klooiblog

Door mux op maandag 29 januari 2007 10:27 - Reacties (3)
Categorie: Techniek, Views: 4.326

How to manage a battery

In this article, we will take a very in-depth look at lithium ion batteries used in cars and how they are managed. This includes balancing, thermal management and capacity estimation. Using this information, we can deduce what Nissan did wrong with their overheating Leaf battery packs

What is a battery?

To be clear, we will be talking about lithium-in batteries in an automotive context - BEV batteries. In the automotive sphere, the term 'battery' is generally used to point at the box that sits under your car - which contains much more than just a big electrochemical cell. A battery typically contains:
  • Electrochemical cells ('cells')...
  • ...grouped together in modules containing 2 or more cells
  • Compression frames around the modules or groups of modules
  • Some thermal management (cooling/heating) system
  • Multiple temperature sensors
  • Voltage sensing on each cell
  • Shunt resistors (used for balancing) on each cell
  • A current sensor for the entire pack
  • A big honking fuse
  • Large contactors (the name for really big relays)
  • A precharge resistor and relay
  • A big steel or aluminum housing
All of these components have an important function and basically none can be left out for a well-functioning automotive battery.

Cell grouping and compression frames
Lithium ion cells, like all other electrochemical cells, store energy in chemical bonds. More precisely; they produce a current (and voltage) when converting one chemical into another, and for lithium-ion chemistries, this reaction can be reversed by reversing the current. But however well this chemistry is optimized, you are still dealing with two different chemicals in the charged and discharged state, and different chemicals tend to have different physical properties. One of these properties is density, or how closely-packed the molecules are, and for the most common types of lithium-ion batteries, the active ingredients in the battery tend to take up more space when the battery is full compared to when it is empty. In other words; the battery expands a little bit when charged.

This has big consequences for longevity. When a battery is charged and it expands, it may not expand the same way everywhere. If we take a pouch-type lithium-ion cell like those used in the Nissan Leaf, the pouch tends to bulge outwards on the large faces when charged. In the middle of the pouch, this means the active material is least dense, while on the edges it is closer together. This causes a lower resistance path to form on the edges of the cell, increasing the rate of discharge there as compared to the middle of the cell and heating up the sides a little more. Moreover, the mechanical movement of the middle of the cell will 'mash' the insides of the cell, causing mechanical damage over time. This is why cells tend to have a big metal cage around them, called a compression frame. This frame makes sure the cells stay flat throughout. They also generally make for easy to assemble packages with relatively few electrical and mechanical connecions. The MUXSAN extender batteries use modules with 12 cells in a compression frame, Nissan uses a few different groupings; cells are grouped 2 at a time and then grouped with 6-16 groupings in a compression frame.

And why do we use many cells and not just one? Well, this all has to do with the types of voltages you need to efficiently deliver the power needed to drive a car. Cars have standardized to about 400V, and with li-ion cells being about 4V per cell, you need roughly 100 cells in series to get to that voltage. Moreover, if you would make one giant cell and even if you could run a car off 4 volts - that cell would have terrible thermal characteristics, likely boiling in the middle even with the sides touching a refrigerated surface. It helps to chop up the battery in many tiny bits to increase the surface area and improve thermals.

Thermal management
There are a lot of misconceptions about thermal management, so let's try to reduce them a bit. Batteries are chemical devices, so they have to obey Arrhenius' law of reaction speed. Roughly speaking, for every 10C increase in temperature, chemical reactions occur twice as fast. In batteries, this means batteries will be better able to accept and provide current at higher temperatures - by quite a lot. Here is a table of the internal resistance of Nissan's 40kWh battery as it changes with SOC and temperature:

https://tweakers.net/ext/f/yBJYQfcsyZp63JTbGDCFtrqe/full.png

Looking at the temperature dependence, the internal resistance doesn't quite reduce by a factor of 2 for every 10 degrees, but for some intervals it isn't that far off. Of course, reaction speed isn't the only thing that determines internal resistance. There is also the resistance of the copper and aluminum conductor plates that form the terminals of the battery and some other more complex physical processes that don't scale exactly according to Arrhenius. Internal resistance isn't one 'thing', it is a representation of all lossy processes inside the cell.

But, well, low internal resistance is a good thing, right? Low internal resistance means lower losses, so a battery should be kept at high temperature? 60 degrees, nice and full, that's clearly good according to this graph?

