How smart is a battery?

The invention of the rechargeable battery in the mid-1800s was a game changer, to say the least. What emerged from this discovery was a new system of energy storage and new ways to use it, allowing for major technological advances across many fields. For nearly 100 years, however, the battery was a black box, a system that was impossible for a user to see inside. This meant that unlike with fuels such as gasoline or kerosine or even lumber, you couldn’t tell by looking how much energy was left to use.  

When considering the battery as an energy source, with electricity as the fuel, it is easy to compare to a gasoline system. 

Let’s think about a lawn mower as an example. For a gasoline lawn mower, we add gasoline to the fuel tank until it is full. You can tell it’s full just by looking in the tank. If you know the size of the container, and you measure how much fuel you’re putting in, you can instantly determine how much fuel was added and how much it previously contained. We can then use the mower until we see the fuel level has dropped to a certain level, at which point we can add more. 

On the other hand, if the same lawn mower was powered by a battery, it would run until the battery is out of energy, but we cannot see exactly when that will be. It is an opaque system. With a battery, we will never have a perfect answer for:

• How much energy the battery has remaining
• If it has sufficient power for a particular use, or 
• The health and condition of the battery.
  

Can We Make a Battery Smarter?

The solution to these unknowns is the smart battery, which integrates communication between the battery, the electronic device powered by the battery, and the user through use of a battery management system (BMS). The BMS gets data from the battery, extrapolates and computes information about its state, and communicates to the user. Although the BMS cannot directly see into the battery, it can give us more information about its state than we had before. 

Determining State of Charge

One of the main things we need to know about batteries is their state of charge. State of charge (SoC) is how much energy is available to use in a battery, much like a volume measurement of a gas tank. 

If the state of charge is 50%, then half the battery’s maximum electricity is available. A 100 kWh battery that is at 60% SoC means that there is 60 kWh of energy available for use. But, while a gas fuel gauge measures liquid flowing in and out of a tank with a known size and minimal losses, it is not so simple with a battery. You can still technically measure electricity going in and going out, but there are losses along the way, and the size of the battery itself may change as it degrades. 

Estimating the SoC is the primary function of the battery management system, and is typically achieved via coulomb counting, which measures the flow of energy in and out of the battery. The accuracy of this method depends greatly on the initial measurements and a good understanding of how much energy is lost along a circuit. Coulomb counting also depends on battery chemistry. In lead acid batteries, for example, this method is quite inaccurate since two big things are not accounted for: current leakage and the fraction of battery charge that is lost as heat waste during use. 

One major advantage of the Li-ion battery is the lack of these two factors. However, even with a lithium ion battery, the fuel gauge can only see the “open circuit voltage,” which is the potential energy of a battery with no device attached to it. The open circuit voltage differs from the operational, closed circuit voltage of a battery (like, when it's in use) because adding in a device introduces resistance, which is energy loss due to friction. To make things even harder, total battery capacity changes over time.

It’s like trying to measure a gas tank that keeps changing size

The biggest issue with coulomb counting is that in order to measure energy in and out, you need to know the initial SoC. Here’s an example. Say you have a bucket that holds an unknown amount of water. If you measure 3 gallons going in and 3 gallons going out, you know the bucket is empty and has a 3 gallon capacity. 

However, say there is already some water in the bottom of the bucket, or some rocks that some kid threw in. You can measure how much water you can add, but unless you know how much was already in there, you won’t be able to know the total bucket capacity. Say the water already in the bucket takes up 20% of the volume, but you estimate it takes up 30%. Then, your calculation of the bucket’s total volume will be off by 10%. The next time you calculate the volume, you will be starting with an error. 

Here’s what that looks like:

1. You have a big bucket with 20 gallons of water in it. However, you are incorrectly guessing it has 30 gallons already in it because you can’t see 
2. You add in water and measure 80 more gallons going in
3. Since you assumed it already had 30 gallons in it, you will calculate the total volume of the bucket to be 110 gallons (30 + 80). But, there were really only 20 gallons, so your estimate is off by 10 gallons. The bucket is actually 100 gallons
4. Next, you measure 60 gallons come out
5. Since you assumed the total capacity was 110 gallons, when you see the 60 come out, you assume there are still 50 gallons in the bucket (110 - 60 = 50). But, there are only really 40
6. You add another 50 gallons to “fill it up” but in fact, the bucket only holds 90 gallons, or 90% capacity. You think it is full, but there are actually 10 more gallons to go. 

‍

The problem of compounding errors in SoC measurements is referred to as ‘drift.’ It’s made worse by battery aging and degradation, which make the total capacity of the battery a moving target. If your initial estimate was off by 10%, after years, the actual capacity of the battery may shrink, leading you to be off by even more. 

Importantly, a SoC gauge that displays 100% implies a fully-charged battery, but is not an indication of the total usable capacity. A battery can be 100 kWh but only have 90 kWh of that capacity be usable. 100% SoC would then mean that only 90 kWh are available. The SoC gauge cannot “see” the gross capacity, only what is available to the user, leading to further errors in the estimation.

Why Does An Accurate Estimation Matter? 

In the case of an EV, it should be obvious why an accurate SoC measurement is desired. As we’ve covered before, range anxiety on lengthy trips will continue to exist until the charging infrastructure is at a point that matches gas stations, and the BMS range calculation is the main tool to combat it. An SoC estimation that is off by 10% could mean a 30 mile miscalculation in range for a 300 mile range vehicle. Worse, if you are overcounting your estimated state of charge, you may have less battery than you expect. Luckily, modern technology has come a long way in terms of accurate state of charge estimates. 

It’s also worth noting that accurate SoC estimation can help the BMS with its own internal tasks that help the battery stay healthy. Cell balancing, or keeping battery cells at roughly the same charge level, is an important maintenance task that the BMS is responsible for, and having a good SoC measurement helps. The BMS also helps avoid over-charging or over-discharging of cells, which could lead to battery damage over time. This too requires knowledge about the current state of charge.   

Written by Brandon August, a lifelong explorer of all things academic. After obtaining an undergraduate physics degree and a doctoral degree in biomedical, he began to explore various professional fields in health and wellness, rideshare work, freelance writing, and day trading. On the recreational side, he has always been involved in the automotive field, owning various vehicles across the years. After a recent move to California, he entered the EV space, purchasing both a Chevrolet Bolt EV and a Bolt EUV for his household.