Knowledge Base

Question: Why is it necessary to balance lithium ion cells in a series configured battery system?

Answer: Most battery chemistries require the cell voltages to stay within certain upper and lower limits in order to either maximize their service life and/or in some cases operate without immediate and/or catastrophic failure.  So for service-life, safety and reliability reasons, it may be important to monitor each of the cell voltages and stop charging for example, when any one of them gets to the upper safe limit and stop discharging when any one of them gets to the lower voltage limit. 

As an imbalanced series string of battery cells is charged and discharged, there will be a difference, a spread, between the upper and lower battery voltage.  If this spread is large enough, the string will deliver a noticeably smaller percentage of its energy content during full discharge cycles.  This is because some of the cells are not being fully charged on the way up, and the other cells are not being fully discharged on the way down. 

If the string is balanced, every cell can be charged to its maximum voltage during recharge, and every cell can be brought to its minimum allowable voltage during discharge.  In this case every cell delivers its full energy to the load. 

In every practical battery string, each of the cells will have different voltages with respect to each other.  These differences stem from variations in the cell manufacturing process and variations in operational conditions.  Tolerances in the electrode material loading, active material make-up, and other factors can lead to capacity and voltage vs. state-of-charge (SOC) differences in each of the cells in the series string.  While operating, variations in cell temperature across the series string can lead to different rates of self-discharge in each of the cells, leading to variations in cell voltages.  Finally, variations in cell performance can grow over time as each of the cells ages differently in response to its environment and physical construction. 

Some battery chemistries need balancing less than others.  In batteries such as Nickel Cadmium, Nickel Metal Hydride, and Lead-Acid there are electrochemical processes which cause self-discharge rates to be proportional to cell voltage.  This will cause the cells to naturally balance over time. 

Lithium Ion batteries and super-capacitors are examples of energy storage devices that have no self-balancing electrochemical processes in them.  If maximum energy is desired over the course of such a string’s service life, some means of balancing is required to make up for cell voltage divergence due to variations in cell construction, environment and aging which are inevitable in every physical and non-ideal battery string. 

Ideally, to maximize deliverable energy from a series string, a balancer would work full-time, and be able to keep the cells balanced in a lossless manner no matter how fast they diverge or how fast the pack is cycling up or down. 

For example, suppose that one of the cells in a five-cell string is weak and during discharge, will reach the minimum cut-off voltage while the strong cells each have 5% of their energy left in them.  Without balancing, the pack would stop discharging when it still has about 4% of its total capacity left.  If however, the string is balanced the discharge will stop with 0% of all the cells' energy remaining when any of the cells reach its minimum cut-off voltage. 

Given that there are real costs, complexities and limitations associated with high-current, low-loss balancing circuits, judicious selection of the right balancing methodology based on real needs is required for every application.  

Question: When we put our PCS into grid-support mode, where it is correcting for instantaneous deviations in voltage, power-factor, or frequency, we notice that the PCS draws "charging" power from the grid, even though the grid conditions are such that it would not need any corrections from the PCS. Why does the PCS draw this power in these modes?

Answer: When the PCS is in a grid-support mode, it has to respond to the grid conditions very rapidly, so it enters an idle mode, similar to car idling its engine with the clutch disengaged- ready to respond at a moments notice.  However, just as in a car, the PCS uses energy to be in this mode, whether it is powering its controls, it cooling loops, occasionally firing its IGBTs in response to spurious line noise, or keeping its gate charge capacitors full with an occasional switching. That’s the best case scenario where you have variable frequency (sliding mode) control of the bridge.  Some are worse as they switch their IGBTs at a fixed frequency of multiple kilohertz, but with very low duty cycle.  

This idling power has to come from either the battery or the grid.  If it comes from the battery, the battery would soon run out of charge and be worthless to the operator.  So, typically, PCS designers adjust their current control so that the idling power comes from the grid so that the battery maintains its SOC.  Sometimes, they employ a “Charge-sustaining” mode during these idling periods, to bring in whatever current from the grid is required to maintain the battery at its present SOC. 

Precedent is set for type of behavior.  Consider a transformer sitting on the line between the PCS and the grid.  Just to keep the core magnetized, the transformer draws a small amount of power from the grid.  If you wanted to avoid this constant power draw, you would disconnect the transformer whenever you are not using it.  However, to use it agin, you would have to go through a time-consuming process of connecting the transformer back to the grid.  Similarly, in a PCS, if you want to remain ready to go to instaneously respond to grid deviations, you have to be in an idle mode, consuming a small but non-trivial amount of power.

Question: My engineering firm is telling me I have to oversize my PCS by 10% to account for line losses to the POI?  is that realistic?  It seems high.

Answer: That's not too far from reality. There are a few losses to consider when sizing a power conversion system (PCS) in any energy storage system with transformers and transmission lines. Transformers especially, add impedance between the PCS and the point of interconnect (POI). Because of this, PCSs must be rated with a power capacity that is higher than the required power at the POI to overcome the effects of these components.

Simplistically, the battery delivers energy to the grid through a PCS and a transformer. The PCS creates ac voltage and current from the batteries dc voltage and current and the transformer converts the PCS’s output ac voltage to the grid’s ac voltage. 

The battery and PCS can be modeled as a single ac voltage source and the transformer can be modeled, for the most part, by two main components, resistance and inductance (neglecting core losses for the moment).

Power flowing into or out of any ac system, whether it be a PCS a transformer, or a grid connection point contains two components: real and imaginary. Real power is that which is doing real work, and Imaginary is that which is flowing back and forth through the wires and not performing any work. Despite their names, they both contribute to real current flowing in and out of power devices and must be dealt with by current-handling devices. Real power is caused by resistive loads that draw current in phase with the voltage from generators. Imaginary power is caused by the presence of capacitances and inductances in an ac system and causes currents to flow 90° out of phase with the voltage.

The mathematical construct to represent these two power components is “complex” power, which is a vector in which the horizontal component (x-axis length) is the real component of power and the vertical component (y-axis) represents the imaginary component. The 90° geometrical configuration correlates to the 90° phase relationship between the currents associated with each. Hence the geometrical angle of any power vector represents the phase of the current or power it is representing.

Using geometry, we can calculate what the various components of power and losses are between the PCS and the POI. The complex power at the POI, is the vector sum of the real power delivered into the grid at the POI, and the reactive “VARs” provided into the grid. Often utilities will require generating assets, like turbines, windmills, PV arrays and batteries, to provide a certain amount of VAR support whenever they call for it. The ratio of real power and the magnitude of complex power is the power factor (abbreviated PF). In general, PF is an indication of how much real work is being done by the current flowing in the lines.

In addition, since the PCS powers the system’s auxiliary power and the transformer’s core magnetization losses while discharging, the real power output must include these extra components.

The total complex power provided by the PCS under full power operation can be geometrically derived from the complex power at the PCS terminals, the real power at the POI, the wiring resistance of the transformer, the series reactive losses in the transformer, and the power factor at the POI, .

As an example, if the grid load is maxed out at 100%, and has a 95% power factor, the complex power at the POI will be approximately 105% of rated real system power. However, if a transformer with 1% per-unit real losses, and 7% reactive impedance is inserted between the PCS and the POI, the complex power at the PCS can be calculated to be another 4% larger than that delivered at the POI, or a total of 9% higher than the real power delivered at the POI. 

Therefore, it is important that the PCS and energy storage system be sized large enough to overcome the real and reactive losses between them and the POI while supplying the required real and reactive power to the grid.