Well, not so fast. IR isn't everything. If only we would have a world where everything is decided by a single number. Higher temperatures means batteries will be more cooperative, but they also degrade faster thermally. A battery is a big bunch of chemical reactions. We prefer the particular reaction that charges or discharges the battery, but many more are possible, just discouraged. At higher temperature, those possibly damaging side-reactions will also go faster and degrade the battery more quickly. This is why there is a bit of an optimum. Around 25C, the battery is both warm enough to be spared of the low reactivity of cold weather and simultaneously cold enough to avoid calendar degradation.

But thermal management isn't just about removing or adding heat to get the temperature to around 25C. As you may know, Tesla heats up their batteries before fast charging - that reduces the internal resistance. But during driving, it keeps the battery colder. Clearly, there isn't one particular 'best' temperature at all times. What really counts, is that the battery is at a consistent temperature! After all, when one part of the battery is at a higher temperature than another - or worse: when one part of a *cell* is at a different temperature from another part - you will get parts of the pack that charge at a different rate from the rest. Or even that degrade differently. This is a big problem in the 40kWh and 62kWh Nissan Leaf, where the top of the pack typically is 5-10C hotter than the bottom. This will cause the pack to degrade and charge unevenly, ruining performance and longevity.

So really, we shouldn't think of thermal management as heating or cooling the battery - we should primarily see it as keeping the entire package at the same temperature throughout. Exactly what temperature is not as important.
quote: practical example: 40kWh Leaf battery losses
The internal resistance causes losses in the cell, and these can be large. Say we have a pack at about 50% charge and 25 degrees C and we accelerate at full power. This is about 400A draw or 200A per cell at an internal resistance of 1.14 mohm per cell or 96x1.14=109.44 mohm. The total losses then are P = I^2R = 2 x 200 * 200 * 0.10944 = 8754W. This is an extreme example, but about 6.7% of the power is wasted in the battery, heating it up.

Well, actually, 6.7% isn't that bad. But this is just the beginning; we're actually doing what batteries like: being discharged at around 25C. How about charging, and how about charging a degraded battery? Here is the degradation characteristic of the 40kWh Leaf battery:

https://tweakers.net/ext/f/7ShD2asuAFo6XspBPkzvEq0K/full.png

You can see that the initial bit of degradation - from 100% to 90% - is not bad at all. The internal resistance only goes up by 9%, which is practically negligible. The real damage occurs when batteries pass about 88% SOH; internal resistance goes up really quick, almost doubling for every 10% additional degradation. A battery at 80% SOH has an internal resistance of around twice that of the original battery. In the industry, a 100% increase in IR is typically the point at which a battery is considered end-of-life. Nissan considers the batteries EOL at 60%, when the IR has risen to 6.3X as much as new.

When discharging the battery, internal resistance is dominated by different effects than when charging the battery. This usually means that the effective internal resistance (or maybe more accurately: the internal losses) when charging are much higher than when discharging. Nissan doesn't give a charging IR table, but they do give a polarization characteristic with this tidbit in it:

https://tweakers.net/ext/f/HmuucDBVL5Vs0vdbMl9jDtEi/full.png

This means the battery has similar losses when discharging at 150A or charging at 62.5A - 2.4x less. This in turn means the internal resistance when charging is approx. 2.4 squared or 5.76x higher, over long time periods.

So let's put all of this together and figure out why the 40kWh Leaf rapidgates (overheats when quick charging). Let's try to charge at 25C again, but with a battery at 90% SOC. This gives us an internal resistance of 109.44 x 1.09 x 5.76=687 mohm. At maximum charging speed - 125A from CHAdeMO or 62.5A per single cell - that translates into 2 x 0.687 x 62.5^2 = 5367W. All of this goes into heating the battery. Say you charge your Leaf battery from empty to full. On average, the fast charger provides 44kW, but in reality the battery only absorbs 44000-5367=38.6kW, so it takes almost exactly a full hour to charge the 38kWh of usable capacity in the battery. In that time, you will have also pumped 5300Wh or 19MJ of heat into the battery!

This heat has to go somewhere, and this is where thermal management comes in. The Leaf has almost no ability to get rid of this heat. At 0.8J/gK of heat capacity and a total weight of about 300kg, the battery will heat up by 1 degree for each 375kJ. 19MJ then means a temperature rise of 50 degrees C! Yes, it is probably nearly impossible to fast charge a Leaf 40kWh battery from empty to full at full speed without it catching fire.


