resistance for the SoC estimation, comparing two balancing algorithms: one inferring the SoC of the cell from the terminal voltage only, and one using the derived OCV (calculated with the use of the known
current and internal resistance). |