Cell balancing
As we saw in the previous chapter and example, even when charging each cell in a battery with exactly the same current, the cells may not really fill up evenly due to temperature differences. Maybe the manufacturing tolerances cause some cells to have a very slightly larger capacity than others. This means the pack will charge and discharge unevenly. But not just that; the strongest cells will suffer less during (dis)charging, so they remain good for longer than the weaker cells. The strong stay stronger, the weak become weaker. This is a runaway process that will ruin a battery pack in the long term.

So it's really important to balance the cells in a pack. This is done very simply; cells at a higher voltage are discharged very slightly and slowly. This will restore them to roughly the same voltage as the other cells, evening out the pack. This is what balancing (or shunt) resistors do in a pack, and it's a vital part of battery management.

Contactors and fuses
For safely handling batteries, especially in a dirty and chaotic environment like a factory or workshop, it is important that the batteries are completely disconnected from the rest of the car whenever the car is off. This is why there are always big, heavy-duty contactor relays inside batteries. These are special relays designed to safely disengage even when large currents are flowing. If you're familiar with welding - that's just a large electrical current flowing through the air. Air can conduct electricity, as can practically anything else if you try hard enough. Believe me, cars and EV traction batteries try REALLY hard.

Automotive contactors have special anti-welding (yes, they can weld stuck) and arc extinguishing features that make them a safe choice for disengaging the pack voltage from the outside. But now we run into another problem; you can't 'just' engage a battery.

In devices like electric motor controllers, there are very large capacitors that will quickly charge up when exposed to the battery voltage. Extremely quickly. Think kilo-amperes, many thousands of amps, if you don't do anything about it. This current can cause damage to wiring and relays, cause massive electromagnetic interference and will generally ruin your day. This is why all automotive batteries also feature a special extra relay and resistor called the precharge relay/resistor. This resistor limits the flow of current when starting up the car, avoiding damage to the electrical system.

Battery voltage, SOC and SOH

You can't have electric energy without both voltage and current. When no current is flowing through a battery, the battery sits at it's so-called open circuit voltage (OCV). This OCV is lowest when the battery is depleted and highest when the battery is fully charged. Most lithium-ion chemistries are empty at 2.5-2.7V and full at 4.2V. In between, it varies but not necessarily linearly. Older types of batteries (e.g. LMO and LCO) have an almost constant voltage from a state of charge between 20-80%, only dipping significantly below that voltage when really empty and only storing a minor amount of energy when charged to higher voltages.

So batteries don't have a fixed voltage. But a lot of batteries have a 'nominal' voltage, often 3.6-3.8V. This is often a voltage chosen to reflect the average voltage over the entire charge range, often a voltage that works well when multiplying the coulombic charge of the battery (Ah rating of the battery) with nominal voltage to get the total energy (Wh rating) for that battery.

Talking about Ah and Wh; these are quite different things. Ampere-hours or Ah(r) tell you for how long you can discharge the battery at a constant current before it depletes. A 25Ah battery can be discharged at 1A for 25 hours. But its voltage, and thus the discharge power, will change significantly during such a discharge.

A more useful rating is the Wh rating of a battery, which tells you for how long you can draw a fixed amount of power from a battery. A 100Wh battery allows you to draw 10 watts for 10 hours, for instance. When nearly full, this translates into, say, 2.4A at 4.2V. When nearly empty, this means 3A at 3.3V. In electric vehicles, we are usually more concerned with power than with current.

As a battery is charged or discharged, the voltage changes. First of all, this is because of internal resistance; a current flowing through any resistance will cause a voltage drop, and this is no different in batteries. But batteries do weird things as you charge or discharge them. They tend to be really good at supplying or taking current for a very short amount of time - a few seconds - but then suddenly their internal resistance rises substantially. In the long term, they can accept or supply progressively less current, even ignoring the fact that the battery will become emptier or fuller. So it is quite hard to estimate exactly how full or empty a battery is based on its voltage. That can change with circumstances.

Additionally, as we saw in the example a while ago, when charging a battery, it may charge slightly better or worse depending on - mostly - temperature. So it is not quite necessarily easier to determine the state of charge of a battery based on how many Ah it has absorbed or spent.

So how do we determine state-of-charge (SOC)? Well, it's a big educated guess. It is, for all intents and purposes, a matter of software tuning and experience. Various battery management systems will do this differently. But they're not magic and they do make errors. For instance, if you mostly drive your car between say 40-80% SOC and never really completely fill it, nor completely empty it, at a certain point the BMS will have to guess how much capacity is left in that bottom 40% or top 20%. It may be a few percentage points off, which has in the past led drivers of electric vehicles by the side of the road even though they had a few percent left according to the car.

The same goes for state-of-health (SOH). This is how much capacity is left compared to the original capacity of the battery, i.e. how degraded the battery has become. Cars like the Nissan Leaf report this with many significant digits, but in reality the error bars on this capacity are quite large and it is largely a guessing game. You can even see this guessing game in action; on typical drives the BMS will just automatically deduct 0.01% of capacity for each charge cycle, and then once every 20-30 cycles it will recalibrate and do a 'proper' measurement, suddenly chopping off 0.3-0.5% of the capacity.

Oh, and your battery's capacity? It's usually measured at really low discharge rates; 0.2C is a common rate. 0.2C means the battery is discharged at 0.2x its capacity rating per hour, aka it takes 5 hours to fully discharge the battery. For the Leaf 40kWh battery, that would mean discharging it at approx. 8kW. In practice, you are almost always either not drawing any significant current or drawing/charging at much more, so this capacity isn't even reflective of real-world capacity.

Degradation mechanisms

Of course, the big question that every EV owner has is: how do I make sure my battery lasts as long as possible? This goes into the way batteries get damaged over time. The two most important factors are temperature and voltage.

Batteries don't like to be hot for a long time. It's actually fine - comparatively - if you live in a temperate or cold climate and occasionally through subsequent fast charges or long trips heat up your battery a lot. These short periods of high temperature aren't that bad. It's being at high temperatures for very long periods of time - e.g. sitting still doing nothing in a desert climate like Phoenix, AZ. This is all down to Arrhenius again; those pesky side reactions that damage the battery from the inside that do their work that much faster at high temperatures.

But that damage doesn't happen quite as fast all the time. You need the second factor: high voltage. Batteries really don't like to be completely full, because voltages over say 4.0V/cell promote damaging chemical reactions in the battery. This is why so many EVs have the option to charge your battery to 80%; this typically leaves the battery at 3.8-4.0V/cell and avoids the 'danger zone' of high voltages. Again, this doesn't mean you should never charge your battery to full. If you know you're about to go on a long trip, by all means plug in the night before and charge to full. But if you want the absolute best longevity out of your battery, use that surplus energy and leave it at 80% or less when the car is sitting still for long periods of time.

Batteries also don't like to be really empty, but for a different reason. Actually, they're quite fine being empty - really empty, just a few percent remaining, at around 3.0V/cell. The problem is that the BMS in the battery still uses a tiny bit of energy - and in some cases the entire car periodically wakes up to do some stuff and uses energy from the battery. This will slowly drain the battery to a dangerously low level. Draining it too much, below about 2.5V/cell, will also cause irrecoverable capacity loss, although not nearly as bad as overcharging it.

Another reason why low SOC isn't a great idea is because older, degraded batteries tend to be very unbalanced at low SOC. Often, one cell may sit at 3.0V while another is 2.7V, so even though on average there is some energy left in the battery - even a small amount of discharge will cause that weak cell to become dangerously low and lose even more capacity. Moreover, the BMS tends to try to balance out the battery whenever it has time to do so. During a drive or charge, it can't compete with the current flows caused by using the car, but when sitting still for tens or hundreds of hours it'll happily balance the pack at whatever point it's at. But remember; at low SOC, the battery is VERY unbalanced, so balancing it at that point will... cause the battery to be very unbalanced when subsequently charged to full! This is the difference between what's called 'top-balancing' and 'bottom-balancing'; balancing the cell voltages when full and empty respectively. To maximize usable capacity and minimize internal resistance, all EVs use top balancing. Leaving a battery very empty for a long period of time will mess that up and causes apparent capacity loss.

So:
  • Charge to 80%
  • Only charge to full when you need the surplus capacity
  • Do not leave the car at <25% for long periods of time
  • Do not leave the car in hot environments for long periods of time
All this being said; use the car for its intended purpose - don't overdo the babying. A well-designed electric car will do much more for battery longevity than these tips.

So, what does Nissan do wrong?

Now we can go through the history of Nissan's Leaf batteries and find out why they are so bad. In the EV space, Nissan's batteries rank among the very worst batteries, in multiple respects. Their early batteries had poor energy density, longevity and charging performance, and the new ones have only partially fixed the energy density issue - making the rest arguably worse. Why?

A lot of this has to do with history, and actually some really good foresight on the part of Nissan. Around the time they started developing the Leaf - circa 2006 - lithium ion batteries for automotive applications barely existed. A123 and Kokam supplied the bulk of high-spec batteries, but even those wouldn't be enough to satiate their ambitions on the EV market. So they chose to build their own battery factory and called it AESC. AESC in turn chose basically the only chemistry available at the time that had good enough performance for an EV - LMO (lithium manganese oxide). And they standardized on a battery shape that would basically fit in any car; a skateboard-type pack with a hump under the front and rear seats to accomodate more volume of cells. From an engineering perspective, all of these choices were great.

For the time it was developed, the first 24kWh battery pack was fine. It didn't have any kind of thermal management, but considering the modest power demand this was actually fine (for an LMO pack). Battery management was quite advanced for the time, using real-time capacity and performance estimation using table lookup. If I think back at what battery management looked like even in literature around that time - when I was just beginning to delve into the topic for my study - they were doing very well indeed.

But around 2010-2011, two new technologies started to become commercially available in large quantities: NCA and NMC li-ion batteries. NCA in particular combined much, much lower internal resistance with similar energy densities to LCO at the time, as well as a more favorable discharge curve for fast charging. Upcoming competitors in the EV space Panasonic and Yuasa immediately began using these chemistries and produced much better products as a result. Nissan? All I can think is that they were tied to AESC and their technology, tweaking their 24kWh battery offering slightly to cope better with high temperatures, but not really going with the times.

Around 2014 they launched the e-nv200 with the exact same battery modules, but with a fan inside the casing to help with evening out internal temperatures and even actively cooling (albeit poorly) during fast charges. The larger battery casing allowed for more air space inside, improving characteristics even though the battery was otherwise absolutely identical - even the firmware. This paid dividends; even quite intensively used e-nv200s with 60kmi/100Mm on the odometer tend to have 85%+ SOH, something almost unheard of in Leafs.

But Nissan couldn't afford the change in battery shape underneath the Leaf, even though they did increase capacities. It took until 2015 when they launched the 30kWh Leaf with NCA cells - well, mostly NCA. The chemistry change alone accounted for the capacity increase, as well as the much improved fast charging. However, this battery did get a little warmer than the 24kWh. It could have definitely benefited from a fan like the e-nv200.

Time moves on, Nissan needed to compete with the e-golf and other EVs in the late 2010s. They brought out the 40kWh Leaf. Same battery casing, higher capacity, so how did they do it? Well, batteries are constructed out of the active material - the stuff that actually stores the energy, electrodes that conduct the electricity out of the cell and packaging - the stuff that keeps all the gubbins in. They thinned out the electrodes and increased the density and volume of active material to get to the higher capacity. This... massively increased energy density and reduced thermal conductivity of the cells.

This is what eventually led to rapidgate. Mind you, other cars WITH thermal management rapidgated - the e-golf for instance! Nissan must have known this and thought it was OK; after all this issue had already existed for 2 years before the new Leaf came out. The higher inherent internal resistance as well as the increased motor power made it a hot house beyond any other car. And compared to the 30 and 24kWh, the battery shell only got fuller and fuller.

Now we're in 2019. Nissan made the 62kWh Leaf. Almost the same case - they made it very slightly taller so it hangs a little lower under the car. Do you think they increased electrode thickness? Do you think they added a fan inside, or even a heat exchanger? Oh, they increased the fast charging speed to 100kW. Can you guess how that ends?

Nissan engineered themselves into a corner with the Leaf. They started out with great engineering choices - at the time. But then they - either through choice or necessity because of their exclusive involvement with AESC - held onto those choices far beyond the point where they made any sense. High internal resistance, disparate temperatures within the pack and resulting rapid aging, relatively outdated battery chemistry choices and missing thermal management aren't a result of Nissan not knowing what they're doing wrong. It's a result of choices made probably 13 years ago coming back to haunt them.

But, with all this knowledge in hand, we can fix it. MUXSAN will try to fix what Nissan couldn't. Make the Leaf great again!